Split (1)

train · ~6.28M rows (showing the first 165k)

added string	created string	id string	metadata dict	source string	text string	version string
2021-01-26T02:15:50.671Z	2021-01-25T00:00:00.000	231699077	{ "extfieldsofstudy": [ "Computer Science", "Physics" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://doi.org/10.1016/j.jcp.2021.110788", "pdf_hash": "cd233bb2c89edba027a151c46096c748a12d2d39", "pdf_src": "Arxiv", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45300", "s2fieldsofstudy": [ "Engineering" ], "sha1": "cd233bb2c89edba027a151c46096c748a12d2d39", "year": 2021 }	pes2o/s2orc	Applying Bayesian Optimization with Gaussian Process Regression to Computational Fluid Dynamics Problems Bayesian optimization (BO) based on Gaussian process regression (GPR) is applied to different CFD (computational fluid dynamics) problems which can be of practical relevance. The problems are i) shape optimization in a lid-driven cavity to minimize or maximize the energy dissipation, ii) shape optimization of the wall of a channel flow in order to obtain a desired pressure-gradient distribution along the edge of the turbulent boundary layer formed on the other wall, and finally, iii) optimization of the controlling parameters of a spoiler-ice model to attain the aerodynamic characteristics of the airfoil with an actual surface ice. The diversity of the optimization problems, independence of the optimization approach from any adjoint information, the ease of employing different CFD solvers in the optimization loop, and more importantly, the relatively small number of the required flow simulations reveal the flexibility, efficiency, and versatility of the BO-GPR approach in CFD applications. It is shown that to ensure finding the global optimum of the design parameters of the size up to 8, less than 90 executions of the CFD solvers are needed. Furthermore, it is observed that the number of flow simulations does not significantly increase with the number of design parameters. The associated computational cost of these simulations can be affordable for many optimization cases with practical relevance. Introduction In computational fluid dynamics (CFD), the Navier-Stokes equations or variations of them are numerically solved in order to simulate fluid flows. At most of the Reynolds numbers associated to environmental flows and relevant to engineering applications, flows are turbulent, see e.g. Ref. [14]. Various approaches have been developed to numerically simulate turbulent flows, ranging from low-fidelity Reynolds-averaged Navier-Stokes (RANS) simulation to high-fidelity approaches including large eddy simulation (LES) and direct numerical simulation (DNS), see Ref. [63]. Moving from RANS toward DNS, the influence of modeling is reduced, the role of numerics becomes more dominant, and above all, the computational cost increases [68]. In many applications where the fluid flows are involved, we need to optimize different parameters in order to meet (within a margin) a certain set of objectives while satisfying a set of constraints, see Refs. [76,17] and the references therein. The aim of the present paper is to illustrate how Bayesian optimization (BO) [67,21], which has been mostly utilized in the field of machine learning and data sciences, can be applied to different types of optimization problems arising in CFD. Previous application of Bayesian optimization to CFD problems is limited to a few studies, see Refs. [73,44,38]; therefore, the diversity of the applications considered in the present study can reveal the flexibility of the method as well as shed light on its pros and cons. To put the work in context, we shortly review the main approaches which have been employed for the purpose of optimization in CFD. Such approaches are iterative and can, in general, be divided into gradient-based, gradient-free, and gradient-enhanced, see Refs. [46,76]. The gradient-based methods are the appropriate choice when the evaluation of the derivatives of the cost function with respect to the design parameters is computationally efficient. At each iteration, the first-order gradient is needed to determine the searching direction and the second-order gradients (i.e. Hessian matrix) or the approximations of them are required to obtain the optimal step size, see e.g. Ref. [46]. Therefore, depending on the approach, a minimum order of smoothness for the objective function is necessary. The gradients can be computed by automatic-differentiation (finite differences), or more efficiently by solving forward or adjoint sensitivity equations. The use of the latter is recommended when the number of design parameters is more than the number of objectives, see [46,28,76]. As its main advantage, the gradient-based optimizations exhibit fast convergence and can handle large numbers of design parameters. However, it is always likely to be trapped by local optima. Besides this, in the cases of unsteady Navier-Stokes equations the need for saving the forward solution fields which are required to solve the adjoint equations may lead to memory issues. To rectify these issues, extra treatments are needed, see [3,29]. Various types of application of gradientbased methods for optimization in CFD can be found in the literature, including shape optimization [41], optimization of initial conditions for fast transition to turbulence [32], and topology optimization in turbulent flows using the RANS approach [17]. In gradient-free approaches, the simulator (here, the CFD solver) is treated as a blackbox and run to evaluate realizations of the response at the samples of the design parameters. This approach which goes back to Box and Draper [7] can be implemented using different methods, see Refs. [33,20]. The simplest method is the grid search, where a set of manually selected samples are tested to find the optimum. Evolutionary algorithms [2], which mimic the nature's survival by iteratively simulating "selection and mutation" of the design parameters, are also considered as an effective approach; see e.g. Ref. [82] where the algorithm is applied to modify the RANS stress-strain relationship. However, the most frequently-used method of this type in CFD is called the response-surface method (RSM), in which a metamodel or surrogate is constructed using a finite set of training data comprised of parameter samples and corresponding responses. The size of the training data set is determined as a balance between the required accuracy of the surrogate and the cost involved in running the simulator. In many CFD applications, Gaussian process regression (GPR), or the Kriging method [57,27], has been employed to construct the surrogate. As detailed in Section 2, GPR is among the non-parametric surrogates, naturally allows for noisy (uncertain) data, and more importantly, can estimate the uncertainties involved in their predictions. Examples of RSM-based optimization can be found, for instance, in Viana et al. [78]. Contrary to the gradient-based methods, RSM is suitable for finding the global optima, but it suffers from the curse of dimensionality. To reduce the required number of samples for highdimensional parameters yet constructing an accurate response surface, gradient information can be added to the training data, see the review by Laurent et al. [35]. In particular, the gradient-enhanced Kriging (GEK) has been extensively used in the last two decades for optimization in CFD, see for instance, [11,34]. The study by Laurenceau et al. [34] showed that at large number of parameters (=45), GEK outperforms RSM-GPR, while at small number of parameters (=6), inclusion of the gradients makes no significant improvement in the computational performance of the optimization. In the present study, the Bayesian optimization based on Gaussian process regression, hereafter referred to as BO-GPR, is employed, see Refs. [21,67]. This approach can be classified among the gradient-free approaches, although a gradientenhanced version of it has also been proposed, see Ref. [84]. In BO, a sequence of samples for the design parameters is taken which converges to the global optimum. Therefore, as a main difference with the RSM approach, the surrogate in BO is sequentially updated. That is, in fact, the clear connection of BO to the Bayesian formalism, see e.g. Ref. [69]. As detailed in Section 2, at each iteration, the decision about the next sample is taken based on combining the exploitation and exploration, where the latter directly takes into account the predictive uncertainty of the surrogate. This active involvement of the uncertainties in the algorithm is another distinguishing characteristic of the BO approach over RSM methods. In general, the predictive uncertainties are a combination of the uncertain CFD data and the constructed surrogate. The use of the BO is a timely choice considering the application of the uncertainty quantification (UQ) [69] in different aspects of CFD and numerical simulation of turbulence over the last decade, see e.g. Refs. [4,48,85,58,59,61]. In fact, most of those studies are classified as UQ forward problems, the outcomes of which can be nicely employed in BO, which is, in fact, an inverse UQ problem. Having the design parameters defined over an admissible space, the BO can be non-intrusively linked to any CFD solver. This provides a great flexibility noting that a CFD solver is not necessarily equipped with adjoint solvers to compute gradients, and in any case, computing the adjoint sensitivities could be expensive. The application of BO in CFD has been very limited. Talnikar et al. [73] developed a parallel version of BO to minimize the drag in a turbulent channel flow simulated by LES. Nabae et al. [44] used BO to optimize the phase-speed of streamwise traveling wave-like wall deformation for drag reduction in a turbulent channel flow simulated by DNS. More recently, BO was employed by Mahfoze et al. [38] to optimize the blowing amplitude and coverage in order to reduce the friction drag to achieve a certain net-power saving in a zero-pressure-gradient turbulent boundary layer (TBL) simulated by DNS. To study different aspects of BO, a diverse set of optimization problems are considered in the present paper. The objectives, number of design parameters, and CFD solver are different in these problems. This paper is structured as follows. In Section 2, a detailed review is given to the theoretical aspects of the adopted Bayesian optimization methodology. In Section 3, BO is applied to optimize the initial conditions of a set of coupled nonlinear ordinary differential equations so that an energy norm at some later time is maximized. As the second example, shape optimization of a laminar lid-driven cavity flow is studied in Section 4, where the objective is to either minimize or maximize the energy dissipation. Another shape optimization which has practical applications, is considered in Section 5. The objective is to optimize the shape of a wall boundary of a channel, so that the streamwise pressure gradient of the turbulent boundary layer (TBL) over the other wall matches a given profile. Section 6 is devoted to the optimization of a spoiler-ice airfoil profile. The parameters controlling the configuration of two spoilers which are located on the pressure and suction sides of the airfoil are optimized so that the distribution of the pressure on the airfoil surface becomes the same as if a real ice was formed on the front part of the airfoil. In the end, concluding remarks are provided in Section 7. Bayesian optimization Consider the set of design parameters q = {q 1 , q 2 , · · · , q d } which are allowed to vary over the admissible space Q ⊂ R d (note that Q is not the set of rational numbers in this paper). Running the simulator, i.e. the CFD code, realizations for the quantities of interest (QoIs) or responses are obtained, which in general can be noisy. The aim of the optimization is to minimize (or maximize) an objective function r(q) which depends on the simulator outputs over space Q, and find optimal parameter q opt , q opt = arg min q∈Q r(q) . (1) Ideally, a functional dependency between the observed values r and parameters q could be established as [57,27], where, ε denotes the observation noise, and without loss of generality it is assumed to be identically distributed for all samples as N (0, σ 2 ). However, the complexities in the flow solver make acquiring a closed form for f (q) infeasible. Alternatively, a surrogate f (q) can be constructed adopting a non-intrusive approach in which the simulator is treated as a blackbox. To this end, a finite set of training data D = {(Q (i) , R (i) )} n i=1 is employed, where Q and R are respectively samples of q and corresponding realizations of r. In general, the number of the training data, n, is limited due to the cost of running the simulator. The two main components of the Bayesian optimization, i.e. the choice of method for constructingf (q), and drawing the sequence of training samples, are now shortly reviewed. The Gaussian process regression (GPR) [57,27] is employed to buildf (q). In this approach,f (q) is assumed to be random for which a prior distribution in the form of a Gaussian process is assumed. A Gaussian process GP is a multi-variate Gaussian distribution defined as, in which the mean and covariance are respectively given by, In these definitions, E[·] denotes the expected value, ε specify the parameters building the noise structure, and β are the hyperparameters in the kernel function which represents the covariance off (q). Given the training data D along with the associated observation uncertainty, the posterior and posterior predictive off (q) (conditioned on D and ε ) can be inferred. Simultaneously, the hyperparameters β are optimized. The posterior-predictive off (q) at a set of test samples Here, without loss of generality, the mean of f (q) in Eq. (3) is assumed to be zero, R = [r (1) , r (2) , · · · , r (n) ], C(Q, Q ) is a n ×n matrix where [C(Q, Q )] ij = c(q (i) , q ( j) ) and c(·, ·) is a kernel function dependent on hyperparameters β. Moreover, C N is the covariance matrix of the noise in the training data. In this derivation, the noise samples are assumed to be independent and identically distributed (iid) as ε ∼ N (0, σ 2 ). A more general case in which the noise is allowed to be observationdependent has been developed by Goldberg et al. [24], and recently applied for the purpose of uncertainty quantification in CFD by Rezaeiravesh et al. [59]. In practice, the observational uncertainties can, for instance, be due to the convergence limits imposed when solving discretized equations and also finite time-averaging in unsteady flows. Including these uncertainties in the BO-GPR requires considering two points. 1. The computed optimum is, in general, uncertain and it may not be possible to obtain a noise-free value. In practice, when maximizing the acquisition function, the expected value of the posterior predictive for the noisy response can be used, see Ref. [26]. 2. The standard acquisition functions, such as expected improvement, Eq. (8), may not provide adequate exploration over the parameter space, therefore, the global optima may not be found or the convergence rate is very slow [77]. Recently, Letham et al. [36] has proposed a modified version of the expected improvement function for handling the noisy data (there are other acquisition functions for noisy data which have been reviewed in Ref. [36]). Implementing such modified acquisition functions for BO-GPR in CFD will be considered in a future extension of the present work. Now, a short overview is given on how the sequence of samples for q can be drawn from Q so that they converge to q opt . At the k-th iteration of optimization the training data set is D 1: . The decision about the next sample q (k+1) is made through maximizing an acquisition function (ACF) α(q; D 1:k ), see e.g. Ref. [21]. The most popular ACF is the expected improvement (EI) which was first suggested by Močkus [40] and then utilized in BO by Jones et al. [31]. The EI for the minimization problem is defined as, where q † = arg min q∈D 1:k r(q), denotes the best estimate for optimum among the k observations (iterations). Note that the improvement I(q) is random since r(q) is so, see Eq. (2). In the cases of using a standard GPR forf (q), Jones et al. [31] derived a closed-form expression for the EI which reads as, where s(q) = √ v(q), and m(q) and v(q) are the mean and standard deviation of the posterior predictive of the GPR at any q, as given by Eqs. (6) and (7), respectively. Further, (ζ ) and ϕ(ζ ) respectively represent the CDF (cumulative distribution function) and PDF (probability density function) of the standard Gaussian distribution for ζ defined by, In Eq. (9), the first term in the summation specifies exploitation, the use of the best value so far, and the second term identifies exploration parts of the Q where the surrogate has highest uncertainty. The ad-hoc parameter ξ can be used in practice as the controller of the exploitation in comparison with exploration. In the present study, the Bayesian optimization algorithm based on GPR is implemented using the Python libraries GPy [25] and GPyOpt [75]. The developed Python codes are non-intrusively linked to the CFD solvers, Nek5000 [19] and OpenFOAM [83] through appropriate bash drivers. As a result, the whole optimization loop is fully automated. All optimizations started from a random sample for q taken from Q. For constructing the kernel in the GPR surrogates, the Matern52 function is used, see e.g. [57,27]. Note that in general, the choice of the kernel function can affect the convergence of the BO-GPR [62]. Our tests showed that Matern52 kernel could lead to a faster convergence for a set of tested problems. To optimize the GPR hyperparameters, the Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm is utilized, see e.g. [46]. In Appendix B, a short discussion is made on the computational time required for BO-GPR and CFD. Recognizing the convergence of the BO and hence imposing a stopping criterion for that can be, in general, not so straightforward, see e.g. [9,66]. In fact, this is a common characteristic of all gradient-free optimizations. In some cases, mon- 2 at the k-th iteration can help. But noting that the exploration guides the samples to different parts of the admissible space Q, obtaining a small difference between the successive samples, especially for large-dimensional parameters, is difficult to achieve in practice. We have observed that a more reliable measure is tracking the best value of the objective, i.e. r(q † ) where q † = arg min q∈D 1:k r(q). If this measure remains unchanged after a sufficiently large n, then we can consider the optimization to be converged (this is called the identification step, according to Ref. [30]). However, the value of n is not known a priori and it is basically limited by the computational cost of the simulations and available computational budget. Despite this, examining the response surface constructed by the surrogate can help identifying parts of Q with insufficient samples at which the predictive uncertainty of the surrogate is high. Guiding the sequence of samples toward those parts can be, for instance, done by adjusting ξ in Eq. (9). Despite this, the cases studied in Sections 5 and 6 are not directly affected by the above discussion, since the optimum value for the objective is the deviation of a computed QoI from a target value, and is therefore a priori known (albeit potentially not exactly achievable given the design space). In Section 4, where this is not the case, monitoring the best value found when sufficiently many samples are drawn will help. An introductory example In many applications in fluid dynamics, in particular hydrodynamic stability, we are interested in finding the optimum initial condition which leads to fastest growth of the unstable modes and hence fastest transition from laminar state to represented by hollow markers. The predicted mean by the constructed GPR is close to the true gain obtained by solving Eq. (11) by the fourth-order Runge-Kutta scheme given the optimum θ . The normalized error between the GPR and true gains measured in L 2 -norm is equal to 0.0102, 0.0142, 0.1360, and 0.4267 for a equals to 10 −4 , 0.9, 0.9999, and 1.0001, respectively. In constructing the GPR, the σ in the observational noise in Eq. (2) was considered to be zero. The shaded areas specify the 95% confidence intervals built around the posterior predictive mean. turbulence. Kerswell et al. [32] have considered this problem through using adjoint-based optimization. As a "toy" problem to examine their approach, Kerswell et al. [32] introduced the following dynamical system comprised of two ordinary differential equations (ODE), with the initial condition x(0) = x 0 . The objective is to find the initial data x 0 parameterized as x 0 = a(cos πθ, sin πθ) (with fixed a) which maximizes x(T ) 2 2 /a 2 , the normalized energy gain at time T . Therefore, the design parameter is q = θ which is allowed to vary over the admissible range Q = [0, 2). Based the above parameterization, all solutions x(t) start from the same distance x 0 2 = a. Depending on the value of a, the trajectories for x(t) will end up at different attractors of the dynamical system at time T . In the gradient-based optimization adopted in [32], the adjoint equation of Eq. (11) is derived using the Lagrange multiplier approach and is then solved backward from t = T to t = 0 in each iteration of the optimization. Here, we apply the BO algorithm explained in Section 2 to find the optimum θ for different values of a adopted in [32]. For all values of a, the BO starts from the same initial samples θ = 0.1 and 1.5. Our further study showed that the initial samples have no impact on the final optimal values, but they can slightly affect the sequence of samples and hence convergence of the BO. As shown in Fig. 1, the computed optima by the BO-GPR approach are accurate (compared to the true values) and the same as those in Ref. [32] where an adjoint method was employed. This is a significant advantage of the BO-GPR method, noting that solving the same problem with the adjoint-based method could be more involved and for some values of a the convergence of the algorithm would be a bit difficult to reach due to numerical instabilities. On the other hand, the convergence in all cases shown in Fig. 1 is achieved with less than 15 samples. The history of the optimization is presented in Figs. A.15 and A.16 in Appendix A. Shape optimization of a lid-driven cavity We now move to problems involving the Navier-Stokes equations. Consider an incompressible two-dimensional cavity flow, where all walls are at rest except the upper wall (lid) which moves with the constant velocity U 0 in the positive horizontal direction, x. The aim is to optimize the shape of the left and right walls so that the energy dissipation is either minimized or maximized with the constraint of keeping the total volume of the fluid confined in the cavity fixed and also retain the length and position of the upper and lower walls. Therefore, the objective is to find the wall shape which either minimizes or maximizes, keeping dx = m fixed. In Eq. (12), S is the strain rate tensor, where S ij = (∂u i /∂x j + ∂u j /∂x i )/2 for i, j = 1, 2 and u i denotes the i-th component of the velocity vector. In order to apply the Bayesian optimization of Section 2, we parameterize the geometry using a third-order polynomial, x = a 0 +a 1ȳ +a 2ȳ 2 +a 3ȳ 3 , wherex = x/l, ȳ = y/h, l is the length of the upper and lower walls, h denotes the height of the side walls, and, x and y specify horizontal and vertical coordinates, respectively. Considering the origin of the coordinates at (x = 0, y = 0), the constraint of keeping l fixed leads to, a 0,L = 0 and a 3,L = −(a 1,L + a 2,L ) for the left wall, and, a 0,R = 1 and a 3,R = −(a 1,R + a 2,R ) for the right wall. As a result, four design parameters are left to optimize, q = (a 1,L , a 2,L , a 1,R , a 2,R ). The admissible space of q is the tensor-product of Q a 1,L = Q a 1,R = [−0.7, 0.7] and Q a 2,L = Q a 2,R = [−0.5, 0.5] which are arbitrarily chosen to allow for different wall shapes to occur. For any sample taken from Q, first the shape of side walls is determined. At this stage, the height of the walls, h, is adjusted to satisfy the constant-volume constraint. Then, the mesh is generated using gmsh [23]. The simulations are performed using Nek5000, an open-source spectral-element-based flow solver developed by Fischer et al. [19]. In Nek5000, the weak form of the incompressible Navier-Stokes equations are integrated in time on Gauss-Lobatto-Legendre (GLL) points through expressing the velocity and pressure fields in terms of Lagrange interpolants of the Legendre polynomials, see the details in Deville et al. [16]. To avoid the discontinuity in the velocity at the top corners where the moving lid and the side walls meet, the smoothing suggested in Ref. [47] is applied. In all simulations, 30 elements in both x and y directions are considered with 10 GLL points in each spatial directions per element. Simulations start at t = 0 from a uniform velocity field and continue until t = 50l/U 0 , which is chosen to ensure the steady-state value for dissipation (12) has been reached. For both minimization and maximization problems, the BO starts from same four initial samples for q that are Then, decision about next sample of the design parameters is taken. This algorithm is repeated until optimal design parameters are obtained. For this example we only consider steady laminar flow, however an extension to turbulent flow is possible in a straightforward way. For a cavity flow at Reynolds number Re = U 0 l/ν = 2000 (ν is the kinematic viscosity), Fig. 2 shows the obtained shapes of the cavity with minimum and maximum energy dissipation. The obtained shapes are consistent with the flow physics, for instance, in case of the maximum dissipation two vortices adjacent to the lid are generated which prevent the fluid underneath to receive a high velocity. The obtained optimal shapes can be compared to results reported by Nakazawa [45] using an adjoint method for optimization, although the setups in the two studies are slightly different. In Ref. [45] the lid velocity was assumed to vary with x (it might be hard to interpret it physically) and the Reynolds number Re was higher than the critical value of about 8000 [6] (although no treatment for turbulence modeling was mentioned). For the case of minimizing the dissipation, the shape in Fig. 2 (top) agrees well with [45]. But we found the shape in Fig. 2 (bottom) has a slightly higher dissipation compared to the maximum-dissipation case reported by Nakazawa [45] in which both side walls were deformed towards the inside of the cavity. This can be an indication for the ability of the Bayesian optimization in exploring different possible geometries and finding the global optimum without quickly getting stuck in a local optimum. Following the discussion in Section 2 regarding the convergence of the BO, the reported optimal shapes in Fig. 2 are based on the parameters which were found to be the best values within a sample size of n = 50. In fact, the parameters for minimizing and maximizing the dissipation were already found at the 20-th and 15-th iterations, respectively. However, the extra samples were taken to ensure the parameter space is sufficiently explored. It is recalled that the definition of the convergence criterion in Bayesian optimization is not straightforward, see [9,66]. Therefore, in order to impose an appropriate stopping criterion, the whole history of the samples has to be considered. The history of the BO-GPR for the optimization cases of the cavity flow is plotted in Figs. A.17 and A.18 in Appendix A. Motivation Studying in a controlled fashion pressure-gradient (PG) turbulent boundary layers (TBLs) is important since they are present in a wide range of industrial applications, for instance, the flow around the curved boundaries such as car bodies, airplanes, airfoils, etc. The so-called Clauser pressure-gradient parameter β [12,13] is widely accepted as one of the most relevant non-dimensional parameters for PG TBLs [64], which is defined as, where δ * , τ w and dP e /dx are, respectively, the displacement thickness, the magnitude of the wall-shear stress and the pressure gradient along the boundary-layer edge in the streamwise direction. Since the history of the PG β(x) has a significant impact on the characteristic of the TBLs, flows with a constant β in the streamwise direction are very relevant for the study of PG TBLs [5,18]. Despite their importance, there are few studies in the literature on flows with constant pressure gradient, mainly due to the difficulty of setting up this configuration with sufficient accuracy. In experiments, a desired β distribution at the edge of a TBL is usually achieved by changing the wall shape of wind tunnels through a trial-and-error process, see e.g. Sanmiguel Vila et al. [64]. This is, in fact, very time consuming, since β is needed to be measured in the streamwise direction at each iteration. Consequently, it is very difficult to achieve a long constant-β region in the streamwise direction. If a proper shape of the wall could be known prior to an experiment, not only time and cost could be saved, but also a more accurate constant-β distribution would be eventually obtained. The objective of this section is to efficiently obtain the optimal shape of the upper wall of a channel using the Bayesian optimization described in Section 2 so that a target β along the edge of the TBL over the lower wall is achieved. We consider both zero-pressure-gradient (ZPG) and adverse-pressure-gradient (APG) TBLs with constant values for the target β. Besides these, an optimization is considered where the target β distribution varies in the streamwise direction. Such distribution is taken from the flow around a NACA0012 airfoil. This demonstrates the applicability of the BO method to a wide variety of target distribution for the pressure gradient β. Problem setup The considered flows have high Reynolds number and are hence turbulent. Therefore, we perform incompressible Reynolds-averaged Navier-Stokes (RANS) simulations using the open-source software OpenFOAM [83]. In particular, the simpleFoam solver is used which is based on the SIMPLE scheme [55] for velocity-pressure coupling. The k − ω shearstress transport (SST) turbulence model introduced by Menter et al. [39] is used, since for TBLs this model has shown better agreement with experimental data compared to other models, see e.g. Vinuesa et al. [79]. Note that the default values are used for the RANS model coefficients. A two-dimensional domain is considered to mimic the conditions that may exist in a wide wind tunnel, see the schematic in Fig. 3. The domain has a fixed-shape flat-plate lower wall and an upper wall whose shape is subject to optimization. The domain length in the streamwise direction and the domain height at the inlet are denoted by L x and L in y , respectively. In particular, we consider L x /δ in 99 = 1000 and L in y /δ in 99 = 40, where δ in 99 is the 99% boundary layer thickness at the inlet. In general, δ 99 is defined as the distance from the wall at which the mean streamwise velocity becomes 99% of the local edge velocity, U e . In the present study, U e represents the local maximum velocity of the lower-wall's TBL. Note that the diagnostic-plot scaling method proposed by Vinuesa et al. [80] is not applicable to RANS due to the lack of information about the instantaneous velocity. Note also that δ in 99 pertaining to the upper and lower TBLs are the same since the inflow condition for the ZPG-TBLs over the upper and lower walls are taken to be the same. The inflow profiles are assigned based on the DNS data of ZPG-TBL from Schlatter and Örlü [65], see the details in Ref. [43]. The upper wall is constructed by using a spline function to smoothly connect the intersection of the channel inlet and the upper wall, i.e. x = 0, y = L in y , and d control points. The control points are equally spaced in the streamwise direction. The heights y of these points are subject to optimization, therefore the (x, y) coordinates of the i-th control point are defined as (iL x /d, q i ) where i = 1, 2, · · · , d, see Fig. 3. When applying the BO of Section 2, for each sample of q a new geometry for the domain is obtained for which a structured hexahedral mesh is generated using OpenFOAM standard meshing function, blockMesh. In order to resolve the near-wall region of the TBL, a fine mesh resolution is needed near the wall. The mesh is constructed in such a way that y + 1 = u τ y 1 /ν remains below unity at the walls. Here, y 1 is the distance of the first computational cell center from the lower wall, ν is the kinematic viscosity, u τ = √ τ w /ρ is the wall-friction velocity with τ w and ρ respectively denoting the wall-shear stress and fluid density. The aim of the optimization is to obtain a pressure-gradient distribution sufficiently close to a given target β t . Therefore, we can formulate a minimization problem whose objective function is defined as an error between the computed β distribution and the target distribution, where, · L 2 denotes a standard L 2 -Lebesgue norm evaluated over the domain of x. Note that regions close to the inlet and outlet of the domain have to be excluded when evaluating r(q) to avoid any inlet or outlet effects on the β distributions, see Fig. 3. The remaining range of x over which the norm in Eq. (14) is computed is referred to as optimization range which may be chosen adequately for each optimization case as explained in Section 5.3. For additional details about the problem setup, refer to Ref. [43]. Optimization As explained in the Section 5.1, optimization is conducted for three different distributions of β: constant values of zero (ZPG) and one (APG), and a non-constant distribution corresponding to the flow around a NACA0012 airfoil. These optimization cases are named as Constant-0, Constant-1 and Wing, respectively, and are discussed in the following subsections. The conditions of these cases are summarized in Table 1. Prior to the optimization, numerical experiments are conducted to decide the number of control points, admissible range of parameters, optimization range, and the total number of iterations. Constant-β distribution For the Constant-0 case, essentially a correction for the displacement effect of the growing boundary layers is expected, which yields a comparably simple shape of the upper wall. Thus we chose to only use 2 points along the length of the 14), the validity of the computed optimal parameters can be examined. The results of the optimization are shown in Fig. 4. It is observed that the global minimum of r is found in the considered admissible space and the associated β distribution is highly accurate (within ±0.005 of the target value of zero). The 95% confidence interval of r(q), which is shown in Fig. 4(d), is smaller close to the optimum because of the higher number of samples. Similar to the Constant-0 case, optimization of the Constant-1 case, i.e. for β t = 1 is conducted. Since the target distribution of β is constant but larger than zero, the optimum shape of the upper wall is expected to be only moderately complex; thus, 4 control points are used. The optimum is found at the 40-th iteration, whose results are shown in Fig. 5. The optimal shape of the upper wall is steeper than that of the Constant-0 case, since the target β t is higher. In the Constant-1 case, the computed optimal β distribution remains almost constant in the optimization range and deviates from the target β t by less than ±0.02. It is noteworthy that since the inflow is taken from a ZPG-TBL, a development length is needed to reach β = 1. This development length is chosen to be 300δ in 99 . We observed that forcing a faster development of the β through imposing a shorter development length, could lead to the separation of the TBL over the upper wall and hence make it more difficult to achieve a constant β for the lower-wall's TBL. As separation is typically unwanted in a wind-tunnel setting, we prefer to have a longer development length avoiding this issue. To validate the results of the constant-β TBLs obtained from the optimization, the streamwise development of the skinfriction coefficient c f and the shape factor H are compared to the reference data, see Fig. 6. The skin-friction coefficient and the shape factor are defined as c f = 2(u τ /U e ) 2 and H = δ * /θ , where δ * and θ respectively denote the displacement and momentum thickness. As the reference data for the constant-β TBLs, the correlations based on the data compiled by Vinuesa et al. [81] and the experimental data from Sanmiguel Vila et al. [64] are adopted. The Reynolds number based on the momentum thickness, Re θ = U e θ/ν, is used as the horizontal axis of the plots in Fig. 6, so that the direct comparison is possible. Clearly, the optimized cases show excellent agreement with the reference data in the region where β is approximately constant (Re θ 3000). The Constant-1 case shows a transition from β = 0 to β = 1 profile because of the development of β in the streamwise direction, see Fig. 5(b). Note that the experimental β distribution of Sanmiguel Vila Note that the range of the vertical axis in Fig. 4(a) and Fig. 5(a) are set differently for illustration purposes. et al. [64] has a variation in the streamwise direction (about ±15%) because of the difficulties to set the constant-β distributions in the experiments, as mentioned in Section 5.1. As a conclusion, not only the applicability of the optimization method to obtain constant-β distributions is confirmed but also the validity of the resulting constant-β TBLs is approved. Non-constant-β distribution The same optimization procedure can be applied when β t is not constant in the streamwise direction. The target β distribution is taken from the flow around a NACA0012 airfoil. Note that due to the geometrical symmetry with respect to the airfoil chord, β distributions of the suction and pressure sides are identical for NACA0012 at zero angle of attack. The distribution of the target β is taken from the numerical simulation by Tanarro et al. [74]. Since the target β is originally defined based on the normalized coordinate x/c, where c denotes the chord length, it has to be mapped to the channel normalized coordinate in the streamwise direction, i.e. x/δ in 99 . Here, the original data are mapped to the first 95% of the channel length, i.e. 0 < x/c < 1 → 0 < x/δ in 99 < 950, to avoid any effect of the outlet boundary. Furthermore, the first 15% of the channel is excluded from the optimization range because the reference data close to the leading edge is not available. According to Ref. [74], this is due to the difficulty of calculating δ 99 with the diagnostic-plot scaling method near the leading edge since the flow is not fully turbulent. To be able to achieve the relatively complex β distribution along the edge of the TBL at the lower wall, the parameterization of the upper wall of the channel is done with 8 control points, see Table 1. The results of the optimization are shown in Fig. 7. The optimum is found at the 52-nd iteration with associated β being found very close to the target distribution. In fact, the largest deviation which occurs around x/δ in 99 ≈ 850 is less than 0.54. This promising result proves that the Bayesian optimization method can also be used for more complex pressure-gradient distributions. This is, in fact, very useful for future wind-tunnel experiments to study wall-bounded turbulent flows with arbitrary β distributions, noting that the common approach in practice is based on trial-and-error, which can be inaccurate and time consuming. The results of the present section have been obtained by RANS, which may have its own deficiency in fidelity compared to e.g. large eddy simulations or even direct numerical simulations. However, the underlying flow simulation method does not affect the performance of the optimization procedure. Therefore, the current results could be even achieved using higher-fidelity methods, or even -if properly implemented -in an experimental setup. Optimization of a spoiler-ice airfoil model Finally, in this section we present a more applied flow case which shows the potential of the BO framework also for industrial flow cases where optimal designs are sought. Motivation When operating in cold climate, the performance of the wind turbines can be reduced by icing [8,72]. In an extreme condition, heavy icing can even lead to a complete stop of the turbine. Icing is also a risky condition for airplane wings since it may induce flow separation at a small angles of attack, which then might lead to stall and consequently loss of lift. Ice accretion is a complex process which depends on both aerodynamics and thermodynamics. The process is affected by many parameters, for instance, ambient temperature, surface properties, relative speed of the airfoil, and the time span over which the icing event takes place. As a result, numerous possibilities for an ice profile exist. Bragg et al. [8] categorized the leading-edge ice on the airfoils into four types: roughness, horn, streamwise, and spanwise-ridge ice. In this study, we focus on the horn and streamwise ice, see Fig. 8. The main characteristic of the horn ice is the presence of the large protuberances which induce a large flow separation downstream of the ice, and hence dramatically affect the aerodynamic performance, including the lift, drag and moment coefficients. On the other hand, streamwise ice is smoothly formed along the streamlines and is less dangerous in the sense of having adverse effects on the aerodynamic performance. Fig. 9. Original iced airfoil used in the present study. The ice is formed on a clean NACA64618 airfoil, which is shown with a black solid line. The original ice shape is illustrated with a red solid line and a filled gray region. Coordinates are non-dimensionalized by the chord length of the clean airfoil, c. Note that the origin of the x coordinate is set as the leading edge of the clean airfoil, and this is the convention throughout Section 6. Modeling an iced airfoil is a challenging task in both CFD and experiments since the ice profile essentially has infinite degrees of freedom, noting that every ice shape is unique and consequently the investigations would be case-dependent. In order to generalize the icing assessments, one needs to find specific characteristics to parameterize the geometry such that various ice profiles can be described with the reduced-order model. The simplified "simulated-ice" concept has been considered as a representative of the actual ice profile. In a simple representation, Papadakis et al. [50][51][52][53][54] approximates the horn-shaped ice from glaze ice conditions as spoilers. As a result, the profile of the horn is reduced to a thin plate and its effect on the aerodynamic properties is parameterized by just its height, angle, and location. In particular, the thin plates which are called the "spoilers" and have the same height as the original ice can be used to reproduce the aerodynamic performance of the ice, see Fig. 8. The resulting model is called the "spoiler-ice" model. Although the height, angle, and location of the spoilers have been considered as important parameters which affect the airfoil aerodynamic performance, see e.g. Refs. [52,51], Tabatabaei et al. [70,71] suggested that the thickness of the spoilers should be also taken into account as an effective parameter. Although parametric studies of the spoiler-ice model have been conducted to investigate associated effects on the aerodynamic performance, see e.g. Refs. [52,51], it is not yet clearly known which configuration of the spoiler is appropriate to accurately represent the aerodynamic performance of an arbitrary given real ice profile. The BO method introduced in Section 2 can be utilized to find the optimal parameters of the spoiler configuration. Although the spoiler-ice concept has been previously used mainly for the horn ice, see e.g. Ref. [53], here we also consider to apply the idea to the streamwise ice. Therefore, an original arbitrary ice shape is chosen so that it has both horn and streamwise ice on the suction and pressure side of the airfoil, respectively. Two spoilers on the suction and pressure sides (the upper and lower spoilers, respectively, shown in Fig. 10(a)) are used to obtain a similar effect of the original iced airfoil. Since a spoiler-ice with the same height as the original ice shape has been used in the previous studies [51][52][53], here the optimization is first conducted with the constraint of keeping the heights of the spoilers fixed. However, it is shown in Fig. 14 that the results of the optimization without such a constraint are in better agreement with those of the original ice. It will thus be shown that a more complete parameter optimization may indeed be relevant for accurate simulations. Problem setup In this section, we describe the setup and implementation of the icing flow case, including the basic simulation, the meshing and optimization loop. Original ice profiles and the spoiler ice As clean airfoil, a NACA64618 profile is considered, which has been used in a 5MW NREL (National Renewable Energy Laboratory) wind-turbine blade. For the purpose of the optimization, an arbitrary original ice shape is required as a reference which is taken from the ice formed on the leading edge of the airfoil studied in Ref. [22], see Fig. 9. The ice on the upper (suction) and lower (pressure) sides of the airfoil is of the horn and streamwise types, respectively. As mentioned in Section 6.1, the flow around the original iced airfoil can be modeled by two spoilers each located on one side of the airfoil. For the purpose of optimization, the configuration of the spoilers is controlled by seven parameters: heights, angles and widths of the upper and lower spoilers, h u , h l , θ u , θ l , w u and w l , respectively, and the displacement of the lower spoiler, (d l ), see Fig. 10(a). The displacement of the upper spoiler is not taken into account since the original ice shape on the suction side is horn ice, which has a clear connection between the position of the horn of the original ice and the position of the spoiler-ice. On the other hand, the pressure side of the original ice is of streamwise type, hence it is not, in general, clear where the spoiler-ice should be located. To rectify this, the displacement parameter d l is needed for the lower spoiler to optimize its position as well as the other parameters. Prior to the optimization, several numerical experiments are conducted to decide the admissible range of the parameters and the total number of iterations. Fig. 10(b) shows the configuration of the spoilers when the design parameters are at either end of their admissible range. Note that the upper limit of the admissible range of the heights, h u and h l , is the same as the highest point of the original ice profile. Numerical simulations Numerical simulations are conducted considering a cross section of an iced-airfoil inside a (virtual) wind tunnel. Twodimensional steady and incompressible RANS are performed using OpenFOAM with the same solver settings and turbulence model as mentioned in Section 5.2. The computational domain, see Fig. 11, is made based on the test section of the Minimum-Turbulence-Level (MTL) wind tunnel, which is a closed-loop circuit tunnel housed at KTH. Additional information about the MTL wind tunnel can be found in e.g. Ref. [37]. The commercial software Ansys ICEM CFD 18.2 [1] is used for generating high-quality meshes in the relatively complex geometries involved in the problem. The mesh is constructed to be sufficiently fine, and in particular y + 1 is kept below 1 all around the airfoil. The resulting total number of computational cells is about 1 million. Note that the number of computational cells slightly differs for the simulations associated with a particular configuration of the spoiler ice. To put the computational cost in perspective, each flow simulation takes about one hour running in parallel on 24 cores of Intel Xeon E5-2690v3 Haswell processors. To measure the similarity of the aerodynamic performance of a spoiler-ice model and the actual ice profile, the pressure coefficient (c p ) distribution around the airfoil is considered, since it is directly related to the aerodynamic characteristics such as the lift and drag coefficients. The objective function in the optimization is thus defined as the L 2 -norm of the error between the computed c p distributions on the suction and pressure sides of the airfoil, and the corresponding distribution where 2 f =1 specifies the summation over the suction and pressure sides of the airfoil and c p,t denotes the c p distribution of the original iced airfoil. Note that the objective function is evaluated only downstream of the original ice, meaning that the optimization range is set to 0.1 < x/c < 1 for both suction and pressure sides of the airfoil. Prior to the optimization, a simulation of the original iced airfoil is conducted to obtain the target pressure-coefficient distribution c p,t . The resulting velocity contours and pressure-coefficient distribution are shown in Fig. 12. Because of the ice formed on the leading edge, there are separation bubbles right after the ice on both suction and pressure sides. Moreover, the c p near the leading edge exhibits sudden changes in the streamwise direction. Note that x/c 0.1 is still on the ice surface (in fact, the ice ends at x/c = 0.086 and 0.0975 on the suction and pressure sides, respectively). Therefore, the optimization range is set to 0.1 < x/c < 1, as represented in Fig. 12. Optimization Optimization of the design parameters is conducted with two configurations: spoilers with fixed height and flexible height. As mentioned in Section 6.1, glaze ice has been simulated by spoiler ice whose height is the same as the highest point of the original ice [51][52][53]. As a result, one optimization is conducted with the constraints of having the heights of the spoilers fixed, and the case is referred to as "fixed-height optimization". Imposed by these constraints, the number of design parameters is reduced to d = 5. Since the height of the spoiler ice is a relevant parameter on the aerodynamic performance [72,15], the second optimization is conducted without the above constraints, and the case is called "flexibleheight optimization". Allowing the height of the spoilers to change, the number of design parameters becomes d = 7. The In Fig. 13, the separation bubbles and vortex structures are illustrated as they appear behind the ice profile and the spoiler ice on both pressure and suction sides. The general structure is in agreement with the previous studies [51,71], where the primary and secondary vortices were observed around the spoiler-ice through numerical simulations. In fact, it is not even necessary that the flows downstream of the original ice and the spoiler ice have similar vortex structures since the primary purpose of using the spoiler-ice is to model the aerodynamic characteristics on an iced airfoil. The flexibleheight optimization successfully yields a reattachment point similar to the original iced-airfoil case, while the reattachment point is pushed further downstream for the fixed-height optimization. This has a direct influence on the c p distribution, see Fig. 14. In fact, the c p distribution of the fixed-height case has four times larger deviation from the target distribution, i.e. r in Eq. (15), compared to the flexible-height case. It is also observed that for the fixed-height optimization, the optimal Fig. 12), and the optimized spoilerice models. The corresponding flow fields are represented in Fig. 13. The solid and dotted lines specify the suction and pressure sides of the airfoil, respectively. The black lines represent the target distribution, c p,t , which is also shown in Fig. 12(b). Note that the optimization range is 0.1 < x/c < 1, which specifies a region downstream of the original ice. The distribution of the clean airfoil is also presented for reference. parameters q opt are located at the borders of the considered admissible space. This means that the global optimum could change value if the admissible space would be different. Recall that the admissible range of the parameters are limited due to the geometrical and meshing constraints. In contrast, for the flexible-height optimization, the optimum q opt is found within the admissible space and leads to a better agreement with the result of the original iced airfoil. Therefore, despite being used in the literature, keeping the heights in the spoiler-ice model fixed is not a sufficiently accurate assumption, and considering also the height as an design parameter leads to higher fidelity in the results. The obtained c p distribution from the flexible-height optimization is almost identical to the target distribution within the optimization range except on the pressure side around x/c = 0.2. The slight deviations are due to the fact that the original ice on the pressure side is of the streamwise type, which is basically more difficult to be represented by a spoiler-ice model. The BO has also shown to be an efficient optimizer even for this applied CFD case. The coupling to the CFD code and meshing software are quite straightforward, and can be done non-intrusively. Note that no derivatives of the target function are needed, which makes the present approach suitable for general CFD applications. Summary and conclusions The Bayesian optimization based on Gaussian process regression (BO-GPR) has been applied to different CFD problems ranging from purely academic to industrially relevant setups, using state-of-the-art simulation methods. The diversity of the examples with different numbers of design parameters helps to better understand the applicability and performance of the BO-GPR approach which has not been frequently used in CFD and turbulent flows despite being a primary choice in other fields, e.g. data sciences. The use of the BO-GPR is straightforward noting that the Bayesian optimization is among the gradient-free approaches and hence only requires forward evaluation of the quantities of interest (QoIs), i.e. running a CFD code. In the BO, a sequentially-updating set of samples is taken from the space of the design parameters, such that the sequence converges to the global optimum after a finite number of iterations. A main advantage of the BO-GPR is its versatility, meaning the algorithm is non-intrusive and can be utilized with any CFD code and for any number of design parameters. The use of a GPR surrogate provides the possibility of estimating the confidence in the values of the cost function over the admissible space of the parameters. What is needed to create the optimization loop is to automatize the whole process through appropriate script-driven interfaces between the BO and CFD codes. The software developed in the present work is based on GPy [25] and GPyOpt [75] libraries and is available online. 1 Although the optimization problems studied in the present study contain as many as 8 design parameters, the maximum number of iterations to ensure obtaining a reliable global optimum is no more than 90. This number should be compared with the number of function evaluations necessary e.g. for adjoint-based optimizations. More interestingly, by changing the number of design parameters from 2 to 8, the number of required flow simulations does not significantly increase. Therefore, the computational effort for BO-GPR is observed to be affordable for many practical applications, especially if the RANS (Reynolds-averaged Navier-Stokes) approach is employed for the simulation of turbulence. However, similar to other gradient-free approaches, the BO-GPR may become inefficient for a large number of design parameters. As shown for instance by Wu et al. [84], the efficiency of BO-GPR can be improved by adding sensitivity information from the gradients of the cost function computed by adjoint methods. Application and analysis of such modifications will be considered in the future extensions of the present study. We demonstrate the BO-GPR approach on two comparably simple problems in Sections 3 and 4 for an analytical test problem and a lid-driven cavity. In both cases, our approach could find the global optimum, as compared to previous liter-ature results of the same cases. The optimization required about 20 iterations for the four-dimensional problem, exploring the complete parameter space in an efficient manner. It can be observed that the Bayesian technique quickly focuses on the regions where the potential optima are located. In addition, the uncertainty information about the optima can be retrieved from the approach. A more physically relevant problem is considered on the example of the shape optimization of the upper wall of a two-dimensional channel. Here, the goal is to obtain a given target pressure-gradient (PG) distributions in a turbulent boundary layer developing over a flat wall, which is achieved by adjusting the location of the upper wall in a variableheight channel flow. The flow is simulated using the open-source finite-volume-based solver OpenFOAM [83] adopting the RANS approach to model the turbulence. Accurate results are obtained for the target pressure gradient regardless of being constant or varying in the streamwise direction. Different numbers of design parameters 2, 4, and 8 are considered for which 25, 50, and 80 simulations are respectively needed to find the corresponding global optimum. The results of the optimum configuration for the constant-PG flows show an excellent agreement with the desired pressure-gradient distribution, and as consequence also with the correlations and experimental data reported in the literature for integral quantities of pressure-gradient boundary layers. In addition, promising results for the non-constant target PG distribution suggests that the BO-GPR approach can be applied to various practical applications including the design of inserts in wind tunnels. This is of special interest noting that the current approach for such designs is based on trial and errors, with limited accuracy and robustness. Finally, we also apply the Bayesian optimization to an industrially-relevant case, in an effort to show that the approach can readily be used in an applied context. The goal is to optimize the configuration of the spoiler ice which is a simplified yet useful model for the actual ice profiles formed on the airfoils in cold conditions. The objective is to match, within a small margin of error, the resulting pressure distribution of an airfoil with modeled ice with that of the same airfoil but with actual ice profile. The RANS simulations using OpenFOAM for design parameters of dimension 5 and 7 are examined for which the maximum number of iterations in the optimization is found to be equal to 50 and 90, respectively. The encouraging results of the optimization prove the applicability of the BO-GPR framework to a relatively complex parameter optimization and also re-confirm the validity of the spoiler-ice models. It is also observed that keeping the heights in the spoiler-ice model fixed, as it is convenient in the literature, would lead to inaccurate flow simulations. Moreover, although the spoiler-ice concept has been mainly used in the literature for modeling the horn ice, see Refs. [51][52][53], the present study reveals that the model can also be applicable to the streamwise ice, conditioned on adopting the optimal parameters. The present study can be extended in several ways. Our future attempts will include the use of scale-resolving approaches for simulation of the turbulent flows, instead of RANS which has been adopted here. In this way, even the effect of time-averaging on transient data can be considered. In addition, larger numbers of design parameters could be inquired, to identify the impact of dimensionality on the complete optimization problem. To make the BO more affordable, at least two strategies can be considered. As it is suggested for instance in Ref. [42], linear or non-linear mappings (feature mappings) can be introduced to map the high-dimensional space of the design parameters to a reduced-dimension space where the optimization of the acquisition function can be performed. As the second remedy, multifidelity modeling (MFM), see Refs. [56,49,60], can be combined with BO. In MFM, an accurate surrogate of the objective function can be constructed combining a large number of low-fidelity CFD simulations with only a few high-fidelity ones. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. These plots are associated with the optimization problems represented in Fig. 2. Fig. A.19. Impact of increasing the number of samples on the surrogate of the objective function for the Constant-0 case in Section 5. The plots are associated with those presented in Fig. 4. Fig. A.20. Convergence of the samples in the BO-GPR for the Constant-0 case in Section 5. The plot is corresponding to the optimizations shown in Fig. 4.	v3-fos-license
2020-10-29T09:02:37.877Z	2020-09-30T00:00:00.000	225127067	{ "extfieldsofstudy": [ "Biology" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://academicjournals.org/journal/AJFS/article-full-text-pdf/64FE1E464904.pdf", "pdf_hash": "149d4b7c0b61530ce82d53053fa2c2d20a1a549c", "pdf_src": "Adhoc", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45301", "s2fieldsofstudy": [ "Agricultural And Food Sciences" ], "sha1": "e6d7038e81d360ad9d259492eeefe224d13aaca7", "year": 2020 }	pes2o/s2orc	Consumer acceptability and nutrient content of Westwood (Cirina forda) larva-enriched Amaranthus hybridus vegetable soups Insects account for the greatest amount of biodiversity in forests with over 1,400 species reportedly eaten as human food, but are the least studied of all fauna. Studies have shown that they may contribute significantly to livelihoods in both rural and urban areas. This study was carried out to assess the consumer acceptability and nutrient content of Cirina forda larva-enriched vegetable soups. Dry C. forda (CF) larva was purchased from a market in Burkina Faso and refrigerated at -4°C. Four vegetable soup samples (plain vegetable soup; Egusi soup; Vegetable soup+CF; and Egusi soup+CF) were prepared traditionally. Dry CF larva and the four vegetable samples were analysed using standard AOAC methods, while acceptability of the soup samples was carried out using 9-point Hedonic scale. Moisture content of CF was 3.98 g while that of soups ranged from 59.78 to 77.14 g /100 g. C. forda larva contained 54.38 g protein and 16.81 g fat which were rich in essential amino acids and unsaturated fatty acids respectively; and high in macro-minerals. Nutrient content of vegetable soups enriched with CF larva were significantly higher (p<0.05), and more acceptable than un-enriched ones; with Egusi soup+CF larva being the most acceptable. C. forda larva is rich in both macro and micronutrients and generally acceptable to consumers. C. forda larva consumption should be popularized as means of improving dietary diversity, nutrient intake and overall health of humans. Key words: Cirina forda larva, edible insects, enriched vegetable soups, consumer acceptability, nutrient content. INTRODUCTION Protein-energy malnutrition is still an important public health issue in the developing countries of the world, especially in Africa, with its attendant problem of morbidity, mortality, stunted growth, and impaired neurobehavioral development in children (Iombor et al., 2017). The common sources of animal protein such as meat and eggs are becoming more and more costly and out of reach of the common man (Headey et al., 2018; *Corresponding author. E-mail: tadepoju@com.ui.edu.ng. Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License Ebenebe et al., 2020), hence, alternative cheap sources of protein of high biological value need to be explored in order to meet human body protein requirements. Insects remain a vital and preferred food and essential source of protein, fat, minerals and vitamins in many developing countries and various cultures throughout the world . Research studies have documented that some edible insects have nutritional value that is comparable with that of meat and fish (Braide et al., 2010); and are often a welcome source of protein in the absence of meat from vertebrates (Sponheimer et al., 2005). Edible insects are important dietary components in many cultures where they contribute significantly to protein, fats, and micronutrient intake of consumers (Akinnawo and Ketiku, 2000;Anvo et al., 2016); and are not used as emergency food to ward off starvation, but included as normal part of the diet whenever available (Adeoye et al., 2014). Insects have played important part in the history of human nutrition in Africa, Australia, Asia and the Americas (Jongema, 2015;Dossey et al., 2016;Koko and Mariod, 2020). Several studies have shown that edible insects contain appreciable amount of proteins of good quality and high digestibility, with beneficial amino acid and fatty acid profile comparable to conventional livestock and fish (Braide et al., 2010;Igwe et al., 2011;Iombor et al., 2017). In some African countries such as Malawi, Zambia and Tanzania, malnutrition in children have been fought through flour made out of dried caterpillars. Pregnant and nursing women as well as anaemic patients also eat caterpillar species high in protein, calcium and iron (Igbabul et al., 2014), hence, the potential of insects needs to be considered in malnutrition alleviation strategies in Nigeria. Insects of nutritional importance include (but not limited to) grasshoppers, caterpillars, beetle grubs and (sometimes) adults, winged termites, Cirina forda larva, and a variety of aquatic insects (Adeoye et al., 2014). Among the edible insect species, C. forda (Westwood) larva has been reported to have the potential to provide substantial amount of protein, minerals and polyunsaturated fatty acids to the diets which are usually deficient in animal protein (Akinnawo and Ketiku, 2000;Adepoju and Daboh, 2013). The larva of this insect is widely consumed in the Northern, Central and South-Western parts of Nigeria (Ogunleye, 2006). The insect larva is processed into dry form and consumed as a delicacy served as snack or as an essential ingredient in vegetable soups along with carbohydrate food in Southern part of Nigeria and many homes in Africa (Omotoso, 2006). Animal sourced foods (ASFs) are often costly and remain out of reach for many low-income households (Headey et al., 2018). Due to changes in environmental factors, dietary changes are urgently required (Springmann et al., 2018); and since insects form a part of traditional food system with high levels of Adepoju et al. 245 energy, quality protein with good amino and fatty acid profile, and variety of essential minerals (Igwe et al., 2011), they should be considered as important food source. Our previous study on consumption pattern of C. forda larva as important source of nutrients confirmed the findings of Omotoso (2006), and revealed that the most preferred mode of consumption of the insect larva is as an additive to vegetable soups (Daboh and Adepoju, 2020). This study was therefore carried out to assess the consumer acceptability and nutrient content of C. forda larva-enriched vegetable soups. Sample collection and preparation Dry C. forda (CF) larva sample was purchased from a local market in Burkina Faso and kept under refrigeration at -4°C. Fresh Amaranthus hybridus vegetable, 'Egusi' (melon seed), pepper, palm oil, onion, maggi cubes and salt used for the study were purchased from Bodija market in Ibadan, Nigeria. A portion of the dry C. forda larvae was ground and labelled as Sample A. Soup preparation The vegetable soups were prepared traditionally (by engaging the service of a local food vendor) at the Dietetic kitchen of the Department of Human Nutrition and Dietetics, University of Ibadan, Ibadan, Nigeria. The vegetable leaves were rinsed with distilled water to eliminate soil and pebbles and then sliced. A total of 350 g C. forda larvae was weighed, rinsed and soaked in hot water to soften a little and was divided into two portions of 175 g each. Sample B: Efo riro soup About 250 ml of palm oil was added to the cooking pot placed on a heater to warm. Ground pepper (10 g) and onions (4 g) were added and fried for 5 min, followed by addition of 4.5 g of salt and two bouillon cubes. Water (250 ml) was then added, stirred together and allowed to simmer for 20 min. Then, 25 g of sliced vegetable leaf was added and allowed to cook for 5 min (Adepoju and Ugochukwu, 2019). Sample C: Efo riro + C. forda larvae enriched soup About 250 ml of palm oil was added to the cooking pot placed on a heater to warm. Ground pepper (10 g) and onions (4 g) were added and fried for 5 min, followed by addition of 4.5 g of salt, two bouillon cubes, and 175 g of C. forda larvae. Water (250 ml) was then added, stirred together and allowed to simmer for 20 min. Then, 25 g of sliced vegetable leaf was added and allowed to cook for 5 min. Sample D: Egusi soup About 250 ml of palm oil was added to the cooking pot placed on a heater to warm. Ground pepper (10 g) and onions (4 g) were added and fried for 5 min, followed by addition of 4.5 g of salt and two bouillon cubes. Water (250 ml) was added, stirred together and allowed to simmer for 20 min. Then, 15 g of ground Egusi and 25 g of sliced vegetable leaf were added and allowed to cook for 5 min. Sample E: Egusi + C. forda larvae enriched soup About 250 ml of palm oil was added to the cooking pot and allowed to warm on a gas cooker. Ground pepper (10 g) and onions (4 g) were added and fried for 5 min, followed by addition of 4.5 g of salt, two bouillon cubes, and 175 g of C. forda larvae. Water (250 ml) was then added, stirred together and allowed to simmer for 20 min. Then, 15 g of ground Egusi and 25 g of sliced vegetable leaf were added and allowed to cook for 5 min. Chemical analyses The dry CF larva sample and the four soups were analysed for their nutrient content using the standard methods of Association of Official Analytical Chemists (AOAC, 2005). Proximate composition determination The moisture content of the samples was determined by hot-air oven method (Plus 11 Sanyo Gallenkamp PLC, UK) at 105°C for 4 h. The crude protein was determined using micro-Kjeldahl method (Method No. 978.04); crude lipid was determined by Soxhlet extraction method (Method No. 930.09). The ash content was determined through incineration in muffle furnace at 550°C for 4 h (Method No. 930.05). Total carbohydrate content was obtained by difference. Gross energy of the samples was determined using ballistic bomb calorimeter. Mineral content determination Potassium and sodium content of the samples were determined by digesting the ash of the samples with perchloric and nitric acids, and then taking the readings on Jenway digital flame photometer/spectronic20 (AOAC, 2005: [975.11]). Phosphorus was determined by vanado-molybdate colorimetric method (AOAC, 2005: [975.16]). Calcium, magnesium, iron, zinc, manganese, and copper were determined by atomic absorption spectrophotometric method (Buck Scientific, Norwalk, UK) and compared with absorption of standards of these minerals (AOAC, 2005: [975.23]. Vitamin A determination Vitamin A was determined through ultraviolet absorption measurement at 328 nm after extraction with chloroform (AOAC Method 960.5 & 974.29, 2005). Calibration curve of vitamin A acetate was made and sample vitamin A concentration estimated as microgram (μg) of vitamin A acetate. Thiamine (vitamin B1) determination Thiamine content of the samples was determined by weighing 1 g of sample into 100 ml volumetric flask with addition of 50 ml of 0.1 M H2SO4 and boiled in a boiling water bath with frequent shaking for 30 min. Five milliliter (5 ml) of 2.5 M sodium acetate solution was added and flask set in cold water to cool contents below 50°C. The flask was stoppered and kept at 45-50°C for 2 h and thereafter made up to 100 ml. The mixture was filtered through a No. 42 Whatman filter paper, discarding the first 10 ml. Ten milliliters (10 ml) was pipetted from remaining filtrate into a 50 ml volumetric flask, and 5 ml of acid potassium chloride solution was added with thorough shaking. Standard thiamine solutions were prepared and treated same way. The absorbance of the samples as well as that of standards was read on a fluorescent UV Spectrophotometer (Cecil A20 Model, USA) at a wavelength of 285 nm. Riboflavin (vitamin B2) determination One gram (1 g) of each sample was weighed into a 250 ml volumetric flask; 5 ml of 1 M HCl was added, followed by the addition of 5 ml of dichloroethene. The mixture was shaken and 90 ml of de-ionized water was added. The whole mixture was thoroughly shaken and was heated on a steam bath for 30 min to extract all the riboflavin. The mixture was then cooled and made up to volume with de-ionized water. It was then filtered, discarding the first 20 ml of the aliquot. Two milliliters (2 ml) of the filtrate obtained was pipetted into another 250 ml volumetric flask and made up to mark with de-ionized water. Samples were read on the fluorescent spectrophotometer at 460 nm. Standard solutions of riboflavin were prepared and readings taken at 460 nm. The sample riboflavin was obtained through calculation. Niacin (vitamin B3) determination Five grams (5 g) of sample was extracted with 100 ml of distilled water and 5 ml of this solution was drawn into 100 ml volumetric flask and made up to the mark with distilled water. Standard solutions of niacin were prepared and absorbance of sample and standard solutions was measured at of 385 nm on a spectrophotometer, and niacin concentration of the sample estimated. Pyridoxine (vitamin B6) determination The vitamin B6 content of the samples was determined by extracting 1 g of sample with 0.5 g of ammonium chloride, 45 ml of chloroform and 5 ml of absolute ethanol. The mixture was thoroughly mixed in a separating funnel by shaking for 30 min, and 5 ml of distilled water added. The chloroform layer containing the pyridoxine was filtered into a 100 ml volumetric flask and made up to the mark with chloroform. Standard solutions of 0-10 ppm of vitamin B6 were prepared and treated in a similar way as samples; and their absorbance measured on Cecil 505E spectrophotometer at 415 nm. The amount of vitamin B6 in the sample was then calculated. Cyanocobalamin (vitamin B12) determination Cyanocobalamin content of the samples was determined by extracting 1 g of sample with distilled water with shaking for 45 min, followed by filtering the mixture. The first 20 ml of the filtrate was rejected, and another 20 ml filtrate collected. To the collected filtrate, 5 ml of 1% sodium dithionite solution was added. Standard cyanocobalamin solutions (0-10 μg/ml) were prepared, and absorbance of sample as well as standards was read on spectronic21D spectrophotometer at 445 nm. Amount of sample cyanocobalamin was then estimated through calculation. Ascorbic acid determination Ascorbic acid in the samples was determined by titrating the aqueous extract of each sample with solution of 2, 6dichlorophenol-indophenol dye to a faint pink end point. Anti-nutrients determination Oxalate content of the samples was determined by extraction of the samples with water for about 3 h and standard solutions of oxalic acid prepared and read on spectrophotometer (Spectronic20) at 420 nm, and amount of oxalate estimated. Phytate was determined by titration with ferric chloride solution (Sudarmadji and Markakis, 1977), while trypsin inhibitory activity was determined on casein and comparing the absorbance with that of trypsin standard solutions read at 280 nm (Makkar and Becker, 1996). The tannin content of the samples was determined by extracting the samples with a mixture of acetone and acetic acid for 5 h, measuring their absorbance and comparing the absorbance of the sample extracts with the absorbance of standard solutions of tannic acid at 500 nm on spectronic20 (Griffiths and Jones, 1977). Saponin was also determined by comparing the absorbance of the sample extracts with that of the standard at 380 nm (Makkar and Becker, 1996). All determinations were carried out in triplicate. Sensory evaluation of vegetable soups The vegetable soups acceptability study was carried out at the Department of Human Nutrition and Dietetics sensory evaluation laboratory, University of Ibadan, Nigeria. The soup samples were assessed for their acceptability using 30 untrained panelists drawn within the University community, who had eaten or known C. forda larva before and were willing to participate. The samples were rated on a 9-point hedonic scale in which the degree to which a product is relished was expressed as: like extremely (9), like very much (8), like moderately (7), like slightly (6), neither like nor dislike (5), dislike slightly (4), dislike moderately (3), dislike very much (2), dislike extremely (1). The panelists were required to observe the sample, taste and score based on colour, taste, aroma, texture and overall acceptability. The panelists were provided with water to rinse their mouths in between sample evaluation, and were instructed to rinse their mouth with water before tasting another sample. Statistical analysis The data obtained from the chemical analyses were expressed as mean and standard deviation of triplicate determinations and subjected to independent t-test using Statistical Package for Social Sciences (SPSS) version 20 (SPSS Inc., USA); and the data obtained from sensory evaluation presented as means of 30 panelists' assessment which was analysed using oneway ANOVA. The level of significance was set at p < 0.05. Proximate composition of C. forda and vegetable soups The proximate composition of C. forda larva and the four vegetable samples is shown in Table 1. The dry C. forda sample (sample A) was high in crude protein (54%) and gross energy content (492.05 Kcal), high in fat, ash, and dietary fibre, moderate in carbohydrate, but low in moisture content (3.98%). Mineral content of C. forda and vegetable soups The C. forda larva contained substantial amount of sodium, potassium, calcium, magnesium and phosphorus, with potassium being the most abundant, followed by ) and copper (0.33 mg 100g -1 ), and low in manganese (0.001 mg100g -1 ) (sample A) ( Table 2). Addition of Egusi to vegetable soup (sample B) significantly increased its mineral content (sample D), (p<0.05). Addition of CF larva to the vegetable and Egusi soups (samples B and D) increased the sodium, potassium, iron zinc and copper significantly (p<0.05), while it decreased the calcium, magnesium and phosphorus content significantly (p<0.05). The vegetable soup (sample B) was rich in magnesium, phosphorus, and calcium, and high in potassium and sodium content. Addition of Egusi to the vegetable (sample D) increased the nutrient content of the product significantly. Also, addition of the insect larva to both vegetable (sample B) and Egusi (sample D) soups significantly increased the sodium, potassium, iron, zinc and copper of the larva-enriched vegetable soups (Samples C and E) (p<0.05). In Table 3, the vitamin composition of C. forda larva revealed that it contains substantial amounts of vitamins, especially vitamins B 12 and C. The larva was low in water soluble vitamins (sample A). The vegetable soup (sample B) was low in vitamins A (0.87), B 1 (0.04), B 2 (0.09), and B 3 (0.71 mg 100 g -1 ), but high in vitamins B 6 (2.11), B 12 (17.01), C (15.73) and E (3.01 mg 100 g -1 ). Addition of ground melon to the vegetable soup significantly improved all the vitamin content of the product (sample D) except vitamin B 12 (p<0.05). Enriching both vegetable and Egusi soups (samples B and D) with CF significantly increased the vitamin content of the products (p<0.05) except vitamin B 12 (samples C and E). Anti-nutrients level of C. forda larva and vegetable soups The phytate and trypsin inhibitor content of C. forda larva were negligible, while oxalate, tannin, and saponin were not detectable (Table 4). However, there was a significant difference (p<0.05) in the trypsin inhibitor content of the samples (p<0.05), with the larvaenriched vegetable Amino acid and fatty acid profiles of C. forda larva The CF larva has good amount of essential amino acids such as lysine, threonine, phenylalanine, leucine, isoleucine, valine, tryptophan and cysteine as well as dispensable amino acids such as glutamic acid, histidine, arginine, aspartic acid and glycine among others, hence, its protein content is of high quality ( Table 5). The C. forda larva is rich in linolenic acid (polyunsaturated), and high in oleic acid (monounsaturated), stearic and palmitic acids among others (Table 6). Linolenic, stearic, palmitic, and oleic acids were very much abundant in fat of the larva. Unsaturated fatty acids constitute 52.27% of the total fatty acids in the larva and 12.15% of this was monounsaturated fatty acid (MUFA). Sensory evaluation of vegetable soups In Table 7a, all the soups were acceptable to the panelists, as none of them was rejected or scored below average. Overall, the control vegetable soup (sample B) Proximate composition of C. forda and vegetable soups In Table 1, the dry C. forda (CF) larva was low in moisture content (3.98%). The low moisture content obtained for the CF larva is in line with the reports of Banjo et al. (2006) and Osasona and Olaofe (2010). The low moisture content is suggestive of its high keeping quality (that is long shelf life), and supportive of the assertion of respondents in our consumption survey carried out earlier that the insect larva can be kept for a long period of time (Daboh and Adepoju, 2020). The low moisture content will prevent microbial contamination. The insect larva contained high amount of protein. The value obtained for the larva protein falls within the range of values reported by Igbabul et al. (2014), similar to 55.41% obtained by Yapo et al. (2017) but lower than 63% reported by Anvo et al. (2016) for Cirina butyrospermi. The variation observed in the biochemical composition of the insect larva may be due to the host tree because the amount of protein varies between insect species and within the same species depending on the nutritional quality of the leaves of the host tree (Banjo et al., 2006;Yapo et al., 2017), the difference in the geographical location where the CF larva was obtained, as well as method of harvesting/processing of the insect larva (Ekpo, 2011;Womeni et al., 2012;Adepoju, 2013). The crude fat content of the insect larva obtained in this study is within the range of values quoted for insects (10 -30%) . The value is higher than the values reported by Osasona and Olaofe (2010) and Igbabul et al. (2014) and 15% reported by Anvo et al. (2016), but lower than the 22.21% reported by Ogunleye (2006). The difference in the crude fat content of C. forda in this study compared with other studies could be due to the difference in the geographical location where the samples were obtained and also, the method of processing (Ekpo, 2011;Womeni et al., 2012;Adepoju, 2013). The moderate fat content of the C. forda larva is beneficial because it may reduce its rate of susceptibility to rancidity. However, the C. forda larva contained a moderate amount of dietary fibre. Dietary fibre helps to regulate the digestive system, aid bowel health and weight management (Rolfe et al., 2009). The value of ash of C. forda larva sample was less than the range of values (2.91 -3.97%) reported by Igbabul et al. (2014) but comparable to that of Akinnawo and Ketiku (2000), Omotoso (2006) and Osasona and Olaofe (2010). The carbohydrate content of the C. forda larva was higher than the range reported by Igbabul et al. (2014). The gross energy content of C. forda sample was very high. This may be due to the fact that at larval stage, the insect feeds much to derive and store the nutrient and energy needed for growth and development during the pupal stage to the adult stage (Chen et al., 2009). This finding is in line with the report of FAO (2004) that reported high energy content in feeding caterpillar flours. The high energy content could also be as a result of its high protein, fat and carbohydrate content. The gross energy content in this study is very similar to those of 492.31 kcal. reported by Yapo et al. (2017) for Cirina butyrospermi. The moisture content of vegetable soup was very high (sample B). The vegetable soup was moderately high in protein content compared with other plant protein (Adepoju and Ugochukwu, 2019). This amount of protein is believed to have been contributed partly by other ingredients used in the preparation of the vegetable soup. The soup was low in fat and carbohydrate content despite the addition of palm oil. Leafy vegetables are generally poor source of lipids and carbohydrates (Adepoju and Ugochukwu, 2019). However, the soup had moderate value of ash and gross energy, and high in dietary fibre. Vegetables are generally good source of dietary fibres. The increase in the dietary fibre of the soups is very beneficial, as dietary fibre promotes a healthy bowel function, helps to control blood sugar level, lowers blood cholesterol levels, and helps in weight management (Rolfe et al., 2009). The significant increase in the ash content of the C. forda larva-enriched soups is an indication of their higher mineral content compared with plain and Egusi vegetable soups. Mineral composition of C. forda larva and vegetable soups The sodium content of the larva is within the range reported for other edible insects (Ajai et al., 2013). The recommended daily dietary allowance of sodium is 2400 mg per day (FAO, 2010). Sodium assists in maintaining the proper acid-balance and in controlling osmotic pressure that develops between the blood and cells due to ionic concentration differences (Paiko et al., 2014). Potassium was the most abundant of all the minerals in C. forda larva. This corroborated the findings of Osasona and Olaofe (2010) who reported potassium to be the most abundant mineral in the flour of CF. The high potassium content is highly beneficial because a high intake of potassium has been reported to protect against increasing blood pressure and other cardiovascular risk (Insel et al., 2007), and it is essential for the functioning of the brain and nerve (Iombor et al., 2017). The potassium-sodium ratio (K:Na) has frequently been used as a diagnostic tool to identify adrenal insufficiency (Iombor et al., 2017). The potassium to sodium ratio in this study is 2.1:1, hence, the larva could be incorporated into diets for the management of hypertension as high potassium intake has been found to lower blood pressure by antagonizing the effect of sodium (Iombor et al., 2017). The calcium content of CF larva was found to be high. The calcium value for the larva in this study was higher than those reported by Akinnawo and Ketiku (2000), Omotoso (2006) and Osasona and Olaofe (2010). The observed difference could be due to the differences in the source of the larva coupled with methods of processing. Calcium is an integral component of the skeleton, about 99% of total body calcium is found in bones and teeth, where it plays a structural role (Insel et al., 2007;Yapo et al., 2017). Calcium is important in the development and maintenance of bones and teeth, blood clotting, nerve impulse transmission, muscle contraction and cell metabolism (Heaney, 2006;Insel et al., 2007). The magnesium content of C. forda larva was higher than those reported by Akinnawo and Ketiku (2000), Omotoso (2006) and Osasona and Olaofe (2010). Magnesium is a co-factor participating in more than 300 enzyme reactions, making it an essential element for the synthesis of carbohydrates, lipids, nucleic acids and proteins, as well as for other actions in different organs of the cardiovascular and neuromuscular systems (Chen and Feng, 2002;EFSA, 2015). The phosphorus content of C. forda larva in this study is higher than the value reported by Paiko et al. (2014) but lower than the 215.54 mg/100 g reported by Omotoso (2006). Phosphorus like calcium is also involved in calcification of bones and teeth. It plays a vital part in the oxidation of nutrients in the form of phosphate groups in ATP (Paiko et al., 2014). C. forda larva contained substantial amount of heme iron. Iron plays an important role as a heme molecule in red blood cells, as it permits oxygen transport (WHO, 2006). It can serve as an antioxidant and can prevent cardiomyopathy and growth retardation (Paiko et al., 2014). Since C. forda larva contained substantial amount of iron, its inclusion in human diets will be beneficial, and as blood building element in anaemic conditions (Rolfe et al., 2009). Micro-minerals such as copper, zinc, and manganese are also contained in substantial amount in C. forda larva. The copper content of C. forda larva is similar to the value reported by Ande (2003). Zinc is important because its deficiency can lead to growth retardation, delayed sexual and bone maturation, skin lesions, diarrhoea, alopecia, impaired appetite, increased susceptibility to infection mediated via defects in the immune system (FAO/WHO, 2001). Vitamin content of Cirina forda larva and vegetable soups Generally, insects have been reported to contain varying degree of both water-soluble and fat-soluble vitamins. C. forda larva was found to contain substantial amounts of B-vitamins, vitamins C and E. Among these, vitamin C was the most abundant. Vitamin C helps in maintaining blood vessels flexibility and improves circulation in the arteries (Alamu et al., 2013). One of the most important benefits derivable from vitamins A, C and E is their role as antioxidants, oxygen free radical scavengers, while that of the B-vitamins is their role as co-enzymes in several enzyme systems of the body (Insel et al., 2007;Alamu et al., 2013). Addition of Egusi and the CF larva led to significant increase (p<0.05) in the vitamin content of the products (samples C, D and E). Anti-nutrients level of C. forda larva and vegetable soups The absence of oxalates, tannins and saponins and negligible phytate content in C. forda is believed to be due to full conversion of the leaf consumed by the larva during the process of its entering the soil before being harvested, as its name implies in Yoruba language (Kanni wolemeaning the larva has entered the soil) (Osasona and Olaofe, 2010). The phytate, oxalate, tannin, and saponin content of C. forda larva in this study was much lower than those reported by Alamu et al. (2013) and Omotoso and Adesola (2018). The difference in the antinutrients content of the larva in this study and past studies might have resulted from the differences in the source of the insect larva. Antinutrients are generally known to reduce the bioavailability of nutrients in the body. High phytate level in human nutrition decreased the availability of some minerals such as calcium, magnesium and iron by formation of insoluble compounds with the minerals (Anuonye et al., 2012), while tannins reduced protein bioavailability when bound to protein by inducing a decrease in solubility and functionality of the protein. Tannins are capable of lowering available protein by antagonistic competition, and can therefore elicit protein deficiency syndrome (Ekop, 2004). Amino acid profile of C. forda larva C. forda larva contained a wide array of both essential and non-essential amino acids (Table 5). The quality of a food protein depends largely on its amino acid component. C. forda larva was rich in essential amino acids. The most predominant amino acids of the larva are lysine, glutamic acid, threonine, aspartic acid, leucine, glycine, isoleucine, tyrosine and phenylalanine. This is similar to the work of Yapo et al. (2017) who also reported most of these amino acids to be the most abundant in C. butyrospermii. Amino acids are important components for healing and protein synthesis processes; any deficiency in these essential amino acids will hinder the recovery process (Zuraini et al., 2006). Glycine together with other essential amino acids such as alanine, arginine and phenylalanine form a polypeptide that promotes growth and tissue healing (Witte et al., 2002;Adeoti et al., 2013). Also, glycine is involved in the transmission of impulses in the nervous system, and alanine strengthens the immune system and prevents buildup of toxic substances in the body (Paiko et al., 2014). The high content of essential amino acid in CF larva is very beneficial, as this implied that its protein is of a high biological value. Addition of CF larva to enrich the vegetables (samples C and E) significantly increased their protein content (p<0.05). Fatty acid profile of C. forda larva The larva of CF contained a high amount of polyunsaturated fatty acid (PUFA), (33.05% linolenic acid and 7.07% linoleic acid). The ratio of polyunsaturated to saturated fatty acids (P/S) has been used widely to indicate the cholesterol lowering potential of a food (Akinnawo and Ketiku, 2000). Akinnawo and Ketiku (2000) reported that a P/S ratio of 0.2 has been associated with high cholesterol level with high risk of coronary heart disorders. The P/S ratio of Cirina forda larva in this study is 0.8, and a ratio as high as 0.8 has been reported to be associated with desirable levels of cholesterol and reduction of coronary heart disease [Akinnawo and Ketiku, (2000)]. Thus, regular consumption of Cirina forda larva could help in intake of healthy fat that can prevent onset of cardiovascular diseases. Sensory evaluation of vegetable soups Although, no significant difference was observed in the scoring of sensory attributes of samples C, D and E (p>0.05), the enriched samples (Samples C and E) had the highest sensory scores. The exoskeleton of insects has a great influence on the texture. Insects are crunchy and the sounds accompanying their eating resemble the sounds of crackers or pretzels (Kouřimská and Adámková, 2016). This may explain why CF-enriched vegetable soups were more preferred and scored higher than either plain or Egusi vegetable soups. The result of sensory evaluation obtained in which the vegetable soups rated higher in all the parameters is similar to the report of Adejoju and Ugochukwu (2019) in which the vegetable sauce was rated higher than ordinary vegetable soup. Conclusion C. forda larva was very rich in protein, high in fat, and Adepoju et al. 253 essential minerals, and very low in anti-nutrients. The amino acid composition of the larva showed that it contained all the essential amino acids needed for human growth in good proportion, hence, the protein content can be considered as a complete protein of high biological value, more so, it is of animal origin. Fat content of the insect larva contained higher amount of unsaturated fatty acid compared with the saturated fatty acid component, showing that it can serve as a good source of healthy fat that is fit for human consumption for promotion of good health. The insect larva and the enriched vegetable soups contained negligible level of antinutrients which cannot pose any threat to nutrient bioavailability from the larva and the vegetable soups, hence, it is believed that the consumption of the insect larva, either as snack or in vegetable soups or sauces is very safe and will promote quality nutrient intake by consumers. The soups enriched with C. forda larva were more acceptable than the plain vegetable or vegetable with Egusi soups. C. forda larva-enriched Egusi soup was the most acceptable soup. Likewise, in terms of nutrient content it was the most nutrient-dense of the soups. Inclusion of C. forda larva in vegetable soups improved both its palatability and nutrient content; therefore, its inclusion in soups and sauces should be encouraged. Also, popularizing the consumption of this insect larva should be encouraged as a means of increasing dietary diversity of the population of the people where the insect larva is readily available. This will help to promote the intake of quality protein and essential minerals among the populace of the host community, and assist in combating protein and micronutrient malnutrition; thereby improving the general health of the people. Also, its consumption could serve as a means of conserving the host tree from going into extinction due to human activities.	v3-fos-license
2022-06-28T05:58:07.839Z	2011-01-01T00:00:00.000	250670670	{ "extfieldsofstudy": [ "Physics" ], "oa_license": "CCBY", "oa_status": "GREEN", "oa_url": "http://cds.cern.ch/record/1301775/files/ATL-INDET-PROC-2010-032.pdf", "pdf_hash": "bbbf99fa7599e8fc1a60b715d68808dcd847813b", "pdf_src": "IOP", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45302", "s2fieldsofstudy": [ "Physics" ], "sha1": "bbbf99fa7599e8fc1a60b715d68808dcd847813b", "year": 2011 }	pes2o/s2orc	ATLAS IBL: integration of new HW/SW readout features for the additional layer of Pixel Detector An additional inner layer for the existing ATLAS Pixel Detector, called Insertable B-Layer (IBL), is under design. The front-end electronics features a new readout ASIC, named FE-I4, which requires new off-detector electronics, currently realized with two VME-based boards: the Back Of Crate module (BOC) implements optical I/O functionality and the ReadOut Driver module (ROD) implements data processing functionality, plus a Timing Interface Module (TIM). This paper presents a proposal for the IBL readout system, mainly focusing on the ROD board. Introduction An additional inner layer for the existing ATLAS Pixel Detector [1], named Insertable B-Layer (IBL) [2], is under design. IBL will allow robust tracking despite effects arising from luminosity, hardware lifetime and radiation. The IBL will also provide improved precision for vertexing and b-tagging to the current detector. IBL front-end electronics features a new readout ASIC (FE-I4 [3]), which requires new offdetector electronics, currently realized with two VME [4] boards: the Back Of Crate module (BOC) implements optical I/O functionality and the ReadOut Driver module (ROD) implements data processing functionality. This document proposes a new architecture for the ROD card, which provides backward compatibility for operation with current Pixel-BOCs and support for a modified architecture of the off-detector readout, with improved performance and a clear separation of data-flow and calibration tasks. Current ROD/BOC system In the current readout system a BOC -ROD pair connects to up to 32 detector channels at 40 Mb/s via optical links (or 8 channels at 160 Mb/s). As shown in figure 1, the BOC provides an optical I/O interface to the Pixel Detector: in particular the BOC is able to transmit configuration and trigger commands to the module front-end electronics and also to receive data from the detector channels and send them to the ROD, which performs formatting and event fragment building. The event fragments are then encapsulated into S-Link packets and routed towards the ATLAS DAQ system through the S-Link [5] on the BOC. 132 ROD and BOC pairs and 132 S-Link boards are used in the ATLAS Pixel Detector. The current ROD board hosts 11 Xilinx FPGAs for data formatting, event fragment building and routing and 5 Texas Instruments Digital Signal Processors (DSPs): one of them, the Master Readout driver card proposal The idea for IBL readout chain is to keep BOC-ROD task subdivision as in the current Pixel readout chain, assigning the data path (optical link interface to the front-end chips and S-Links) to the BOC and part of the data processing to the ROD and part to an external computer farm. The two boards will benefit from up-to-date programmable devices and technologies in order to increase system performances and to have a more compact system. A description of the proposed BOC is available at [6], while this paper will focus on ROD details. The aim of the new 9U VME ROD card is to process data coming from 16 IBL modules for a total number of 32 FE-I4 data links at a rate of 160 Mb/s each. This number of channels accounts for a total I/O bandwidth of 5.12 Gb/s that requires the use of 4 S-Links since no data reduction is foreseen. This leads to a data I/O bandwidth as large as 4 times the current ROD board. The proposed architecture for the new ROD card (see figure 2) is based on the use of modern FPGA devices, while no DSP chip is foreseen. The main idea is to perform on-board event fragment building and histogramming, while sending histograms (or even raw data) to a computer farm for fitting operations. In more details, the new ROD executes the data taking in order to obtain the perpixel calibration of noise and charge response by accumulating the per-pixel occupancies, sums of Time Over Threshold 1 (ToT) and sums of ToT squared parameters. Histograms are created, saved on RAMs and eventually transferred via Gbit Ethernet to an off-line high-performance computer farm. Gbit Ethernet links will be used in order to avoid bottlenecks on the data transfer experienced with the VME bus. The execution of data fitting using computers allows an improved flexibility for fit code development, since more convenient tools are available compared to the DSP environment. Another idea is to transfer the current ROD Master DSP functions to a PowerPC 2 core inside a programmable device, in order to have an architecture without DSPs: this has the great advantage of leading to a single design and simulation environment within which both logic and PowerPC software can be verified together. This should allow a reduction in the design and debug time, if compared to having two different design systems for FPGAs and DSPs. In more details the foreseen architecture is based on one Virtex5 FPGA with PowerPC capabilities implementing the ROD controller function, plus 2 Xilinx Spartan6 programmable devices implementing event builder and histogrammer blocks. These commercial devices also allow the reuse of most of the VHDL code that has been designed to implement the firmware on the current ROD card for the ATLAS pixel detector. The next paragraphs provide more details about the main firmware blocks under design. ROD controller The ROD controller (see figure 3) is a Virtex-5 programmable device with an embedded PowerPC core performing all the basic ROD control operations. The PowerPC runs a software which performs system control and non real-time functions, while the remaining FPGA logic performs all the real-time functions. The PowerPC provides FE-I4 configuration to a command interface block using a serial port with data received from the VME host or from an external PC through a Gbit Ethernet connection, that can be used as a higher speed alternative to the VME bus. It also drives the calibration runs by providing software triggers through the same serial port and controlling that the data taking is working properly. The PowerPC core also drives the setup buses used to read or control all the configuration registers on the ROD and BOC boards; a bus arbiter block controlled from the VME host decides whether the PowerPC is controlling the setup buses or if they are controlled by the VME host itself. The VME host is able to access the processor memory when needed in R/W mode. Some of the real-time functions are the decoding of trigger information from the TIM board and consequent distribution to the BOC board and the routing of the busy signal to the TIM in order to stop the generation of triggers when the ROD is not able to receive any of them. Event builder block The event builder block shown in figure 4 has the main purpose of reading and buffering data from 8 FE-I4 chips through the BOC by means of a FIFO-like interface: as soon as the event builder block recognizes that the BOC buffer has data inside, it starts reading data on two 8 bit buses at 80 Mb/s and pushing them into two FIFOs. Data are then popped out and a frame is produced in which each event contains a header (along with L1ID + BCID information coming from the ROD controller), data from the 8 FE-I4s in an ordered manner, plus a trailer containing a data error summary. Such frames are then sent along 3 different pathways: they are stored in a FIFO before being transferred towards one S-Link, they are sent to the Histogramming block and also towards the Gbit Ethernet block. Histogrammer In order to calibrate the detector, a number of histograms have to be collected during detector operation: this task is accomplished on the histogrammer block, which perform calculations during the calibration tasks, such as threshold scan and ToT scan. The goal is to accumulate per-pixel occupancy values, sums of ToT and sums of ToT 2 parameters in order to create histograms that are saved on-the-fly on a very fast memory available inside the FPGA chips or on external synchronous static RAMs and eventually transferred via Gbit Ethernet to an off-line high-performance computer, where the fit processing is executed. The logic required inside the FPGA to perform histogramming is simple and requires little logic resources apart from memory. Figure 5 shows the block diagrams for threshold scan requiring a 1-byte wide adder per pixel and for ToT scan with the additional adders from the sum of ToT (12 bits) and the sum of ToT 2 (16 bits). The internal memories are dual-ported and capable of reading and writing one word per clock period at the same time. The two Gbit Ethernet links allow to reach an overall maximum data bandwidth of 220 MB/s for sending the histograms towards the external PC farm. This value is much higher than the 4 MB/s bandwidth supported by the VME bus in the current system. At the same time the usage of some GHz processors in the PC farm allows to speed-up the fit operations, if compared to the 220 MHz DSPs used in the current ROD board. JINST 6 C01018 The communication protocol with the external PC through the Gbit Ethernet link is a custom made protocol in which the ROD board needs to receive a continuous feedback from the PC. In particular the ROD board transmits block n after receiving from the PC the acknowledge that all previous blocks at least up to block n − 2 have been correctly received. In case of packets loss a timeout mechanism is foreseen. Conclusions A new architecture for the IBL off-detector readout is proposed with major modifications to the current BOC-ROD boards in order to manage a higher data bandwidth and achieve better performances. In particular a ROD card without DSPs is presented with the innovative approach of using a PowerPC core for high-level operations, a histogrammer based on FPGA logic blocks and the use of a PC farm for fit operations. The presented design has also the important advantage of allowing the designers to use a single design and simulation environment both for FPGA logic and the embedded processor software, leading to a simpler and more manageable system.	v3-fos-license
2023-02-18T14:54:23.840Z	2018-04-19T00:00:00.000	256947796	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/s41598-018-24709-0.pdf", "pdf_hash": "fca001f702ab12b3b52f16bddef027890e14f166", "pdf_src": "SpringerNature", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45303", "s2fieldsofstudy": [ "Medicine", "Biology" ], "sha1": "fca001f702ab12b3b52f16bddef027890e14f166", "year": 2018 }	pes2o/s2orc	Amino Acid Signature in Human Melanoma Cell Lines from Different Disease Stages Cancer cells rewire metabolism to sustain high proliferation rates. Beside glycolysis and glutaminolysis, amino acids substitute as energy source, feed fatty acid biosynthesis and represent part of the secretome of transformed cells, including melanoma. We have therefore investigated acetate, pyruvate and the amino acid composition of the secretome of human melanoma cells representing the early slow (WM35, WM278, WM793b and VM21) and metastatic fast (A375, 518a2, 6F and WM8) growth phase in order to identify possible signalling components within these profiles. Proliferation assays and a principle component analysis revealed a stringent difference between the fast and slow growing melanoma cells. Moreover, upon inhibition of the mevalonate pathway, glutamic acid and alanine were identified as the central difference in the conditional media. A supplementation of the media with glutamic acid and the combination with alanine significantly accelerated the proliferation, migration and invasion of early stage melanoma cells, but not metastatic cells. Finally, the inhibition of the mevalonate pathway abolished the growth advantage of the melanoma cells in a time dependent manner. Taken together, these data corroborate a stage specific response in growth and aggressiveness to extracellular glutamic acid and alanine, indicative for microenvironmental signalling of individual amino acids. Tumour signalling and progression is strongly dependent on the tumour microenvironment which comprises components like the extracellular matrix, surrounding stromal cells and signalling molecules including secreted proteins 1 . In melanoma immune checkpoint inhibitors were evaluated for the first time to highlight the microenvironment as a therapeutic battlefield for the immune system to attack transformed cells 2 . Moreover, metabolic reprogramming in response to oncogenic stimuli has been elucidated as an adaption mechanism to scope with hypoxia, acidosis and cellular stress in the tumour microenvironment 3,4 . Decoupling of the mitochondrial tricarboxylic acid (TCA) cycle from cytosolic glycolysis allows cancer cells to establish a flexible adaptation to the conditions of the microenvironment by glycolysis and glutaminolysis 5,6 . On the crossroads of glycolysis and glutaminolysis, acetyl-CoA has been established to play a crucial role in cancer cell progression by feeding fatty acid synthesis and the mevalonate pathway 7 . The activation of the mevalonate pathway is therefore essential for a rapid proliferation of transformed cells and inhibition associated with cell cycle arrest and the induction of apoptosis [8][9][10][11][12] . Conversely, an activation of the mevalonate pathway is triggered by mutant p53 or Myc and thereby favours the conjecture that pharmacological inhibition by statins may serve as a therapeutic concept 7,12,13 . This assumption is further supported by the finding that the dysregulation of the mevalonate pathway promotes transformation 14 . Using statins is a proper tool to trigger the mitochondrial pathway of apoptosis in various cancer cells 9,10,15 . Interestingly, human metastatic melanoma cells are highly susceptible to statin induced apoptosis, while cells from the radial growth phase and primary human melanocytes are virtually insensitive 8,16 . It is therefore anticipated that fast proliferation rates are in favour of mevalonate pathway inhibition and thereby may use a switch from glucose utilisation to glutamine 7 . Recently, amino acids other than glutamine were responsible for the majority of proliferative cell mass 17 . Amino acids substitute as energy source, feed lipid biosynthesis and represent part of the secretome of transformed cells, including melanoma. However, little is known whether extracellular 1 amino acid profiles correlate with specific growth behaviour of defined melanoma cell lines. Melanoma are heterogeneous tumours with different subpopulations characterized by distinct doubling times 18 . We have therefore investigated the amino acid composition as well as acetate and pyruvate of the secretome of human melanoma cells representing early slow growth phase and rapid growth phase of metastatic cells. Making use of subsequent multivariate data analysis, namely principle component analysis (PCA) and partial least squares (PLS) regression enabled to elucidate significant changes in the amino acid composition of media in a time and stage dependent manner. Further analyses of proliferation, migration and invasion confirmed a crucial role for glutamic acid to support enhanced cell growth and aggressiveness in early stage melanoma cells. Inhibition of the mevalonate pathway abrogated the growth advantage and thereby underlined the importance of the mevalonate pathway in melanoma progression. Finally, the underlying mechanisms and potential therapeutic implications of our findings were discussed. Results Deviation in amino acid profiles characterize melanoma cells of different stages. Human metastatic melanoma cells (Fig. 1B) grow significantly faster than WM35, WM278, WM793b and VM21 cells from the early radial and vertical growth phase of primary tumours, i.e. within 48 hours proliferation was not significantly enhanced in slow growing cells (Fig. 1A). This biological criterion was used throughout this manuscript to distinguish between the two growth types of melanoma cells. Expression patterns of transcription factors like microphthalmia-associated transcription factor (MITF) and inversely correlated receptor tyrosine kinases like AXL have been implicated in staging of melanoma with respect to progression and resistance 19 . However, the expression levels of MITF in various melanoma cell lines are highly variable and correlation to other receptor tyrosine kinases may be also implicated in acquired drug resistance 20 . WM793b cells were selected from primary melanoma which lack metastatic potential in a SCID murine xenograft model 21 . Accordingly, based on the proliferation velocity depicted in Fig. 1, WM35, WM278, WM793b and VM21 cells were classified as slow growing melanoma cells while the others (A375, 518a2, WM8 and 6F cells) were termed fast growing and from metastatic origin. Importantly, inhibition of the mevalonate pathway by Simvastatin significantly reduced proliferation in all cell lines (Fig. 1). Next, we asked the question, whether it is possible to discriminate melanoma cell lines with different proliferation behaviour on the basis of extracellular amino acid profiles? The method of choice is principle component analysis (PCA), which enables to identify the principle components (e.g. individual amino acids or patterns of amino acids) out of a complex data set to explain biological differences between cells or treatments. Therefore, the conditional media of WM35, WM793b and A375 melanoma cells were used for HPLC analysis to determine acetate, pyruvate and amino acid profiles (Supplementary Table S1). The cells were treated in the absence and presence of Simvastatin for 24 and 48 hours. After normalization of the data to the cell count, principle component analysis (PCA) was performed in order to assess differences in the amino acid profiles according to cell line, incubation time and treatment via dimensionality reduction. As the difference between cell lines in acetate and pyruvate secretion is quite obvious due to different proliferation velocity, only amino acids were considered for pattern recognition in order to prevent overlaying effects by the organic acids. In the case of identical profiles for all cells and treatments the data sets would cluster in the centre of the plot around the coordinate origin. Clearly, this is not the case and a marked clustering was observed for A375 cells versus slowly growing cell lines WM35 and WM793b (Fig. 2). Stage dependent clustering of the cell lines was still retained after inhibition of the mevalonate pathway with Simvastatin. Moreover, clustering was also dependent on incubation time, indicating that metabolic rates determined for amino acids may provide further insights. The difference in amino acid concentrations of conditional media between the 24 and 48 hour values was used to determine metabolic rates of individual amino acids in order to improve comparability of the experimental results. PCA was performed after calculation of the specific rates, i.e. the change in amino acid (and organic acid) concentration per cell and hour, indicating uptake or secretion. The data dimension could be reduced to four principle components (PCs) explaining in total 98.4% of the variance in the data. PC1 mainly represents the differences between Simvastatin treated and untreated cells of the slow growing cells, whereas PC2 shows the differences of A375 cells from the WM35 and WM793b cells, respectively (Fig. 3A,B; Supplementary Fig. S1). The metabolic correlation for this clustering can be seen in the loading plots (Fig. 3C,D). Mainly, metastatic A375 cells seem to take up more acetate, alanine and glutamic acid than slow growing cells, because the PC2 values for A375 cells are highest (Fig. 3A). Another interesting finding of the metabolite analysis in conditional media is obtained from WM793b cells and can also be extracted from the PCA model shown in Fig. 3. Each PC of the model describes a different variance in the dataset. Analysing the score plot of PC2 versus PC4 shows again the differences of A375 cells along PC2. Additionally, a separation of the WM35 and WM793b cells along PC4 can be observed (Fig. 4A). According to the loading plot ( Fig. 4B) this separation correlates with proline, i.e. proline seems to be taken up in higher amounts by WM793b cells and to a lesser extent by WM35 cells. This finding may correlate to the observation by others that proline de novo synthesis is an important downstream feature of glutaminolysis in melanoma cells compared to melanocytes and contributes to redox homeostasis 5,22,23 . Impact of mevalonate pathway inhibition. The correlation between statin treatment and cell growth has already been described in the Figs 1, 2 and 3. The inhibition of the HMG-CoA reductase leads to a significant inhibition of metastatic cell growth which can already be seen after 24 h in A375, 518a2 and 6F cells. The uptake of glutamic acid and alanine not only determines the metabolic differences between the growth velocities, but also the metabolic adaptation after Simvastatin treatment in A375 cells (Fig. 5A,B). Untreated A375 cells take up glutamic acid and alanine. In contrast, the uptake of these two amino acids is not only inhibited in the presence of Simvastatin, it even leads to secretion of alanine and glutamic acid, indicated by a positive value for PC2 (Fig. 5B). The metabolic benefit of glutamic acid uptake by metastatic cells may fuel the energy producing TCA cycle in an anaplerotic reaction 3 . Additionally, alanine functions as a glucogenic amino acid supporting gluconeogenesis. The inhibition of the HMG-CoA reductase by Simvastatin may lead to an accumulation of acetyl-CoA, which is fuelled in the TCA cycle instead of the lipid synthesis. This rewiring of acetyl-CoA could be the reason for the increased secretion of alanine in statin treated A375 cells and may contribute to statin induced apoptosis in particularly fast growing metastatic cells 10,16 . In addition to the amino acid profiles, the activity of caspase 3, 8 and 9 was determined after 24 h and 48 h, to combine the induction of apoptosis with metabolic adaptation (Supplementary Table S1). Via PLS regression, a correlation between specific amino acids rates (Gln, Glu, Ser, Gly, Ala, Cys-Cys and Lys) and caspase 3 activity could be established for the metastatic A375 cells (Fig. 5C,D). Based on the PLS regression the induction of caspase 3 by simvastatin could be predicted and significantly separated from untreated A375 cells (Fig. 5C) and thereby indirectly confirmed previous observations for Simvastatin induced apoptosis 8,10 . Summing up, the hypothesis is supported that alanine and/or glutamic acid enhance the growth characteristics of melanoma cells. Growth effects of glutamic acid and alanine on melanoma cells. In order to confirm a biological effect of alanine and/or glutamic acid at concentrations determined in conditional media (Supplementary Table S1), the proliferation of slow and fast growing melanoma cell lines was analysed and normalized to proliferation of untreated cells (Fig. 6). Clearly, glutamic acid or the combination of glutamic acid and alanine stimulated proliferation in slow growing WM35, WM278, WM793b and VM21 cells significantly, while alanine per se had no or only a weak effect. Conversely, in the fast growing metastatic cells (A375, 518a2, 6F and WM8) the amino acid combinations had no growth stimulatory effect (Fig. 6A). Upon inhibition of the mevalonate pathway by Simvastatin this growth stimulatory effect was completely abrogated in all melanoma cells (Fig. 6). However, in some melanoma cells (A375, 518a2 WM35 and WM793b cells) proliferation is completely inhibited after 48 hours in the presence of simvastatin. Interestingly, under these conditions caspase 3 activation is significantly activated in metastatic cells, while in early growth phase melanoma cells threshold levels were reached, which were significant only for WM278 cells (Supplementary Fig. S2). In order to corroborate growth stimulation of glutamic acid and alanine, migration was investigated in a wound healing assay (Fig. 7). Again, the stage dependent differences in proliferation were mirrored in delayed gap closure of WM35 and WM793b cells (Fig. 7). Gap closure was significantly faster in WM35 and WM793b cells by addition of glutamic acid either in the absence or presence of alanine, while migration of A375 cells was not influenced. Again, Simvastatin significantly inhibited migration of all three melanoma cell lines (Fig. 8A). However, the effect was less pronounced in WM35 and WM793b cells. The addition of glutamic acid, alanine or a combination was not effective in accelerating migration (Fig. 8B-D). Finally, we investigated the invasion of melanoma cells to expand amino acid effects on aggressiveness and malignancy. Within 24 hours invasion of slow growing melanoma cells (WM35, WM278, WM793b and VM21) was significantly enhanced in the combination of glutamic acid and alanine (Fig. 9B). In contrast, invasion of fast growing metastatic melanoma cells was not altered by glutamic acid and alanine addition. Although invasion of A375 cells was significantly enhanced by the combination of glutamic acid and alanine, these data are considered inconclusive, since glutamic acid alone significantly reduced invasiveness (Fig. 9A). Again, inhibition of the mevalonate pathway significantly inhibited invasion of metastatic melanoma cells, while in slow growing melanoma cells this was only seen in WM278 cells (Fig. 9). Taken together, these data confirm that an inhibition of the mevalonate pathway arrests proliferation, migration and invasion of melanoma cells, which is not compensated by amino acid rescue metabolism like glutaminolysis. Discussion In the present study we have demonstrated that: (i) amino acid profiles from conditional media of melanoma cells from different growth stages are significantly different in their composition; (ii) the differences are of biological significance and corroborate acceleration of proliferation, migration and invasion by glutamic acid in slow growing melanoma cells, but not in metastatic cells; and (iii) inhibition of the mevalonate pathway overcomes the growth stimulatory effect of glutamic acid and alanine, highlighting the stringent dependency on this pathway in melanoma cells. A central issue of transformed cells, including melanoma, involves high glycolytic flux, which transforms glucose to lactate 3 . Stimulation of this pathway branches also into the pentose phosphate pathway and the serine synthesis pathway 24 . The latter is of particular interest in melanoma because the first reaction in this pathway is catalysed by phosphoglycerate dehydrogenase (PHGDH), which is amplified in melanoma 25 . Of note, a decrease in PHGDH expression reduced proliferation not only in melanoma cells but also in breast cancer cells, and 4). Significance vs. control treatment was performed with one-way ANOVA and post hoc Student-Newman-Keuls test and indicated with asterisks (p < 0.05, p < 0.01, *p < 0.005). Significance vs. L-alanine or L-glutamic acid were indicated with # or §, respectively ( # p < 0.05, ## p < 0.01, § p < 0.05; n.s. denotes not significant). SCIEnTIfIC RepORtS \| (2018) 8:6245 \| DOI:10.1038/s41598-018-24709-0 heterologous overexpression in non-transformed cells predisposes cells to transformation 24,25 . Although these studies link a glycolytic enzyme to proliferation of cancer cells, serine or glycine were not identified as discriminators for differences in growth behaviour in the melanoma cells investigated in our study. Under hypoxic conditions tumour cells accumulate lactate, the so called Pasteur effect 26 . Although we have not determined lactate, we could measure high levels of acetate in the conditional media. It has been previously shown that acetate is the main contributor to the formation of acetyl-CoA to fuel fatty acid synthesis 27 . However, lipid synthesis starting from acetyl-CoA branches also into the mevalonate pathway, which is responsible for the synthesis of integral intermediates for proper tumour growth and progression 7 . We have previously shown that statins reduce the dolichol levels in human neuroblastoma cells and thereby reduce the glycosylation of integral membrane proteins like the ATP-binding cassette transporter (ABCB1; P-glycoprotein) 11,15 . Another crucial link between tumorgenesis and glucose consumption has been elucidated by Cheng et al., showing that increased glucose uptake leads to N-glycosylation of the sterol regulatory element-binding protein (SREBP) cleavage-activating protein (SCAP) and consequently in lipogenesis via activation of SREBP-1 28 . Indirectly, our results confirm the crucial role of the mevalonate pathway in proliferation, migration and invasion (Figs 1, 6, 8 and 9). Simvastatin significantly inhibited the proliferation and migration of melanoma cells, which was not compensated by the co-application of glutamic acid and alanine. Interestingly, invasion was also significantly suppressed by Simvastatin in all cell lines, except WM35 cells (Fig. 9). Blockade of invasion might be explained by inhibition and reduction of matrix metalloproteinases in non-transformed endothelial cells and cancer cells including melanoma cells [29][30][31] . Although epidemiological analyses have revealed a beneficial effect of statins in human cancers including melanoma, opposite reports exist 7,32,33 . Importantly, statins have been discussed as potential candidates for chemoprevention, alone or in combination with other drugs, like metformin or cyclooxygenase inhibitors 7,12 . Nevertheless, statins are well tolerated but were evaluated for cholesterol lowering in cardiovascular risk patients, thus other derivates may exist with improved anticancer action. Mitochondrial glutaminolysis overcomes the decrease of TCA cycle metabolites under hypoxic conditions 34 . The glutamine uptake and consecutive activation of the isocitrate dehydrogenase 1 provides sufficient acetyl-CoA to support lipogenesis 3,6 . In this study, we have identified glutamic acid together with alanine as the primary determinates for discrimination between melanoma cells from different growth stages in a multivariate analysis. Glutamic acid is known to be taken up by non-transformed cells and tumor cells 35 . However, a specific function or an extracellular signaling cascade are not attributed to this amino acid. Interestingly, glutamic acid enhanced proliferation, migration and invasion are only observed in slow growing melanoma cell lines, but not in metastatic melanoma cells (Figs 6-9). An explanation of this phenomenon is currently not available. Nevertheless, dualistic effects have been described for the transcriptional coactivator peroxisome proliferator-activated receptor-gamma coactivator 1 α (PGC1α), the master regulator of mitochondrial biogenesis 36 . High levels of PGC1α, found in the radial growth phase of melanoma were assigned to promote growth and survival, while low levels were associated with invasion and metastasis. The authors conclude that signaling inputs including nutrients may guide changes in PGC1α and thereby switch from survival and proliferation to invasion and metastasis 36 . Alanine is considered a glycogenic amino acid, but a possible role for secreted alanine is currently not available. In our experiments, the biological effects of alanine alone were not significant in any melanoma cell line tested, indicating that the glycogenic effect is of minor importance to proliferation and migration (Figs 6 and 7). Nevertheless, secretion of alanine has been previously shown for melanoma cells but not for melanocytes 5 . Besides, we could identify an enhanced uptake of proline by slow growing, non-metastatic melanoma cells (Fig. 4). A source for extracellular proline in vivo is provided by its abundance in collagen, which is part of the extracellular matrix and therefore crucial for the tumour microenvironment 23 . Conversely, proline may be used as a source for ATP synthesis by collagen breakdown during tumour cachexia. Nevertheless, intracellular proline biosynthesis has been previously described in mitochondrial glutaminolysis or alternatively from cytosolic ornithine in WM35 and WM 793 cells 22 . Here we used multivariate data analysis as a tool, suitable to identify crucial amino acids from media of melanoma cell lines of different stages with specific biological characteristics. Our results confirm for the first time that amino acid profiles from conditional media are capable to distinguish between stage dependent growth and treatment. Hence, changes in amino acid composition of media from melanoma cells derived from different disease states reflect a stage dependent adaptation with functional consequences on proliferation, migration, aggressiveness and survival. Thus, identification of glutamic acid mediated signalling may open pharmacological targeting in early melanoma stages. Cell Proliferation. Melanoma cells were seeded in a 24-well plate at a density of 3.5 × 10 4 cells per well. Materials After an overnight recovery, cells were incubated with indicated compounds. The number of attached (living) cells was determined in duplicates at given time points. Attached cells were washed with PBS, detached with trypsin and resuspended in medium (500 µl) for automated cell counting (Luna II cell counter, Logos Biosystem, Villeneuve-d' Ascq, France). Data from three independent experiments were pooled. As a comparison, proliferation was also determined in DMEM-high glucose medium for all cell lines described above (data not shown). Scratch assay. Melanoma cells were seeded in a 12-well plate at a density of 7 × 10 5 cells per well. After an overnight recovery, the cell monolayer (70-80% confluence) was scratched off with a plastic pipette tip (200 µl) on the bottom of each well. Cells were washed twice with PBS and treated with compounds indicated in the figure legends. At the indicated points of time four pictures were taken from each well at labelled orientation points and three wells were measured for each condition. Three independent experiments were carried out. The percentage of cell-free surface was calculated by the TScratch software tool (CSElab, ETH Zürich, Switzerland) and normalized to the gap at the beginning of the experiment (t (0) ). Invasion assay. Membranes of Boyden chambers were coated with type I collagen (Thermo Scientific ® , Waltham, MA, USA; 300 µg/ml in PBS/NaOH) and kept overnight at 37° under aseptic conditions. Covered membranes were washed with PBS and 3,510 4 cells per well were allowed to adhere for 6 hours. Thereafter, the indicated compounds were added to the outer chamber at 37 °C in a 5% CO 2 humidified atmosphere for 24 hours. Thereafter, cells from the inner membrane were removed and invaded cells from the outer membrane were fixed (4% PFA/PBS), washed (PBS) and stained with Hoechst 33342 dye (20 µM). Membranes were mounted on a glass cover slip and 5 images were taken from one membrane using a fluorescence microscope Axiovert 200 M (Zeiss ® ; Oberkochen, Germany) at a magnification of 100x. Cell counting was performed with ImageJ software (www. imagej.net) and corrected for the individual proliferation rates of the cells. Data were normalized to the invasion of cells under control conditions. Quantification of amino acids and metabolites. Quantification of substances in the cell-free microenvironment was performed via HPLC analysis using an Ultimate 3000 (Thermo Fisher Scientific, USA) system equipped with a pump (LPG-3400SD), a split-loop autosampler (WPS-3000 SplitLoop), a column oven (Col. Comp. TCC-3000SD) and a fluorescence detector (FLD-3400RS). Chromeleon 7.2 was used for the control of the device as well as for the quantification of the peak areas. Chromatographic separation of amino acids was achieved with a reversed phase column (Agilent Eclipse AAA, 3 × 150 mm, 3.5 µm), a guard column (Agilent Eclipse AAA, 4.6 × 12.5 mm, 5 µm) and a gradient using eluent (A) 40 mM NaH 2 PO 4 monohydrate pH 7.8 and eluent (B) MeOH/ ACN/ H 2 O (45/45/10, v/v/v) 38 . The protocol was run with a flowrate of 1.2 mL min −1 , the column oven temperature was set to 40 °C and the injection volume was 10 µL. As most amino acids have no fluorophore in their structure an in-needle derivatization step was performed using 0.4 M borate buffer, 5 mg mL −1 ortho-phthaldialdehyde (OPA) in 0.4 M borate buffer containing 1% of 3-mercaptopropionic acid (3-MPA), 2.5 mg mL −1 fluorenylmethyloxylcarbonyl chloride (FMOC) and 1 M acetic acid for pH adjustment. In order to guarantee sample quantification despite the derivatization step, every sample and standard were spiked with 25 mM sarcosine and 25 mM norvaline both in 0.1 M HCl as internal standards. Primary amines and norvaline were detected at Ex 340 nm/Em 450 nm and secondary amines and sarcosine were detected at Ex 266 nm/Em 305 nm. Additionally, quantification and calibration were performed with an external standard including 22 amino acids. Acetate and Pyruvate were separated with an Aminex HPX-87H (Bio-Rad) column and guard column Aminex HPX 87-H. The method was run isocratically with 0.1% TFA in H 2 O at 50 °C for 60 minutes with a flowrate of 0.6 mL min −1 . Organic acids were detected at 210 nm and quantified by external calibration. All chemicals used for HPLC were of highest quality or HPLC grade quality. For the chromatographic method ultra-pure water was used, produced with a Milli-Q system from Merck Millipore (Billerica, USA). Calculation of specific rates. Specific rates for amino acids and organic acids were calculated from 24 h and 48 h values (Supplement Table S1), in order to allow the comparison of the data. The following formula was used: X is representing the cell count, dc i /dt is the change of the concentration of substance i over time and q i is the specific rate of substance i. A positive specific rate of substance i shows a secretion of i by the cells, whereas a negative rate represents an uptake of i by the cells (e.g. see PC4 in Fig. 4B; proline uptake by WM793b cells. (GraphPad Prism ® software). Pairwise comparison was done with Mann-Whitney Rank Sum Test or Students T-test. A value of p < 0.05 was considered as statistically significant. Multivariate data analysis was performed using Umetrics SIMCA 4.0 software (Umeå, Sweden). Data were normalized and mean-centred before principle component analysis (PCA) or partial least squares (PLS) regression. Models were generated for all data points or after rate calculation between 24 h and 48 h of incubation. PCA and PLS were used as statistical methods for the deconvolution of correlations in big datasets. The here presented dataset (Supplementary Table S1) includes more than 20 variables (amino acids and organic acids). In order to evaluate the linkage between these variables and the observations (i.e. the cell lines with and without Simvastatin treatment at different time points) statistically, the data set was restructured along the highest variance in the data. This new structure is represented by the generated principal components (PC) for PCA or latent variables (LV) for PLS. Interpretation of the new data structure is possible with the score plot, showing the distribution of the observations between two PCs. As example, we get 4 PCs, where each PC represents another variance in the dataset (e.g. differences in the uptake and secretion of amino acids). The highest variance is represented by PC 1, then PC 2, etc. If a score plot of PC 1 versus PC 2 is created the observations (e.g. cell lines with or without treatment) are distributed in this new coordinate system according to their differences that are illustrated by PC 1 and PC 2 (e.g. Supplementary Fig. S1). Depending on these differences clustering of the observations in the score plot can be observed. In order to understand which PC represents which variance, the loading plot has to be evaluated. The loading plot shows the distribution of the variables (e.g. amino acid rates) in the new coordinate system, hence, between PCs and gives information about the variables responsible for variance/clustering of the observations (e.g. Fig. 3D) 39 . A high value for a variable in the loading plot for PC 1 means that observations that cluster along PC 1 show a high rate/concentration of this variable. Hence, hypotheses about correlations between metabolic behaviour and cell lines and their treatment can be generated. The main difference between PCA and PLS is that PCA just reduces a dataset (e.g. amino acid rates of different cell lines) whereas PLS additionally allows correlations of this dataset with a response (e.g. caspase activity; Fig. 5). Hence, we can assume which variable (amino acid) correlates with higher or lower caspase activity in certain cell lines by the so-called coefficient plot. The model created via the PLS algorithm is then evaluated by plotting the observed values against the predicted values by the model (Fig. 5C). The residuals show how much the observed values vary from the model (Fig. 5D). Both plots give good estimation of the model quality.	v3-fos-license
2021-08-27T16:35:08.238Z	2021-07-07T00:00:00.000	238801760	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "http://www.scirp.org/journal/PaperDownload.aspx?paperID=111276", "pdf_hash": "d0c2bb3d16aebcc7d5533e372c8a5f2301840935", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45304", "s2fieldsofstudy": [ "Medicine", "Biology" ], "sha1": "040e39f92fb54ef490eb0ee4b604426ec283374a", "year": 2021 }	pes2o/s2orc	Indications, Interpretation and Clinical Consequences of Tuberculin Skin Tests in Resource Limited Settings Objective: to evaluate the policy of TST testing in Suriname. As there is no gold standard to diagnose latent tuberculosis infection (LTBI), the tuberculin skin test (TST) is used to diagnose LTBI. However, internationally, the cut-off values of the TST are not uniform and depend on local tuberculosis (TB) ep-idemiology and guidelines for test initiation. In Suriname, where currently several indications exist for TSTs, cut-off values are set at 5 mm or 10 mm, depending on the age and/or medical history of the patient. LTBI classification is performed by pulmonologists primarily based on the American Thoracic Society targeted TB testing guidelines. Method: retrospective analysis of outpatient TST data between 2011 and 2019 from Suriname’s sole pulmonary medicine clinic. Result: 1373 patients were evaluated. 590 patients were from the screening group of whom 253 had a positive TST result, 46 of whom were classified as LTBI. In the contact tracing group of 649 patients, 616 had a positive TST, 352 of whom were classified as LTBI. In the medical condition group of 134 patients, 96 had a positive TST, 38 of especially among patients who do not disclose TB risk factors. In our opinion the TST cut-off value for such patients in Suriname and other similar settings could be set at 15 mm. We also promote that for all patients with a TST result of 15 mm or greater, offering IPT should be considered (after excluding active TB). Introduction Infection of humans by tuberculosis (TB) bacilli can either result in no disease symptoms and spontaneous healing, active disease (with pulmonary TB being the most common and infective form), or latent TB infection (LTBI). The latter, non-infective form [1] [2], is the major source of new active TB cases [3] [4] [5] [6]. In 2014, some 1.7 billion people were estimated to have LTBI, equivalent to 23% of the world population [4]. LTBI is defined by WHO as "a state of persistent immune response to stimulation by Mycobacterium tuberculosis antigens with no evidence of clinically manifest active TB". The lifetime risk of LTBI reactivation in healthy individuals ranges from 5% to 15% but may increase depending on concomitant risk factors and comorbidity, eventually resulting in the potential development of active TB. As such, people living with HIV (PLHIV) have about a 100-fold increased risk of developing active TB [7]. As there is no gold standard to diagnose LTBI [3] [5] [7], TB contact screening, Interferon Gamma Release Assays (IGRAs), and tuberculin skin tests (TSTs) are used in clinical practice to evaluate LTBI. The IGRA has a better predictive ability but is expensive compared to the TST, which in turn has the disadvantage of requiring multiple visits to the test facility [8]. According to criteria of the American Thoracic Society (ATS), skin indurations of 5 mm or 10 mm are classified as a positive TST result depending on the patient's age, HIV status, TB exposure, chest X-ray findings, comorbidities, injection drug use, and occupation. A TST result of 15 mm or greater is considered positive in persons with no risk factors for TB [9] [10] [11]. Suriname, a low-and middle-income country in the South American tropics, can be considered a low TB incidence country (defined as a TB incidence rate of less than 100 per 100,000) [7], with an estimated TB incidence of 24 per 100,000 in 2000, 46 per 100,000 in 2010, and 29 per 100,000 in 2019 [12]. In 2012, the National Tuberculosis Program (NTP) launched the first national TB guidelines on behalf of the Ministry of Health. Per these guidelines, the cut-off values for TSTs were set at 5 mm for PLHIV and children under 5 years of age and 10 mm for all other persons [13]. For decades, Bacillus Calmette-Guerin (BCG) vaccinations have not been administered in Suriname. As it turns out, the BCG vaccination program carried out from 1955 to 1956 among 56,000 of Suriname's inhabitants was the last mass vaccination program. Nonspecific tuberculin sensitivity in Suriname seems to be frequent, as reported by Bleiker and van Erpecum from 1962 to 1965 and by van Weissenbruch in 1979. Between 1966 and 1971, screening for TB using chest X-ray among 60,301 people resulted in signs of TB for only 60 people. Also, from 1965 to 1974, a tuberculin survey among 156,000 people revealed only 88 positive cases [14]. The population of Suriname was 379,607 in 1970 [15]. Nontuberculous Mycobacteria (NTM) are commonly present in Suriname [16]. The WHO encourages countries to develop and implement national TB guidelines at their own discretion [7]. As such, all patients with TB in Suriname should be registered with the NTP, a government workforce that is responsible for TB contact tracing and public health measures regarding TB. Since IGRAs are not available in Suriname, the TST is used to identify LTBI. Other than for TB contact tracing, in Suriname the TST is sometimes required for a work permit or for admission to a nursing home. In addition, countries like the USA and Belgium require a negative TST result or LTBI treatment from Surinamese nationals for the purposes of emigration or study. Having utilized the TST cut-off values of 5 mm and 10 mm for many years, we had the impression of over designation of positive TST results, as a consequence of the NTP guidelines. As such we were curious to evaluate how the TST results, of TB contact tracing patients, screening patients, and patients with a medical condition other than TB, were dealt with in clinical practice and if adjustments needed to be made to the TST policy of Suriname. Our investigative parameters were a) the reason a TST was performed; b) the numerical value and designation of the TST; and c) the policy pursued by the treating physician and eventually the pulmonologist. TST Testing Algorithm and Patient Workup Indication: to evaluate tuberculosis exposure, the TST can be ordered by any physician practicing in Suriname, which is performed by the NTP with purified protein derivative (PPD RT23/SSI Copenhagen Denmark) ( Figure 1). NTP personnel measure the skin induration in mm and designate the TST result as negative or positive, according to the NTP guidelines. Tested persons with a positive TST result are advised to consult a medical specialist for additional evaluation, namely a pulmonologist for adults and a pediatrician for children. Interpretation: for the classification of a positive TST test result a detailed anam- nesis is taken at the pulmonology clinic and a chest X-ray is performed, which is evaluated for pulmonary lesions like consolidation, broadened hila, granulomas and miliary TB [17]. Subsequently the pulmonologist categorizes the TST as done in case of 1) screening (for a work permit, admittance to a nursing home, study abroad, or emigration); 2) TB contact tracing; or 3) an additional test to rule out Clinical consequence: no follow-up is conducted by the pulmonologist if a person has a negative TST result and is without an anamnesis or signs of TB. In the TB contact tracing group and the medical condition group, patients classified as LTBI [9] by the pulmonologist are offered isoniazid preventative treatment (IPT), whereas patients with a positive TST but not classified as LTBI are assigned to a "wait and see, no IPT" policy for 2 years-as this is the most likely time for TB reactivation to occur [5]-except in the presence of comorbidities such as diabetes mellitus (DM), HIV, cancer, and chronic renal failure. No follow-up or "wait and see, no IPT" policy is applied to screening group patients who have a positive TST result (as designated by the NTP) but who were not classified as LTBI by the pulmonologist. Patients classified as LTBI are offered 6 months of IPT and are monitored by pulmonologist for adverse reactions, liver function tests-aspartate aminotransferase (AST) and alanine transaminase (ALT)-and treatment adherence. IPT is completed after 6 months of isoniazid intake, during which the patient should have no signs of active TB. If an LTBI patient declines IPT or does not complete IPT, follow-up evaluations are offered for a period of 2 years [5]. Study Protocol We compiled outpatient data and conducted a study regarding the TST results of patients-who had a TST done at the NTP between January 1 st , 2011 and December 31 st , 2019 and were referred to Suriname's sole pulmonary medicine clinic at the Academic Hospital Paramaribo (children were mostly referred to the pediatrician). Follow-up data of these patients were collected until December 31 st , 2020. Excluded from analysis were patients who had been treated for TB or presented with active TB and persons who had no numerical value for their TST result. The extracted data were anonymized and categorized into three TST indication groups (i.e., the screening, TB contact tracing, and medical condition groups), and included: sex, age, TST indication and TST result, co-morbidities known to be risk factors for TB [7], and the interventions undertaken by the pulmonologist. Statistics Statistical analyses included the calculation of follow-up durations using the midyear date of the year in which the TST was performed (the exact dates on which TSTs were carried out were unavailable) and the end date of the study period. Differences between TST groups-which are based on the reason for the test: screening, TB contact tracing, and medical condition-on selected variables were analyzed with chi-square tests for categorical variables. Statistical analyses were performed using SPSS version 20.0 (computer software; IBM Corp., 2011). For significance testing, all alpha levels were set at 0.05. Ethics Our study was performed with anonymized patient data from our own clinic and was approved by the Ministry of Health. Results During the study period, 1384 patients with a TST result were referred. Excluded from analysis were 11 patients who had no numerical TST result. Data from 1373 patients were analyzed. Mean age at the time of referral was 37.3 years; the average follow-up time for all positive patients was 4.3 years. Our study cohort included 96 children, of whom 11 were under the age of 5 years; 85 were between 5 and 15 years. Based on the NTP cut-off values, 408 patients (29.7%) had a negative TST result (including 24 patients from the TB contact tracing group with a TST result between 5 mm and under 10 mm; see Table 1 and Figure 2 IPT was prescribed to 325 (74.5%) of the LTBI patients, of whom 284 (87.4%) completed treatment. The remaining 529 patients (54.8%) with a positive TST result, of whom 91 were lost to follow up after the first visit to the pulmonologist, were not classified as LTBI by the pulmonologist, although 174 (32.9%) had a TST result that was 15 mm or greater. In the latter group, 142 patients were not prescribed IPT by judgement of the pulmonologist, while the other 32 patients with a TST result of 15 mm or greater did not return to the pulmonologist after their first consultation. The reason a TST was performed, the numerical value of the TST, and the classification of the TST result by the pulmonologist are shown in Figure 2. In the screening group of 590 patients (43% of the study cohort), a total of 46 patients (7.8%) were classified as LTBI, but 64 patients (out of 93 with a TST result of 15 mm or greater) were not classified as LTBI and were assigned to a wait-and-see policy. Also, in the screening group, 159 patients had a TST result between 10 mm and under 15 mm, of whom 17 (10.7%) were classified as LTBI. In the TB contact tracing group of 649 patients (47.3% of the study cohort), 352 patients (54.2%) were designated as LTBI. Even in the TB contact tracing group, 93 patients (out of 276 with a TST result of 15 mm or greater) were not classified as LTBI and were assigned to the wait-and-see policy. In the medical condition group of 134 patients (9.8% of the study cohort), 38 patients (28.4%) were classified as LTBI (see Table 2). During follow-up, three patients with a TST result between 10 and 14 mm developed active pulmonary TB between 3 to 7 years after their first visit to the pulmonologist. Two patients were from the TB contact tracing group and had been classified as LTBI and were prescribed IPT, which was rejected by one patient and discontinued by the other patient due to elevated liver enzymes, alanine amino transferase (ALT) and aspartate aminotransferase (AST) (over two times the upper limit on two consecutive occasions that were at least 1 week apart). The third patient was from the medical condition group and had been treated for non-TB pneumonia in 2015 and presented with pulmonary TB in 2018. After presenting with active TB disease, these three patients were successfully treated with first line tuberculostatics [18]. Regarding progression to TB, three TB cases correspond to an incidence rate of 0.7 per 1000 person-years among those with a positive TST result. Regarding risk factors for TB, HIV was reported in 10 patients (0.7%) with 6 patients being on antiretroviral therapy (ART), of which 5 could be prescribed IPT; 55 patients (4%) with a positive TST result had DM, 25 were classified as LTBI, 17 completed IPT; 8 female patients had cancer (0.6%), 6 cases of breast cancer and two cases of ovary cancer of whom two were prescribed IPT; and 6 patients (0.4%) had chronic renal failure, of whom two were prescribed IPT. In our database, we came across three foreign patients with a TST result of 15 mm who had been BCG vaccinated in their home country. An additional 14 patients were noted with nontuberculous mycobacteria (NTM) in their sputum, with their TST result ranging from 0 mm to 21 mm. Other (LTFU, IPT discontinued, etc.) 11 28.9% Notes. † A chi-square test of independence was used to compare the proportion LTBI to that of non-LTBI cases within each TST group-χ 2 (2, N = 965) = 308. Discussion and Conclusion In our study cohort, 4 out of 10 patients had a TST done for screening purposes, of whom less than 8 out of 100 were ultimately classified as LTBI. Regarding the patients with a positive TST result, more than half were not classified as LTBI according to the ATS guidelines. While the ATS recommends a TST cut-off value of 5 mm for recent contacts of TB case patients (9), and 24 patients in our cohort did meet this criterion, their TST results between 5 mm and under 10 mm were designated TST negative and not classified as LTBI by pulmonologists because their anamnesis yielded no indication of overt exposure to TB and because of the abundant prevalence of NTM [16] in tropical Suriname. None of these patients developed active TB during the following 1 to 9 years (until the end of 2020), but latent TB infection is still possible [5]. On the other hand, these positive TST results may have been the result of cross-reactions with NTM [19]. Incorrect administration of PPD or TST reading errors could also play a role, but this seems less likely. A positive TST result due to BCG vaccinations may be ruled out because mass BCG vaccinations in Suriname were suspended in 1956. In our opinion, the high proportion of positive TST results combined with the relative paucity of progression to active TB cases during follow-up-especially among those in which, based on clinical judgment, it was decided not to prescribe IPT-is due to the arbitrarily opted cut-off value of 10 mm for the TST. In all three indication groups, there were patients with a TST result of over 15 mm that were not designated as LTBI but were submitted to the wait-and-see policy. The reason for this approach could not be clarified. However, none of these patients developed active TB. The three patients with a positive TST result who did progress to active TB, years after initial presentation, could be cases of reinfection and not necessarily reactivation [20]. Still a more comprehensive policy in the approach of these patients is warranted and therefore included in the general approach that we propose below. The TST remains a valuable tool in the diagnosis of LTBI, especially in resource-limited settings and in the absence of advanced diagnostic means. We observed that there are still indications for performing TSTs in Suriname that are not supported by ATS or other guidelines. The requirement of a TST before being allowed to emigrate to another country is not subject to our decision making, and therefore cannot be abandoned, but the cut-off value for the designation of a positive result should be raised to 15 mm in people with no TB risk factors. Given the low TB incidence, requesting a TST for elderly home admissions or work permits is not indicated and should in our view be abandoned. In such cases anamnesis, physical examination, and chest X-rays should suffice to rule out infectious tuberculosis. The cut-off value of 5 mm for people with HIV should be maintained and followed by an examination of sputum using the GeneXpert test. If this test is negative, initiation or continuation of ART should be advised. This advice is based on our finding that most HIV infected TB patients in Suriname have low CD4 counts and/or are not using ART [21]. The cut-off value of 5 mm should also be used and designated positive for those in recent contact with TB patients. For patients with medical conditions the cut-off value should be related to TB exposure and illness being a risk factor for TB, as advocated by the ATS [9]. All patients with a TST result of over 15 mm should be considered infected, but before the diagnosis of LTBI is considered and IPT is offered, it is important to verify that such a patient does not have active TB. If IPT is contra-indicated, rejected, or terminated due to side effects, patients should be followed by the pulmonologist and properly instructed to report symptoms that could indicate active TB. Regarding children, the very low number of children under 5 years of age with a positive TST result in our cohort is probably because children are preferentially referred to the pediatrician and does not allow for any conclusions or suggestions about alterations to the algorithm applied to this age group. In view of our results, we recommend the TST cut-off value to be set at 15 mm for people with no risk factors for TB, based on the following data: Suriname is a low TB incidence country, with an abundant prevalence of NTM and historical nonspecific tuberculin reactivity, where progression to active TB-in persons with a TST test designated positive by the current NTP guidelines-is very low. The strength of our study is the sample size of the population that could be evaluated during a prolonged testing period. We assume that our study cohort is representative for the tested population in Suriname since people with a positive TST result are almost exclusively referred to the pulmonologist. A limitation of this study is that while our outpatient clinic is the only pulmonary medicine clinic in Suriname, we may have missed patients with a positive TST result who did not visit our clinic after referral by the NTP, particularly people living in remote areas of the country with limited access to health care. Another limitation of our study concerns patient follow-up. Over half of patients with a positive TST result, including one in four patients classified as LTBI, were lost to follow-up or had no follow-up appointment with the pulmonologist. We may have missed people developing active TB who moved abroad and people who have died of TB without being diagnosed as such. In conclusion, considering the incidence of TB and NTM in Suriname, our study emphasizes that TST results should be interpreted depending on the reason the TST was requested-in line with TB risk assessments and in conjunction with the incidence of TB and NTM-as advocated by the ATS. Diagnosis of LTBI should be made considering the TST in relation to regional TB incidence, patient exposure to TB and chest X-ray abnormalities compatible with undiagnosed TB. Adherence to this policy may lead to healthcare cost reductions and could result in less anxiety and less inconvenience for patients. In our opinion the TST remains a valuable tool in the diagnosis of LTBI, especially in resource-limited settings and in the absence of advanced diagnostic means. As such, our results probably may be of value to communities with a comparable TB incidence.	v3-fos-license
2023-02-09T16:13:48.759Z	2023-02-01T00:00:00.000	256686117	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1422-0067/24/4/3227/pdf?version=1675696867", "pdf_hash": "9aa16611f9cee6b30145d791c35a5c6e17542af3", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45305", "s2fieldsofstudy": [ "Biology", "Medicine" ], "sha1": "cb94377d08a9e1ceada0cf3d161570312fd901db", "year": 2023 }	pes2o/s2orc	Bone Marrow-Derived Mesenchymal Stem Cell Implants for the Treatment of Focal Chondral Defects of the Knee in Animal Models: A Systematic Review and Meta-Analysis Osteoarthritis remains an unfortunate long-term consequence of focal cartilage defects of the knee. Associated with functional loss and pain, it has necessitated the exploration of new therapies to regenerate cartilage before significant deterioration and subsequent joint replacement take place. Recent studies have investigated a multitude of mesenchymal stem cell (MSC) sources and polymer scaffold compositions. It is uncertain how different combinations affect the extent of integration of native and implant cartilage and the quality of new cartilage formed. Implants seeded with bone marrow-derived MSCs (BMSCs) have demonstrated promising results in restoring these defects, largely through in vitro and animal studies. A PRISMA systematic review and meta-analysis was conducted using five databases (PubMed, MEDLINE, EMBASE, Web of Science, and CINAHL) to identify studies using BMSC-seeded implants in animal models of focal cartilage defects of the knee. Quantitative results from the histological assessment of integration quality were extracted. Repair cartilage morphology and staining characteristics were also recorded. Meta-analysis demonstrated that high-quality integration was achieved, exceeding that of cell-free comparators and control groups. This was associated with repair tissue morphology and staining properties which resembled those of native cartilage. Subgroup analysis showed better integration outcomes for studies using poly-glycolic acid-based scaffolds. In conclusion, BMSC-seeded implants represent promising strategies for the advancement of focal cartilage defect repair. While a greater number of studies treating human patients is necessary to realize the full clinical potential of BMSC therapy, high-quality integration scores suggest that these implants could generate repair cartilage of substantial longevity. Introduction Osteoarthritis is a chronic disease, primarily affecting the elderly or obese population, involving diffuse degeneration of articular cartilage in load-bearing joints [1,2]. Acquiring focal cartilage defects through traumatic injuries is another risk factor for developing the condition, one which commonly plagues young, active patients. While a minority of cartilage defects may regress, most increase in size in otherwise healthy people [3]. Given that current, less invasive treatment options cannot ultimately prevent the need for joint replacement [4], these injuries pose the risk of requiring early joint replacement and multiple subsequent revision surgeries in young patients [5]. Given the significant individual and organizational costs posed by knee arthroplasty [6] and the projected increase in the incidence of revision surgeries [7], a method 3.2 to 6 mm in diameter and 1.5 to 5 mm in depth in studies using rabbits. One study with horses as subjects involved treatment of defects 15 mm in diameter [32]. Seven studies used BMSCs harvested from the tibial bone marrow [22,24,[26][27][28][29]31], four from the ilium [23,25,32,33], and one from the femoral metaphysis [30]. One study used periosteum-, synovium-, adipose-, and muscle-derived MSCs as comparators [30]. The number of cells per implant varied from an order of magnitude of 10 5 [29] to 10 11 [26][27][28] in the studies which recorded this parameter. Protein scaffolds or gels were commonly based on collagen [22,24], hyaluronic acid [23,25], glycolic acid [26][27][28], or caprolactone polymers [31,33]. Exceptions included one study using platelet-enhanced fibrin [32] and another using demineralized bone as a scaffold [30]. Most studies implanted undifferentiated BMSCs into the chondral defects, while a minority investigated BMSCs differentiated into chondrocytes in vitro prior to implantation [27,28,33]. The timing of the final sacrifice of the animal subjects was at least three months for most studies and ranged between two [24] and twelve months [32]. Except for one [25], all studies showed that BMSCs generated superior integration results in comparison to cell-free controls; five of these showed a statistically significant difference [26][27][28]30,33]. Other comparators included MSCs from different sources [22,30]. Wakitani et al. used collagen gel embedded with MSCs for defect repair [22]. There was no difference in the quality of implant integration when comparing periosteum-derived MSCs (PMSCs) and BMSCs at 24 weeks. In contrast, Li et al. demonstrated superior integration of implant and host cartilage when using BMSCs [30]. After embedding all MSC types in demineralized bone, they demonstrated better outcomes in comparison to periosteum-, adipose-, synovium-, and muscle-derived MSCs. This difference was statistically significant (p < 0.05). Studies also differed regarding the composition of scaffolds or gels. Fan et al. directly compared the use of scaffolds composed of poly-(lactic-co-glycolic acid) (PLGA) with hybrid PLGA-gelatin/chondroitin/hyaluronate (PLGA-GCH) scaffolds [26]. Integration did differ between the groups at 24 weeks, the time of final sacrifice. The PLGA-GCH group showed better quality integration in comparison to the PLGA group. While the difference was not statistically significant, the former showed significantly better integration in comparison to the empty control cohort. A sufficient number of studies assessed integration comparably and were, therefore, amenable to subgroup analysis on the basis of scaffold type (Figure 3). Analysis demonstrated a statistically significant difference in favor of PLGA-based scaffolds (p = 0.001). While it was not valid to include Zhou et al.'s study in subgroup analysis, their study investigating polyglycolic-acid-hydroxyapatite (PLGA-HA) scaffolds demonstrated better integration in comparison to the cell-free scaffold and empty control groups. However, these differences were not statistically significant. Wakitani et al. used collagen gel embedded with MSCs for defect repair [22]. There was no difference in the quality of implant integration when comparing periosteum-derived MSCs (PMSCs) and BMSCs at 24 weeks. In contrast, Li et al. demonstrated superior integration of implant and host cartilage when using BMSCs [30]. After embedding all MSC types in demineralized bone, they demonstrated better outcomes in comparison to periosteum-, adipose-, synovium-, and muscle-derived MSCs. This difference was statistically significant (p < 0.05). Studies also differed regarding the composition of scaffolds or gels. Fan et al. directly compared the use of scaffolds composed of poly-(lactic-co-glycolic acid) (PLGA) with hybrid PLGA-gelatin/chondroitin/hyaluronate (PLGA-GCH) scaffolds [26]. Integration did differ between the groups at 24 weeks, the time of final sacrifice. The PLGA-GCH group showed better quality integration in comparison to the PLGA group. While the difference was not statistically significant, the former showed significantly better integration in comparison to the empty control cohort. A sufficient number of studies assessed integration comparably and were, therefore, amenable to subgroup analysis on the basis of scaffold type (Figure 3). Analysis demonstrated a statistically significant difference in favor of PLGA-based scaffolds (p = 0.001). While it was not valid to include Zhou et al.'s study in subgroup analysis, their study investigating polyglycolic-acid-hydroxyapatite (PLGA-HA) scaffolds demonstrated better integration in comparison to the cell-free scaffold and empty control groups. However, these differences were not statistically significant. Limited study numbers and variability in the recording of results precluded the completion of further subgroup analyses on the basis of, for example, scaffold or gel composition, cell density, or timing of sacrifice. Furthermore, the impacts of varying such factors were rarely investigated directly in the selected studies. Three studies did directly compare in vitro and in vivo differentiation methods [27,28,33]. Wu et al.'s study used BMSCs embedded in polycaprolactone polymeric films [33]. Both the mixed phenotype (undifferentiated and pre-differentiated) and bilayered pre-differentiated BMSC implants exceeded the integration quality observed in the undifferentiated group. In the case of the bilayered differentiated group, this difference was statistically significant (p < 0.05). The other two studies also showed statistically significant differences in the integration quality achieved by in vitro and in vivo differentiated BMSCs [27,28]. However, this difference was in favor of the in vivo group. Other Histology Scores Scores pertaining to repair tissue morphology and staining characteristics are represented in Table 1, as are the total histological scores. Morphology and matrix staining both showed similar trends to integration, with the performance of BMSC interventions exceeding that of cell-free scaffolds and empty controls across all studies. This difference was statistically significant for four of the ten studies making this comparison [23,28,30,31]. Furthermore, increased post-operative time was associated with an improvement in these scores. While total histological scores were similar between two BMSC cohorts and their respective cell-free comparators [26,28], the remaining majority of BMSC interventions consistently showed superior total scores in comparison to cell-free interventions or controls. This difference was statistically significant for six of the nine studies making this comparison [23][24][25][26][28][29][30][31]. Again, all studies showed a sustained and improved total score as the time between the intervention and sacrifice increased. Li et al. compared the performance of MSCs from various sources loaded on a demineralized bone scaffold [30]. They demonstrated that the performance of BMSCs with regard to repair tissue cell morphology and staining properties exceeded that of periosteum-, adipose-, synovium-, and muscle-derived MSCs to a statistically significant degree. The same is true for the total scores. In contrast, Wakitani et al.'s comparison between BMSCs and PMSCs did not show a difference in any of these categories [22]. While no study directly compared wholly different polymer scaffolds, Fan et al. showed that the hybridization of PLGA and a gelatin/chondroitin/hyaluronate component led to improved outcomes [26]. On assessment of cell morphology and matrix staining, the PLGA-GCH/MSC group showed significantly better results compared with the PLGA/MSC group at 24 weeks (p < 0.05). Total scores also differed significantly at 12 and 24 weeks (p < 0.05). Three studies directly compared undifferentiated and pre-differentiated BMSCs [26,28,33], with inconsistent results. Two of these showed better morphology, staining, and total scores for in vivo differentiated BMSCs in comparison to in vitro differentiated groups when using poly-lactic-co-glycolic acid (PLGA) scaffolds [26,28]. Twice the difference in total scores was statistically significant, and on one occasion, morphology and staining were significantly better in vivo compared to in vitro differentiation. Conversely, using polycaprolactone scaffolds, Wu et al. demonstrated that pre-differentiated BMSCs yielded superior performances in comparison to cells undifferentiated at the time of implantation [33]. Both the mixed phenotype (undifferentiated and pre-differentiated) and bilayered pre-differentiated BMSC implants had higher total scores than the undifferentiated group. Morphology and staining were not assessed. Risk of Bias Assessment Overall, there were only low or some concerns with regard to bias in the included studies ( Figure 4). The lack of an explicit statement regarding whether study subjects were randomized into each treatment arm mostly contributed to concerns. In addition, experimenters carrying out the interventions unavoidably were aware of whether animals received an implant or the cartilage defect remained empty. Otherwise, almost all studies included outcomes for all or nearly all participants, and most had blinded investigators for histological scoring. studies (Figure 4). The lack of an explicit statement regarding whether study subjects were randomized into each treatment arm mostly contributed to concerns. In addition, experimenters carrying out the interventions unavoidably were aware of whether animals received an implant or the cartilage defect remained empty. Otherwise, almost all studies included outcomes for all or nearly all participants, and most had blinded investigators for histological scoring. Discussion This review aimed to investigate the integration of native and implant cartilage, utilizing the results of animal studies using BMSC-seeded implants for the repair of focal cartilage defects of the knee. The quality of this integration is an important factor in determining the effectiveness of implants in repairing chondral defects [34]. However, conclusions regarding cartilage-cartilage integration in human knees following the use of MSC-seeded implants have been limited by a lack of clinical studies and small sample sizes [35]. The small number of human studies investigating MSC-based therapies have Discussion This review aimed to investigate the integration of native and implant cartilage, utilizing the results of animal studies using BMSC-seeded implants for the repair of focal cartilage defects of the knee. The quality of this integration is an important factor in determining the effectiveness of implants in repairing chondral defects [34]. However, conclusions regarding cartilage-cartilage integration in human knees following the use of MSC-seeded implants have been limited by a lack of clinical studies and small sample sizes [35]. The small number of human studies investigating MSC-based therapies have demonstrated promising results with regard to both integration and clinically observable outcomes [36]. None of the studies included in this review directly compared MSCs to ACI. However, MSC implant therapies have been shown to outperform the more wellestablished ACI technique in humans, in terms of both functional parameters and the appearance of repair tissue on MRI [37]. The potential benefits of MSC therapies warrant further investigation. The relative abundance of animal studies necessitated that these became the focus of this review. Twelve studies were included, reporting quantitative results of histological outcomes following BMSC implantation. Overall, BMSC-seeded implants achieved reasonably good quality integration. Pooled integration scores obtained through meta-analyses equated to more than one of the two sides of the implant under investigation that achieved integration with native cartilage. Furthermore, BMSC implants had better integration scores compared to controls, cell-free implants, and MSCs from other sources. Serial measurements also demonstrated that integration quality improved over time. Interestingly, the use of a PLGA scaffold was found to achieve better quality integration in comparison to other implants to a statistically significant degree. Despite these promising results, heterogeneity in interventions has made it difficult to draw conclusions on the best implant engineering techniques, a problem mirrored in the literature [38]. Furthermore, long-term follow-up to determine the functional benefit of MSC implants need to be addressed if these animal studies are ultimately to inform human studies and clinical practice. MSC Source The heterogeneity of MSC phenotypes is a well-documented issue in studies investigating their use in cartilage regeneration [37]. The success of implant integration relies on a multitude of biomechanical and joint environmental factors, many dependent on the tissue-engineering processes involved in cultivating, differentiating, culturing, and seeding MSCs onto the implants [34]. To eliminate some of this variety, this review focused solely on bone marrow-derived MSCs. Owing to their relatively well-developed and common use in the repair of animal cartilage [39] and in human trials [40], BMSCs are also likely to be the most translatable cell type for human studies. Nonetheless, the 12 studies did include three different bone sources of BMSCs. The tibial bone marrow was the most common (7/12 studies), followed by the iliac crest (4/12 studies), and one study used the femur. Therefore, heterogeneity is not eliminated, as BMSCs cultivated from different areas possess different properties which may predispose them for better cartilage regeneration [38]. For instance, Hermann et al. showed that iliac crest chondrocyte cells were superior to femoral head cells in chondrogenic potential [41]. Comparisons between different MSC sources in our included studies could not be made as there was insufficient comparably reported data to perform subgroup analyses. Further investigation with larger sample sizes comparing MSCs from different sites will be essential for the future selection of the most appropriate source for cartilage repair. The selection of MSCs through the expression of cell surface markers may play a future role in reducing heterogeneity even further. Studies investigating this have shown that a variety of different markers are expressed even amongst MSCs cultivated from the same area of the body [38]. The studies in this review did not quality control MSCs by cell surface marker expression, but this has been investigated in human clinical trials. Akgun et al. used flow cytometry to select MSCs for cluster of differentiation 105 (CD105), CD73, and CD90 and lack surface expression of CD45, CD34, CD14 (or CD11 b), CD79 a (or CD19), and Human Leukocyte Antigen-DR (HLA-DR) [37], following the standards stated in the Mesenchymal and Tissue Stem Cell Committee of the International Society for Cellular Therapy. Additional novel surface markers are now being used to characterize MSCs, with certain markers showing an association with the cell's ability to regenerate cartilage [38]. This means obtaining a molecular profile will be especially pertinent if future studies are to determine which profiles are associated with the best integration. In addition to the MSC source, several papers investigated if factors such as differentiation status [27,28,33] or transfection of genes [24] would affect integration. Again, there was an insufficient number of papers to perform subgroup analyses. Future follow-up studies with larger sample sizes are be needed to establish how important these factors are to cartilage integration. Implant Composition Another source of heterogeneity was in the composition of implants in which MSCs were embedded. These were largely composed of either protein or organic acid polymers. In our preceding systematic review, mainly focused on the use of ACI in humans, we found through subgroup analysis that collagen scaffolds were associated with significantly better outcomes across multiple clinical scoring systems when compared to other scaffold types [4]. Integration, assessed using magnetic resonance imaging (MRI), was also superior, although this was not statistically significant (p = 0.06). Given these findings, we investigated whether scaffold composition had any bearing on integration outcomes when using BMSCs in animal studies. Subgroup analysis showed that the use of PLGA scaffolds was associated with a statistically significant difference in integration quality when compared to other scaffold types. The findings of subgroup analyses in this and our previous paper demonstrate that the choice of polymer used for implant composition has an important influence on the outcome of regenerative cell therapies. Different scaffolds possess differing mechanical properties and have variable effects on cell expansion, differentiation, and retention within implants [42,43]. Graft survival may also be influenced by immunoreactivity to synthetic materials, like in the case of PLGA, which needs to be balanced against the potentially inconsistent quality of more benign natural materials such as collagen. Combinations of biomaterials also exist in numerous possible permutations. For example, combining PLGA with fibrin and HA may be more effective than PLGA alone in promoting chondrogenesis [42]. Among the studies included in our review, the best integration was achieved with the PLGA-gelatin/chondroitin/hyaluronate (PLGA-GCH) hybrid scaffold combined with gelatin microspheres impregnated with transforming growth factor-β1 [36]. Therefore, studying the performance of different scaffold types is as significant as the choice of cells used. The large-scale assessment of various combinations of cell type, core scaffold composition, and added components, ideally in human patients, represents a priority for future research in this domain. In summary, heterogeneity in techniques has made it difficult to select the best processes for maximizing integration, a problem which must be resolved by future human trials. Translation to Clinical Practice This review focused on animal studies due to the relative lack of human studies investigating MSC implant integration [4]. Not only are such results more widely reported by animal studies, but these papers also offer insight into histological appearances of cartilage repair, which is rarely the case in human studies. There are obvious barriers to the translation of these animal studies into clinical outcomes. Differences in cartilage geometrical and mechanical properties between different species is a limitation when extrapolating the findings. An additional value of human studies is afforded by being able to assess functional outcomes, which are possibly the highest priority for patients themselves. Comments on functional outcomes were limited in the included studies. In two studies, it was observed that the animals did not display an abnormal gait following the procedure [22,29]. This was despite immediate post-operative weight bearing, an important difference compared to human studies, which commonly involve external fixation and gradual weight-bearing regimens [22,44]. Bornes et al. [45] summarize studies reporting level IV evidence regarding the efficacy of BMSCs in humans. All of these demonstrated either a return to baseline levels of activity or improvements in validated clinical scores. Furthermore, a phase I randomized controlled trial comparing BMSCs embedded in atelocollagen scaffolds to microfracture demonstrated improved International Knee Documentation Committee (IKDC) scores for both techniques [46]. While the BMSC group showed a sustained improvement at the two-year follow-up, the IKDC scores of those treated with microfracture demonstrated a subsequent decline before one year, which continued until the final follow-up. Gobbi et al. [47] compared the efficacy of bone marrow aspirate concentrate (BMAC), containing BMSCs, and MACI, both using a hyaluronan scaffold. Improving clinical scores were seen from baseline to two years post-operation. Participants treated with BMAC showed a continued improvement between two years and final follow-up, while the MACI group showed declining function. Significant improvements in functional scores after the implantation of BMAC were seen in a further two studies conducted by the same group [48,49]. These studies exemplify the importance of long-term follow-up data for demonstrating the safety and long-term efficacy of interventions for cartilage repair, another feature lacking from animal studies. Within the twelve included studies, only one had a follow-up duration longer than one year [32]. The importance of these data is signified by our previous review, in which the longevity of implants was called into question [4]. Declining trends in imaging scores were sometimes only appreciable several years after treatment with MACI. Data from the MRI assessment of cartilage repair following implantation of BMSCs in humans are limited in availability but have shown encouraging results. Gobbi and colleagues [47][48][49] have consistently demonstrated high-quality repair following at least three years of follow up after the implantation of BMAC on polymer scaffolds. Imaging from a large majority of their patients has shown complete or near complete defect filling and complete integration with adjacent host cartilage. While the authors did not statistically correlate clinical scores with a quantitative assessment of the repair, using the Magnetic Resonance Observation of Cartilage Repair Tissue (MOCART) score [36], for example, they commented that good quality repair, as seen on MRI, was consistent with promising clinical outcomes. Search Algorithm This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA guidelines) [50]. A comprehensive literature search was conducted from conception to October 2022 using the following databases: (1) PubMed, (2) Embase, (3) MEDLINE, (4) Web of Science, and (5) CINAHL. The detailed search strategy can be found in Supplementary Table S1. This review was registered in the International Prospective Register of Systematic Reviews PROSPERO (CRD42022343052). Studies were uploaded onto the Rayyan website [51], where titles and abstracts were independently screened by EL and IEE before a subsequent full-text screen. A third (VL) and fourth reviewer (WK) were consulted for unresolvable disagreements. Inclusion and Exclusion Criteria Using the Population, Intervention, Comparison, Outcome, Study type (PICOS) model [52] as a guide, we formulated our inclusion and exclusion criteria for study selection (Table 3). Table 3. PICOS inclusion and exclusion criteria for study selection. Domain Inclusion Criteria Exclusion Criteria Population Animal subjects with surgically-created focal chondral defects of the knee joint. Studies involving human or cadaveric subjects. Studies involving animals with diffuse osteoarthritis models. Ex vivo, in vitro, or in silico studies. Intervention Studies using implants consisting of autologous or allogeneic mesenchymal stem cells seeded into an organic scaffold or gel which were surgically introduced into a focal chondral defect. Studies involving cell-free therapy, cell therapy without open surgical implantation, cells which are not bone marrow-derived, or other cell types such as chondrocytes, except as comparators. Studies using xenogeneic stem cells. Domain Inclusion Criteria Exclusion Criteria Comparison Studies comparing the use of scaffolds/gel implants to cell-free scaffolds/gels, cell therapy without implantation, or empty controls. None. Outcome Studies which quantitatively report the quality of integration of the regenerated and native cartilage, assessed histologically. Studies reporting outcomes of joints other than the knee. Studies without a quantitative scoring system. Studies not involving a histological assessment. Studies which do not specify integration scores. Study Type English articles with full text available. Sample size greater than 10 animals. Case reports, case series with fewer than 10 subjects, review articles. Search Results The structured search, using five databases, yielded 948 papers in total ( Figure 5). After removal of duplicates, 829 articles remained, of which 773 were excluded following title and abstract screening. Of the remaining 56 studies which underwent full-text screening, 12 were eligible for inclusion. Data Extraction Extraction of the data was independently performed by EL and IEE, using third and fourth reviewers (WK and VL) to resolve disagreements. A standardized table created in an Excel spreadsheet was populated with extracted data including: 1. Study characteristics and demographics including study design, animal type, cohort size, cartilage defect location, defect size, and time of sacrifice. 2. Type of intervention including source of BMSC and scaffold composition. Data Extraction Extraction of the data was independently performed by EL and IEE, using third and fourth reviewers (WK and VL) to resolve disagreements. A standardized table created in an Excel spreadsheet was populated with extracted data including: 1. Study characteristics and demographics including study design, animal type, cohort size, cartilage defect location, defect size, and time of sacrifice. 2. Type of intervention including source of BMSC and scaffold composition. 3. The cluster of differentiation (CD) molecule profile of cells. 4. Primary outcome measures regarding integration of the implant and native cartilage, assessed by histological scoring systems. 5. Secondary outcomes including total histological scores, data regarding the cartilage morphology, and qualitative descriptions of macro-and microscopic cartilage characteristics. Data Analysis Histological scores were recorded in a format defined by the grading scale used in each paper. These were namely original or modified versions of the Wakitani [22] and O'Driscoll [23] grading scales. Occasionally, papers use unnamed scales. These largely assessed similar constructs, including cartilage morphology, staining characteristics, underlying subchondral bone quality, surface regularity, defect fill, integration, and degenerative changes. Integration quality was usually assessed on a three-point scale: (a) both edges integrated, (b) one edge integrated, or (c) neither edge integrated. Points given (0, 1 or 2) depended on whether ascending or descending scores were assigned to a better or worse outcome. When comparable scoring systems were used, outcomes of integration of regenerated and host cartilage were pooled in meta-analyses. Total histological, cartilage morphology, and staining scores were considered pertinent to our understanding of the quality of cartilage repair and were, therefore, extracted. Due to variations in the recording of these assessments, they were not pooled in meta-analysis but analyzed qualitatively for trends. Meta-analyses were carried out using RStudio version 4.0.5. for continuous data. The estimator reported by Hozo et al. was used where the standard deviation was not provided in the manuscript [53]. Higgins and Thompson's I2 statistic and Cochran's Q test were used as measures of heterogeneity [54,55]. Prediction intervals were also included to provide a range into which future studies' effect size can be expected to fall into. Assessing Risk of Bias Risk of bias assessments were carried out independently by EL and IE, and VL was consulted for unresolvable disagreements. The Cochrane RoB 2.0 tool was used to assess the randomized trials according to its five domains [56]: (1) bias from the randomization process; (2) bias due to deviations from the intended interventions; (3) missing outcome data; (4) bias in measurement of the outcome; and (5) bias in selection of the reported result. These domains were each assessed as having low risk, some concerns, or high risk of bias, and an overall risk was determined. Results of the assessments were presented using the robvis package in RStudio [57]. Conclusions This review collated results of integration outcomes following the repair of focal cartilage defects of animal knees using BMSC-seeded implants. BMSC-seeded implants achieved good quality integration with the native cartilage. Studying integration in animals allows the establishment of a baseline by which investigations of BMSC-based treatments in humans can be informed. We also hope to reiterate future areas of focus for the optimization of this promising therapy. Future studies could focus on the standardization of techniques to reduce heterogeneity, namely cell source, implant composition, and molecular characterization of MSCs, to achieve an understanding of the best combination of these various factors.	v3-fos-license
2020-10-06T13:35:29.761Z	2020-09-29T00:00:00.000	222148635	{ "extfieldsofstudy": [ "Medicine", "Computer Science" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1424-8220/20/19/5582/pdf", "pdf_hash": "4bb0799fbbef37af2a83939e92d40e5ab8db8a8c", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45313", "s2fieldsofstudy": [ "Engineering", "Computer Science" ], "sha1": "b86249f2ed2cacff5bb5fd4cd89d4984ab5e6655", "year": 2020 }	pes2o/s2orc	Driver Monitoring for a Driver-Centered Design and Assessment of a Merging Assistance System Based on V2V Communications Merging is one of the most critical scenarios that can be found in road transport. In this maneuver, the driver is subjected to a high mental load due to the large amount of information he handles, while making decisions becomes a crucial issue for their safety and those in adjacent vehicles. In previous works, it was studied how the merging maneuver affected the cognitive load required for driving by means of an eye tracking system, justifying the proposal of a driver assistance system for the merging maneuver on highways. This paper presents a merging assistance system based on communications between vehicles, which allows vehicles to share internal variables of position and speed and is implemented on a mobile device located inside the vehicle. The system algorithm decides where and when the vehicle can start the merging maneuver in safe conditions and provides the appropriate information to the driver. Parameters and driving simulator tests are used for the interface definition to develop the less intrusive and demanding one. Afterward, the system prototype was installed in a real passenger car and tests in real scenarios were conducted with several drivers to assess usability and mental load. Comparisons among alternative solutions are shown and effectiveness is assessed. Introduction In recent years, studies have focused on the design and development of Advanced Driver Assistance Systems (ADAS) because human error is considered to be a major factor in traffic accidents, estimated to be 94% compared to other causes such as vehicles, the environment, and unknown critical reasons [1]. Some estimates [2] ensure that road accidents and fatalities would be reduced by up to 90% thanks to assistance systems. While it is true that many of these technologies have made driving safer and more comfortable, reducing road accidents [3], there are many complex scenarios in which risk situations can happen that have not been covered yet. Situations, such as merging, lane departure, and roundabouts, require a very manual and intuitive control for the driver, where there is also a lack of knowledge of the intentions and maneuvers of other vehicles and where the same situation will never occur due to the driving style of each driver. This paper proposes a system for providing information to the driver in merging maneuvers, which is a safe-critical situation. A driver assistance system is proposed, based on the knowledge of other vehicle positions using vehicular communications. Developing an appropriate interface is a key and duration of gaze fixations [19,20]. Studies have shown that merging situations demand a higher cognitive load than usual [21][22][23]. Considering this state-of-the-art, an application for merging assistance was developed in this work, based on Vehicle-to-Vehicle (V2V) communications, and parameters of the cognitive load were studied by means of an eye tracking system. This system provides the driver information to take accurate decisions on how to perform the maneuver. The system proposes the speed the vehicle should reach based on road and traffic conditions. The developed application was tested in a simulator and validated in real tests using V2V communications, compensating the shortcomings found in previous studies, such as the lack of tests in a real environment [15,17,18] and the lack of an interface systematic design, as well as the lack of evaluation of usability and acceptability by the user, which is quite common in human-machine interface (HMI) prototypes. Due to the relevance for safety of these maneuvers and the high cognitive load that could generate to drivers, this work tries to select the most convenient interface and to corroborate its usability and usefulness as well as overall driver satisfaction. This paper has as its starting point previous research on the same subject [21][22][23], in which driver attention on safety-critical scenarios have been tested and assessed. Section 3 shows a preliminary justification of the need of assistance systems in the safe-critical scenario of merging lanes. Section 4 explains the assistance system definition, based on V2V communications, the control architecture, and the algorithm that controls it. Section 5 corresponds to the design of the HMI and the tests carried out on the simulator for the choice of the interface to be integrated into the assistance application, considering objective and subjective measures (Satisfaction, Usefulness, Usability, and Mental Load) obtained from a sample of drivers. Section 6 shows the real tests carried out with the merging assistance system so its impact could be assessed. Finally, Section 7 provides the main conclusions. Justification A survey on behavior and sensations in merging carried out during the experiments corroborates the facts that support the previous statements. This survey served both to know the stress level perceived by the drivers and the style and behavior of these in merging situations into another road. Being a short online survey of 33 questions, it was possible to obtain a wide variety of profiles. A total of 99 participants responded to the survey, of which 53 were men and 46 women, aged between 18 and 72 years (mean = 37.68 and SD = 11.02), with a driving experience of up to 48 years (mean = 17.07 and SD = 11). About 90% of them regularly drive a car. There was a similar number of respondents at different levels of distance traveled per year (less than 5000, between 5000 and 10,000, between 10,000 and 20,000, and more than 20,000 km/year). The correlation between the different results of the survey was analyzed, obtaining the dimensionless value r. The value r oscillates between −1 and +1, being the signs for the positive or negative linear correlation. Table 1 summarizes the most noteworthy correlations. The main results obtained conclude that the variable years of driving experience has a negative linear relationship with the stress experienced in different merging scenarios. In the same sense, it correlates with the driver's age. This suggests that there is a tendency for drivers with more driving experience, as well as older drivers, to have lower levels of perceived stress in merging situations and for those with less experience, as well as younger drivers, to have higher perceived stress. Both variables (experience and age) are closely linked, so the experience variable is the most relevant in this context. There is a tendency for drivers with higher levels of stress to wait until the end of the lane to start braking, to be riskier, and to start merging earlier if they notice vehicles behind them start with the maneuver first and use the horn on the highway more frequently. On the other hand, drivers are usually less risky if they carry passengers in their vehicles. In any case, most drivers report that they rarely, or only sometimes, feel stress when they merge (approximately 70-80%); these stress levels are higher when the driver arrives without sufficient speed at the end of the lane, visibility conditions are poor, or they find a truck or bus in the lane where they wish to merge (50-60% recognize feeling a medium or high level of stress). On the other hand, approximately half of the participants recognize that if you reach merging with enough speed, the rest of the vehicles ease the maneuver. Only 15% of drivers stop before the acceleration lane signal if they cannot merge, 35% stop after the signal, and 50% stop at the end of the acceleration lane. Approximately half of the drivers do not stop at the beginning of the lane even if they have a short acceleration lane and do not see the exit clearly. However, the majority of interviewees (more than 80%) facilitate the merge of other vehicles onto the road on which they travel when conditions permit. Mainly, drivers when merging are looking at the vehicles on the road, the vehicle in front, and the length of the acceleration lane. With these results, we can conclude that the stress variable is associated with different undesirable behaviors of drivers on the road when they merge; for example, waiting until the end of the lane to brake when they cannot merge or being riskier when other drivers behind the driver want to merge before them. This relationship is probably not causal, but it seems appropriate to analyze how far a system to assist merging can improve and facilitate these maneuvers. That is why a merging assistance system is proposed, which the driver can consult while performing the maneuver and provides information on the vehicles on the road where it is intended to merge. The system should advise in terms of speed how much to accelerate or decelerate to perform the maneuver safely, either in front or behind the vehicle. Assistance System Definition The purpose of the assistance system is to help drivers while merging, providing them information about the right moment to enter the main road or how to perform this maneuver in terms of speed control (accelerating or decelerating, depending on the situation). It has been implemented in a smartphone that serves as the driver interface and implements the algorithm that handles the information from V2V communications and from a digital map. Although using a smartphone while driving is illegal in many countries, it must be noted that, in this system, this device only plays the role of the human-machine interface and cannot be manipulated during the driving task. Any previous configuration of the system must be introduced before starting the trip. This section provides the description of the technological parts of the assistance system. Hardware and Information Sources Vehicle-to-X (V2X) communications allow information exchange between vehicles, so the environment is converted into a cooperative system, which gives the vehicle a wider horizon of information, so that it can anticipate the decision making. Some C-ITS experiments, such as [24,25], employed vehicle-to-infrastructure (V2I) communications to display road information panels to the driver inside the vehicle by means of a human-machine interface (HMI). Several experiments also used vehicle-to-vehicle (V2V) communications for the exchange of information between vehicles [26,27]. Other solutions can be seen in [28,29], where V2I communication can be understood as a network of crossroads, traffic and construction signs, speed limits, and traffic signal information to automated vehicles. Similarly, V2V communication allows vehicles sharing their own information [30,31]. By integrating V2V and V2I communication with automated vehicle systems, an efficient "cooperative driving" network can be developed [28,32]. For these purposes, the research team has developed communication modules that could be placed both in the vehicles and the infrastructure [33][34][35][36]. The Decentralized Environmental Notification Messages (DENM) provided by a G5 On-Board Unit (OBU) installed in the vehicle are shared by all vehicles in a certain area and, optionally, these communication modules could be connected to Road Side Units (RSU) linked to the Traffic Management Centers of the Directorate General of Traffic (DGT TMC). Furthermore, the OBU incorporates a Global Navigation Satellite System (GNSS) NV08C-CSM chipset with an update frequency of 5 Hz. These OBUs are installed in vehicles in order to provide the information of absolute positioning and speed to other vehicles in the surroundings. Vehicles in a merging lane use this information as described in the following subsection. Furthermore, a digital map with previously included information of the merging lanes is also used and specific map-matching algorithms are used for vehicle positioning on the map [37]. Algorithm The merging algorithm indicates the need to accelerate, decelerate, or enter the main road and is more intuitive and simpler than the one presented in [18], for example. Figure 1 illustrates the position of the vehicles involved in an instant in which the algorithm is trying to decide how Vehicle 1 could enter the main road, considering that d i is the distance between vehicle i and the end of the merging lane, and v i is the speed of vehicle i. The algorithm is implemented in a mobile application with an iOS system. For these purposes, the research team has developed communication modules that could be placed both in the vehicles and the infrastructure [33][34][35][36]. The Decentralized Environmental Notification Messages (DENM) provided by a G5 On-Board Unit (OBU) installed in the vehicle are shared by all vehicles in a certain area and, optionally, these communication modules could be connected to Road Side Units (RSU) linked to the Traffic Management Centers of the Directorate General of Traffic (DGT TMC). Furthermore, the OBU incorporates a Global Navigation Satellite System (GNSS) NV08C-CSM chipset with an update frequency of 5 Hz. These OBUs are installed in vehicles in order to provide the information of absolute positioning and speed to other vehicles in the surroundings. Vehicles in a merging lane use this information as described in the following subsection. Furthermore, a digital map with previously included information of the merging lanes is also used and specific map-matching algorithms are used for vehicle positioning on the map [37]. Algorithm The merging algorithm indicates the need to accelerate, decelerate, or enter the main road and is more intuitive and simpler than the one presented in [18], for example. Figure 1 illustrates the position of the vehicles involved in an instant in which the algorithm is trying to decide how Vehicle 1 could enter the main road, considering that di is the distance between vehicle i and the end of the merging lane, and vi is the speed of vehicle i. The algorithm is implemented in a mobile application with an iOS system. The algorithm output is based on previous studies in which an intelligent speed adaptation (ISA) system was developed [38]. That system informed the driver about the speed to be acquired when reaching a road section, and suggested the driver, in qualitative terms, how much to accelerate or decelerate when approaching that section. In the case of the merging assistance system, the driver should receive similar information of how to adapt his speed to merge onto the main road in safe conditions. The main premises are as follows: 1. The safety margin between vehicles must not be less than two seconds because of previous studies on driver reaction time; 2. The maximum speed of the road in the acceleration lane must not be exceeded in any case; 3. The assumed acceleration and deceleration limits are 2 m/s 2 and 4 m/s 2 , respectively. The reason for defining a safety margin in terms of time is because higher speeds require larger distances to brake. Several studies maintain that a driver reacts, in the worst case, with a reaction time of 1.5 s to a surprise event, as an object that moves suddenly on the driver's route [39]. However, other studies found larger reaction times [40]. This is why, in a conservative way, a two seconds reaction time is chosen. The acceleration and deceleration levels have been stated considering stateof-the-art values. In [41], a collision warning system is implemented, where three levels of sensitivity The algorithm output is based on previous studies in which an intelligent speed adaptation (ISA) system was developed [38]. That system informed the driver about the speed to be acquired when reaching a road section, and suggested the driver, in qualitative terms, how much to accelerate or decelerate when approaching that section. In the case of the merging assistance system, the driver should receive similar information of how to adapt his speed to merge onto the main road in safe conditions. The main premises are as follows: 1. The safety margin between vehicles must not be less than two seconds because of previous studies on driver reaction time; 2. The maximum speed of the road in the acceleration lane must not be exceeded in any case; 3. The assumed acceleration and deceleration limits are 2 m/s 2 and 4 m/s 2 , respectively. The reason for defining a safety margin in terms of time is because higher speeds require larger distances to brake. Several studies maintain that a driver reacts, in the worst case, with a Sensors 2020, 20, 5582 6 of 19 reaction time of 1.5 s to a surprise event, as an object that moves suddenly on the driver's route [39]. However, other studies found larger reaction times [40]. This is why, in a conservative way, a two seconds reaction time is chosen. The acceleration and deceleration levels have been stated considering state-of-the-art values. In [41], a collision warning system is implemented, where three levels of sensitivity are established at the driver's choice. The value of 4 m/s 2 for deceleration has been taken, corresponding to the high sensitivity level to maintain greater distances. For the value of acceleration, Prestl [42] considered the range of accelerations for an intelligent system of this type must be between −2 m/s 2 and 1 m/s 2 to avoid problems in the traffic flow in case of an inappropriate decision. The control algorithm calculates the constant acceleration/deceleration that is required for Vehicle 1 in the acceleration lane in order to merge onto the main road where Vehicle 2 is moving at a constant speed. This acceleration calculation as a function of time t, given by Equation (1), is executed continuously so the speeds of both vehicles are constantly updated. The decision tree shown in Figure 2 starts considering that both vehicles maintain their speeds and the variable that determines the required action is the relationship between how long they take to reach the end of the merging lane (t 1 and t 2 ), given by Equation (2). −2 m/s 2 and 1 m/s 2 to avoid problems in the traffic flow in case of an inappropriate decision. The control algorithm calculates the constant acceleration/deceleration that is required for Vehicle 1 in the acceleration lane in order to merge onto the main road where Vehicle 2 is moving at a constant speed. This acceleration calculation as a function of time t, given by Equation (1), is executed continuously so the speeds of both vehicles are constantly updated. The decision tree shown in Figure 2 starts considering that both vehicles maintain their speeds and the variable that determines the required action is the relationship between how long they take to reach the end of the merging lane (t1 and t2), given by Equation (2). (2) In case Vehicle 1 reaches that point earlier than Vehicle 2 (considering the safety margin), no action is required for merging behind Vehicle 2 (and Vehicle 1 could even accelerate). In case the first condition is not fulfilled, the algorithm decides whether merging is executed in front of or behind Vehicle 2. The first scenario is possible if the maximum acceleration amax and maximum speed vmax are not exceeded, to guarantee the safety margin; estimating the final speed of Vehicle 1 is done as per Equation (3). * Otherwise, merging must be executed behind Vehicle 2. In this case, if the difference between t1 and t2 is smaller than the safety margin T, a deceleration is required; but, if not, Vehicle 1 could accelerate up to a maximum level. It must be noticed that, in case the algorithm decides that merging in front of Vehicle 2 is not possible, but merging behind it is feasible, the same flow chart is repeated with the vehicle behind Vehicle 2 in its lane (Vehicle 3), to verify that merging in front of it is safe enough. This process is repeated until a safe maneuver is found. In case Vehicle 1 reaches that point earlier than Vehicle 2 (considering the safety margin), no action is required for merging behind Vehicle 2 (and Vehicle 1 could even accelerate). In case the first condition is not fulfilled, the algorithm decides whether merging is executed in front of or behind Vehicle 2. The first scenario is possible if the maximum acceleration a max and maximum speed v max are not exceeded, to guarantee the safety margin; estimating the final speed of Vehicle 1 is done as per Equation (3). Otherwise, merging must be executed behind Vehicle 2. In this case, if the difference between t 1 and t 2 is smaller than the safety margin T, a deceleration is required; but, if not, Vehicle 1 could accelerate up to a maximum level. It must be noticed that, in case the algorithm decides that merging in front of Vehicle 2 is not possible, but merging behind it is feasible, the same flow chart is repeated with the vehicle behind Vehicle 2 in its lane (Vehicle 3), to verify that merging in front of it is safe enough. This process is repeated until a safe maneuver is found. The control algorithm used has a low computational load and it works correctly in real time, which is an advantage over other similar studies. Interface Design Human-machine interfaces must be designed in the most ergonomic way, both physically and mentally, to avoid being a distraction for the driver, but in many cases, when designing an ADAS, usability and user satisfaction are not usually evaluated and this fact could lead to warnings, misunderstanding, confidence losses, or a usefulness perception of the system. Furthermore, as the scenario involves a high attentional load, this HMI assessment is even more crucial. Some authors, such as Biondi [43], have developed a scale to assess the interface of several different assistance systems. As premises for this study and for the construction and subsequent evaluation of the different proposed interfaces, the Commission Recommendations of the European Communities of 26 May 2008, relating to safe and efficient in-vehicle information and communication systems, were followed [44]. In [45] are summarized some of the desirable features on the devices in the vehicles. For example, no average glance duration should be greater than 1.2 s and the device must not affect the vehicle control, neither to driver workload nor its situational awareness, attracting the person's attention only if necessary. The importance that the device need not only be safe and efficient but also accepted by the user, in other words, that it is perceived as useful and pleasant and that its use does not require any specific training, is pointed out in [46]. Furthermore, the designs chosen are based on the research group's own experience in designing and validating interfaces for safe driving on mobile devices in previous applications [38,47]. Interface Location In a previous study of the authors [22], two different scenarios of traffic density were shown, one involving a high attentional load for the driver (HAL) and another one of low attentional load (LAL). After this study, in [21,23], a Wilcoxon signed-rank test to paired-samples data was carried out for pupil diameter in merging situations and in baseline situations, whose values for the right eye were Z = −2.38 and p = 0.016 and for the left eye were Z = −2.1, p = 0.036. These values represent that there are statistically significant differences between both situations. The fixations were also analyzed using heat maps, with a remarkably hot zone common to both rear mirrors. When the subjects performed the merging maneuver, a large percentage of their fixations were located on the upper-inner part of the mirror in more than half of the tests conducted. This conclusion serves as a precedent for the proposal of the merging assistance system that is placed at the bottom of the vehicle A-pillar, in order to support the driver during the maneuver. HMI Design Proposals The proposed interfaces are shown below. Some of them are derived from the basic design and others are based on the field of aeronautics, in which variometers indicate to the pilot whether it is ascending or descending into an air mass. In summary, the proposed designs are based on a set of bars or circumferential segments, as can be seen in Figure 3. From the proposed interfaces, a team of experts belonging to the Department of Psychology of the Complutense University of Madrid finally chose three, Interfaces 1, 3, and 6, due to their operability and ease of understanding. Interface 1 was chosen as the simplest of those of that typology; Interface 3 is chosen because, in addition to the information provided by Interface 1 and the similarity with Interface 2, it adds the direction, which can point the side where to merge; and, finally, despite its similarity to Interfaces 4 and 5, Interface 6 was chosen due to its similarity to the speedometer of a vehicle control panel, to which drivers are used to and implies a completely different concept compared to the previous ones. Interface Selection in Driving Simulator In order to know the acceptability of these HMIs, both technically and socially, and to obtain a robust validation, several tests have been proposed in a driving simulator. The aim of these tests is to find out which interface design is less distracting for the driver and has more acceptance by the participants. The chosen interface would be implemented in the final system for the real driving tests. Instrumentation and Data Acquired Tobii Pro Glasses 2, a state-of-the-art eye tracking device consisting of glasses and a software controller connected to the eye tracker via an HDMI cable, were used for monitoring the driver activity. The glasses include two IR sensors and two cameras in each eye, which allows acquiring gaze and pupil data. In addition, a camera placed in the center and facing to the front recorded the video of the scene, in such a way that the software returns an image of the visual field of the driver and superimposes the point the driver is looking at in real time. The tests took place in a driving simulator in the laboratory. The eye data were recorded at 50 Hz with a spatial accuracy of 0.63° at 1.5 m. The transmission protocol is the one provided by Tobii [48], which operates on UDP. To analyze ocular data, Tobii Pro Lab software was used. In addition, data were collected of the driver's actions in the simulator. To evaluate the effect of the amount of light on the pupil, with and without an interface, a Digital Illuminance Meter model ISO-TECH 1332 A was used. From the proposed interfaces, a team of experts belonging to the Department of Psychology of the Complutense University of Madrid finally chose three, Interfaces 1, 3, and 6, due to their operability and ease of understanding. Interface 1 was chosen as the simplest of those of that typology; Interface 3 is chosen because, in addition to the information provided by Interface 1 and the similarity with Interface 2, it adds the direction, which can point the side where to merge; and, finally, despite its similarity to Interfaces 4 and 5, Interface 6 was chosen due to its similarity to the speedometer of a vehicle control panel, to which drivers are used to and implies a completely different concept compared to the previous ones. Interface Selection in Driving Simulator In order to know the acceptability of these HMIs, both technically and socially, and to obtain a robust validation, several tests have been proposed in a driving simulator. The aim of these tests is to find out which interface design is less distracting for the driver and has more acceptance by the participants. The chosen interface would be implemented in the final system for the real driving tests. Instrumentation and Data Acquired Tobii Pro Glasses 2, a state-of-the-art eye tracking device consisting of glasses and a software controller connected to the eye tracker via an HDMI cable, were used for monitoring the driver activity. The glasses include two IR sensors and two cameras in each eye, which allows acquiring gaze and pupil data. In addition, a camera placed in the center and facing to the front recorded the video of the scene, in such a way that the software returns an image of the visual field of the driver and superimposes the point the driver is looking at in real time. The tests took place in a driving simulator in the laboratory. The eye data were recorded at 50 Hz with a spatial accuracy of 0.63 • at 1.5 m. The transmission protocol is the one provided by Tobii [48], which operates on UDP. To analyze ocular data, Tobii Pro Lab software was used. In addition, data were collected of the driver's actions in the simulator. To evaluate the effect of the amount of light on the pupil, with and without an interface, a Digital Illuminance Meter model ISO-TECH 1332 A was used. Procedure The experiment was carried out in a Faraday cabin. Participants sat in front of a projector screen where three driving simulations of 5 min each were projected and in which they carried out the task of simulated driving. In each simulation, a different interface design was incorporated, which was presented counterbalanced by the participants. Before starting the experiment, participants were instructed to drive naturally and follow the interface indications. In addition, each of them configured the operation of the system as they understood as more intuitive. Regarding the data analysis, the design was within-subject. Each participant performed three experimental situations, one for each type of interface. For each one, the following variables and indexes were measured: (A) system acceptance (Satisfaction, Usefulness, and Usability); (B) mental load (through a subjective measure, Rating Scale Mental Effort (RSME), and a physiological measure, the Pupillary Dilation); (C) eye measurements (total time looking at the interface and number and average duration of fixings); and (D) measurement of interface tracking (percentage of tracking of the interface according to its state in speed and direction). After each experimental condition, where a different interface was shown, the Systems Acceptance questionnaire of [49] was answered by the participant. This questionnaire consists of nine 5-point rating-scale items, ranging from −2 to 2. These items score on two scales: Usefulness of the system and Satisfaction. In addition, the participants answered the System Usability Scale [50] and the Rating Scale Mental Effort [51]. At the end of the three driving simulations, they also answered the I-Driving Scale [52] and some questions about demographic information, as well as whether or not they would implement the application on their mobile if it was free. The average duration of each session was about twenty minutes. Visual information is acquired during eye fixations [53]. A longer duration of the fixations in one interface versus another would indicate that the former requires a longer time to process the information and thus involves a greater distraction from the main driving task. However, in this context, the number of fixations, or the total time looking at the interface, informs of better tracking of it. In this sense, we must remember that the interface state is changing, indicating continuously the speed required to perform a safe merging. With a dynamic interface, the driver needs to analyze the current state of the interface, react, return his gaze to the road, and look again at the interface to check the new status. In an interface that works properly, more fixations but of brief duration are expected, indicating that the driver is not distracted and that it is easy to process the information presented. For each dependent variable, a unifactorial Repeated Measures Analysis of Variance (ANOVA) in the interface design was performed. The adjustment of the Pairwise Comparisons was performed by the Bonferroni method. For each variable, three contrast conditions were considered (Interface 1, Interface 3, and Interface 6), except for the pupil diameter variable, in which the condition of the pupil diameter in the situation of driving without any interface was introduced in the analysis. In this last analysis, furthermore, a Helmert contrast was carried out, allowing the comparison between the pupil diameters when driving without the interface (baseline) versus the pupil diameter when the interface is active. Measures of Mental Load Regarding the results in the RSME, statistically significant differences in the mental load that requires the follow-up of the different interfaces were found, F (2;44) = 5.57, p = 0.007, η2 = 0.2. The drivers evaluated that Interface 6 supposes a greater effort than Interfaces 1 or 3, and no differences between these last two were found. The results can be seen in Figure 5. Measures of Mental Load Regarding the results in the RSME, statistically significant differences in the mental load that requires the follow-up of the different interfaces were found, F (2;44) = 5.57, p = 0.007, η2 = 0.2. The drivers evaluated that Interface 6 supposes a greater effort than Interfaces 1 or 3, and no differences between these last two were found. The results can be seen in Figure 5. For the pupil diameter variable, a statistically significant difference between the different conditions was found, with F (3; 57) = 19.68, p < 0.001, η2 = 0.51 for the left pupil and F (3; 57) = 17.31, p < 0.001, η2 = 0.48 for the right pupil. These differences occur between driving without the interface (baseline) and when the interface is active, being clear that the pupil dilates more in the second case. However, there are no statistically significant differences between the conditions with the different interfaces. Probably, these differences are reflecting the mental load difference performing a merging situation versus driving performing any other maneuver (baseline) (Figure 6), rather than following the interface advice. It must be considered that the pupil diameter is not affected by light conditions in these tests because the conditions in the laboratory do not change. For the pupil diameter variable, a statistically significant difference between the different conditions was found, with F (3; 57) = 19.68, p < 0.001, η2 = 0.51 for the left pupil and F (3; 57) = 17.31, p < 0.001, η2 = 0.48 for the right pupil. These differences occur between driving without the interface (baseline) and when the interface is active, being clear that the pupil dilates more in the second case. However, there are no statistically significant differences between the conditions with the different interfaces. Probably, these differences are reflecting the mental load difference performing a merging situation versus driving performing any other maneuver (baseline) (Figure 6), rather than following the interface advice. It must be considered that the pupil diameter is not affected by light conditions in these tests because the conditions in the laboratory do not change. For the pupil diameter variable, a statistically significant difference between the different conditions was found, with F (3; 57) = 19.68, p < 0.001, η2 = 0.51 for the left pupil and F (3; 57) = 17.31, p < 0.001, η2 = 0.48 for the right pupil. These differences occur between driving without the interface (baseline) and when the interface is active, being clear that the pupil dilates more in the second case. However, there are no statistically significant differences between the conditions with the different interfaces. Probably, these differences are reflecting the mental load difference performing a merging situation versus driving performing any other maneuver (baseline) (Figure 6), rather than following the interface advice. It must be considered that the pupil diameter is not affected by light conditions in these tests because the conditions in the laboratory do not change. Regarding other measures used, such as eye measurements and measurement of interface tracking, no statistically significant differences in the study depending on the interface design were found. A summary table of these measures' analyses with the values of their mean and standard error of the mean is shown in Table 2. Discussion and Interface Design Selection According to the results obtained in terms of the eye measurements and measurement of interface tracking, it is not possible to conclude a different operation of the three evaluated interfaces. This fact is reasonable, considering that the three interfaces are the result of a previous selection. However, there is a difference between the user acceptance of Interfaces 1 or 3 versus Interface 6, in the sense that the first two are considered more satisfactory, more usable, and more useful than Interface 6. These characteristics are fundamental when choosing or starting up a new system in vehicles. According to [49], a prerequisite for the introduction of new in-vehicle technology is the acceptance because it is unproductive to invest effort in designing and building an intelligent co-driver if the system is never switched on, or is even disabled. Considering jointly the results for the two variables of mental load (RSME and Pupillary Diameter), it can be concluded that following Interfaces 1 and 3 is more effortless than Interface 6. Therefore, although the difference in the pupil diameters is not statistically significant, it follows the same trend as the effort evaluation. On the other hand, the pupil diameter results reveal that, when a driver is in a merging situation, the mental load is greater than in an ordinary driving situation. Given all the results, Interface 6 is discarded to be implemented on the final system and Interface 3 is chosen for the next stage, due to the fact that this interface reports similar results as what Interface 1 provides, but provides more information. Implementation of the Final System on a Real Vehicle For the correct validation of the merging assistance interface, it was implemented in a mobile device placed in a vehicle to perform tests in real driving conditions in which attentional measures were analyzed recording eye movements, measures of mental load, and measures of acceptance of the system. Methodology In order to assess the suitability of the driver assistance system, the experiments from real driving were evaluated in the same conditions as in the driving simulator, by means of eye tracking and a questionnaire that summarizes aspects of usability, acceptability, and mental load, but a questionnaire on mental effort was introduced. This questionnaire is the National Aeronautics and Space Administration's Task Load Index (NASA-TLX) [54], which is widely used and has achieved solid arguments in human factors research [55,56]. It is also used today to assess subjective workload and cognitive effort for a task projected on an HMI as in [57][58][59]. The application is installed on an iPhone device that connects with the vehicle module via WiFi. Instrumentation and Data Acquired A vehicle with automatic gear shift to free the driver from the process of changing gears was equipped with the smart phone placed near the vehicle A-pillar, with the merging assistance application installed. This vehicle and the vehicles driving along the main road were equipped with OBU C-ITS G5 communication modules. The assistance application generates only visual warnings based on the speed information and the positioning collected by the GNSS on a digital map. To assess driver behavior, apart from acquiring actions on the vehicle, the same eye tracking system was used as in the previous section, Tobii Pro Glasses 2, and the same variables were analyzed. Procedure The real driving tests were carried out on a stretch of the M−45 motorway near Madrid (Spain), with three signaled acceleration ramps. Before performing the real tests, drivers had to complete the first part of the NASA mental effort questionnaire, in which they subjectively evaluate the workload of a task. Then, tests were performed with the instrumented vehicle and the assistance system to help them in the merging maneuvers. The average duration of each session was approximately fifteen minutes driving and five minutes more to answer the scales. It must be considered that these mobile devices located inside the vehicle can both aid the drivers or distract them from the main driving task [60]. In order to evaluate the functioning of the interface, various attentional measures were recorded by means of (A) measures of the assistance system acceptance; (B) measures of mental load; and (C) measures of eye movements. Different measures of system acceptance were taken for each participant at the end of the driving tests, answering the System Usability Scale [50] and the Systems Acceptance Questionnaire of [49]. In order to evaluate the mental load using the assistance system, ratings were recorded on the NASA Task Load Index [54], and on the Rating Scale Mental Effort [51], as well as the pupil diameter (during merging maneuvers with and without the assistance system). In order to evaluate eye behavior, the following parameters were recorded: the average number and duration of fixations to the interface; the average duration of the first fixation; the number and average duration of looks at the interface (a look is defined as each time the gaze remains on the interface, with one or more fixations before looking away from it); and the average total duration of the gaze on the interface (i.e., the average total time spent looking at the interface each time it appears or the sum of the duration of the looks). The average duration of ocular fixations is considered as a distraction measure. A greater duration of the fixations to the interface would imply a greater time to process the information. However, in this context, the number of fixations or the total time looking at the interface report a better tracking of the interface. In this sense, it should be remembered that the interface is continuously changing with the situation and this fact could involve more frequent fixations but of short duration, even with a proper interface. Due to the small sample size, the results cannot be analyzed as in the previous section, where tests were carried out on a driving simulator. For the data analysis, a Wilcoxon signed-rank test to paired-samples data was carried out to verify whether there are statistically significant differences between the pupil dilations in both eyes during merging with and without the system. This test was considered because it is not possible to determine whether the sample comes from a population with a normal distribution. This test could be considered the equivalent of the t-test for paired-samples but operates with ranges instead of means. Some correlations between the variables are also reported using the Pearson (p) correlation coefficient. Measures of Satisfaction, Usefulness, and Usability Results of the mean and standard deviation of the tests carried out under real driving conditions are shown below. The results for the Measures of Satisfaction, Usefulness, and Usability are only descriptively comparable, as shown in Table 3. Mental workload measurements have been evaluated using the RSME and NASA-TLX questionnaires, which are highly used to evaluate the task load in different fields. In Table 4, results of the mean and standard error are shown. Regarding the eye tracking measures, no statistically significant differences were found between pupil size during incorporations with or without system function, neither for the pupil of the left nor right eye (Z = −0.594, p = 0.552 and Z = −0.664 p = 0.507, respectively). These results support the hypothesis that in both situations the mental load is the same, probably due to the effort required by the merging situation. These results seem to indicate that the use of the assistance system does not imply an over-exertion for drivers (Table 5). Correlations have been made between all the measurements obtained during the real driving tests, finding statistically significant correlations for some of them. The variables analyzed are shown in Table 6 and the results obtained using the Pearson's correlation coefficient (ρ) in Table 7: These correlations are interpreted as follows: On the one hand, when a driver is more satisfied with the system, he perceives it as more usable and mentally less demanding. In addition, the more useful the system is evaluated, the greater the number of looks at the interface, which implies greater tracking. However, the worst the assessment of the mental workload involved in the tracking the system, the higher the number of fixations to the interface, probably reflecting a greater need for information processing. Finally, it is interesting to underline the strong association between pupil size and scores on the RSME scale for both measures of mental load, one being a physiological measure and the other a subjective measure, making the latter a valid tool to assess mental workload in real traffic situations. Discussion In the system acceptance measures, it can be observed that the average values obtained in the three scales are over half of the score, highlighting more usability and slightly less satisfaction and usefulness. Regarding the measures of mental load, in the same way as in the measures of acceptance, the analysis should be only descriptive. The values obtained in both questionnaires are below half of the score, the result of the RSME being considerably lower than that of NASA TLX. These results indicate that users rated the interface performance as effortless, an important feature when implementing a new technology in a vehicle. The measures obtained from the visual tracking system show that the driver assistance system for merging situations does not distract the driver [44]. The number of looks at the interface for each merging maneuver (M = 8.56; SD = 3.22) is considered low-their average duration was 492.22 milliseconds (SD = 181.62)-which is a favorable result for the desirable characteristics defined by [45]. The total time looking at the system in each merging maneuver was also short, only 4.07 s (SD = 2.31), so the results are considered satisfactory. Conclusions Merging maneuvers are quite safe critical and assistance systems that could help the driver to perform them in a safer and more satisfactory way. This paper presents the definition of the system from a technological point of view, but it is mainly focused on the proper HMI design in order to enhance acceptance and effectiveness, performing a proof of concept of the driver's behavior when using it. The merging assistance system presented in this paper is based on communications between vehicles and its interface shows the driver, in a passive way, the speed recommendations to perform the maneuver in the safest possible way. The intentions of the main road driver are clarified thanks to the decision algorithm based on the position and speed variables of both vehicles. In this design, certain premises were followed, such as the recommendations of the Commission of the European Communities concerning safe and efficient in-vehicle information and communication systems [44], the desirable characteristics of in-vehicle devices according to [45], and the considerations mentioned in [46]. The system was validated not only in a driving simulator but also in a real driving environment, in which vehicles exchange information using V2V communication modules. Thanks to an eye-tracking system and several system acceptability surveys performed by drivers, it was possible to select a driver assistance interface for merging situations. The proposed method has proved its value in distinguishing the best interfaces without any default consideration and combining several criteria, such as satisfaction, usefulness, usability, and mental effort. The results from the driving simulator and real vehicles show that the selected interface is well accepted by the users. Comparisons of the final system with other similar studies are not possible without replicating those developments as the test environment and conditions are different (for example, driving simulator tests and real driving tests are not directly comparable). The results presented in this paper try to show the whole design process with the drivers' assessment at each stage, and this fact is the main difference with previous assistance systems for the same problem and the pillar for obtaining such good results in usability and low mental load. Finally, a smartphone was selected for the HMI device because it provides flexibility in the design and integration process, but future integration of the interface is intended to be within the vehicle controls or integrated in the structure of the vehicle, not as a separate device, to improve the visual conditions of the driver position. This new configuration would provide even better results because mobile devices may distract drivers more easily.	v3-fos-license
2016-05-12T22:15:10.714Z	2015-02-01T00:00:00.000	17816213	{ "extfieldsofstudy": [ "Medicine", "Biology" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/cddis2014585.pdf", "pdf_hash": "f4fbb8e7b47d8ca2674b36d7f61f2428e4c28ee3", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45314", "s2fieldsofstudy": [ "Biology" ], "sha1": "f4fbb8e7b47d8ca2674b36d7f61f2428e4c28ee3", "year": 2015 }	pes2o/s2orc	F-box protein Fbxl18 mediates polyubiquitylation and proteasomal degradation of the pro-apoptotic SCF subunit Fbxl7 Fbxl7, a subunit of the SCF (Skp-Cul1-F-box protein) complex induces mitotic arrest in cells; however, molecular factors that control its cellular abundance remain largely unknown. Here, we identified that an orphan F-box protein, Fbxl18, targets Fbxl7 for its polyubiquitylation and proteasomal degradation. Lys 109 within Fbxl7 is an essential acceptor site for ubiquitin conjugation by Fbxl18. An FQ motif within Fbxl7 serves as a molecular recognition site for Fbxl18 interaction. Ectopically expressed Fbxl7 induces apoptosis in Hela cells, an effect profoundly accentuated after cellular depletion of Fbxl18 protein or expression of Fbxl7 plasmids encoding mutations at either Lys 109 or within the FQ motif. Ectopic expression of Fbxl18 plasmid-limited apoptosis caused by overexpressed Fbxl7 plasmid. Thus, Fbxl18 regulates apoptosis by mediating ubiquitin-dependent proteasomal degradation of the pro-apoptotic protein Fbxl7 that may impact cellular processes involved in cell cycle progression. Y Liu 1 , T Lear 1 , Y Zhao 1 , J Zhao 1 , C Zou 1 , BB Chen 1 and RK Mallampalli* , 1,2 Fbxl7, a subunit of the SCF (Skp-Cul1-F-box protein) complex induces mitotic arrest in cells; however, molecular factors that control its cellular abundance remain largely unknown. Here, we identified that an orphan F-box protein, Fbxl18, targets Fbxl7 for its polyubiquitylation and proteasomal degradation. Lys 109 within Fbxl7 is an essential acceptor site for ubiquitin conjugation by Fbxl18. An FQ motif within Fbxl7 serves as a molecular recognition site for Fbxl18 interaction. Ectopically expressed Fbxl7 induces apoptosis in Hela cells, an effect profoundly accentuated after cellular depletion of Fbxl18 protein or expression of Fbxl7 plasmids encoding mutations at either Lys 109 or within the FQ motif. Ectopic expression of Fbxl18 plasmid-limited apoptosis caused by overexpressed Fbxl7 plasmid. Thus, Fbxl18 regulates apoptosis by mediating ubiquitindependent proteasomal degradation of the pro-apoptotic protein Fbxl7 that may impact cellular processes involved in cell cycle progression. It is estimated 480% proteins undergo ubiquitin-dependent degradation either by the proteasome or lysosome. 1 The ubiquitylation of a target protein occurs in a carefully orchestrated manner by an enzymatic cascade involving an E1 ubiquitin-activating enzyme, an E2-conjugating enzyme and an E3-ubiquitin ligase. 2,3 Among hundreds of E3 ligases, the SCF superfamily has emerged as a class of E3 proteins that modulates diverse biological processes. 4 The SCF complex contains a catalytic core including Skp1, Cul1 and an F-box protein. [5][6][7] The carboxyl terminus of Cul1 recruits the small RING protein Rbx1, which bridges the E2 enzyme to the E3 ligase, and the amino terminus of Cul1 binds to Skp1 and a variable F-box protein as a substrate receptor. 8 Based on the pivotal roles of SCF E3 ligases in cellular function through the elimination of substrates, SCF components have recently attracted significant attention on cancer therapeutic drug development and potential to modulate inflammation and innate immunity. 9,12 F-box proteins contain two domains: an NH 2 -terminal F-box motif and a carboxyl terminal leucine-rich repeat (LRR) motif or tryptophan-aspartic acid (WD) repeat motif, which are identified within two subclasses: Fbxl or Fbxw proteins, respectively 13 and. 14 The third subclass of F-box proteins are recognized as Fbxo (F-box only) proteins contain no identifiable LRR or WD motifs. The F-box motif binds to Skp1, whereas the LRR/WD motif is a receptor-like motif that recognizes diverse substrates for ubiquitin ligation and subsequent degradation. 15 With the increasing characterization of F-box proteins and its cognate substrates, this has led to an improved understanding of how SCF complexes have multiple roles in cell cycle progression, gene transcription, circadian oscillation and inflammation. 16 However, with the exception of a few SCF components, the substrates for the majority of~70 F-box proteins remain largely unknown. In addition the mechanisms that regulate the abundance of the majority of F-box proteins remain unclear and this represents an area of active investigation. It is known that some components within the SCF apparatus are targeted by the ubiquitin ligase anaphase-promoting complex or by autoubiquitylation within their own SCF complex. 17,18 Recent studies also show that one F-box protein may target another F-box protein for ubiquitylation to mediate protein degradation thereby impairing the SCF activity. 12 We have previously characterized the role of a relatively new F-box protein, Fbxl7, on cell cycle progression. Fbxl7 within the SCF complex targets Aurora A for polyubiquitylation and proteasomal degradation leading to mitotic cell cycle arrest. 19 Additionally, Fbxl7 was reported to be associated with an ovarian cancer gene and its expression is linked to steroid responsiveness in human subjects with asthma. 20,21 Although these data suggest a role for Fbxl7 in controlling cell proliferative activity and viability, further studies investigating both its biological role and molecular expression are needed. Here we identified that Fbxl7 is itself ubiquitylated and degraded via the proteasome. Interestingly, we identified that an orphan F-box protein, Fbxl18, targets Fbxl7 for its ubiquitylation using distinct molecular signatures that leads to its disposal in cells. Ectopically expressed Fbxl7 displayed pro-apoptotic activity in Hela cells, an effect abrogated by co-expression of Fbxl18. These observations suggest a molecular interplay between two SCF components that govern availability of a biologically important pro-apoptotic protein. Results Fbxl7 undergoes proteasomal degradation. High levels of Fbxl7 in cells appear to critically regulate mitotic arrest; however, the lifespan of this F-box protein in cells is unknown. We first measured the protein stability of endogenous Fbxl7 in MLE12 cells. As shown in Figures 1a and b, cells treated with the protein synthesis inhibitor cycloheximide (CHX) indicated that the half-life of Fbxl7 is~1 h. To identify the degradation pathway involved in Fbxl7 decay, MLE12 cells were also treated with the proteasome inhibitor (MG132) or a lysosome inhibitor (leupeptin) prior to CHX treatment. MG132 treatment, instead of leupeptin, resulted in accumulation of immunoreactive Fbxl7 in cells, which suggests that the proteasome, rather than the lysosome is involved in degradation of Fbxl7. Fbxl7 is polyubiquitylated. To test if Fbxl7 is modified by ubiquitylation, we examined steady-state levels of Fbxl7 protein after overexpression of cells with a plasmid encoding ubiquitin. The levels of endogenous Fbxl7 decreased after expression of increasing amounts of ectopically expressed HA-tagged ubiquitin (Figure 1c). In co-immunoprecipitation (co-i.p.) studies, we also detected an interaction between endogenous Fbxl7 and ubiquitin (Figure 1d). We further confirmed the binding of Fbxl7 to ubiquitin after expression in cells with Fbxl7-V5 and HA-ubiquitin (HA-Ub) plasmids under MG132 treatment (Figure 1e). Collectively, these results indicate that Fbxl7 degradation is mediated by the ubiquitinproteasome system. Fbxl18 targets Fbxl7 for its ubiquitylation and degradation. Recent studies indicate that some F-box proteins target others for their ubiquitylation and degradation. 12 Thus, we investigated the interaction between Fbxl7 and randomly chosen F-box proteins by co-immunoprecipitation. Fbxl7 specifically bound to Fbxl18 (Supplementary Figure 1A). We further screened ability of ectopically expressed plasmids encoding~50 F-box proteins from L, W and O families that might mediate Fbxl7 degradation. As shown in Figure 2a, randomly chosen V5-tagged F-box proteins in L family were ectopically expressed in MLE12 cells. Interestingly, only Fbxl18 overexpression decreased endogenous Fbxl7 protein levels. Additionally, we observed that overexpressed F-box proteins from W and O families displayed no effect on endogenous Fbxl7 protein abundance, as shown in Supplementary Figures 1B and C, respectively. We further observed that overexpression of Fbxl18 plasmid decreased endogenous Fbxl7 protein levels in a dose-dependent manner ( Figure 2b). However, transfection of cells with increasing amounts of Fbxl16 plasmid had no effect on endogenous Fbxl7 levels. Particularly, MG132 treatment abolished the Fbxl7 decrease induced by overexpression of Fbxl18 ( Figure 2c). In co-i.p. studies we detected the association of endogenous Fbxl18 with Fbxl7 ( Figure 2d) in Hela cells. We further confirmed that Fbxl18 mediated the polyubiquitylation of Fbxl7 using in vitro ubiquitylation assays, and we also excluded the possibility that Fbxl7 might display autoubiquitylation activity ( Figure 2e). Additionally, we investigated Fbxl18-mediated polyubiquitylation using a neddylation inhibitor MLN4924 (Figure 2f). Neddylation-deficient Cul1-Rbx1 largely decreased the polyubiquitylation of Fbxl7, in support that Fbxl7 polyubiquitylation is dependent on a (Figure 2g). In separate studies knockdown of Fbxl18 with shRNA-stabilized Fbxl7 protein levels in a time-dependent manner (Figures 2h and i). These results indicate that Fbxl18 mediates Fbxl7 polyubiquitylation and proteasomal degradation. Lys 109 is an Fbxl7 ubiquitin acceptor site. To localize the ubiquitin receptor site within Fbxl7, we ectopically coexpressed individual Fbxl7 plasmids encoding lysine point mutations with an Fbxl18 plasmid in MLE12 cells. Mutants whose protein levels decreased with overexpressed Fbxl18 were excluded from further analysis (data not shown). We then assessed protein stability of the remaining mutants by exposure to CHX treatment after Fbxl7 cellular expression. We identified that only transfection of cells with a Lys 109E mutant stabilized the decay of Fbxl7, compared to the wildtype or other lysine mutants as shown in Figure 3a. Further, in MLE12 cells co-transfected with Fbxl18 and Fbxl7 plasmids, overexpressed Fbxl18 plasmid dose-dependently decreased wild-type Fbxl7 levels, but had no effect on Fbxl7 K109E protein levels ( Figure 3b). To exclude an amino acid charge effect on protein stability, we further compared the protein turnover rate after expression in cells with plasmids encoding additional Lys 109 substitutions. As indicated in Figures 3c and d, expression in cells of plasmids encoding three distinct Lys 109 point mutants resulted in stabilization of Fbxl7 proteins levels compared with cellular expression of wild-type Fbxl7 plasmid. Here, the addition of a tag to the protein extended the lifespan of wild-type Fbxl7 versus endogenous Fbxl7 (Figure 1). In the in vitro ubiquitylation assays, the expressed Lys 109 point mutants displayed resistance to SCF Fbxl18 -induced polyubiquitylation versus wild-type Fbxl7 (Figure 3e). We also compared the polyubiquitylation levels between wild-type Fbxl7 and the K109R mutant through ubiquitin pull downs in MLE12 cells. The lysine 109 mutation within Fbxl7 dramatically decreased Fbxl7 polyubiquitylation (Figure 3f). These results demonstrate that the ubiquitin acceptor site Lys 109 is essential for Fbxl7 decay through ubiquitin-proteasomal processing. for an additional 2 h before boiling in PBS containing 2% SDS. The lysates were next diluted in TBS to a final 0.2% SDS concentration before pull down using a rabbit ubiquitin antibody. The co-i.p. products were processed for immunoblotting with mouse V5 antibody and mouse ubiquitin antibodies, respectively Fbxl18 regulates Fbxl7-induced apoptosis Y Liu et al Fbxl7, we constructed a series of NH 2 -terminal and C-terminal truncated Fbxl7 expression plasmids, as schemed in Figure 4a. In Hela cells, endogenous Fbxl18 interacted with all the Fbxl7 C-terminal deletion mutants. However, loss of the NH 2 -terminal region containing the F-box domain largely abolished the interaction between Fbxl7 and Fbxl18 (Figure 4b). We then focused on the Fbxl7 NH 2 -terminal region and further localized the Fbxl18 interacting region within Fbxl7 via co-immunoprecipitation (Figures 4c and e). Figures 4d and e demonstrate that an amino-terminal region spanning residues 63-69 harbors a putative Fbxl18 binding site within Fbxl7. Indeed, by generating and expressing plasmids encoding specific Fbxl7 point mutants in cells followed by co.ip., we observed that an FQ motif within Fbxl7 was essential for Fbxl18 interaction with Fbxl7 (Figure 4e). To further confirm that the FQ motif is the interaction site between Fbxl18 and Fbxl7, we employed a biotin-labeled peptide binding assay and detected in vitro synthesized V5-tagged Fbxl18 bound to wild-type Fbxl7 peptides. However, a Fbxl7 FQ-AA mutant largely exhibited a decrease in its association with Fbxl18 ( Figure 4f). We also compared Fbxl7 protein stability after expression in cells of a plasmid encoding either wild-type or an FQ mutant. As shown in Figure 4g, cellular expression of the FQ mutant plasmid resulted in an Fbxl7 variant that exhibited extended lifespan in cells versus the wild-type Fbxl7 protein. Moreover, similiar to wild-type Fbxl7, this FQ mutant displayed ability to retain association with other components within the SCF complex ( Figure 4h) and to mediate degradation of a known substrate, Aurora A, for proteasomal degradation (Figure 4i). These results strongly suggest that an Fbxl7 FQ motif functions as the molecular recognition site for Fbxl18 binding. The binding capacity of Fbxl7 through this motif does not impact its effect on substrate degradation. 19 Here, we observed that overexpressed Fbxl7 induces apoptosis. The expression of Fbxl7 plasmid in cells led to appearance of a larger percentage of apoptotic cells in both a dose and timedependent manner, as determined by propidium iodine staining compared with the cells transfected with a control vector (Figures 5a and b). These data indicate that Fbxl7 is pro-apoptotic. Fbxl18 counteracts Fbxl7-induced apoptosis. As shown above, overexpression of Fbxl7 plasmid causes apoptosis. We next investigated if Fbxl18 differentially modulates Fbxl7induced apoptosis. Cellular depletion of Fbxl18 with three different Fbxl18 shRNAs caused increased apoptosis, compared with effects of scrambled shRNA-treated cells (Figure 6a). However, the apoptosis induced by knockdown Fbxl18 was largely rescued by the additional knockdown of Fbxl7 ( Figure 6b). Further, as shown in Figure 6c, double transfection of the Fbxl7 FQ mutant and Fbxl18 plasmids showed a impaired cellular proliferation, compared with the cells transfected with Fbxl7 wild-type and Fbxl18 plasmids or a control vector. These data suggest that the Fbxl7 docking site mutant protein that lacks ability to interact with Fbxl18 harbors an extended lifespan in cells to induce apoptosis. Further, Fbxl18 counteracts the pro-apoptotic activity of Fbxl7 in a biological model of cell injury. Discussion Fbxl7 is a highly evolutionally conserved protein sharing a 98% sequence identity between mouse and humans, and 495% identity among other species. However, Fbxl7 is a relatively uncharacterized protein with regard to its biological role and molecular regulation in cells. Whether Fbxl7 is an oncoprotein or a component within a complex that triggers mitotic arrest leading to cell death requires further study. Our data demonstrate potent activity of this SCF component as it is sufficient to elicit apoptosis in some cells. This effect is relevant as high-level Fbxl7 protein expression in cells, left unchecked, could promote sustained tissue injury through activation of apoptotic programs. This is important given that cell proliferative activity must overcome the activities of proapoptotic factors to ensure optimal healing during tissue injury. Here, we demonstrate that Fbxl7 protein levels are controlled by the actions of the SCF Fbxl18 complex, perhaps serving as a feedback control mechanism to limit high-level expression of this pro-apoptotic factor. We have demonstrated that Fbxl7 harbors a molecular signature that recruits Fbxl18, identified the acceptor site for Fbxl18 activity within its substrate, and present data suggesting that levels of Fbxl18 affect the ability of cells to remain viable from Fbxl7-induced apoptosis. The data are also the first to characterize the molecular behavior of Fbxl18 in cells and raise the possibility for molecular targeting of an Fbxl18-\|Fbxl7 → apoptosis pathway. Fbxl7 is a relatively short-lived protein (t 1/2~1 h) through SCF Fbxl18 -dependent polyubiquitylation and subsequent proteasomal degradation. However, 1 h after cycloheximide treatment, Fbxl7 protein levels remains stable in cells (Figures 1a and b). One possibility could be that Fbxl7 interacts with ubiquitin-specific peptidases. For example, Fbxl7 directly associates to USP1 and USP45, and indirectly is linked to USP12. 22 These deubiquitylation enzymes could counteract SCF Fbxl18 to remove ubiquitin from Fbxl7 to prevent its degradation. More importantly, the Fbxl7 steady-state levels remain detectable with some degree of stabilization following its initial rapid decay that appears to be caused by the accelerated degradation of Fbxl18 based on our observations (data not shown). Although the half-life of endogenous Fbxl7 is short, the addition of a C-terminal V5 tag extended its half-life to~6 h. This might result from a conformation change induced by the tag or that exogenous Fbxl7 exceeded the endogenous SCF Fbxl18 ubiquitylation machinery capacity to impair its function. The underlying mechanism for this disparity is under further investigation. The molecular factors that regulate Fbxl18 levels and potential sensing of Fbxl7 abundance also remain obscure. Fbxl18 may be prone to autoubiquitylation or other modifications and subsequent degradation. Our data suggest that Fbxl18 levels rapidly fluctuate in cells but when elevated, they are linked to stability of Fbxl7 to regulate the apoptotic program. We speculate that there may exist a negative feedback loop driven by Fbxl18 that could be essential to keep a dynamic equilibrium of Fbxl7, which could be critical for execution of vital roles by multiple substrates downstream Fbxl7. The identification of these Fbxl7 substrates in addition to Aurora A also requires further study. The interaction of F-box protein and its substrates must be tightly controlled. In response to a stimuli, F-box proteins must recruit substrates rapidly and specifically to the catalytic core for ubiquitylation and degradation. For example, phosphorylation of the substrate resulting in a phosphodegron is one of the most common recognition signals for F-box proteins, which either accelerates or inhibits the interaction of F-box protein and substrates. In response to a specific stimuli, kinases such as glycogen synthase kinase 3β regulate biological function by manipulating protein stability through phosphorylating substrates. 23,24 Recently, acetylation is also considered as an important post-translational modification to regulate protein stability. 25 Additionally, other interaction factors also have a vital role in mediating F-box protein to recognize their substrates. 26,27 Interestingly, another F-box protein, Fbxl2 that impacts lipogenesis, innate immune responses and cell cycle progression targets its substrates through an IQ motif on its substrates. 28 In this study, we identified that a similar FQ motif within Fbxl7 is the major docking site for recognition by Fbxl18. We cannot completely exclude the possibility that another minor domain within Fbxl7 might also serve as a recognition signal to recruit Fbxl18 because a series of N-terminal deletions do not totally eliminate the interaction between Fbxl18 and Fbxl7 (Figures 4b-f), and C-terminal truncated Fbxl7 mutants showed slightly different binding intensities to Fbxl18. Whether this FQ motif-based interaction mechanism is unique or undergoes any modification responding to stimuli, is currently under further investigation. The results suggest that some members of the F-box family harboring LRR domains may be recruited to their substrates through these signatures. The ability of Fbxl18 to recruit pro-apoptotic Fbxl7 for ubiquitylation and proteasomal degradation may be of fundamental importance in human disease. For example, aside from neoplasia, several disorders including emphysema, traumatic injury and aging are characterized by an imbalance between cell proliferative activities and apoptosis. The ability of Fbxl18 to protect cells from apoptosis induced by Fbxl7 might provide an opportunity to design small molecule Fbxl18 agonists to enhance tissue repair and regeneration, or alternatively design novel therapeutics that impair Fbxl18:Fbxl7 interaction to stimulate apoptosis in tumors. These data are, however, too preliminary to conclude that Fbxl18 is an anti-apoptotic molecule until larger numbers of Fbxl18-substrate pairs are characterized in multiple systems. We are currently investigating the identities of additional SCF Fbxl18 substrates to better interpret its role in apoptosis. Additionally, the Fbxl7 sequence is highly conserved among species. Specifically, Fbxl7 in mammals is homologous to Grr1 in yeast Saccharomyces cerevisiae. Grr1 is among the beststudied F-box proteins in yeast, which has a vital role in cell cycle regulation by recruiting Cln2. 29 Besides Aurora A, we expect to identify additional kinases or cell cycle regulators under regulation of Fbxl7 to further elucidate the pro-apoptotic behavior of Fbxl7. Cloning and mutagenesis. PCR-based approaches were applied to clone different F-box proteins into pcDNA3.1D/v5-his (Invitrogen) for expression in mammalian cells using appropriate primers. All mutants were constructed using PCR-based approaches or site-directed mutagenesis (Agilent). Transfection. All plasmids were delivered into MLE12 cells or BEAS-2B cells using nucleofection or FuGENE HD (Hela cells), following the manufacturers' protocols. Cellular expression of GFP-tagged plasmids using these methods was achieved at 490% efficiency. Immunoblotting and Immunoprecipitation. Cell lysates (normalized to total protein concentration) were subjected to SDS-PAGE, electrotransferred to membranes, and immunoblotted. For ubiquitin pull down, the freshly collected cells in PBS containing 2% SDS were boiled at 100°C for 10 min. The lysates were diluted with TBS to a final concentration with 0.2% SDS containing 5 μM ubiquitin aldehyde. 1 mg total cellular proteins were immunoprecipitated with 2 μg ubiquitin rabbit antibodies at 4°C for 3-4 h followed by the addition of 30 μl of protein A/Gagarose for an additional 1 h at 4°C. The precipitated complex was washed three times with 0.5% Triton X-100 in PBS and analyzed by immunoblotting with an enhanced ECL system. For other immunoprecipitations, 1-mg cell lysates in PBS containing 0.5% Triton X-100 were incubated with 2-μg-specific primary antibodies for 3-4 h at 4°C followed by the addition of protein A/G-agarose and three washes with lysis buffer before immunoblotting. Apoptosis analysis. Transfected cells were incubated with FITC-Annexin V and propidium iodide for 15 min following manufacturer's protocols (Biolegend, San Diego, CA, USA). FACS samples were analyzed with the AccuriC6 system with De Novo Software. For proliferation assays, Hela cells were transfected with indicated plasmids. Cells were cultured in 35 mm dishes for up to 48 h; at indicated time points, cells were collected and stained with trypan blue. Viable cells were then counted and quantified. Statistical analysis. Statistical comparisons were performed using an ANOVA or an unpaired t-test with Po0.05 indicative of significance.	v3-fos-license
2018-05-02T08:49:08.829Z	2018-05-01T00:00:00.000	13756710	{ "extfieldsofstudy": [ "Medicine", "Physics" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/s41598-018-25283-1.pdf", "pdf_hash": "707bf96124dfc490c37214c391182c37ace0f165", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45316", "s2fieldsofstudy": [ "Physics" ], "sha1": "707bf96124dfc490c37214c391182c37ace0f165", "year": 2018 }	pes2o/s2orc	Accurate estimation of a phase diagram from a single STM image We propose a new approach to constructing a phase diagram using the effective Hamiltonian derived only from a single real-space image produced by scanning tunneling microscopy (STM). Currently, there have been two main methods to construct phase diagrams in material science: ab initio calculations and CALPHAD with thermodynamic information obtained by experiments and/or theoretical calculations. Although the two methods have successfully revealed a number of unsettled phase diagrams, their results sometimes contradicted when it is difficult to construct an appropriate Hamiltonian that captures the characteristics of materials, e.g., for a system consisting of multiple-scale objects whose interactions are essential to the system’s characteristics. Meanwhile, the advantage of our approach over existing methods is that it can directly and uniquely determine the effective Hamiltonian without any thermodynamic information. The validity of our approach is demonstrated through an Mg–Zn–Y long-period stacking-ordered structure, which is a challenging system for existing methods, leading to contradictory results. Our result successfully reproduces the ordering tendency seen in STM images that previous theoretical study failed to reproduce and clarifies its previously unknown phase diagram. Thus, our approach can be used to clear up contradictions shown by existing methods. In this study, we present a new approach to constructing a phase diagram by combining our recently proposed theory 16,17 with two-dimensional STM images. Since our approach directly determines underlying many-body interactions only from the geometrical information of any-sized objects at thermodynamic equilibrium within a given accuracy, we can clarify that underlying interactions play an essential role in stabilizing the measured microscopic structure. Thus, when contradictions occur between ab initio calculations and experiments, it is expected that our approach, which does not require any thermodynamic information, can bridge this gap. To demonstrate the validity of our approach, the stacking-fault interface of Mg-Zn-Y alloy, in which L1 2 -type Zn 6 Y 8 clusters are arranged, is a suitable target. We herein show the interface phase diagram of the 18R-type LPSO phases in Mg-Zn-Y alloy, compare our results with previous studies, and successfully reproduce the short-range-order (SRO) of clusters seen in the STM image, which the previous work failed to reproduce quantitatively. Results and Discussion On the right-hand side of Fig. 1, we show a typical STM image of the distribution of Zn-Y clusters in the stacking-fault interface (see Methods for sample preparation). Since the STM is only sensitive to the topmost surface atoms, one can directly determine the individual positions of the clusters 18 . Before the evaluation of the SRO, we corrected the distortion of STM images, which originates from thermal drift. In particular, we calculated the vectors between all clusters and selected those corresponding to the 6th-nearest-neighbor (6NN) distance. STM images were corrected by linear transformation such that 6NN vectors had directions and magnitudes that were as correct as possible. Once we eliminated the effects of thermal drift, we could measure the relative positions between any pairs of clusters. We counted the number of cluster-cluster, cluster-vacancy, and vacancy-vacancy pairs in the field of the STM image up to a distance of 25NN and evaluated SRO. Energy coefficients of mNN pairs in Eq. (2), V m , as determined by SRO from STM images (see Methods) are shown in Fig. 2(a). We also shows a schematic illustration of some mNN pairs at the interface in Fig. 2(b). Since in 18R-type LPSO, the interplane interactions are relatively weak 19 , this method is valid for reproducing the characteristics of an STM image, ignoring interplane interactions and regarding the interface as an isolated two-dimensional system, i.e., all interactions in Fig. 2 are in-plane interactions. In order to confirm validity of V m in Fig. 2(a), we investigated the pair-formation energy of a single mNN (m = 6, 7, 8) cluster-cluster pair with an L × L two-dimensional triangular lattice under a periodic boundary condition. The pair-formation energy of ΔE(m, L) is defined as Mg cluster Here, E Mg denotes the total energy of a configuration filled with Mg, and E cluster denotes total energy of a configuration consisting of a single cluster and Mg atoms on the rest lattice points in a considered cell. The definition of Eq. (1) corresponds with pair interactions used in the previous DFT work 14 and reflects actual mNN cluster-cluster pair interactions. In Fig. 3, ΔE(m, L) is shown for m = 6-8 and L = 5-40. Figure 3 shows that positive values of ΔE(6, 5), ΔE(7, 6) and ΔE(8, 7) imply that 6-8NN cluster-cluster interactions behave repulsively under very small simulation cells, i.e., very dense cluster compositions. This tendency is similar to the result obtained by the previous DFT work 14 , which estimated interactions using a simulation cell with fully arranged clusters. Meanwhile, under dilute cluster compositions where L ≥ 10, ΔE(6, L) becomes strongly negative unlike ΔE(7, L) or ΔE(8, L). Negative value of ΔE(6, 40) in Fig. 3, which means 6NN cluster-cluster pairs are favorable, seems to contradict with the previous DFT result 14 , which held that 6NN cluster-cluster pairs are unfavorable. However, since strong oscillation of ΔE(m, L) in L ≤ 10 means that clustercluster interactions depend on the cluster composition, we actually clarify that the most probable reason for this disagreement is originate from insufficient consideration of dependence on cluster compositions in the previous DFT work 14 . Therefore, the 6NN cluster-cluster interaction behaves attractively for dilute clusters. Although the previous result 14 successfully showed qualitative landscapes of the radial distribution function of nanoclusters, it could not reproduce the quantitative microscopic ordering tendency seen in STM images (solid Fig. 1). This is because the previous DFT work determined cluster-cluster interactions by dense-cluster simulation cells, whereas we found that interactions behave differently between dense and dilute cluster concentrations, as shown in Fig. 3. Now, let us confirm that our multiscale interactions in Fig. 2(a) can capture the characteristics of SRO of clusters at the interface. Figure 4 shows thermodynamically averaged SRO at equilibrium, as obtained by Monte Carlo simulation (see Methods for a more detailed procedure), and this SRO corresponds to the radial distribution function of clusters. The strong SRO of 6NN pair means that the number of Mg-Mg and cluster-cluster pairs are relatively larger than that of Mg-cluster pairs, i.e., clusters are arranged so as to be at each other's 6NN position. Our SRO landscape agrees well with the radial distribution functions obtained by previous DFT and experimental work, especially for the peaks in 6, 9, and 15NN pairs. For further discussion, we took a snapshot at the interface in the simulation and compared it with an STM image in Fig. 1. From the STM image, we can clearly confirm chain-like cluster ordering, unlike the previous DFT work, which showed a uniform arrangement of clusters. As above, our snapshot quantitatively captures the features of STM images and supports the validity of our multiscale interactions. The Mg 1−x (Zn 6 Y 8 ) x interface phase diagram obtained by different three STM images is presented in Fig. 5(a), which shows the ordering tendency in the cluster-rich phase (Fig. 5(b)) through a first-order order-disorder phase transition. Since, the interplane interactions are relatively weak in 18R-type LPSO 19 , it is valid to regard the interface of LPSO as an isolated two-dimensional system. As a result, the closed circle in Fig. 5(a), showing the point where the STM image in Fig. 1(b) is obtained, indicates that Fig. 1(b) is not a single phase but a order-disorder phase coexistence state. In summary, we suggested a new approach based on microscopic geometric information for constructing phase diagrams in cases where this is difficult for conventional simulation methods. Using STM images, we demonstrated our approach through the interface of 18R-type Mg-Y-Zn LPSO. We clarified the contradictory information concerning cluster arrangements presented by experiments and simulations and successfully reproduced an interface phase diagram consistent with that obtained by experiment. Our approach is expected to become a powerful tool for modeling isolated systems subjected to experimental observations such as STM. Methods Sample preparation for STM images. The directional solidification process was applied to a Mg 85 Zn 6 Y 9 ingot. The ingot was annealed at 773 K for 168 h and quenched with water. We confirmed the presence of a diffraction spot corresponding to 18H-type LPSO by TEM observation. Samples for STM observation were prepared by cutting the alloy ingot, typically into a reed shape of 8 mm × 5 mm × 0.5 mm. After the introduction of the sample into the preparation chamber of the low-temperature ultrahigh-vacuum STM (Unisoku 1200), we cooled the sample with a liquid-nitrogen flow for 10-15 min. Finally, we cleaved the sample using a pushing rod in the UHV chamber. The cleaved sample was immediately transferred into a pre-cooled observation chamber. All STM observations were carried out at liquid-nitrogen temperature (up to 77 K). STM itself has enough spatial resolution to resolve each atomic position on the surface. However, due to some noise and thermal drift, the positions of the clusters determined by our method have some uncertainty. To demonstrate the accuracy of our method, we plotted relative positions of the clusters measured from all the clusters in a single STM image in Fig. 6. As clearly seen, the relative positions of the clusters construct a lattice; which indicate that the accuracy of positions is good enough to estimate SRO. Generalized Ising model. In recent alloy studies with ab initio calculations, the generalized Ising model 20 (GIM) has often been used for constructing a coarse-grained Hamiltonian, which captures the characteristics of a system within a given accuracy. Since we are interested in how clusters would be arranged at the interface with Mg solvent, for simplicity, it is natural to consider the clusters as coarse-grained atoms, i.e., Mg and the center of cluster as +1 and −1 Ising-spin variables, respectively. In GIM, the total energy of a given configuration, σ, is expanded by a set of complete and orthogonal basis functions, { } k ξ , with indices where V k denotes an energy coefficient and k is an index specified as a symmetrically nonequivalent figure such as 1NN and 2NN pairs. In particular for a binary system, ξ k corresponds to the average product of spin variables s α on k over all symmetrically equivalent k in an Ising-spin configuration where α denotes a site on a given lattice and N k denotes the number of k in the configuration. V k is defined as the inner product between total energy and ξ k Here 〈\|〉 denotes inner product for configuration space, ρ k 0 denotes a normalized constant for the inner product, and the term in 〈〉 denotes over all possible configurations except for pair k (i.e., σ N−k ). Transformation from the second to the third line means that summation is taken in terms of possible configurations on pair k (i.e., σ k ) instead of a summation over all possible configurations of σ. This transformation explicitly shows us that when a given system includes many-body interactions should include the many-body interactions except for that of k. Therefore, even though in this study pair-wise ks are only considered, our method correctly treats essential many-body interactions. Short-range-order in our simulation. We calculated the SROs of clusters using the Metropolis algorithm with a 100 × 100 two-dimensional triangular lattice under fixed composition. Note that, to explicitly consider the size of the nano-sized clusters without loss of validity, interactions of 1-3NN pairs are configured as having a very high so as not to overlap with each other. Determination of many-body interactions from a single STM image. We first assume that the measured structure obtained by STM is in thermodynamic equilibrium. Then, the expectation value of the kth coordination of the structure at temperature T, ξ k (T), can be given in the framework of classical statistical mechanics as where summation is taken over all possible configurations σ, and Z denotes a partition function. Under these conditions, we have recently derived the relationship 16,17 between the structure, ik ik ave B where Q ave represents a linearly averaged configuration and S ik denotes an element of the covariance matrix for the configurational density of states in ξ ξ ( , ) i k two-dimensional space for a non-interacting system. Both of these quantities can therefore be known a priori without any information about energy or temperature. From Eq. (6), we can thus directly determine the energy coefficients defined in Eq. (4) from a measured structure using the equation In this study, the interface is regarded as a Mg-cluster pseudo-binary alloy. To construct the alloy phase diagram, a semi-grand-canonical (SGC) ensemble is more often used than grand-canonical or canonical ensemble. In an SGC ensemble, compositions can vary under the number of atoms, N, kept fixed, i.e., we handle the difference of chemical potentials for each constituent element, µ µ µ = − cluster M g in spite of handling compositions. The total energy, E SGC , is defined as E E Nx SGC µ = − , and the free energy, φ, is defined as φ = −k T Y ln B , with Boltzmann constant k B and the partition function in the SGC ensemble Y defined as: where W denotes density of states (DOS). Note that Helmholtz free energy F in the canonical ensemble is related to the free energy φ in the SGC ensemble through Legendre transformation: We can obtain x using partial differentiation by interpolating φ for each chemical potential:	v3-fos-license
2018-04-03T02:25:19.370Z	2016-05-26T00:00:00.000	15881025	{ "extfieldsofstudy": [ "Chemistry", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0156529&type=printable", "pdf_hash": "b5bbb87a3db2e1065868bff873ceac88320980bf", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45317", "s2fieldsofstudy": [ "Biology" ], "sha1": "b5bbb87a3db2e1065868bff873ceac88320980bf", "year": 2016 }	pes2o/s2orc	Construction of Large-Volume Tissue Mimics with 3D Functional Vascular Networks We used indirect stereolithography (SL) to form inner-layered fluidic networks in a porous scaffold by introducing a hydrogel barrier on the luminal surface, then seeded the networks separately with human umbilical vein endothelial cells and human lung fibroblasts to form a tissue mimic containing vascular networks. The artificial vascular networks provided channels for oxygen transport, thus reducing the hypoxic volume and preventing cell death. The endothelium of the vascular networks significantly retarded the occlusion of channels during whole-blood circulation. The tissue mimics have the potential to be used as an in vitro platform to examine the physiologic and pathologic phenomena through vascular architecture. Introduction Tissue engineering has led to in vitro construction of tissues/ organs and may ameliorate the limited supply of organs for transplantation. Currently only a few types of organs such as skin [1,2], bladders [3,4], and tracheas [5,6] have been engineered for clinical application. Unlike living tissues in vivo, cell viability and functions cannot be sustained in the core of dense engineered tissue in vitro [7][8][9] because diffusion alone supplies nutrients and oxygen to cells within engineered tissue; the lack of adequate mass transport leads to necrotic cell death in the core. The limitation of diffusion becomes increasingly critical as the volume and cell population of engineered tissue increase. Thus bioengineers are seeking ways to incorporate microcirculation into engineered tissues [10][11][12][13][14][15][16]; methods include induction of angiogenesis by biomolecular cues, and formation of microvasculature. The introduction of microfabrication technologies to tissue engineering has allowed construction of perfusable channels in engineered tissues [13,[17][18][19]; this approach has the potential to achieve stable mass transport from the initiation of the cell culture. To construct functional channels for this purpose, emphasis should be placed on developing endothelized channels in an engineered tissue for metabolite transport in vivo. Soft lithography, micromachining, and micromolding technologies have been utilized to fabricate two-dimensional (2D) endothelized channels with biomaterials [13,17,18]. 3D printing of carbohydrate fibers and rapid casting have been used to fabricate endothelized lattice channels in 3D hydrogel in the present of living cells [19]. To realize the complex branching patterns of vascular networks, 3D printing technology based on layer-by layer deposition has enabled the construction of 3D vascular networks with various branching patterns and angles [20,21]. However, cells should be repopulated after completing the structure fabrication because the process entails use of cytotoxic organic solvents. Thus, challenges remain in controlling the position of repopulated cells to form tissues with spatially distinct endothelium on the lumen of the porous monolithic structure fabricated by the technology. The objective of this study was to use stereolithography (SL) based on photopolymerization [22] to construct 3D functional vascular networks in tissue mimics, and to establish a basis for large-volume tissue regeneration. Although SL technology requires use of photocurable materials, a wide range of biomaterials can be used with indirect method [23]. The work led to a welldesigned molding process and successful fabrication of scaffolds using PLGA, PLLA, PCL, chitosan, alginate, and bone cement with the resolution of 50~70 um. Here, we modified indirect SL technology to enable formation of an inner-layered fluidic network in a porous scaffold by introducing a hydrogel barrier on the luminal surface. The scaffold containing inner-layered fluidic networks was turned into a tissue mimic containing vascular networks by seeding it separately with human umbilical vein endothelial cells (HUVECs) and human lung fibroblasts (HLFs). The effects of artificial vascular networks on oxygen delivery and cell viability were assessed under perfusion culture, and the early performance of vascular networks beyond 24 h in a physiological environment was investigated using whole-blood perfusion as a surrogate for transplantation. Design of a microfluidic network system for a porous scaffold We designed 3D fluidic network models for a cylindrical scaffold (10 mm in diameter × 10 mm in length) based on an algorithm introduced previously [8]. At bifurcation lesions, the parent branch and its two daughter branches lay in the same plane and the branch opening half-angles did not exceed p 4 , which is the maximum value in the physiologically relevant range. The inlet diameter was set to be 2 mm and the daughter diameters were determined using Murray's law [24]. The same numerical analysis was performed to evaluate oxygen transport by 3D fluidic networks in the 3D large-volume scaffold. Fabrication of Scaffolds with Inner-layered Fluidic Network Indirect-SL fabrication [8,23] was proposed to construct dual-pore scaffolds having designed pores and local pores together by combining SL technology and sacrificial molding process. A sacrificial mold having an inverse shape of global pores is fabricated from an alkali-soluble photopolymer using the SL technology. Then the mold is filled with biomaterials and local pores are formed by traditional methods such as phase inversion and salt leaching techniques. Finally, the sacrificial mold is removed and designed global pores and irregular local pores are formed within a structure. In this study, the indirect-SL fabrication technique was modified to construct fluidic networks having a collagen inner layer in a porous scaffold (Fig 1). As described in detail previously, first, projection image data were created from computer-aided design models of the designed fluidic networks [25]. Second, the shape of the fluidic networks was fabricated from an alkali-soluble photopolymer (44 wt% N,N-dimethyl-acrylamide, 44 wt % methacrylic acid, 12 wt% poly(vinyl pyrrolidone)) [26] using a projection-based microstereolithography (pMSTL) system. Third, the fabricated structure was dipped in a collagen solution (2 v/v%, Koken atelocollagen implant, Koken, Japan) and the collagen was cross-linked in N-hydroxysuccinimide(NHS)/N-(3-dimethylaminopropyl)-N'-ethylcarbodiimide (EDC) solution in ethanol (NHS/EDC at 1:1 each 10 mg/ml in 95% ethanol). Fourth, polycaprolactone (PCL; Polysciences, Inc., Warrington, PA, USA; Mw 43,000-50,000, 20% w/v) solution in chloroform was mixed with sodium chloride particles which were sieved through a 300-μm mesh, and the mixture was poured into the mold and placed in pure isopropyl alcohol to remove the Large-Volume Tissue Mimics with 3D Vascular Networks chloroform. Then the sacrificial mold and salt were dissolved using 0.5 N NaOH for 8 hours. The dissolution time and the effect of residues of sacrificial mold were investigated in our previous work [23] and this protocol was established based on the results. Because the PCL surface is hydrophobic and lacks active sites to immobilize collagen molecules, the surface of the collagen layer and porous scaffold are in contact without any interfacial anchoring. Hydrolysis of PCL by NaOH during the step of removing the sacrificial mold can break the ester bonds in PCL; this process introduces active sites on the surface of PCL by producing carboxylic groups and hydroxyl groups [27,28]. After removing the mold in NaOH solution, the structure was immersed in EDC/NHS solution again to induce amide bonds between the PCL scaffold and the collagen layer. The final PCL structure was carefully washed with distilled water three times for total 24 h. To observe the surface, scaffolds were sputter-coated with gold and images were taken with scanning electron microscopy (SEM; Hitachi SU-6600; Hitachi, Tokyo, Japan). To observe the inner architecture, scaffolds were cross-sectioned with a cryostat (Leica, Wetzlar, Germany) and images of bifurcation lesions were taken using micro-CT. Assessment of Collagen Denature by Fourier transform infrared (FT-IR) spectroscopy In the fabrication process (Fig 1), collagen is exposed in several solvents such as isopropyl alcohol, NaOH, and chloroform. To determine whether the collagen was denatured by the solvents, infrared spectra were obtained for the raw and processed collagen. The raw collagen film was prepared after crosslinking in EDC/NHS solution. The processed collagen film was prepared by immersing the raw collagen in isopropyl alcohol for 2 d, NaOH solution for 8 h, and chloroform for 2 d one after another. Spectra were obtained using a Fourier transform infrared spectrometer (M1200, MIDAC Co. Ltd). Construction of 3D Artificial Tissue Mimics The porous scaffold containing inner-layered fluidic networks was turned into a tissue mimic containing vascular networks by seeding it with HUVECs and HLFs cells in two steps. HUVECs and HLFs were purchased from ATCC and labeled with DiI (red) and DiO (green) dyes, respectively, in accordance with the manufacturer's (Molecular probes, Invitrogen) instructions. Briefly, cells were dissociated by treatment with trypsin, then resuspended in 1 ml medium. The cell suspensions were mixed with 5 μl of DiI or DiO cell labeling solutions, then incubated at 37°C for 30 min. First, a suspension of HUVECs was injected into the inlet of the fluidic network until the suspension filled the whole network. After shaking and incubating for 3 h at 37°C and 5% CO 2 , HUVECs were attached to the luminal surface of the fluidic network. Second, HLFs were seeded onto the scaffolds surrounding the fluidic network by using a vacuum-aided seeding technique [29] to homogeneously distribute the cells across the surface and thickness of the scaffold. Then the scaffolds were loaded in perfusion chambers (Fig 2a) and cultured for 6 days in the perfusion system (Fig 2b). The perfusion system composed of a peristaltic pump, medium reservoir, gas permeable silicon tube as a gas exchanger, and chamber. The PET chamber was designed to have the same diameter with scaffolds and plugged with silicon rubbers which were connected to tubes for perfusion. The pimonidazole conjugation assay was used to detect hypoxic cells in the tissue mimics. Pimonidazole is reductively activated in hypoxic cells and forms stable covalent adducts with thiol (sulphydryl) groups in proteins, peptides and amino acids. According to the manufacturer's protocol, the cells having those adduct are detected by immunochemical means. First, a pimonidazole was added to the cell culture media at day 3 and incubated for 36 h. The tissue mimics were fixed for 30 min in 10% formalin, frozen, and embedded in O.C.T. compound embedding medium (Tissue-Tek, Sakura Fineteck Inc, Torrance, CA, USA). Then, 4-μm sections were prepared from the frozen block with a cryostat. In the process, the collagen layer was somewhat detached from the scaffold surface but maintained the shape of fluidic network. Samples were mounted with DAPI and images were taken under a fluorescence microscope (LX71; Olympus, Tokyo, Japan). Cell apoptosis was detected using an assay for degradation of DNA (DeadEnd Fluorometric TUNEL System, Promega). Cell-seeded scaffolds were washed and fixed in 4% paraformaldehyde according to the manufacturer's protocols. After preparing 4-μm sections as described above, fixed cells were permeabilized with Triton X-100, then labeled with the Terminal Deoxynucleotidyl Transferase (TdT) enzyme. Apoptotic cells were identified by green fluorescence using a fluorescence microscope. Images were acquired from three different positions for the analysis. Statistical Analyses Statistical analyses were performed via one-way analysis of variance with a post hoc Tukey test using MINITAB version 17 (State College, PA, USA). Differences between groups were considered statistically significant at P < 0.05. Whole Blood Circulation as a Surrogate Model of Transplantation To investigate the patency of artificial vascular networks in physiological environment, we adopted whole-blood perfusion as a surrogate for transplantation [30]. According to the IRB regulations established by Korean Ministry of Health and Welfare, this research involving the use of anonymous human tissue specimens is exempt from the requirement for IRB approval. The whole-blood perfusion was performed in the perfusion culture system (Fig 2), using whole blood as a perfusate instead of culture medium. Fresh blood was collected in vacutainer tubes containing sodium citrate (BD, USA) from a healthy adult volunteer free of aspirin or other drugs that could bias the results. The blood was taken specifically for this study. After 6 d of culture, the culture medium was washed away with phosphate buffer saline and whole blood was circulated for 24 h, during which it was replenished three times. The patency of artificial vascular networks was observed by the alteration of flow rate. The tissue mimics were fixed after 24 h of whole-blood circulation and 10-μm sections were prepared for macroscopic images and hematoxylin and eosin (H&E) staining. Optimal Design of the Fluidic Network System based on the Oxygen Transport Simulation As the number of bifurcations in the 3D fluidic network increased, the oxygen distribution (Fig 3a and 3b) became increasingly uniform, and calculated resident cell volume decreased and non-hypoxic volume increased (Fig 3c); 96.5% of the cell resident volume became non-hypoxic at "Bifurcation: 4". Thus, further bifurcation was judged to be unnecessary and "Bifurcation: 4" was selected as the model for scaffold fabrication. Scaffolds with Inner-layered Fluidic Network Porous scaffolds with 3D fluidic network were successfully fabricated using the combination of indirect-SL technology and salt leaching. Because the PCL and salt were mixed in the same ratio as in the previous study [8,31], we assumed that the porosity and hydraulic permeability of the scaffold except the fluidic network part are 67 ± 2.9% and 1.51 ± 0.074 × 10 −12 m 2 , respectively. Pores were formed in the spaces occupied by sodium chloride crystals. Although the sodium particles were sieved through a 300-μm mesh, the pores sizes are irregular as shown in Fig 4c. This is the inherent limitation of salt leaching method. Channels were successfully fabricated at each bifurcation (Fig 4a) and the measured diameters were 150 μm~210 μm less than the designed diameters (Table 1) due to shrinkage of the 3D fluidic network mold during collagen coating. The micro-CT image shows the bifurcating channels in the porous scaffold (Fig 4b). Because the plane of bifurcation varies with position, the representative image was taken from the "Bifurcation:4" model. The channel lumens were covered with thin collagen layers and the pores were unexposed on the lumen (Fig 4c). FT-IR spectra (Fig 4d) of films show the N-H stretching vibration (3330-3310 cm -1 ) that is the major characteristic peak of collagen, and also an amide I band (1640-1660 cm -1 ), an amide II (1535-1550 cm -1 ), and an amide III (1230-1270 cm -1 ) band. Compared with the spectra of raw collagen, the peaks of processed collagen are present at similar positions; therefore the solvents did not significantly denature it during the fabrication process. Engineered-Tissue Containing Vascular Networks To develop artificial vascular networks from the fluidic networks, endothelization on the luminal surface was attempted. The collagen layer was assumed to have two functions on the luminal surface. First, the collagen layer can improve endothelization [32]. PCL, which is the material of the porous scaffold, is hydrophobic and lacks active sites for biomolecule immobilization and cell attachment. A collagen layer on the PCL surface can improve the initial cell attachment and promote formation of a confluent endothelial layer. Second, the collagen layer can prevent endothelial cell migration from the luminal surface into the porous scaffold. The pore size of the scaffold is~300 μm, whereas the size of endothelial cell is~10 μm; therefore, cells could infiltrate and migrate through pores on the luminal surface when cell suspension is injected into the fluidic network for seeding. The collagen layer can be a barrier between endothelial cells and porous scaffold. After 6 d of perfusion culture most endothelial cells were on the luminal surface, whereas fibroblasts were distributed in the porous scaffold region (Fig 5a-5c). Some endothelial cells were present in the porous scaffold region (Fig 5d), possibly due to overflow of cell suspension during the injecting process or leakage through holes of the collagen layer. Stained image (Fig 5e) from the cross section of the 3D vascular network shows a similar cell distribution on the luminal surface. Because of the opacity of PCL, the formation of adherence junctions between endothelial cells in the form of 3D border lines could not be seen easily, but the VE-Cadherin staining from the cross section of 3D vascular network shows evidence of an endothelial cell lining with intracellular junctions (Fig 5f), which defines the endothelial barrier functions [33,34]. Functions of Vascular Networks in Tissue Mimics To directly investigate whether hypoxia induced cell death, we compared the percentages of apoptotic cell death and hypoxic cells in two separate areas, R lumen and R scaffold in the cross sections at the 1 st bifurcation. R lumen and R scaffold were defined as the area within and outside 1 mm from the channel lumen, respectively. The percentage of apoptotic cell death significantly increased to 41.2% in R scaffold where the ratio of hypoxic cell was 78.2% (Fig 6). During whole-blood perfusion, the chamber containing tissue mimics without collagen layer (VN non-collagen layer ) filled with bubbles soon after perfusion and the flow rate became extremely slow, but the chamber containing tissue mimics with collagen layer (VN collagen layer ) started to fill with bubbles after 24 h (images are not shown). Macroscopic pictures of cross sections of VN non-collagen layer and VN collagen layer demonstrate that the channels of VN non-collagen layer started to be occluded at the 1 st bifurcation and were almost occluded at the 2 nd bifurcation (Fig 7a and 7b). In contrast, the channels of VN collagen layer were perfectly patent at the 1 st bifurcation and half of channels were patent even at the 3 rd bifurcation (Fig 7d and 7e). H&E staining images definitely reveal that the occlusion was caused by red blood cells, which are the primary factors of thrombosis (Fig 7c and 7f). The formation of bubbles during perfusion was assumed to be induced by blood flowing thorough the micropores in the scaffolds instead of through the occluded channels. Discussion We have performed a series of studies to develop tissue mimics containing functional vascular networks which are designed based on oxygen transport simulation [8,31]. First, the effective diffusion coefficient in a cell-seeded scaffold was experimentally measured [31]. Then a procedure to design an effective fluidic network system for an engineered tissue was proposed based on the coefficient [8]. And the reliability of the procedure was demonstrated by experiments using scaffolds containing the 2D microfluidic network system. Herein, we eventually developed a large-volume scaffold containing 3D functional vascular network that allows stable mass transport within the scaffold. This study not only adapts our strategy to design effective fluidic network based on numerical analysis in 3D, but also shows the technical development to construct tissue mimics containing endothelium as artificial vascular networks. Contrary to other reports [13,19], cells were repopulated after completing the fabrication process due to the use of NaOH solvent for removing the sacrificial mold. However, the results demonstrated that cells could be evenly repopulated in the large-volume scaffold with vacuum-aided seeding [31], and that spatially-distinct endothelium could be formed on the lumen of the porous monolithic structure. Moreover, the technology can fabricate 3D channels with various branching patterns and angles, can use a wide range of biomaterials, from synthetic polymers to natural polymers, and can be combined with other technologies such as gas foaming and phase inversion to form micropores in the scaffold. In our previous study [8], the distance from the surface exposed to medium to the point at which pimonidazole staining first occurred was 1.2mm. This value was set to be the limit of non-hypoxic area and the volume of a cylindrical scaffold (10 mm in diameter × 10 mm in Large-Volume Tissue Mimics with 3D Vascular Networks length) was determined to exceed this limit as well as to be close to the maximum volume which can be fabricated with our current system. Highly branched fluidic networks can densely pervade a region, but the increase in channel volume is accompanied by a decrease in cell residence volume in a scaffold [8]. Moreover, decreasing channel width increases the risk of thrombus formation when the network is surgically connected to the host vasculature and blood flows through the channel in vivo. Therefore the design of 3D vascular networks was determined from the aspect of compromise between mass transfer efficiency and volume loss/ thrombogenic potential. In this study, the thinnest channel diameter was designed as 790 μm (fabricated result: 570 μm), but the previous study on the development of indirect-SL technology [23] suggests that channels with diameters of several tens of microns (i.e., the size of an arteriole) could be easily fabricated with this technology. And the data regarding GPC and mechanical properties of PLGA scaffolds fabricated by the indirect SL technology showed no significant changes in biomaterial properties [23]. Considering PCL degrade more slowly due to the presence of five hydrophobic-CH2 moieties in its repeating unit, we expect that the fabrication procedures do not significantly affect the mechanical properties of PCL as well. The significant increase of apoptotic cell death in R scaffold represents that the cell death was induced in response to hypoxia. However, although the percentage of hypoxic cells was 78.2% in R scaffold , the percentage of apoptotic cell death was only 41.2%. Severe and prolonged hypoxia may initiate apoptosis, whereas cells often adapt to acute and mild hypoxia and survive [35]. In this study, the positive hypoxic staining indicated that cells were alive under hypoxic conditions (oxygen concentration < 14 μM) but it could reveal neither the severity nor duration of hypoxia. The patency of VN collagen layer was maintained distinctly longer than VN non-collagen layer , but occlusion occurred after 24 h of whole-blood circulation in VN collagen layer . This difference was presumably caused by the absence of endothelium on the lumen and perturbed hemodynamics in VN non-collagen layer . The thin collagen layer plays a role as a barrier on the lumen and helps endothelial cells to reside on the lumen rather than inside the porous scaffolds (Fig 5c and 5d). This endothelium makes the lumen surface to be anti-thrombogenic. However, PCL exposed on the lumen of VN non-collagen layer is not considered to contribute to the rapid thrombosis formation because it is a well-known bio-inert polymer [36]. Instead, the porous and rough surface perturbed hemodynamics such as shear, collision of blood element with the wall, and prolonged contact, which resulted in thrombus formation [37,38]. And the bubbles formed in VN non-collagen layer generated extremely high air-blood interfaces and aggravated thrombosis [39]. However, the results from the whole-blood perfusion indicate that the artificial vascular networks cannot function as long-term mass transport channels in vivo. The artificial vascular networks must be combined with induction of microvessel formation by biomolecular cues and vascular cells to reduce the time required for regeneration of large-scale multicellular organs, thereby reducing the risk of thrombus formation. This study focused on maximizing cell viability rather than specific functions by introducing fluidic channels in scaffolds. Fibroblasts are suitable for this purpose due to the ease of handling, abundant data about the metabolic properties which have been identified by researchers [31,40]. However, further study is required to investigate the effect of oxygen transport on cellular functions and potential for regenerating specific organs such as liver, lung and kidney. Conclusion This study demonstrated that the modification of indirect-SL technology made it possible to construct artificial vascular networks in a porous scaffold by introducing hydrogel barrier on the luminal surface. This fabrication technology has the advantages of indirect-SL such as construction of 3D freeforms, wide selectivity of biomaterials, and compatibility with various other methods. The artificial vascular networks functioned as channels for oxygen transport, reducing the hypoxic volume and preventing apoptotic cell death. Furthermore, the endothelium of the vascular networks significantly retarded the occlusion of channels during wholeblood circulation. As the technology matures, the tissue mimics can be used as in vitro platforms to examine the physiologic and pathologic phenomena through vascular architecture. Furthermore, incorporation of microvessel formation in response to biomolecular cues and vascular cells, may allow large-scale formation of multicellular organs and reduce the risk of thrombus formation.	v3-fos-license
2021-05-17T13:19:50.254Z	2021-05-17T00:00:00.000	234683739	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.frontiersin.org/articles/10.3389/fmech.2021.685519/pdf", "pdf_hash": "b5e4147537da0237d2029417ee0b1ba48ee49ab3", "pdf_src": "Frontier", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45318", "s2fieldsofstudy": [ "Biology", "Engineering" ], "sha1": "b5e4147537da0237d2029417ee0b1ba48ee49ab3", "year": 2021 }	pes2o/s2orc	Numerical Investigation of Axonal Damage for Regular and Irregular Axonal Distributions Recently, various researches have revealed the importance of the investigations performed for evaluating mechanical properties and damages of the brain tissues while dealing with the production of surgical ligaments and helmets. Therefore, it is vital to study the structure of the brain both experimentally and numerically. By experimental tests, despite being costly, it is almost impossible to establish stress distribution in micro scale, which causes injury. Micromechanical predictions are effective ways to assess brain behavior. They can be applied to compensate for some experimental test limitations. In this work, a numerical study of the axonal injury in different heterogeneous porcine brain parts with different axon distributions under quasi-static loading is provided. In order to produce a heterogeneous structure, axons are distributed in regular, semi-regular, and irregular patterns inside the representative volume element. To accurately examine the brain tissue time-dependent behavior, a visco-hyperelastic constitutive model is developed. Also, axonal damage is studied under different conditions by applying different levels of load and rate. Because of geometrical complexities, a self-consistent method was applied to study the damage in higher volume fractions of the axon. The results reveal that the regions of the brain enjoying a regular axon distribution would have higher strength. In addition, among the two influential load and loading rate parameters, the brain tissue in all regions shows more sensitivity toward the applying load. INTRODUCTION Any alteration in brain function, caused by an external force is defined as traumatic brain injury (TBI) (Menon et al., 2010). Direct impact effect of external force, quick acceleration (Smith and Meaney, 2000) and deceleration, infiltration of external objects such as gunshot and exposure to the blast wave (Chafi et al., 2010) would lead to TBI (Maas et al., 2008). TBI could be called one of the well-known public health problems. Every year, 1.7 million people experience TBI (Langlois et al., 2006). Falls resulting in 35.2% of TBIs come first when discussing the causes of TBI, motor vehicle traffic, struck by/against events, and assaults is on the next place (Faul et al., 2002(Faul et al., -2006. Since TBI leads to both death and disability of individuals, and it is also more common among the youth (75% in males), it imposes high costs on society (Maas et al., 2008). Axonal damage due to brain impact and TBI can cause axonal death and disorientation (Peter and Mofrad, 2012). In most cases, symptoms of brain injury are invisible to the eye (Tanielian et al., 2008). Hence, to get a better explanation of the nature of TBI, simulation of the injury under different conditions could be helpful. Considering the vast studies performed on damage, their impact on mechanics and the advances made in finite element (FE) simulation in recent years, assuming the brain tissue as a mechanical matter, computer simulations could be employed to study TBI. Dealing with brain tissues, material properties and injury criteria (i.e. mechanical behavior of the tissue) are the two most important parameters to be considered. Recently, several procedures have been provided to study the mechanical properties of the brain tissue. Atomic force microscopy (AFM) (Magdesian et al., 2012) is one of the most vastly used devices to characterize the brain tissue, especially single neuron mechanical behavior (Bernick et al., 2011). Magdesian et al. (Magdesian et al., 2012), performed an experimental test on a single neuron employing AFM. They found out that the axons have time dependent behavior. In another study, critical pressure to axonal injury and reversibility of the damage were investigated in hippocampal axons and dorsal root ganglion (DRG) (Harper and Lawson, 1985;Caffrey et al., 1992) neurons under gradual forces. Apart from AFM, Micropipette Aspiration (Schmid-Schönbein et al., 1981), Scanning Force Microscopy (SFM) (Christ et al., 2010), Transmission Electron Microscopy (TEM) (Labus and Puttlitz, 2016), Magnetic Resonance Elastography (MRE) (Anderson et al., 2016;Testu et al., 2017;Schmidt et al., 2018;Weickenmeier et al., 2018), as well as Digital Image Correlation (DIC) technique (Libertiaux et al., 2011) are other common devices used for studying mechanical behavior of the brain tissue. Considering the experimental test limitations such as preparing the sample and the lack of access to the test setups, numerical methods have been developed to predict the mechanical behavior of the material. Therefore, many different constitutive models were presented to study brain tissue in macro and micro scales. Wright et al. (Wright et al., 2013) performed a macroscale simulation of brain tissue using MRI data while applying an acceleration curve on the model. They found out that rotational acceleration has a more sensible effect than translational acceleration on injury response. Additionally, by employing numerical methods, researchers were able to provide a detailed record of the phenomenon happening in different conditions, while there was a huge difficulty for in vitro experiments. While performing micro scale experimental tests on the tissue is still challenging Karami et al., 2009;Yousefsani et al., 2018), characterization of a single neuron (Goriely et al., 2015) is one of the most common micro scale simulations of the brain tissue. Abolfathi et al. (Abolfathi et al., 2009) implemented numerical tests to study axonal undulation and neuron volume fraction impacts on brain tissue mechanical properties. It is found out that undulation has a significant effect on material stiffness in the axon longitudinal direction, while axon volume fraction has an overall influence on the tissue behavior. Based on a similar structure of brain tissue and composites, common concepts, algorithms, and failure metrics of micromechanics could be considered in biomechanics for modeling the tissue. Representative volume element (RVE) is a helpful tool to simulate and study the effective properties of a macroscopic material (KAZEMPOUR et al., 2018;Sheidaei et al., 2019). Appropriate selection of the RVE size is an important factor to have an accurate effective behavior prediction (Hashin, 1983). In such a system, strain (Bain and Meaney, 2000;Ganpule et al., 2013) and strain energy (Goriely et al., 2015) are two micromechanical parameters which could be employed to predict the brain injury level. The unknown coefficients of the employed constitutive model could be determined through optimization functions , while these functions are also beneficial for studying and unraveling the problems of human body tissues (Chavoshnejad et al., 2018). Miller et al. (Miller et al., 1998) considered strain, stress, and negative pressure parameters as the necessary injury criteria to determine more vulnerable regions of white and gray brain tissue matters. In fact, when a point is damaged, its surrounding area experiences higher loads, which results in a higher risk of injuries. This necessitates a damage function that can model the axonal injuries more realistically. Goriely et al. (Goriely et al., 2015) advantaging from continuum theories, presented a damage criterion for axons which is a function of strain energy or strain. To wrap it up, studies performed on conventional injuries made a great deal of effort to demonstrate vulnerable regions of the tissue, however, they mostly fail to provide damage distribution, damage initiation and injury evolution. In this study, the RVE definition was employed to micromechanically study the brain tissue. Axons and Extra Cellular Matrix (ECM) are considered as the constituent parts of the tissue in the RVE. A visco-hyperelastic constitutive model presenting damage function has been utilized as the material model [cf. Goriely et al. (Goriely et al., 2015)]. The model along with the injury criteria is assigned to the RVE. Therefore, a VUMAT ABAQUS subroutine including damage parameters, i.e. two viscoelastic terms and a hyperelastic term, has been developed to define the RVE behavior. Axons were distributed irregularly in a 3D RVE, and were analyzed under different strain and strain rates. Since irregular distribution would lead to geometrical complexity, the issue is resolved employing the self-consistent approximation method. The damage and material models are found to be suitable for studying the mechanical behavior of the brain tissue, while the accuracy of RVE simulation could be enhanced by taking advantage of the abovementioned technique. Representative Volume Element Construction Due to the dimensions of axons compared to the tissue level of the brain, numerical analysis of the tissue including all of its components would lead to high computational costs. Representative Volume Element (RVE) is one of the micromechanical applicable tools to decrease the computational costs of the numerical simulations while records macro scale behavior. RVE is the smallest size of a heterogeneous material which is large enough to describe the material behavior in macro size (Kanit et al., 2003). To this aim, axons, and Extra Cellular Matrix (ECM) are simulated in a 3D RVE to study axonal injury. Considering the geometrical complexity, for studying the axonal injury, undulated axons were distributed in the RVE. Since the distribution of axons in various regions of the brain tissue has both regular and irregular patterns, the corresponding isotropic and anisotropic behavior of these parts should be carefully observed. It is worth mentioning that the regions of brain with regular axonal distribution can be divided into two completely regular and semi-regular parts. In the former, axons are aligned with each other and in the latter, anisotropic behavior would be predictable. To simulate all the three different regions (two with anisotropic behavior and one being isotropic), a computational MATLAB script has been developed to distribute the axons in the RVE. The volume fraction of the axons, reference axon distances and shape as well as RVE dimensions Minimum distances between axons are given as an input to the script, using random rotational and translational matrixes, it creates the axons, and checks the intersection between them. Finally, reference axon coordinates, translation and rotation matrixes would be saved as output files. Figure 1 shows the construction of the RVEs in different regions of the brain. Constitutive Model In the current investigation, axonal damage rate in three different parts of the brain is studied. To reduce the calculation expenses, Representative Volume Element (RVE) has been defined and used. Therefore, the RVE size is required. The importance of true size specification of RVE is the prevention of probable influences of its geometric dimensions on the deduced results which is in contradiction with the definition of RVE. Diverse constitutional models are presented to predict the behavioral traits of brain and other soft tissues in which elastic, viscoelastic, hyperelastic and visco-hyperelastic (Miller, 1999;Sahoo et al., 2014;Samadi-Dooki et al., 2018;Voyiadjis and Samadi-Dooki, 2018;Kazempour et al., 2019;Chavoshnejad et al., 2020) models can be mentioned. In the current study, the elastic constitutional model is utilized to examine the RVE size, thereafter isotropic and anisotropic behavior in different parts of brain are thoroughly investigated. First, it is assumed that the RVE has an orthotropic elastic behavior and the required loadings were exerted accordingly. where Ε i , G ij and ] ij are Young modulus, shear modulus and Poisson ratios in different directions, respectively. Therefore, six tests including normal and shear tests in different directions are conducted. Through a comparison of the coefficients of the orthotropic model, the isotropic and anisotropic behavior of different parts of the brain are studied. The corresponding elastic constants for the axon and ECM are summarized in Table 2. Isotropic Constitutive Model After obtaining the optimum size for RVE and examining its behavior in various alignments, a visco-hyperelastic model and offered damage model by Goriely et al. (Goriely et al., 2015) are used to study the axonal damage. Assuming an isotropic nature (Wang and Wineman, 1972) for axon, strain energy W e would be a function of invariants of left or right Cauchy-Green deformation tensor (Holzapfel, 2000). When self-consistent method is used for isotropic parts of the brain, ECM can be modeled as an isotropic material. Therefore: Where I i s are invariants of either the left or right Green deformation tensor (B or C). Hence, where B, C, J, W iso e and W vol e are deviatoric parts of left and right Cauchy-Green deformation tensor, the square root of I 3 , deviatoric, and volumetric parts of the strain energy, respectively. Since the responses vary in different physical conditions such as variation of the strain rate, temperature, strain levels, and other parameters, a visco-hyperelastic constitutive model has been developed to consider the time-dependent behavior of the brain tissue. Similarly, the stress is also divided into two timedependent and elastic parts (Holzapfel, 2000). Where σ e and σ v i are quasi-static and viscous stress, respectively. In addition, the viscous stress is further divided into two parts including deviatoric stress and volumetric stress. Therefore, the strain energy for the time-dependent part contains a deviatoric and a volumetric energy. Bergström and Boyce (Bergström and Boyce, 1998) studies showed that quasi-static stress equations could also be applied for deviatoric part of the viscous response. As for the strain energy function, Neo-Hookean strain energy model (Eq. 9) is employed to include deviatoric parts of both the elastic and viscous stresses. It is noteworthy that Neo-Hookean coefficients have different values for each of the elastic and viscous parts. Moreover, for the viscous part of the stress, as mentioned earlier, two relaxation time constants are assumed so as to account for both fast and slow loading velocities. where θ i is the relaxation time constant. Moreover, the strain energy function related to the volumetric part of the energy is expressed with a well-known relation as Eq. 11 (Horgan and Murphy, 2009;Zhang et al., 2014;Whitford et al., 2018). where K, K ′ and K } are the volumetric modulus of elastic and viscous parts which are considered to be equal in this study. Finally, this model is numerically implemented as a VUMAT subroutine for studying the brain tissue. Anisotropic Constitutive Model Since, a self-consistent approximation is used to increase the volume fraction of the axon, for the RVEs that are used in anisotropic regions, an anisotropic behavior for the ECM is employed. Eq. 12 shows the behavior of the anisotropic regions. where I 1 and I 3 are invariants of either the left or right Green deformation tensor (B or C) and I 4 is the pseudo-invariant of C and a 0 ⊗a 0 (Holzapfel, 2000) which is given by: Note that invariant I 4 is equal to the square of the stretch in the fiber direction a 0 (Holzapfel, 2000) defined for incompressible material such as brain tissue (Libertiaux et al., 2011;Fereidoonnezhad et al., 2020). Similar to the isotropic model, the strain energy could be separated to volumetric and deviatoric parts. Similarly, time-dependent parameters are added to the anisotropic model. Therefore, the stress is calculated using Eq. 6 and Eq. 7. Although stresses in isotropic and anisotropic models could be obtained by the same logic, the difference comes from the point that the anisotropic model has invariant I 4 parameter. Additionally, are the first and second invariants of the deviatoric part of left Cauchy-Green deformation tensor and is deviatoric part of square of the stretch in the fiber direction a 0 (Holzapfel, 2000). The models are implemented in VUMAT subroutines for a stress analysis in ABAQUS. Therefore, discretization of the models is needed in small time increments. Eq. 6 could be discretized the following way, Considering Eq. 10 and assuming the small time increments, the numerical stress relation could be obtained. Where Eq. 16 is a recursive relation to calculate the stress in every time increment based on the strain in the current time increment and the previous one's. In order to implement Eq. 16 in VUMAT subroutine, Cauchy stress must be calculated (Holzapfel, 2000). Using the provided derivation and separation relations by Holzapfel (Holzapfel, 2000), Cauchy stress would be obtained as: It is noteworthy that in the isotropic model, all of the I 4 invariant coefficients are equal to zero. Damage Criteria One of the forms of traumatic brain injury is Diffuse Axonal Injury (DAI) (Faul and Coronado, 2015), that occurs when the brain rapidly moves in the skull. Axons are the long connecting fibers which would suffer a cut in high accelerations and decelerations. Many parts of the brain are vulnerable to such an injury, therefore investigation of the DAI in different parts of the brain becomes of a high importance. Goriely et al. (Goriely et al., 2015) in 2015 studied axonal damage and presented damage criteria based on strain energy. where W follows Eq. 3 for the isotropic model and Eq. 12 for the anisotropic model. Moreover, d is the scalar damage variable which would be changed from zero to one (zero for intact axons and one for completely injured ones). The multiply of damage variable in undamaged stored energy W would reduce it to the injured stored energy ψ. Since the stresses related to the damaged material must be calculated from the injured strain energy, Eq. 20 would be translated to: where σ is the stress in the injured material. By substituting Eq. 17 in Eq. 21: Likewise the equation of Goriely et al. (Goriely et al., 2015), the proposed damage variable is a function of initial damage d 0 , maximum amount of damage d ∞ , damage increasing controller ξ, damage threshold κ and damage history variable κ which records the maximum elastic energy W. κ 0 is the initial damage threshold and W(s) shows the stored energy during the loading history −∞ < s < t. Finite Element Modeling In order to construct the geometric model of RVE in ABAQUS software, a python script has been written in which reference axonal coordinates, generated translation and rotation matrixes by MATLAB script is read. Through this script, axons would be created, and dispersed in the RVE. In further steps, python code specifies appropriate material for the axon and ECM, distinguishes the existent contacts and creates a complete connection among them. For the geometric discretization of ECM and axons, C3D4 and C3D8 elements have been employed respectively. To move forward, a tensile load needed to be exerted, hence, a loading model presented by Shaoning has been used. In this model, three orthogonal facets have symmetric boundary conditions being constrained in normal direction to the same face and one of three remained faces is subjected to a vertical tension. Zero boundary conditions in various loadings have been demonstrated in Table 1 employing notations given in Figure 2. Subsequently each RVE is studied while subjected to two loading Frontiers in Mechanical Engineering \| www.frontiersin.org May 2021 \| Volume 7 \| Article 685519 modes with different sizes and rates. In order to exert shear load, one of the faces is fully constrained and the corresponding free parallel face is loaded in the shear direction. Following various tests and for providing homogeneous results on macroscopic level, volume averaging is used in microscopic scale. The volume average stress and strain are obtained according to Eq. 25: Where Ω m , σ m and ε m are the total volume of RVE, local microscopic stress and strain, respectively. Due to the geometrical intricacies of the RVE, increasing axonal volume fraction leads to the rise of calculation expenses. Indeed, this increase makes axons geometrically closer to each other leading to geometrical complexity then discretization and RVE meshing problems would be generated. To solve this problem, a self-consistent method should be applied. Regarding selfconsistency theory, heterogeneous composites are able to be simulated by homogeneous material, in this manner, average mechanical constants for homogenous material should be presented. In the aforementioned method, homogenized coefficients of a heterogeneous composite would be applied on the basis matrix. In case of further axonal distribution in this matrix, the increase in volume fraction will be closely observed. In order to apply self-consistency estimation, PSO function is exerted in a MATLAB code form. Normal and shear stresses according to time are introduced as input data to the function. By employing an optimization function, a homogenized basis matrix on the fitted graphs by a second axonal distribution is obtained, based on which a new RVE with higher volume fraction is created. Subsequently, the MATLAB code runs the python code in which ABAQUS software begins to analyze the matrix structural model by using the VUMAT subroutine. Finally, after finishing the analysis, a python code gives a time correspondent stress chart. MATLAB reads the python output for the produced homogenized matrix coefficients and compares them with heterogeneous composite data. This procedure goes on until the graphs fully fit each other with acceptable tolerances. Appendix Figure A1 briefly demonstrates the applied procedure. RESULTS AND DISCUSSION In this section, proceeding to the selection of RVE size, as an important parameter affecting the whole results, their behavior has been analyzed. The resultant isotropic or anisotropic behaviors have been validated through various distributions of the axons in the medium. Consequently, the effect of volume fraction is also investigated to see whether it implies any changes in the predicted behavior of the RVE. Following the validation of the selected RVE size, a more real condition has been applied to the model (i.e. the material and damage criteria). Considering a more practical situation, the effects of volume fractions, loading rates, and loads have been studied on the damage propagation. Employing Elastic Model to Investigate Representative Volume Element Size, Isotropy and Irregular Distribution Mechanical properties should be independent of RVE size. A wrong decision of the RVE size would lead to less independence of mechanical properties. Therefore, RVE size would be concerned as one of the most important parameters in predicting effective attributes of heterogeneous materials. Considering the impact of boundary conditions on the results, on one hand, calculation of the outputs by using a smaller RVE size could not be acceptable. On the other hand, by enlarging the RVE, the calculation expenses will grow dramatically, therefore, RVEs are generated in different sizes (500, 700, 900, 1100 micrometers) with volume fraction of 5%. The applied attributes of RVEs are displayed in the Table 2. Subsequently, studying RVEs according to the elastic model, the optimum size has been selected through analyzing their effective elastic modulus (further details are provided in the following sections). The effective elastic modulus for various sizes of RVEs with the same volume fraction quantity are displayed in Figure 3. According to Figure 3, increasing RVE size to more than 900 mM brings no significant changes in the results, indicating an appropriate RVE size equal to 900 mM, in this study. It can also be understood that the results are independent of RVEs size, and the 900 mM RVE size leads to an optimized computation expense. Due to the random distribution of axons, for RVE creation, it is necessary to examine their behavior, especially for irregular and semi-regular distribution sections. Thus, an orthotropic behavior is considered for the RVEs with a size of 900 mM, summarized in Table 3. According to the obtained results for loading RVEs in various directions, RVEs with irregular distribution have an isotropic response and RVEs with regular distribution show anisotropic behavior. Therefore, the mechanical behavior of the RVEs with isotropic behavior is predictable by two tests and the behavior of the RVEs which behave anisotropically could be predicted by six tests. Due to the anisotropic behavior of the RVE with regular distribution, subsequent studies for these regions are performed with anisotropic constitutive models. However, in order to ensure the isotropic behavior of the RVEs with the irregular distribution of the axons, it is necessary to examine the other two conditions. The effect of different random distributions on mechanical behavior and the study of the RVE in different volume fractions are the states which should be considered for the investigation. In order to investigate the effects of different random distributions, three RVEs with various distributions were created and studied in 5% axonal volume fractions. The results show independent behavior of the axonal distribution for the RVEs (Table 4). Axonal volume fraction influence on isotropic behavior is also studied by the creation of three RVEs with volume fraction of 5, 10, and 15%, subjected to a tensile strain of 3%. RVEs were subjected to static loading. In order to increase the volume fraction, self-consistent approximation was used. The results show that enhancing axons volume fraction leads to an increase in the effective elastic modulus. Table 5 presents the results of the effects of volume fraction on the mechanical properties of the RVE. As can be seen in Table 4, elastic modulus in different volume fractions is almost equal for perpendicular directions, which again confirms the isotropic behavior of the representative volume element. In Figure 4, 3% strain is applied to the RVE as a static load. The results show that axons experience higher stresses than ECM. Therefore, axons play a more important role in the brain injury investigation compared to ECM. It can be interpreted that the micromechanical study of the stress distribution in the brain tissue is one of the most important issues in the TBI study. Assuming the elastic behavior in this section, the importance of micromechanical analysis is shown. Therefore, in this study, for achieving a more realistic result, a visco-hyperelastic model is developed. Employing Visco-Hyperelastic Model for a More Realistic Response of the Brain Tissue Javid et al. (Javid et al., 2014) performed an experimental test on the brain tissue in which mechanical properties of axon and ECM were presented separately. In the present study, the outcomes of these tests were used to extract the coefficients of the viscohyperelastic model. To evaluate the developed constitutive model, a RVE with 52.7% axonal volume fraction is generated. In the next step, 3% strain is applied on the created RVE and the results were compared with the experimental outcomes represented by Javid et al. (Javid et al., 2014). The results are highly consistent with the experimental results and could be used in the present study. Table 6 indicates the obtained material coefficients of the RVE components. For a better illustration, the results obtained by Javid et al. (Javid et al., 2014) along with the mean volumetric stress of the RVE calculated in the present study are demonstrated in Figure 5. In order to investigate axonal damage in different regions of the brain, RVEs were created with different volume fractions in different axonal distributions. RVEs were subjected to two various strains in two different loading rates, and the RVEs responses were evaluated with respect to time. Strain loads of 10 and 15% were applied in tensile and shear modes at times of 0.5 and 1s in a quasi-static manner on the RVEs. Figure 6 shows the loading conditions. It is also important to note that micromechanical analysis is time-consuming, especially in quasi-static and dynamic loadings, in which the used constitutive model has a timedependent behavior. In the present study, due to intangible variation of stress over time after 6 s, the analysis was performed up to 6 s. In Figure 7, the variations of the mean volumetric stress in the loading direction are illustrated for 15% volume fraction of axons under 15% tensile strain in a region with irregular distribution. Figure 7 demonstrate the variations of the mean stress of the RVE versus time with volume fraction of 15% in the region with a semi-regular distribution. As can be seen, the stress increases up to the 0.5 and 1 s at different loading rates. Due to the viscous behavior in the RVE, when the strain reaches a constant value, a decrease in stress would be observed. Moreover, due to the loading rate behavior in the brain tissue (Pervin and Chen, 2009), increasing the loading rate leads to a firmer response in the tissue (Rashid et al., 2014). It should be noted that in accidents, the loading rate on the brain tissues is usually high, which indicates that the brain tissue is more vulnerable. These figures present a suitable response of the rate dependency of the developed constitutive model which could be very useful when studying the rate dependency of the brain tissue. Figure 8 shows the mean volumetric stress of the axons against time for axons in both non-damaged and damaged states. It is found out that in damage presence, the experienced stresses by the axons are reduced, indicating a decrease in their stiffness and strength. Moreover, Figure 9 shows the maximum experienced stress by the axons in both damaged and non-damaged modes. According to the figure, the RVE with regular distribution of axons expresses higher stiffness than the semi-regular and irregular distribution modes. Stress also drops in the regular, semi-regular and irregular distribution mode by 10.7, 9.9, and 7.8%, respectively. Due to the full axon alignment along the loading direction in regular distribution, the largest reduction would occur in this section coming from a higher strain energy experienced by the axons. According to the damage model presented in Eqs 23, 24, due to the dependency of the damage on the strain energy, the damage starts to grow from the loading onset. Therefore, the damage amount could be compared between loading, loading rate and different volume fraction. Figure 10A shows the stress of the axons and the RVE at 10% volume fraction and 15% strain load, and different loading rates in the non-applied mode of the damage model. The results show that the experienced stress by the axons is higher than that of ECM ). In addition, as the strain rate rises, the rate of increase in the maximum experienced stress in the axons grows greater than that of ECM, indicating an increase in crash vulnerability ( Figure 10B). Moreover, the following figure shows the experienced stress in the RVE at 10% strain and 10% volume fraction in the non-applied state of the damage model. According to Figure 11, studying different distributions of axons would reveal that the RVE with regular distribution shows a higher stiffness due to the axon presence in loading direction. This suggests that those regions in the brain in which the axons have a more regular orientation express stiffer behavior. In addition, by increasing the loading rate, regions of the brain which have a more regular axon distribution reveal a higher raise in the stiffness. Figure 12 displays the maximum mean experienced stress for three different axon distributions, 10% strain load and 10% volume fraction at two different loading rates. According to the figure, in all loading rates, brain tissue regions having a regular axonal distribution express larger stiffness. In addition, at higher loading rates, the damage amount increases, indicating the sensitivity of the brain tissue to the loading rate. Figure 13 shows the changes in the maximum experienced mean stress by the axons at 15% load and 5 and 10% volume fraction in the presence of the damage model, and without the model. As shown in the figure, by rising the axon volume fraction, the brain tissue in all regions shows stiffer behavior. Moreover, in constant loading, by increasing volume fraction, stress-drops rate decreases during injury, indicating a grow in tissue stiffness in all areas Karami and Shankar, 2011). In addition, in a constant volume fraction, in the brain regions where axons have a more regular distribution, by applying the damage model, the stress-drop rate increases. This phenomenon indicates more energy is required to apply a constant strain to these areas. According to Eqs 23, 24, the damage parameter is a function of strain energy. An enhancement in absorbed strain energy leads to an increase in the damage coefficient, and subsequently in the stress reduction. However, the areas of the tissue having a regular distribution of axons still show higher strength in case of injury. It should be noted that due to the low volume fraction of axons and the proximity of the axon properties and ECM, the variation rate in tissue behavior is not noticeable. Figure 14 shows the maximum experienced mean stress by axons at 5% volume fraction, 10 and 15% strain. As can be seen, the experienced stress by axons increases in all axonal distributions while enhancing the load. Furthermore, the rate of increase for the brain tissue is higher when dealing with the regular distribution of axons, which indicates higher stiffness in these areas. Moreover, as the load increases, the drop amount in stress increases in all areas, indicating a rise in axonal damage. Figure 15 and Figure 16 show the von Mises stress distribution on the axons and the homogenized stress for the ECM, respectively. The RVE is under the strain of 15%, which shows the volume fraction of 5% in 0.5 s for the irregular distribution of axons. According to the following figure, the maximum average stress on the ECM is approximately 291 Pa, while the experienced stress at the same time for the axons is between 693 and 1353 Pa. It is clear that the stress observed in ECM is less than that of axon, confirming previous results under different loading conditions. It can be concluded that axons experience more stress, and are more vulnerable. Karami et al. (Karami et al., 2009) and Javid et al. (Javid et al., 2014) also came at the same conclusion. Moreover in all experiments, the FIGURE 14 \| Maximum mean experienced stress by axons with 5% VF under 10 and 15% strain including the damage and non-damage modes. FIGURE 15 \| Distribution of the maximum stress on axons of a RVE with 5% volume fraction on which 15% strain is applied at 0.5 s (the results should be multiplied by 1e12 Pa). Frontiers in Mechanical Engineering \| www.frontiersin.org experienced stress in neurons is higher than that in ECM . CONCLUSION In the present study, the brain tissue behavior and axonal damage extent in different loads, volume fractions and loading rates were investigated for different parts of the brain tissue in regular and irregular axonal distributions. For this purpose, a MATLAB code was developed which creates axons in different distributions; then, the generated coordinates by the Python code were entered into ABAQUS, and the aforementioned analysis was performed on the tissue in this software. It should be noted that due to the small size of the axons at the tissue level, RVE definition has been used to analyze the tissue. Subsequently, the elastic model has been utilized to investigate the optimal RVE size, and avoid the effects of boundary conditions. Furthermore, the amount of tissue anisotropy has been studied in different distributions. For this, a visco-hyperelastic model has been developed to examine the load to texture, loading rates and different volume fractions. It is found out that by increasing the loading rate, the experienced stress by axons increases, indicating that the tissue is vulnerable to high loading rates. Moreover, as the load increases, the experienced stress by the axons rises, while the amount of tissue sensitivity to the load becomes much greater than the loading rate. In regular axon distribution, for instance, a 1.5 times increase in load, causes the maximum experienced stress increase by 1.33 times, while by doubling the loading rate, the enhancement of the axonal stress would be 1.17 times higher. This shows the importance of efforts to reduce the load on the brain tissue. In addition, the experienced stress by axons under the same loading conditions is greater than the ECM, which emphasizes the necessity of the existence of DAI. The results of this study could be useful in the production of helmets, surgical robots, and all the tools which are used for preventing or treating the brain injuries. DATA AVAILABILITY STATEMENT The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.	v3-fos-license
2023-08-26T16:08:51.816Z	2023-08-01T00:00:00.000	261149428	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2072-666X/14/8/1635/pdf?version=1692591082", "pdf_hash": "affffd90dbeb0aaa354352e04dfb536f6dfb0dab", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45319", "s2fieldsofstudy": [ "Materials Science", "Engineering" ], "sha1": "f3f1c3bccc7ec9b9195a293d41c73ff92f7b446c", "year": 2023 }	pes2o/s2orc	Study on the Effect of Cold Deformation and Heat Treatment on the Properties of Cu-Ag Alloy Wire The effects of various drawing parameters and annealing processes on the structure and properties of Cu-Ag wires, containing 1 wt% silver, were investigated using specialized equipment including fine wire-drawing machines, very fine wire-drawing machines, heat treatment equipment, tensile testing machines, microcomputer-controlled electronic universal testers, resistance testers, and scanning electron microscopes. The results revealed that continuous drawing of Cu-1%Ag alloy wires led to elongation of the grains, resulting in a uniform and tightly fibrous microstructure. Moreover, the tensile strength of the alloy wire increased from 670 MPa to 783.9 MPa after a single pass with a deformation of 14%. Subsequently, when the wire was drawn at a speed of 500 m/min, the tensile strength further increased to 820.1 MPa. After annealing the Փ0.08 mm Cu-1% Ag alloy wire, an increase in annealing temperature up to 500 °C resulted in the wire’s tensile strength decreasing from 820.1 MPa to 377.5 MPa. Simultaneously, the elongation increased from 1.94% to 15.21%, and the resistivity decreased from 1.931 × 10−8 Ω·m to 1.723 × 10−8 Ω·m. Additionally, when annealing was conducted at a rate of 80 m/min, the wire resistivity dropped to 1.635 × 10−8 Ω·m. Introduction The rapidly expanding microelectronics industry has presented significant challenges in chip packaging technology and materials.These challenges encompass a range of requirements, such as achieving enhanced levels of hermeticity and integration, improving heat dissipation capabilities, and ensuring the reliability of packaged chips.As the industry continues to evolve, meeting these demands has become crucial for advancements in microelectronic devices and systems [1,2].Wire bonding is widely recognized as the most popular chip interconnection technology in microelectronic packaging due to its mature process, excellent bonding performance, high reliability, and cost-effectiveness [3][4][5][6].The bonding wire, serving as the core material of the package, plays a crucial role in connecting the pins and silicon wafers, as well as transmitting electrical signals.Currently, there is a demand for finer wire diameters and higher strength in bonding wire technology.Additionally, it is essential to maintain good plasticity, stable chemical properties, and high conductivity.In comparison to gold-and silver-based alloy wires, copper-silver alloy wires offer high strength, high hardness, and cost-effectiveness, while still preserving excellent electrical and thermal conductivity [7][8][9][10][11]. With the continuous development of integrated circuits and chips, there is a growing demand for copper-silver alloys with high strength and high conductivity.These alloys are sought after due to their ability to meet the requirements of modern-day technology and provide enhanced performance in various applications.The Cu-1 wt%Ag alloy with 98% cold-rolled depression prepared by Jianlan Chen in combination with intermediate heat treatment obtained a comprehensive tensile strength of 956 MPa and a relative electrical conductivity of 72% [12]; Lin-Sheng Tang used vacuum casting to prepare Cu-4.5 wt%Ag alloy, through the "casting-peak aging-cold drawing" process, the strength of coppersilver alloy wire reached 830 MPa, an electrical conductivity of 85% IACS [13].Yanjun Zhou et al. prepared Cu-3.5Ag alloy using "solid solution-ageing-cold deformation" and "solid solution-cold deformation-ageing" processes, which resulted in a tensile strength of 417.2 MPa and a conductivity of up to 95.5% IACS [14].Kexing Song et al. investigated the dynamic deformation behavior of Cu-20Ag alloys and concluded that the increase in strain rate promoted Ag precipitation, hindered the movement of dislocations, and increased the stress and yield strength of Cu-Ag alloys [15].By comparing and analyzing the properties of Cu-Ag alloys with different compositions, Fei Cao et al. provide references and ideas for the preparation of ultra-high strength, high conductivity and high Ag content Cu-1 wt%Ag alloys [16].Mingwang Xie used multi-pass drawing combined with intermediate annealing process to prepare a high-strength, high-conductivity Cu-20Ag alloy with a strength of 1175 MPa and an electrical conductivity of 80.4% IACS [17].Varol Temel et al. used a new Cu-Ag alloy prepared by chemical plating and hot pressing and found that the hardness and density of the material increased with the proportion of silver-plated copper powder in the copper powder [18]; Choi Eun-Ae et al. designed a thermomechanical treatment process for subeutectic Cu-12Ag alloys to obtain their tensile strength up to 1189 MPa and electrical conductivity up to 71.9% IACS [19].In Cu-Ag alloy wires, the electrical conductivity of Ag content in 3-28% is relatively poor, and Ag content of 0.1% will lead to poor tensile strength [20], so the study of Cu-1 wt%Ag alloy wires is of great significance for the preparation of high-performance bonding wires.Most of the previous research has focused on copper-silver alloy wires with diameters exceeding 0.1 mm.However, there has been relatively limited investigation into microfine copper-silver alloy wires with diameters below 0.1 mm.This paper aims to fill this research gap by focusing on studying copper-silver alloy wires with smaller diameters.The objective is to explore the influence of various drawing parameters and annealing processes on the properties and microstructure of these wires. Test Materials and Equipment The experiments conducted in this study utilized Cu-1 wt%Ag wire with a silver content of 1 wt%.The wire initially had a diameter of 0.25 mm in its unannealed state.The initial parameters of the wire were as follows: a tensile strength of 670 MPa, an elongation of 1.86%, and a resistivity of 1.922 × 10 −8 Ω•m.The necessary equipment for conducting these tests is listed in Table 1.The test material is an 8.0 mm diameter Cu-1Ag alloy rod prepared by vertical leaded vacuum melting directional solidification continuous casting machine.The 8.0 mm diameter alloy rod was drawn to a 3.0 mm diameter by the large-wire-diameter drawing machine, then to a 0.9 mm diameter by the medium-wire-diameter drawing machine with multiple dies, and finally processed to a 0.25 mm diameter alloy wire by the small-wire-diameter drawing machine with multiple dies.The principle of the continuous casting machine is shown in Figure 1.The continuous casting parameters employed are as follows: the melting temperature is 1250 • C, the cooling temperature is 25 • C, the traction speed is 80 mm/min, the traction time is 0.5 s, and the stopping time is 0.5 s. Micromachines 2023, 14, 1635 3 of 16 multiple dies, and finally processed to a 0.25 mm diameter alloy wire by the small-wirediameter drawing machine with multiple dies.The principle of the continuous casting machine is shown in Figure 1.The continuous casting parameters employed are as follows: the melting temperature is 1250 °C, the cooling temperature is 25 °C, the traction speed is 80 mm/min, the traction time is 0.5 s, and the stopping time is 0.5 s.The wire-drawing die primarily consists of a die sleeve and a die core.The die sleeve serves to provide force support for the core during the drawing process of high-strength wire, preventing the core from rupturing.The conventional dimensions of the die sleeve are Փ25 × 8 mm.On the other hand, the die core is comprised of four components: the entrance area, compression area, sizing area, and exit area.Natural diamond is the material used for the die core.The wire-drawing die is depicted in Figure 2. The entrance area is typically designed with a circular arc, allowing for easy entry of the wire into the working area.The angle of the circular arc is usually set to 60°.As the wire material progresses through the compression area, the drawing fluid creates a lubricating layer on the surface of the wire material, initiating plastic deformation.In this region, selecting an appropriate working cone angle, denoted as 2α, is crucial for determining the magnitude of the drawing force.The choice of working cone angle should follow the principle that, for a higher compression ratio and harder drawing material, a smaller working cone angle is preferred.For copper wire drawing, a cone angle of 16°-18° is commonly employed.The primary purpose of the sizing area is to achieve the final dimensions of the wire material.Its length is generally 0.5-1.0 times the diameter of the wire.The exit area serves as the final section where the material exits the die hole.It plays a role in protecting the sizing zone and the wire material, with an exit angle ranging from 45° to 65°.The wire-drawing die primarily consists of a die sleeve and a die core.The die sleeve serves to provide force support for the core during the drawing process of high-strength wire, preventing the core from rupturing.The conventional dimensions of the die sleeve are Φ25 × 8 mm.On the other hand, the die core is comprised of four components: the entrance area, compression area, sizing area, and exit area.Natural diamond is the material used for the die core.The wire-drawing die is depicted in Figure 2. The entrance area is typically designed with a circular arc, allowing for easy entry of the wire into the working area.The angle of the circular arc is usually set to 60 • .As the wire material progresses through the compression area, the drawing fluid creates a lubricating layer on the surface of the wire material, initiating plastic deformation.In this region, selecting an appropriate working cone angle, denoted as 2α, is crucial for determining the magnitude of the drawing force.The choice of working cone angle should follow the principle that, for a higher compression ratio and harder drawing material, a smaller working cone angle is preferred.For copper wire drawing, a cone angle of 16 • -18 • is commonly employed.The primary purpose of the sizing area is to achieve the final dimensions of the wire material.Its length is generally 0.5-1.0 times the diameter of the wire.The exit area serves as the final section where the material exits the die hole.It plays a role in protecting the sizing zone and the wire material, with an exit angle ranging from 45 • to 65 • . In the wire-drawing process, the amount of deformation is a crucial factor that significantly affects the wire's performance.Excessive deformation can result in higher deformation resistance, making the process more difficult and increasing the likelihood of wire breakage during drawing.Furthermore, it can also lead to higher hardness of the wire after drawing.On the other hand, insufficient deformation will require a greater number of drawing passes, reducing drawing efficiency.In this specific test, the wire was initially drawn to a size of 0.25 mm.Subsequently, it was further drawn to 0.135 mm using different drawing passes and a single-pass strain.Finally, the wire was drawn to 0.08 mm using different drawing rates, with varying levels of strain as indicated in Table 2.The drawing strain can be calculated using the following formula: l 0 is the length of the wire before drawing (mm); l 1 is the length of the wire after drawing (mm). According to the principle of constant volume, and considering that the cross-section of the test wire is circular, the formula can be deformed to A 0 is the cross-sectional area of the wire before drawing (mm 2 ); A 1 is the cross-sectional area of the wire after drawing (mm 2 ); D 0 is the diameter of the wire before drawing (mm); D 1 is the diameter of the wire after drawing (mm). icromachines 2023, 14, 1635 In the wire-drawing process, the amount of deformation is a crucial fa nificantly affects the wire's performance.Excessive deformation can result formation resistance, making the process more difficult and increasing the wire breakage during drawing.Furthermore, it can also lead to higher ha wire after drawing.On the other hand, insufficient deformation will requ number of drawing passes, reducing drawing efficiency.In this specific test initially drawn to a size of 0.25 mm.Subsequently, it was further drawn to 0. different drawing passes and a single-pass strain.Finally, the wire was draw using different drawing rates, with varying levels of strain as indicated in drawing strain can be calculated using the following formula: l0 is the length of the wire before drawing (mm); l1 is the length of the wire after drawing (mm).According to the principle of constant volume, and considering that the of the test wire is circular, the formula can be deformed to In the heat treatment test, a diameter of 0.08 mm Cu-1Ag alloy wire in the SD-40 continuous heat treatment equipment for heat treatment, heating for electric heating, annealing tube material for quartz glass, an annealing tube length of 1200 mm, a temperature fluctuation range of 2 • C, heat treatment equipment using an angular displacement sensor to control the tension of the wire, and a tension range of 0.01-0.10N were used.The annealing equipment is shown in Figure 3.In the different annealing rate tests, the annealing rate refers to the speed at which the alloy wire passes through the annealing tube and is the rate at which the wire is wound during the annealing heat treatment. Test Method The initial Cu-1 wt%Ag wires with a diameter of 0.25 mm underwent multi-pass drawing tests to achieve alloy wires with a diameter of 0.135 mm.The drawing process followed the test scheme presented in Table 3, with single-pass drawing amounts of 18%, 14%, and 10%, respectively, at a drawing speed of 500 m/min.Sample intercepts were taken at each pass to test the tensile strength, elongation, and resistivity of the alloy wires.Furthermore, metallographic samples were obtained from the final passes to observe the effect of single-pass deformation on the microstructure of Cu-1 wt%Ag bonded wire.For subsequent tests, the wire that underwent a single-pass deformation of 14% was selected.The test procedure is outlined in Table 4.In this test, the Փ0.135 mm alloy wire was subjected to three different drawing speeds: 300 m/min, 500 m/min, and 700 m/min.Each pass involved a single-pass drawing of 15%, resulting in a final alloy wire with a diameter of 0.08 mm.Samples were intercepted from each pass to measure the tensile strength, elongation, and resistivity of the wire.Additionally, metallographic samples were obtained from the final pass to observe the impact of the drawing rate on the microstructure of the Cu-1 wt%Ag bonded wire.In the above test, the wire with a drawing rate of 500 m/min was selected for annealing heat treatment according to the test procedure outlined in Table 5.The 0.08 mm wire was annealed at temperatures of 250 °C, 300 °C, 350 °C, 400 °C, 450 °C, 500 °C, and 550 °C, Test Method The initial Cu-1 wt%Ag wires with a diameter of 0.25 mm underwent multi-pass drawing tests to achieve alloy wires with a diameter of 0.135 mm.The drawing process followed the test scheme presented in Table 3, with single-pass drawing amounts of 18%, 14%, and 10%, respectively, at a drawing speed of 500 m/min.Sample intercepts were taken at each pass to test the tensile strength, elongation, and resistivity of the alloy wires.Furthermore, metallographic samples were obtained from the final passes to observe the effect of single-pass deformation on the microstructure of Cu-1 wt%Ag bonded wire.For subsequent tests, the wire that underwent a single-pass deformation of 14% was selected.The test procedure is outlined in Table 4.In this test, the Φ0.135 mm alloy wire was subjected to three different drawing speeds: 300 m/min, 500 m/min, and 700 m/min.Each pass involved a single-pass drawing of 15%, resulting in a final alloy wire with a diameter of 0.08 mm.Samples were intercepted from each pass to measure the tensile strength, elongation, and resistivity of the wire.Additionally, metallographic samples were obtained from the final pass to observe the impact of the drawing rate on the microstructure of the Cu-1 wt%Ag bonded wire.In the above test, the wire with a drawing rate of 500 m/min was selected for annealing heat treatment according to the test procedure outlined in Table 5.The 0.08 mm wire was annealed at temperatures of 250 • C, 300 • C, 350 • C, 400 • C, 450 • C, 500 • C, and 550 • C, respectively, with an annealing rate of 100 m/min.Samples were collected after the annealing process to measure the tensile strength, elongation, and resistivity of the wire.The mechanical and electrical properties of the Cu-1 wt%Ag bonded wires were studied at different annealing temperatures.Metallographic samples were also taken from the wires that underwent annealing at different temperatures.These samples were observed using scanning electron microscopy (SEM) to analyze the evolution of the material microstructure under changing temperature conditions.The annealing rate test was conducted using a wire material with a diameter of 0.08 mm and a drawing speed of 500 m/min.The test followed the protocol outlined in Table 5, which included annealing rates of 20 m/min, 40 m/min, 60 m/min, 80 m/min, 100 m/min, and 120 m/min.The annealing temperature for the heat treatment test was consistently maintained at 500 • C.After each stage of the test, samples were carefully collected and subjected to thorough analysis, measuring key properties such as tensile strength, elongation, and resistivity.This comprehensive examination allowed for a detailed investigation into how the material's properties evolve at various annealing rates. The SEM samples were prepared as follows: The intercepted samples were put into the molds and poured with epoxy resin to solidify naturally.After that, the sanding operation was performed on the polishing machine using 600#, 800#, 1000#, 1200#, and 2000#SiC sandpaper, and the surface of the sandpaper was rinsed with water stream throughout.After each stage of polishing, the specimen was rotated by 90 • , and each stage of polishing was subject to the absence of the last scratch on the surface.After the above grinding steps, the surface of the specimen was free of obvious scratches.The specimen underwent rough and fine polishing using 1.0 µm and 0.5 µm diamond suspension sprays, respectively.Following the fine polishing, the specimen's surface exhibited a smooth texture without any visible scratches when observed under a microscope.A corrosion solution was prepared by mixing 5 g of copper chloride with 100 mL of ammonia solution.This ready-to-use corrosion solution was applied for a duration of 5-10 s.Subsequently, the specimen was thoroughly cleaned using distilled water and anhydrous ethanol, followed by gentle blowing to ensure complete dryness.The cleaned surface of the specimen was then coated with a layer of gold before being placed under a scanning electron microscope.Various magnifications were used to observe the microstructure of the specimen, allowing for the analysis of variations and distinctions in microstructure at different levels of deformation, drawing rates, and annealing conditions. Effect of Different Deformation on the Microstructure of Cu-1 wt%Ag Wire Figure 4 presents SEM images of a Cu-1 wt%Ag wire with a diameter of 0.25 mm after undergoing three different single-pass deformations during the drawing process.In Figure 4a, the initial microstructure of the wire appears striped, clearly displaying the presence of grain boundaries.Figure 4b-d depict the wire after being drawn with varying single-pass deformations.In each case, the grains are further elongated in the stretching direction, resulting in a fibrous microstructure that is neatly arranged along the drawing direction.Additionally, the interfaces between different phases become straighter, and the overall distribution becomes more uniform and compact.It is worth noting that there are no significant differences in the fibrous grain morphology among the wires drawn using the three different deformation schemes.However, the scheme involving a single-pass deformation of 10% exhibits a more uniform grain size distribution.According to the Cu-1 wt%Ag phase diagram, the solid solubility of Ag in Cu at room temperature is only 0.1 wt%.In the as-cast sample of Cu-1 wt%Ag, a portion of Ag is dispersed within the Cu matrix.During the drawing and deformation process, the grains gradually undergo breakage and refinement, leading to the Ag grains being drawn into a fibrous structure distributed along the axial direction. Micromachines 2023, 14, 1635 7 of 16 the three different deformation schemes.However, the scheme involving a single-pass deformation of 10% exhibits a more uniform grain size distribution.According to the Cu-1 wt%Ag phase diagram, the solid solubility of Ag in Cu at room temperature is only 0.1 wt%.In the as-cast sample of Cu-1 wt%Ag, a portion of Ag is dispersed within the Cu matrix.During the drawing and deformation process, the grains gradually undergo breakage and refinement, leading to the Ag grains being drawn into a fibrous structure distributed along the axial direction. Effect of Different Deformation on the Performance of Cu-1 wt%Ag Wire The drawing process induces lattice distortion, leading to the formation of an elastic stress field around dislocations.Consequently, Ag atoms in the vicinity of dislocations tend to migrate, thereby generating additional stress around dislocations and contributing to the improved strength of the Cu-1 wt%Ag bonded alloy wire.According to metal-electron theory, the disruption of the crystal lattice integrity results in scattering, which is the fundamental cause of resistance in metals.The lattice distortion and dislocation entanglement occurring during the drawing process disrupt the crystal lattice integrity, thereby enhancing the scattering effect on metal electrons.During continuous multi-pass drawing, interfacial scattering primarily influences the resistivity of the Cu-1 wt%Ag wire.This is because the average diameter of the fibrous grains decreases, leading to an increase in material resistivity.Additionally, as the strain increases, the solid solution of Ag atoms within the copper matrix also increases.This causes certain Ag atoms to dissolve into the copper matrix, weakening the scattering effect at the phase interface.Consequently, the resistivity of the wire material decreases during the drawing process. Based on the variation curves of mechanical properties in Figure 5a-d, the tensile strength of the Cu-1 wt%Ag wire after drawing is similar for different deformation levels.The tensile strengths are approximately 784.63 MPa, 783.9 MPa, and 782.58 MPa, respectively.Overall, the tensile strengths of all three samples increase with drawing.In the case of a single deformation of 18%, the material's tensile strength reaches 777.86 MPa at a Effect of Different Deformation on the Performance of Cu-1 wt%Ag Wire The drawing process induces lattice distortion, leading to the formation of an elastic stress field around dislocations.Consequently, Ag atoms in the vicinity of dislocations tend to migrate, thereby generating additional stress around dislocations and contributing to the improved strength of the Cu-1 wt%Ag bonded alloy wire.According to metalelectron theory, the disruption of the crystal lattice integrity results in scattering, which is the fundamental cause of resistance in metals.The lattice distortion and dislocation entanglement occurring during the drawing process disrupt the crystal lattice integrity, thereby enhancing the scattering effect on metal electrons.During continuous multi-pass drawing, interfacial scattering primarily influences the resistivity of the Cu-1 wt%Ag wire.This is because the average diameter of the fibrous grains decreases, leading to an increase in material resistivity.Additionally, as the strain increases, the solid solution of Ag atoms within the copper matrix also increases.This causes certain Ag atoms to dissolve into the copper matrix, weakening the scattering effect at the phase interface.Consequently, the resistivity of the wire material decreases during the drawing process. Based on the variation curves of mechanical properties in Figure 5a-d, the tensile strength of the Cu-1 wt%Ag wire after drawing is similar for different deformation levels.The tensile strengths are approximately 784.63 MPa, 783.9 MPa, and 782.58 MPa, respectively.Overall, the tensile strengths of all three samples increase with drawing.In the case of a single deformation of 18%, the material's tensile strength reaches 777.86 MPa at a strain of around 0.7.After this point, the tensile strength remains stable, while the elongation increases from 1.8% to 2.022%.For a single deformation of 14%, the material's tensile strength reaches 762.3 MPa at a strain of about 0.7.It then stabilizes and further increases to 783.9 MPa at a strain of 1.23.The elongation also rises from 1.8% to 2.111%.With a single deformation of 10%, the tensile strength of the Cu-1 wt%Ag wire reaches a maximum value of 803 MPa at a strain of 1.1, representing a 20% increase compared to the initial wire.Further deformation reduces the tensile strength to 783.9 MPa, while the elongation increases from 1.8% to 2.158%.This results in an approximately 17% increase in tensile strength compared to the initial sample.During multiple drawing passes, the wire structure becomes fibrillated, increasing the deformation stress and filament strength.The solid dissolution of Ag atoms in the matrix causes localized distortion points.This distortion increases the resistance to crystal slip, making deformation more difficult.As plastic deformation progresses, internal defects in the wire multiply, dislocations rapidly proliferate, and dislocation density continuously increases.The grains elongate and deform along the drawing direction, leading to a continuous increase in tensile strength. Micromachines 2023, 14, 1635 8 of 16 strain of around 0.7.After this point, the tensile strength remains stable, while the elongation increases from 1.8% to 2.022%.For a single deformation of 14%, the material's tensile strength reaches 762.3 MPa at a strain of about 0.7.It then stabilizes and further increases to 783.9 MPa at a strain of 1.23.The elongation also rises from 1.8% to 2.111%.With a single deformation of 10%, the tensile strength of the Cu-1 wt%Ag wire reaches a maximum value of 803 MPa at a strain of 1.1, representing a 20% increase compared to the initial wire.Further deformation reduces the tensile strength to 783.9 MPa, while the elongation increases from 1.8% to 2.158%.This results in an approximately 17% increase in tensile strength compared to the initial sample.During multiple drawing passes, the wire structure becomes fibrillated, increasing the deformation stress and filament strength.The solid dissolution of Ag atoms in the matrix causes localized distortion points.This distortion increases the resistance to crystal slip, making deformation more difficult.As plastic deformation progresses, internal defects in the wire multiply, dislocations rapidly proliferate, and dislocation density continuously increases.The grains elongate and deform along the drawing direction, leading to a continuous increase in tensile strength.As strain increases, the interface between the eutectic and Cu matrix absorbs the dislocation substructure.Subsequently, the dislocation migrates to the interface area between the Cu solid solution and eutectic fiber.The overall dislocation density decreases, and the level of interfacial strengthening reaches a higher point.As a result, the material properties change more gradually. Figure 5e,f illustrate the resistance change curves of Cu-1 wt%Ag wire at different deformation levels.In the case of 18% deformation, the resistivity exhibited a relatively smooth pattern during the drawing process.However, a significant decrease in resistivity (to 1.882 × 10 −8 Ω•m) occurred at a strain of approximately 0.7, followed by a sustained growth trend.The resistivity reached its minimum values at a strain of 0.28 and 0.4 (1.886 × 10 −8 Ω•m and 1.899 × 10 −8 Ω•m, respectively) when the deformation level was reduced to 14% and 10%.Subsequently, the resistivity slowly increased and remained stable. Effect of Different Drawing Rates on the Microstructure of Cu-1 wt%Ag Wire Comparison of Figures 4 and 6 shows that the microstructure of the longitudinal section of the alloy wire tends to be increasingly fibrous with the continuation of drawing, although the effect of different drawing rates on the microstructure of the material is not obvious, but due to the continuous deformation of multiple passes, the grain refinement and torsion deformation are extremely serious, and the spacing of the fibrous tissue layers decreases gradually, and Figure 6 exhibits an obvious fiber-like structure.The fibrous organization becomes increasingly compact because, when the fibrillation of the material is completed, the increase in strain does not change the morphology of the fibers but reduces the fiber grain diameter. 9 of 16 As strain increases, the interface between the eutectic and Cu matrix absorbs the dislocation substructure.Subsequently, the dislocation migrates to the interface area between the Cu solid solution and eutectic fiber.The overall dislocation density decreases, and the level of interfacial strengthening reaches a higher point.As a result, the material properties change more gradually. Figure 5e,f illustrate the resistance change curves of Cu-1 wt%Ag wire at different deformation levels.In the case of 18% deformation, the resistivity exhibited a relatively smooth pattern during the drawing process.However, a significant decrease in resistivity (to 1.882 × 10 −8 Ω•m) occurred at a strain of approximately 0.7, followed by a sustained growth trend.The resistivity reached its minimum values at a strain of 0.28 and 0.4 (1.886 × 10 −8 Ω•m and 1.899 × 10 −8 Ω•m, respectively) when the deformation level was reduced to 14% and 10%.Subsequently, the resistivity slowly increased and remained stable. Effect of Different Drawing Rates on the Microstructure of Cu-1 wt%Ag Wire Comparison of Figures 4 and 6 shows that the microstructure of the longitudinal section of the alloy wire tends to be increasingly fibrous with the continuation of drawing, although the effect of different drawing rates on the microstructure of the material is not obvious, but due to the continuous deformation of multiple passes, the grain refinement and torsion deformation are extremely serious, and the spacing of the fibrous tissue layers decreases gradually, and Figure 6 exhibits an obvious fiber-like structure.The fibrous organization becomes increasingly compact because, when the fibrillation of the material is completed, the increase in strain does not change the morphology of the fibers but reduces the fiber grain diameter. Effect of Different Drawing Rates on the Properties of Cu-1 wt%Ag Wire The tensile strength of Cu-1 wt%Ag wire increased from 783.9 MPa to 806 MPa, 820.1 MPa, and 817.1 MPa after drawing at rates of 300 m/min, 500 m/min, and 700 m/min, respectively.Figure 7a,b depict the changes in tensile strength of Cu-1 wt%Ag wire during the drawing process at different rates.The variation in tensile strength was similar for different drawing rates: it exhibited an increasing trend followed by a plateau as the strain increased, reaching its peak at a strain of 0.6 in all cases (815.3MPa, 839 MPa, and 828.85 MPa). Effect of Different Drawing Rates on the Properties of Cu-1 wt%Ag Wire The tensile strength of Cu-1 wt%Ag wire increased from 783.9 MPa to 806 MPa, 820.1 MPa, and 817.1 MPa after drawing at rates of 300 m/min, 500 m/min, and 700 m/min, respectively.Figure 7a,b depict the changes in tensile strength of Cu-1 wt%Ag wire during the drawing process at different rates.The variation in tensile strength was similar for different drawing rates: it exhibited an increasing trend followed by a plateau as the strain increased, reaching its peak at a strain of 0.6 in all cases (815.3MPa, 839 MPa, and 828.85 MPa).During the wire-drawing process, as plastic deformation continues, the crystal lattice structure be-comes distorted.This distortion leads to the generation of numerous dislocations, vacancies, and other defects at the boundaries between individual grains.These defects have a scattering effect on the free electrons present in the material, causing an increase in resistivity.Additionally, the Cu-1 wt%Ag material contains second-phase Ag particles.These Ag particles interact with the dislocations generated during plastic deformation.Such interactions further contribute to an increase in resistivity within the wires.The presence of these second-phase particles introduces additional obstacles for the movement of free electrons, leading to an overall increase in resistivity.Huadong Fu et al. pointed out that the refinement of fibers due to plastic deformation leads to an increase in interfaces with higher tensile strain, resulting in a decrease in electrical conductivity [21]. Micromachines 2023, 14, 1635 10 of 16 electrons present in the material, causing an increase in resistivity.Additionally, the Cu-1 wt%Ag material contains second-phase Ag particles.These Ag particles interact with the dislocations generated during plastic deformation.Such interactions further contribute to an increase in resistivity within the wires.The presence of these second-phase particles introduces additional obstacles for the movement of free electrons, leading to an overall increase in resistivity.Huadong Fu et al. pointed out that the refinement of fibers due to plastic deformation leads to an increase in interfaces with higher tensile strain, resulting in a decrease in electrical conductivity [21].After being drawn at three different rates, the Cu-1Ag wire exhibits minimal changes in tensile strength.At this stage, significant work hardening occurs within the Cu-1Ag wire.The presence of Ag increases the concentration limits of dislocations and subgrain boundaries in the Cu solid solution, resulting in a higher degree of hardening compared to pure copper wire [22].In the initial stages of drawing, cold deformation results in an increased dislocation density.Work hardening and dislocation strengthening contribute to a rise in tensile strength.As the deformation progresses to a certain extent, the movement of dislocations becomes more difficult, leading to saturation of dislocation density and minimal changes in grain size.Consequently, the effect on tensile strength becomes less pronounced.Zhu Xiao et al. showed that, with the continuation of the drawing, in the stage of large deformation, the tensile strength of Cu-Ag alloys will slowly increase [23].As the strain increases, the solid solubility of Ag within the Cu matrix also increases.When the strain reaches a certain threshold, a portion of the secondary Ag phase undergoes back dissolution.This results in a decrease in the area of the Ag/Cu phase boundary and subsequently reduces the scattering effect on electrons.Consequently, the wire resistivity decreases when the strain reaches 0.75. The mechanical properties, electrical properties and microstructure of the alloy wires after drawing with different deformation behaviors have certain trends, which will be compared with the data of the heat-treated alloy wires and the analysis of the organization After being drawn at three different rates, the Cu-1Ag wire exhibits minimal changes in tensile strength.At this stage, significant work hardening occurs within the Cu-1Ag wire.The presence of Ag increases the concentration limits of dislocations and subgrain boundaries in the Cu solid solution, resulting in a higher degree of hardening compared to pure copper wire [22].In the initial stages of drawing, cold deformation results in an increased dislocation density.Work hardening and dislocation strengthening contribute to a rise in tensile strength.As the deformation progresses to a certain extent, the movement of dislocations becomes more difficult, leading to saturation of dislocation density and minimal changes in grain size.Consequently, the effect on tensile strength becomes less pronounced.Zhu Xiao et al. showed that, with the continuation of the drawing, in the stage of large deformation, the tensile strength of Cu-Ag alloys will slowly increase [23].As the strain increases, the solid solubility of Ag within the Cu matrix also increases.When the strain reaches a certain threshold, a portion of the secondary Ag phase undergoes back dissolution.This results in a decrease in the area of the Ag/Cu phase boundary and subsequently reduces the scattering effect on electrons.Consequently, the wire resistivity decreases when the strain reaches 0.75. The mechanical properties, electrical properties and microstructure of the alloy wires after drawing with different deformation behaviors have certain trends, which will be compared with the data of the heat-treated alloy wires and the analysis of the organization evolution law to find the process parameters for the preparation of high-performance Cu-1 wt%Ag wires; in addition, the study of the cold deformation is of great significance for the prediction of future changes in the properties of alloy wires under the same drawing conditions. 3.5.Effect of Different Heat Treatment Temperatures on the Microstructure of Cu-1 wt%Ag Wire Higher heat treatment temperature helps grain growth, while lower temperature is conducive to grain refinement.The annealing temperature range was set to 250-550 • C in order to explore the influence of different temperatures on the microstructure and properties of alloy wire in the heat treatment process, to ensure that the recrystallization phenomenon of the alloy wire can fully occur, and to find the temperature range of the occurrence of the recrystallization phenomenon. Figure 8a illustrates the microstructure of unannealed Cu-1 wt%Ag wires.At this stage, the Cu-1 wt%Ag wire exhibits a fibrous structure, with the diameter of the fibers being noticeably smaller at the wire's edges compared to its center.This variation in diameter is primarily caused by the uneven distribution of forces during the drawing process.The edge portion of the wire comes into contact with the die first and experiences both shear deformation and extrusion from the die simultaneously.As a result, the deformation in the edge region is more severe compared to other areas of the wire. Effect of Different Heat Treatment Temperatures on the Properties of Cu-1 wt%Ag Wire Figure 9a shows the variation of mechanical properties of 0.08 mm diameter Cu-1 wt%Ag bonded wires at different heat treatment temperatures.The unannealed Cu-1 wt%Ag bonded wire exhibited high tensile strength of 820.1 MPa and low elongation of 1.94%.Following annealing at 250 °C, the Cu-1 wt%Ag wire displayed a slight change in mechanical properties, with a tensile strength of 798.14 MPa and elongation of 2.4%.As the annealing temperature increased from 300 °C to 350 °C, significant changes in the properties of the Cu-1 wt%Ag wire were observed compared to the initial sample.These changes included a decrease in tensile strength to 723 MPa and an increase in elongation to 2.84%.Further increasing the annealing temperature, the tensile strength of the Cu-1 Figure 8b presents the microstructure of Cu-1 wt%Ag wire in the annealed state at 400 • C. The addition of Ag raises the recrystallization temperature, which causes the Cu-1 wt%Ag wire to begin undergoing reversion.As a result, the dislocation density decreases.However, due to the relatively low heat treatment temperature, a significant number of deformed regions still exist within the wire.These regions exhibit a fibrous structure aligned along the drawing direction. Figure 8c displays the microstructure of Cu-1 wt%Ag wire after annealing at 500 • C. With increasing annealing temperature, the wire exhibits enhanced recovery, resulting in the release of distortion energy and the precipitation of a secondary Ag phase in close proximity to the fiber phase.Simultaneously, the elevated temperature induces intensified movement of dislocations, vacancies, and other lattice defects.These defects are gradually absorbed by a significant number of internal stresses, leading to their gradual release.Consequently, the deformation structure gradually disappears, initiating the process of recrystallization. Figure 8d illustrates the microstructure of Cu-1 wt%Ag wire in the annealed state at 550 • C. Subsequent to annealing at this temperature, there is an enhanced propensity for the release of distortion energy, leading to intensified movement of dislocations and vacancies with a further reduction in their density.During this stage, recrystallization initiates in the Cu-1 wt%Ag wire, resulting in gradual grain growth.The microstructure exhibits diminished visibility of the deformation organization, and some coarsened grains become apparent.As the level of recrystallization progresses, phase interfaces undergo complete activation, and there is pronounced migration of grain boundaries.Consequently, the fiber structure tends to vanish and redistribute, facilitating the growth of grains.In the heat treatment test at different temperatures, the annealing temperature below 250 °C, the microstructure and properties of alloy wires have no obvious changes, but the value of tensile strength and resistivity has a tendency to decline, and the value of the elongation has a tendency to increase, although the annealing temperature is low, but the alloy wires also have a certain performance impact; When the annealing temperature is 250-500 °C, the performance of the alloy wire has the same change trend as that of low annealing temperature, and the change is more obvious.This suggests that lower annealing temperatures may cause a slight change in properties, while annealing at temperatures in the range of 250-500 °C may change the properties of the alloy wires more significantly. Effect of Different Annealing Rates on the Properties of Cu-1 wt%Ag Wires Lower annealing rates usually increase the toughness and ductility of the material but may decrease the hardness and strength.Higher annealing rates may increase hardness and strength but may decrease toughness.The annealing rate is set to control the time of annealing, to study the performance of alloy wires at different annealing rates, and to find the annealing rate parameter for the excellent performance of alloy wires in conjunction with experimental data. Figure 10a illustrates the impact of different heat treatment rates on the mechanical properties of Cu-1 wt%Ag bonded wires.For annealing rates ranging from 20 to 120 m/min, the tensile strength of the Cu-1 wt%Ag wires gradually increases from 348.3 MPa In the Cu-1 wt%Ag wire, after undergoing significant deformation, the cross-sectional area experiences a rapid reduction.This reduction leads to an increase in the density of grain boundaries per unit cross-sectional area.Additionally, an abundance of vacancies, dislocations, and other defects accumulate at the grain boundaries, forming an isolation layer rich in crystalline defects.As a result, the electron scattering effect is enhanced within the wire. In Figure 9b, the electrical properties of a Cu-1 wt%Ag bonding wire with a diameter of 0.08 mm are shown at different heat treatment temperatures.The unannealed Cu-1 wt%Ag wire exhibits a higher resistivity of 1.931 × 10 −8 Ω•m.This increased resistivity is attributed to the presence of defects such as dislocations and vacancies within the material.When the wire is subjected to heat treatment below 250 • C, the resistivity decreases slightly to 1.904 × 10 −8 Ω•m.At this temperature, the microstructure of the wire does not undergo significant changes compared to its unannealed state.As the annealing temperature is increased to 300 • C, intense movement of point defects, such as vacancies and interstitial atoms, takes place.Vacancies migrate towards the grain boundaries and neutralize with interstitial atoms.This results in a decrease in the density of point defects, leading to a decrease in resistivity to 1.840 × 10 −8 Ω•m.Between 300 • C and 450 • C, the dislocation density gradually decreases, and the fibrous structure begins to disappear.This reduction in dislocation density and fibrous structure contributes to a further decrease in resistivity.The resistivity of the wire decreases to 1.78 × 10 −8 Ω•m within this temperature range.As the annealing temperature continues to increase, the reversion effect becomes more prominent, leading to the disappearance of the fibrous structure in large quantities.Recrystallization of the material initiates, albeit at a slow rate.As a result, the ability of the material to scatter electrons decreases.The resistivity of the wire continues to decrease, and after annealing at 550 • C it reaches 1.718 × 10 −8 Ω•m. In the heat treatment test at different temperatures, the annealing temperature below 250 • C, the microstructure and properties of alloy wires have no obvious changes, but the value of tensile strength and resistivity has a tendency to decline, and the value of the elongation has a tendency to increase, although the annealing temperature is low, but the alloy wires also have a certain performance impact; When the annealing temperature is 250-500 • C, the performance of the alloy wire has the same change trend as that of low annealing temperature, and the change is more obvious.This suggests that lower annealing temperatures may cause a slight change in properties, while annealing at temperatures in the range of 250-500 • C may change the properties of the alloy wires more significantly. Effect of Different Annealing Rates on the Properties of Cu-1 wt%Ag Wires Lower annealing rates usually increase the toughness and ductility of the material but may decrease the hardness and strength.Higher annealing rates may increase hardness and strength but may decrease toughness.The annealing rate is set to control the time of annealing, to study the performance of alloy wires at different annealing rates, and to find the annealing rate parameter for the excellent performance of alloy wires in conjunction with experimental data. Figure 10a illustrates the impact of different heat treatment rates on the mechanical properties of Cu-1 wt%Ag bonded wires.For annealing rates ranging from 20 to 120 m/min, the tensile strength of the Cu-1 wt%Ag wires gradually increases from 348.3 MPa to 392.7 MPa.This increase in tensile strength is relatively slow within this annealing rate range.At annealing rates between 80 and 120 m/min, the wires are in the recovery stage.During this stage, the internal work hardening caused by the drawing process is gradually eliminated, and the degree of recovery becomes more pronounced as the annealing rate decreases.As a result, the elongation of the wires increases from 14% (at 120 m/min) to 16.3% (at 80 m/min).When the annealing rate is 60 m/min, the material reaches its peak elongation of 17.28%.At this point, the material has fully reverted internally and starts to recrystallize.However, at an annealing rate of 20 m/min, the elongation decreases to 12.1%.This reduction in elongation is attributed to the slow annealing rate, which leads to severe grain coarsening in the material. Figure 10b presents the changes in resistivity of Cu-1 wt%Ag bonded wire for different heat treatment rates.With a decrease in annealing rate, the overall resistivity of Cu-1 wt%Ag wire tends to decrease.At an annealing rate of 120 m/min, the resistivity of Cu-1 wt%Ag wires is relatively high at 1.728 × 10 −8 Ω•m.This is because the fast annealing rate does not allow sufficient time for the wires to fully recover.As the annealing rate decreases, the resistivity of the material gradually reduces.At an annealing rate of 20 m/min, the resistivity reaches a lower value of 1.632 × 10 −8 Ω•m.This slower annealing rate allows for improved recovery and leads to a decrease in resistivity. Figure 10b presents the changes in resistivity of Cu-1 wt%Ag bonded wire for di ent heat treatment rates.With a decrease in annealing rate, the overall resistivity of C wt%Ag wire tends to decrease.At an annealing rate of 120 m/min, the resistivity of C wt%Ag wires is relatively high at 1.728 × 10 −8 Ω•m.This is because the fast annealing does not allow sufficient time for the wires to fully recover.As the annealing rate creases, the resistivity of the material gradually reduces.At an annealing rate of 20 m/ the resistivity reaches a lower value of 1.632 × 10 −8 Ω•m.This slower annealing rate al for improved recovery and leads to a decrease in resistivity. Conclusions (1) After a single-pass drawing with deformation amounts of 18%, 14%, and 10%, the grains gradually transformed into a fibrous morphology, and Ag fiber grain organization was observed within the Cu-1 wt%Ag wires.The addition of the Ag element increased the deformation resistance of the wires.Specifically, the tensile strength of the Cu-1 wt%Ag wires increased from 670 MPa to 784.63 MPa, 783.9 MPa, and 782.58 MPa for deformation amounts of 18%, 14%, and 10%, respectively.Additionally, the resistivity increased from 1.922 × 10 Figure 6 . Figure 6.Longitudinal microstructure of Cu−1 wt%Ag wires after different drawing rate tests. Figure 6 . Figure 6.Longitudinal microstructure of Cu−1 wt%Ag wires after different drawing rate tests. Figure Figure 9a shows the variation of mechanical properties of 0.08 mm diameter Cu-1 wt%Ag bonded wires at different heat treatment temperatures.The unannealed Cu-1 wt%Ag bonded wire exhibited high tensile strength of 820.1 MPa and low elongation of 1.94%.Following annealing at 250 • C, the Cu-1 wt%Ag wire displayed a slight change in mechanical properties, with a tensile strength of 798.14 MPa and elongation of 2.4%.As the annealing temperature increased from 300 • C to 350 • C, significant changes in the properties of the Cu-1 wt%Ag wire were observed compared to the initial sample.These changes included a decrease in tensile strength to 723 MPa and an increase in elongation to 2.84%.Further increasing the annealing temperature, the tensile strength of the Cu-1 wt%Ag wire decreased by 197.1 MPa, reaching 530 MPa after annealing at 450 • C. Additionally, the elongation increased to 4.3%.Continuing to raise the annealing temperature resulted in further property changes in the Cu-1 wt%Ag wire.Compared to the sample annealed at 450 • C, the tensile strength decreased by 210 MPa to 319.5 MPa when annealed at 550 • C, while the elongation significantly increased to 20.2%. Figure 9 . Figure 9. Mechanical and electrical properties of Cu−1 wt%Ag wires at different annealing temperatures ((a) mechanical properties; (b) electrical properties). Figure 9 . Figure 9. Mechanical and electrical properties of Cu−1 wt%Ag wires at different annealing temperatures ((a) mechanical properties; (b) electrical properties). ( 1 ) After a single-pass drawing with deformation amounts of 18%, 14%, and 10% grains gradually transformed into a fibrous morphology, and Ag fiber grain org zation was observed within the Cu-1 wt%Ag wires.The addition of the Ag elem increased the deformation resistance of the wires.Specifically, the tensile streng the Cu-1 wt%Ag wires increased from 670 MPa to 784.63 MPa, 783.9 MPa, and 78 MPa for deformation amounts of 18%, 14%, and 10%, respectively.Additionally resistivity increased from 1.922 × 10 −8 Ω•m to 1.959 × 10 −8 Ω•m, 1.923 × 10 −8 Ω•m, 1.923 × 10 −8 Ω•m, respectively.Overall, the wires exhibit superior performance w subjected to a single-pass deformation of 14%.(2) Following the experiments conducted with various drawing rates, the te strength of Cu-1 wt%Ag wires increased from 783.9 MPa to 806 MPa, 820.1 MPa, 817.1 MPa, respectively.Meanwhile, the electrical resistivity rose from 1.923 × Ω•m to 1.927 × 10 −8 Ω•m, 1.931 × 10 −8 Ω•m, and 1.954 × 10 −8 Ω•m, respectively.Nota when employing the drawing rate of 300 m/min, the resulting wires consistently hibited low resistivity and showcased smooth performance transitions throug the drawing process.(3) The resistivity of the Cu-1 wt%Ag wires exhibited minimal changes prior to ann ing temperatures of 250 °C.At an annealing temperature of 300 °C, the Cu-1 wt% bonded wires commenced reversion, with a more pronounced effect observed a 350 °C.At 450 °C, the resistivity of the Cu-1 wt%Ag wire decreased to 1.78 × 10 −8 Ω Further increasing the temperature to 550 °C resulted in a substantial reductio the fiber phase, accompanied by gradual recrystallization of the material.Howe the recrystallization rate was relatively slow, leading to a continued decrease in material's overall ability to scatter electrons.Consequently, the resistivity stea declined, reaching a final value of 1.718 × 10 −8 Ω•m.The Cu-1 wt%Ag wires exhib enhanced performance following annealing between 500 °C and 550 °C.(4) Within the annealing rate range of 20 to 120 m/min, the tensile strength of C wt%Ag wires exhibited a gradual increase, ranging from 348.3 MPa to 392.7 MP an annealing rate of 60 m/min, the material underwent complete internal rever Figure 10 . Figure 10.Mechanical and electrical properties of Cu−1 wt%Ag wires at different annealing rates ((a) mechanical properties; (b) electrical properties). −8 Ω•m to 1.959 × 10 −8 Ω•m, 1.923 × 10 −8 Ω•m, and 1.923 × 10 −8 Ω•m, respectively.Overall, the wires exhibit superior performance when subjected to a single-pass deformation of 14%.(2) Following the experiments conducted with various drawing rates, the tensile strength of Cu-1 wt%Ag wires increased from 783.9 MPa to 806 MPa, 820.1 MPa, and 817.1 MPa, respectively.Meanwhile, the electrical resistivity rose from 1.923 × 10 −8 Ω•m to 1.927 × 10 −8 Ω•m, 1.931 × 10 −8 Ω•m, and 1.954 × 10 −8 Ω•m, respectively.Notably, when employing the drawing rate of 300 m/min, the resulting wires consistently exhibited low resistivity and showcased smooth performance transitions throughout the drawing process.(3) The resistivity of the Cu-1 wt%Ag wires exhibited minimal changes prior to annealing temperatures of 250 • C. At an annealing temperature of 300 • C, the Cu-1 wt%Ag bonded wires commenced reversion, with a more pronounced effect observed above 350 • C. At 450 • C, the resistivity of the Cu-1 wt%Ag wire decreased to 1.78 × 10 −8 Ω•m.Further increasing the temperature to 550 • C resulted in a substantial reduction in the fiber phase, accompanied by gradual recrystallization of the material.However, the recrystallization rate was relatively slow, leading to a continued decrease in the material's overall ability to scatter electrons.Consequently, the resistivity steadily declined, reaching a final value of 1.718 × 10 −8 Ω•m.The Cu-1 wt%Ag wires exhibited enhanced performance following annealing between 500 • C and 550 • C. (4) Within the annealing rate range of 20 to 120 m/min, the tensile strength of Cu-1 wt%Ag wires exhibited a gradual increase, ranging from 348.3 MPa to 392.7 MPa.At an annealing rate of 60 m/min, the material underwent complete internal reversion and initiated recrystallization, resulting in a peak elongation of 17.28%.With a decrease in the annealing rate to 20 m/min, the recrystallized grains began to grow, leading to a significant reduction in elongation by 12.1% compared to the 60 m/min annealing rate.Concurrently, the resistivity of the Cu-1 wt%Ag wire increased from 1.632 × 10 −8 Ω•m (at 20 m/min) to 1.728 × 10 −8 Ω•m (at 120 m/min).Based on these findings, the optimal annealing rate for Cu-1 wt%Ag wire falls within the range of 60-80 m/min. Table 1 . Mainly used equipment. Table 2 . Variation of wire strain in the drawing tests. Table 3 . Drawing tests with different deformation. Table 4 . Different drawing rate tests. Table 3 . Drawing tests with different deformation. Table 4 . Different drawing rate tests. Table 5 . Different annealing process tests.	v3-fos-license
2021-07-27T00:05:15.200Z	2021-05-29T00:00:00.000	236412207	{ "extfieldsofstudy": [ "Business" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2071-1050/13/11/6135/pdf", "pdf_hash": "6598bdb23e2eccfee65b8d88ac68275c5b706dd6", "pdf_src": "MergedPDFExtraction", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45321", "s2fieldsofstudy": [ "Business" ], "sha1": "3afbafe07359ce581f82f8d1f5a57c5e10152e2d", "year": 2021 }	pes2o/s2orc	Enacting Transitions—The Combined Effect of Multiple Niches in Whole System Reconﬁguration : The environmental and social issues caused by agricultural and food distribution practices call for a profound reconﬁguration of the agri-food system. This paper is aimed at contributing to a better understanding of the way such a reconﬁguration may be fostered. Building on recent developments of transition studies that analyze whole system reconﬁgurations, it proposes a pragmatist, whole system approach to examine the socio-political dimension of sustainability transitions. Based on the ethnographic and longitudinal study of a unique case of (territorial) agroecological transition in France, it identiﬁes the mechanisms involved in a transition and the way actors enacted them. It characterizes required prior, incremental system changes, and stresses the role of multiple niches that inﬂuence simultaneously the various components of the agri-food system. From an action-oriented perspective, these results suggest that transitions may be fostered by: (1) supporting the diffusion of an alternative technological paradigm within the regime that niches may be congruent with; (2) stimulating the development of a diversity of radical innovations related to the various dimensions of the agri-food system and fostering their interactions with the regime; and (3) moving from a technology-driven approach of innovation towards an emphasis on organizational innovations that foster the rebalancing of power relations. consumption, research, development, education, public policies and regulation). It is performed through the prism of the three variables presented above: (1) these actors’ technical, social, marketing, and/or organizational practices; (2) their representations; (3) their interactions. These variables allow identifying groups of individuals and organizations that share common sets of representations and whose practices are aligned; they allow characterizing distinct sociotechnical conﬁgurations (Figure 1). The regime is the dominant conﬁguration, i.e., the largest one. Niches are alternative conﬁgurations, composed of actors whose practices and representations radically diverge from the regime. Introduction In order to tackle the environmental, social, and health issues caused by agricultural and food distribution practices, the agri-food system urgently needs to be profoundly reconfigured to shift towards sustainability. This paper is aimed at contributing to a better understanding of the way such a shift may be fostered by examining the way actors may act on socio-technical developments and eventually trigger a sustainability transition. The multi-level perspective (MLP) is a fundamental heuristic underpinning transition studies, which considers that societal functions such as food provision are fulfilled by systems made of co-evolving social and technical elements. The MLP conceptualizes transitions as system innovations, and it distinguishes three analytical levels: the regime, corresponding to the set of rules supporting the dominant socio-technical system, which maintains actors on a given technological trajectory and exerts strong lock-in effects; niches, small networks of actors developing radical innovations and testing new rules around these innovations; and the landscape, the exogenous socio-technical context actors cannot directly influence [1][2][3][4]. It suggests that various radical innovations are developed in niches, until one of them appears as the most promising, and that it is the interaction between this niche, the regime, and the landscape that possibly triggers a socio-technical transition. Over the past fifteen years, transition studies have provided inputs to policy reflection in numerous countries and industries (e.g., [5,6], and the incorporation of insights from a variety of disciplines and theoretical backgrounds into this heuristic has allowed for the refining of the several dimensions involved in transitions. However, due to the complexity of the processes involved, understanding "why and how some niches set in motion transformational change [ . . . ] while others fail" [7] remains an unresolved challenge. To contribute to this understanding, this paper asks: how can actors at a local level have a transformative influence on socio-technical developments? Transition studies have been criticized for focusing on technological dimensions [8] and overlooking the role of actors in transition dynamics [9,10]. Numerous contributions have then been made to shed light on agency (i.e., actors' capacity to act). Geels and Schot ([3], p. 414), building on Giddens' structuring theory [11], argue that actors may both reproduce and change the rules and invite analysts "to zoom in on actors". This invitation has paved the way for several refinements: Elzen et al. [12] analyze how normative pressure exerted by outsiders can influence regimes; Geels and Verhees [13] show how framing struggles between incumbent and alternative discourses performed on public stages impact cultural legitimacy; and other authors analyze the trajectories and roles of specific actors, including farmers (e.g., [14]), intermediary actors (e.g., [15]), and civilsociety (e.g., [16]). Various typologies have been proposed to analyze the role of actors in transitions (for a review, see [17]), and of their evolving interests, values, rationales, and strategies [18][19][20]. Answering calls to account for the inherent political dimension of socio-technical dynamics [21,22], Grin et al. [23] demonstrate that transitions imply change in power relations. Several conceptual frameworks have since been mobilized or created to analyze shifts in power relations [24,25], including by Avelino and Wittmayer [26], who claim analyzes should focus on levels of dependencies between actors, and Rossi et al. [27], who suggest transitions imply more nuanced power relations between the various actors of the socio-technical system. Furthermore, Wittmayer et al. [28] show that fundamental changes in the roles of actors and their inter-relations are a key feature of transition processes. Those various contributions fruitfully identify some specific dynamics and levers through which actors manage to influence some regime rules. Nevertheless, in a perspective of operationalizing the transition to sustainable food systems, a comprehensive review of the mechanisms involved in regime shifts and of the way actors manage to operate these mechanisms is still needed [29]. Recent research has addressed this gap by operating the complementary movement of 'zooming out', enlarging the focus to whole system reconfigurations. Geels's [30] longitudinal analysis of the UK passenger mobility system points out the importance of examining multiple regime dynamics (auto-mobility, train, bus, and cycling)and multiple niche-innovations simultaneously to give full account of the diverse interacting change mechanisms involved in socio-technical developments. McMeekin et al. [31] develop an approach for whole system analysis and produce a typology of change mechanisms based on the case of the low-carbon transformation of the UK electricity system. However, as stressed by these authors, the trade-off of this 'zooming out' strategy is the difficulty to conduct a sharp analysis of social change processes; the socio-political dimension of whole system reconfigurations has yet to be fully explored. This paper is aimed at contributing to such exploration by looking at a unique case of agroecological transition in France, in the Drôme valley, to understand how actors can foster whole system reconfigurations. It is organized as follows. Section 2 develops an analytical framework to tackle agency in a whole system approach of sustainability transitions. Section 3 details the methodology designed to apply the framework to the Drôme valley case, as well as the analyzed material. Section 4 presents the socio-technical evolution of the agri-food system of the Drôme valley from 1970 to 2015 and evidence of whole system reconfiguration. Section 5 stresses the transition mechanisms that have allowed some actors to bring about a regime shift in this region. The concluding section synthesizes the paper's contributions and discusses their implications for an operationalization of sustainability transitions. A Pragmatist, Whole System Approach to Tackle Agency in Transitions In line with Geels's [30] and McMeekin et al.'s [31] work, the present whole system approach reconnects with the MLP's systemic foundations to address: (1) multiple regime dynamics, meaning that, for the agri-food regime, all food chains and all production systems are included in the analysis; (2) multiple niche-innovations; and (3) multiple landscape dynamics that influence the agri-food regime. This paper builds on Geels and Schot's definition of niche and regime as two analytical levels, corresponding to socio-technical configurations (i.e., heterogenous arrangements of coevolving social and technical elements) that differ in size and stability [3]. The regime is the dominant socio-technical configuration, stable and underpinned by a set of rules. Niches are smaller and less stable socio-technical configurations whose rules radically differ from the ones at stake in the regime. These definitions leave much space for interpretation. Depending on scholars' analytical choices, a technological evolution might be interpreted as an incremental change or as a regime shift [9]. The need for reflexivity in the empirical use of the MLP [32] is not always met, which has produced a number of studies focusing on innovations that do not radically diverge from the dominant system [33], mistakenly giving incremental innovations the status of niche and restricting transitions to the broad dissemination of an alternative technology with lower environmental impact. Examples in the agricultural sector are studies on innovations aimed at mitigating the environmental impact of intensive farming, which certainly induce changes in farmers' practices and at the system level, but actually contribute to the adaptation and reproduction of the unsustainable, dominant agri-food system. Which rules are we talking about? How do we assess their radical nature? Despite being a key concept in the MLP, the notion of rules is rather imprecise and has hardly been operationalized [34]. Rules are "what lies underneath the activities of actors" ([1] p. 31) and encompass a wide variety of heterogeneous elements such as beliefs, social norms, world views, lifestyles, consumption patterns, users' expectations, technological paradigms, research agendas, heuristics, problem definitions, values, policies, regulations, and contracts [1][2][3][4]8,35]. This heterogeneity makes the empirical study of rules difficult. The proposition here is to consider that what lies underneath the actors' activities is their representations, and that their activities consist of a set of practices and interactions. On this basis, saying that niches' rules radically differ from the regime's amounts to saying that the representations, practices, and/or interactions developed by niche actors deviate (i.e., engage in an antagonistic dynamic [24]) on most points from those in the regime. Representations are here broadly defined, encompassing all elements that influence actors' practices and interactions: the values they uphold; the way they define their role (their objectives, missions, strategies, and the criteria they choose for evaluation) and the others' roles; the problems, issues, and solutions they identify; their vision of the future; and their vision of agriculture, of rural development, of organic farming. The notion of representation thus encompasses the various variables that previous work on agency in transitions have focused on and proven useful (Section 1), which allows grasping the complex and evolving rationality of actors. Practices are the actual practices implemented by the actors, the building blocks of their activities that contribute to the societal function under study. For the present analysis, the practices directly and indirectly related to the production, buying, selling, storing, transportation, processing, marketing, and consumption of food products that are implemented by the various actors of the agri-food, socio-technical system will be considered. Practices are both constituent parts of the socio-technical system and proxy indicators of the actors' implicit representations. As they reflect the actors' scope of action, they are also proxy indicators of the power relations at stake. Interactions are all the formal and informal relations between actors of the socio-technical system, e.g., collaboration, coordination, knowledge exchange, and discussion. Geels ([1] p. 31) recommends that "the analyst should first demarcate her object of analysis and then operationalize the analytical levels from the MLP". Inferring the analytical levels from the analysis allows avoiding two other epistemological pitfalls, Sustainability 2021, 13, 6135 4 of 21 which I believe have so far obscured the role of actors in transitions. The first pitfall is to define social groups ex ante. According to Pinch and Bijker [36], a social group is a group of individuals who share the same set of meanings attached to an artefact. It may therefore be convenient to distinguish various social groups within a social category, if the individuals composing this category do not share the same set of meanings. The MLP theoretically bases analyses on social groups [37], but actors are generally considered in monolithic, pre-determined categories (e.g., producers, consumers, public authorities, firms) (for a review of categories and typologies in the literature, see [28]). Consequently-and this is the second pitfall-niche and regime actors are designated ex ante, based on whether they do or do not use an alternative technology or contribute to its development. For instance, scholars working on organic farming considered organic actors as niche and non-organic actors as regime (e.g., [38,39]). Yet farmers' motivations to convert into organic farming are diverse; some convert for reasons of ill health or economic expediency, while others are engaged in a philosophical reflection on mankind's relation to nature. The technical and marketing options they choose relate to these motivations and have different transformative potentials [14]. Various forms of organic farming coexist, ranging from highly diversified and self-supporting farming systems to highly specified and input-intensive farming systems. These farming systems are implemented by diverse social groups, embedded in various supply chains, and supported by different networks of actors-some contributing to the reproduction of the system and others inventing, for instance, new distribution channels or consumption patterns. To account for this complexity, the proposition is to conduct a 'flat' longitudinal analysis, i.e., without defining social groups or analytical levels ex ante, and to infer them from the empirical findings. In a whole system approach, the analysis of socio-technical dynamics takes into account all the actors of the agri-food system (i.e., involved in land allocation, input supply, agricultural production, processing, retail, consumption, research, development, education, public policies and regulation). It is performed through the prism of the three variables presented above: (1) these actors' technical, social, marketing, and/or organizational practices; (2) their representations; (3) their interactions. These variables allow identifying groups of individuals and organizations that share common sets of representations and whose practices are aligned; they allow characterizing distinct sociotechnical configurations ( Figure 1). The regime is the dominant configuration, i.e., the largest one. Niches are alternative configurations, composed of actors whose practices and representations radically diverge from the regime. believe have so far obscured the role of actors in transitions. The first pitfall is to define social groups ex ante. According to Pinch and Bijker [36], a social group is a group of individuals who share the same set of meanings attached to an artefact. It may therefore be convenient to distinguish various social groups within a social category, if the individuals composing this category do not share the same set of meanings. The MLP theoretically bases analyses on social groups [37], but actors are generally considered in monolithic, pre-determined categories (e.g., producers, consumers, public authorities, firms) (for a review of categories and typologies in the literature, see [28]). Consequently-and this is the second pitfall-niche and regime actors are designated ex ante, based on whether they do or do not use an alternative technology or contribute to its development. For instance, scholars working on organic farming considered organic actors as niche and non-organic actors as regime (e.g., [38,39]). Yet farmers' motivations to convert into organic farming are diverse; some convert for reasons of ill health or economic expediency, while others are engaged in a philosophical reflection on mankind's relation to nature. The technical and marketing options they choose relate to these motivations and have different transformative potentials [14]. Various forms of organic farming coexist, ranging from highly diversified and self-supporting farming systems to highly specified and input-intensive farming systems. These farming systems are implemented by diverse social groups, embedded in various supply chains, and supported by different networks of actors-some contributing to the reproduction of the system and others inventing, for instance, new distribution channels or consumption patterns. To account for this complexity, the proposition is to conduct a 'flat' longitudinal analysis, i.e., without defining social groups or analytical levels ex ante, and to infer them from the empirical findings. In a whole system approach, the analysis of socio-technical dynamics takes into account all the actors of the agri-food system (i.e., involved in land allocation, input supply, agricultural production, processing, retail, consumption, research, development, education, public policies and regulation). It is performed through the prism of the three variables presented above: (1) these actors' technical, social, marketing, and/or organizational practices; (2) their representations; (3) their interactions. These variables allow identifying groups of individuals and organizations that share common sets of representations and whose practices are aligned; they allow characterizing distinct socio-technical configurations ( Figure 1). The regime is the dominant configuration, i.e., the largest one. Niches are alternative configurations, composed of actors whose practices and representations radically diverge from the regime. To analyze the interactions between these socio-technical configurations, as well as the associated changes in stability and power relations, the concepts of enrolment and interessement from pragmatist sociology are mobilized (Table 1). It should be specified that both concepts, as well as the 'flat' empirical approach, are incorporated as analytical tools to avoid the above-mentioned pitfalls, with no intention, however, to question the multi-level conceptualization. Enrolment is the process by which an actor successfully redefines other actors' roles, goals, orientations, motivations, or interests [40]-i.e., their representations-resulting in the building and breaking of alliances. The founders of the MLP already mobilized this concept to describe processes of niche development, and despite the ontological controversies this generated [8,9,41], further combining pragmatist sociology and MLP has proven relevant to analyze niche-regime interactions [42,43]. The originality of the present approach is twofold. First, it considers the building and breaking of alliances in a whole system approach instead of taking as a starting, focal point one single niche innovation. This enlarged perspective on enrolment processes allows characterizing changes in power relations and mechanisms of (de)stabilization of socio-technical configurations. As their set of alliances enlarges, actors grow more powerful. As more actors are enrolled in one sociotechnical configuration, the latter stabilizes [44] due to the alignment of the representations of the various actors involved [40], as well as of their practices. In turn, other configurations are destabilized as they are deprived of those actors and of their human, material and financial resources, and the remaining actors are weakened. The second originality of the present approach is to incorporate the concept of interessement. Interessement is the process by which one actor enforces others into new roles and activities [45]. In the present approach, it is used to highlight the more subtle changes in power relations that occur when one actor enforces changes in practices and representations of another actor, but without the practices and representations of both actors becoming aligned. Although the socio-technical configuration the enforced actor is part of is not destabilized, the set of practices and representations it is composed of is (partially) modified. Hence, the concept of interessement contributes to a more comprehensive view of socio-institutional changes, emphasizing novel aspects of agency. Table 1 presents how the key concepts, variables, and mechanisms of the analytical approach are linked. A Qualitative and Inductive Methodology This whole system approach was used to analyze the case of the Drôme valley, a rural territory located in southeast France (around 2200 km 2 , 102 municipalities, 54,000 inhabitants), chosen for its unique development of organic farming (OF). There, OF represents one third of farmers, versus 6% at the national scale, and involves a large diversity of local actors, including processing, marketing, and retail organizations, as well as local authorities who elaborated an ambitious local development policy called Biovallée. This suggests that, locally, actors have built a new socio-technical system around OF, which was treated as a research question when fieldwork started. The research is conducted at the local scale to ensure the feasibility of combining a 'zooming out strategy' with a detailed analysis of socio-political aspects. Extending Lawhon and Murphy's argument that focusing on dynamics through which regimes that are differentially embedded in a diversity of local contexts is necessary to understand "why progress towards sustainability proceeds in a spatially uneven manner" ( [46] p. 362), the assumption here is that an in-depth whole-system analysis of such an 'unlocked' form of embeddedness may provide generic insights on agency. The local agri-food system is hence approached as a local variation of the regime and considered in its dynamic, multi-scale (local, regional, national, international) dimension. The socio-technical evolution of the local agri-food system was analyzed from the late 1960s, when OF first appeared in this valley, until 2015. In a whole-system approach, all the local actors who have an influence on, or aim to influence, agri-food dynamics were taken into account: farmers, consumers, local authorities, the local agricultural school, civil-society organizations, and the main agricultural economic operators (input suppliers, processors, farmers' marketing cooperatives and other agri-businesses, small and corporate food retailers, restaurants, canteens). All food chains of the valley were included: goat and laying hen rearing, aromatic and medicinal plants, fruits, vine, cereals, and vegetable growing. As some components of the agri-food system only exist at other scales, also included were departmental and regional organizations such as farmers' associations, research and experimentation stakeholders, wholesalers, and the Chamber of Agriculture-referred to as 'the Chamber' from this point on (Chambers of Agriculture are professional, agricultural associations and they are the main players for training and extension services in France). Where they existed, alliances with more remote actors were also taken into account. The broader dynamic context (landscape level) was accounted for at regional, national, European, and international scales in terms of agricultural and rural development policies, economic environment and agricultural prices, and cultural changes related to environmental and agricultural issues. The practices, representations, and interactions of the above-mentioned actors were examined (illustration for some of them in Table 2) longitudinally. On this basis, different socio-technical configurations, composed of actors whose practices and representations are aligned, were identified. Practices Farmers: agricultural, training, experimentation, and marketing practices Farmers' cooperatives: infrastructure building, investments, experimentation programs and designs, HR management (sales commission, coupling of technical and commercial functions), purchasing and marketing practices, organization of grain collection and share of OF Local authorities: design, content, implementation, and assessment of Local development and Agricultural policies; distribution of financial, material, and human resources Chamber of agriculture: content, speakers, and target population of training courses; design and experimentation programs; agricultural and rural development programs; financial, material, and human resource allocation Retailers: sourcing, buying, organizational and marketing practices Representations Representation of their own role (their objectives, missions, strategies, and the criteria they chose for evaluation), and of the others' roles; the problems, issues, and solutions they identify; their vision of the place and future of agriculture, of rural development, of conventional and alternative farming, of the "good" agriculture and agri-food system. Interactions Presence/absence of relations between actors, nature (e.g., supplier/client or subsidized organization/funder) and functioning (decision-making processes). Fine granularity was preferred to homogeneous treatment, which is why not all actors were equally considered. First, a few key actors were examined in-depth: the two community of municipalities of the valley that elaborated the flagship project Biovallée (2009)(2010)(2011)(2012)(2013)(2014)(2015), which targeted 50% OF; a farmers' cooperative, which gathers two thirds of local farmers and is 70% organic-a unique positioning for a French grain cooperative; and four initiatives shown, during the first year of fieldwork, to play a pivotal role in reconfiguration dynamics. On this basis, the trajectories (i.e., the co-evolution of the interdependent practices, representations, and interactions) of each of these actors or networks were thoroughly analyzed. Through the lens of these trajectories, data on the practices, interactions, and representations of the other actors was collected. Using writing as an analytical process [47], these trajectories were then weaved together into a monograph of the local agri-food system. Data collection was based on an ethnographic study and archival work. The ethnographic study (2012-2015) includes observations of 59 interaction situations to analyze representations and interactions in present time (e.g., agricultural committees and working meetings of local authorities; training sessions organized by the Chamber and the Organic farmers group; project steering meetings, general meetings, and board meetings of various local, departmental, and regional organizations; public events) and 24 comprehensive interviews to obtain actors' interpretation of past and present events. Thirty interviews, conducted in 2011-2013 by colleagues, were also integrated into the analysis (Table 3). Such a high number of interviews is justified by the diversity of actors included in the whole system approach and the length of the study period. For the purpose of the longitudinal analysis, an extensive documentary work was also completed. More than 700 documents dating from 1969 to 2015 were qualitatively analyzed (project documents; guidance documents; minutes of general meetings, board meetings, and working meetings; minutes of agricultural committees; of steering committees; articles from general and specialized press; newsletters and communication tools, grey literature). This was done in order to gather data on past controversies, negotiations, conflicts, and projects, including failed ones, as well as quantitative data such as agricultural statistics, demographics, national and European financing, local and regional budget allocations, and the development of national and international agricultural prices. Due to the heterogeneity of the data sources and of the data itself, manual coding was preferred to computer processing. Triangulating multiple types of data allowed deconstructing the reinterpretation of past events actors may make in present time and retracing the processes of socio-technical evolution as precisely as possible. The resulting analytical narrative is summarized in the next section (in which, unless indicated otherwise, all data and analyses come from the author's Ph.D. thesis [48]). The Agroecological Transition in the Drôme Valley (1969-2015): A Two-Step Process "[In the agricultural secondary school], I was formatted, like all the kids back then, that is to say to productivism, farm expansion [ . . . ] and fertilizers, agrochemicals, etc. So, we came out very, very formatted. And when I settled into farming, I was in that frame of mind. And I remember, I had settled for two months I think, and my uncle [ . . . ] came to see me, and he came with a fertilizer salesperson. In the countryside, dealers always come along with a farmer because it appears like a guarantee. [ . . . ] He said to me: "Oh, this land hasn't seen fertilizers for a long time, you should buy NPK, potash, nitrates and all. Otherwise, you'll have no grass for your animals". [ . . . ] Neither did the veterinary know about phyto[therapy] nor veterinary homeopathy, so we did work with antibiotics, it was extremely traditional. On top of that came the farm technicians, who were totally in a vision . . . when you settled, to get the Young Farmer Grant [ . . . ] you had to fit in a framework for investment, to buy equipment, to invest tremendously. So, I was totally into this vision [ . . . ] with the advice from the technician of the Chamber of Agriculture, with the training of the agricultural secondary school. And the social context too, because it is just as strong, as significant as the good advice from professionals." (Former breeder, who settled into farming in the Drôme valley in 1972) As illustrated by this producer's testimony, in the 1970s in the Drôme valley, one single socio-technical configuration gathered producers, veterinary and extension services, agrochemical industries locally represented by independent salespersons, and the agricultural school around a vision of "good agriculture" based on mechanization and the intensive use of chemical inputs. This socio-technical configuration also included supply and marketing farmers' cooperatives and local authorities. As such, it can be inferred that, as elsewhere in France and western Europe, the local agri-food system was aligned with the paradigm of "Agricultural modernization". As it combined a large set of heterogeneous elements, this sociotechnical configuration was highly stable. In this socio-technical configuration, the agricultural school and the Chamber played a major role because their strategies and activities strongly shaped those of the other actors. Local authorities (i.e., the two communities of municipalities of the valley) played a minor role, limited to the implementation of national agricultural policies, that encouraged agricultural restructuring to expand the size of farms, mechanization, intensification, and the development of income-generating products to maintain farming activities and thereby combat rural depopulation. It is in this context that OF emerges. It is developed both by innovative newcomers and local farmers [49] who feel at odds with intensification and convert to OF to translate their set of values into a set of practices [50]. Most newcomers belong to the second "backto-the-land" movement, which took place after 1975 in France, and yearn for living in the countryside and entering the existing social and economic networks [51]. They join traditional agricultural organizations such as farmers' cooperatives and experimentation groups, which locally fosters the development of OF. Step 1: Differentiation of Two Configurations within the Regime The creation of the organic label in 1985 incites the four local farmers' cooperatives located in the upstream part of the valley called Diois, to elaborate a project, to structure organic supply and marketing chains for organic farmers to access the inputs they need, and to add value to their products. To cover the investments for infrastructures, the cooperatives need to increase the share of organic sales (OF only represents a few percent in the early 1990s), so the project includes conducting experiments to develop their staff's advisory skills. The cooperatives convince the community of municipalities of Diois (CCD) to support their project. In a context where rural development policies take a quality turn [52], and where locally depopulation is a major issue, OF appears as a way to maintain agriculture by better rewarding producers and creating a positive image of the area. This alliance is critical as it allows benefiting from the support of departmental authorities (in France, departments and regions are the next administrative levels above communities of municipalities) that have direct connection with EU agents and manage to get a special envelope from the European Union. These credits act as a driving force: the cooperatives receive funds that national and regional institutions would not have granted otherwise, and the high rates of funding incite them to develop a more ambitious program. The initial six-year project is deployed for ten years and gives visibility to Diois, which becomes recognized as an innovative area at the national scale. The influx of funding has other symbolic impacts: it increases the legitimacy of organic farmers, who stop being perceived as eccentrics, and the credibility of OF as a way forward for the whole local agriculture. This translates into the conversion of Diois' agricultural leaders and, consequently, of more farmers. Additionally, it fosters the enrolment of CCD, as indicated by the title of its guidance document in 1995: "Organics, the future of Diois", and by its recruiting of the coordinator of the project in the late 1990s, which proves the institutionalization of this vision of OF and of the role of CDD as a legitimate stakeholder in relation to agricultural matters. The inter-cooperative project has a structuring effect both technically and commercially, and OF reaches 10% in 10 years. It also has a significant impact on the activities and representations of the other actors of the dominant system (interessement). To receive funding from the inter-cooperative project, the Chamber starts providing advisory services to organic farmers (which it had refused to do up to that point), and the agricultural school starts proposing training on OF; OF, which had hitherto been an invisibilized model to the major players, is put on their agenda. In turn, in the 2000s, in the Val de Drôme downstream area, where farming is more intensive than in Diois, many farmers, processors, and other cooperatives also start to engage in organics. This is of course also fostered by changes at the landscape level consisting of an increasing demand for organics, but the dynamics is much stronger in the valley than elsewhere, and the detailed analysis [48] shows that it results directly from the shift in activities and representations of the major agricultural players of the area. This momentum is maintained with the combined actions of various actors. For instance, in the early 2000s, the grain cooperative invests in a seed platform to produce organic seeds adapted to local conditions (an imperative need no longer fulfilled by seed-bearers). It merges with the supply cooperative and redefines its advisors' routines, disconnecting their salary from product sales, focusing their activity on extension services and developing their skills on organics (trials on high value crops and advice on farm seeds to foster farmers' self-sufficiency). Additionally, it buys products that are farmed organically but not yet certified at the price of organic products, to support farmers who chose to convert (during the first three years of conversion, organic farmers are not allowed to sell their products as organics, which is a major obstacle given that conversion is generally associated with significant yield loss). The other cooperatives and CCD also continue supporting the development of OF (for a detailed analysis, see [48]). These combined actions allow OF to rise from 10 to 25% in a few years in Diois. As OF develops, its image evolves. The Chamber starts acknowledging its technical interest by recruiting new organic advisors, and from 2007 on, organizing a professional fair on alternative and organic techniques. This legitimizes OF as a sound technical model and fosters conversions in other parts of the valley. However, although the Drôme Chamber becomes the number one for OF, OF remains a marginal issue in its guidance documents; recruiting organic advisors is not related to a profound shift in representations, it is a way to preserve its hegemony in the local agri-food system (interessement). All organic chains of the valley experience a boom in the 2000s. The aromatic and medicinal plants chain is particularly emblematic. The outstanding quality of its products achieves international recognition, and companies settle in the valley to process these products, generating hundreds of employment opportunities [53]. Consequently, the community of municipalities of Val de Drôme (CCVD) starts considering OF as an economic drive (interessement). It proposes CCD to elaborate a joint project at the scale of the whole valley, taking organics as the foundation of an endogenous development. In 2008, they seize a regional funding opportunity that calls for innovative pilot projects with an integrated approach to sustainable development. This encourages them to develop a more ambitious and integrated project: the Biovallée program, aimed at making the Drôme valley the European region of reference for sustainable development. The objectives for agriculture are far more ambitious than the national goals prevailing at that time (Table 4) and put OF to the fore as a model; CCVD has been enrolled and its vision and policies are now aligned with those of CCD. In parallel, the agricultural school, threatened with closure, makes a strategic move towards OF. It decides to specialize in organics, and it eventually does avoid closure. In summary, a first transformation of the local agri-food system starts in the early 1990s. A new socio-technical configuration emerges within the regime as a result of the changes in practices of some farmers and of the inter-cooperative project. This new sociotechnical configuration progressively aggregates increasing social and technical elements and becomes more stable. The practices and representations of the various actors of the local agri-food system co-evolve for two decades. This results, in the late 2000s, in the regime being composed of two socio-technical configurations, which partially integrate the proposal of OF (partially in the sense that neither embraces the proposal of OF in all its technical, environmental, economic, and social dimensions, as it is defined by the International Federation of Organic Agriculture Movements or in the Charter of the French organic movement). Of the two configurations, one is aligned with the paradigm of "Agricultural modernization", and the other gathers farmers, the Diois cooperatives, the two local authorities, and the agricultural school, around a new technological trajectory and a set of representations that can be referred to as the paradigm of "Ecological modernization". "Ecological" because OF is considered as a way forward for all local agriculture, and "modernization" because it does not question the functioning of the global dominant system (without reference to the ecological modernization literature). This is the first step towards a further, more profound reconfiguration of the local agri-food system. Step 2: Development of a Third Configuration Outside and in Interaction with the Regime The Biovallée program is launched in 2009, but local authorities face difficulties regarding its implementation. Their strategy is to stimulate strong growth in OF through the installation of agro-industries, encouraging a cluster dynamic. They also have a partnership with the Chamber to develop training and create an experimental platform on OF and alternative farming techniques. However, collaborations with mainstream actors fail to sustain this strategy, with several projects aborting. For instance, the Chamber creates an experimental platform, but the experiments are designed following conventional principles (input-intensive monocrops, highly mechanized technical operations, etc.) and consequently do not answer organic farmers' needs for technical references. Additionally, despite strong demand stemming from the Biovallée program, the Chamber decides not to take charge of organizing food chains for school catering. A distribution platform is created by a major distributor of organic products, with strong financial support from Biovallée, but closes after two years because the company considers school catering as not being profitable enough. Another objective of Biovallée is the creation of farm incubators in economic activity areas for developing small-lot, high added value agriculture. Encouraging young farmers' setting-up is one of the Chamber's missions, and the fact that it undertakes no action in relation to that objective proves again its resistance to the vision of the Biovallée program. It is in this context that local authorities are asked for funding by actors that develop other ways of supporting OF: • Carline, a cooperative organic store willing to increase its share of local products; • Agricourt, a community association gathering parents and farmers willing to create a distribution platform to supply school canteens with fresh, local products from small-scale agriculture; • The departmental goat farmers' union, willing to develop an innovative experimental program on herbal veterinary treatments (phytotherapy) in collaboration with a veterinarian and a pharmacist; • Compagnons de la Terre, a community association already running a small, organic farm incubator in the area and gathering ex-employees of the agricultural school, farmers, and integration associations, which is asked by local authorities, after the Chamber declined, to create a farm incubator for Biovallée on a 16-hectare farm they had newly acquired for that purpose. Thanks to Biovallée funding, these initiatives develop. Carline supports the emergence or growth of various local organic food chains (e.g., meat, cheese, flour and bread, vegetables) by setting prices in cooperation with local producers, applying lower margins to local products (and even lower margins to new entrants in farming), investing in equipment (e.g., refrigerated showcases) and infrastructure (purchase of a 160 m 2 retail space), and developing projects to better coordinate (via pooled cropping plans or equipment) and enhance the technical skills of the small-scale farmers it works with. Thanks to these developments, it manages to supply 10% of the local population with organic products from local and small-scale farmers, based on shared governance between consumers, producers, and its employees, who jointly design an alternative economic model and its Ethics charter. This leads it to gradually consider itself as a legitimate player in local development. This representation is now institutionalized, as it became a member of the CCD agricultural committee, where local agricultural policies and grant allocations are discussed. Agricourt also supports the growth of local food chains, both organic and non-organic, with the creation of a distribution platform. It started in the President's garage with a private vehicle, and now has dedicated premises with a cold room and refrigerated lorry that supplies nurseries, schools, and private and company restaurants in the valley. Shared governance between consumers, producers, processors, and restaurants allows the development of fair and ecologically-sound marketing practices (e.g., prices and cropping plans collectively discussed, long-term agreements, market price list displaying only local products, partnership with an organic cooperative sharing a similar Ethics charter to complement local production when lacking), thanks to which, 60% of the meals served in the schools of the valley are prepared with local products from small-scale, mostly organic, fairly remunerated farmers. By comparison, during the same period in France, 25% of school canteens propose organic products, generally once a month (2/3) or once a week (1/3), and only half of these canteens report favoring local food chains (Agence bio, 2015). In the goat phytotherapy project, farmers make, test, and collectively assess treatments made from on-farm resources, to explore solutions to technical bottlenecks they encountered. The goat farmers' union turned to Biovallée after its project was rejected by the regional experimentation committee, which considered that phytotherapy was not a promising prospect. Despite certain reservations about the scientific soundness of the project, Biovallée funds were granted. The fact that none of the funding dedicated to research in the Biovallée program had been used was certainly an important factor. The project generated new knowledge on herbal treatments and interested FIBL, a worldleading research institute on OF. The resulting partnership brought credibility to the project and legitimacy to the topic and eventually lead to its integration into the regional research agenda by the regional experimentation committee, as well as into the national training offer (the project served to create a training module on phytotherapy provided by the national agricultural training agency). Compagnons de la Terre supported a dozen new entrants in farming, with professional guidance and provision of land and farm equipment, supplemented as needs arose by storing, processing, and marketing facilities. Then, acknowledging incubated young farmers' persisting difficulties to access land due to rejection by conventional farmers of both OF and new entrants in farming, they also built, together with CCVD, a new system for farmland management: a public land tenure fund, which favors the setting up of organic new farmers and enables them to take ownership of the farmland after two years of testing their activity. This farmland management system is implemented in coordination with SAFER, the Land Use and Rural Settlement Agency. All four initiatives can be considered as socio-technical niches, as they are small networks of actors that develop new rules and practices that radically differ from the regime's. The analysis of their trajectories, which can only be roughly summarized here to conform with the format of an article (for a more detailed analysis of the initiatives' trajectories, see [42,48]), shows that the coevolution of practices, representations, and interactions within each niche and in interaction with local authorities leads to the construction of alternative models (economic for Carline and Agricourt, land management for Compagnons de la Terre, and related to animal-health management for the goat farmers' union) and of radically alternative visions of food and farming issues, as well as of the associated network of relevant actors. This set of representations can be referred to as the paradigm of "Radical ecologization", given that its representation of good agriculture is totally different from the one in the dominant paradigm. In this setting, good agriculture is a peasant, small-scale, autonomous agriculture, making little or no use of chemical inputs while being respectful to the environment. It produces healthy food, offers good working conditions for farmers and workers, and implies reconnection and empowerment of consumers and producers. These niches have little interactions; the paradigm of "Radical ecologization" is supported by a network as conceived by Thomas Hughes [54]: a seamless web. By receiving awards (from Fondation de France for Compagnons de la Terre and Fondation Crédit Coopératif for Agricourt) and support from international institutions (from FIBL for the goat farmers' union), as well as a lot of media attention (Figure 2), these initiatives help legitimize the locally controversial Biovallée program. Sustainability 2021, 13, x FOR PEER REVIEW ecologization" is supported by a network as conceived by Thomas Hughes [54] less web. By receiving awards (from Fondation de France for Compagnons de la T Fondation Crédit Coopératif for Agricourt) and support from international ins (from FIBL for the goat farmers' union), as well as a lot of media attention (Figure initiatives help legitimize the locally controversial Biovallée program. Contrasting with the disengagement of incumbent business and agricultura these niche initiatives bring about solutions that contribute to the Biovallée object demonstrate their feasibility-this was contested by many conventional agricul tors, as was their relevance for the development of the area. The values and rep tions they convey are gradually shared by a wider range of local actors, includ authorities. A shift in discourse can be noticed in 2013: while they had, up to th promoted an organic cluster's dynamic based on the development of agro-indu ganic food supply chains with the intention of exporting food products from th as well as promoting the use of organic or local products indifferently in school c local authorities start advocating the use of fresh, local products from small-scale (who are mostly but not exclusively organic) in school canteens, the relocation of Contrasting with the disengagement of incumbent business and agricultural actors, these niche initiatives bring about solutions that contribute to the Biovallée objectives and demonstrate their feasibility-this was contested by many conventional agricultural actors, as was their relevance for the development of the area. The values and representations they convey are gradually shared by a wider range of local actors, including local authorities. A shift in discourse can be noticed in 2013: while they had, up to that point, promoted an organic cluster's dynamic based on the development of agro-industrial organic food supply chains with the intention of exporting food products from the valley, as well as promoting the use of organic or local products indifferently in school canteens, local authorities start advocating the use of fresh, local products from small-scale farmers (who are mostly but not exclusively organic) in school canteens, the relocation of the food system, and the reconnection between its various actors. This creates a conflict with the Chamber, whose representative, opposed to this vision, claimed loudly during one Biovallée agricultural committee that "agricultural policies must be in the farmers' hands!" and almost slammed the door. The agricultural school is also deeply influenced. In the same period, it develops several modules on farm autonomy (farm-saved seeds, animal phytotherapy, etc.), in line with the "Radical ecologization" paradigm. Further fundamental changes come about within the regime. In the socio-technical configuration fostering "Ecological modernization", the cooperatives continue their technological trajectory. For instance, in 2008, the local grain cooperative sets a target of 100% organics. At this time, organic crops represent 30% of the collection but generate 45% of the revenue, so in a competitive environment where small cooperatives tend to disappear [55], OF represents a solution for local farmers to keep control of their cooperative tool. It initiates the creation of a cooperatives' union for cereals marketing with neighbor cooperatives, allowing it to manage larger volumes and improve their knowledge and bargaining power in the market of organic cereals. It then engages in a strategy of downstream vertical integration; investing in a feed-mill, it turns organic to process its members' production and invests in an organic egg-firm that valorizes the feed produced in the mill to ensure that added-value is distributed to the farmers. It does not intentionally seek to change the dominant system (most grain and eggs are sold in long circuits), yet it creates new practices and new forms of coordination that give farmers and their cooperative a greater control of the supply chain. Thanks to their alliance in the Biovallée program, the two local authorities increase their budget three-fold for agriculture and thereby strongly enhance their capacity to orient the other actors' activities, bringing about critical changes in the dominant sociotechnical configuration related to the paradigm of "Agricultural modernization". The Chamber, whose budget significantly depends on project financing, is forced to actively contribute to the development of OF and to the Biovallée objectives, despite the ongoing conflict with local authorities. It works with the cooperatives on the testing of organic varieties, supports farmers who wish to convert, and organizes demonstration days and training sessions on organic techniques. On one of these demonstration days, funded by Biovallée, a local company of agricultural machinery showcases self-engineered equipment for mechanical weeding; various local actors join the dynamics around OF, and the artefacts and practices they build foster interessement and enrolment of more farmers. This further legitimizes OF and provokes a massive adoption of some OF practices by conventional farmers. For instance, nowadays 50% of all local grain farms use mechanical weeding instead of chemicals. In summary, enrolment processes allowed for the emergence and development of two alternative socio-technical configurations within the local agri-food system: one within the regime, mainly driven by farmers' cooperatives and aligned with a paradigm of "Ecological modernization", and one outside the regime, driven by several niche initiatives and aligned with a paradigm of "Radical ecologization". They lead to the alignment of the visions and practices of many actors with these two alternative paradigms, which in turn, through interessement processes, caused OF to be placed on the agenda of all local agri-food actors and to become a structuring element of the local agri-food system (Figure 3). Sustainability 2021, 13, x FOR PEER REVIEW 15 of 24 Discussion The longitudinal analysis of the local agri-food system brings evidence that, although OF still represents a minority of farmers, a profound reconfiguration of the socio-technical system (i.e., a transition) is ongoing in the Drôme valley. This transition follows a much more complex dynamic than the MLP archetype, as it results from the interaction between, not two, but three socio-technical configurations. As we will now see, accounting for this complexity allows the identification of the mechanisms that have allowed a transition to take place in this region and how actors have enacted them. Interessements and Enrolments Causing Regime Destabilisation and Reconfiguration The destabilization of the existing regime starts with the engagement of the farmers' cooperatives on a new technological trajectory and their enrolment of CCD. The representations and practices of these regime actors progressively align around a paradigm of "Ecological modernization", and these actors start considering OF as a way forward for all local agriculture. This provokes the unlocking of technical and governance regime elements, as well as beliefs, roles, and business models, identified by Turnheim and Geels [56] as signals of full destabilization. As such, these changes are significant and fundamental. These actors may, however, be considered as remaining within the regime, as neither their representations nor their novel practices and interactions challenge the dominant system-not at the beginning at least (Figure 4). Discussion The longitudinal analysis of the local agri-food system brings evidence that, although OF still represents a minority of farmers, a profound reconfiguration of the socio-technical system (i.e., a transition) is ongoing in the Drôme valley. This transition follows a much more complex dynamic than the MLP archetype, as it results from the interaction between, not two, but three socio-technical configurations. As we will now see, accounting for this complexity allows the identification of the mechanisms that have allowed a transition to take place in this region and how actors have enacted them. Interessements and Enrolments Causing Regime Destabilisation and Reconfiguration The destabilization of the existing regime starts with the engagement of the farmers' cooperatives on a new technological trajectory and their enrolment of CCD. The representations and practices of these regime actors progressively align around a paradigm of "Ecological modernization", and these actors start considering OF as a way forward for all local agriculture. This provokes the unlocking of technical and governance regime elements, as well as beliefs, roles, and business models, identified by Turnheim and Geels [56] as signals of full destabilization. As such, these changes are significant and fundamental. These actors may, however, be considered as remaining within the regime, as neither their representations nor their novel practices and interactions challenge the dominant system-not at the beginning at least (Figure 4). This is an essential first step, because some incumbent actors' visions become compatible with more radical representations, laying the foundations for further change. As they implement strategies to support the development of OF through the setting up of new forms of coordination and new practices, some actors manage to redefine their own role and to enlarge their capability within the local agri-food system: CCD becomes a legitimate stakeholder in relation to agricultural matters; the local farmers cooperatives develop advisory activities and increase their control over the supply chain. This impacts incumbents' roles: the Chamber is no longer the only local actor who can determine local agricultural policies and agendas and offer agricultural extension services. As suggested by Wittmayer et al. [28], the creation of new roles, and the alteration of existing ones, provokes regime destabilization. This destabilization leads to a transformation of the selec- This is an essential first step, because some incumbent actors' visions become compatible with more radical representations, laying the foundations for further change. As they implement strategies to support the development of OF through the setting up of new forms of coordination and new practices, some actors manage to redefine their own role and to enlarge their capability within the local agri-food system: CCD becomes a legitimate stakeholder in relation to agricultural matters; the local farmers cooperatives develop advisory activities and increase their control over the supply chain. This impacts incumbents' roles: the Chamber is no longer the only local actor who can determine local agricultural policies and agendas and offer agricultural extension services. As suggested by Wittmayer et al. [28], the creation of new roles, and the alteration of existing ones, provokes regime destabilization. This destabilization leads to a transformation of the selection environment, with newly developed practices and representations being more compatible with radical innovations. As such, it is preparatory and necessary to the more profound reconfiguration provoked by the enrolment of regime actors into the alternative socio-technical configuration that fosters "Radical ecologization" (second step). As shown in Figure 5, enrolments make some actors move from the dominant socio-technical configuration to the others, as their practices and representations gradually evolve according to their needs, and to opportunities, and under the influence of external factors, resulting in fine to a greater number of actors in alternative configurations ( Figure 5). Contrary to the stretch-and-transform process described by Smith and Raven [57], where a niche restructures the mainstream selection environment in ways favorable to its development, here the process is contingent and internal and does not result from a niche's deliberate action to transform the regime. The results, in line with Fuenfschilling and Truffer's [58] assumption, show that the transformation of the selection environment through the emergence, within the regime, of an alternative paradigm that niches may be congruent with, is the critical, first step of the reconfiguration process. Combined Effect of Multiple Niches Various niches interact with the regime in the recent period. As described in Section 4, each niche aggregates various elements (infrastructures, practices, representations, governance arrangements) in an alternative model and impacts various components of the local agri-food system (consumption, production, processing, marketing, research, extension, and/or public policies). For instance, Carline influences: farming practices by providing remunerating outlets for small-scale, local, organic farmers; consumption practices by giving access to organic food, often in bulk, at a reasonable price for all inhabitants of Diois; public policies by being a member of the Diois agricultural committee; market, as it developed local food chains; and symbolic meanings by building a viable, alternative economic model. However, if each niche influences various components of the socio-technical system, they each have a significant impact on only one or two components: Carline works with only a few dozen local producers but supplies a tenth of Diois' population; the phytotherapy project only involves fifteen farmers but it manages to put the topic of phytotherapy on the agenda of both the regional body for research on goat farming and the national training fund; the farm incubator has only welcomed a dozen project holders so far, but it strongly impacts access to land with the creation of a new farmland management system; Agricourt also only works with a few dozen local producers but it allows 60% of the meals served in school canteens to be made out of fresh products coming from local, small-scale, fairly remunerated farmers. Unlike the MLP's archetypal suggestion that transitions are triggered by one-niche regime interaction, the transition ongoing in the Drôme valley is provoked by multiple niches, corresponding to distinct components of the socio-technical system, which are aligned with the same paradigm and co-exist and exert pressure, simultaneously, not in a coordinated manner but in the same direction, on the various dimensions of the regime. These results call for revisiting the MLP diagram [1], as presented in Figure 6. agricultural issues and OF. The circles are the actors. Line thickness represents the actors' influence capacity (i.e., power). The white rectangles show the new practices that actors develop. Red arrows show the circulation of actors from one configuration to another along time. Combined Effect of Multiple Niches Various niches interact with the regime in the recent period. As described in Section 4, each niche aggregates various elements (infrastructures, practices, representations, governance arrangements) in an alternative model and impacts various components of the local agri-food system (consumption, production, processing, marketing, research, extension, and/or public policies). For instance, Carline influences: farming practices by providing remunerating outlets for small-scale, local, organic farmers; consumption practices by giving access to organic food, often in bulk, at a reasonable price for all inhabitants of Diois; public policies by being a member of the Diois agricultural committee; market, as it developed local food chains; and symbolic meanings by building a viable, alternative economic model. However, if each niche influences various components of the socio-technical system, they each have a significant impact on only one or two components: Carline works with only a few dozen local producers but supplies a tenth of Diois' population; the phytotherapy project only involves fifteen farmers but it manages to put the topic of phytotherapy on the agenda of both the regional body for research on goat farming and the national training fund; the farm incubator has only welcomed a dozen project holders so far, but it strongly impacts access to land with the creation of a new farmland management system; Agricourt also only works with a few dozen local producers but it allows 60% of the meals served in school canteens to be made out of fresh products coming from local, small-scale, fairly remunerated farmers. Unlike the MLP's archetypal suggestion that transitions are triggered by one-niche regime interaction, the transition ongoing in the Drôme valley is provoked by multiple niches, corresponding to distinct components of the socio-technical system, which are aligned with the same paradigm and co-exist and exert pressure, simultaneously, not in a coordinated manner but in the same direction, on the various dimensions of the regime. These results call for revisiting the MLP diagram [1], as presented in Figure 6. Rebalancing of Power Relations The actors, in line with alternative paradigms ("Ecological modernization" and "Radical ecologization"), enhanced their own capabilities, which indicates that power relations have been altered. The longitudinal analysis shows that, along with the reconfiguration of the socio-technical system, a change in power relations comes about, which occurs through iterative processes of vision and practice adjustment that actors provoke through enrolment processes. In the 1990s, the alignment of more and more numerous and powerful actors with the paradigm of "Ecological modernization" forces the Chamber and the agricultural school to consider OF within their scope and to modify their activities accordingly. Deeper reconfiguration subsequently occurs through the interactions between the four niches and the regime. Within these niches, the radical nature of innovation lies beyond technology; it lies in the challenging of the organization-of the architecture, in McMeekin et al.'s words [31], of the dominant agri-food system. Agricourt and Carline both call for reconnecting producers and consumers and for defining new business practices, giving greater control of the food chain to producers and consumers. The project of the goat farmers' union designs a strategy of animal health management, which gives back to farmers the control over input production, knowledge, and prescription. Compagnons de la Terre acts to widen the management of farmland to civil society and local authorities, facilitating the setting up of atypical entrants in farming. In these niches, actors do not develop technological novelties; they develop radically new visions of what should be the relevant network of actors, and accordingly, new forms of coordination and governance. As suggested by Rossi et al. [27], they are enabling relational environments where power reconfiguration occurs and transformative power [24] develops. The enrolment of local authorities is critical as it prompts niche actors to develop an even more integrated vision of the role of agriculture in local development and of the desirable form(s) of governance. It additionally enables these visions to become a structuring element of the local agri-food system. Consequently, existing governance networks (the network of actors managing farmland, the experimentation network for goat farming, and the governance bodies of local food chains) are influenced or complemented accordingly. This results in the widening of the governance of the agri-food system to include actors who were previously excluded, e.g., small-scale farmers and non-agricultural actors, such as civil society, consumers, and local authorities, and in the weakening of the Chamber. These findings suggest that the local scale allowed overcoming lock-in effects that predominate at broader scales by facilitating the formation of ground-breaking collaborations (the inter-cooperative project, the multi-actor niche-networks) and articulating complementary perimeters of action. This important topic needs deeper elaboration, which will be provided in another article. Conclusions The analytical approach developed in this paper allows exploration of the sociopolitical dimension of whole system reconfigurations and hence highlights novel aspects of agency in transitions. The analysis of the system reconfiguration processes at stake in the Drôme valley sheds light on the way actors created and activated levers for a sustainability transition of the local agri-food system. It points to three main mechanisms that allowed a regime shift to come about: (1) the emergence and development of an alternative socio-technical configuration within the regime; (2) the combined effect of multiple niche-innovations that impact the various components of the socio-technical system, simultaneously and in the same direction; and (3) the rebalancing of power relations and the redesign of local agri-food governance resulting from the interactions between these niches and the alternative configuration within the regime. It shows that these mechanisms were enacted by actors through enrolment and interessement processes, and stresses the importance of spill-over effects. These findings challenge the assumption commonly underlying research on agency in transitions that profound changes at the regime level are intentionally enacted by actors who have an interest in changing the regime, who build alliances and implement strategies to that end. These results build on and articulate various previous contributions from transition studies, thereby confirming the strength of the MLP heuristic. They reinforce the view that accounting for multiple regime dynamics and multiple niche-regime interactions is critical to develop further understanding of transition processes [29]. They contribute to a further characterization of regime destabilization processes and of the incremental system improvements required to foster sustainability transitions [30]. They also suggest that transitions require more than single niche scaling-up or niche accumulation processes. They stress the importance of the emergence of an alternative paradigm within the regime and the role of multiple, unrelated niches impacting the various dimensions of the agri-food system simultaneously. From a transformative perspective, these results suggest that transitions may be fostered by: (1) supporting the diffusion of an alternative technological paradigm within the regime that niches may be congruent with; (2) creating the conditions to stimulate the coexistence and development of a diversity of radical innovations relating to all the various components of the socio-technical system; and (3) moving from a technologydriven approach of innovation towards an emphasis on organizational innovations that foster the rebalancing of power relations. By showing the paramount importance, in this transition process, of the change in power relations that occurred chiefly through linking agricultural issues to wider socioeconomic and political issues-such as fairness, multi-stakeholder decision-making, community life-the present analysis challenges the assertion that a comprehensive understanding of system reconfiguration can be developed with approaches focusing primarily on environmental sustainability, without tackling "wider socio-economic problems such as poverty, inequality, problems in democratic accountability" [59]. It calls for the embracing of a systemic view of sustainability in transition studies, so as to shed light on, investigate, and question the foundational socio-economic and political dimensions of incumbent sociotechnical systems and develop a better understanding of their reconfiguration processes. One important limitation of this research is that it is based on the empirical analysis of a single, local case. This raises the question of the generic nature of its results, particularly given the strong specificities of this case (not a highly productive area, presence of many agricultural productions and chains, long-term inter-municipalities dynamics). Further research could investigate the genericity of the results by looking at other regional dynamics. The spatial dimension of the studied processes would also deserve a profound analysis [60,61], which will be developed in another paper. Funding: This research was funded by INRA and Rhône-Alpes Region, and benefited from financial support from Dynrurabio (ANR-10-STRA-0009) and Era-Net HealthyGrowth projects. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: The data presented in this study are available in [48].	v3-fos-license
2024-05-16T06:17:55.254Z	2024-05-14T00:00:00.000	269768985	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journal-inflammation.biomedcentral.com/counter/pdf/10.1186/s12950-024-00389-8", "pdf_hash": "5c6185a2bee17230c2b0692efdfb08746da0ba72", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45322", "s2fieldsofstudy": [ "Biology", "Medicine" ], "sha1": "4236d63df9bd7db3695534a98f8a6f39ba678422", "year": 2024 }	pes2o/s2orc	Histone H3 posttranslational modified enzymes defined neutrophil plasticity and their vulnerability to IL-10 in the course of the inflammation Background Neutrophils are a heterogeneous population capable of antimicrobial functions associated with pre-activation/activation and tissue regeneration. The specific polarisation of immune cells is mediated by the modification of ‘chromatin landscapes’, which enables differentiated access and activity of regulatory elements that guarantee their plasticity during inflammation No specific pattern within histone posttranslational modifications (PTMs) controlling this plasticity has been identified. Methods Using the in vitro model of inflammation, reflecting different states of neutrophils from resting, pre-activated cells to activated and reducing tissue regeneration, we have analysed 11 different histone posttranslational modifications (PTMs), PTM enzymes associated with remodelling neutrophil chromatin, and H3K4me3 ChIP-Seq Gene Ontology analysis focusing on the processes related to histone PTMs. These findings were verified by extrapolation to adequate clinical status, using neutrophils derived from the patients with sepsis (systemic septic inflammation with LPS-stimulated neutrophils), neuromyelitis optical spectrum disorders (aseptic inflammation with pre-activated neutrophils) and periodontitis (local self-limiting septic inflammation with IL-10-positive neutrophils). Results Physiological activation of neutrophils comprises a pre-activation characterised by histone H3K27ac and H3K4me1, which position enhancers; direct LPS exposure is induced explicitly by H3K4me3 which marked Transcription Start Site (TSS) regions and low-level of H3K9me3, H3K79me2 and H3K27me3 which, in turn, marked repressed genes. Contrary to antimicrobial action, IL-10 positively induced levels of H3S10p and negatively H3K9me3, which characterised processes related to the activation of genes within heterochromatin mediated by CHD1 and H3K9me3 specific demethylase JMJD2A. IL-10 protects changes within histone PTMs induced by TNF or LPS that affected H3K4me3-specific methyltransferase SETD1A and MLL1. Neutrophils previously exposed to inflammatory factors become unvulnerable to IL-10 because previous LPS stimulation interrupts TSS regions marked by H3K4me3 of CHD1 and JMJD2A genes. Therefore, LPS-activated neutrophils are disabled to induce CHD1/JMJD2A enzymes by IL-10, making this process irreversible. Because transcription of JMJD2A and CHD1 also depends on TSS positioning by H3K4me3, neutrophils before LPS stimulation become insensitive to IL-10. Conclusion Neutrophils, once pre-activated by TNF or directly stimulated by LPS, become insensitive to the anti-inflammatory effects of IL-10, and vice versa; IL-10 protects neutrophils against these proinflammatory stimuli. This phenomenon is responsible for disturbing the natural process of resolving inflammation and tissue regeneration. Graphical Abstract Supplementary Information The online version contains supplementary material available at 10.1186/s12950-024-00389-8. Introduction Neutrophils, as the fastest-reacting cells in response to pathogens, with loosely arranged nuclear chromatin facilitating access of transcription factors to DNA in a relatively short time after stimulation, are supposed to be 'ready' for rapid initiation of gene expression related to immune response [1].High antimicrobial potential acts as a double sword, as uncontrolled activation of neutrophils can damage surrounding tissues or lead to increased mortality because of cachexia during activation in the bloodstream [2][3][4].For this reason, neutrophil physiological activation is under control of two steps, which prevent them from uncontrolled initiation beyond inflammatory sites and with adequate force to pathogens.The first step in blood vessels prepares neutrophils for a more effective response.It is initiated by a low concentration of proinflammatory cytokine/ chemokine or components of the complement pathway by signals from the ongoing inflammation [5].The second one is the direct response to adequately recognised pathogens via pathogen molecular pattern manner.Our previous studies also discovered other mechanisms regulating the immune response orchestrated by neutrophils [6].This phenomenon, called 'infection tolerance, ' is initiated by neutrophil-LPS-stimulated Treg interaction, which results in the repolarisation of neutrophils to IL-10-positive cells, which, in the second phase, induces other IL-10-secreting neutrophils in a paracrine manner.We postulate that this two-step reaction results in massive neutrophil apoptosis, occurring during resolving acute inflammation or replacing the innate immunity to acquire one.It is worth emphasising that this mechanism can occur without proinflammatory factors observed within prolonged chronic inflammation or direct (without process of pre-activation) LPS stimulation [7].The process of silencing neutrophil antimicrobial properties is essential in tissue regeneration after infections, as prolonged inflammation leads to tissue damage [8].Continuing our studies, we discovered that despite IL-10-induced apoptosis, it can also switch the profile of cytokines/chemokines/ growth factors desired in resolving inflammation via non-canonical NF-kB pathway with simultaneous canonical NF-kB pathway inhibition [9].IL-10-stimulated neutrophils are characterised by a significant secretion of vascular endothelial growth factor (VEGF) and CXC chemokines.VEGF performs its function by stimulating extracellular matrix remodelling, enhancing endothelial cell proliferation and migration, and increasing vascular permeability, resulting in fibrinogen leakage into the perivascular environment [10].In turn, CXC chemokines (CXCL1, CXCL2, CXCL5, CXCL8, CXCL10, MIF CCL19) associated with induction of angiogenesis by ligation of the CXCR2 recruit and moderate the function of macrophages as well as dendritic cells [11].It has been postulated that neutrophils are actively involved in resolving inflammation by transparent tissues from infiltrated leucocytes and creating a specific environment desired for tissue architecture and function repair [12,13]. The described specific 'polarisation' of immune cells is mediated by the modification of chromatin, which enables differentiated access and activity of regulatory elements that guarantee their plasticity during inflammation [14].The nucleosome consists of H2A, H2B, H3, and H4 histones susceptible to numerous post-translational modifications (PTMs) along their entire length.The correlation between certain histone modifications and specific gene expression led to the concept that they directly 'instruct' transcriptional activation or repression.The combinations of histone modifications are controlled by histone-modifying enzymes that determine diverse DNA scaffold processes, including gene expression [15]. Functional specialisation of the cell and epigenetic 'code' are closely linked.Cildir and colleagues demonstrated a positive correlation between histone H3 lysine K4 trimethylated (H3K4me3) and histone H3 lysine K27 acetylated (H3K27ac) in mast cells upon immunoglobulin E-mediated cross-linking of the IgEε receptors [16].Gene transcription is controlled by enhancers that interplay with gene promotors as gene-distal regulatory elements, increasing gene transcription and characterised by the presence of H3K27ac histones [17].Contrary to enhancers, transcriptional start sites (TSSs) are located mainly within H3K4me3 domains [18].Our previous studies have revealed that circulating neutrophils of HIV-infected subjects compared to healthy controls are characterised by high levels of H3K4me3 histone and low H3Ac.Peripheral neutrophils in HIV-infected individuals are characterised by impairment of their antimicrobial properties, including the ability to chemotaxis, phagocytosis, and oxidative burst potential.ChiP-Seq H3K4me3, bioinformatic analysis of Gene Ontology, additionally revealed significant changes within target genes related to the cell activation, cytokine production, adhesive molecule expression, NF-kB transcription factor in canonical pathways, and also target genes within histone remodelling via upregulation of methyltransferase.Therefore, we conclude that the primary source of neutrophil dysfunction in HIV patients is interplayed between H3Ac and H3K4me3, which results in disturbing gene positioning associated with antimicrobial properties [19].This data proposed that neutrophil pre-activation, which prepares neutrophils to respond more intensely to pathogens, can be related to H3K27ac, contrary to their direct stimulation mediated by H3K4me3 positioning. In this study, we have focused on two processes: the physiological overriding mechanism, which reflects the shifting nature of neutrophils from cells ready to neutralise pathogens to cells reducing inflammation by initiating tissue regeneration after IL-10 exposure and the pathological mechanism disturbing this phenomenon by prolongated pre-activation observed in patients suffering from autoimmune disease and after direct activation by LPS observed during sepsis. Using the in vitro model of inflammation, we have analysed 11 different histone PTMs (marked inter alia: active promotors, transcribed regions, repression genes, enhancer regions), PTM enzymes associated with remodelling neutrophil chromatin, and H3K4me3 ChIP-Seq Gene Ontology analysis focusing on the processes related to histone PTMs.Finally, this data was validated by comparison to adequate clinical status corresponding to the pathological neutrophil condition observed in Neuromyelitis Optica Spectrum Disorders (NMOSD) as an autoimmune disease, E.coli-induced sepsis vs. healthy volunteers.Periodontitis-derived circulating neutrophils were used as a physiological response of neutrophils to IL-10. Study design In the first step of our investigation, using isolated circuiting neutrophils from healthy volunteers, we defined mechanisms involved in neutrophil TNF-pre-activation, LPS direct stimulation and IL-10-induced suppressor/tissue regeneration neutrophils.We made two assumptions: the first, that the antimicrobial and regenerative properties should be excluded as processes that do not co-occur; the second follows from the first, that the prior action of pro-inflammatory factors precludes the activity of IL-10 and vice versa -IL-10 prevents pre-and activation of neutrophils.To demonstrate this process, we stimulated neutrophils in a different order of TNF, LPS or IL-10. 2 × 10 6 of neutrophils were incubated in vitro without stimulation, and in the presence of 100ng/ml ultrapure LPS from E. coli (serotype R515, Alexis Biochemicals), 100ng/ml human TNF (recombinant, expressed in E. coli; Sigma-Aldrich, St. Louis, MO, USA), and 100ng/ ml human IL-10 (expressed in Sf21 insect cells, Sigma-Aldrich, St. Louis, MO, USA) in RPMI 1640 for 5 h (5% CO 2 , 37 °C, humid atmosphere).In the second step, we verified the thesis by extrapolating obtained data to an adequate clinical status where circulating neutrophils are permanently exposed to pro-inflammatory factors and characterised by prolongated life spine: neutrophils isolated from the blood of sepsis patients as LPS-stimulated [20], Neuromyelitis Optica Spectrum Disorders as an autoimmune disease with pre-activated neutrophils [21] as the control of physiological response-neutrophils vulnerable to IL-10, circuiting neutrophils from periodontitis patients were used [7].Isolated neutrophils were incubated for 5 h in RPMI under similar conditions as neutrophils used in the in vitro model.There were four independent replicates for each experiment in the in vitro model.To validate results from ChIP-Seq analysis obtained from the in vitro model, three representative patients with functional test parameters closest to the mean values of their group were classified to ChIP-Seq analysis. Assumptions for PTM selection First, we selected modifications that seemed important in changing the nature of neutrophils from pro-to anti-inflammatory cells after IL-10 stimulation in a relatively short time (10 min).Therefore, we focused on PTMs marked active promotors and transcribed regions, assuming that neutrophils must be ready to differentiate the pro-vs.anti-inflammatory signal by triggering genes that engage chromatin-modifying enzymes.This process can be mediated by positioning the TSS region (H3K4me3) and active promotors (H3K36me3, H3K79me2) of transcribed regions.Simultaneously, we suspect that modification of transcription suppressor occurs to hide pro-inflammatory genes (H3K9me3, H3K27me3) and PTMs responsible for initiating genes within heterochromatin (H3S10p).Second, we selected modifications enhancing antimicrobial neutrophil properties observed during neutrophil TNF-preactivation.We suspect such neutrophils' properties must be mediated by H3K27ac -PTM, described as a transcription enhancer.Third, as neutrophils leaving the bone marrow are already prepared to neutralise pathogens, we assume that LPS stimulation directly positions TSS regions by H3K4me3. Neuromyelitis Optica Spectrum Disorders (NMOSD) Three patients with confirmed AQP4-IgG seropositive NMOSD, fulfilling the 2015 Wingerchuk criteria, were selected for ChIP-Seq analyses [22].All the patients were without systemic steroids or other anti-inflammatory drugs for at least three months before the study. Sepsis Six patients with sepsis developed at least two symptoms from the following criteria: body temperature above 38℃ or below 36℃; pulse rate > 90/min; respiratory rate > 20/ min or PaCO 2 < 32 mmHg; leukocyte > 12,000/µL or < 4000/µL were classified for investigations [23].Patients who had not received antibiotics for at least three months before the study and with positive microbial tests for E. coli in the blood were selected for ChIP-Seq analyses. Periodontitis Twelve patients with generalised stage III and IV periodontitis were selected for this study.The criteria of the Classification of Periodontal and Peri-Implant Diseases and Conditions 2017 [24] were used for periodontitis staging and classified for further investigation.Patients who had not received antibiotics or anti-inflammatory drugs for at least three months before the study were selected for ChIP-Seq analyses. All participants of the study were diagnosed and recruited at the Department of Neurology (patients with NMOSD), at the I Department of Urology (patients with sepsis) and at the Department of Periodontology and Oral Mucosal Diseases (patients with periodontitis and healthy controls), all from Medical University of Lodz. Neutrophil isolation 20 mL of the whole blood on lithium heparin anticoagulant was collected from sepsis, NMOSD, periodontitis patients and HC.Neutrophils were purified by negative selection by microbeads, which allowed the removal of DCs, B cells, monocytes, macrophages, activated T cells, and activated NK cells (MACSxpress Whole Blood Neutrophil Isolation Kit, Miltenyi Biotec GmbH, Germany).Residual erythrocytes were lysed using 2mL ammonium chloride Lysing Reagent (BD Biosciences) for 5 min.The final purity of the neutrophil population was assessed by flow cytometry using CD14-PE (clone M5E2), CD15-FITC (MMA), and CD16-PECy7 (3G8, all from BD Pharmingen) mAbs.Flow cytometric analysis of the isolated cell population showed that the percentage of CD15 high CD16 + CD14 − neutrophils was > 98%.The level of contaminating CD14 + CD15 + monocytes was about 0.4%, and CD15 + CD16 − eosinophils was < 0.1% after isolation, as we proved in our previous studies [7,9,19]. Phagocytosis and reactive oxygen intermediates (ROI) production With some modification, ready-to-use kits were used to analyse neutrophil phagocytosis and ROI production (Bursttest and Phagoburst, OrphoGen Pharma, Heidelberg, Germany).Briefly, isolated neutrophils were stimulated by different LPS, TNF and IL-10 constellations.Each stimulator was added to a 500 mL suspended cell for 10 min, 37 °C, in a water bath.The total time of neutrophil stimulation in this part of the experiment was 20 min.After incubation, samples were cooled down in an ice bath for 10 min.The ability of neutrophils to phagocytosis was performed on E. coli conjugated with FITC (3.3 × 10 8 bacteria/mL) and presented as the mean fluorescence intensity acquired from intercellular FITC signal.The ROI production was performed with or without restimulation by PMA (phorbol 12-myristate 13-acetate, 0.162 µM, 10 min) and presented as the mean intensity of 1,2,3-dihydrorhodamine (1,2,3-DHR), whose reduction depends on intracellular ROI concentration.Samples were analysed within 30 min using flow cytometry (BD LSRII, FACSDiva ™ ). H3K4me3 ChIP-Seq analysis Chromatin immunoprecipitation (ChIP) Neutrophils stimulated by LPS, TNF or IL-10 (in vitro model) and isolated from patients: NMOSD, sepsis, and periodontitis (neutrophils stimulated in vivo) were used to analyse DNA coupled with H3K4me3 histone.ChIP was carried out in neutrophils according to the manual of Magna ChIP ™ A/G Chromatin Immunoprecipitation Kit (Merck Millipore).Cells were fixed with 1% formaldehyde in RPMI solution for 10 min at RT and quenched with 10x glycine in 5-minute incubation at RT to stop the fixation.After washing with cold PBS, cells were treated sequentially with 1x Protease Inhibitor Cocktail II, Lysis Buffer with Protease Inhibitor Cocktail II, and Protease Inhibitor Cocktail II with Nuclear Lysis Buffer.The supernatant was removed, and the cell pellet was resuspended in a Nuclear Lysis Buffer.Sonication (10 cycles; 30 s. "ON" 30 s. "OFF") was done using Bioruptor ® Pico Sonicator (Diagenode, Belgium).The obtained chromatin was spun at a minimum of 10,000 x g at 4 °C for 10 min to remove insoluble material.Each immunoprecipitation required the addition of Dilution Buffer and Protease Inhibitor Cocktail II. 25 µL of the diluted chromatin as 'Input' was saved at 4 °C for further proceeding.Chromatin immunoprecipitation was performed with the use of the set of antibodies: Normal mouse IgG (negative control), anti-RNA Polymerase II (clone CTD4H8) as positive control and anti-trimethyl-Histone H3 (Lys4) (MC315, Merck Millipore) mAbs.Both antibodies were recommended for use in the ChIP-Seq technique [25].Immunoprecipitation reactions were incubated overnight at 4 °C with rotation.DNA was eluted and purified using spin columns.The DNA concentrations of obtained samples were measured by Qubit 4 Fluorometer (Ther-moFisher Scientific). Library preparation and NGS sequencing Double-stranded DNA was generated from a singlestranded fraction of ChIPed DNA using NEBNext ® Ultra ™ II Non-Directional RNA Second Strand Synthesis Module (E6111S, New England Biolabs).The reaction was performed with random primers from NEBNext ® RNA First Strand Synthesis Module (E7525, New England Biolabs).Libraries for sequencing were prepared using NEBNext ® Ultra ™ II DNA Library Prep Kit for Illumina ® (E7645L, New England Biolabs).Single-end sequencing with a read length of 75 bases (SE75) was performed with NextSeq550 (Illumina) to obtain at least 20 million reads per sample that could be mapped to the human genome [26]. Bioinformatic methodology of the ChIP-Seq analysis In the first stage, the quality of the raw sequence reads was checked using the FASTQC software (version: 0.11.8).Next, all reads were subjected to the adapter, and quality filtering (minimum quality (-q 25), minimum length (-m 15) using the Cutadapt tool (version: 1.18) in NextSeq reads mode.Trimmed reads were aligned to the reference genome (GRCh38) using the Bowtie2 (version: 2.2.9) in the single-end mode.Duplicated reads were located and tagged using the Picard MarkDuplicates tool (version: 2.18.4).Reads with low mapping quality score (MAPQ < 10) were removed from downstream analysis with the Samtools software (version: 1.6).Protein binding sites identification in the previously prepared BAM files was performed with the MACS2 (Model-based Analysis of ChIP-Seq) software (version: 2.1.0)in narrow peak mode [27].Subsequently, identified peaks were annotated using annotatePeaks.plfrom Homer software (version: 4.11.1, hg38 annotation library).Functional enrichment analysis for various categories (molecular function, biological process, cellular component and pathways interaction) was also executed [28].To find enriched motifs in ChIP-Seq peaks, the findMotifsGenome.plprogram from Homer software (version: 4.11.1) was used.The quantitative assessment of ChIP-Seq quality was checked by applying the ChIPQC package (version: 1.21.0) from R Bioconductor (version: 3.6.0).Differentially enriched sites between experimental conditions were identified using the DiffBind package (version: 2.12.0) from R Bioconductor (version: 3.6.0). ChIP-Seq library quality control analysis and assessment of ChIP-Seq quality were provided in our previous studies [9]. Statistics Arithmetic means and standard deviations were calculated for all parameters.A statistical analysis of differences was performed using the one-way ANOVA (comparison between 'n.s.' , LPS-and TNF-stimulated neutrophils) or the two-way ANOVA test (analysis of IL-10 on LPS stimulation or TNF-preactivation processes).Tukey's test was used for multiple comparisons as a post hoc test when statistical significance was identified in the ANOVA test.The statistical comparison between clinical groups (sepsis, NMOSD and periodontitis patients) and healthy volunteers was performed using the Student-t test.p ≤ 0.05 was considered as the significant difference. IL-10 protects neutrophils against pro-inflammatory stimuli In the first set of experiments, we analysed the influence of IL-10 on two main processes: reactive oxygen production after LPS stimulation and the ability of E. coli to be phagocytosed.To reflect the natural process at different times of inflammation, we assumed that neutrophils are susceptible to pro-inflammatory factors in the early period of inflammation, but in the final period, they should be sensitive to anti-inflammatory IL-10.First, we have demonstrated the phenomenon of pre-activation.Neutrophils previously exposed to TNF for 10 min were characterised by significantly greater ROI production after 10 min of LPS or PMA stimulation (Fig. 1A, B) and more efficient E. Coli phagocytosis process vs. non-preactivated cells (Fig. 1C). The pre-activation phenomenon is spectacularly demonstrated with the ROI production after PMA stimulation (81% increase in neutrophils previously exposed to TNF (Figure 1B).This effect is not simply additive, as TNF-acting alone did not stimulate ROI production.Surprisingly, we didn't observe the preactivation phenomenon during phagocytosis as no differences have been observed between LPS acting alone vs. TNF + LPS (Fig. 1C).It is worth adding that in our previous investigation, we noted a slight, but statistically significant, increase in the ROI production after TNF-preactivation [9].This discrepancy is the result of different ways of interrupting the TNF activation.In previous manuscripts, we used only one of each factor for 10 min and after that samples were washed with ice-cold RPMI to stop the reaction and remove stimuli.In this experiment, TNF-preactivated neutrophils had to wait for 10-12 min.on the ice for the rest of the samples: TNF + IL10, IL10 + TNF and then were rinsed with ice-cold RPMI. Next, we have focused on the CD11b/CD18 heterodimers (MAC-1) as the most prevalent member of β2 integrin specifically expressed on neutrophils [29].Both, adhesion molecules CD11b and CD18, increased their expression within 10 min of exposition on TNF, LPS or IL-10.The most spectacular increase has been observed during TNF exposition and subsequent in stimulation with LPS.This effect can be additive as both stimuli increased CD11b as well as CD18 molecules to a similar degree (Supplementary Fig. 1A, B). Following, we have focused on the effectiveness of IL-10 in preventing neutrophils from pre-and activation by proinflammatory stimuli estimated by the ability to ROI production, phagocytosis as well as CD11b/CD18 molecule adhesion expression.We noted that 10 min of IL-10 neutrophil stimulation made them insensitive to TNF-preactivation or direct LPS stimulation.This phenomenon is observed in all estimating parameters.IL-10 had no such properties if neutrophils were previously exposed to TNF or LPS.In Fig. 1 the right part of each graph presents ROI production and phagocytosis, while in Supplementary Fig. 1A and, B right panels of each graph and low panels demonstrated the effect of IL-10 on CD11b/CD18 expression. Another important molecule responsible for adhesion, transendothelial migration (TAM, diapedesis) and neutrophil preactivation is L selectin (CD62L).Loss of CD62L expression is used as a gold standard to assess neutrophil activation which can correspond with increased CD11b/CD18 expression [30].As expected, a 10-minute exposition of TNF and LPS decreased CD62L on the surface of neutrophil.Unlike CD11b/CD18 expression, we did not observe the effects of IL-10 on shedding CD62L (Supplementary Fig. 1C upper and middle panels).Next, we have focused on the effectiveness of IL-10 in preventing the shedding of CD62L molecules induced during TNF preactivation and LPS activation.We noted that IL-10 stimulation made neutrophils insensitive to the shedding of CD62L by TNF or LPS stimulation.Similar to the tests described above, IL-10 had no such properties if neutrophils were previously exposed to TNF or LPS (Supplementary Fig. 1C low and upper panels). These experiments elucidate why neutrophils, in acute and chronic inflammatory conditions, become insensitive to IL-10 until the inflammatory agents are removed.In both sepsis induced by Gram-negative bacteria and autoimmune diseases such as NMOSD, circulating neutrophils are persistently activated or preactivated and are not susceptible to the anti-inflammatory effects of IL-10 even though high concentrations are observed in serum (sepsis, not surviving patients) or in serum and cerebrospinal fluid in NMOSD [31,32]. We speculate that such crucial changes in neutrophil function must occur at the genomic level by blocking and/or initiating more transcription genes.Therefore, we analysed 11 histone PTMs that characterised heteroand/or euchromatin. LPS-modified neutrophil histone H3K4 via activation of lysine methyltransferase SETD1A and MLL1 As we have shown above, LPS-stimulated neutrophils were characterised, among others, by high H3K4me3 levels, which, in turn, were diminished by previous exposure to IL-10.This phenomenon suggests a prevailing role for H3K4me3 in IL-10 protective effect.Therefore, we have focused on the enzymes referred to as H3K4 'writers' [33].Histone H3 lysine can be modified by one, two or three methyl groups, and its levels and distribution reflect the combined action of methyltransferases and demethylases.In humans, H3K4 is methylated by the lysine methyltransferase SETD1A/B and MLL1-4 [34].Using ICC methods, we have noted the high expression of SETD1A and MLL1 proteins after LPS but not TNF or IL-10 exposition (Fig. 3A, upper panel of each graph).The discrepancy between the outcomes of transcription and translation probably results from the different dynamics of both processes.TNF-preactivation which is characterized by intensified SETD1A translation, but not their transcription, is physiologically justified as the process associated with preparing neutrophils for antipathogen response.In turn, the direct exposition of neutrophils to LPS results in the intensification of both processes simultaneously. The analysis of the colocalisation rate confirmed the appearance of these enzymes in the nucleus (Supplementary Fig. 3).The transcription analysis of these enzymes revealed a statistically significant increase in mRNA expression of SETD1A but not MLL1 after exposition on TNF or LPS to 'n.s.' neutrophils (Fig. 3A, upper panel).This set of experiments suggested that the induction mechanism of H3K4me3 PTMs by SETD1A depends on the positioning itself via a positive feedback mechanism contrary to MLL1, whose level is predefined and does not require transcription.The experiment with IL-10 exposition revealed that this cytokine prevents neutrophils from high expression of SETD1A and MLL1 resulting from LPS stimulation.IL-10 diminished SETD1A and MLL1 protein expression but only mRNA for SETD1A.The protective effect of IL-10 occurs only if IL-10 first acts on neutrophils before LPS or TNF-stimulation (Fig. 3A, upper right panels).This data suggests a mechanism of IL-10 preventing LPS stimulation mainly dependent on the SETD1A gene positioning and acting only when removing pro-inflammatory stimuli. IL-10 rearranges chromatin by CHD1 and JMJD2A induction The crucial aspect of the overriding mechanism reflects the changing nature of neutrophils from cells ready to neutralise pathogens into cells that reduce inflammation and start tissue regeneration after IL-10 exposition.We speculated that such changes in neutrophil function, through the activation of initially inactive genes, should result from changes at the histone rearrangement level that engage chromatin remodelling enzymes triggering genes within heterochromatin.We have selected two enzymes that we thought to be involved in neutrophil polarity reversal: chromodomain helicase DNA-binding protein 1 (CHD1) and histone demethylase (JMJD2A/ KDM4A).CHD1 cooperates with H3K4me3 as most likely associated with the rapid activation and positioning of transcription genes; in turn, JMJD2A, an enzyme that demethylates H3K9me3 positing genes within heterochromatin [35,36].The ICC analysis revealed high expression of CHD1 in neutrophils exposed to IL-10 during pre-activation by TNF as well as pre-activation and subsequent activation by LPS.It is noteworthy that prior LPS stimulation results in a blockade of CHD1 expression under the influence of IL-10.We have not observed any changes within CHD1 mRNA expression in any of the experimental setups (Fig. 3B).In the same experiment series, we noted high expression of JMJD2A in protein and mRNA levels in neutrophils exposed to IL-10.ICC analysis revealed increased expression of JMJD2A in the neutrophil nucleus exposed to IL-10 (Fig. 3C).The analysis of TNF or LPS influence on this process revealed that only neutrophils exposed first to IL-10 characterised with JMJD2A high expression, while additional exposure to LPS or TNF did not affect JMJD2A expression.In turn, neutrophils previously pre-activated by TNF or stimulated by LPS abort the influence of IL-10 on JMJD2A expression.A similar relationship was observed at the mRNA level, suggesting that induction of JMJD2A by IL-10 and blocking this process by TNF or LPS occur at the transcription level (Fig. 3C).The colocalisation of CHD1 and JMJD2A within DNA using z stock-confocal analysis has confirmed a high colocalisation rate in IL-10-stimulated neutrophils (Supplementary Fig. 4).It is worth noting that the location of CHD1 is on the periphery opposite to JMJD2 which is localised mainly in the centre of nuclei.This remark suggested acting these enzymes within different A/B segments of chromatin.Compartment A is active and located in the interior nuclear space, whereas compartment B is inactive and located in the nuclear lamina [37].The high expression of CHD1 within compartment B confirms our assumption that this enzyme is involved in triggering inactive genes within heterochromatin. Neutrophils stimulated by IL-10, LPS, or TNF-regulated gene-specific profiles of PTMs via H3K4me3 In our previous studies, using H3K4me3 ChiP-Seq analysis, we identified various transcriptional start sites (TSSs) associated with the plasticity of human neutrophils exposed to TNF, IL-10, or directly stimulated by LPS.In this study, we have used our previously obtained data from ChiP-Seq of H3K4me3 to analyse target genes and gene ontology (GO) related to PTMs [9]. First, we have focused on the H3K4me3 peak breadth distribution, giving us an overall view of positioned genes ready for transcription and requiring additional regulatory transcription DNA binding proteins within the TSS regions.The ChiP-Seq analysis of H3K4me3-enriched nucleosomes allowed us to distinguish two types of peaks within the TSS regions: sharp-narrow peaks (< 1 kb) positioned near the TSS of Exon1, not requiring additional DNA binding proteins and broad-peaks (> 4 kb) being under control by them [38].In our study, we have noted significant changes in peak distribution within H3K4me3 defined in MACS2 analysis.The average breadth of H3K4me3 peaks after TNF and IL-10 exposure was significantly shortened compared to non-stimulated or LPS-stimulated neutrophils (Supplementary Fig. 5A).The detailed analysis of several peaks, divided into 'sharp' and 'broad' , indicated that this phenomenon resulted from significantly increased sharp peaks (Supplementary Fig. 5B).This data suggests that TNF pre-activation and repolarisation of neutrophils by IL-10 introduce them as highly specialised cells that can prepare specific genes for transcription. To determine the effect of H3K4me3-marked histone on remodelling chromatin, we have performed a Gene Ontology analysis of processes engaged in PTMs.We have discovered that neutrophils, regardless of the type of stimuli used, are characterised by high positing 'Chromatin organisation' (GO:000631250) compared to non-stimulated neutrophils (Fig. 4A).The detailed analysis of target genes within this term revealed that there are 35 target genes, which are selectively positioned by IL-10 stimulation, 10 by TNF and 8 by LPS (Fig. 4B and Supplementary Table 1A).TNF pre-activated neutrophil, opposite to IL-10 or LPS stimulation, positioned genes within term related to histone acetylation: 'Histone acetyltransferase complex' (GO:0000123), 'Histone acetyltransferase activity' (GO:0004402) and 'DNA-binding transcription factors' (GO:0140297) (Supplementary Fig. 6).Detailed analysis of particular genes within terms 'Histone acetyltransferase complex (GO:0000123)' and 'Histone methyltransferase complex' (GO:0035097) revealed that more genes are positioned by H3K4me3 regardless of the stimuli used (Supplementary Table 1B).In turn, LPS stimulation was characterised by positioning target genes within two processes: 'Histone H3K4-trimethyltransferase activity' (GO:0080182) and 'Histone H3K4 demethylase trimethyl H3K4 specific' (GO:0034721).The comprehensive analysis within the term 'Histone H3K4-trimethylation' highlighted that all target genes are located in the common set (9 from 15 all in term) and 6, which characterised activated neutrophils regardless of the kind of stimuli (Fig. 4B right graphs).Contrary to LPS or TNF, IL-10 stimulated positioning target genes within terms 'Histone demethylase activity H4K9 specific' (GO:0032454) (Fig. 4A and Supplementary Fig. 6).The detailed analysis within this term highlighted that all target genes are located in the common set (8 from 11 all in term), 2 are characterised by positioning activated neutrophils regardless of stimulation (Fig. 4B) and one: KDM4D specific form for IL-10 or TNF exposition. Next, we have focused on the ChiP-Seq H3K4me3 peak density readings within the TSS region of genes MLL1, SETD1A, CHD1 and JMJD2A.LPS-stimulated neutrophils, contrary to 'n.s.' , IL-10, TNF exposition was characterised by high peak density within Exon 1 coded MLL1 and SETD1A.Neutrophils exposed to IL-10 were characterised by high peak density within CHD1 and JMJD2A (Fig. 4C).ChiP-Seq, together with ICC and mRNA analysis of these enzymes, revealed an overriding mechanism related to neutrophil polarisation during different stages of inflammation. IL-10 stimulation triggered gene positioning via H3K4me3 and coded CHD1 and JMJD2A-enzymes remodelling chromatin to initiate transcription of gene hidden within heterochromatin and/or marked by H3K9me3.Contrary to IL-10, LPS stimulation does not require such chromatin rearrangement; genes positioned by H3K4me3 coded methyltransferase enzymes amplify anti-microbial properties by direct gene positioning within H3K4me3 (probably in a positive feedback manner).In turn, TNF-preactivation prepares neutrophils for a more effective response against pathogens by engaging target genes associated with binding transcription factors to DNA and target genes within acetylation of is the process that dominates regardless of whether neutrophils are exposed to pro-or anti-inflammatory factors.Pro-vs.anti-inflammatory stimuli differences are determined via H3K4me3 'de positioning' methylation of histone H3K9.The green arrow points to the demethylation of H3K9 as the process triggered upon exposure to IL-10.The red one points to H3K4me3 trimethylation as the parent process of neutrophil polarisation regardless of stimuli.Data are presented as mean values calculated from four independent experiments.B Binding site overlap in the GO term 'Chromatin Organization' , 'Histone H3K4-trimethylation' and 'Histone demethylase activity H3K9 specific'; n.s, non-stimulated.The analysis of target genes in the GO terms Chromatin organisation are presented in Supplementary Table 1.C The comparison of peak density within PTM enzymes revealed high signal within TSSs (Exon 1) of MLL1 and SETD1A in neutrophils stimulated by LPS and increased density within CHD1 as well as JMJD2A in neutrophils exposed to IL-10 H3K27ac responsible for positioning genes susceptible to enhancers. Validation of In Vitro data to clinical status Finally, we have verified the obtained data by their extrapolation to adequate clinical status, using neutrophils derived from patients with sepsis (systemic septic inflammation with LPS-stimulated neutrophils), NMOSD (aseptic inflammation with pre-activated neutrophils), and periodontitis (local self-limiting septic inflammation with IL-10-positive neutrophils). In the first comparison, ICC analysis of enzyme expression has shown a correspondence between the in vitro model and a particular disease.SETD1A and MLL1 revealed a high expression and colocalisation within DNA in sepsis patients, corresponding with the in vitro model of LPS-stimulated neutrophils (Fig. 5A).In turn, neutrophils isolated from periodontitis were characterised by high expression of CHD1 and JMJD2A, which corresponds with the in vitro model of neutrophils exposed to IL-10 (Fig. 5B).Surprisingly, we have also observed high expression of MLL1 in NMOSD and periodontitis patients, but without explicit colocalisation Fig. 5 A Circulating sepsis neutrophils, similar to the in vitro LPS model, induced inflammation characterised by high expression of SETD1A and MLL1 and their colocalisation to the nucleus.B In turn, periodontitis neutrophils, similar to the in vitro IL-10 model, stimulated HC neutrophils with high expression of JMJD2A and CHD1 (A and B upper panels).The example of ICC analysis and confocal 3D projection confirmed colocalisation of SETD1A and MLL1 within the nucleus in sepsis patients, while colocalisation of JMJD2A and CHD1 within the nucleus mainly in periodontitis patients (white arrows) (A and B right panels).Data are presented as mean ± SD calculated based on 3 cases of patients with NMOSD, 6 with sepsis and 12 with periodontitis within DNA, which may be related to MLL1 and SETD1A synthesis, but with limited capacity to colocalisation into cell nucleus (Fig. 5A, right panel). Next, we have focused on GO processes positioned by H3K4me3 and associated with chromatin organisation that emerged in vitro experiments.We have noted that high positioning target genes characterised neutrophils isolated from patients with NMOSD within terms 'Histone acetyltransferase complex' (GO:0000123) and ' Acetyltransferase activity' (GO: 0004402), which corresponds to TNF-preactivated neutrophils.In turn, neutrophils isolated from sepsis patients highly positioned genes associated with 'Histone methyltransferase activity' (GO:0008168), in particular, 'Histone methyltransferase activity H3K4me3 specific' (GO:00422800), which corresponds with in vitro model of LPS-stimulated neutrophils.Similarly, neutrophils isolated from patients with periodontal diseases highly positioned genes associated with 'Histone demethylase activity' (GO:0032454), particularly 'H3K9 specific' (GO:0031056).Regardless of the type of disease, all clinical groups were characterised by high positioning of target genes determined by 'Chromatin organisation' (GO:0006325) (Fig. 6A).Although we noted a high correlation in this point of GO analysis, the particular target gene analysis within this term shows significant differences compared to the in vitro model.The vena plot analysis within terms: 'Chromatin organisation' , 'Histone H3K4me3 trimethylation' (GO:0080182) and 'Histone demethylase activity H3K9 specific' (GO:0032454)' revealed that more target genes are located in the common compartments representing target gene positioning whether the neutrophils are isolated from HC, NMOSD, sepsis or periodontitis patients (Fig. 6B).Moreover, the compartment analysis within the 'Chromatin organisation' term also pointed to the target genes that are specific for NMOSD (4 genes), sepsis (25 genes) or periodontitis (4 genes) (Fig. 6B).Among these genes, we noted those that additionally correlate with the in vitro model: KDM4D (Lysine Demethylase 4D -JMJD2D) for periodontitis patients, AURKA (aurora kinase A) for NMOSD and HLTF (Helicase Like Transcription Factor) for sepsis patients (Supplementary Table 2 green font).In the next step of the bioinformatic analysis, we have focused on comparing peak density within TSSs (Exon 1) of PTM enzymes: MLL1, SETD1A, JMJD2A and CHD1.In this comparison, we have noted only partial correspondences with the in vitro inflammation model.Especially, the high signal of MLL1, SETD1A, but also CHD1 and JMJD2A is observed in the course of sepsis which did not correlate with LPS stimulation resulting in MLL1 and STED1A high signal only.Instead, we did not observe CHD1 in the course of NMOSD which did not correspond with TNF-preactivation in the in vitro model.This difference is probably due to the limitation of the inflammation model.In the course of sepsis, neutrophils are exposed to other factors, including proinflammatory cytokines, heat shock proteins, and complement components (C5a, C3a).These factors probably affect neutrophils simultaneously.In turn, in the course of NMOSD, neutrophils are preactivated mainly by C5a, but not TNF, which probably engages different pathways for their preactivation [21]. High density within CHD1 as well as JMJD2A is observed in periodontitis which accurately corresponds with the in vitro model-neutrophils exposed to IL-10 (Fig. 6C).To summarise, this data may suggest that changes within the target genes of these enzymes are quantitative; thus, gene expression depends on the number of positioned genes within DNA but is not associated with positioning the new one. Finally, we have compared peak breadth distribution within genes positioned by H3K4me3, and we noted a decrease in the average breadth as well as more 'sharp-narrow' peaks in patients with NMOSD and periodontitis, which corresponds with TNF-preactivated and IL-10-stimulated neutrophils (Supplementary Fig. 5C, D). In the last step of validating the in vitro model to adequate clinic status, we examined the level of two PTMs we had expected to be remarkably characterised by preactivated neutrophils and these exposed to IL-10.For pre-activated neutrophils, we have examined the level of H3K27ac responsible for positioning enhancers.For IL-10 stimulation, we estimated the level of H3S10p, which, in turn, was related to triggering genes within (See figure on next page.)Fig. 6 A Gene Ontology analysis of neutrophils isolated from NMOSD, sepsis and periodontitis patients confirmed that H3K4me3 positions target genes within the main processes associated with chromatin organisation.NMOSD neutrophils are characterised by positioning target genes related to histone acetyltransferase activity that corresponds with TNF-preactivated neutrophils (blue arrows), sepsis neutrophils positioned target genes associated with histone methyltransferase, especially H3K4 specific what correspond with LPS-stimulated neutrophils (red arrows).In turn, periodontitis neutrophils positioned target genes within the term 'Histone demethylase activity' , particularly H3K9 (green arrows), which correlates with IL-10-stimulated neutrophils.B Binding site overlap in the GO terms 'Chromatin organisation' , 'Histone H3K4 trimethylation' and 'Histone demethylase activity H3K9 specific' .C The comparison of peak density within PTM enzymes revealed high signal within TSSs (Exon 1) of MLL1 and SETD1A in sepsis circulating neutrophils, high density within CHD1 as well as JMJD2A in periodontitis and sepsis which reflects in vitro model of IL-10-stimulated, but not in LPS-stimulated neutrophils heterochromatin.Due to the limited number of cells obtained from patients, we have used ICC despite the dot blot method.We have noted high expression of H3K27ac in neutrophils isolated from NMOSD patients, which characterised pre-activated neutrophils.In turn, neutrophils isolated from periodontitis were characterised by a high level of H3S10p, which corresponds with neutrophils exposed to IL-10 (Supplementary Fig. 7).In our previous investigation using the same technique, clinical samples and in vitro model, we had shown that neutrophils isolated from sepsis patients, opposite to HC, were characterised by a high level of H3K4me3, which corresponds with the in vitro model of LPS-stimulation [9]. Discussion One of the crucial questions raised in our research concerns the overriding mechanism that reflects the changeable nature of neutrophils from cells ready to neutralise pathogens to those reducing inflammation and initiating tissue regeneration after IL-10 stimulation.As we have shown in our previous studies, the neutrophil polarisation to immunosuppressive, restricted inflammatory cells is associated with initiating gene transcription dependent on alternative NF-kB pathway activation with simultaneous inhibition of classical NF-kB pathway activation.In the current research, we have shown that this process results from the reduction in the properties of neutrophils to E. coli phagocytosis, ROI production and CD11b/CD18 expression after TNF preactivation or LPS stimulation.In our opinion, it is physiologically justified, as maintenance of neutrophils' anti-infection potential is high energy and unnecessary for cytokine/chemokine or growth factor synthesis during tissue regeneration [9].We speculated that such a significant change, which completely alters their function, can go through the activation of primarily inactive genes.It should result from significant changes at the level of histone rearrangement, engaging chromatin remodelling proteins and gene silencing within heterochromatin.From the four protein subfamilies related to chromatin remodelling (ISWI, CHD, SWI/SNF and INO80), we designated chromodomain helicase DNA-binding protein 1 (CHD1) as the enzyme considered an effector of active histone modification by specific recognition of H3K4me3 residues produced by Set1 [39].Therefore, CHD1 is regarded as a chromatin remodeler, strongly associated with transcription and nucleosome turnover downstream of the TSS [35].Johnsen and colleagues show that CHD1 is required to induce osteoblast-specific gene expression, extracellular-matrix mineralisation and ectopic bone formation in vivo, which directed this molecule on the high potential in cell polarity [40].In our study, we noted that neutrophils exposed to IL-10 were characterised by high CHD1 protein and mRNA expression and high NGS density reading within the TSS region of CHD1 genes marked by H3K4me3.Another molecule that we suspected to be related to IL-10-induced neutrophil polarity is histone demethylase JMJD2A, also known as KDM4A, which promotes gene transcription by antagonising H3K9 specific Lys9 residues histone H3 methylation marked heterochromatin [41,42].Similarly to CHD1 analysis, we have noted that neutrophils exposed to IL-10 were characterised by high JMJD2A protein and mRNA expression as well as high NGS density reading within the TSS region of JMJD2A genes marked by H3K4me3.Opposite to CHD1, JMJD2A seems to be not involved in neutrophil TNF-preactivation. LPS added to neutrophils previously exposed to IL-10 prevents inducing CHD1 and JMJD2A by IL-10.As both enzymes are-actively transcribed and require TSS positioning by H3K4me3, alteration within poisoning H3K4me3 by stimulating neutrophils with LPS blocks active transcription of CHD1 and JMJD2A by IL-10 stimulation.This hypothesis is also moderately supported by studies of circulating neutrophils isolated from sepsis, NMOSD and periodontitis patients.Neutrophils isolated from periodontitis that respond correctly to IL-10 were characterised by high CHD1 and JMJD2A expression (ICC and ChiP-Seq H3K4me3 results), opposite to sepsis-derived neutrophils, with no changes within these enzymes.We have also noted high CHD1 expression in NMOSD neutrophils, which corresponds with the result from the in vitro model of TNF pre-activation.Investigating further mechanism that prevents the action of IL-10 in an ongoing inflammatory reaction, we have drawn attention to PTMs induced by direct exposure to LPS.LPS-stimulated neutrophils, on the one hand, were characterised by lower levels of H3K9me3, H3K79me2 and H3K27me3-the profile of histone modification responsible for gene repression, on the other hand, by high levels of H3K4me3-PTMs, which position active transcribed genes within the end of Exon1.In our opinion, this is not a natural process of eliminating pathogens because first, these cells must be pre-activated by a low concentration of pro-inflammatory factors.The pre-activation process probably prevents neutrophils from being activated in the blood, gives additional time for chemotaxis, diapedesis, and enhances processes involved in antimicrobial activity.This theory is confirmed by our analysis of functional tests (ROI production and phagocytosis), where neutrophils entering blood vessels are already prepared for a response in contact with the pathogen; however, this response is less intense than before TNFpreactivation.Pathological, direct LPS activation can be observed in the periphery of sepsis patients.The analysis of neutrophils isolated from blood patients with sepsis revealed the correlation with the in vitro model within the expression of MLL1 and SETD1A, as well as analysis of GO processes positioned by H3K4me3 related to chromatin remodelling.In addition, we have also presented an in vitro model that direct LPS stimulation 'blocks' IL-10-induced suppression of ROI production, phagocytosis, CD11b/CD18 and CD62L adhesion molecules expression.A series of in vitro experiments and neutrophils isolated from blood sepsis patients prompted us to hypothesise that neutrophils directly exposed to LPS trigger the fast pathway of gene transcription through the involvement of H3K4me3-specific methyltransferases.This process is irreversible, avoiding chromatin rearrangement by IL-10.Probably, this phenomenon has severe clinical implications in the course of the second phase of sepsis, during compensatory anti-inflammatory response syndrome (CARS), as neutrophils isolated from the blood of these patients were characterised inter alia: abnormal high number immature neutrophils, delayed apoptosis and uncontrolled synthesis of pro-inflammatory factors despite of high concentration of anti-inflammatory IL-10 in serum [43,44]. On the other hand, the proper reaction associated with eliminating pathogens preceded by pre-activation can also lead to pathological conditions.In the course of autoimmune diseases such as NMOSD, diabetes mellitus type 1 (DM1), and rheumatoid arthritis (RA), circulating neutrophils are permanently pre-activated with enhanced properties to phagocytosis, ROI production, high cell surface adhesion of molecule expression in ex vivo stimulation as well as prolongated life spine.Similar to the case of sepsis, the high concentration of IL-10 observed in the neutrophil environment, does not prevent their preactivation and its negative effects.A high concentration of anti-inflammatory IL-10 is observed in serum rheumatoid arthritis RA; high in serum and cerebrospinal fluid in NMOSD, high in culture supernatants LPS-stimulated neutrophils isolated from DM1 patients [45][46][47].This study revealed that H3K4me1 and H3K27ac, which characterised patterns of PTMs for enhancers, mark TNFpreactivated neutrophils.Furthermore, we have shown that the process of H3K27 acetylation (H3K27ac), but not H3K4 monomethylation (H3K4me1), is triggered by positioning genes within H3K4me3.We concluded that H3K27ac enables neutrophils to have a much stronger and more targeted cellular response to proinflammatory cytokines arriving at the site of inflammation.This thesis, based on an in vitro model pre-activation, is reflected in neutrophils isolated from blood NMOSD patients which are characterized by high levels of H3K27ac. IL-10-stimulated neutrophils are characterised by PTM-marked active promotors, transcribed regions and enhancers.Although this modification sometimes coincides with pre-activation or LPS activation, the level of IL-10-induced PTMs is the most diverse.Opposite to TNF or LPS, it is additionally characterised by high phosphorylation of histone H3 at serine 10 (H3S10p).H3S10 phosphorylation is involved in chromatin compaction during mitosis, meiosis, and chromatin relaxation after transcription activation [48].Although H3S10P is referred to as a marker of mitosis, it can also function as an 'open-chromatin factor' in interphase for a subset of genes, allowing many different elements to access the chromatin, keeping them in a more open state to enable transcription [49,50].The H3S10p cooperated with histone H3 at lysine 9 (H3K9me3) via heterochromatin protein beta (HP1b), a family of heterochromatic adaptor molecules implicated in both gene silencing and supra-nucleosomal chromatin structure during mitosis.This inhibition phenomenon of the ability of H3K9me3 to bind HP1b by H3S10p is responsible for heterochromatin reorganisation [51].As in our experiment, we have observed high levels of H3K9me3 and H3S10p in IL-10-stimulated neutrophils, and we speculated that this modification could also be responsible for the activation of genes ordinarily unavailable during cell activation by proinflammatory factors.This phenomenon is probably essential for protecting tissue regeneration by restricting inappropriate inflammation-promoting genes. The neutrophil TNF-preactivation or IL-10 stimulation results in a specific-ultimate gene reorganisation for transcription also reflected in peak lengths within Exon 1 of H3K4me3 histone.H3K4me3-marked nucleosomes are detected as sharp-narrow peaks flanking the TSSs at the end of the first exon, strictly correlating with transcriptional activity.In our study, we have revealed that this kind of peak significantly increases with TNF pre-activation or after IL-10 exposure, which, in our opinion, proves the extremely polarised state of neutrophils.In turn, unstimulated neutrophils characterised by H3K4me3-marked nucleosomes were detected as broad peaks associated with many histone modifications.Genes in broad H3K4me3 peaks are involved in cell plasticity and their polarisation to specialised cells.This process is responsible for the adaptation of the cell to changing environmental conditions and seems critical in the polarisation of immunocompetent cells during inflammation.The broad peaks within H3K4me3 are the purpose of the action of enhancers, super-enhancers and transcription factors [38].Therefore, the large number of broad H3K4me3 peaks proves that resting neutrophils can make rapid polarisation in one of the two possible opposite manner: maximising response to pathogens initiated by TNF-preactivation or starting neutrophil pro-tissues regeneration triggering by IL-10.On the other hand, we did not observe any changes within H3K4me3 peak distribution in LPS-stimulated neutrophils, which can indicate that neutrophils passing into the bloodstream do not require such changes within H3K4me3.This hypothesis is reflected in the clinic, the average breadth and number peaks of H3K4me3 decrease in NMOSD, and there is no difference in sepsis patients within breadth peaks of H3K4me3. The significant role of nuclear chromatin modifying enzymes during various periods of inflammation is evidenced by paired comparison to monocytes as the immune cells with shared ontology [52].The comparative analysis of eQTL (expression quantitative trait loci) maps of neutrophils and monocytes from 4 to 8 individuals by the BLUEPRINT consortium, revealed that they have substantial overlap in regions of the genome that are methylated and similar profile histone modifications [52,53].The differences in gene expression within both populations probably result from differences within enhancers and PTM enzymes.This hypothesis is confirmed by research with the ontology approach.J. Knight groups analysed enhancer maps generated by CAGE-Seq in FANTOM [54,55] and revealed that neutrophils opposite to monocytes are much more enriched in the enhancer region [52], therefore they are more sensitive to environmental change and adapting to it more variously.These data also elucidate the phenomenon that both cells in similar environments, at the same time, exert different biological effects within periods of immunity.This phenomenon is additionally confirmed in the clinic.The same group of scientists revealed that the allele at rs2240335 within PADI4 affects rheumatoid arthritis susceptibility by increased expression of PADI4 in neutrophils but reduced in monocytes.This opposite mRNA expression between both leukocytes is the result of the marked region within PADI4 by H3K27ac and H3K4me1 in neutrophils but not in monocytes [52].These studies confirm our hypothesis that in autoimmune diseases permanently preactivated neutrophils which are characterized by a high level of H3K27ac and H3K4me1.Although we used neutrophils isolated from patients suffering from NMOSD as an example of autoimmune disease, we also noted high levels of H3K27ac.As we presented in our other studies, NMOSD C5a-preactivated neutrophils reverse astrocyte glutamate pump into extracellular space, as a consequence of pathological glutamate removal from extracellular space [21]. Conclusions Our investigation, directed at two main phenomena, brings us closer to understanding the physiological nature of neutrophils in tissue regeneration and their pathological role in sepsis and autoimmune diseases.First, the profile of PTMs determined the actual status of neutrophils during infection.Newly emerging neutrophils in the periphery are characterised by constitutively high expression of H3K4me3, which possess TSS regions associated with antimicrobial properties and genes coding histone PTM enzymes responsible for adapting neutrophils to the current inflammation state.Second, IL-10-induced JMJD2A and CHD1 can encourage genes desired in the regeneration of tissues after infection and prevent neutrophils from LPS-activation or TNF-preactivation.This phenomenon elucidates that the signal from the inflammation site can completely change the character from preprepared to neutralise pathogens with high antimicrobial potential into neutrophils with desired properties in tissue regeneration.LPS stimulation triggers a positive feedback mechanism that enhances the synthesis of enzymes that maintain histone-H3-lysine-4 methylation on transcriptionally active promoters (SETD1A and MLL1) by positioning H3K4me3 itself.Because transcription of JMJD2A and CHD1 also depends on TSS positioning by H3K4me3, neutrophils before LPS stimulation become insensitive to IL-10.The second conclusion is that these changes are irreversible.Neutrophils pre-activated by TNF or directly stimulated by LPS become insensitive to the antiinflammatory effects of IL-10, and vice versa; IL-10 protects neutrophils against proinflammatory stimuli.There are practical implications in a new treatment approach to patients with sepsis, that should take into account medical interventions already at the stage of myelopoiesis before neutrophil exposure to LPS. Fig. 1 Fig. 2 Fig.1IL-10 protects neutrophils against pre-and activation provided they had not been previously exposed to pro-inflammatory agents such as TNF or LPS.A The direct exposition of TNF, LPS, IL-10 on ROI production and the effect of IL-10 on the neutrophil pre-and activation process.The double-headed red arrow demonstrates the pre-activation phenomenon by TNF amplifying antimicrobial neutrophil properties.The low panel presents one of four independently performed experiments.B To better demonstrate the protective effect of IL-10 on the pre-activation, neutrophils were additionally stimulated by PMA, which directly activated NADPH oxidase.C The exposition of TNF, LPS, IL-10 on E. coli phagocytosis and the effect IL-10 on TNF-preactivated and LPS-stimulated neutrophils.The double-headed red arrow demonstrates the influence of TNF, LPS, IL-10 and TNF + LPS on neutrophil E. coli phagocytosis (left part of the graph).The right part of the graph (demarcated by a dashed line) demonstrates the protective effect of IL-10 on TNF-pre-activated and LPS-activated neutrophils and the disruption of this process due to previous short-term exposure to TNF or LPS.The rights panels present the most representative example of four independent experiments (See figure on next page.) Fig. 3 Fig. 3 IL-10 prevents LPS direct stimulation and TNF-preactivation of neutrophils by regulating PTM-modified enzymes.A Increases in expression of MLL1 and SETD1 proteins, being the result of direct LPS effect (left part of the graph), are neutralised by IL-10 (right part of each graph).The protective effect of IL-10 occurs only when IL-10 first acts on neutrophils.The upper panels presented average protein levels and low panel mRNA expression ± SD. * -statistically significant differences within: n.s.; TNF, LPS, IL-10 or TNF + LPS samples.# -statistically significant differences within IL-10 + TNF, TNF + 10, IL-10 + LPS or LPS + IL-10.A separate statistical comparison within protein levels is marked with a dashed line.B CHD1 is overexpressed during direct exposure of neutrophils to IL-10 or TNF, but not LPS.Co-stimulation of neutrophils with IL-10 and LPS disturbs CHD1 expression regardless of the order of LPS administration.C IL-10, opposite to LPS or TNF exposition, triggered JMJD2A Fig. 4 Fig.4 Neutrophils stimulated by IL-10, LPS, or TNF-regulated gene positioning specific profile of PTMs via H3K4me3 during different states of inflammation (pre-activation, activation or suppression).A Gene Ontology analysis of the Biological processes responsible for chromatin organisation mediated methylation and demethylation within histone H3 by positioning H3K4me3 methylation.Methylation of H3K4me3 is the process that dominates regardless of whether neutrophils are exposed to pro-or anti-inflammatory factors.Pro-vs.anti-inflammatory stimuli differences are determined via H3K4me3 'de positioning' methylation of histone H3K9.The green arrow points to the demethylation of H3K9 as the process triggered upon exposure to IL-10.The red one points to H3K4me3 trimethylation as the parent process of neutrophil polarisation regardless of stimuli.Data are presented as mean values calculated from four independent experiments.B Binding site overlap in the GO term 'Chromatin Organization' , 'Histone H3K4-trimethylation' and 'Histone demethylase activity H3K9 specific'; n.s, non-stimulated.The analysis of target genes in the GO terms Chromatin organisation are presented in Supplementary Table1.C The comparison of peak density within PTM enzymes revealed high signal within TSSs (Exon 1) of MLL1 and SETD1A in neutrophils stimulated by LPS and increased density within CHD1 as well as JMJD2A in neutrophils exposed to IL-10	v3-fos-license
2024-02-29T14:12:51.424Z	2024-02-27T00:00:00.000	268041963	{ "extfieldsofstudy": [ "Biology" ], "oa_license": "CCBY", "oa_status": null, "oa_url": null, "pdf_hash": "9303d2777908282687dd7820a40db1522025214a", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45323", "s2fieldsofstudy": [ "Biology", "Engineering", "Medicine" ], "sha1": "e8fa5e6295e72bc7a6aa13b2bfc7fa6628e7f6ad", "year": 2024 }	pes2o/s2orc	Directed biomechanical compressive forces enhance fusion efficiency in model placental trophoblast cultures The syncytiotrophoblast is a multinucleated structure that arises from fusion of mononucleated cytotrophoblasts, to sheath the placental villi and regulate transport across the maternal–fetal interface. Here, we ask whether the dynamic mechanical forces that must arise during villous development might influence fusion, and explore this question using in vitro choriocarcinoma trophoblast models. We demonstrate that mechanical stress patterns arise around sites of localized fusion in cell monolayers, in patterns that match computational predictions of villous morphogenesis. We then externally apply these mechanical stress patterns to cell monolayers and demonstrate that equibiaxial compressive stresses (but not uniaxial or equibiaxial tensile stresses) enhance expression of the syndecan-1 and loss of E-cadherin as markers of fusion. These findings suggest that the mechanical stresses that contribute towards sculpting the placental villi may also impact fusion in the developing tissue. We then extend this concept towards 3D cultures and demonstrate that fusion can be enhanced by applying low isometric compressive stresses to spheroid models, even in the absence of an inducing agent. These results indicate that mechanical stimulation is a potent activator of cellular fusion, suggesting novel avenues to improve experimental reproductive modelling, placental tissue engineering, and understanding disorders of pregnancy development. www.nature.com/scientificreports/Syncytin, cytokines, proteases, growth factors and oxygen levels play a well-established role in successful fusion [19][20][21] .Conversely, the biochemical factors encountered during diseases including intrauterine growth restriction, gestational diabetes, and pre-eclampsia are known to disrupt fusion and impair villous formation 22 .While the effects of these cues on fusion processes have been well-studied, the impact of biomechanical factors on fusion, particularly those that arise during villous development, remains less clear. Mechanical forces arising from budding morphogenesis have been established to play an important role in directing cell function in the gut 23 and lungs 24,25 for example; and are known to affect fusion in systems such as muscle 26 and bone 27 .Such stress patterns can spontaneously arise during 3D tissue morphogenesis [28][29][30] to directly impact cell function 31 .In placental models specifically, endogenous mechanical stress patterns have been computationally predicted to arise from trophoblast proliferation and fusion 32 , and such endogenous stresses have been experimentally shown to affect trophoblast fusion via models that tune substrate stiffness and colony sizes [33][34][35] .Hence, it seems likely that the spatially dynamic biomechanical stress profiles that arise during villous development could play an important role in the fusion process. In this work, we ask whether fusion is associated with altered biomechanical stresses and show that sites of fusion in a trophoblast model of monolayer fusion are accompanied by spatial patterns of mechanical stress consistent with those expected from the development of bud-like projections.We then recreate these patterns of dynamic stress and demonstrate that the application of externally generated, spatially directed compressive stresses, similar to those that have been computational predicted to occur during budding morphogenesis of the placental villous tree, enhances fusion efficiency.We extend these findings to a 3D spheroid culture model, and demonstrate that low levels of compressive stress can enhance fusion even without a biochemical induction method of forskolin 36 . Increased compressive radial stresses arise at sites of induced fusion in culture While the mechanical impact of processes such as proliferation and apoptosis have previously been understood in biological systems, the mechanical features that accompany a trophoblast fusion event have not been previously studied.Here, using traction force microscopy (TFM), we aimed to quantitatively measure the stresses associated with stochastic fusion events observed after forskolin-induction in a classical choriocarcinoma cell model system of trophoblast fusion (Figs.1A and 2A).A live monolayer of BeWo cells was imaged under forskolin stimulation, and assessed for fusion by identifying those regions positive for Syndecan-1, a marker for successful fusion 37,38 .In this way, regions exhibiting fusion can be distinguished from those that remain mononucleated, and the history of mechanical stresses in those regions can be quantified. As expected, we observed stochastic appearance of small fused (syncytial) patches across the fenestrated monolayer, amidst non-fused (mononucleated) regions (Fig. 2B; overall fusion efficiency of 38 ± 3% across 3 independent samples).We then considered stresses encountered during the prior 48 h around syncytial and mononucleated areas, and quantified the radial and tangential components of these stresses in relation to the centroid of the syncytial or mononucleated patch (Fig. 1B-D; Supplementary Fig. S1).This approach allowed us to quantify compression and tension in the radial (inwards/outwards) direction (radial stresses or S R ) as well as circumferential stresses around the edges (tangential stresses or S T ). Changes in both radial and tangential traction stresses were observed within ~ 6 to 12 h after induction of fusion and then stayed relatively constant for the duration of the experiment (Fig. 2C,D,F,G).To quantify the mechanical stress levels without capturing this transition period, stresses from 12 to 48 h were averaged for comparison purposes (Fig. 2E,H).A statistically significant increase in compressive radial stresses (directed into the region of interest) was observed around syncytialized regions, but not in mononucleated sites of equivalent area (Fig. 2E).Similarly, an increase in tangential stresses (directed around the region of interest) was observed (Fig. 2F-H) for syncytialized regions.This pattern of biaxial compression towards a radial region matches the stresses predicted to arise during villous bud formation due to local proliferation of cells during morphogenesis 17,34 , suggesting that the process of fusion itself may also contribute to sculpting the villous bud in situ.However, whether these forces that arise in situ are merely correlated with, or a necessary component of the fusion process remains unclear and can only be addressed by recreating these dynamic mechanical stresses in culture. Externally applied uniaxial compression does not affect fusion efficiency To externally apply dynamic mechanical stresses to cells in monolayer culture, we utilized a commercially available elastomeric platform (CellScale MechanoCulture FX) to apply an externally defined strain-based deformation over time.The substrate strains are then expected to apply stresses to the adherent cells, which can then respond and remodel in response to the applied load.We first tested external uniaxial compressive strains, which should apply an appropriate load without matching the spatial patterns of stress observed in the stochastic fusion stress measurement experiments.A loading profile of 10% compression was selected based on the displacements observed in the traction force microscopy experiments; and applied in steps over two days (Fig. 3A,B).As negligible fusion was observed for BeWo cells in standard media, we added forskolin to all cultures for these experiments to compare the relative change in fusion efficiency due to compression.Interestingly, we found that uniaxial compression did not have any effect of fusion efficiency compared to a mechanically static control, based on analysis of the fraction of nuclei in fused syncytia, as assessed by the presence of an intact E-cadherin junctional network (Fig. 3C-E).We further confirmed that the fraction of nuclei in regions positive for Syndecan-1 expression did not change between mechanically static and uniaxial compression conditions (Fig. 3F-H).This would suggest that compression along a single axis is not sufficient to induce fusion.This is perhaps due to cell reorganization along an orthogonal axis to relieve the unidirectional stresses applied within the cell sheet.If true, this would further suggest that the spatial pattern of force is a key parameter in achieving high levels of fusion in this model. Equibiaxial compression increases fusion efficiency whereas equibiaxial tension does not In order to apply mechanical compression and tension along both the radial and circumferential directions of a circular elastomeric cell culture substrate (Fig. 4A), we used a custom iris-like equibiaxial stretching system 39 .To control for possible confounding effects of substrate stiffness in these experiments, we fabricated a stiffnessmatched custom silicone mixture in a petri dish to serve as a static control.No statistically significant differences in mechanical stiffness were observed between the commercially prefabricated stretch dishes and cast silicone elastomer static petri dish samples (Supplementary Fig. S2A-E), but this is likely due to the considerable variation observed across the samples tested.However, it is important to note that since substrate stiffnesses greater than 50 kPa have been shown to induce tissue culture plastic-like morphological phenotypes in BeWo cells 33 , these substrates can be considered sufficiently stiff for comparative purposes.Together, this data demonstrates that the control conditions can be considered as stiffness-equivalent substrates so as to isolate the effects of externally applied mechanical stress on the cell cultures.Furthermore, to control for any potential confounding effects of www.nature.com/scientificreports/cellular density or space restriction in affecting fusion, we adjusted the initial seeding densities such that the final nuclear densities after mechanical deformation were similar (Supplementary Fig. S2F).Similar to the uniaxial compression experiment, BeWo cells were subjected to 10% equibiaxial strains in either compression or tension over 2 days (Fig. 4B).A statistically significant increase in fusion was observed by analysis of both E-cadherin expression (Fig. 4C-F; Supplementary Table S1), and Syndecan-1 expression (Fig. 4G-J; Supplementary Table S2) only in equibiaxial compression conditions, and not in equibiaxial tension conditions.Taken together, these results confirm that for mechanical stimulation to enhance fusion, the stress fields must recreate both the directional (compressive) and spatial (biaxial) stresses as those predicted computationally 34 and experimentally observed to arise spontaneously during stochastic fusion in monolayer culture (Fig. 2C,F). 3D compression confirms mechanical forces from compression positively influence fusion To determine whether our findings extend to 3D culture models, we investigated the effect of compression on syncytiotrophoblast formation in a spheroidal BeWo model.Trophoblast organoids have recently emerged as important tools to study various aspects of placental biology including self-assembly, expansion and differentiation 15,16,35,40 .However in these models, fusion takes place at the core of the spheroids which is opposite to the in vivo situation 15,16 .We have previously shown that the spheroid core is subjected to compressive stresses generated by a 'skin' of tension formed on the spheroid surface 28 .Hence, we wondered whether the state of internal compression present in trophoblast organoids might drive internal syncytialization and sought to manipulate the internal compression in such systems. To experimentally manipulate compressive stresses in 3D culture, we subjected BeWo spheroids to osmotically induced compression (Fig. 5A).Briefly, long-chain molecules in the cell culture media such as dextran are able to alter the osmotic pressure of a solution.Biological objects such as spheroids permit water transport but prevent dextran movement.This establishes an osmotic pressure gradient across the tissue boundary, through which the tissue undergoes compression in an isotropic stress field.This has been previously shown to significantly affect biological systems through processes such as reduced proliferation 41 or Wnt-mediated stem cell renewal 42 .This technique has the advantage of generating mechanical stresses without complex equipment or structures that is often required for physical manipulation of 3D cultures. To quantify the degree of compression generated by a known osmolar imbalance, we first determined the compressive strain exhibited by a polyacrylamide hydrogel of known stiffness (G = 5.15 ± 0.34 kPa, as assessed by shear rheometry) in response to 25 and 50 mg/mL of dextran.Finite element modelling was then used to determine the stresses required to achieve these deformations (1.85 ± 0.67 kPa or ~ 2 kPa, and 2.86 ± 2.50 kPa or ~ 3 kPa for 25 and 50 mg/mL, respectively).These values are consistent with other studies 43 . BeWo spheroids were formed in polyacrylamide micropockets 44 and maintained in culture for two days with and without both forskolin and externally applied compression (Fig. 5B-E).In all cases, spheroidal tissues exhibited some degree of additional compaction over two days, likely due to continued internal remodelling 28 .www.nature.com/scientificreports/However, this compaction was markedly increased in spheroids subjected to dextran compression without forskolin treatment (Fig. 5F).Analysis of nuclear shape in immunostained sections of these tissues indicated that all nuclei were intact and normal (Supplementary Fig. S3), confirming that compression was achieved without inducing apoptosis or necrosis within the spheroidal tissue.Analysis of fusion via E-cadherin junctional analysis was not possible in these models due to the densely packed nuclei, and so we evaluated fusion efficiency via whole-mount light sheet fluorescent analysis of Syndecan-1 expression (Fig. 5G-H), and further confirmed our findings with immunostained tissue sections (Supplemental Fig. S3).Although it remains a possibility that Syndecan-1 expression can occur without fusion, our comparative findings (Figs. 3, 4) suggest that this is unlikely.To obtain an unbiased estimate of the amount of Syndecan-1 expressed within the spheroids, the total integrated fluorescent intensity throughout the spheroid was assessed and normalized against spheroid volume.While the least fusion was observed in the condition without compression or forskolin, we noted that BeWo cells did exhibit some degree of fusion even without forskolin induction, in contrast with our studies of fusion in monolayers which required forskolin.We also noted that ~ 2 kPa of compression was sufficient to significantly increase fusion in standard cultures, whereas ~ 3 kPa stress did not have a similar effect, suggesting that low levels of compression optimally impact fusion.This seems reasonable as too high a compressive stress must adversely affect cell function and activity.Furthermore, at these levels of compression, we achieved fusion efficiencies statistically similar to those achieved via forskolin induction.Surprisingly, while intracellular levels of β-hCG did increase significantly with both compression and forskolin-induction, they did not match the levels of fusion inferred by Syndecan-I expression, consistent with other reports suggesting that intracellular β-hCG levels do not scale with fusion efficiency 33,45 .This data does however confirm that compression affects fusion in 3D, even in the absence of forskolin induction. Discussion While mechanical forces are now well-established to play an important role in directing cell function 46 , the specific relationships between forces arising during development and the corresponding specification of cell behaviour are highly diverse across organ systems.Placental villous tree formation is a multifaceted and dynamic process, and understanding the precise mechanical contributions and impact of the various processes occurring during villous morphogenesis remains challenging.Here, we demonstrate that the process of fusion itself is associated with specific patterns of local stress at the site of fusion.These findings are consistent with other studies of cell fusion in muscles 47 in which the highest contractile stresses were observed at the tips of cells undergoing differentiation.Interestingly, the patterns of stress observed in trophoblast fusion are similar to those that would drive morphogenetic budding in the placental bed.We then demonstrate that applying qualitatively similar patterns of external compressive stress can enhance cell fusion efficiency, in both 2D and 3D culture models.These results together suggest that a positive feedback loop exists between trophoblast fusion and mechanical remodelling that act together to sculpt a villous tree structure with a continuous syncytialized monolayer. We also demonstrate that in 3D cultures, externally applied low levels of compressive stresses can potentially achieve similar levels of fusion as via chemical induction with forskolin.Although BeWo choriocarcinoma cells are a frequently used model for studies of trophoblast fusion, non-physiological chemical induction is often considered essential for fusion in this model.Speculatively, our findings therefore could suggest that downstream effects of chemical stimulation may be physically responsible for fusion.For example, forskolin induction increases production of cyclic adenosine monophosphate (cAMP) which is also known to regulate mechanical contractility in the heart 36,48 , and might establish pro-fusion mechanical conditions.This suggests the intriguing possibility that recently-developed trophoblast stem cell models 14,49 that exhibit higher frequencies of spontaneous fusion may present distinct mechanical behaviours and architectures that make fusion more efficient. Our study does have some significant technical considerations that limit interpretation of results.Specifically, analyzing fusion in 3D is limited to a volumetric estimate of total Syndecan-1 expression, which although frequently used in 3D models may not necessarily be definitive to assess fusion efficiency within a spheroid.Furthermore, quantitatively recreating the stress patterns measured via TFM and predicted via existing computational models cannot be achieved with current commercially available systems.2D monolayer mechanical stimulation systems are strain-controlled, rather than stress-controlled, and rely on transfer of deformation patterns from the elastomeric culture substrate to the adherent cells.Conversely, osmotic compression of 3D systems is stress-controlled, but cannot create specific spatially defined deformation patterns.Hence, although our findings correlate well with each other and are conceptually unified, establishing quantitative stress responses could be an important next step for this work.This would require novel technological developments in cell mechanobiology stimulation platforms, perhaps using 3D engineered hydrogel model systems to apply local stresses to living tissue 50 . More broadly, this specific study does have conceptual limitations that should be considered carefully.First, all experiments were performed with the BeWo choriocarcinoma cell line, which although a well-established model to study fusion processes specifically, may not accurately capture other aspects of placental biology.This is appropriate for a first study because we focus explicitly on establishing external mechanical stresses as fundamental drivers of fusion, but extending this work towards other trophoblast stem cell types would be an important next step.Second, the 2D experiments performed here examine fusion between adjacent cells in culture.Typically, maintenance of the syncytiotrophoblast requires fusion through the basal surface of an established syncytium, which is experimentally quite challenging to recreate.While we can conclude that in vitro fusion is impacted by mechanics, whether this translates directly to in vivo contexts such as syncytial maintenance or to primary syncytialization remains unclear.Finally, establishing causative mechanistic relationships between external mechanical stresses and fusion is particularly challenging.Unlike molecular systems, "external www.nature.com/scientificreports/mechanics" cannot be specifically and precisely inhibited or targeted without affecting a wide variety of cellular processes.Nevertheless, establishing the mechanosensitive pathways that affect fusion would be an important next step in identifying actionable mechanisms to manipulate fusion as needed.In summary, our results demonstrate that directed external mechanical stresses in the form of compression, such as those that would arise when a small patch of cells fuse, might be an important factor in efficient fusion in vitro.We also explicitly show this in spheroids where greater fusion was observed under low compressive stresses even in the absence of forskolin, demonstrating that mechanical stimulation is likely a potent stimulator of fusion and can be utilized as a tool to impact in vitro levels of fusion.Broadly, this work provides a better mechanobiological understanding of trophoblast fusion as a fundamental biological process, which can ultimately be leveraged to improve fusion efficiency in in vitro models and potentially be used as a novel therapeutic target for placental dysfunction in vivo. Methods Unless otherwise stated, all cell culture materials and supplies were purchased from Fisher Scientific (Ottawa, ON) and chemicals from Sigma Aldrich (Oakville, ON). Cell culture-monolayers BeWo human placental choriocarcinoma cells (ATCC; CCL-98) between passages 20-25 (for monolayers) or passages 4-6 (for spheroids) were cultured in 10% foetal bovine serum (FBS) and 1% antibiotic-antimycotic in Dulbecco's Modified Eagle Media (DMEM).BeWo cells were seeded at ~ 120,000 cells/cm 2 to form a continuous monolayer sheet of cells.Seeding density was adjusted for mechanical compression/tension experiments such that the final nuclear density after mechanical deformation of the substrate was comparable.Media was supplemented with 24 µM forskolin (CAS: 66575-29-9) to induce fusion, with media replaced every 24 h.For traction force microscopy (TFM) experiments, an excess volume of media was used to avoid disrupting the automated imaging for a media exchange. Cell culture-spheroids A custom stamp block with a 1 mm cone at the surface was adapted from a previous technology 44 , and designed to fit a 48 well plate using Fusion 360 (Autodesk) and printed in PR57-K black prototyping resin (Colorado Photopolymer Solutions) using the Autodesk Ember DLP 3D printer.To create micropockets, Loctite AA3525 was used as an adhesive to attach polyacrylamide gels to the bottom of the well plate 51 .The adhesive was incubated for 1.5 h in RO water to limit toxicity, after which a polyacrylamide mixture corresponding to 25.6 kPa shear stiffness 52 was added to the bottom of the well and covered immediately with the stamp block.After curing for 8 min, the block was removed and left in PBS solution containing 1% antibiotic-antimycotic for at least 3 days and with the solution changed every 24 h to remove any residual chemicals that might be harmful to cells.On the day of experimentation, microwells were exposed to 365 nm UV light for 1-2 h and then incubated in complete culture media.50,000 cells were pipetted into the well containing the micropocket and the well plate was centrifuged at 400 RCF for 10 min, after which the well plate was incubated at 37 °C for 1-2 days to enable spheroids to form. Traction force microscopy We have previously determined polyacrylamide hydrogel formulations that result in gels with stiffnesses that mimic native healthy placental tissue mechanical properties 33,53 .For TFM experiments, prepolymer solutions of the 3.9 kPa native placental stiffness containing 0.5% volume of 0.5 μm diameter carboxylate-modified fluorescent beads in PBS (FluoSpheres, Invitrogen, Catalog: F8812) were prepared as previously described 34 .To functionalize the gel surface, the gels were activated twice with 0.1 mg/ml of the photoactivable bifunctional crosslinker N-sulfosuccinimidyl-6-[4'-azido-2'-nitrophenylamino] hexanoate (sulfo-SANPAH, ProteoChem); and incubated with an excess amount of 80-100 µl of bovine Collagen I (0.1 mg/ml in phosphate buffered saline or PBS, Life Technologies) on a parafilm sheet at 4 °C overnight.Such TFM substrates prepared on 12 mm glass coverslips were attached to the bottom of a 24 well plate using a drop of cured polydimethylsiloxane (PDMS, 10:1 prepolymer: curing agent, Dow Sylgard 184) and exposed to 365 nm UV light for at least 45 min to facilitate quick attachment and to prevent detachment of gels, after which gels were washed in sterile PBS and immediately used for experiments. TFM substrates were incubated in complete DMEM for 2 h at 37 °C and then with BeWo cells, which were allowed to attach and form a uniform monolayer over 24 h.Cultures were induced to fuse by adding 24 µM of forskolin, and fluorescent/phase contrast images of the uppermost layer of embedded beads were captured every 30 min using an automated fluorescent microscope (20 × objective, EVOS m7000, ThermoFisher Scientific) over 48 h after induction. TFM image analysis To analyse the collected TFM image datasets, template alignment, particle image velocimetry (PIV), and Fourier transform traction cytometry (FTTC) ImageJ plugins were used as per previous protocols 54 .Briefly, both stressed and relaxed fluorescent bead images were combined into a stack and aligned to account for experimental drift.The bead displacements were estimated by PIV (advanced) following an iterative procedure where the interrogation window was made progressively smaller (128 × 128 pixels, 64 × 64 pixels, 48 × 48 pixels; correlation threshold: 0.6) to produce an ultimate displacement field grid of ~ 14.8 μm × 14.8 μm.Traction force fields could then be reconstructed with the FTTC plugin using values: pixel = 0.3086 μm, Poisson's ratio = 0.457, and Young's Modulus = 3900 Pa, along with default values for remaining inputs.This analysis was repeated for each 30 www.nature.com/scientificreports/interval image over 48 h using a custom ImageJ macro.The first time point at t = 0 (time of forskolin induction) was used as the reference image to determine the relative traction stresses that arise during forskolin-induced.Fused syncytialized regions were identified by analysis of Syndecan-1 37,38 immunofluorescent expression in fixed cultures after 48 h of forskolin induction.Only regions that were 75-80% fused were selected for analysis.These regions of interest (ROIs) were enlarged by 80 pixels, to empirically include 1-2 cells outside the region of fusion to include their influence on the stress patterns.A corresponding ROI of the same area but for a non-fused mononucleated region (25-30% fusion efficiency) was also selected for analysis.Since the precise timepoint(s) at which fusion occurs within each ROI is unknown, all time-points in the 48-h fusion window were analyzed for traction stresses, capturing the history of stresses at specific sites.This analysis therefore allows us to retroactively quantify stresses that had occurred during the fusion process.Using a custom Matlab code, traction stress vectors were then decomposed into radial stresses (oriented inwards/outwards to the areal centroid of the ROI), and tangential stresses (oriented perpendicular to the radial vector).For quantification purposes, the average radial and absolute tangential stresses from t = 12 to 48 h was calculated for each ROI to compare between syncytial vs. mononucleated regions. Uniaxial compression stimulation experiments Uniaxial compression experiments were performed with a commercial mechanical stimulation platform (CellScale MechanoCulture FX) using a 24 well silicone membrane platform.The silicone platform was prestretched to 10% uniaxial strain, coated with 0.1 mg/ml bovine Collagen I by incubating at 37 °C for 2 h, seeded with cells in the stretched condition and left undisturbed for 24 h to form a sheet.Similar steps were performed on a separate static 24 well silicone membrane platform to be used as a control.A mechanical stimulation regimen consisting of 5% uniaxial compression over 6 h and no movement for 18 h was repeated twice for a total of 2 days, so that an overall 10% uniaxial compressive strain could be achieved.Media containing forskolin was changed daily.After 2 days of stimulation, cells were fixed, immunostained, and imaged. Equibiaxial compression and tension stimulation experiments Both equibiaxial compression and tension experiments were performed using a custom iris-like stretchable device system that has been previously described in multiple stretching applications 39,55 .In order to establish a static control to compare against the commercial pre-fabricated stretchable dishes used for the stimulation experiments, a silicone membrane with similar properties to that of the stretchable dishes was replicated in a 35 mm petri dish by mixing equal parts of Silicone Elastomer Parts A and B (Elkem LSR-4305 Elastomer, Product Code: A-221-05; Factor II, Inc.) and cured in an oven at 70 °C for 2 h after degassing.The Young's modulus of both the static control petri dishes and stretchable dishes were verified using rheometry (see Supplementary Methods S1).Both the stretchable dishes and the static control petri dishes containing silicone were washed with double distilled water (ddH 2 O) and treated with 30% sulfuric acid for 15 min at room temperature.Both platforms were again rinsed with ddH 2 O and then treated with 1% (3-aminopropyl) triethoxysilane (APTES) for 2 h in a 70 °C oven.Both platforms were again rinsed with ddH 2 O and stored in a 4 °C refrigerator for up to a month prior to use in experiments. On the day of experimentation, both platforms were treated with 1% glutaraldehyde solution for 15 min at room temperature after which both platforms were rinsed with ddH 2 O, sprayed with 70% ethanol, and taken into a biological safety cabinet for sterility with subsequent steps.The stretchable dish was connected to a computer to be pre-stretched using a custom LabVIEW based software to 10% strain from the base minimum area (for compression experiments) or pre-stretched to the required base minimum area (for tension experiments).Both platforms were coated with 0.05 mg/ml Collagen I by incubating at 37 °C for 2 h.After rinsing with sterile PBS, cells were seeded in both platforms and left undisturbed up to 24 h to form a monolayer.Static control cultures were seeded at increased or decreased cell densities such that the number of cells per unit area after mechanical compression/tension was similar.A mechanical stimulation regimen consisting of 5% equibiaxial compression (or tension) over 6 h and no movement for 18 h per day was repeated twice for a total of 2 days, so that an overall 10% equibiaxial compressive (or tensile) strain could be achieved.Media containing forskolin was changed daily.At the end of 2 days, cells were fixed, the bottom of stretchable dishes were cut out with a scalpel, and then immunostained and imaged.Experiments were repeated for 3 independent stretchable dishes and static petri dishes with results normalized to the corresponding control for comparison.Due to limitations in the number of stretching platforms available, it was necessary that repeated experiments be performed at different passage numbers.Since passage number is known to affect BeWo cell fusion efficiency, a fold-based analysis was used to normalize fusion efficiency within each passage number. Osmotic compression of 3D cultures Spheroids were used for fusion-compression experiments under the combinatorial conditions of with/without forskolin and with/without dextran, where dextran (Mw = 500 kDa, Dextran Products; Polydex Pharmaceutical) was added separately as 25 mg or 50 mg of dextran per ml of complete DMEM media.Media was changed every 24 h and timelapse images at time points t = 0, 3, 6, 24, 30 and 48 h were acquired to assess compaction though compressive strains.After 48 h, samples were washed with PBS and then fixed overnight for immunostaining or histology to confirm findings (see Supplementary Methods S2). COMSOL modelling: determination of spheroid compression due to dextran The isotropic compression applied on the spheroids due to the concentration of dextran used was calculated using a model similar to previous work 28 .In brief, disc-shaped polyacrylamide gels were fabricated by casting polyacrylamide prepolymer gel solution (18.75% v/v 40% Acrylamide, 2.70% v/v 2% Bisacrylamide, 68.40% v/v PBS, 0.15% v/v TEMED and 10.0% v/v APS) into 12 mm molds under a constant nitrogen gas flow for 15 min.After gelation, disc gels were washed with PBS 3 times and stored at 4 °C to allow the gels to swell.After 7 days, the rheological properties of disc gels were measured through rheometry (see Supplementary Methods S1).We have previously demonstrated that this dextran chain length formulation does not penetrate polyacrylamide gels 28 .The change in size of the gels due to dextran induced osmotic compression was measured using dextran solutions with concentrations of 25 mg/ml and 50 mg/ml in PBS.To do so, images of the disc bulk gel were acquired using the EVOS m7000 microscope at 4X magnification before and 30 min after addition of Dextran solution at 37 °C, and the corresponding change in diameter was calculated in ImageJ (NIH). To quantify the compressive forces generated by the dextran, an inverse finite element simulation was made in COMSOL Multiphysics 6.0 (Comsol Inc., Burlington, MA, USA).Disc shaped geometries were generated in a 2D axisymmetric model, and a pressure boundary condition was set on the top and side while a roller condition was placed on the bottom surface.A free triangular mesh was generated with predefined extremely fine element size giving an average element quality above 0.9.Discs were prescribed linearly elastic properties with a Young's modulus measured by rheometry (see Supplementary Methods S1) and a Poisson ratio estimated at 0.457 from literature 56 .Using the experimentally measured pre-and post-compression gel sizes, the applied pressure was determined. Microscopy All samples of equibiaxial compression or tension experiments were imaged on an inverted fluorescent Olympus microscope (Olympus, IX73) outfitted with an sCMOS Flash 4.0 Camera and Metamorph software (version 7.8.13.0) for brightfield and corresponding fluorescence filters (blue for Hoechst 33,258, red for E-Cadherin and green for Syndecan-1).For TFM experiments, the uniaxial compression experiment, spheroid compression time lapse experiment and spheroid histology sections, samples were imaged on an inverted EVOS m7000 microscope for brightfield and corresponding fluorescence filters (blue for Hoechst 33258, green for Syndecan-1 and red for fluorescent beads or E-cadherin).Images were acquired at random locations at 20X magnification for stimulation experiments and spheroid sections, at fixed positions at 20X magnification stored in memory over 48 h at 30-min intervals for TFM experiments, and per well at 4X magnification over 48 h at time points t = 0, 3, 6, 24, 30 and 48 h for the spheroid compression time lapse experiment. Whole spheroid samples were imaged on a customized SCAPE 2.0 light sheet microscopy system 57 in a topdown format, outfitted with a water dipping objective lens (20X 1.0 NA) and an Andor Zyla 4.2 + CMOS camera.Samples were individually encased in 1% agarose (UltraPure™ Agarose, Invitrogen) in a 35 mm petri dish prior to imaging.Slices of the entire sample were obtained at 10 μm spacing under 9.3X magnification, using a 488 nm excitation laser with 100 ms exposure and 10 mW power at source settings (green for Syndecan-1) and a 561 nm excitation laser with 100 ms exposure and 8 mW power at source settings (red for β-hCG). Image analysis All images were analysed in FIJI ImageJ (NIH).For fusion analysis, cells expressing Syndecan-1 or a colony containing 2 or more cells sharing the same membrane boundary expressing E-cadherin were considered to be fused.The total number of nuclei (T) was counted either manually or using the ParticleSizer plugin in ImageJ 58 and used to calculate nuclear density over a fixed area of 0.34 mm 2 whereas the number of nuclei in a fused syncytia (F), the number of syncytia (S) and the number of nuclei expressing Syndecan-1 (FS) were counted manually.Based on the usage of E-cadherin or Syndecan-1 expression for analysis, fusion efficiency was calculated as described below, based on similar techniques previously reported 34,37,59 : For analysis of fusion in spheroids, the laser illumination settings, camera capture settings, and brightness and contrast settings of all slices for all conditions were uniformly and consistently maintained for Syndecan-1 and β-hCG image collection respectively.To avoid sampling issues associated with selecting a single optical section of an intact spheroid, and to avoid out-of-plane light collection from affecting the analysis, we calculate the integrated fluorescent signal intensity for the entire spheroid, and normalized this against the volume of each spheroid (measured by analyzing the area of each spheroid in each slice of the imaging stack).This value was used as a proxy for fusion levels with arbitrary units, and was calculated as follows: Compressive strains in timelapse images of spheroids subjected to forskolin and/or dextran compression was calculated by finding the effective change in diameter of the spheroid at each time point (assuming a spherical structure) compared to that at t = 0 h.Immunostained sectioned spheroid images were created by overlaying all channels (nuclei, Syndecan-1 and E-cadherin) with brightness and contrast values fixed independently for each channel. Statistical analysis Statistical analysis was performed using JASP, version 0.16.3 60 .Independent Student's t-test was used for the uniaxial fusion analysis and comparison of TFM stresses, one-way ANOVA with post-hoc Tukey's test for pairwise comparisons was performed for equibiaxial compression and tension fusion and nuclear density analyses whereas two-way ANOVA with post-hoc Tukey's test for pairwise comparisons was performed for spheroid fusion analyses.In all cases, p ≤ 0.05 was considered statistically significant. Figure 1 . Figure 1.Techniques used for traction force microscopy (TFM) analysis.(A) Schematic illustrating the TFM experiment performed with sheets of fused and non-fused BeWo cells on a polyacrylamide gel.Arrows indicate the change in position of fluorescent beads that is used to calculate stresses.(B) Stress vectors S were decomposed into radial S R (blue) and tangential S T (orange) components based on the areal centroid (x C , y C ) of a fused region.(C,D) Schematic illustrating an idealized stress field of all the correspondingly resolved radial and tangential stress vectors, respectively within and around a region of interest (ROI). Figure 2 . Figure 2. Traction force microscopy (TFM) performed on fusion induced BeWo monolayers.(A) Schematic depicting a sheet of cells undergoing fusion over a period of 48 h; fused regions represented in green.(B) Representative fluorescent image of a sheet of BeWo cells, showing a fused syncytial region (yellow outline) and a non-fused mononucleated region (red outline).Scale bar is 100 µm; green: Syndecan-1, blue: nuclei.(C,D) Radial traction stresses and (F,G) tangential traction stresses over 48 h, for both syncytial and mononucleated regions, respectively.Each line represents one tracked region.(E) Radial traction stresses and (H) tangential stresses, averaged from hours 12 to 48 for both syncytial and mononucleated regions.(Data presented as box plot distributions; the central line is the median, the box represents 25th to 75th percentile (lower and upper quartiles) and whiskers represent min/max values; n = 6-7 regions; p < 0.05 by independent Student's t-test). Figure 3 . Figure 3.Effect of uniaxial compression on fusion in BeWo cells, induced with forskolin (24 µM) for 2 days.(A) Schematic illustrating the 10% uniaxial compression experiment performed along with a graph (B) showing the actual regimen over 48 h.(C,D) Fluorescent images of fused BeWo cells stained for E-cadherin junctional proteins, and (F,G) Syndecan-1 on static vs. 10% uniaxial compression over 2 days.Scale bar is 100 µm; green: Syndecan-1, red: E-cadherin, blue: nuclei.(E,H) Quantification of fusion efficiency calculated using E-cadherin and Syndecan-1, respectively.(Data presented as mean ± standard deviation; n = 5 independent samples, ns not significant, p = 0.916 and p = 0.788, respectively by independent Student's t-test). Figure 4 . Figure 4. Effect of equibiaxial compression and tension on fusion in BeWo cells, induced with forskolin (24 µM) for 2 days.(A) Schematic illustrating the 10% equibiaxial compression/tension experiment performed along with a graph (B) showing the actual regimen over 48 h.(C-E) Fluorescent images of fused BeWo cells stained for E-cadherin junctional marker expression and (G-I) Syndecan-1 expression on static control, 10% uniaxial compression, and 10% equibiaxial tension substrates over 2 days.Scale bar is 100 µm; green: Syndecan-1, red: E-cadherin, blue: nuclei.(F,J) Quantification of averaged normalized population fused calculated using E-cadherin and Syndecan-1 expression.(Data presented as mean ± standard deviation; n = 3 independent experiments with 10 image viewfields each, sampled at random locations, p < 0.05 by one-way ANOVA).	v3-fos-license
2019-04-30T13:08:41.210Z	2018-11-29T00:00:00.000	139618609	{ "extfieldsofstudy": [ "Materials Science" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2075-4701/8/12/1000/pdf", "pdf_hash": "0465cf63d3b651bcfb743d32f5ff897d33d33b76", "pdf_src": "MergedPDFExtraction", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45326", "s2fieldsofstudy": [ "Materials Science" ], "sha1": "9078d888523618041757090680b354532c4a3035", "year": 2018 }	pes2o/s2orc	Characterization of the Microstructure, Local Macro-Texture and Residual Stress Field of Commercially Pure Titanium Grade 2 Prepared by CONFORM ECAP The paper investigated the residual strain and stress distribution, microstructure, and macro-texture along the transverse direction of commercially pure titanium grade 2 samples prepared by the CONFORM ECAP technique. This method belongs to the severe plastic deformation methods; hence, it could be assumed that residual stress fields would be present in the work-pieces. Residual stresses cannot be directly measured; thus, neutron diffraction measurements, Electron back-scatter diffraction (EBSD) investigations, and local X-ray macro-texture measurements were performed in different regions of the sample to determine the data for the residual stress calculation. The calculation was based on the modified Kröner model. Neutron diffraction strain scans and residual stress calculations revealed that symmetrical residual strain and stress gradients with compression character were present in the axial and hoop direction after one and two passes. Asymmetric distribution of the residual strains and stresses remained after the third pass of the CONFORM ECAP. EBSD investigations showed that after the first pass, significant grain refinement occurred; however, further passes did not cause any dramatic grain refinement. X-ray texture measurements revealed that local macro-texture was dependent on the number of passes of the CONFORM ECAP and on the investigated area in the samples. Introduction Titanium and its alloys have a wide range of applications due to possessing high strength, low density, excellent corrosion resistivity, and nontoxicity [1]. For many years, the materials have been playing an important role in medicine, where a large number of implants are titanium-based. In biomedical application, the desired mechanical properties are usually achieved through alloying. However, some of the alloying elements are believed to be toxic to the human body, which limits the elements usage as biocompatible materials. Thus, the scientific community is constantly seeking Materials and Methods CP Ti grade 2 was studied in this work. The chemical composition of the samples can be found in Table 1. Work-pieces in the form of 2 m long cylindrical rods with a diameter of 10 mm were subjected to the C-ECAP process as described in reference [9] using COMTES FHT a.s. company's specially designed machine, as in reference [17]. Specimens passed 1×, 2×, and 3× through the C-ECAP machine on a route denoted as "A" during the multiple C-ECAP processes described in reference [18]. Then, 80 mm long samples were cut from the rods after the mechanical treatments. Neutron Diffraction Neutron diffraction (ND) measurements were carried out on the samples after the C-ECAP treatment at the Nuclear Research Institute,Řež, Czech Republic, on the HK4 beamline. The beamline HK4 is equipped with a strain scanner SPN-100 diffractometer supported by a 2D position sensitive detector (PSD) and a Si (111) monochromator. The complete description of the instrument can be found in References [19,20]. The diffracted neutron wavelength was 2.13 Å and square shaped slits with 3 mm side length were used. The center of the 2D PSD was set to a fixed value 2θ D = 55 • during the experiments. This 2θ D angle allowed the observation of reflections 1011 and (0002), and the nominal gauge volume (NGV) was ∼32.96 mm 3 . Samples were placed on the xyz stage, which allowed both the motion in three perpendicular directions, i.e., x, y, and z, as well as the rotation by an arbitrary angle. Samples were scanned using ND, where the sampled gauge volume was moved in the coordinate system of the sample. The scanning was performed along a straight line by 1 mm shifts, which intersected the center of the circular cross-section. This line lay in the transverse direction in the conventional ECAP notation. Three scans were performed along the same line on each sample: (1) diffraction vector → q R was parallel with the radial direction of the sample; (2) diffraction vector → q H was parallel with the hoop direction; (3) diffraction vector → q A was parallel with the axial direction. The scheme of the scanning is drawn in Figure 1. Neutron Diffraction Neutron diffraction (ND) measurements were carried out on the samples after the C-ECAP treatment at the Nuclear Research Institute, Řež, Czech Republic, on the HK4 beamline. The beamline HK4 is equipped with a strain scanner SPN-100 diffractometer supported by a 2D position sensitive detector (PSD) and a Si (111) monochromator. The complete description of the instrument can be found in References [19,20]. The diffracted neutron wavelength was 2.13 Å and square shaped slits with 3 mm side length were used. The center of the 2D PSD was set to a fixed value 2θD = 55° during the experiments. This 2θD angle allowed the observation of reflections (101 ̅ 1) and (0002), and the nominal gauge volume (NGV) was ~32.96 mm 3 . Samples were placed on the xyz stage, which allowed both the motion in three perpendicular directions, i.e., x, y, and z, as well as the rotation by an arbitrary angle. Samples were scanned using ND, where the sampled gauge volume was moved in the coordinate system of the sample. The scanning was performed along a straight line by 1 mm shifts, which intersected the center of the circular cross-section. This line lay in the transverse direction in the conventional ECAP notation. Three scans were performed along the same line on each sample: (1) diffraction vector ⃗⃗⃗⃗ was parallel with the radial direction of the sample; (2) diffraction vector ⃗⃗⃗⃗ was parallel with the hoop direction; (3) diffraction vector ⃗⃗⃗⃗ was parallel with the axial direction. The scheme of the scanning is drawn in Figure 1. The differential form of Bragg's Equation (1) was used for calculation of the lattice strain ℎ of the particular crystallographic plains as follows. The differential form of Bragg's Equation (1) was used for calculation of the lattice strain ε hkl of the particular crystallographic plains as follows. d 0 hkl and θ B0 hkl are the interplanar distance of the hkl planes of the stress-free reference and the corresponding Bragg angle. Furthermore, d hkl and θ B hkl are the interplanar distance of the hkl planes under load and the corresponding Bragg angle. The ND scanning started at a distance of -3 mm and ended at distance of +3 mm from the center of the samples. Omission of the measurement points at the edges of the specimen (positions ±4 mm) was given by the fact that at these points the NGV was not completely filled with material, which resulted in anomalous θ B hkl shifts, and consequently in spurious strains [21]. EBSD Specimens for the microstructural analysis with the EBSD technique were cut from the samples used at the ND measurements. Samples were prepared by standard metallographic grinding on SiC papers (from 320 to 1200 grit), followed by three-step vibratory polishing for 24 h (Alumina 0.3 µm, Alumina 0.05 µm, and Colloidal Silica 0.04 µm-each for eight hours). Finally, the specimen surfaces were ion-beam polished on a Leica EM RES102 system (Leica Mikrosysteme, Wetzlar, Germany). The investigated surfaces were perpendicular to the axial direction. Microstructure analysis based on the EBSD technique was performed using a scanning electron microscope FEI Quanta 200 FX (Thermo Fisher Scientific, Brno, Czech Republic). Working distance was 13 mm with accelerating voltage 15 kV. The scan was conducted with a 0.1 µm step size. The scanned areas (left and right periphery, and the central part) were chosen with the aim to characterize the microstructure in the ND irradiated sample volume. X-ray Texture The X-ray PANalytical XPert MRD (PANanalytical, Almelo, The Netherlands) diffractometer using CuKα radiation with polycapillary optics in the primary beam was employed for acquiring the (0002), 1010 , and 1011 pole figures. The orientation distribution functions (ODF) and the full pole figures were determined using MTEX 4.4 Free and open source software toolbox for MatLab (MathWorks, Natick, MA, USA) [22]. Calculation of the Residual Stresses The calculation of the residual stresses in the three perpendicular directions (axial, hoop, and radial) was based on the results acquired by the EBSD, X-ray texture, and ND measurements. The calculation was done using the software IsoDEC version from 9/21/2016 (NIST Center for Neutron Research, Gaithersburg, MD, USA) as in reference [23], where the computation was based on the model proposed by Kröner [24] and modified by Bollenrath [25] and Behnken [26]. Owing to the assumption of grain-interactions, the model allowed for the use of specific grain shapes to calculate the right diffraction elastic constants. Moreover, the crystallographic preferred orientation (texture) of the sample given by the orientation distribution function (ODF) was also included in the model. Neutron Diffraction and Residual Strain The residual strains were calculated using Equation (1), where θ B hkl angles were determined from the neutron diffraction measurements. The average of the measured θ B hkl angles between points −3 and 3 in the axial direction of the initial sample were taken as the stress-free reference (θ B0 hkl ), because the angle variation through the sample was small in contrast to the hoop and radial directions. The residual strains in the radial, hoop, and axial directions are shown for each sample in Figure 2a-c. The sample after the first C-ECAP pass was denoted as C1, then after the second pass as C2, and after the third pass as C3. , and radial (c) direction of the samples after the first, second, and third C-ECAP pass. Axial Direction It was obvious that in the axial direction, after the first pass, the peripheral regions were almost strain-free whilst in the center of the sample, large strain with compression character was present. This behavior resulted in a large strain gradient, where the strain changed from ~−17 με to ~−600 με (on the left side of After the second pass (Figure 2a-C2), the compression strain increased in the periphery, whereas in the center part it had slightly decreased. Consequently, the strain gradient was less pronounced. In general, the shape of the strain distribution through the samples in the axial direction, after the first and second pass, was a "V" form. After the third pass (Figure 2a-C3), the compression strain reached a lower value in the center of the sample and at the same time the strain had increased in the periphery. The shape of the overall strain distribution lost the "V" form, and it took on a rather wave-like shape with the lowest compression strain in the center of the sample. The reason was that during the third C-ECAP pass, the alignment of the work-piece, as a consequence of the previous severe deformation, was not precise. Consequently, the stress distribution during the processing was uneven, which caused the asymmetric residual strains with wave-like form in the axial direction. Axial Direction It was obvious that in the axial direction, after the first pass, the peripheral regions were almost strain-free whilst in the center of the sample, large strain with compression character was present. This behavior resulted in a large strain gradient, where the strain changed from ∼−17 µε to ∼−600 µε After the second pass (Figure 2a-C2), the compression strain increased in the periphery, whereas in the center part it had slightly decreased. Consequently, the strain gradient was less pronounced. In general, the shape of the strain distribution through the samples in the axial direction, after the first and second pass, was a "V" form. After the third pass (Figure 2a-C3), the compression strain reached a lower value in the center of the sample and at the same time the strain had increased in the periphery. The shape of the overall strain distribution lost the "V" form, and it took on a rather wave-like shape with the lowest compression strain in the center of the sample. The reason was that during the third C-ECAP pass, the alignment of the work-piece, as a consequence of the previous severe deformation, was not precise. Consequently, the stress distribution during the processing was uneven, which caused the asymmetric residual strains with wave-like form in the axial direction. In summary, the first pass introduced a large strain gradient in the axial direction toward the center of the sample. The increasing number of passes (2x-3x passes) increased the compression strain in the peripheral regions, whilst decreasing the compression strain in the center part. Hoop Direction In the hoop direction, all of the values showed a compression strain character (Figure 2b). It was obvious that after the first and second pass, the strain distribution followed the same shape through the sample (Figure 2b-C1, C2). The difference was in the higher values of the compression strain after the second pass. After the third pass (Figure 2b-C3), the C-ECAP process resulted in a decrement of the compression strain (increasing the interplanar distance) on the left side of the sample, whilst on the right side of the sample, the interplanar distance decreased (compression character was more pronounced) with respect to the previous pass. Radial Direction The residual strain distributions through the samples in the radial direction after the first and second pass showed a very homogenous nature with the compression character ( Figure 2c-C1, C2). After the first pass, the values were distributed around −200 µε. Only at the right peripheral region of the sample was the strain closer to the stress-free reference. After the second pass, the residual strain values were distributed around −400 µε; however, at the left peripheral region, the compression strain had increased. The third pass resulted in an inhomogeneous residual strain distribution compared to the previous passes (Figure 2c-C3). In the right peripheral region, the compression strain slightly decreased to a level similar to that after the first pass, whilst in the left peripheral region, the strain distribution showed alternating behavior. In summary, the C-ECAP process resulted in a relatively symmetric distribution of the residual strains in all directions after the first and second pass. The largest strain gradient was in the axial direction (Figure 2a), whilst in the radial direction (Figure 2b), the strain was homogenously distributed. This symmetry was interrupted by the application of the third C-ECAP treatment on the material. Microstructure: EBSD Results The observed grain orientation map, pole figures, grain size distribution, and fractions of the low-and high-angle grain boundaries (LAGB and HAGB) of the initial state, after 1, 2, and 3 C-ECAP passes are shown in Figures 4-7. Samples after the C-ECAP treatment were studied in three different regions across the diameter, schematically shown in Figure 3. In this paper, the notation used at strain scanning by neutron diffraction to describe the cylindrical sample reference system that was used instead of the conventional axis notation used in material engineering (normal-ND, transverse-TD, and rolling direction-RD). In the pole figures, the vertical axis represented the hoop direction (Hoop), the horizontal axis represented the radial direction (Radial), and the normal of the pole figure plane represented the axial (Axial) direction. In the initial state, the average grain sizes were in the µm domain (5-7 µm) and the grain shape was approximately equiaxed in all the regions (Figure 4a,b). In peripheral region (a), the (0001) basal planes were parallel to the axial direction, forming four distinct maxima, which were between the hoop and radial directions. The 1010 prismatic planes formed several maxima in the peripheral region: one perpendicular to the axial direction and then two which were ∼60 • inclined from the axial direction of the sample. This so-called partial fiber texture is typical for extruded hexagonal materials [27,28]. In the central region (b), the texture was completely different. The (0001) basal planes formed two distinct maxima which were ∼40 • inclined from the axial direction. Furthermore, in the peripheral region, the fraction of the HAGB was less than in the center of the sample, whilst the fraction of the LAGB did not depend on the measured region. In the initial state, the average grain sizes were in the μm domain (5-7 μm) and the grain shape was approximately equiaxed in all the regions (Figure 4a,b). In peripheral region (a), the (0001) basal planes were parallel to the axial direction, forming four distinct maxima, which were between the hoop and radial directions. The (101 ̅ 0) prismatic planes formed several maxima in the peripheral region: one perpendicular to the axial direction and then two which were ~60° inclined from the axial direction of the sample. This so-called partial fiber texture is typical for extruded hexagonal materials [27,28]. In the central region (b), the texture was completely different. The (0001) basal planes formed two distinct maxima which were ~40° inclined from the axial direction. Furthermore, in the peripheral region, the fraction of the HAGB was less than in the center of the sample, whilst the fraction of the LAGB did not depend on the measured region. After the first C-ECAP pass, significant grain refinement occurred in all the three regions, where the maximum grain size distribution was at 0.8 µm. However, other local maxima could be found at 2.7-3.7 µm (Figure 5a was formed by a mixture of elongated and equiaxed grains. The dark areas in the EBSD maps were places with a confidence index (CI) < 0.1. Most probably, the higher dislocation density caused poor diffraction patterns in these areas. The local micro-texture in the center (b) and in the right peripheral region (c) was the same, whilst in the left peripheral region (a) it was slightly different. In positions (b) and (c), the (0001) basal planes formed three distinct maxima: along the hoop axis and inclined by 45 • -55 • from the axial direction, and then two on the radial axis perpendicular to the axial direction. The first mentioned texture component was also observed in other magnesium alloys processed by classical ECAP [29]. In the left peripheral region (c), there were two distinct maxima near the edge of the (0001) pole figure, between the hoop and the radial direction. However, these maxima were blurred, compared to the sharp maxima in positions (b) and (c), and part of the maxima area included even a 45 • -60 • inclination from the axial direction. The maxima in the 1010 pole figures in position (a) and (b) had the same location, whilst in position (c), there was no sharp and unique maximum. The 1011 pole figures showed that these planes had an almost homogenous distribution; however, in position (a), there was a local minimum in the surroundings of the axial direction. In addition, after the first C-ECAP pass, the sample exhibited a higher fraction of HAGBs in all the three regions than that in the initial state, which is typical for materials processed by ECAP [30,31]. After the first C-ECAP pass, significant grain refinement occurred in all the three regions, where the maximum grain size distribution was at 0.8 μm. However, other local maxima could be found at 2.7-3.7 μm (Figure 5a), 2.5-3.5 μm (Figure 5b), and 2 μm (Figure 5c). The microstructure was formed by a mixture of elongated and equiaxed grains. The dark areas in the EBSD maps were places with a confidence index (CI) <0.1. Most probably, the higher dislocation density caused poor diffraction patterns in these areas. The local micro-texture in the center (b) and in the right peripheral region (c) was the same, whilst in the left peripheral region (a) it was slightly different. In positions (b) and (c), the (0001) basal planes formed three distinct maxima: along the hoop axis and inclined by 45°-55° from the axial direction, and then two on the radial axis perpendicular to the axial direction. The first mentioned texture component was also observed in other magnesium alloys processed by classical ECAP [29]. In the left peripheral region (c), there were two distinct maxima near the edge of the (0001) pole figure, between the hoop and the radial direction. However, these maxima were blurred, compared to the sharp maxima in positions (b) and (c), and part of the maxima area included even a 45°-60° inclination from the axial direction. The maxima in the (101 ̅ 0) pole figures in position (a) and (b) had the same location, whilst in position (c), there was no sharp and unique maximum. The (101 ̅ 1) pole figures showed that these planes had an almost homogenous distribution; however, in position (a), there was a local minimum in the surroundings of the axial direction. In addition, after the first C-ECAP pass, the sample exhibited a higher fraction of HAGBs in all the three regions than that in the initial state, which is typical for materials processed by ECAP [30,31]. According to grain size distributions of the sample after the second C-ECAP pass (Figure 6a-c), the maxima of the distributions were at the same position as they had been in the case of the sample after the first pass. On the other hand, there were no other notable local maxima, indicating a more homogeneous microstructure, although, the grain orientation maps showed some larger grains. From the (0001) pole figures, the maxima of the main texture component were blurred and ∼40 • -75 • inclined from the axial direction in all the three regions. If we compared the 1010 pole figures in positions (a), (b), and (c), it could be noticed that the distribution of these planes was very similar. Owing to these features, the sample after the second C-ECAP pass exhibited the same microstructure in positions (a), (b), and (c). The only difference between the investigated regions was in the fraction of the HAGBs, which was, in respect to level and distribution, the same as in the sample after one C-ECAP pass. According to grain size distributions of the sample after the second C-ECAP pass (Figure 6a-c), the maxima of the distributions were at the same position as they had been in the case of the sample after the first pass. On the other hand, there were no other notable local maxima, indicating a more homogeneous microstructure, although, the grain orientation maps showed some larger grains. From the (0001) pole figures, the maxima of the main texture component were blurred and ~40°-75° inclined from the axial direction in all the three regions. If we compared the (101 ̅ 0) pole figures in positions (a), (b), and (c), it could be noticed that the distribution of these planes was very similar. Owing to these features, the sample after the second C-ECAP pass exhibited the same microstructure in positions (a), (b), and (c). The only difference between the investigated regions was in the fraction of the HAGBs, which was, in respect to level and distribution, the same as in the sample after one C-ECAP pass. In Figure 7a-c the microstructural characteristics of the left peripheral, center, and right peripheral regions of the sample after the third C-ECAP pass are shown, respectively. From the grain size distributions, it was obvious that further grain refinement had occurred. Maxima were around 0.5 µm in all the three regions and the grains were equiaxed. In the grain orientation maps, large dark areas with CI < 0.1 appeared, indicating high dislocation densities. The maxima of the (0001) and 1010 pole figures were nearly at the same position in all the three investigated regions. Nevertheless, they were slightly different compared to the sample after two C-ECAP passes. The (0001) planes in position (a), (b), and (c) formed blurred maxima, inclined 55 • -85 • from the axial direction. The maxima of the 1010 pole figure were in the axial directions. In addition, the fraction of the HAGBs had increased in all the regions (a), (b), and (c), compared to the previous cases. In summary, the sample after the first C-ECAP pass exhibited a heterogeneous local micro-texture. The microstructure contained a mixture of elongated and equiaxed grains. However, there was a slight difference in the grain size distribution between the central region and the peripheral regions. Nevertheless, the fraction of the HAGBs had globally increased compared to the initial state. After the second C-ECAP pass, the sample exhibited a relatively homogenous microstructure from the point of view of local micro-texture and grain shape. There was no difference between the investigated regions. The fraction of the HAGBs remained at the same level as it was after one C-ECAP pass. The third C-ECAP pass resulted in the most homogeneous microstructure and it further increased the fraction of the HAGBs. In addition, the character of the distribution of the HAGBs remained the same, i.e., it increased from left to right through the sample. Black areas on the grain orientation maps revealed the inhomogeneous dislocation density at particular regions of the samples after the first and third pass. CP Ti primarily deforms by a slip on basal (0002), prismatic 1010 , or pyramidal 1011 planes and by twinning on particular planes [32][33][34]. Moreover, it was shown that twinning in CP Ti played an important role as one of the main deformation modes during conventional ECAP treatment at elevated temperatures [35][36][37][38]. However, twins were not observed in our samples in the cross-section, which did not necessarily mean that they were not present. Many conditions can influence the twin activity and density in hcp materials. In reference [38], the authors showed that the maximum twin density was achieved at 623K ECAP deformation temperature of CP Ti, whilst at 473 K, the twin density was almost one-third of the maximum density. At an 873 K pressing temperature, the twin density was 10 times lower and the twin width was below 0.1 µm. In reference [39], Palán et al. simulated the conditions which prevailed in the C-ECAP machine. The authors results indicated that the temperature, the strain rate, and the velocity distribution were very inhomogeneous in the die. Thus, for example, the temperature in the die could reach 800-850 K where the twin density and width reached the minimum according to reference [38]. Our EBSD observations were conducted at a 0.1 µm step size, which was approximately the width of the expected twins at around 800-850 K deformation temperature. Hence, the EBSD measurement evaluation software would not recognize the twins, rather it would report an error for the measured pixel instead. Furthermore, the initial grain size could also influence the twinning activity as it was reported in reference [40] for the Mg alloys and for Ti [41] (including the strain rate). Our starting material had an average grain size of 5-7 µm, whilst in the above-mentioned experiments where twinning was reported, the average grain size of the starting materials was several times larger than ours. Thus, the relatively small initial grain size could probably have suppressed the formation of larger twins, which could be observed by our experimental conditions. X-ray Texture Measurements The results of the X-ray texture measurements of the samples after one, two, and three passes in the three different regions of the sample (left peripheral-(a), center-(b), and right peripheral-(c)) are shown in Figures 8-10. The axis notation was identical to the axis notation at the EBSD measurements. From the simulated (0002) pole figures at the peripheral regions (Figure 8a,c) after the first C-ECAP pass, the (0002) basal planes parallel to the axial direction formed maxima on the edge of the (0002) pole figures. These components would be indicated as "parallel" components in the following text. Moreover, (0002) planes formed other maxima inclined at ~50°-60° from the axial direction and distributed between the hoop and radial axis. However, the arrangement was in the opposite sense and it was rotated to hoop axis by ~120°-130° around the axial direction. This texture component would be noted as the "50°-60°" component. The texture component "50°-60°" was also observed in our EBSD investigation ( Figure 5). The same texture component was seen in the (0002) pole figure in the central region (Figure 8b), but it lay on the hoop axis. In this region, this was the only notable maximum, and parallel components were not present. The only notable maxima in the (101 ̅ 0) pole figures were inclined at ~25°-35° from the axial direction; however, the overall distribution of the maxima formed a ring around the axial direction. The position arrangement of the maxima around the axial direction was in the opposite sense, as it was present at the (0002) pole figures. The maxima formed by (101 ̅ 0) were in an axial direction independently from the measured region; however, the random distribution of the planes inclined by 35°-55° from the axial direction forming a ring in the peripheral regions. The most intense maxima of the (0002) and (101 ̅ 1) pole figures were in the central region ( Figure 8b) and at the same time, the least intense maximum of the (101 ̅ 0) pole figure was also found here. From the simulated (0002) pole figures at the peripheral regions (Figure 8a,c) after the first C-ECAP pass, the (0002) basal planes parallel to the axial direction formed maxima on the edge of the (0002) pole figures. These components would be indicated as "parallel" components in the following text. Moreover, (0002) planes formed other maxima inclined at ∼50 • -60 • from the axial direction and distributed between the hoop and radial axis. However, the arrangement was in the opposite sense and it was rotated to hoop axis by ∼120 • -130 • around the axial direction. This texture component would be noted as the "50 • -60 • " component. The texture component "50 • -60 • " was also observed in our EBSD investigation ( Figure 5). The same texture component was seen in the (0002) pole figure in the central region (Figure 8b), but it lay on the hoop axis. In this region, this was the only notable maximum, and parallel components were not present. The only notable maxima in the 1010 pole figures were inclined at ∼25 • -35 • from the axial direction; however, the overall distribution of the maxima formed a ring around the axial direction. The position arrangement of the maxima around the axial direction was in the opposite sense, as it was present at the (0002) pole figures. The maxima formed by 1010 were in an axial direction independently from the measured region; however, the random distribution of the planes inclined by 35 • -55 • from the axial direction forming a ring in the peripheral regions. The most intense maxima of the (0002) and 1011 pole figures were in the central region ( Figure 8b) and at the same time, the least intense maximum of the 1010 pole figure was also found here. The result of texture measurements on the sample after the second C-ECAP pass was visible in Figure 9a-c. In the peripheral regions ((a) and (c)), the simulated (0002) pole figures showed that "50°-60°" components were almost exactly at the same position as they were in the sample after the first pass, whilst "parallel" components were slightly rotated. In the central region (Figure 9b), the "50°-60°" component was rotated counterclockwise to the hoop axis by ~20° around the axial direction, with respect to the position of the maximum after the first pass, and its intensity increased as well. The inclination of the maxima from the axial direction in the (101 ̅ 0) pole figures was the same as they were in the first pass, but they were rotated clockwise by ~25°-35° around the axial direction compared to previous positions. The maxima in the (101 ̅ 1) pole figure were still exactly in the axial direction; however, higher density regions inclined by ~35°-50° from the axial direction appeared in the peripheral regions ((b) and (c)). The result of texture measurements on the sample after the second C-ECAP pass was visible in Figure 9a-c. In the peripheral regions ((a) and (c)), the simulated (0002) pole figures showed that "50 • -60 • " components were almost exactly at the same position as they were in the sample after the first pass, whilst "parallel" components were slightly rotated. In the central region (Figure 9b), the "50 • -60 • " component was rotated counterclockwise to the hoop axis by ∼20 • around the axial direction, with respect to the position of the maximum after the first pass, and its intensity increased as well. The inclination of the maxima from the axial direction in the 1010 pole figures was the same as they were in the first pass, but they were rotated clockwise by ∼25 • -35 • around the axial direction compared to previous positions. The maxima in the 1011 pole figure were still exactly in the axial direction; however, higher density regions inclined by ∼35 • -50 • from the axial direction appeared in the peripheral regions ((b) and (c)). (Figure 10c), except the clear maxima in the axial direction, the distribution of the (101 ̅ 1) planes inclined by ~30°-40° from the axial direction formed blurred regions with a higher intensity in the (101 ̅ 1) pole figure (c). The same held for the (101 ̅ 1) pole figure in the central region (b); however, three regions could be spotted with more clear edges. The above-discussed results can help to illustrate the preferred orientation in the different regions of the samples. It could be observed, that the C-ECAP mechanical treatment resulted in different textures in the different regions. However, the "50°-60°" component was present in all the The above-discussed results can help to illustrate the preferred orientation in the different regions of the samples. It could be observed, that the C-ECAP mechanical treatment resulted in different textures in the different regions. However, the "50 • -60 • " component was present in all the regions, the rotation with respect to the radial axis was dependent on the region and the number of passes. The theoretical shear plane normal was inclined by 45 • from the extrusion-axial direction due to the C-ECAP design. If we assumed, for example, after the first pass in the central region, that the main deformation mode was a dislocation slip in the basal (0002) planes in direction 1120 , then theoretically the "50-60" component in the (0002) pole figure should be inclined exactly by 45 • from the axial direction and fully located on the hoop axis. However, in our results, the maximum in the (0002) basal pole figure appeared at a ∼55 • inclination. This deviation may be explained citing the results of Prangnell et. al. [42], where the authors showed with the help of finite element modelling that the filling of the outer corner of the die in conventional ECAP was dependent on whether the friction between the material and the die wall was present. The present outer curvature of the die corner could theoretically change the real shear plane, which resulted in a slightly different observed macro-texture from the theoretically predicted one of the material. The material flow in the die was also simulated in reference [43], where it turned out that its distribution was highly dependent on the value of the friction coefficient. Furthermore, as discussed at the end of Section 3.2, different conditions prevailed in different parts of the C-ECAP machine. In reference [39] (Figure 2), it was shown that in the corners and the adjacent areas, the strain rate and velocity distribution was practically zero. This may be additional evidence that the real shear plane can be different from the predicted one ( Figure 11). regions, the rotation with respect to the radial axis was dependent on the region and the number of passes. The theoretical shear plane normal was inclined by 45° from the extrusion-axial direction due to the C-ECAP design. If we assumed, for example, after the first pass in the central region, that the main deformation mode was a dislocation slip in the basal (0002) planes in direction 〈112 ̅ 0〉, then theoretically the "50-60" component in the (0002) pole figure should be inclined exactly by 45° from the axial direction and fully located on the hoop axis. However, in our results, the maximum in the (0002) basal pole figure appeared at a ~55° inclination. This deviation may be explained citing the results of Prangnell et. al. [42], where the authors showed with the help of finite element modelling that the filling of the outer corner of the die in conventional ECAP was dependent on whether the friction between the material and the die wall was present. The present outer curvature of the die corner could theoretically change the real shear plane, which resulted in a slightly different observed macro-texture from the theoretically predicted one of the material. The material flow in the die was also simulated in reference [43], where it turned out that its distribution was highly dependent on the value of the friction coefficient. Furthermore, as discussed at the end of section 3.2., different conditions prevailed in different parts of the C-ECAP machine. In reference [39] (Figure 2), it was shown that in the corners and the adjacent areas, the strain rate and velocity distribution was practically zero. This may be additional evidence that the real shear plane can be different from the predicted one ( Figure 11). Figure 11. Simulated strain rate, as in reference [39]. In summary, it can be concluded that the activity of the particular slip system depended on the number of passes and the relative position within the specimen. After the first pass in the central region, the basal slip system 〈112 ̅ 0〉 (0002) was the dominant slip mode. However, in the peripheral regions, the prismatic slip system 〈112 ̅ 0〉 (101 ̅ 0) and pyramidal slip system of the first order 〈112 ̅ 0〉 (101 ̅ 1) were activated, which could be deduced from the positions of the maxima of the measured pole figures. In the first case, i.e., the basal slip in the central region after the first pass (Figure 8b), the maximum of the "50-60" component in the (0002) pole figure lay fully on the hoop axis and was inclined by 55° from the axial direction. Thus, the (0002) basal planes were parallel with the mentioned other shear plane and preferably oriented for shear deformation in the vertical direction on the shear plane. In the second case, i.e., the prismatic and pyramidal first order slip in the periphery (Figure 8a,c), the maxima of the "50-60" component of the (0002) pole figure was rotated by ~120° from the hoop axis on each side. The maxima of the (101 ̅ 0) pole figures were inclined by ~30° from the axial direction and rotated by ~60° from the hoop axis. This arrangement of texture maxima allowed slipping 〈〉 dislocations in the prismatic and in pyramidal planes [44] in the direction towards the center of the work-piece. This was due to the fact that in peripheral regions, the shear direction had a normal component to the edge of the die beside the vertical one. Consequently, Figure 11. Simulated strain rate, as in reference [39]. In summary, it can be concluded that the activity of the particular slip system depended on the number of passes and the relative position within the specimen. After the first pass in the central region, the basal slip system 1120 (0002) was the dominant slip mode. However, in the peripheral regions, the prismatic slip system 1120 1010 and pyramidal slip system of the first order 1120 1011 were activated, which could be deduced from the positions of the maxima of the measured pole figures. In the first case, i.e., the basal slip in the central region after the first pass (Figure 8b), the maximum of the "50-60" component in the (0002) pole figure lay fully on the hoop axis and was inclined by 55 • from the axial direction. Thus, the (0002) basal planes were parallel with the mentioned other shear plane and preferably oriented for shear deformation in the vertical direction on the shear plane. In the second case, i.e., the prismatic and pyramidal first order slip in the periphery (Figure 8a,c), the maxima of the "50-60" component of the (0002) pole figure was rotated by~120 • from the hoop axis on each side. The maxima of the 1010 pole figures were inclined by ∼30 • from the axial direction and rotated by ∼60 • from the hoop axis. This arrangement of texture maxima allowed slipping a dislocations in the prismatic and in pyramidal planes [44] in the direction towards the center of the work-piece. This was due to the fact that in peripheral regions, the shear direction had a normal component to the edge of the die beside the vertical one. Consequently, the stress fields of the different types of dislocations (basal, prismatic, and pyramidal) differently affected the residual strain distribution in particular regions. Obviously, a very large amount of basal dislocations was generated during the first pass owing to the initial texture in the central region (Figure 4b). This resulted in a large compression residual strain in the axial direction (Figure 2a-C1). After the second pass, the prismatic and pyramidal dislocations also appeared beside the basal dislocations due to the rotated "50 • -60 • " component. This coexistence of basal, prismatic, and pyramidal dislocations in the central region could be observed after the third pass as well. In the periphery, only the prismatic and pyramidal dislocations were active after the first and second pass. They generated a stress field, which acted in the hoop direction mainly after the first C-ECAP pass (Figure 2b). However, after the second pass, they acted in the axial direction as well (Figure 2a,b). It could be assumed, that the grain size influenced the dislocations effects in the different directions. After the third pass, they were only active in the right periphery as can be seen from the maxima of the (0002) and 1010 pole figures (Figure 10a,c). Residual Stress Calculation Rresults The residual stress was calculated using the method described in Section 2. EBSD and X-ray texture measurements were conducted in the three regions (left and right peripheral regions, and the central region) of the samples; hence, the calculation was provided for each of the three investigated regions and passes with separate ODFs. The grain shapes for particular regions and passes were determined from our EBSD results in the cross section and from reference [39] in the longitudinal section of the samples. The grain size and shape did not depend significantly on the region in the sample for particular passes, as seen in Section 3.2. Thus, it could be assumed that the same statement was valid as well for the axial direction. Points −3, −2 belonged to the left peripheral region, −1, 1, 1 to the central region, and 2, 3 to the right peripheral region. The results of the calculation are seen in Figure 12a-c. the stress fields of the different types of dislocations (basal, prismatic, and pyramidal) differently affected the residual strain distribution in particular regions. Obviously, a very large amount of basal dislocations was generated during the first pass owing to the initial texture in the central region ( Figure 4b). This resulted in a large compression residual strain in the axial direction (Figure 2a-C1). After the second pass, the prismatic and pyramidal dislocations also appeared beside the basal dislocations due to the rotated "50°-60°" component. This coexistence of basal, prismatic, and pyramidal dislocations in the central region could be observed after the third pass as well. In the periphery, only the prismatic and pyramidal dislocations were active after the first and second pass. They generated a stress field, which acted in the hoop direction mainly after the first C-ECAP pass (Figure 2b). However, after the second pass, they acted in the axial direction as well (Figure 2a,b). It could be assumed, that the grain size influenced the dislocations effects in the different directions. After the third pass, they were only active in the right periphery as can be seen from the maxima of the (0002) and (101 ̅ 0) pole figures (Figure 10a,c). Residual Stress Calculation Rresults The residual stress was calculated using the method described in section 2. EBSD and X-ray texture measurements were conducted in the three regions (left and right peripheral regions, and the central region) of the samples; hence, the calculation was provided for each of the three investigated regions and passes with separate ODFs. The grain shapes for particular regions and passes were determined from our EBSD results in the cross section and from reference [39] in the longitudinal section of the samples. The grain size and shape did not depend significantly on the region in the sample for particular passes, as seen in section 3.2. Thus, it could be assumed that the same statement was valid as well for the axial direction. Points −3, −2 belonged to the left peripheral region, −1, 1, 1 to the central region, and 2, 3 to the right peripheral region. The results of the calculation are seen in Figure 12a-c. It could be seen from the stress distributions (Figure 12a-c), that the distributions followed almost the same shape as the strain distributions from Section 3.1 (Figure 2a-c). In the center of the sample C1 in the axial direction (Figure 12a), a relatively large, −63 MPa, stress prevailed, as was expected from the strain measurements. In the left peripheral region, the C1 sample exhibited +14 MPa tensile stress, whilst the right peripheral region was practically stress-free in the axial direction in contrast to the measured compression strain (Figure 2a). It was interesting to see the change of sign between the strain and the stress at points −3, −2, and 2 mm. At points −3, −2, and 2 mm, the strain was −17, −39, and −82 µε, respectively, but the stress was +14, +14, and +2 MPa. The same behavior was spotted at point −3 in the C2 sample ( Figure 12a) and at points 2 and 3 mm in the sample C3 in the radial direction (Figure 12c). Sample C2 exhibited a very symmetric distribution of axial stresses around the center of the sample with the maximum of compression stress at −37 MPa. The compression stress increased from the edges towards the center of the sample C2. The C3 sample showed a large, −63 MPa, stress at the right peripheral point −3 mm. In the central region (−1, 0, and 1 mm) and in the right peripheral region (1 and 2 mm) the residual compression stress decreased. In the hoop direction (Figure 12b), samples C1 and C2 showed a very symmetric distribution of stresses around the center; however, it was in opposite progress than it was in the axial direction. These samples had the lowest value of compression stress in the center and it increased towards the edges. Sample C1 seemed to be stress-free in the center. Sample C3 showed a linearly increasing compression stress from the left to the right edge in the hoop direction. The residual compression stresses did not show any gradient in the radial direction for samples C1 and C2, though at point −3 mm it was slightly increased (Figure 12c), and they were distributed at around −18 MPa. Additionally, a very inhomogeneous distribution could be seen at sample C3. Its maximum compression stress was at point −1 mm. From this point, the stress decreased towards the right edge of the sample until its sign changed between points 1 and 2 mm. The right edge of the sample exhibited tensile stress at about +20 MPa. Conclusions Neutron diffraction measurements revealed the residual strain distributions of the samples after one, two, and three passes of the C-ECAP machine. Symmetric strain gradients were present in the case of samples after first and second pass in the axial and hoop directions. In the radial direction, there was no gradient and also the strain distribution was constant through the sample. The third pass broke the gradient symmetry in the axial and hoop direction. Moreover, in the radial direction, the constant strain distribution was interrupted as well. EBSD observations confirmed the grain refinement effect of multiple pass C-ECAP treatment; however, after the second pass, the grain refinement was not as significant as after the first pass. Deformation twins were not observed, though twins with nano-sized width may have been present. Thus, it is most likely that a dislocation slip was responsible for the deformation of the Ti during the C-ECAP treatment. This concept was supported by the experimental pole figures from the EBSD measurements and the local X-ray macro-texture measurements. In addition, the macro-texture of the samples was dependent on the investigated region, which could indicate the activation and combination of different slip systems as the function of the position in the work-piece. The calculation of the residual stress distributions showed that compression residual stresses prevailed in the specimens after the C-ECAP process and they followed the shape of the distributions of the residual strains. Author Contributions: K.M. and G.N. conceived and designed the experiments; J.P. and M.D. prepared the C-ECAPed materials; C.H. performed the neutron diffraction experiments; G.N. prepared the samples for EBSD and X-ray texture measurements; K.H. studied microstructure of the samples using EBSD; P.C. performed the X-ray texture measurements; G.N. and K.M. analyzed the data; G.N. wrote the paper.	v3-fos-license
2018-04-03T01:29:57.893Z	2012-12-21T00:00:00.000	22098229	{ "extfieldsofstudy": [ "Biology", "Medicine" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "http://www.jbc.org/content/288/7/4538.full.pdf", "pdf_hash": "723c38bdb4fd5196e88132c65601af4aa71b95ca", "pdf_src": "Highwire", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45328", "s2fieldsofstudy": [ "Biology", "Medicine" ], "sha1": "070899e3d74466c49d483f28dff60b7a29de4433", "year": 2012 }	pes2o/s2orc	Identification of the Low Density Lipoprotein (LDL) Receptor-related Protein-1 Interactome in Central Nervous System Myelin Suggests a Role in the Clearance of Necrotic Cell Debris* Background: LRP1 is a scavenger receptor involved in the clearance of apoptotic cells and myelin vesicles. Results: Novel ligands for LRP1 were discovered in CNS myelin by affinity purification combined with proteomics. Conclusion: Some ligands are intracellular proteins, suggesting a function for LRP1 in the clearance of necrotic debris. Significance: LRP1 mediates the removal of cellular waste and could maintain homeostasis. In the central nervous system (CNS), fast neuronal signals are facilitated by the oligodendrocyte-produced myelin sheath. Oligodendrocyte turnover or injury generates myelin debris that is usually promptly cleared by phagocytic cells. Failure to remove dying oligodendrocytes leads to accumulation of degraded myelin, which, if recognized by the immune system, may contribute to the development of autoimmunity in diseases such as multiple sclerosis. We recently identified low density lipoprotein receptor-related protein-1 (LRP1) as a novel phagocytic receptor for myelin debris. Here, we report characterization of the LRP1 interactome in CNS myelin. Fusion proteins were designed corresponding to the extracellular ligand-binding domains of LRP1. LRP1 partners were isolated by affinity purification and characterized by mass spectrometry. We report that LRP1 binds intracellular proteins via its extracellular domain and functions as a receptor for necrotic cells. Peptidyl arginine deiminase-2 and cyclic nucleotide phosphodiesterase are novel LRP1 ligands identified in our screen, which interact with full-length LRP1. Furthermore, the extracellular domain of LRP1 is a target of peptidyl arginine deiminase-2-mediated deimination in vitro. We propose that LRP1 functions as a receptor for endocytosis of intracellular components released during cellular damage and necrosis. In the central nervous system (CNS), fast neuronal signals are facilitated by the oligodendrocyte-produced myelin sheath. Oligodendrocyte turnover or injury generates myelin debris that is usually promptly cleared by phagocytic cells. Failure to remove dying oligodendrocytes leads to accumulation of degraded myelin, which, if recognized by the immune system, may contribute to the development of autoimmunity in diseases such as multiple sclerosis. We recently identified low density lipoprotein receptor-related protein-1 (LRP1) as a novel phagocytic receptor for myelin debris. Here, we report characterization of the LRP1 interactome in CNS myelin. Fusion proteins were designed corresponding to the extracellular ligand-binding domains of LRP1. LRP1 partners were isolated by affinity purification and characterized by mass spectrometry. We report that LRP1 binds intracellular proteins via its extracellular domain and functions as a receptor for necrotic cells. Peptidyl arginine deiminase-2 and cyclic nucleotide phosphodiesterase are novel LRP1 ligands identified in our screen, which interact with full-length LRP1. Furthermore, the extracellular domain of LRP1 is a target of peptidyl arginine deiminase-2-mediated deimination in vitro. We propose that LRP1 functions as a receptor for endocytosis of intracellular components released during cellular damage and necrosis. Multiple sclerosis (MS) 2 is an autoimmune disease in which CNS myelin is destroyed and the survival of myelin-producing oligodendrocytes is compromised (1). The etiology of MS is poorly understood, but, in some cases, oligodendrocyte apoptosis is detected without the involvement of immune cell infiltrates (2). Thus, the failure to clear apoptotic cells and cellular debris may initiate inflammation and the autoimmune response characteristic for MS (3). Low density lipoprotein receptor-related protein-1 (LRP1) is a scavenger receptor with key roles in the phagocytosis of myelin debris (4) and apoptotic cells (5,6). LRP1 is a member of the LDL receptor gene family, first recognized as a receptor for apolipoprotein E and the serum protease inhibitor ␣ 2 -macroglobulin (7,8). To date, Ͼ40 different ligands for LRP1 have been identified, including proteases, growth factors, heat shock proteins, extracellular matrix proteins, and foreign toxins (9). LRP1 is a two-chain receptor in which the ligand-binding ␣ chain is entirely extracellular and non-covalently associated with the membrane-spanning ␤ chain (9). Upon binding to LRP1 on the cell surface, LRP1 ligands are internalized, dissociate in acidified endosomes, and are delivered to lysosomes, whereas LRP1 is recycled to the cell surface (10). LRP1 also mediates endocytosis of cell surface receptors (11) and, in this manner, participates in the regulation of cell signaling events (12,13). In addition to its function as a phagocytic receptor for myelin, LRP1 also participates in myelin-mediated inhibition of axon formation (14). The goal of this study was to characterize the interactome of LRP1 in myelin. We designed fusion proteins corresponding to the second and fourth clusters of complement-like repeats (CCRII and CCRIV), two major extracellular ligand-binding domains in LRP1 (15). CCRII or CCRIV fusion proteins were incubated with myelin protein extracts in the presence or absence of RAP, an LRP1 ligand binding inhib-itor (16), and associated partners were identified by tandem mass spectrometry. Using this approach, we have identified Ͼ70 LRP1 ligands in CNS myelin. Validation studies were performed for two myelin-specific LRP1 ligands: cyclic nucleotide phosphodiesterase (CNP, also called CNPase) and peptidyl arginine deiminase-2 (PAD2). CNP is an enzyme highly expressed in oligodendrocytes, with functions in tubulin polymerization and oligodendrocyte process outgrowth (17). CNP is an abundant component of myelin, accounting for up to 4% of the total protein content (18). Although myelination is normal in CNP knock-out mice, they succumb to axonal loss and neurodegeneration due to impaired communication between cells (19). Furthermore, CNP is a potential auto-antigen in MS, as CNPreactive T cells can be found in MS patients with active disease (20,21). Another potential auto-antigen in MS is the myelin basic protein (MBP), which we previously identified as a ligand for LRP1 (4). MBP is a substrate for our second chosen target for validation, enzyme peptidyl arginine deiminase 2, or PAD2 (22). PAD2 catalyzes the conversion of protein arginine residues into citrulline, in a reaction called deimination or citrullination (23). Similar to CNP, PAD2 is a component of the myelin sheath and is the only member of the PAD family expressed in the healthy CNS (23,24). MS patients contain higher levels of MBP citrullination, which is thought to contribute to MBP degradation and loss of myelin sheath integrity (25)(26)(27)(28). Similarly, MBP citrullination and PAD2 activity correlate with disease in animal models of MS (29,30). Our studies demonstrate that both CNP and PAD2 interact with LRP1 CCRs as well as with the full-length LRP1. Finally, we show that LRP1 can be citrullinated by endogenous PAD2 in vitro and that citrullination decreases the endocytic function of LRP1. Interestingly, many of the newly identified LRP1 ligands were intracellular proteins, indicating a novel role for LRP1 in the clearance of cellular debris, potentially released during necrosis. We further demonstrate for the first time that LRP1 regulates the removal of necrotic cells and cellular debris using model T cells and oligodendrocytes. Necrotic waste is harmful to surrounding cells and is generated during catastrophic events such as trauma and hypoxia or when the apoptotic cell clearance system is saturated (31). Furthermore, necrotic debris has been shown to initiate a proinflammatory response by macrophages (32). During MS, improper clearance of apoptotic cells and necrotic debris would thus generate an inflammatory environment, which would further fuel disease severity. Taken together, the data presented here suggest that, by mediating cellular and myelin-associated debris removal after cell damage, LRP1 may inhibit the immune response generated by excessive cellular debris at sites of inflammation. Cloning, Expression, and Purification of CCRII and CCRIV-CCRII and CCRIV from human LRP1 were cloned into pFuse-rFC2 (Invivogen, San Diego, CA) as described in the literature (15) to generate Fc-fusion proteins. CCRII includes EGF repeat 4 and complement repeats 3 to 10 (LRP1 amino acids 806 to 1184). CCRIV spans complement repeats 21 to 31 (LRP1 amino acids 3332 to 3778). CCRII and CCRIV were expressed in CHO-K1 cells and purified from the culture supernatants on protein A-agarose resin (GE Healthcare). CCR Pulldowns and Proteomics-Myelin vesicles were purified from mouse brains as described previously (35). Myelinassociated proteins were solubilized in radioimmune precipitation assay buffer, containing 100 mM Tris, pH 7.5, 150 mM NaCl, 1% Triton X-100, 0.5% deoxycholate, and 0.1% SDS, supplemented with 1 mM CaCl 2 and complete protease inhibitor mixture (Roche Applied Science). Protein extracts (2 mg) were incubated with 1 pm of CCRII, CCRIV, or Fc with or without 20 pm of GST-RAP for 16 h at 4°C. CCRII, CCRIV, and Fc were recovered by adding protein A-agarose beads for 1 h at 20°C. After extensive washing with radioimmune precipitation assay buffer, proteins were either digested with trypsin or eluted with SDS sample buffer for SDS-PAGE and immunoblot analysis. Trypsin digestion was performed in the presence of Protease-MAX surfactant as described by the manufacturer (Promega, Madison, WI). Proteins associated with CCRII, CCRIV, and Fc were identified by LC-MS/MS as described previously (36). Cloning, Expression, and Purification of PAD2-His, GFP-RAP, and GST-CNP-Murine PAD2 open reading frame (ORF) was cloned into pET-30a(ϩ) vector (Novagen). Human RAP ORF was cloned downstream of GFP into pET-30a(ϩ). pET-30A(ϩ) allows bacterial expression and contains a histidine tag for affinity purification. Murine CNP ORF was cloned into pGEX-2T (GE Healthcare) to allow bacterial expression and GST tag-mediated purification. PAD2-His and GST-CNP were purified by affinity chromatography using the Profinia purification kit and chromatography system (Bio-Rad). Immobilization Binding Studies-Purified BSA, rat LRP1, or fibronectin (25 nM) (Sigma) were diluted in PBS and absorbed on ELISA plates for 12 h at 4°C. Wells were blocked for 1 h with PBS containing 1% BSA. PAD2-His (25 nM) was added to the wells for 1 h at 20°C. The wells were washed, and retained PAD2-His was detected using anti-PAD2 serum, in an ELISA format using ABTS as a colorimetric substrate. Alternatively, purified BSA, GST, GST-CNP, or GST-RAP (350 nM) were diluted in PBS and absorbed on ELISA plates for 12 h at 4°C. Wells were blocked for 1 h with PBS containing 1% BSA. Human shed LRP1 (50 nM) was added to the wells for 1 h at 20°C. The wells were washed with PBS, and retained LRP1 was detected with antibody 8G1, in an ELISA format using ABTS as a colorimetric substrate. Citrullination Assay-RAW 264.7 cell extract and brain protein extract were prepared in ice-cold extraction buffer (50 mM HEPES, 150 mM NaCl, 1% Triton X-100, 10% glycerol) containing complete protease inhibitor mixture without EDTA (Roche Applied Science). Equal amounts of protein extracts were incubated with 10 g of PAD2-His for 3 h at 37°C in the presence or absence of 10 mM CaCl 2 . LRP1 was recovered by adding 10 g of GST-RAP and glutathione-agarose beads (33). Glutathioneagarose beads were washed with the extraction buffer. Proteins were eluted with SDS sample buffer for SDS-PAGE and immunoblot analysis. For cell surface citrullination, LRP1-expressing and -deficient fibroblasts were treated for 6 h at 37°C with 150 nM recombinant PAD2 in DMEM containing 0.3% BSA. Cell surface proteins were labeled with biotin and purified as described (36). GFP-RAP Endocytosis-LRP1-positive (PEA10) and LRP1deficient (MEF2) fibroblasts were treated for 3 h at 37°C with 150 nM recombinant PAD2 in DMEM containing 0.3% BSA. After washing, cells were treated with 500 nM of GFP-RAP for 3 h at 37°C. After washing with PBS, cells were detached with trypsin and washed twice with PBS containing 5% FBS. Cells were analyzed by flow cytometry on a BD FACSCalibur (BD Biosciences). Data were plotted using the FlowJo software (Tree Star). Phagocytosis of Necrotic Cells-LRP1-positive (PEA10) and LRP1-deficient (MEF2) fibroblasts were labeled with CFSE according to the manufacturer's instructions (Invitrogen). To prepare necrotic cells, Jurkat or N20.1 cells were heat shocked at 55°C for 20 min and then incubated for 4 h at 37°C. Necrosis was confirmed by microscopy using trypan blue exclusion, or by flow cytometry using annexin V-FITC and 7-amino-actinomycin D staining, according to manufacturers instructions (eBioscience). Necrotic and live cells were labeled with Cypher5 (10 M, GE Healthcare) for 20 min at 37°C in DMEM, and excessive dye was quenched by resuspending the cells in DMEM containing 10% FBS. Necrotic and live cells were then added to the CFSE-labeled fibroblasts in DMEM containing 10% FBS for 2 h at 37°C (10 to 1). After washing with PBS, cells were detached with trypsin and washed twice with PBS containing 5% FBS. In experiments with BV2 cells, necrotic and live cells were added for 30 min at 37°C (4:1). BV2 cells were distinguished from N20.1 by staining with CD11b-Alexa Fluor 488 antibody. Cells were analyzed by flow cytometry on a BD FACSCalibur (BD Biosciences). Data were plotted using the FlowJo software (Tree Star). RESULTS Preparation of CCRII and CCRIV-To identify LRP1 ligands present in CNS myelin, we prepared two fusion proteins corresponding to CCRII and CCRIV of human LRP1 (Fig. 1A). These two CCRs mediate the binding of LRP1 ligands, with the possi-FIGURE 1. CCRII and CCRIV bind LRP1 ligands. A, schematic representation of LRP1 and CCRII and CCRIV constructs. B, Fc, CCRII, and CCRIV were incubated with GST-RAP as described under "Experimental Procedures," and GST-RAP was recovered by incubation with glutathione-agarose beads. Samples and purified proteins were analyzed by Coomassie staining. C, Fc, CCRII, and CCRIV were incubated with myelin protein extract in the presence or absence of GST-RAP. Fc and CCRs were recovered by adding protein A-agarose beads. Samples were analyzed by Coomassie staining. ble exception of ␣ 2 -macroglobulin (9,15). CCRII and CCRIV were expressed as soluble Fc-fusion proteins in CHO-K1 cells (Fig. 1A) and tested by examining binding of GST-RAP in a pulldown assay. RAP binds to LRP1 and blocks the binding of most known LRP1 ligands (9). Equivalent amounts of CCRII, CCRIV, or Fc (control) were incubated with GST-RAP, followed by incubation with glutathione-agarose beads. After washing, samples were separated by SDS-PAGE alongside the purified proteins, and the gel was stained with Coomassie Blue. CCRII and CCRIV, but not Fc alone, bound to GST-RAP (Fig. 1B). Next, equivalent amounts of CCRII, CCRIV, or Fc were incubated with CNS myelin protein extracts in the presence or absence of excess GST-RAP. Fc fusion proteins were recovered by pull down with protein A-agarose beads. As shown in Fig. 1C, several protein bands are purified from myelin with CCRII and CCRIV, but not with Fc (arrowheads). Treatment with GST-RAP blocks the binding of these proteins to CCRs (Fig. 1C). These results demonstrate that the recombinant Fc-fusion proteins, CCRII and CCRIV, are functional in ligand binding and may be used to identify LRP1 ligands. Identification of LRP1 Ligands in CNS Myelin by Tandem Mass Spectrometry-CCRII, CCRIV, or Fc were incubated with CNS myelin protein extracts. Associated proteins were recovered by protein A-agarose pulldown as described above. As a control, CCRII, CCRIV, and Fc were preincubated with a 20-fold molar excess of GST-RAP to inhibit ligand binding. Preparations were trypsin-digested, and the resulting peptides were identified by mass spectrometry using an LTQ-Orbitrap. As an additional control, fusion proteins also were analyzed without incubation with myelin extracts to identify ligands derived from CHO-K1 cells, which may co-purify with the CCRs. Using the described method, 72 proteins were identified as ligands for CCRII and CCRIV (supplemental Table 1). As expected, LRP1 peptides were present only when CCRs were analyzed and not Fc. LRP1 peptide abundance was not affected by the presence of myelin protein or by a large excess of GST-RAP, indicating that these reagents did not influence CCRbinding to protein A-agarose. LRP1 partners identified in our screens were separated into two categories: myelin-specific LRP1 ligands (Table 1 and Fig. 2A) and CHO-K1 cell proteins co-purifying with CCRs (Table 2 and Fig. 2B). When CCRII and CCRIV were used to identify myelin-specific LRP1 ligands (Table 1 and Fig. 2A), we confirmed the binding of MBP and myelin-associated glycoprotein, two proteins that we have previously reported as ligands for LRP1 (4,14). The myelin proteins, CNP and PLP, also were identified as ligands for LRP1 using our screen. Surprisingly, myelin-specific CCR partners included many intracellular proteins. We identified members of the septin family (septins 2, 4, 7, and 8) and microtubule-associated proteins, Mtap1a and Mtap1ab, as novel LRP1 ligands (Table 1 and Fig. 2A). As shown in Table 2 and in Fig. 2B, 10 proteins co-purified at significant levels with CCRII and CCRIV when the fusion proteins were expressed in CHO-K1 cells. This list includes known LRP1 ligands such as RAP, matrix metalloproteinase-9, and tissue inhibitor of metalloproteinase 2 (37,38) together with previously unidentified LRP1 ligands, such as the proteases matrix metalloproteinase-19 and legumain, and the extracellular matrix proteins SPARC and collagen 5 (col5a2). As anticipated, following GST-RAP treatment, the spectral counts for RAP binding to CCRII and CCRIV increased from 10 to 71 and from 15.2 to 46.2, respectively. This result shows that exogenous RAP can efficiently bind to the CCRs. In both categories of LRP1 ligands, comparable binding to CCRII and CCRIV was most frequently observed. However, we were also able to identify proteins that bind specifically to CCRII or CCRIV, but not both. Collagen 5 (Col5a2), PAD2 or Padi2, and MYO18 (Myo18a) bind only to CCRII, whereas legumain, tissue inhibitor of metalloproteinase 2, and FHL1 (four and a half LIM domains 1) bind only to CCRIV (Fig. 2B). Taken together, our results suggest that the binding sites for a variety of LRP1 ligands reside within a specific domain of LRP1, allow- FEBRUARY 15, 2013 • VOLUME 288 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4541 ing inhibitor design for targeted disruption of specific LRP1ligand interactions. PAD2 and CNP Are Novel LRP1 Ligands in Myelin-To validate the results obtained using proteomics we chose a candidate approach. We selected two myelin-specific proteins, CNP and PAD2, as targets for validation, due to the proposed involvement of these proteins in neurodegeneration and MS. CNP is a major protein component of myelin, an early marker of oligodendrocyte myelination (39), and a potential auto-antigen in MS (20,21). The proposed function of CNP is to mediate the compaction of myelin by organizing the tubulin network (40). PAD2 is a member of the PAD family of Ca 2ϩ -dependent enzymes, which catalyze the post-translational modification of protein arginines, in a process called deimination or citrullination (24). PAD2 expression and the level of citrullination correlate with active disease in both human MS and the animal models of MS (25)(26)(27)(28)(29)(30). We first confirmed binding of PAD2 and CNP to CCRII and CCRIV by immunoblot analysis. As shown in Fig. 3A, PAD2 in whole CNS myelin interacted with CCRII, but not with CCRIV or Fc, confirming the results of our proteomic analysis (Fig. 3A). CNP bound to both CCRII and CCRIV but not the control protein Fc (Fig. 3B). GST-RAP treatment completely inhibited association of both PAD2 and CNP with the CCRs (Fig. 3). We next determined whether recombinant PAD2 and CNP bind to full-length LRP1. We expressed PAD2 as a His-tagged fusion protein and CNP as a GST-tagged fusion protein in bacteria, as described under "Experimental Procedures." Protein preparations were first tested for purity by SDS-PAGE. As shown in Fig. 4, recombinant PAD2 and CNP migrated as single bands with an apparent size of 80 and 60 kDa, respectively, as anticipated (Fig. 4A). Binding of PAD2-His and GST-CNP to LRP1 was then evaluated using an ELISA format. Purified LRP1, fibronectin, and BSA were immobilized in microtiter plates. Recombinant PAD2-His bound to LRP1, but did not bind to purified fibronectin or BSA under the same conditions (Fig. 4B). To study binding of LRP1 to CNP, BSA, GST, GST-CNP, and GST-RAP were immobilized in microtiter plates. Shed LRP1, which was purified from human plasma, was then incubated with the immobilized proteins. Binding of shed LRP1 to GST-CNP wells was significantly increased compared with binding to BSA or GST alone (Fig. 4C). Shed LRP-1 binding to immobilized GST-RAP was used as a positive control. LRP1 Is Citrullinated by PAD2-Protein citrullination converts positively charged arginine residues into the neutral residue, citrulline, which can substantially alter protein structure and function (23). To test whether LRP1 is a target for PAD2mediated citrullination, we performed in vitro citrullination assays. RAW 264.7 cell protein extracts were incubated with or without PAD2 in the presence or absence of Ca 2ϩ . LRP1 was then recovered by affinity precipitation with GST-RAP, and samples were subjected to immunoblot analysis. As is shown in Fig. 5A, when cell extracts were incubated with PAD2 in the presence of Ca 2ϩ , LRP1 citrullination was readily detected. As expected, LRP1 was precipitated under all conditions. Other proteins also were citrullinated by PAD2, as determined by immunoblotting for citrullinated proteins in the RAW 264.7 cell extracts. As a control, we determined that levels of LRP1 and Grp78 were present in equal quantity in the initial cell extracts incubated with PAD2. LRP1 Ligands in CNS Myelin with recombinant PAD2 and Ca 2ϩ . LRP1 was recovered by affinity precipitation with GST-RAP and protein citrullination was examined. As is shown in Fig. 5B, in the absence of added Ca 2ϩ , we did not detect citrullination of LRP1. However, when we added Ca 2ϩ to the brain extracts, to activate endogenous PAD, LRP1 citrullination was observed. Because PAD2 is the only member of the PAD family expressed in the healthy CNS, it is reasonable to conclude that endogenous PAD2 citrullinates LRP1 (24). Adding recombinant PAD2 to mouse brain extracts, in the presence of Ca 2ϩ , induced robust LRP1 citrullination. (Fig. 5B). In the CNS extracts, LRP1 was one of many proteins that are citrullinated by PAD2, as determined by immunoblot analysis for citrullinated proteins in protein extracts (Fig. 5B). Next, we sought to determine whether extracellular PAD2 can citrullinate cell surface proteins. LRP1-positive and -deficient fibroblasts were treated with recombinant PAD2 for 6 h at 37°C. Cell surface proteins were labeled with membrane impermeable biotin and recovered by streptavidin affinity precipitation. Affinity precipitates were then analyzed by immunoblot for the presence of citrullination and LRP1. As is shown in Fig. 6A, PAD2 induces a robust citrullination of cell surface proteins. Furthermore, cell surface citrullination is increased in LRP1 positive cells (Fig. 6A). LRP1 was only detected in LRP1 positive cells, as expected. We further tested whether PAD2-mediated citrullination of cell surface proteins affects the endocytic function of LRP1. LRP1-positive and -deficient fibroblasts, previously treated with recombinant PAD2 for 3 h at 37°C, were incubated with GFP-RAP for 3 h (Fig. 6B). After extensive washing, GFP fluorescence was assessed by flow cytometry as a measure of LRP1mediated endocytosis. No fluorescence was detected with LRP1-deficient cells, as expected (Fig. 6B). PAD2 treatment induces a small but significant decrease (12 Ϯ 2%, n ϭ 4) in GFP-RAP association with LRP1-positive cells. Therefore, our data suggests that citrullination of cell surface proteins by PAD2 can decrease the endocytic function of LRP1. LRP1 Is a Phagocytic Receptor for Necrotic Cells-Multiple LRP1-specific ligands identified in our proteomic screen are myelin-enriched cytoplasmic proteins, including components of the cytoskeleton. We thus hypothesized that LRP1 could be a novel receptor involved in the clearance of necrotic cells and cellular debris. To test this hypothesis, we prepared two types of necrotic cells by heat shock treatment: the T cell-like Jurkat cell line and the oligodendrocyte cell line N20.1 (34). Four hours post induction of necrosis, Ͼ95% of the cells are trypan bluepositive, indicating the loss of plasma membrane integrity (data not shown). Furthermore, necrotic Jurkat cells were 95% positive for both annexin-5 and 7-amino-actinomycin D when analyzed by flow cytometry (Fig. 7A). Finally, necrotic N20.1 cells showed ruffling of the plasma membrane when compared with live cells by phase contrast microscopy (Fig. 7B). Necrotic Jurkat cells were labeled with CypHer5, a pH-sensitive fluorescent dye (41), and added to CFSE-labeled LRP1 positive and deficient fibroblasts. After 2 h of incubation at 37°C, engulfment of necrotic cells was analyzed by flow cytometry. As shown in Fig. 7, LRP1-positive cells are phagocytosing necrotic Jurkat cells more efficiently than LRP1-deficient cells (2.7-fold increase, p Ͻ 0.0001, n ϭ 3; Fig. 7, C and D). On the TABLE 2 Proteins co-purifying with CCRII and CCRIV in CHO-K1 cells The gene products below were identified by tandem mass spectrometry as partners for LRP1 CCRs with (ϩ) or without (Ϫ) RAP. Mean spectral counts and S.D. are shown (n ϭ 3). Gene name Fc other hand, engulfment of live Jurkat cells was not significantly different between LRP1-positive and -deficient fibroblasts. In a complimentary approach, when LRP1-positive fibroblasts were pretreated with an LRP1 antagonist (GST-RAP), engulfment of necrotic cells was inhibited to the levels observed in LRP1-deficient cells (supplemental Fig. 1A). To next examine the function of LRP1 in the phagocytosis of CNS myelin debris, we repeated necrotic cell internalization experiments using the microglial cell line BV2 as model phagocytes and the oligodendrocyte cell line N20.1 as model necrotic cells (34,42). When BV2 were pretreated with GST-RAP, phagocytosis of necrotic oligodendrocytes was inhibited by 35% (p Ͻ 0.001, n ϭ 6; Fig. 7, E and F), whereas no significant difference was detected with live oligodendrocytes following GST or GST-RAP treatment. Finally, endocytosis of necrotic oligodendrocytes was also inhibited in LRP1-deficient fibroblasts (supplemental Fig. 1B). Together, our results demonstrate that LRP1 binds intracellular proteins present in myelin. We also describe a new function for LRP1 as a phagocytic receptor for necrotic cells and cellular debris that could be present in the CNS. DISCUSSION Proper clearance of dead cells is critical for tissue homeostasis. Defects in apoptotic cell clearance can lead to secondary necrosis and release of cellular components into the extracellu- , and LRP1 were adsorbed onto plastic wells and incubated with purified PAD2-His. PAD2 binding to the immobilized phase was detected in an ELISA format, using anti-PAD2 serum, as described under "Experimental Procedures." *, p Ͻ 0.001. C, BSA, GST, GST-CNP, and GST-RAP were adsorbed onto plastic wells as described under "Experimental Procedures" and incubated with soluble human LRP1. LRP1 binding to the immobilized phase was detected in an ELISA format, using LRP1 ␣-chain-specific antibody 8G1. , p Ͻ 0.001. FIGURE 5. LRP1 is a substrate for PAD2-mediated citrullination. A, RAW264.7 cell extracts were incubated with or without purified PAD2 and CaCl 2 , as described under "Experimental Procedures." LRP1 was then purified using GST-RAP pulldown. Samples and input were analyzed by immunoblotting (IB) using LRP1 ␣-chain-specific antibody or with an antibody specific for citrullinated proteins. Grp78 immunoblot was used as a control of protein load. B, brain protein extract was incubated with or without CaCl 2 , or with CaCl 2 and purified PAD2, as described under "Experimental Procedures." LRP1 was then purified using GST-RAP pulldown. Samples and input were analyzed by immunoblotting using LRP1 ␣-chain-specific antibody or with an antibody specific for citrullinated proteins. lar environment, generating an inflammatory stimulus that could lead to the development of autoimmune diseases, including MS (3,43). LRP1 was recently reported to function as a phagocytic receptor for apoptotic cells and myelin debris (4 -6). Although LRP1 was originally described as a scavenger receptor for extracellular proteins (9), it is now clear that LRP1 function is more complex than just mediating endocytosis (9). LRP1 can regulate cell signaling events and functions in inflammation and neurogenesis (12,44). The goal of the present study was to characterize LRP1 ligands present in CNS myelin. The LRP1 ligand-binding domains CCRII and CCRIV were prepared as fusion proteins and used in an affinity-based proteomics screen. Using this approach, we identified Ͼ70 ligands for LRP1. These ligands could be separated into two categories based on their cellular localization: extracellular and intracellular proteins. Some of the proteins were identified because of their ability to co-purify with the CCRs during expression and purification (Table 2 and Fig. 2B). This category included known ligands for LRP1, such as RAP, matrix metalloproteinase-9, and tissue inhibitor of metalloproteinase 2 (16,37,38). The most abundant LRP1 ligand discovered in the extracellular protein category is a member of the FGF receptor family: Golgi glycoprotein 1 (GLG1)/cysteine-rich fibroblast growth factor receptor/E-selectin ligand 1. GLG1 function remains elusive but studies using GLG1 knock-out mice suggest a role during development as an FGF18 receptor (45). Furthermore, GLG1 is an E-selectin ligand involved in leukocyte extravasation during inflammation (46). By mediating the removal of GLG1 from extracellular spaces, LRP1 may modulate leukocyte migration. We also identified membrane bound myelin-specific proteins during our screen. We confirmed the interaction of MBP and myelin-associated glycoprotein with LRP1 (Table 1 and supplemental Table 1) (4,14) and also detected PLP1 as LRP1 ligands ( Table 1). The most surprising result we obtained by proteomics screening was the identification of numerous intracellular LRP1 ligands in myelin (Table 1). These included cytoskeleton proteins such as septin 2, 4, 7, and 8; MTAP1a and 1b; and ROCK. MTAP and septins are known components of myelin (47)(48)(49). As myelin wraps multiple times around the axon, the formation and maintenance of the myelin sheath is dependent on the cytoskeleton, probably explaining the abundance of cytoskeletal proteins in myelin preparations (50). We selected two intracellular proteins as candidates for validation as LRP1 binding partners: CNP and PAD2. CNP is a major component of myelin, with a potential role in myelin formation and compaction (40). Clearance of CNP by LRP1 may be critical because it has been identified as a potential auto-antigen in MS (21), and auto-reactive populations of CNP-specific T cells were identified in MS patients with active disease (20). Similarly, PAD2 is a common component of the CNS myelin (24), and its up-regulation is thought to be an early marker of demyelinating disease due to the reduced stability of the citrullinated protein components of the myelin sheath (25). We demonstrate here that LRP1 is a potential regulator of A, LRP1-positive or -negative fibroblasts were incubated with or without purified PAD2, as described under "Experimental Procedures." Cell surface proteins were labeled with membrane impermeable biotin and recovered by streptavidin affinity purification. Samples and input were analyzed by immunoblotting using LRP1 ␤-chain-specific antibody or with an antibody specific for citrullinated proteins. Tubulin immunoblot (IB) was used as a control of protein load. B, LRP1-positive or -negative fibroblasts were pretreated with or without purified PAD2 and then treated with GFP-RAP for 3 h. After washing, cell-associated GFP fluorescence was analyzed by flow cytometry. Results are normalized to the difference of mean GFP-RAP fluorescence between treated LRP1-positive cells and untreated LRP1-positive cells (n ϭ 4; , p Ͻ 0.001). ns, not significant. demyelination via interaction with CNP and PAD2. Furthermore, LRP1 can itself be post-translationally modified, at least in vitro, by endogenous PAD2 present in brain extract. We also demonstrate that LRP1 function in endocytosis is regulated by extracellular PAD2-mediated citrullination. Whether PAD2mediated citrullination affects endocytosis of only a subset, or all LRP1 ligands, remains to be determined. Finally, we demonstrate here for the first time a new function for LRP1 in clearing necrotic debris. LRP1 has been previously described as a phagocytic receptor for apoptotic cells (5,6). In this study, we demonstrate that LRP1 function was required for optimal engulfment by both non-professional (fibroblasts) and CNS professional (microglia) phagocytes of necrotic T cells and necrotic oligodendrocytes. Removal of dead and dying oligodendrocytes and infiltrating or resident activated lymphocytes could be of critical importance in neuroinflammatory states such as MS, as a means of removal of potential auto-antigens and danger signals. In conclusion, our study demonstrates that LRP1 specifically associates with numerous proteins in myelin and is a novel phagocytic receptor for necrotic cells. To our knowledge, we are the first to describe that LRP1 can interact with a variety of cytoplasmic proteins through the extracellular domain. Our results strongly indicate that the function of LRP1 extends beyond regulation of extracellular matrix and cell surface proteins and is consistent with the proposed function of LRP1 in mediating the removal of apoptotic cells and necrotic debris during inflammatory diseases of the CNS, including MS. Further elucidation of the mechanisms by which LRP1 may regulate inflammatory diseases could pave the way for the development of therapeutic treatments, perhaps based on CCRII and CCRIV, which would be utilized to inhibit inflammation and the autoimmune response. . LRP1 is a phagocytic receptor for necrotic cells. A, annexin 5/7-amino-actinomycin D (7AAD) staining of live and necrotic Jurkat cells. B, phase contrast microscopy of live and necrotic N20.1 cells. C, CFSE-labeled LRP1 positive or negative fibroblasts were incubated with live or necrotic CypHer-labeled Jurkat cells for 2 h. After washing, fibroblast-associated CypHer fluorescence was analyzed by flow cytometry. D, results are normalized to the difference of CypHer median fluorescence between LRP1-positive cells incubated with necrotic cells and untreated LRP1-positive cells (a representative of two independent experiments is shown; , p Ͻ 0.0001). E, BV2 cells were incubated with live or necrotic CypHer-labeled N20.1 cells for 30 min. After washing, BV2 were stained with CD11b-Alexa Fluor 488 antibody and BV2-associated CypHer fluorescence was analyzed by flow cytometry. F, results are normalized to the difference of CypHer median fluorescence between GST treated cells incubated with necrotic cells and untreated BV2 cells (a representative of six independent experiments is shown). *, p Ͻ 0.001. ns, not significant.	v3-fos-license
2019-04-22T13:11:10.217Z	2018-02-28T00:00:00.000	125361590	{ "extfieldsofstudy": [ "Mathematics" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "http://www.journalijar.com/uploads/605_990_IJAR-21939.pdf", "pdf_hash": "99bd4bd0d5e1150f389773b3d627bf3057d763d4", "pdf_src": "Adhoc", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45331", "s2fieldsofstudy": [ "Economics" ], "sha1": "f7831ad054c76248a380bcbab9918f6d5433b3ff", "year": 2018 }	pes2o/s2orc	AN ASSESSMENT OF CHALLENGES FACED BY LIVELIHOOD & GROWTH ORIENTED ENTERPRISES IN AKSUM TOWN. The major findings of the study indicated that the impact of challenges faced by enterprises, order of challenges priority, its severity and similarity of livelihood and growth oriented enterprises. Livelihood and Growth oriented enterprises are part in SMSE"s. Micro, small entrepreneurs eagerly looking for new shines through these enterprises but most of the micro, small level entrepreneurs dreams are roaming in the darkness. Their desires are buried by various challenges. Some road side vendors, micro level enterprises are struggling by lack of personnel, business skills and some are continuing their enterprises with lack of capital, technology, and shortage of other resources. In this study, researcher tried to assess the challenges faced by livelihood, growth oriented enterprises in Aksum town. The study mainly focused to identify the severity of various types of challenges which are showing negative impact on enterprises. This study collected primary information through 120 respondents [60 from livelihood and 60 from growth oriented] by quota sampling technique under convenient sampling method. Secondary information gathered from Government offices in Tigray central zone, various journals and books. Data was analyzed through percentages, ranks and correlation with use of SPSS (Spatial Package for Social Sciences) 16.0. The major findings of the study indicated that the impact of challenges faced by enterprises, order of challenges priority, its severity and similarity of livelihood and growth oriented enterprises. …………………………………………………………………………………………………….... Introduction:- Charles Harvie, identifies two distinct types of micro-enterprise: (1) livelihood enterprises, and (2) growth-oriented micro-enterprises 1 . In these enterprises, entrepreneurs are struggling with different crises which are destroying their livelihood. Entrepreneurs are entering into SMSE"s with their immortal passions but finally regretting to overcome the hurdles in the high competitive world. In one side of the coin, Micro small and medium enterprises are showing radiant path to the survivalists, business entrepreneurs, small and medium investors. People those who have low skills and less capital they can easily start street business. Under growth oriented businesses, people can start opportunity enterprises named as small scale family enterprises that could be located in intermediate sector. The basic objective of these enterprises, maximize profits and accumulated wealth by taking risk. Here, basic capital, managerial skills, human resource, product specializations and technological requirements are necessary to start and continue the business 2 . 671 Without knowing the severity of problem identifying solutions are difficult to small investors, entrepreneurs as well as government. In this study, researcher studied and tried to assess the survival problems in MSMEs by the classification of two parts. One is livelihood enterprises and second is growth oriented enterprises. Before this, various researchers defined MSME based on labour and capital. In that, again they divided service and industrial sector. Researcher observed in the Aksum town, several entrepreneurs are coming and very short period closing their businesses. Not only in micro level or street side enterprises, some of middle level enterprises are also same and most of the enterprises are not upgrading to next level. Lack of capital, lack of skill, lack of business communication, lack of administration skills etc., are showing causes by entrepreneurs but no one is expressing the common challenges and severity of challenge to wind up the business. Finally the researcher recognized that the importance to assess the severity of challenges in MSME"s in the Aksum town. Objectives of the study:- The study has both general and specific objectives, General Objective: assess the challenges faced by lively hood & growth oriented enterprises in Aksum town. Limitations of the Study:- Researcher did not consider all type of MSMEs in service and manufacturing sector. This study concentrated Micro, small level of enterprises. Medium level enterprises which capital is more than 1.5 million in manufacturing and 500,000 in service sector not considered. Particularly this study collected information from which enterprises are available in the Aksum town. Respondent"s age also permitted between 20 to 60 years. Literature Review:-Definition of MSME's:-Analysis of different SMEs definitions worldwide reveal that it is very difficult to arrive at a common definition. In fact one study by Auciello (1975) in 75 countries found more than 75 definitions was used in the target countries. This demonstrates very well that there is no common accepted definition of SMEs. Depending on the country and industry, business size, assets and products, the definitions will continue to vary. Even though there is no common acceptance, some definitions about MSMEs mentioned below.. (NCR, South Africa) Survivalist enterprise:- The income generated is less than the minimum income standard or the poverty line. This category is considered pre-entrepreneurial, and includes hawkers, vendors and subsistence farmers. In practice, survivalist enterprises are often categorised as part of the micro-enterprise sector. Micro-enterprise:- The turnover is less than the value added tax (VAT) registration limit. These enterprises usually lack formality in terms of registration. They include, for example, spaza shops, minibus taxis and household industries. They employ no more than 5 people. Very small enterprise:-These are enterprises employing fewer than 10 paid employees, except for the mining, electricity, manufacturing and construction sectors, in which the figure is 20 employees. These enterprises operate in the formal market and have access to technology. Small enterprise:- The upper limit is 50 employees. Small enterprises are generally more Established than very small enterprises and exhibit more complex business practices. 672 Medium enterprise:-The maximum number of employees is 100 or 200 for the mining, electricity, manufacturing and construction sectors. These enterprises are often characterised by the decentralisation of power to an additional management layer 7 . Livelihood enterprises, which do not generate much employment and are unlikely to grow, but whose development and growth as a whole can generate more employment as well as alleviate poverty. These can start with micro level capital and more comfortable to road side businesses named street Vendors. (Charles harvie,2004)Street vendors are businesses made up of a single person or a household based with family members helping to sell the products (Butale, 2001;Harvie, 2003, Ohiokpahei, 2003 8 ,. In addition, (Bharan, 2004) 9 states that a street vendor is a person who offers goods and services for sale to the public without having a permanent built up structure but with temporary static structure or mobile stall. Challenges & failures of MSMEs:- According to the 2007 Informal Sector Survey (2009), many street vendors enterprises in Botswana are faced with the following challenges; non-payment of goods and services supplied on credit to their customers; high competition, lack of managerial skills which were the most constraints among operators, lack of space for business operations and lack of credit facilities 10 . In the United States, over 50% of small businesses fail in the first five years. Globally, the top ten causes of failure in small businesses are: lack of skills and experience in operating businesses, insufficient capital, poor physical location, poor inventory management, over-investment in fixed assets, poor credit arrangements, personal use of business funds, unexpected growth, still competition, and low sales ( Ericson and Pakes predicted that small firms die more often than their large counterparts in the same industry 18 . About 80% of Small and medium enterprises are stifled because of poor financing and other associated problems 19 Banks may avoid providing financing to certain types of SMCFs, in particular, start ups and very young firms that typically lack sufficient collateral, or firms whose activities offer the possibilities of high returns but at a substantial risk of loss. These include policies regulations, inadequate financial infrastructure, firm regulations, trade regulations, tax regulations, changing government policies, tax rates, corruption, labour regulations, cost of capital, and keen competition for limited opportunities (Uriyo 2004) 20 . Most of the studies mentioned challenges and failures of MSMEs but these all identified failure reasons without categorization of MSMEs. Based on MSMEs definition, there are minimum 3 categories like micro, small and medium size enterprises. If we observe clearly, micro level indicates very small business size like street vendors, small enterprises are representing half of micro level remaining half to the growth level like general stores, welding shops etc, and finally medium level represents wholesale, manufacturing sectors etc., these level of enterprises capital, manpower, technology are different compare with remaining enterprises. Solving for these variances, the study particularly classified the enterprises in to two categories viz., Livelihood and growth oriented enterprises which are invested less than 50,000 birr for livelihood and more than 50,000 birr for growth oriented. Data Collection Methods:- Researcher collected information through primary and secondary data. Primary data refers to the information obtained firsthand by the researcher on the variables of interest for the specific purpose of the research. Here, researcher used questionnaire with formal, informal questions and by the support of language translator, conducted field trips for collecting the information from individual Entrepreneurs and enterprise managers those who running livelihood and growth oriented businesses. Target Population:- The population of this study was covered Livelihood and growth oriented enterprises in the Aksum town. These are subparts of SMSEs. Purposively researcher separated enterprises by two categories viz., which are less than 50,000 birr and greater than 50,000 birr. In this study, Researcher collected information from livelihood enterprises those capital less than 50,000 birr like Cafeteria, small restaurant, beauty salons, internet café, vegetable, fruit sellers and small general stores etc. and also considered growth oriented enterprises those capita is more than 50,000 birr as well as representing to the manufacturing sector like Metal works, wood work including furniture, jewellery, whole sale and retail trade, restaurants etc. Sampling size and technique:- Researcher for his convenient purpose randomly conducted 150 interviews with Individual owners, managers or partners of enterprises. 12 different natures of enterprises he opted based on availability. Systematically, 10 to 12 questionnaires distributed to every nature of business. Out of 150 completed questionnaires, only 142 were accepted because some of the entrepreneurs not responded well. In these 142 questionnaires, 82 respondents belong to livelihood, remaining 60 from growth oriented. Then researcher selected the quota sampling technique for similar 674 comparison of the information. Finally 60 from livelihood and 60 from growth oriented accepted. Remaining 22 questionnaires in livelihood, removed 1 from every 4 questionnaires the rest of 2 questionnaires removed without any sampling technique. Data Analysis techniques:- This study used the descriptive design with descriptive techniques. The researcher arranged questionnaire by 3 parts viz., demographic, livelihood and growth oriented enterprises. Here, demographic questions are common for all respondents remaining questions separated based on capital size. The survey comprised questionnaire with mostly closed questions. The respondents answers were collected through questionnaire by the survey instrument using a 5-Likert scale where 5 = was strongly Agree, 3= Neutral and 1= strongly disagree. Open ended questions were used at the end of the closed ended questions to get additional information not captured in the closed ended questions. To assess the challenges, researchers used percentages, Ranks, arguments and opinions were computed. For simplicity, percentages of corresponding responses for each variable were discussed. Responses of respondents were electronically coded and saved for analysis using Statistical Package for Social Sciences (SPSS) 16. Findings And Discussion:- The researcher collected demographic information from the respondents in the Aksum town. In this study their education qualification also helpful to assess the current situation of enterprises. Here, out of 120 respondents 20% are uneducated in that, lively hood enterprises 14.17% [4.17+10]. Almost 25[12+13] women out of 120 studied below 5 th standard, these percentage was indicated 20.83. Lack of educational awareness, lack of free education facilities, family burdens, less financial support etc., are main causes to them and most of women migrated from near villages because of unhealthy agricultural environment. 33.33% respondents studied between 6 th to 12 th grades. In these, 25 respondents only in growth oriented enterprises. These percentages are indicating those who have money, and staying in Aksum town they gave priority to education and utilized local educational institution services. Graduation and above courses also most of the respondents completed from growth oriented sector only. Surprisingly, both type of enterprises marital status comparatively similar. Almost 40% of entrepreneurs were unmarried and divorced. In case of livelihood, unmarried % is high. Respondents were expressed their poor financial condition is main cause. Most of the divorced respondents were not taken diverse legally but together not living with their personal disparities. One thing should be appreciate to the respondents those who married, because maximum married respondents are working and co-operating each other to develop their financial and wealth conditions. Table 3 is disclosing what kind of business activities are doing by livelihood and growth oriented enterprises. Because, capital, human resource, infrastructure and technical support etc., are essential to run any type of enterprise. With the help of table 3, researcher tried to provide clarity of business selection by the livelihood & growth oriented enterprises. In the above table, 63.33% businesses are running only by livelihood enterprises. These all are running very small businesses and their capital less than 10,000 ETB in that, some of doing their sales on footpath without any permanent roof. These people started business with below 5,000ETB. Here most of the entrepreneurs are individuals they don"t have any human support. The remaining 35.67%[100%-63.33] invested amount between 20,000 to 50,000 ETB and maintaining 2 or 3 workers. Those who have little bit financial source, human support, managerial skills etc., are not selecting 1 to 5 types of businesses. They mentioned, these are less profit and social dereorganisation. Even though some of entrepreneurs are running these type businesses with more profits under growth oriented enterprises like "Kudda Juice point". 676 Serial No. 6 to 12, 80% businesses [1 to 5 only 20%] are running by growth oriented here livelihood proportion very less. These enterprises requires more than 50,000ETB. Some of invested more than 1 million also. Their human power and other resources are also very high compare with lively hood. Here, male participation is more. Restaurants, furniture sellers, and construction materials proportion is more almost 50% of enterprises doing transactions by by these only. Under livelihood enterprises also running these businesses but with less capital and lack of human resource they are not showing any impact on market. Table 4 is disclosing 10 different types of challenges which are facing by Livelihood enterprises. Respondents responded clearly based on every challenge. Researcher also arranged all challenges in proper sequence. Primarily, first 5 challenges representing their internal conditions remaining are belongs to external environment. 73.33% [40+33.33] respondents agreed that, lack of education, lack of business skill are major challenge to livelihood enterprises. Only 11% respondents are disagreed because these enterprises are secondary source to them, their family people are doing other works and they also studied above 10 th grade. As per their opinion, by the education, business skills will not increase, these skills will come from the circumstances or from their ancestors but most of the livelihood entrepreneurs are doing businesses without any family support and struggling for day today surviving. 75%[48.33+26.67] respondents agreed that lack of family support is major challenge because most of these enterprises running by ladies only. Those who took diverse, who neglected by man support, who doesn"t know any other work, who are not physically strong, orphans are attracting to these type of businesses. When coming to the lack of financial support, more than 86%[36.67+50] agreed that lack of financial support is one of the biggest hurdle to start and continue the enterprise. In that, 50% respondents strongly agreed, more over no one opposed. So, it is a clear indication that all livelihood entrepreneurs are struggling for financial sources. No financial institute is supporting them because they do not have any collateral assets for mortgage and no one ready to give guarantee behalf of them. Some micro-financial institutions are providing services but some of street vendors they don"t have permanent house address also. These entrepreneurs are taking hand loans for short term with high interest rates almost 12% to 20%. Finally, they are failing to re-pay the loan and losing their small business shelter also. While coming to the iv, v challenges most of the enterprises are not using any high technical equipment and not participating in the productive activities. Because of that, almost 50% respondents disagreed that the challenge of 677 poor technology. Some of restaurants, bread makers, juice maker"s are maintaining good quality but those who are depending on entry pro trade, fancy and clothe sellers etc., are agreed that good quality raw material is not available in the whole sale market because of that our production also quoting less standards. Lack of market information and lack of transportation are not that much related to livelihoodd enterprises because, they are purchasing huge quantity of items as well as not involving in to exports. These enterprises are concentrating only in local market with local facilities. Weekly or monthly once these are going near towns individually and purchasing small quantity of items. In the above table, respondents also similarly expressed their opinion some of agreed and some of disagreed. Another challenge is Correct location of the enterprises here also respondents gave mixed result 56% are agreed and above agreed remaining expressed disagree and neutral opinions because, these entrepreneurs sales almost constant and they can start their business any residential area or in the main market location based on demand of product. Because of that they did not particular about exact location. Another major challenge of livelihood enterprises is competition. 73.33% respondents were selected agreed and strongly agreed option and no one strongly disagreed. Here customers are limited and their requirements also in the particular frame no one purchase more items and also less proportion of new customers. These enterprises size is small but more people those who are in middle class and below middle class are involving in the same type of businesses. Example, Bus station area in Aksum, more than 20 general stores are available because, these are became common to all. Any special qualification and skill no need more over minimum income generating purpose these are comfortable. Finally last challenge but not least is that government involvement in livelihood enterprises. 72.33% respondents were under disagreed and strongly disagreed options. Because, government charging very small registration fee from them and no other conditions are implementing. Few of these are not taking any permission from the government and doing their business transactions. Table 5 provides the information about challenges facing by Growth oriented enterprises. Here also same challenges we have considered which we discussed in the previous. Bet here respondents shared different experiences with same type of challenge some time completely differ with livelihood enterprises. In the growth oriented enterprises, only 36.67% respondents accepted the first challenge. Most of the respondents expressed without education running the enterprise is very difficult but especially in growth oriented enterprises no one enter without educational knowledge. Lack of education is not a major challenge because most of the 678 respondents were completed their school level. This is quite opposite response compare with livelihood. Their most of the respondents accepted lack of education is major challenge but here respondents disagreed with their opinion. Here respondents have some special business skills, marketing skills, more over they have good awareness on their business. Some respondents mentioned lack of family support also one of the challenge. When they are going for out sourcing of employee services, most of the times failing to get proper commitment in the work. If family members are takes the responsibility of enterprises there is a scope to earn more. When we come to the lack of financial support, 90%[70+20] of respondents agreed that this is a major challenge. High collateral guarantee and high bank interest rates are showing major cause. Frequently financial institutions are not providing proper loans to them for Purchasing equipments, machinery and raw material. Business size and individual properties are not that much sufficient to get proper loans from financial institutions. Any long time scarcity appeared in demand and supply, they cannot survive from the financial hurdles. Another challenge is technical and power problem facing by growth oriented enterprises. Especially for timber & furniture, metal engineers are suffering with lack of technical equipments. Quality machines and new technology is not available in their work. Still they are depending on manual skills. Power problems also interrupting them but few only mentioned that. Poor productivity and technical problems are interlinked challenges in growth oriented enterprises. 60% [46.67+33.33] of respondents agreed poor productivity is also one of the biggest problems to the enterprises. Because of raw material scarcity, unskilled labour, limited technology, some productive oriented enterprises like bread making factories, furniture manufactures, and metal workers are producing less quality product. Especially in furniture and metal works they are not getting effective finishing like machine made. More bargaining nature of the customers, high competitions is pressuring them to reduce the selling price. Finally, they are making less quality items. 53.33% respondents for lack of Market information, 50% for lack of transportation and 63% for wrong place selection & misuse of premises agreed. In present days, mobile internet, television and radio services are available all over the world. In case of Aksum town, some owners are not using internet services and not watching Televisions as regular habit. Based on that, enterprises are not getting updated information about market rates and other information of the products. Most of the enterprises are not in right place. They invested huge amount for failed to attract the customers. Super markets, restaurants are facing these types of challenges. Another challenge is high competition environment. General stores, restaurants, furniture, metal businesses and beauty salons are facing this problem. Almost 80% of enterprisers agreed this. Finally, government and tax problems agreed by 2/3 respondents. Most of the respondents are not comfort with high rate of taxes, Government rules, and controlling, inspection also very high. Under table-6, researcher compared both enterprise challenges based on spearman"s rank correlation. This study contains almost ten common challenges in livelihood as well as growth oriented enterprises. Researcher gave serial numbers like i, ii, iii, ....and compared every challenge in both enterprises and also compared rank deviations. This comparison strongly shows the equality in challenge iii named as "lack of capital, financial sources and saving culture". In both enterprises, this challenge stood on 1st place. Either big or small, enterprises are struggling more for financial sources. As our early discussion, livelihood enterprises mostly depending on unstructured markets for collecting the amount with high interest rates and growth oriented enterprises are also struggling with collateral arrangements. Respondents, expressed same opinion on V th and Viii th challenges named as poor productivity and wrong place of selection. In the ranking table, both places came 5 th and 6.5 places. These ranks are not occupied from 1 to 4 th ranks. Even though, this study must give priority to challenges as per their ranks. Respondents expressed quite opposite opinion on challenge (i) and (x). Most of the livelihood enterprises were agreed that lack of education and business entrepreneur skills are major challenge to distress the entrepreneurs. 679 Remaining enterprises are completely ditching the livelihood entrepreneurs with new business strategies. Based on this 1 st challenge occupied 3.5 rank in livelihood entrepreneurs. Another end, growth oriented entrepreneurs gave 10 th rank to the same challenge. These respondents completely differed with livelihood enterprises. As per their opinion, most of the entrepreneurs have minimum educational background and minimum business skills. Growth oriented enterprises are facing financial, technical and other challenges but not educational backgrounds. In case of (xth) challenge [Government and tax challenges], livelihood entrepreneurs gave 10 th rank and growth oriented enterprises gave 3.5 th rank. Government rules and regulations are not showing any impact on livelihood enterprises but growth oriented enterprises are struggling with government rules, licence system and tax barricades. Remaining ranks are little bit similar in both enterprises. Finally, the Co-relation shows 0.605 between two enterprises. This result shows that, high positive relations in both enterprise challenges. Conclusion:- An assessment of challenges faced by livelihood & growth oriented enterprises is a most impertinent research to identify the practicality and real situation in Aksum town. Under the Ethiopian circumstances millions of entrepreneurs surviving by livelihood businesses. New industrial policies in Ethiopia attracting the new investors and day by day thousands of growth oriented entrepreneurs are coming. Most of the researchers found different challenges but identification of challenges not only sufficient, Finding the impact and range of challenges are also plays an important role. Researcher have been discussed various challenges in livelihood and growth oriented enterprises and found the possible reasons. Aksum town is a small town and huge investment high productivity enterprises are not available even though respondents expressed their views as similar to SMSE"s. This study succeeded to provide ranks to different types of challenges which are facing by enterprises and also study useful to estimate the severity of challenges. Based on this study, iii, ii, vii, viii, ix th challenges are disclosing the high impact of challenges. Government, Investors and entrepreneurs can easily estimate their strategies about the enterprises. Ranks and rank deviations were clearly explained the difference between livelihood and growth oriented enterprises surviving struggles.	v3-fos-license
2022-11-22T16:10:42.680Z	2022-11-20T00:00:00.000	253755372	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1080/23322039.2022.2143772", "pdf_hash": "a676454c8b59631098c77935462c4532f91df842", "pdf_src": "TaylorAndFrancis", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45332", "s2fieldsofstudy": [ "Economics" ], "sha1": "590ac7f66ef2f4d46b0614e9e2e1c738bb13cf29", "year": 2022 }	pes2o/s2orc	Impact of foreign direct investment on socio-economic development in belt and road countries Abstract This study is an attempt to empirically investigate the impact of foreign direct investment (FDI) on socio-economic development in the “Belt & Road” initiative (BRI) countries. It employed panel data for the period 2005 to 2018. This study used three socio-economic development indicators, life expectancy, GNI per capita, and human capital development, as outcome variables. This study found a long-run association between FDI and socio-economic development. The baseline regression results indicate that FDI does not contribute to socio-economic development in BRI countries and these findings are robust with Fully Modified Ordinary Least Square (FMOLS). However, when taking into account the region-wise analysis it is found that FDI has a positive and significant impact on poverty reduction in South Asia (SA) and Central and Western Asia (CWA) region and a negative impact in Central and East Europe (CEU) and Middle East North Africa (MENA) and it has no effect on poverty reduction in Sub-Sahara Africa (SSA) and East Asia Pacific (EAP) region. FDI has a positive effect on improved health facilities in SA and CWA region and it has a statistically insignificant or negative impact in the rest of the region. Finally, there is a positive and significant impact of FDI on human capital development in CEU and EAP region and negative association in MENA, however, it has no role in SA, CWA and SSA region. This finding postulates an idea that foreign direct investment is not a black and white mechanism. It is recommended that governments of BRI countries must prioritize public health facilities, social safety programs, poverty reduction initiatives, and human capital development strategies when looking for foreign direct investment in the country. Introduction Socio-economic development is a process of social and economic development in society. It is measured as life expectancy, GDP, literacy level, and level of employment in society. Sustainability is a broad subject and it applies to various sectors. It has many aspects but the social aspect of sustainability focused on welfare at the grassroots (Olawumi, 2018). It has three pillars economy, environment, and society. The origin of social sustainability belongs to resource mobilization from developed nations to developing nations. Currently, many countries are focusing on sustainable development (UNGA 2019). This planet faces many challenges to deal with these challenges United Nations (UN) proposed seventeen sustainable development goals. The foremost and first goal is the reduction in poverty (welfare) in developing nations through resource mobilization. Conventional argument broadly assumes that mankind's well-being is the linear function of economic growth. However, this supposition has been studied by many scholars (Pyatt, 1987a;Escobar, 2000). But Sen (1995) performed cross-country analysis and concluded that Sri Lanka and China have low GNP but life expectancy is higher in these countries as compared to Brazil, Gabon, and South Africa in 1992. Furthermore, they suggested that economic growth is a necessary condition but it is not a sufficient condition to boost up overall welfare of the nation. Based on the findings of Sen, (1995) another scholar (Stglitz, 2006) claimed that due to the narrow focus on the gross domestic product other important factors such as health and education can be overlooked. Economic growth is one of the major determinants of socio-economic development therefore, higher sustainable economic growth has been seen as one of the major objectives of all policy makers around the world. Although, GDP growth is considered necessary but not a sufficient condition for human development and it does not guarantee desirable human development globally. Economic growth is measured through GDP however a consequence in one sector within the multifaceted dimension of actual development. In a broader sense, it is considered a toonarrow measure of human progress. So, measuring welfare through the gross domestic product may be misleading (Krugman, 1995). Socio-economic development catches diverse aspects of social welfare that include other social and physical indicators of liberty and quality of living standards like assessable and affordable health facilities, living standards, access to modern medical information and knowledge are the needs of social economic development. A broad debate among policymakers and economists on finding the conclusive indicator for the social and economic development of a nation resulted in constructing the new indexes and measures. The United Nations Development Programs (UNDP) developed the human development index in 1991 which computes the living standard and it provides a valuable instrument for comparative assessment of countries in terms of various indicators including literacy, education, quality of life, and average life expectancy. The human development index got tremendous popularity among policy makers and academicians when it was recognized as a rating tool from 1990 to the present (Ranis et al., 2000;Streeten, 1999). There is another strongly debatable concept in the field of macroeconomics called foreign direct investment. The origin of this debate goes back to following the product cycle theory of Vermon's 1966, the internalization theory of Buckley 1985 and the theory of competitive advantage Porter, 1990, and are the best examples of the important contribution to academic debates and constitution of theoretical grounds for the significance of foreign direct investment. Based on these theories and background foreign direct investment has increased significantly for the last many decades. Even the perceived risk of foreign direct investment to the sovereignty of the host countries has not been adequate to impede the incentive for developing nations to catch the attention of multinational enterprises. So this progress has led to the fast growth of foreign direct investment inflow, particularly to developing nations. This growing importance of foreign direct investment has resulted in various empirical research exploring the effect of FDI on the economy. Owing to the growing importance and magnitude of the effect of foreign direct investment on skill and knowledge (Hansen & Rand,), the competitiveness of local enterprises (Sun & He, 2014) technology development (Kemeny, 2010) and stock value has been studied intensively (Huang & Shiu, 2009). Many research articles explored the impact of foreign direct investment on economic development and the impact of FDI on economic growth. The majority of the results showed that foreign direct investment has a progressive effect on economic growth both for developing and developed nations (Iamsiraroj & Ulubaşoğlu, 2015;Omri et al., 2014). Foreign direct investment net inflow is considered an important channel to mobilize the resources in developing nations (Gohou & Soumaré,). In 1990 private sector led by FDI contributed 75% of external capital flows to in BRI countries. Maryla & Dominique (2019) studied the economic, poverty, and environmental impact of the BRI countries. Yu, Zhao, et al. (2020) found that China's BRI enhanced China's export to forty-five BRI countries and thirty non-BRI top trading countries. Du and Zhang (2018) found that China's BRI has enhanced China's cross boarder investment. Total green factor productivity of Chinese provinces has been improved through BRI. Liu and Xin (2019) and Foo, Lean, & Salim, (2019) studied the impact of BRI on international trade in the ASEAN region. Sun et al. (2019) examined the economic effects of the Belt and Road Initiative in BRI counties countries. Ma (2022) examined growth effects on economic integration in BRI countries. Hussain, Li et al. (2022) examined the dynamic linkages between corruption tourism and carbon emission in BRI countries. Sattar, Tolassa et al., (2022) examined the environmental effects of China's ODI in South Asian BRI countries. Hussain, Li et al. (2022b) evaluated the impact of energy and the environment on economic growth in BRI countries. Bilal Khan, Huobao et al. (2019) examined the effects of FDI on poverty reduction in Pakistan and Liao et al., (2020) studied the impact of international development aid on FDI in BRI countries but the impact of FDI on socio-economic development in BRI has not been conducted in the prior studies. As per authors' understanding it is the pioneer attempt to examine the impact of FDI on socio economic development in BRI countries. As foreign direct investment can impact macroeconomic indicators including economic growth, human capital, employment, capital development and it can affect the social welfare through technological spillover, human capital and industrial structure. Therefore, the basic question of this research paper: does foreign direct investment impacts on socio economic development in BRI countries? The finding of the study reports that FDI does not contribute in social economic development in BRI countries however, region wise analysis have significant differences. It is recommended that governments of BRI countries should focus on improving health facilities, human capital development and poverty reduction strategies by implementing the social safety programs and poverty alleviation strategies and these areas should be on top priority when bring FDI in the country. The rest of the article is arranged as 2. Literature Review, 3. Methodology, Data and Variables 4. Results 5. Discussion & Conclusion. Literature Review The history of foreign direct investment (FDI) in developing nations originated after World War II, and the main motive behind this startup was political instead of economic. This motive shifted to developed nations that provided subsidies, economic incentives, and fiscal benefits to developing nations. This link has different dimensions first, the association between foreign direct investment and poverty can be described in terms of economic and social sides. The social aspect is that it helps the government to overcome poverty. The reason behind this is that it provides jobs, strengthens skills, and restores technological improvement. As per the earlier viewpoint, the economic side describes technological progress as the key factor for sustainable economic development and it has a significant effect on societal improvements (Solow 1956a). Nowadays attention has been shifted to improving human capital. As per endogenous growth theory technology and human capital play a vital role in development and these indicators are considered major contributors to sustainable economic development. Socio-economic development is considered the primary factor behind human capital development which excites our major interest to investigate how foreign direct investment impacts human capital development. Second, this impact may be indirect or direct. The private sector has a spillover effect through backward connection between indigenous suppliers, foreign direct investment, and domestic sourcing which generates linkages between FDI and domestic companies that improve the export capacity of domestic firms. Likewise, modern technology creates effective competition in the market and it has a positive significant spillover effect on the welfare and economic growth. Another channel for FDI is employment opportunities but it is only effective if the employment ratio is significantly more than foreign direct investment-related unemployment. Thus foreign direct investment has an enormous impact on the welfare of the labor-intensive segment. But these benefits depend on the type of FDIs. But policy measures are also important to increase FDI inflow. It has been seen that at the macro level FDI improves economic growth if the attitude of revenue transfer is positive. In this situation, the association will be indirect. Additionally, the host country's economic level also impacts this association. For example, this impact can be achieved through skilled labor, efficient supplies, and cultural, political, social, and resource endowment characteristics of the FDI's receiving country (Sumner, 2005). The association between human capital development and FDI is discussed in the literature in terms of economic growth and welfare. Sen, (1998) studied that welfare economics focused more on societal and economic parameters rather than a gross domestic product which shows a paradigm move in the course of evaluating the overall progress of an individual's quality of life. Sen (1992) declare that "while the economic study has often focused on goods and income to measure a person's misery, deprivation and benefit there is required to transfer concentration to things that persons have reasons to worth intrinsically." It is broadly accepted that economic growth makes a constructive contribution towards human progress if this supposition is right then it is predicted that foreign direct investment has an indirect impact on human well-being. Although, this assumption of a perfect positive association between welfare and economic growth has been criticized by Anand & Sen, (2000). A study conducted by Bruno, Ravallion, Squire et al., (1996a) claimed that economic growth provides benefits to all players of society but could not decrease income inequality automatically. The impact of income expansion on income inequality belonged to the opening state of imbalance in the nation. The countries whose income inequality level is high economic growth may be increase but the poverty level also increase and make the socioeconomic development worsen (Gohou & Soumaré,). If economy development is not pro-poor then it does not encourage wealth redistribution in the society it negatively impacts the welfare and makes inequality worse. Arcelus et al., (2005) analyzed the effect of FDI on human development using the human development index score for low and middle-income countries for the period 1975-1995 and found a positive effect. Similarly Arcelus et al., (2005) studied the impact of foreign capital inflow on three indicators of the human development index including wealth, education, and life expectancy, and concluded that the impact of foreign capital inflow on socio-economic development depends on many factors the important factor is a return to scale. The general assumption about studying the association between foreign direct investments on economic growth is that economic growth improves the welfare in developing countries and overall conclusion about this phenomenn is mixed. Many studies show that FDI improves economic growth. Most of the research indicators show that FDI increases economic growth. The difference in the results could be the reason for different econometric techniques and lack of harmonized and comprehensive data sets. Many studies explored the relationship between foreign direct investment and economic growth using different econometric methods. Results are mixed. Beugelsdijk, Smeets, Zwinkels et al., (2008a) evaluated the horizontal and vertical foreign direct investment in forty-four countries for the period from 1983-2003, and found a significant and positive impact in developing countries but no vertical or horizontal effect in these countries. A study conducted by Lee & Chang, (2009) used a panel correction and panel co-integration model using annual data from 1970 to 2002 in thirty-seven countries and found long relationship is strong and the short-run relation is weak short. However, a study used cross-sectional data from 91 countries for the period 1975 to 2005 by applying threshold level regression and concluded that the development of the financial market has a positive effect on foreign direct investment and without improvement of the domestic financial market has no significant effect on foreign direct investment (Azman-Saini et al., 2010). Similarly, another study conducted (Dutta & Roy 2011) investigated the impact of political risk on the relationship between financial development and foreign direct investment in 97 countries using panel data and concluded that a positive relationship between financial development and foreign direct investment becomes negative beyond a specific level and the political factors changed this level of financial growth. Moreover, this threshold level has a high score during the political stability in the country, and parallel existence of political stability and financial development is difficult for the significant effect of foreign direct investment on economic growth. Omri et al., (2014) investigated the causal relationship between foreign direct investment, CO 2 emission, and economic growth by applying a dynamic and simultaneous equation model using panel data from 54 countries from 1990 to 2011. The data set is distributed into sub-samples Central Asia, Latin America, North Africa, Middle East, Sub-Sahara, and Caribbean countries. The result shows that there is bidirectional causality in all regions except North Africa and Europe. Another study employing panel data from 49 countries for the period 1974 to 2008, result indicate that foreign direct investment in higher stocks leads to higher productivity growth, and usage of modern technology also increased in developing countries (Baltabaev, 2014a). Pegkas, (2015) investigated the effect of foreign direct investment on economic growth in Eurozone countries for the period 2002 to 2012 by employing DOLS and FMOLS methods results showed that there is a long-run association between foreign direct investment stocks and economic growth. Furthermore, the result also shows that foreign direct investment positively impacts the Eurozone countries. Jude & Levieuge, (2017) used smooth panel regression investigated conditional association among the institutional quality, foreign direct investment, and economic growth and the finding showed that foreign direct investment has a positive impact on economic growth beyond a specific threshold level of institutional quality it is also concluded that institutional reforms also have a better impact on economic development. Barrel & Nahhas, (2018) investigated the impact of bilateral direct foreign investment and market integration in OCED countries and results showed that distance, single market and a single market affect the bilateral foreign direct investment inflow in the region Likewise, Feils & Rahman, (2011) established the hypothesis that regional associations among the countries increased inward foreign direct investment and many members of the association attract more FDI inflows. However geographic distance and cultural, institutional efficiency, and market size have diverse effects on the outsider nations. On contrary, a study discussed the foreign direct investment-led hypothesis in twenty-eight developing nations applying the co-integration model. Results showed that there is no evidence of the short and long-run impact of foreign direct investment on economic growth (Herzer et al., 2008). A study conducted by Herzer, (2012) argues that there is a negative effect of foreign direct investment on economic growth in developing nations. Another study showed that the impact of foreign direct investment on economic growth in 05 SARRC countries is also not significant (Basnet & Pradhan, 2014a). Finally, Iwasaki & Tokunaga, (2014) concluded that Central & Eastern European countries examined the impact of foreign direct investment on economic growth and concluded that there is no solid evidence between foreign direct investment and economic growth. Very few studies used the human development index to check the impact of foreign direct investment on socio-economic development Sharma & Gani, (2007) examined the impact of foreign direct investment on human development in low and middle-income nations for the period 1975 to 1999 and used the fixed effect model. Results showed that there is a positive impact on human development. The more recent study conducted by Gohou & Soumaré, () investigated this association between FDI and welfare in the African region. Results showed that there is a positive and significant effect on poverty alleviation but significant differences in the African region. Moreover, foreign direct investment has a good effect on welfare in poor countries as compared to rich countries. used the proportion of the poverty line and reexamined this relationship and found that there is a significant effect of foreign direct investment on poverty reduction in the African region. A study conducted by Mirza & Giroud, (2004) said that international corporations are the major source of investment in Vietnam. Yet Vietnam needs subsidies in the value chain to gain an important competitive advantage over other countries in the region. Another study using panel data showed the significant and positive impact of FDI on poverty (Hung 2005). A study was conducted in Pakistan to check the impact of foreign direct investment on poverty reduction and the result suggested that foreign direct investment reduces poverty in Pakistan (Shamim, Azeem, et al., 2014a). Similarly Ahmad et al., (2018) concluded a bidirectional casual association between FDI and economic expansion in ASEAN nations. However Tambunan, (2005) suggested that foreign direct investment has a significant positive impact on poverty but there is needed more effort for technological spillover and tax collection from multinational organizations. conducted a study in Taiwan and concluded that continuous economic improvement is the key force to reducing poverty in the short and long run. Perera & Lee, (2013) conducted research in 09 Asian countries to check the impact of institution quality, and foreign direct investment on income inequality and poverty, and the finding showed that economic growth reduces poverty. From the above literature it is found that no empirical study has been conducted on socio-economic development in "Belt & Road" countries. To bridge this gap in the literature, this study is the pioneer attempt which is going to empirically investigate the impact of FDI on socio economic development in BRI countries. Methodology, Data, and Variables The purpose of this is to empirically investigate the impact of foreign direct investment on socioeconomic development in BRI countries. The study sample consisted of the duration 2005-2018. Foreign direct investment net inflow (as a percentage of GDP) is used as the independent variable. Three dependent variables are used to estimate socio-economic development i.e., gross national income per capita is used as a proxy variable for poverty reduction. Life expectancy at birth is used to capture the improved health facilities in BRI countries and school enrollment primary (% of gross) is used to capture human capital development four sets of control variables were used in this study including economic and policy variables, infrastructure development indicators, institutional quality, and risk factors and all these variables are used in various studies followed by Table 1. The following equation is used to estimate the impact of FDI on socio-economic development. According to United Nations Development Program (UNDP) human deployment index report, three indicators have been used as proxy variables including gross national income per capita, life expectancy at birth, and school enrollment (% of primary) as dependent variables to capture socioeconomic development (SED) effects where FDI (foreign direct investment net inflow) is the independent variables and X is sets of control variables which includes economy and policy, institutional quality and political risks factors in each country and regions, β 0 is the intercept and β 1 and β 2 are respective coefficients and ε it is the error term. This study employed the cross-sectional test (CSD) to determine whether the variables are crosssectional dependent or independent followed by Pesaran, (2004). The CSD issue occurs due to rapid growth of financial integration, economic and globalization, unobserved common shocks, and because of augmentation in the measurement of trade liberalization. Therefore, it is necessary to use the CSD test for the panels. Mathematically, it is presented as CSD = ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi Where pij presents the correlation between each pair of residual series got from regression estimates. Before estimating the co-integration test, it is essential to check the stationary of the model. It is also necessary to investigate the presence of unit roots for its long-run relationship of independent and dependent variables. For the co-integration estimation, the order of the integration might be similar for all the variables. This study used (Levin et al., 2002) unit root test to control the issue for non-stationary data because the regression result will be spurious or biased if variables are nonstationary. After examining the stationary in the variables the further step is to investigate the long-run relationships among the variables. This study adopted Pedroni, (2004). Mathematically it is presented as Freedom House Control The severe problems of endogeneity and autocorrelations can create nuisance issues and bias the findings of the estimated coefficients from the panel data least square regression, therefore this work employed the fully modified ordinary least square (FMOLS). This technique is adopted because all variables are stationary at their first difference form. So, the cointegration test can be used for testing the existence of a long run relationships among the variables. In case of the existence of cointegration among the variables, then the FMOLS estimator can be used to estimate the long-run equilibrium coefficient. In this study, cointegration exists between variables, therefore, FMOLS is used. Mathematically it is presented as Results The purpose of this study is to empirically investigate the impact of foreign direct investment on socio-economic development in BRI countries. Descriptive statistics is significant to define vital characteristics of the data. Table 1 and 2 presents descriptive statistics of the data. The mean value of dependent variables GNI per capita, School enrollment, and life expectancy is referred to as 18,706, 12.35 and 5.284 minimum values describe as 1059, 31, and 54.35, and maximum values are reported as 132,440, 150, and 83.15 respectively and the standard deviation is reported as 21,569, 12.35 and 5.283 respectively and main independent variables foreign direct investment net inflow reports their respective mean, standard deviation, minimum and maximum values 4.795, 6651, −41.51 and 54.65 respectively. However, other economic and policy variables and institutional quality variables also report their respective values. Before performing the formal regression analysis, this study checked certain assumptions. The kernel Distribution Plot is used to check the normality of the data. Figure 1 indicates that data is normally distributed. For the multicollinearity issues in the data, this study used variance inflation factor (VIF) and the mean value of the VIF score is 2.78 which falls below 5 and suggests that there is no collinearity issue in the data. Breuch-Pagan test is used to check the heteroskedasticity and the null hypothesis of Breuch-Pagan test is H0: constant variance for all cases and the results of Table 3 report that p-values of three dependent variables that correspond to the Chi-Square test statistic are less than 5% level of significance (0.0037, 0.0010 and 0.000 < 0.05) and conclude that heteroscedasticity exists in the data. There are several ways to fix this issue, including the transformation of the response variable, the use of weighted regression, and the use of robust standard errors. Robust standard errors are more "robust" to the problem of heteroscedasticity and tend to provide a more accurate measure of the true standard error of a regression coefficient therefore this study used robust standard errors to handle the heteroscedasticity issue in the data. In panel data analysis the cross-sectional dependence test is used as a fundamental setup before determining the unit root. The cross-sectional dependence test is used by many researchers in their empirical studies (such as Rauf et al., 2018;Dong et al., 2019), consequently, this study adopted Pesaran, (2004). The null hypothesis is H0; that there is no cross-sectional dependence. The finding reported in Table 4 portrays that there exists cross-sectional dependence. Sometimes macroeconomic measures have a unit root problem. In that situation, the finding of the regression estimate is typically misleading or spurious. Therefore, to investigate the stationary of the variables this study used the Levin-Lin Chu Unit Root test, and the results of the unit root test (Levin et al., 2002) Table 5 indicate that all variables are stationary at the level and their 1 st difference form. Before examining the long-run association among the indicators entire variables should be co-integrated. This study used Pedroni, (2004) and the results of Table 6 indicate that test statistics of Modified Phillips-Perron t, Phillips-Perron t, and Augmented Dickey-Fuller t are statistically significant which confirms that long-run association exists among the variables. Panel Data Ordinary Least Square is used to estimate the baseline regression results. Table 7, column-1-2, indicates that the impact of foreign direct investment (FDI) on gross national income per capita in BRI countries is negative or insignificant when using the total sample and when controlling the year fixed in effects which suggests that FDI plays no role in poverty reduction in BRI countries. Similarly, column 3-4 also produces the same results suggesting that foreign direct investment has no role in improving human capital development. Finally, columns 5-6 indicate that the impact of foreign direct investment in improving the health facilities in BRI countries is also lacking. However, government spending in BRI countries in enhancing the gross national income per capita is positive and significant which suggests that government efforts and their public expenditure can reduce poverty in BRI countries but public spending on attaining human capital is not attracting. Finally, government spending on improving health facilities contributes positively. The effect of trade openness and landlocked countries on social and economic development is not encouraging. The infrastructure indicators like mobile phone users and access to electricity contribute positively to enhancing the gross national income per capita and healthier working population and the level of institutional quality of BRI countries contributes positively to social economic development. Finally the political risk factors i.e. political rights and civil liberties in BRI countries have no role in social economic development except the political rights contribute positively to gross national income per capita. To investigate the impact of foreign direct investment on gross national income per capita which is used as a proxy variable for poverty reduction. This study divides the total sample into six subsamples. Table 8 columns 1-2 shows that there is a positive and significant impact of FDI on gross national income per capita in South Asia (SA) and Central and Western Asia (CWA) region. However, it has a statistically significant negative impact in Central and East Europe (CEU) and Middle East North Africa (MENA) region and it has no effect on gross national income per capita in Sub Sahara Africa (SSA) and East Asia Pacific (EAP) region which suggest that FDI contributes in reducing poverty in SA and CWA region. When taking into account the economic and policy variables Govt. spending contributes significantly to poverty reduction in SA, CEU, MENA, and SSA and negative or insignificant role in CWA and EAP regions. Trade openness does not have an Robust t-statistics in parentheses * p < 0.01, p < 0.05, * p < 0.1 Table 8. Results Impact of FDI on Gross National Income Per Capita VARI. (1) SA encouraging role in reducing poverty in all regions. Infrastructure development indicators like mobile phone users show a positive impact on GNI per capita in SA, CWA, CEU, and MENA regions, and access to electricity also has a positive impact on GNI in all regions except CWA. Institutional quality indicator like CPI also has a positive effect on enhancing GNI per capita which as a result reduce poverty in all regions except SSA. The risk factors indictors like political rights increase gross national income per capita in MENA and EAP and civil liberties only contribute to increasing income level in SA, and CWA and SSA region. Table 9 columns 1-2 shows that there is a positive and significant impact of FDI on life expectancy in South Asia (SA) and the CWA region and it has a statistically insignificant or negative impact in the rest of the region. When taking into account the economic and policy variables Govt. The results of Table 10 show that there is a positive and significant impact of FDI on human capital development in CEU and EAP region and a negative association in MENA however, it has no role in SA, CWA, and SSA. The economic and policy variables indicate that Govt. spending contributes significantly to human capital development in CWA and rest of the regions it has no role. The infrastructure development indicators like mobile phone users contribute positively to human capital development in CWA, CEU, MENA regions only and access to electricity shows a positive impact on human capital development in SA, MENA regions, and the rest of the regions it has no impact. The institutional quality indicator like CPI do not contribute to human capital development significantly. The risk factors indicators like political rights contribute significantly in EAP, SSA, MENA, and CWA regions however, civil liberties do not contribute to human capital development in any region. Robustness Checking To check the robustness of the baseline results this study used Fully Modified Ordinary Least Square (FMOLS) and the results of Table 11, indicate that impact of foreign direct investment (FDI) on gross national income per capita in BRI countries is negative or insignificant when using total sample and when controlling the year fixed in effects which suggest that FDI playing no role on poverty reduction in BRI countries. Similarly column 3-4 also produce the same results suggesting that foreign direct investment has no role in improving human capital development. Finally, column-5-6 indicates that the impact of foreign direct investment in improving the health facilities in BRI countries is also lacking and these findings are in line with Table 7. However, Government spending in BRI countries remains useful in reducing poverty in increasing the gross national income per capita but public spending on improving human capital and improving health facilities is not encouraging. Trade openness and landlocked countries do not have a significant effect on social economic development indicators. Finally, Government spending on improving health facilities contributes positively. The infrastructure development indicators like mobile phone users and access to electricity contribute positively to increasing the gross national income per capita and health facilities for a healthier working population and the level of institutional quality of BRI countries contributes positively to social economic development. Finally the political risk factorsi .e. political rights and civil liberties in BRI countries have no role in social economic development except the political rights contribute positively to gross national income per capita. The results of FMOLS Table 12 show that there is a positive and significant impact of FDI on gross national income per capita in South Asia (SA) and Central and Western Asia (CWA) region and a negative or insignificant effect on the rest of the variables. And these findings are robust with the baseline regression results. The economic and policy indicator like Govt. spending Robust t-statistics in parentheses * p < 0.01, p < 0.05, * p < 0.1 contributes significantly to poverty reduction in SA, CEU, MENA, and SSA and negative or insignificant role in CWA and EAP regions and these findings support the baseline results. Trade openness does not have an encouraging role in reducing poverty in all regions. Infrastructure development indicators like mobile phone users show a positive impact on GNI per capita in SA, CWA, CEU, and MENA regions, and access to electricity also has a positive impact on GNI in all regions except CWA. The institutional quality indicator like CPI also has a positive effect on enhancing GNI per capita which as a result reduce poverty in all regions except SSA. The risk factors indictors like political rights increase gross national income per capita in MENA and EAP and civil liberties only contribute to increasing income level in SA, and CWA and SSA region an these findings are also robust and in line with Table-9. Similarly, FMOLS Table 13 indicates that FDI has a significant impact on life expectancy in South Asia (SA) and CWA region and it has a statistically insignificant or negative impact in the rest of the region which imply that improved health facilities produce healthier working population as a result income level increased and rest of the regions do not get a strong effect. (1) SA Finally, human capital is considered a critical factor in the socioeconomic development of a country. The FMOLS results of Table 14 show that there is a positive and significant impact of FDI on human capital development in CEU and EAP region and a negative association in MENA however, it has no role in SA, CWA and SSA and these results are robust with the baseline results. The economic and policy indicators portray that Govt. spending has the potential to contribute significantly to human capital development in CWA and in the rest of the regions it has no role. The infrastructure development indicators like mobile phone users contribute positively to human capital development in CWA, CEU, MENA regions only and access to electricity shows a positive impact on human capital development in SA, MENA regions, and the rest of the regions it has no or negative impact on human capital development. The institutional quality variable like CPI no or negatively contribute to human capital development EAP. The risk factors indicators like political rights contribute significantly in EAP, SSA, MENA, and CWA regions and have a negative or insignificant effect on SA and CEU region however, civil liberties negatively impact human capital development in all regions except SSA. Discussion and Conclusion This study evaluates the impact of foreign direct investment on socioeconomic development in Belt and Road Countries. We analyze it in two ways first we check the impact of foreign direct investment (FDI) on socioeconomic development using a total sample of BRI countries. Gross national income per capita is used as a proxy variable for poverty reduction, education has used as a proxy to examine the effect of FDI on human capital development in BRI countries and life expectancy is also used as a proxy variable to determine the effect of FDI on improved health facilities and these variables are used as dependent variables. Furthermore, four sets of control variables like economic and policy variables, infrastructure development indicators, institutional quality, and risk factors including political risk and civil liberties are used as independent variables. Second, we do a heterogeneous analysis by dividing the total sample into six regions South Asia (SA), Central and West Asia (CWA), Central and East Europe (CEU), Middle East North Africa (MENA), Sub Sahara Africa (SSA) and East Asia Pacific (EAP). Panel Data has been used for the period (2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018). A kernel distribution plot is used to check the normality of the data and it is found that the data is normal. Variance inflation factor (VIF) is used to detect multicollinearity issues in the data and it found no collinearity issue. Breuch-Pagan test is used to detect the heteroscedasticity issue in the data and results report that a heteroscedasticity issue exists in the data. Therefore, the robust standard error is used to fix the heteroscedasticity issue . Pesaran, (2004) is used to detect crosssectional dependence and it found cross-sectional dependence. (Levin et al., 2002) Levin-Lin Chu Unit Root test is used to check the stationary in the variables and results found that all variables are stationary at a level and its first difference form. Pedroni, (2004) cointegration test is used to find a long-run relationship among the variables and the results show that there exists a long-run association among the variables. A panel data regression model is used for baseline results and a Fully Modified Ordinary Least Square (FMOLS) is used for robustness checking. Baseline regression results indicate that impact of foreign direct investment (FDI) on gross national income per capita in BRI countries is negative or insignificant when using the total sample and when controlling the year fixed in effects which suggests that FDI plays no role in poverty reduction in BRI countries. But empirical literature is not conclusive on the fact that FDI contributes to poverty reduction and some studies prove that FDI does not contribute to poverty reduction. The Current finding contrast and match with Bilal Khan, Huobao et al. (2019) who found that FDI contributes positively to reducing poverty in Pakistan. Akinmulegun, 2012) found the insignificant effect of FDI in reducing poverty, similarly Ali et al., (2010) Huang et al., (2010) found an inverse association between FDI and poverty alleviation. Ganic, (2019) examined the nexus between poverty reduction and FDI in 12 European countries and the findings conclude that Balkan region there is a positive effect between FDI and poverty reduction and Central Europe Robust t-statistics in parentheses * p < 0.01, p < 0.05, * p < 0.1 region there is a negative and insignificant effect of FDI on poverty alleviation. Similarly, some past studies provide mixed effects (Soumare, 2015;Ucal, 2014;Shamine et al., 2014;Gohou &Soumare, 2012, Fowowe and and found a positive effect of FDI on poverty alleviation. Whereas, Ali et al., (2010), Huang et al., (2010) found a negative effect of FDI on poverty reduction. Similarly Fayyaz Ahmad, Muhammad et al., (2019) examined the effect of FDI on welfare in SAARC and ASEAN countries, and this study significant positive effect on poverty reduction in Asia but it has significant differences in various regions in Asia. Furthermore, this study found that foreign direct investment has also no role on human capital development in BRI countries. The reason is that mainly FDI comes for development work like the construction of Roads, rail networks, energy and bridge construction, and transportation systems. The impact of foreign direct investment on life expectancy or improving the health facilities in BRI countries is also lacking because many BRI countries are developing countries but due to lack of funds education and health sectors are not top agenda however, sustainable development goals mainly focused on poverty reduction, access to education improve living standard and adequate health facilities to every citizen of a nation. The negative effect of foreign direct investment on life expectancy shows that public health facilities are worsening in OBOR countries. these findings are also supported by Herzer & Nagel, (2015). Government spending in BRI countries increases gross national income per capita which suggests that Government efforts and their public expenditure can reduce poverty in BRI countries but public spending on attaining human capital is not attracting. Finally, Government spending on improving health facilities contributes positively. The effect of trade openness and landlocked countries on social and economic development is not encouraging. The infrastructure indicators like mobile phone users and access to electricity contribute positively to enhancing the gross national income per capita and healthier working population and the level of institutional quality of BRI countries contributes positively to social economic development. Finally the political risk factors i.e. political rights and civil liberties in BRI countries have no role in social economic development except the political rights contribute positively to gross national income per capita. When taking heterogeneous analysis the results indicate that the impact of FDI on reducing poverty in South Asia (SA) and Central and West Asia (CWA) is positive however, in Central and East Europe (CEU) and Middle East North Africa (MEAN) it has a negative effect but it has no effect in Sub Sahara Africa (SSA) and East Asia Pacific it has no impact on reducing poverty. There is a positive and significant impact of FDI on life expectancy in South Asia (SA) and Central and West Asia (CWA) region and it has a statistically insignificant or negative impact in the rest of the region. Finally, it has been seen a positive and significant impact of FDI on human capital development in Central and East Europe (CEU) and (EAP) region and a negative association in (MENA) and it has no role in (SA, CWA, and SSA) regions. It is concluded that these findings are heterogeneous. This study has several policy implications for the Governments of BRI countries. Socio-economic development in BRI countries can be improved by establishing multi-level socio-economic development strategies like poverty reduction strategies to uplift the people from poverty by providing basic health and welfare facilities like access to drinking water, improvement in health quality, improvement in compulsory school education enrollment, development of industries and business in rural areas, women empowerment through information technology, and improvement in technical education, labor skills, expansion in urban markets and improvement in the agriculture sector. The development of infrastructural facilities including Roads, Rails, Internet connection, mobile phone utilization, and access to social media networks in BRI can also assist to improve socio-economic development in BRI countries. Many BRI countries are developing countries and they are facing energy shortages and load-shedding issues like Pakistan due to the non-availability of electricity many industries, local markets, shopping malls, and shops are not working at their full capacity, therefore, BRI countries should focus on energy poverty. Finally, these countries have low wage rates therefore it is suggested to increase the minimum wages as per international standards. People's welfare can be improved by introducing comprehensive sustainable development poverty reduction strategies and social safety programs. Like the Pakistani Government introduced an Ehsass Program in 2019 which is a complete social safety and poverty reduction strategy that provides, conditional and non-conditional direct cash transfers, provides targeted subsidies and it increases nutritional and health coverage. It has multi-dimensional initiatives. The Ehsaas program increases efficiency and transparency, and it has a great social impact (Michael & Muqeet, 2022). Charitable organizations and markets can play a supplementary role by providing volunteer services and resources to reduce poverty in the country. Finally, there should be political rights, civil liberties access, and freedom of speech that provide opportunities for individuals to put their basic welfare needs and demands before the Governments of their respective countries for a better living standard. It is recommended that policymakers must prioritize public health facilities, social safety programs, and poverty reduction strategies and educate the people by providing technical education, training, and skills when they looking for foreign direct investment in the country. The limitation of this study is that it used panel data for the period 2005 to 2018 and used fully Modified Ordinary Least Square (FMOLS). Further study can be conducted to conduct to analyze the policy effects of the "Belt & Road" initiative on socio-economic development in BRI countries using the Difference-in-Difference (DID) Model or Propensity Score Matching Difference-in-Difference (PSMDID).	v3-fos-license
2021-01-07T09:04:55.352Z	2020-12-31T00:00:00.000	242484554	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GREEN", "oa_url": "https://www.researchsquare.com/article/rs-136524/latest.pdf", "pdf_hash": "5a1d9ea5d2a95e50c91b9f631e41e88f09363c45", "pdf_src": "ScienceParsePlus", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45334", "s2fieldsofstudy": [ "Biology", "Medicine" ], "sha1": "a616d1f653518fab9ca98a0fb860c5ed8ac16dab", "year": 2020 }	pes2o/s2orc	Ezrin regulated myoblast differentiation/fusion and muscle ber specialization through PKA-NFAT-MyoD/MEF2csignalling pathway Backgorund:Neuromuscular diseases are a kind of nervous system diseases that have a high disability rate.Ezrin’ role in skeletal muscle has not been identied. This study aims to conrm the effect and mechanism of Ezrin on myoblast differentiation and fusion, myotube size, and myober type. Method:By using immunoassaying and western blot analyses, Ezrin, MyHC,MEF2c, MyoG, PKAα/β/γ, PKA reg Iα, PKA reg IIβand NFATc1-c4 were detected in myoblast cells treated with Ad-Ezrin or Ad-shEzrin. Real-time PCR were used to evaluate MyoD, Myf5, MyHC-I , MyHC-IIa/b and MyHC-IIx in myoblast cells. PKA inhibitor H-89 or PKAreg I activator N 6 -Bz-cAMP were added into medium to conrm their relationship between Ezrin and PKA during myoblast differentiation/fusion. In vitro, Ad-NFATc1/c2 or Ad-shNFATc3/c4 were respectively transfected into C2C12 cells, myoblast differentiation/fusion, myotube size and myober type were assessed by using immunostaining of MyHC, MEF2c and MyoG. In vivo, transfection of Ad-Ezrin into gastrocnemius and soleus muscles for 7 days, the numbers of MyHC-1 postivemyobers were analyzed after immunostaining of MyHC-1. Results: Ezrin expression were time-dependently increased during myoblast differentiation/fusion. Knockdown of Ezrin by shRNA delayed myoblast differentiation and fusion in a time dose-dependent pattern, as shown by immunostaining of MyHC. Conversely, over-expression of Ezrin by adenovirus time-and dosage-dependently promoted myoblastdifferentiation/fusion, and muscle ber specialization characterized by increased MyHC I and MyHCIIa/b. Forced expression of Ezrin did not alter PKA, and PKAreg II α levels, but altered the levels of PKAreg I α/β, Myf5 and MyoD, and leading to the accumulation of MyoG+/MEF2c+ nuclei. By contrast, Ezrin knockdown signicantly decreased the PKA reg I/II ratio and MyoG+/MEF2c+ nuclei. The PKA inhibitor H-89 remarkably abolished the benecial effect of overexpressingEzrin on the numbers of MyHC+ myotubes and MyoG+/MEF2c nuclei. These opposite changes mediated by knocking down Ezrin were almost eliminated by PKAreg I activator N 6 -Bz-cAMP. Furthermore, over-expression of NFATc2 or knockdown of NFATc4reversed the inhibitory effect of Ezrin knockdown on myoblast differentiation/fusion, resulting in the recovery of the numbers ofMyoG+/MEF2c+ nucleiin3-nuclei + myotubes. Meanwhile, overexpression of Ezrin specically induced type I muscle ber specialization, which was associated with increased levels of NFATc1/c2. Furthermore, in vivo transfection ofAd-Ezrin into gastrocnemius and soleus muscles increased the numbers of MyHC-1 postivemyobers. By contrast, knockdown of NFATc4resulted in the recovery to normal levels of MyHC-2b in Ezrin-knockdown myoblast cells, attributingtoregainingMyoDand MEF2c expression. Conclusions: Ezrin trigger myoblast differentiation and fusion, myotube size, and alters muscle ber specialization through PKA-NFAT-MyoD/MEF2C signalling pathway. Introduction Neuromuscular diseases are a kind of nervous system diseases that have a high disability rate, causing long-term mental and economic burden to the society, family, and the individual 1 . Muscle biopsy pathology is the most effective for the diagnosis of myopathy. Studies found that the pathological basis of muscular atrophy and muscle weakness is the degeneration and necrosis of muscle bers. Skeletal muscles have the ability to regenerate to prevent the loss of muscle mass and maintain normal shape and function 2 . It is, therefore, crucial to study the characteristics of muscle ber regeneration and their associatedfactors for the prevention and cure of neuromuscular diseases 3 . Muscle satellite cells conduce to the need for physiological self-renewal and the repair of pathological injury 4 . In ammatory reactions are one of the pathologies that result in myopathy,includingpolymyositis and dermatomyositis 5 . A recent study has shown that a more robust in ammatory area within the skeletal muscle demonstrated features of high Ezrin expression, a member of the Ezrin/radixin/moesin (ERM) proteins family. Published data supplied evidence that Ezrin could play a crucial role in transferring extracellular signal molecules into the skeletal proteins, activating the corresponding signal pathways, regulating the cell morphology, adhesion, phagocytosis, movement, and angiogenesis [6][7][8][9][10] . However, the role and mechanism of Ezrin in muscle satellite cells are still unclear. Traditionally, the activated protein kinase A(PKA) has been linked to the unique phenomenon of myoblast differentiation/fusion and myotube formation, ascribing to the alteration in PKA regulatory subunit I (PKA RI) under normal differentiation condition 11 . Our previous study has shown that it abated the ratio of PKA RI/RII in myoblast cells, resulting in the postponement of myoblast differentiation and fusion 12 . Further evidence shows that ERM proteins act as PKA-anchoring proteins and sequester PKA close to its target proteins for their effective phosphorylation and functional regulation 13 . NFATs (Nuclear factor of activated T cells) activation mediated by PKA plays crucial role in myoblast differentiation and fusion, myotube size, and altered muscle ber specialization 4,13 . In this study, knockdown of Ezrin by shRNA reduced the numbers of MyoG/MEF2C-positive cells and myotube number/size while decreasing the ratio of PKA RI/RII, causing the increased expression of NFATc2/c3/c4, suggesting that Ezrin triggered myoblast differentiation and fusion, myotube size, and altered muscle ber specialization through PKA-NFAT-MEF2C signalling pathway. Method C2C12 myoblast culture and differentiation induction C2C12 myoblast cells were inoculated in 75-cm 2 culture dishesand cultured with proliferation medium (PM) containing high glucose DMEM (HG-DMEM) supplemented with 10% FBS at 37℃and 5% CO 2 . When the con uence of the cells reached 75%, the PM was replaced with differentiation medium (MD) containing HG-DMEM supplemented with 2% horse serum (HS) to induce C2C12 myoblast cell differentiation. Traits of myotube formation from myoblast differentiation were observed daily under a microscope 14 . Overexpression or Knockdown of Ezrin in vitro Construction of Ezrin overexpression and short hairpin RNA (shRNA) adenoviral vector were prepared as previously described 15 . To con rm the role of Ezrin on myoblast cells,Ad-Null, Ad-shEzrin, or Ad-Ezrin (1 × 10 9 pfu) were added into the corresponding culture dishesa day before adding the differentiation medium. These cells were then replaced with the differentiation medium for further observation. Overexpression or Knockdown of NFATs in vitro Construction of NFATc1/c2 overexpression adenoviral vector were prepared as previously described 15 . The gene accession number of overexpressing-NFATc1/c2 is NM_172390 and NM_173091, respectively. Construction of NFATc3/c4 short hairpin RNA (shRNA) adenoviral vector were prepared as previously described 15 . These overexpression adenoviral vectors containing Ad-NFATc1, Ad-NFATc2, Ad-shNFATc3 and Ad-shNFATc4were obtained from Vigenebio.To con rm the role of NFATc3 or NFAtc4 on myoblast cells, the addition of Ad-shCtrl, Ad-shNFATc3, or Ad-shNFATc4 (1 × 10 9 pfu) into correspondingculture dish one day before the ISO was performed. And then these cells were replaced with differentiation medium for further observation. Immuno uorescence Staining C2C12 myoblast differentiotion was determined by immuno uorescence staining. Primary monoclonal and polyclonal antibodies against MEF2C (#5030s, 1:200, CST), MyHC (sc-20641, 1:150, Santa Cruze) were added into each well in every group and incubated for 12 h at 4 °C. The cells were washed with PBS 3 times for 15 min and incubated with appropriative uorescent dye-labeled secondary antibodies (Jackson Lab, 1:500, USA) at 25℃ for 2 h. The nuclei were stained with DAPI (Molecular Probes). The images for each group were photographed under Nikon 80i uorescence microscope 16 . Myoblast Differentiation After myoblast cells were treated under DM for the indicated time, the differentiated myoblast cells were stained for MyoG or MEF2C using the primary polyclonal antibody MyoG (sc-12732, 1:150, Santa Cruze) or MEF2C (5030S, 1:400, CST) and appropriative TRITC-labeled secondary antibody (Jackson Lab, 1:500, USA). The nuclei were stained with DAPI. C2C12 myoblast cells with only 1-2 nucleuseswithin a cellular structure were evaluated withMyoG or MEF2C staining.TheMyoG + or MEF2C + cells were de ned as the differentiated cells that did not fuse to form myotubes. Myoblast cells with 3 or morenucleusesin the structure of a cell were de ned as myotubes. The number of double-positive nuclei under high power eld (HPF, 50 µm) were analyzed afterdouble staining of MyoG/DAPI or MEF2C/DAPI. Two individuals who did not know the results evaluated the images using Image J (Java) software (National Institutes of Health, USA). Myoblast Fusion And Myotube Morphology The differentiated myoblast cells were stained for MyHC with the primary polyclonal antibody MyHC (rabbit anti-mice antibody, sc-20641, 1:150, Santa Cruze) and appropriative TRITC or FITC-labeled secondary antibody (Jackson Lab, 1:500, USA). C2C12 myoblast cells with only 1-2 nucleuses within a cellular structure were evaluated by MyHC staining, indicating that the MyHC + cells were de ned as the differentiated cells without mutually fusing to myotube. Myoblast cells with 3 or more nucleuses in the structure of a cell were de ned as myotube. The nuclei were stained with DAPI. To analyze myotube size, we divided the cells into 2 groups, including short myotube with 3 ~ 5 myoblast fusion and long myotube with more than 5 myoblast fusions. Morphology was assessed by myotube length, area (grouped less than 200 µm and more than 200 µm), and the number of myotubes (grouped 3 5 nuclei or more than 5 nuclei myoblast fusion) under high-power magni cation 15,17 . Two individuals who did not know the results evaluated the images using Image J (Java) software (National Institutes of Health, USA). Quantitative RT-PCR Total RNA from C2C12 myoblast cells was obtained using TRIzol (Invitrogen, Life Technologies) and transcribed into cDNA using the SuperScript II cDNA kit (Invitrogen, Life Technologies). Quantitative PCR was carried out using SYBR green PCR master mix (Thermo Fisher Scienti c, Applied Biosystems, CN) in Real-Time PCR System (RotorGene 6000, Qiagen, Germany). The transcript levels of the gene of interest in each group were normalized to GAPDH levels 18 . The primers used are listed in Table 1. Table 1 The sequences of primers of qPCR. Gene Forward Reverse qPCRs were performed to identi ed satellite cell differentiation and muscle bers traits by using the speci c primers of satellite cell differentiation markers including MyoD and MyoG, type I muscle ber makers like MyHC1, and type II muscle ber makers such as MyHC2a, MyHC2b, and MyHC2X. Statistical analysis Data of quantitative and semi-quantitative analysis presented are mean ± SD.Paired or unpaired Student's t-test determined statistical signi cance between the two groups.One-way ANOVA was used to compare the results for more than two experimental groups to specify the differences between groups.P < 0.05 is considered meaningful. Results Ezrin is expressed in C2C12 cells during myoblast differention and fusion To con rm the role of Ezrinin C2C12 myoblast cells, we determined the changes ofEzrin expression during myoblast differentiation and fusion by western blot. We found that the expression ofEzrin gradually increased, reaching its peak on day 4 of myoblast differentiation (Fig. 1).This expression is accompanied byC2C12 myoblast cell alterations, that time-dependently differentiated into mature muscle cells,formingmyotubes characterized by MyHC positive staining under differentiation medium containing 2% HS-DMEM. These results indicated that Ezrin could play an essential role in myoblast differentiation and fusion. MYOG/MEF2C Involved Ezrin-mediated Myoblast Differentiation And Fusion Since MyoG and MEF2C play a crucial role in the initiation and later of myoblast cells differentiation 20 , we used MyoG and MEF2C staining to con rm the relationship between Ezrin and myoblast differentiation. Our results showed that following treatment of Ad-Ezrin,the number and percentage of MyoG + nuclei were higher than that of MEF2C + nuclei within the myoblast during C2C12 myoblast differentiation (Fig. 3A-2G). Knockdown of Ezrin by shRNA markedly reduced the percentage of MyoG + and MEF2C + nuclei cells in C2C12 myoblast cells (Fig. 3A-2G).Moreover, the percentage of MEF2C + nuclei was lower than that of MyoG + nuclei in Ezrin-knockdownC2C12 myoblast cells (Fig. 3C-2G).These results indicated thattheknockdown of Ezrincouldinhibit the initiation and late phase of myoblast differentiation through MyoG and MEF2C, especially MyoG. Ezrin regulated myoblast differentiation and fusion through PKA signalling pathway Previous results have shown that PKA and PKAreg I/II ratio play crucial roles in controlling myoblast differentiation and fusion 11,12 . To con rm if Ezrin's role in myoblast differentiation and fusion was involved in PKA signalling pathway, we used western blot to detect PKA signaling (Fig. 4A-4G). Our results revealed that, the overexpression of Ezrin did not alter PKAreg II levels, but it signi cantly increased the levels of PKAα,PKAreg Iα,and PKAreg Iβ, resulting in an increased PKAreg I/II ratio. By contrast, knockdown of Ezrin by shRNA signi cantly reduced PKAreg Iαand PKAreg Iβlevels, but it did not alterPKAreg II levels, resulting in a decreased PKAreg I/II ratio (Fig. 4A-4G). PKA Involved Ezrin-mediated Myoblast Differentiation And Fusion PKA activity is found to have effects on myoblast differentiation and fusion 11,12 . Combining with the above results that the overexpression or knockdown of Ezrinaffected myoblast differentiation and fusion by altering PKA activity, we treated C2C12 cells with PKA inhibitor (H-89). We found that PKA inhibitor (H-89)abolished the bene cial role of Ezrin-mediated C2C12 myoblast differentiation and fusion. By contrast, both PKA activator and cAMP analogues reversed the inhibitory effects of Ezrin knockdown onC2C12 myoblast differentiation and fusion (Fig. 5A-5D). These results indicated that the effect of Ezrinduring myoblast differentiation and fusion could be associated with PKA signalling. Ezrin regulated myoblast differentiation and fusion through PKA-MyoG/MEF2C signalling pathway Indeed, myotube is formed by the fusion of differentiated myoblasts, which is characterized by three (3 + ) or morenuclei in the structure of a cell 15,17 . As shown in Fig. 6A-6J, we found that knockdown of Ezrin by shRNA markedly decreased the percentage of MyoG + and MEF2C + nuclei in less than 3 cells and 3 + myotubes, and these effects could be abolished by PKA activator. By contrast, overexpression of Ezrin substantially increased the number of MyoG + or MEF2C + nucleus in less than 3 and 3 + myotubes. However, the speci c changes were almost entirely cancelled by the PKA inhibitor (Fig. 6C-6J). These results indicated that Ezrin participated in C2C12 myoblast differentiation and fusion with coordination of MyoG and MEF2C. Ezrin regulated myoblast differentiation and fusion through NFATc2/c4-MyoG/MEF2C signaling pathway Since MyoG and MEF2Cinvolved in the ontrol of the initiation and later stage of myoblast differentiation, respectively 26 . As shown in Fig. 8A and sFigure3A-3F, we found that knockdown of Ezrin by shRNA markedly decreased the numbers and percentage of MyoG + and MEF2C + nuclei in less than 3-nuclei cells and 3-nuclei + myotubes, and these effects could be abolished by Ad-NFATc2 or Ad-shNFATc4. In addition, Ad-NFATc1 or Ad-shNFATc3 reversed the numbers of MEF2C + nuclei in 3-nuclei + myotubes. These results indicated that Ezrin participated in C2C12 myoblast differentiation and fusion with coordination of MyoG and MEF2C, which were associated with NFATc2/c4, at least in part. Discussion In this study, we made three novel observations. Firstly,we found that theEzrin expression has a timedynamic characteristic during myoblast differentiation and fusion.Secondly, Ezrin signi cantly controlled myoblast differentiation and fusion. And lastly, Ezrin regulated C2C12 myoblast differentiation, fusion, and myotube type formationviaPKA RI-NFAT-MyoD/MEF2C signalling pathway. Ezrin belongs to the ERM family of proteins that play structural and regulatory roles in the assembly and stabilization of plasma membrane interactions through their ability to interact with transmembrane proteins and the cytoskeleton 13 . As one of ERM proteins, Ezrinactivated the intracellular signal pathways through transferring extracellular signal molecules into actin cytoskeleton, affecting several key cellular processes, including membrane dynamics, cell adhesion, cell survival, motility, and determination of cell shape [6][7][8][9][10] . Indeed, these cellular processes were associated with myoblast differentiation and fusion 21 . Forthe rst time, we found that Ezringradually increased during myoblast differentiation/fusion, and rapidly decreased when formed myotubes approached and/or reached maturity. Furthermore, overexpression of Ezrin signi cantly accelerated the process of myoblast differentiation and myotube formation. Conversely, this process was delayed by the knockdown ofEzrin. These results suggest that Ezrin could act as a crucial factor during skeletal muscle regeneration and restoration in physiological and pathological conditions. Previous studies have shown that the overexpression of PKA inhibited myogenic differentiation, contributing to HDAC4 phosphorylation and the transcriptional repression of muscle-speci c genes by the myogenic regulators Myf-5 and MyoD 22,23,24,25 . Furthermore, ERM proteins,includingEzrin, are reported toact as PKA-anchoring proteins and sequester PKA close to its target proteins for their effective phosphorylation and functional regulation 13 . However, our results did not show anyapparentchanges in expressions of Myf-5 and MyoD in overexpressingEzrin in myoblast cells, but showed increased numbers of MyoGandMEF2C-positive myoblast cells and myotubes. Actually, MyoG and MEF2C play an essential role in the initiation and later stage of myoblast differentiation, respectively 26 .Recent reports have also associated PKA RI and RII to myoblast differentiation and myotube formation 11,12 . Of interest, we found that overexpressingEzrin markedly increased PKA RI levels, leading to an increased PKA RI/RII ratio, accompanied by an acceleration of myoblast differentiation and fusion. Furthermore, the knockdown of Ezrin, resulting in a lower PKA RI/RII ratio, inhibitory effects observed during myoblast differentiation/fusion could be reversed by the PKA RI activator. More importantly, PKA RI activator almost completely recovered the numbers of MyoG or MEF2C positive myotubes. Therefore, Ezrin could promotemyoblast differentiation/fusionviaPKA RI/RII-MyoG/MEF2C signalling pathway. Existing data have shown that NFATs act as a crucial player in myoblast differentiation and fusion, especially myo ber speci cation [26][27][28][29][30][31][32] . Moreover, NFATs activities are frequently regulated by the PKAcalcineurinsignalling pathway during cell differentiation 33,34,35,36 . Indeed, NFATc2 primarily controlled myoblast recruitment and myoblast fusion 37,38,39,40 . Similarly, we found that the inhibitory effect of myoblast fusion by knockdown of Ezrin could be restored by the overexpression of NFATc2, but not NFATc1. In line with stimulatory effect of constitutively active NFATc1(caNFATc1) on MyHC-1 expresison 41 , the increased NFATc1 levels by overexpression of Ezrin induced MyHC-1 expression. However, enhanced NFATc1 levels by Ad-Ezrin did not decrease MyHC-2b as caNFATc1 did, but keeping its normal levels. Meanwhile, Ad-Ezrin reduced NFATc3/c4 levels in myoblast cells, but maintaining MyHC-2a expression, which was different from the inhibitory role of knockdown for NFATc3 and NFATc4 in the expression of MyHC-2a 30 . These different changes could be related with the increased levels of NFATc2 by Ad-Ezrin, because NFATc2 acted as important role in MyHC-2a and MyHC-2X expressions 37,40,41 . In a word, altered NFATs signaling by Ezrinaffectedmyoblast differentiation/fusion, especiallytype-1 and type-2amuscle ber specialization. In relation to muscle ber type specialization, slow muscle specialization and fast to slow myo ber-type switch need the coordination of MEF2C and MyoD 39,40,41,42,43 respectively. In this study, the overexpression ofEzrinincreased the MEF2C levels within the nucleus while decreasingthelevels of MyoD in myoblast cells, resulting in the increased MyHC-1 expressions, accompanied by normal levels of MyHC-2a, MyHC-2b and MyHC-2X. The knockdown of Ezrinreduced MEF2C + myoblast cells numbers while increasing MyoDexpressions, leading to special changes that the levels MyHC-2a and MyHC-2b were obviously increased. Previous reports have shown that NFATc2 could perform speci c role in MyHC-2a and MyHC-2X expression with synergistic effect of MyoD while NFATc1 promoted MyHC-1 expression through inhibiting MyoD signaling 37,40 . More, NFATc4 has shown the inhibitory role in myoblast fusion 38 , and appears to mostly contribute to fast muscle ber formation, especially MyHC-2b 30 . The overexpression of NFATc4 affects the differentiation process by decreasing the expression of late differentiation markers, including MEF2c, and impairs myotube formation 30,38 . We found that these similar effects mediated by the knockdown of Ezrin could be reversed by knockdown of NFATc4. Thus, Ezrincould regulate myoblast fusion and type IIb muscle ber specialization through NFATc4-MEF2C signalling pathway. Declarations Ethics approval and consent to participate Not applicable. Competing interests The authors declare that they have no competing interests. Availability of Data and material Please contact corresponding author for data requests.	v3-fos-license
2018-04-03T04:30:09.459Z	2012-03-28T00:00:00.000	17913800	{ "extfieldsofstudy": [ "Biology", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "http://downloads.hindawi.com/journals/jo/2012/806382.pdf", "pdf_hash": "558039b6832808c6d400369b94f98bc9d4ad6570", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45337", "s2fieldsofstudy": [ "Biology" ], "sha1": "79a50132bf37e350b51d2e0544468ecb47230ff2", "year": 2012 }	pes2o/s2orc	Modulation of Telomeres in Alternative Lengthening of Telomeres Type I Like Human Cells by the Expression of Werner Protein and Telomerase The alternative lengthening of telomeres (ALT) is a recombination-based mechanism of telomere maintenance activated in 5–20% of human cancers. In Saccharomyces cerevisiae, survivors that arise after inactivation of telomerase can be classified as type I or type II ALT. In type I, telomeres have a tandem array structure, with each subunit consisting of a subtelomeric Y′ element and short telomere sequence. Telomeres in type II have only long telomere repeats and require Sgs1, the S. cerevisiae RecQ family helicase. We previously described the first human ALT cell line, AG11395, that has a telomere structure similar to type I ALT yeast cells. This cell line lacks the activity of the Werner syndrome protein, a human RecQ helicase. The telomeres in this cell line consist of tandem repeats containing SV40 DNA, including the origin of replication, and telomere sequence. We investigated the role of the SV40 origin of replication and the effects of Werner protein and telomerase on telomere structure and maintenance in AG11395 cells. We report that the expression of Werner protein facilitates the transition in human cells of ALT type I like telomeres to type II like telomeres in some aspects. These findings have implications for the diagnosis and treatment of cancer. Introduction As progressive loss of telomere DNA is associated with senescence [1], maintenance of telomere function is essential for indefinite cell proliferation. Most cancer cells rely on expression of telomerase for suppression of telomere shortening [2]. However 5%-20% percent of cancers maintain telomeres by the alternative lengthening of telomere (ALT), a recombination-based mechanism [3]. Telomere maintenance mechanisms are a potential prognostic indicator [3] and promising target in cancer diagnosis and therapy [4][5][6]. Increasing evidence supports that Werner protein (WRN), a RecQ helicase and exonuclease, plays a direct role in telomere maintenance [7] and promotion of tumor cell growth [8]. WRN epigenetic silencing in human cancers leads to hypersensitivity to treatment with a number of chemotherapeutic drugs [9]. Germline mutations in the WRN gene cause an autosomal recessive disorder, Werner syndrome (WS). WS is characterized by symptoms suggestive of premature aging and by the development of mesenchymal neoplasms [10]. Strikingly, the ALT mechanism is more prevalent in tumors arising from tissues of mesenchymal origin, such as osteosarcomas, than in those of epithelial origin [11]. It has been suggested that the telomere-telomere recombination in WRN-deficient, telomere dysfunctional cells promotes escape from senescence and engagement of the ALT pathway [12]. Werner protein also colocalizes with telomeres in human ALT cells [13]. S. cerevisiae cells that lack functional telomerase undergo telomere attrition and lose viability [14]. Rare cells escape senescence and two types of survivors arise. Type I ALT survivors have telomeres that have a tandem array structure. The repeat unit in the array consists of a subtelomeric Y element containing an ARS (yeast origin of replication) associated with short telomeric TG 1−3 repeats. This repeat unit is amplified as a tandem array structure at chromosome termini. Type II survivors have little or no amplification of Y elements, but instead have long, heterogeneous, TG 1−3 repeats extending up to several kilobase pairs (kbp) [15,16]. The generation of type I cells depends on expression of proteins involved in recombination, including RAD52 and RAD51. Type II cells depend on expression of Sgs1, the S. cerevisiae RecQ family helicase, in addition to recombination proteins RAD52 and RAD50 [13,17]. WRN can complement Sgs1 deficiency in type II ALT cells of S. cerevisiae [17]. Sgs1 deletion also facilitates the generation of survivors that grow independent of Rad52. Although tlc1rad52 yeast does not form survivors of telomere dysfunction, tlc1rad52sgs1 triple mutants readily generated survivors [18]. Nearly all of the human ALT cell lines analyzed to date have characteristics similar to that of S. cerevisiae type II ALT. Most human ALT cells have long and heterogeneous telomeres, ranging from 2 to 20 kb within an individual cell and have ALT-associated promyelocytic leukemia bodies (APBs) [19]. APBs contain the constitutive components of promyelocytic leukemia bodies, telomere DNA, and the proteins involved in DNA replication and recombination including RAD51, RAD52, RAD50, and WRN [3]. One immortalized human cell line has a "tandem array" telomere structure similar to that of type I ALT in yeast [20,21]. AG11395 is an SV40 T-antigen transformed, immortalized fibroblast cell line derived from an individual diagnosed with Werner syndrome [22]. It does not contain APBs and lacks the Werner syndrome protein. The chromosome termini of AG11395 consist of a repeat unit containing 2.5 kb of SV40 DNA and a variable amount of TTAGGG telomere sequence repeats. The SV40 DNA integrated into the telomere in this cell line contains the regulatory regions, which include the origin of replication and the early and late promoter sequences [21]. This cell line offers a unique system to investigate the role of the WRN protein and tandem array telomeres in human ALT telomere maintenance. Here we determine whether telomere maintenance in AG11395 involves a functioning SV40 origin of replication and we describe the effect on type I like structures of expression of WRN protein. Cell Lines and Culture. AG11395, WI38 75.1, and GM847 were obtained from the Coriell Cell Repository (Camden, NJ). HeLa and COS cells were obtained from the American Type Culture Collection (Manassas, VA). Cells were grown in 5% CO 2 under conditions recommended by the supplier. Quantitative Polymerase Chain Reaction (qPCR). To determine whether in AG11395 the SV40 origin of replication integrated at telomeres is functioning, we cotransfected pRLSV40 (Promega. Madison, WI) that has SV40 origin of replication and pRLCMV (Promega. Madison, WI) that does not contain an SV40 origin of replication in a 1 : 1 ratio in AG11395. The ratio of pRLSV40 to pRLCMV was assessed by qPCR, with primers specific to each plasmid using Express SYBR Green qPCR reagents (Invitrogen) in a Mx3000P thermal cycler (Stratagene, La Jolla, CA). -GGT GAG CCA GGC CAT CAC-3 Amplification conditions for all qPCR were 10 minutes at 95 • C, followed by 30 cycles of 30 seconds at 95 • C followed by 60 seconds at 60 • C. Western Blots. Ten micrograms of protein extract were resolved on an 8% SDS-PAGE gel and transferred to HyBond ECL membrane (New England Nuclear, Boston, MA). Affinity purified polyclonal rabbit anti-Werner and monoclonal mouse anti-actin antibodies (Abcam, Cambridge, MA) were used at a dilution of 1 : 1000. Western blotting was performed as described [23]. Telomerase Enzymatic Assays. Telomerase enzymatic activity was detected using the TRAPeze telomerase detection kit (Chemicon, Temecula, CA). Quantification of Telomere Length by Quantitative Fluorescent In Situ Hybridization (Q-FISH). Telomere lengths were quantified in arbitrary fluorescence units (AFUs) using a peptide nucleic acid (PNA) probe directly labeled with a Cy3 (Applied Biosystems, Foster City, CA) as previously described [25]. Cells were harvested by centrifugation at 200 × g. The pellet was resuspended in hypotonic KCl (0.067 M) for 5 min at 37 • C and then fixed and washed in methanol/acetic acid (3 : 1). Q-FISH was performed as described by Poon and Lansdorp [25]. Briefly, fixed metaphase nuclei were dropped onto slides, hybridized with PNA probe, and images of telomeres and chromosomes were captured using digital fluorescence microscopy. The images were analyzed with the TFL-Telo program to detect the telomere-specific fluorescence signal. To compensate the inter-specimen hybridization fluctuation, plasmids containing 1.6 kb, and 800 bp telomere sequence were used as control. To check any variation in lamp or alignment of the optics, we measured the integrated fluorescence intensity (IFI) of 0.1 μm beads for normalization between different experiments [25]. All experiments within a set in which quantitative length comparisons were made were dropped, hybridized, and analyzed in parallel at one time to minimize effects of probe integrity, hybridization efficiency or drift in the performance of the optical system. Chromosome Orientation-FISH (CO-FISH). Telomere sister chromatid exchange (T-SCE) was performed using chromosome-orientation-specific fluorescence in situ hybridization (CO-FISH). CO-FISH procedure was performed as described in [26]. Briefly, during S-phase, Bromodeoxyuridine (BrdU) and Bromodeoxycytidine (BrdC) nucleotides are incorporated into the newly synthesized DNA during one cell division. The BrdU-substituted DNA strands are degraded by Hoechst dye, UV, and exonuclease treatments. The parental G-rich and C-rich strands are visualized on the single-stranded sister chromatids by specific probes labeled with Cy3 (against the G-rich strand) or FITC (against the C-rich strand). Combined IF/PNA FISH. Colocalization of PML protein with telomeres in WRN expressing AG11395 was confirmed by IF/FISH, with IF performed as described above. After incubation with secondary antibody, the immune complexes were cross-linked with 4% paraformaldehyde, dehydrated and processed for PNA FISH, as described for Q-FISH above. AG11395 Does Not Support the Replication of the SV40 Origin of Replication. The telomeres of AG11395 contain an SV40 sequence, which includes a wild-type sequence origin of replication [21]. One possible function of the nontelomere repeat sequence integrated into the telomeres of cell line AG11395 could be to provide a functional origin of replication integrated at telomeres. The Y element found integrated into the telomeres of type I survivors of telomereinduced senescence in S. cerevisiae contains a yeast autonomously replicating sequence (ARS) [27]. The ARSs in Y elements are nonfunctional in the proximity of telomere sequences [28,29]. To assess whether any SV40 origin of replication is functional in cell line AG11395, we determined whether a plasmid containing a wild-type origin of replication would be replicated when transfected into this cell line ( Figure 1). We co-transfected pRLSV40, which contains a functional SV40 origin of replication, and pRLCMV, which does not contain an SV40 origin of replication. Replication of the SV40 origin containing plasmid is measured as an increase in the ratio of pRLSV40 to pRLCMV DNA, as measured by qPCR. In addition, we tested whether inhibitors of replication of SV40 were present in the cell lines by additionally cotransfecting pBRSV40, which expresses wild-type SV40 T-antigen. As carrier DNA in those experiments was not including pBRSV40, plasmid pEGFP was used, allowing for visual monitoring of transfection efficiency (Figure 1(a)). SV40 origin containing plasmid replication was assessed in 3 cell lines: AG11395, WI38 75.1 (an SV40-transformed, T-antigen expressing, ALT fibroblast cell line), and COS (which efficiently supports replication directed by the SV40 origin). Demonstrating that this methodology for measuring plasmid replication is functioning, we observed a substantial increase in the pRLSV40/pRLCMV ratio in COS cells, both with and without T-antigen provided in trans by transfection of plasmid pBRSV40 (Figure 1(a)). Neither AG11395 nor WI38 75.1 supported replication of the SV40 origin containing plasmid without expression of wild-type T antigen. The small decrease in the pRLSV40/pRLCMV measured in the cell lines not replicating the SV40 origin containing plasmid may reflect slight differences in the quality of the plasmid preparations used for transfection. WI38 75.1, but not AG11395, was capable of supporting the replication of the SV40 origin containing plasmid when T-antigen was provided by co-transfection of pBRSV40. These results show that SV40 origins of replications are not functional in AG11395 cells. A short pulse (30 min) of EdU (5-ethynyl-2 -deoxyuridine) in S phase cell does not show a significant difference in the incorporation of EdU pattern in AG11395 and WI38 75.1 cells (Figure 1(b)), again confirming that SV40 origins in the telomeres of AG011395 cells are not functional. [30]. WRN also colocalizes with telomeres in human ALT type II cells [13]. Given these observations, we investigated the possible role of WRN in transition from a "type 1" like ALT telomere structure into a "type II" like ALT telomere structure in human cells. We cloned the full-length WRN coding sequence in retroviral vector pLXIN. AG11395 cells were infected with empty and WRN expressing pLXIN. Three WRN expressing clones (WRN1, WRN2, and WRN3) and three empty-vector clones (EV1, EV2, EV3) were selected for analysis. WRN expression was confirmed by western blot (Figure 2(a)) and indirect immunofluorescence (IF) (Figure 2(b)). We then determined the localization of WRN in AG11395. Indirect Immunostaining of WRN and TRF1 revealed colocalization of WRN with TRF1 (data not shown). Exogenously We further confirmed WRN colocalization with telomere DNA by detecting the telomere with peptide nucleic acid fluorescent in situ hybridization (PNA FISH) combined with WRN IF (Figure 2(c)). tracts in AG11395. We quantified telomere repeats by quantitative PNA FISH in control and WRN-expressing clones (Figure 2(d)). We found that all three WRN expressing clones show increased telomere repeat signals at telomeres at population doubling (PD) 24 ( Table 2). The WRN expressing clones AG11395 WRN1, WRN2, and WRN3 had mean telomere sequence lengths of 807, 771 and 919 AFUs, as comp-ared to the mean telomere sequence lengths in empty vectorinfected AG11395 cells EV1, EV2, and EV3 that measured 577, 547, and 449 AFUs, respectively. SV40 Sequences Are Lost with Increasing Population Doubling (PD) after WRN Expression in AG11395. In AG11395, the repetitive structure at telomeres contains both telomere and SV40 sequences [21]. In yeast, the transition from type I ALT to type II ALT telomere structures involves loss of a tandem array telomere structure and appearance of long tracts of telomere repeats [17]. After observing an increase in telomere sequence with expression of WRN protein in AG11395, we determined whether SV40 sequences were decreasing. We quantified the SV40 sequence by quantitative polymerase chain reaction (qPCR) at PD 24, 54, and 70 ( Figure 3). All three WRN expressing clones showed a trend of decreasing SV40 sequence with increasing PDs when compared with the empty vector-infected clone at the same PD ( Table 2). As analyzed by FISH, SV40 sequences detectable in AG11395 colocalize with telomere sequences [20]. The overall loss of SV40 sequences as measured by qPCR therefore indicates a loss of SV40 sequences from telomeres. ALT Associated PML Bodies (APBs) Are Present in WRN Expressing AG11395 Cells. APBs are characterized by the colocalization of the PML protein with telomere repeat binding factors in ALT cells and have been used as morphologic marker for the presence of ALT telomere maintenance [19]. The frequency of cells containing APBs is increased when cultures are enriched for cells in the G2/M phase of the cell cycle [32]. AG11395 cells do not contain APBs [20]. Instead, they have nuclear aggregates containing SV40 large T antigen and many of the same components as APBs. After WRN expression in AG11395, we observed few spots of TRF1 and PML colocalization in rare cells (data not shown). Follow- (Table 2). Expression of hTERT Restores Telomerase Activity in AG11395. AG11395 cells lacked detectable telomerase activity as measured by the telomere repeat amplification protocol (TRAP) assay ( Figure 5(a)) and lack expression of human telomerase reverse transcriptase (hTERT) mRNA by quantitative reverse-transcriptase PCR (data not shown). Telomerase activity can be reconstituted in some type II ALT cells by expression of hTERT [33]. To investigate if the ALT type I like cells can reconstitute human telomerase reverse transcriptase (hTERT) activity and alter telomere maintenance we produced AG11395 expressing pBABE-puro-hTERT [34]. Telomerase activity was assayed by TRAP [35] at PD10. AG11395 cells express telomerase and become TRAP positive ( Figure 5(a)). Expression of Telomerase in AG11395 Cells Lengthens Telomeres, Decreases STLs, and Induces Loss of SV40 Sequences. Expression of exogenous hTERT in ALT type II cells results in lengthening of the shortest telomeres [33]. WRN-deficient fibroblasts have higher frequency of short telomeres and STLs [31]. Telomerase preferentially acts on short telomeres in WS fibroblasts and thus reduces number of STLs [31]. Expression of telomerase in AG11395 cells resulted in lengthening of short telomeres (Figure 5(b)). As expected, Journal of Oncology the extension of short telomeres decreased the frequency of STLs (Table 1). We also quantified SV40 sequences in hTERT expressing AG11395 cells by qPCR. Unlike WRN expression that caused a slow, gradual reduction in SV40 sequence (Table 2) with increasing PDs, we found a greater initial decrease in SV40 sequence at PD10 with no significant additional change over 80 PDs (Figure 5(c)). Expression of WRN or Telomerase Did Not Change the Rate of Telomere Sister Chromatid Exchange (T-SCE) in AG11395. To understand the effect of WRN expression on 8 Journal of Oncology telomere sister chromatid exchange we employed CO-FISH. We did not observe a significant change in the T-SCE in WRN expressing cells when compared to empty vector-transfected AG11395 cells ( Table 2). The rate of T-SCE in empty vector-transfected, WRN-expressing and telomerase-expressing AG11395 cells is 10.4%, 11.9%, and 9.4%, respectively. Discussion WS patients exhibit an increased incidence of mesenchymal cancers compared with the general population, but little is known about the genetic pathways perturbed in these tumors because few WS tumor-derived cell lines exist [10]. The increased incidence of mesenchymal cancers in WS patients is intriguing because many human sarcomas maintain telomeres by the ALT pathway [10]. When immortalized G5 mouse Terc −/− Wrn −/− clones are injected subcutaneously into severe combined immunodeficiency mice, aberrant telomere recombination coupled with the strong selective pressure to maintain telomere length in the absence of telomerase results in the activation of ALT tumors [12]. These ALT tumors appeared to be type II, but the possibility of ALT type I tumor formation by WRN-deficient cells cannot be ruled out. It has been suggested that in S. cerevisiae ALT type I is generated by the selection of cells in which successive recombination events have taken place in subtelomeric regions [36]. WRN deficiency can facilitate the sub-telomeric recombination in human cells due to the extensive telomere sequence loss during telomere replication [31] and homologous recombination [37] and possibly by increased aberrant recombination between nonidentical (homeologous) DNA sequences [30,37]. Survivors of such events may generate ALT that maintains telomeres through a mechanism analogous to the type I survival pathway observed in sgs1, telomerase-null yeast. The Werner mutant, SV40-immortalized ALT cell line AG11395 maintains telomeres in a mechanism similar to the type I survival pathway of yeast [20,21]. This cell line offers a unique system to investigate the role of WRN deficiency and tandem array telomeres in human ALT telomere maintenance. Cohen and Sinclair tested the effect of reintroduction of Sgs1 into type I yeast survivors. This resulted in telomeres with telomere sequence tracts extended by about 300 bp and with decreased numbers of Y elements, but it did not fully convert the telomeres to a type II structure [17]. Our results with reintroduction of WRN in AG11395 cells, with increased numbers of telomere sequence repeats and decreased SV40 sequences present at chromosome ends, are consistent with experiments of reintroduction of Sgs1 in yeast Typ1 ALT [17]. These data support the hypothesis that Sgs1 and WRN promote telomere-telomere recombination that results in extension of telomere sequence and reduction of non-telomere sequence, possibly, by decrease in aberrant recombination between nonidentical (homeologous), DNA sequences and telomere sequence loss. The proposed role of WRN in transforming type I ALT like telomeres to type-IIlike telomeres are summarized in Figure 6. Autonomously replicating sequences (ARSs) in Saccharomyces cerevisiae have been characterized as both origins of DNA replication and as chromatin repressors/silencers. Some (ARSs) that functions in a plasmid are silent as replication origins in their natural chromosomal context. The origin and the silencer activities of ARS are modulated by position and chromatin environment. The silencing side of ARSs is surfaced in the subtelomere regions, where they enhance and extend the repression signals emitted by the telomere. The effect of sequence elements or chromatin structure on the function of SV40 origin of replication has also been reported [38]. To avoid the sequence dependency, well-characterized plasmids containing the SV40 origin have been extensively used to study SV40 origin function in mammalian cells in the presence of large T antigen. AG11395 maintains telomeres that have a tandem array structure that contains SV40 sequences, including the SV40 origin of replication, and telomere repeats. To determine if the SV40 origins are functional in AG11395 cells we used a plasmid containing the SV40 origin of replication. We found that unlike WI38 75.1 AG11395 does not support replication of the SV40 origin of replication, even if wild-type T-antigen is provided in trans. The most extensively reported reasons for failure of SV40 origin to support replication of plasmid are cellular inhibitory factors [39] or dominant negative T antigen [40]. For example, WT1 encoded by the Wilms' tumor gene inhibits replication by binding to SV40 origin. Many cellular factors, including the cellular late origin binding protein (LOB) protein (and possibly additional origin-specific proteins), interact with T antigen to prepare the SV40 origin for replication [41]. AG11395 cells may be missing or contain mutated cellular factor(s) required for normal functioning of the SV40 origin of replication. Alternatively, T antigen functions as an oligomer [42]. Some mutant T antigens can be complemented in vivo by providing a wild-type T antigen in trans, while others exhibit a strong trans-dominant-negative phenotype [40]. Although we do not have any direct evidence, but it is possible that the T antigen expressed in AG11395 cells has a dominant negative phenotype for function at the SV40 origin of replication. This would most simply explain the failure to stimulate replication at the SV40 origin in AG11395 when wild-type T antigen was expressed. The role of nonfunctional Y elements to yeast type I survivors and the role of the SV40 origin of replication in AG11395 telomeres remain unclear. They extend the net length of telomere with interspersed nontelomeric sequences and may serve as chromatin repressors like the ARS in Y element of type I ALT telomere in yeast. A small number of AG11395 cells express APBs after WRN expression. Endogenous WRN colocalizes with PML protein in ALT cells [13]. However, there is no simple correlation between ALT activity level and the number of APBs or APB-positive cells. The proportion of APB-positive cells can be greatly increased by methionine starvation, by DNAdamaging agents, or by restoring p53 function in ALT cells [43,44]. In each of these cases, the treatments that induced APBs also caused growth arrest. An ectopically expressed GFP-WRN partially colocalized in large nucleoplasmic foci with the endogenous PML protein in UV-irradiated U-2OS cells [45]. PML shell formation starts from the site of PML-WRN contact and proceeds until the PML layer is formed on entire surface of a WRN body [46]. At present it is not clear how WRN facilitates APB formation. One possibility is that WRN is modulating an arrest phenotype. For example, Werner syndrome cells escape hydrogen-peroxide-induced cell proliferation arrest [47]. We anticipate that, given the expression of T antigen, AG11395 cells do not arrest in G1 phase of cell cycle. However, it is possible that restoring WRN facilitates the arrest of few cells accompanied by the induction of APBs in G2/M phase following spontaneous oxidative damage. The expression of telomerase in AG11395 resulted in a rapid expansion of telomere sequences and in a decrease in the frequency of STLs. Although the expression levels of WRN protein in the clones that were selected for analysis varied significantly, we did not observe any significant correlation in STL or telomere sequence gain with the level of WRN expression. We observed rapid loss of SV40 sequences by PD 10 that persisted, but did not increase, over 80 PDs. As ALT is a recombination-based mechanism, the rapid loss of SV40 sequences is consistent with increased deletions occurring during homologous recombination in the absence of WRN [37]. That no further loss of SV40 sequences was observed after 10 PDs in the presence of increased telomere repeats is consistent with the presence of telomere repeat tracts protecting the residual SV40 repeats and is in accord with the more recent observation that telomerase acts at the majority of telomeres in the cell at each cell division [48]. If telomerase only acted on the shortest telomeres in the cell, a continued decline in the presence of SV40 sequences at telomeres would be expected as the longer SV40 containing telomeres over time reach the "critical" length and are subsequently extended by telomerase. Conclusion Our results indicate that the Werner protein plays a fundamental role in determining the telomere maintenance mechanisms in transformed cells. In the human WRN mutant cell line AG11395, exogenous expression of Werner protein facilitates the transition of ALT type I like telomeres to type II like telomeres. The increased incidence of mesenchymal cancers observed in WS patients may relate to the nature of the telomere maintenance program activated in these tumors. Besides giving valuable diagnostic and prognostic information, telomere maintenance mechanisms are a potential therapeutic target for cancer treatment. The status of Werner protein expression in tumor cells may be helpful in designing therapeutic strategies.	v3-fos-license
2020-01-09T09:07:23.660Z	2020-01-04T00:00:00.000	214246497	{ "extfieldsofstudy": [ "Materials Science" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.26656/fr.2017.4(3).282", "pdf_hash": "303ef55bcee8207ace1bcd9e71a923f2deeb7778", "pdf_src": "MergedPDFExtraction", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45338", "s2fieldsofstudy": [ "Agricultural and Food Sciences" ], "sha1": "2483059706202b86770f8565f8edafef82d1b85f", "year": 2020 }	pes2o/s2orc	Characterization of modified cassava flour (mocaf)-based biscuits substituted with soybean flour at varying concentrations and particle sizes Mocaf can be used as an alternative raw material for making complementary food because it has high carbohydrate content. However, the protein content is low so that it is necessary to substitute other ingredients which have high protein content such as soybean. The objective of this study was to characterize the mocaf-based biscuits containing soybean flour at concentrations of 10%, 15%, and 20%, and particle sizes of 420, 250, and 177 μm. Pasting properties of composite flours were evaluated in terms of peak viscosity, breakdown viscosity, setback viscosity, final viscosity, and pasting temperatures, while physicochemical properties of mocaf-based biscuit and microstructures were investigated in terms of moisture, ash, protein, fat, carbohydrate, calorie contents, color, hardness, and fracturability. Higher concentrations of soybean flour were associated with increased ash, protein, and fat content, as well as hardness. Moreover, the hardness of biscuits varied significantly depending on the particle sizes of soybean flour. Finally, the highest protein contents were achieved using 20% soybean flour with a particle size of 420 μm. Introduction In general, complementary foods on the market are biscuits and instant porridge. Biscuits are small baked products made principally from flour, sugar, and fat (Manley, 1998) and have a long shelf life due to their low moisture contents. Although wheat flour is most commonly used for making biscuits, it is not produced in Indonesia, and alternative ingredients such as cassava flour are used to make biscuits as a complementary food. Modified cassava flour (mocaf) is a fermentation product of cassava; among many agricultural commodities produced in Indonesia, cassava production reached 19,053,748 tons in 2017 (BPS, 2018). The disadvantage of mocaf is its low protein content, at 1.77% (Afifah and Ratnawati, 2017), to meet the minimum protein content requirement of 6% in complementary foods (BSN, 2005). Alternative ingredients with high protein content and production in Indonesia are needed, one of them is soybean. Soybean production in Indonesia is relatively high, with an annual production of 538,253 tons in 2017 (BPS, 2018). Moreover, the protein content of soybean flour is 40.94% (Ratnawati et al., 2019). Particle size is considered one of the most important physical properties of powders because it affects flowability (Abu-Hardan and Hill, 2010). Specifically, small particles have higher cohesiveness, reflecting greater contact area and stronger intermolecular forces between particles (Landillon et al., 2007). Furthermore, dough rheology is influenced by particle sizes and their distributions (Moreira et al., 2014;Ahmed et al., 2016), and differences in particle sizes can be exploited to give different characteristics to food products, especially bakery products. Previous studies have been reported on complementary foods based on wheat flour substituted with Dumbo catfish flour and soybean protein isolates (Mervina et al., 2012); soybean flour, arrowroot starch, and sweet potato flour (Zulfa and Rustanti, 2013); arrowroot starch, soybean flour, and sweet potato flour (Aini and Wirawani, 2013). In other studies, non-wheat flour-based complementary foods have been made from maize, soybeans and moringa leaves (Odinakachukwu et al., 2014); maize, millet and moringa leaves (Arise et al., 2014); millet, sorghum, pumpkin and amaranth seed flour (Simwaka et al., 2017). The use of mocaf as raw material for making complementary food is still rarely done, so this study is required to determine the characteristics of mocaf-based biscuits where soybean flours with varying particle sizes were used as substitutes at varying concentrations. Materials Mocaf was obtained from UKM Harapan Jaya, Subang, West Java, Indonesia. Soybean (Glycine max) was purchased from a local market at Subang. Soybeans were washed and soaked in water at 60°C-70°C for 3 hrs, then dehulled and dried at 50°C for 12 hrs. Particle sizes of soybean flours were reduced using a disk mill and a sieve until particle sizes of 420, 250, and 177 µm was achieved. Other ingredients included banana (Musa acuminata), egg yolk, powdered sugar, baking powder, unsalted butter, and lecithin. Preparation of composite flour Composite flour was made by weighing mocaf and soybean flour according to the composition in Table 1. After that, composite flour was mixed using a dry mill and then stored in polypropylene (PP) plastic bags for further analysis. Preparation of mocaf-based biscuits Biscuits were made in the Pilot Plant Bakery of the Research Center for Appropriate Technology, Indonesian Institute of Sciences, Subang, West Java, Indonesia. Biscuit formulations are shown in Table 1. Unsalted butter, powdered sugar, egg yolk, and lecithin were mixed together using a high-speed mixer until they expanded. Banana puree and baking powder were then added and stirred to homogeneity using a lowspeed mixer. Subsequently, soybean flour and mocaf were added and mixed by hand until a smooth dough was produced. Biscuit dough was sheeted to a final thickness of 7 mm and baked for 10 mins in an oven at 150°C. Biscuits were then inverted and baked for 20-30 min at 100°C. After cooling, the biscuits were placed in polypropylene (PP) plastic bags and were stored at ambient temperature for further analysis. Pasting properties of composite flours Pasting properties of composite flours (mocaf and soybean flours) were analyzed using a Rapid Visco Analyzer (RVA-TecMaster, Macquarie Park, Australia). Suspensions of 3.5 g (14% wb) of flour in 25 g of distilled water were stirred at 50°C (160 rpm) for 1 mins, then heated from 50°C to 95°C for over 7.5 mins and maintained at 95°C for 5 mins. Suspensions were then cooled from 95°C to 50°C for over 7.5 mins and incubated at 50°C for 2 mins. Parameters were measured on the Visco-amylogram: peak viscosity (PV), breakdown viscosity (BV), final viscosity (FV), setback viscosity (SV) and pasting temperature (PT). Evaluation of mocaf-based biscuits Physicochemical analysis of samples was performed to determine proximate, calorie contents, color and textural properties. Proximate analysis was performed according to the Indonesian National Standard (BSN, 1992) procedures and included determinations of moisture, ash, and crude fat contents using Soxhlet extraction. Protein of biscuit was analyzed using a DuMaster protein analyzer (DuMaster D-480, Buchi, Switzerland). Total carbohydrate content was calculated by subtracting percent moisture, ash, protein, and fat contents from 100% (100-(% moisture + % ash + % protein + % fat)). Calorie content was calculated using the Atwater conversion factors for proteins (4 kcal/g), carbohydrates (4 kcal/g), and lipids (9 kcal/g), as reported by Osborne and Voogt (1978). Color of biscuit was measured using a Chromameter (NH310, China). All determinations were performed in three replicates. Color characteristics were recorded as L* values of 0-100 representing dark to light, a* values representing degrees of redness to greenness, and b* values representing degrees of yellowness to blueness. Textural properties was analyzed in terms of hardness and fracturability using a TA.XTPlus texture analyzer (Stable Micro System, Surrey, UK). A threepoint bending rig (type HDP/3PB) was used to cut samples after placement on base beams that were 4 cm apart. Compression strengths was measured using the following conditions: test mode, compression; test speed, Samples B2-B4 were made using flour with a particle size of 420 µm, samples B5-B7 were made using flour with a particle size of 250 µm, and samples B8-B10 were made using flour with a particle size of 177 µm. Microstructures of biscuit was analyzed using a Scanning Electron Microscope (SEM, Hitachi SU3500). Prior to SEM analysis, samples were placed on SEM holders and coated with gold under vacuum conditions. Sample images were taken at 2500× magnification with an accelerating voltage of 10 kV (Blaszczak et al., 2004). Statistical analysis Data in tables are presented as averages from triplicate analysis. Significant differences in multiple comparisons were identified using analysis of variance (ANOVA), followed by Duncan tests for significance at 5%. Pasting properties Pasting properties of mocaf and composite flours (mocaf-soybean flours) were determined using RVA (Table 2). In this study, mocaf (B1) had the highest peak, breakdown, and final viscosity, but the pasting temperature for mocaf was lower than that of composite flours containing soybean flour. The peak viscosity of B1 was significantly different (p<0.05) from that of composite flour. The peak viscosity of composite flour tended to decrease with increasing soybean flour concentration. It is due to the peak viscosity of soybean flour (19.17 cP) lower than mocaf (4,755 cP), so that the composite is made the peak viscosity will decrease (Afifah and Ratnawati, 2017;Ratnawati et al., 2019). The addition of soybean flour in composite flours led to increased protein and fat contents. Accordingly, the protein and fat can inhibit interactions between starch granules and limit the swelling of starch, leading to changes in viscosity (Du et al., 2013;Hamid et al., 2015). The results in this study similar with a previous study were conducted by Julianti et al. (2017), that showed the addition of soybean flour in composite flour consist of sweet potato flour and maize starch can decrease the peak viscosity of these blends. Furthermore, the addition of soybean flour with fine particles (177 µm) caused the greater peak viscosity than the addition of soybean flour with coarse particles (420 µm). This result in line with the previous study was conducted by Ahmed et al. (2015), the peak viscosity increased in very fine particles of water chestnut flour (1,172 to 1,218 BU). According to Adebowale et al. (2008), high breakdown viscosity is associated with increased susceptibility of flour to withstand heating and shear stress during cooking. The breakdown viscosity of B1 differed significantly (p<0.05) from those of B2, B3, B4, B7, and B10. The increasing level of soybean flour in composite flour can be decreased the breakdown viscosity. This relates to the fiber content of composite flour. Ratnawati et al. (2019) showed that the dietary fiber of composite flour substituted by 40% soybean flour (18.53%) higher than the dietary fiber of mocaf (9.58%). The hydrophilic group in the fiber will form hydrogen bonds with water thereby reducing the amount of water that can be absorbed by the starch granules (Julianti et al., 2017). The final viscosity of B1 was significantly different (p<0.05) from other samples, it tended to decrease with increasing soybean flour addition. Similarly, smaller particle sizes of flours were associated with decreased final viscosity. The final viscosity was decreased due to the fat contained in soybean flour which can inhibit the swelling of the starch granules (Dautant et al., 2007). The setback viscosity of B1 was not significantly different (p>0.05) with B4 and B6 samples, but significantly different (p<0.05) with other composite flours. The setback viscosity also decreased with the addition of soybean flour. In the previous study was conducted by Asante et al. (2013) There were no significant differences in pasting temperature of mocaf and composite flour (Table 2), except mocaf and B4 sample. The composite flour had higher pasting temperature than mocaf. These observations are similar to those reported by Ocheme et al. (2018), who showed that higher pasting temperatures with increasing groundnut protein concentrate (GPC) reflect higher water absorption capacity of the blends with higher GPC contents. Physicochemical properties In evaluations of physicochemical properties of biscuits (Table 3), moisture contents of samples ranged from 4.22% to 6.46%. The Indonesian National Standard (BSN, 2005) tolerates a maximum of 5% moisture in baby biscuits, and those made from the flour blends B1, B2, B7, B8, and B9 met this standard, whereas the other biscuit had higher water content. The ash, protein, fat, and total calorie contents of the biscuit samples containing soybean flour were higher than those of the control. The ash content ranged between 1.52-2.32%, and was within the Indonesian National Standard (BSN, 2005) those maximum content of ash i.e 3.5%. This standard also regulates that the minimum content of protein in complementary food is 6%. The biscuits in this study with soybean flour addition have a protein content that is in accordance with the standard. The highest protein content was found in the B4 biscuit (14.27%), and the lowest protein content was B1 biscuit (4.07%). Therefore, control biscuits made from mocaf not fulfill the Indonesian National Standards. The fat content of the present biscuits ranged from 17.54% to 24.20%, reflecting significant contributions of soybean flour to the fat content of biscuits. Soybean flour was known to have high-fat content i.e 25.01% (Ratnawati et al., 2019). In this study, the biscuits produced not fulfilling Indonesian National Standards (BSN, 2005), it is due to the fat content exceeded 18%. According to the Indonesian National Standard (BSN, 2005), calorie content of biscuits is required to contain at least 4 kcal/g or 400 kcal/100 g. In this study, the calorie content of all samples were ranged 463.02-488.54 kcal/100 g, fulfilled the minimum energy content requirements. Color parameters of food products are important because they affect consumer acceptance. The results in this study showed that the lightness values (L) of biscuits decreased with soybean flour contents (Table 3) and ranged between 45.46 and 55.08. Higher L values indicate a brighter appearance of biscuits. The soybean flour substitutions increased the protein content of the present biscuits and were negatively correlated with lightness, indicating major roles of Maillard reactions in color formation (Chevallier et al., 2000). Laguna et al. (2011) suggested that proteins are subject to Maillard reactions when baked, leading to the development of brownish colors and decreased lightness values. The present color values followed a similar trend to that reported by Mieszkowska and Marzec (2016), who showed that the addition of chickpea flour to short-dough biscuits decreases L* values from 80 to 77.9. Table 3), and these were inversely proportional to yellowness values (b) of biscuits, which increased with concentrations of soybean flour, reflecting the yellowish color of soybean flour. Mieszkowska and Marzec (2016) similarly showed that the addition of chickpea flour to short-dough biscuits increases b values from 23.9 to 28.6. Textural properties are important qualities of biscuit products as they influence consumer acceptance. In this study, the result showed that the addition of soybean flour increased the hardness of mocaf-based biscuits at all concentrations. Biscuit fracturability also tended to decrease with increasing soybean flour content. Mocaf biscuits had hardness values of 216~358% of the control (100% mocaf), indicating harder textures. Arun et al. (2015) previously identified dough components that affect the hardness of biscuits and showed interactions between protein, fat, carbohydrates, and starch contents. Similarly, Mieszkowska and Marzec (2016) indicated that the addition of 20% chickpea flour increases the hardness values of biscuits from 24.7 to 35.2 N. Conclusion The present analysis show that higher concentrations of soybean flour in flour composites are accompanied by increased ash, protein, and fat contents, and lead to increased hardness of the biscuits. Although particle sizes of soybean flour significantly affected the hardness, the highest protein contents were achieved with 20% soybean flour with a particle size of 420 µm.	v3-fos-license
2018-10-23T01:11:20.328Z	2014-04-25T00:00:00.000	53395044	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nepjol.info/index.php/JPN/article/download/10296/8370", "pdf_hash": "d51a053a6e69e68afee860717598c142315bb605", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45340", "s2fieldsofstudy": [ "Medicine" ], "sha1": "d51a053a6e69e68afee860717598c142315bb605", "year": 2014 }	pes2o/s2orc	Smear technique for intraoperative diagnosis of central nervous system neoplasms Correspondence: Dr. Srijana Shrestha, MD Consultant pathologist, Grande City Clinic and Diagnostic Services, Jamal, Kathmandu, Nepal E-mail: srijanak82@gmail.com Background: Smear cytology has become increasily popular as an alternative to frozen section for the rapid diagnosis of most of central nervous system lesions. The aim of this study was to assess the utility of smear technique for the rapid diagnosis in the neurosurgical biopsies and to compare the smear cytological features with the final histopathological examination. technique is superior in displaying abnormal cellularity, nuclear and cytoplasmic details and occasionally even tissue architecture. 1,2 of smear technique safeguards the surgeon being in the wrong location, prevent sampling purely necrotic or reactive tissue, thereby avoiding a need for second anesthesia and invasive procedure.In cases of radical excision of diffusely infiltrating gliomas where margins are not obvious macroscopically, the intraoperative smear technique can be useful to define the margins. 37 Smear technique requires a very small amount of tissue, as small as 0.1cm for its preparation.With increase use of stereotactic technique in neurosurgical field, there is limitation of size if biopsy which might be inadequate for other technique such as frozen section.6] MATERIALS AND METHODS This was a prospective interventional study conducted in BPKM Cancer Hospital for a period of one year.It included 60 consecutives cases that underwent for open craniotomy and burr hole biopsy in the neurosurgical department.The biopsy samples were transferred in isotonic saline.Smears were prepared by placing 1-2 mm of biopsy material on at one edge of a clean, dry and labeled slide and crushing with another slide with just enough pressure to spread the tissue into a thin film.The smear was immediately immersed in 95% ethanol and stained with Papaniculaou stain.Remaining biopsy was submitted for paraffin sections.Cytological features were studied in detail, radiological and operative findings were compared with cytological findings.All Smear cytological diagnoses were compared with histological findings.The tumors were classified according to the World Health Organization Classification of CNS. RESULTS Overall diagnostic accuracy of smear technique for rapid diagnosis achieved in the present study was 88%.Complete correlation was considered for the cases in which the intraoperative cytological diagnosis was same as the histological diagnosis including the grading.Total discrepancy between cytological and histological diagnosis was in five cases (Table 1).The misdiagnosis in cytology were mainly due to sampling error and the lack of histological architecture.Four cases of gliomas showed partial discrepancy.Partial Discrepancy was considered for the cases where the grading of the tumor not same in cytology and histology diagnosis.In a total of 60 cases, Gliomas (51.6%) were the most frequently occurring tumor (Table 2). DISCUSSION Inherent soft nature of CNS tissue of brain tissue and high water content renders poor quality frozen sections.A cytological examination has shown to be great value as an alternative method in intraoperative consultation of CNS pathology.Intraoperative consultation of brain lesions are requested to differentiate neoplastic from reactive lesions; to differentiate metastatic from primary lesions; to estimate the degree of malignancy and to determine the tumor margins. 7,80][11] Most of the gliomas were easily identified in the smears.Presence of Fibrillary background in smear were very useful features to diagnose gliomas (fig. 1) .Regarding grading of tumor in smear, Marshall et al have mentioned in his study that the high grade tumors are liable to undergraded in smear because necrotic tissue which is of the hallmark for high grade tumor in histology would be purposely avoided when selecting the small portion of biopsy to be smeared. 12n the present study, four cases were undergraded due to lack of all the features.As Gliomas could be heterogeneous, many studies have also concluded that it would be unwise to grade the tumors in every cases in rapid diagnosis. 1,7,10,12 the study conducted by Jennifer et al, the most common and distinctive findings in 23 cases of pilocytic astrocytoma were markedly elongated bipolar cells, Rosenthal fibers and eosinophillic granular bodies. 13All these features were evident in our case of pilocytic astrocytoma (fig.2). Similarly, cytological features of Oligodendrogliomas were very distinct in our cases as in the study of Roessler K et al. 14 Tumor cells were moderately pleomorphic, rather in monolayer sheets, well spreaded, with uniform oval nuclei and conspicuous nucleoli (fig.3).But, histological features such as perinucler halos and chicken wire like vascular channels were not evident in the smear. 10 Two cases of ependymoma were correctly diagnosed in the smear (fig.4).Poorly formed rosettes, mildly pleomorphic glial cells and calcifications were evident in the smears .In contrast to present study; Sidawy et al felt difficulty in distinguishing Ependymoma.The author emphasized the importance of definitive diagnosis of ependymoma, because a surgeon will try to go for complete resection in case of Ependymoma whereas in Diffuse Astrocytoma, surgery usually is terminated in favor of radiotherapy. 10,11sinterpretation of Glioblastoma as metastatic tumor was seen in one case due to presence of bizarre cells, the lack of fibrillary background and lack of architecture.Similar error was encountered in the study of Mouriquand et al.In our study, Out of Seven metastatic lesion, one case of the metastatic tumor was misdiagnosed as Glioblastoma due to presence of abundant gliofibrillary background in the smear. 8,9 h case of Hemagioblastoma and lymphoma and two cases of pituitary adenoma were easily identified in the smear.Discohesive round cells lying discretely in absence of fibrillary background with presence of lymph granular bodies in toluidine blue stain were the features of lymphoma.Pituitary adenoma showed cellular smear consist of round cells with salt and pepper chromatin (fig.5).Meticulous correlation with clinical and radiological features helped to achieve high accuracy in the present study. Meningiomas were the second most common tumors in our studies (fig.6).In Most of the cases, meningothelial cells were easily identifiable.Meningothelial cells were oval to round containing vesicular nuclei with conspicuous nucleoli and ill defined wispy cytoplasm.Psammoma bodies, whirling pattern, microcalcification, intracytoplasmic inclusions were readily appreciated in the smears as mentioned in the study of Kobayashi S et al.Two cases of atypical Meningioma were misinterpretated as low grade gliomas due to hyperchromatic irregular cells and prominent fibrillary in the background. 15 CONCLUSION The current study also found smear /squash technique to be accurate and reliable for the rapid diagnosis of CNS tumors.It is highly recommended technique and can be safely practiced as an alternative to frozen sections.Lastly, the goal of the pathologist while using this technique in intraoperative set up is to give sufficient preliminary information to optimize the surgery.	v3-fos-license
2024-03-02T06:17:35.061Z	2024-02-29T00:00:00.000	268083233	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": null, "oa_url": null, "pdf_hash": "003ce22cbacc19fd3a11829335c8fab08c84c724", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45341", "s2fieldsofstudy": [ "Environmental Science", "Biology" ], "sha1": "0e05ee7fdda7c8dae39d29a79e9e4bd1f1db0b4c", "year": 2024 }	pes2o/s2orc	Response of Eurasian otters (Lutra lutra) to underwater acoustic harassment device sounds Seal scarers (or acoustic harassment devices, AHDs) are designed to deter seals from fishing gear and aquaculture operations, as well as to prevent seals from entering rivers to avoid predation on valuable fish. Our study investigated the potential effects of AHDs on non-target species, specifically the Eurasian otters (Lutra lutra), by testing the reaction of two rehabilitated otters to simulated AHDs sounds at 1 and 14 kHz, with a received sound intensity of 105–145 dB re 1 µPa rms. The 1 kHz sounds were used to investigate alternative frequencies for scaring seals without scaring otters. The otters reacted to both 1 and 14 kHz tonal signals when retrieving fish from a feeding station 0.8 m below the surface. Their diving behaviour and time to extract food progressively increased as sound intensity increased for all tested sound levels. Notably, the sound levels used in our tests were significantly lower (40–80 dB) than the source levels from commercial AHDs. These findings highlight the importance of caution when using AHDs in river and sea habitats inhabited by otters, as AHDs can change their behaviour and potentially result in habitat exclusion. www.nature.com/scientificreports/Directive 27 , deliberate disturbance of otters is prohibited, and activities that may compromise their conservation status should be avoided. Whereas most AHDs operate at frequencies above 10 kHz, harbour seal behaviour has also been shown to be affected by signals centred at 1 kHz 28 .If AHDs could operate efficiently at lower frequencies, this would most likely reduce impact on non-target species, such as otters, that have a hearing sensitivity in air tapering off below 4 kHz 21 .As there are no studies on underwater hearing of Eurasian otters, it is therefore interesting to measure the response of otters not only to currently used AHDs, but also to signals of lower frequency emphasis. Here we investigated if Eurasian otters under human care responded to underwater tonal signals with a frequency of 1 and 14 kHz.The sounds were played back at various sound levels to two diving and foraging otters in an enclosure.We examined whether the otters' response to the sounds were influenced by sound level and frequency, and if there were any indications of individual differences between the two animals or emerging habituation in their reactions. Methods Experiments were carried out at AQUA Aquarium and Wildlife Park (Silkeborg, Denmark) over 11 days during May-August 2019.Two adult otters were included based on availability: a 12-year-old male and a 5-year-old female, both brought into the facility as abandoned pups.Their weight during the experiments is unknown.Both otters were accustomed to aerial noise from the park's guests and to underwater noise from the water circulation pump of their enclosure.The freshwater pool was irregularly shaped with a depth of 0.3-1.6 m, with concrete walls and a bottom covered by sand and rocks (Fig. 1).Due to its irregular shape, the pool was not notably reverberant during playback.The water temperature was 15-22 °C.The underwater background noise level was measured regularly at different locations and various depths (A-E, Fig. 1) using a SoundTrap ST202HF (Ocean Instruments, Inc.; sampling rate 48 kHz; 16 bits; preamplifier gain high; clipping level 176 dB re 1 µPa p, determined by relative calibration 29 .The spectral density of background noise and playback signals were measured in Matlab (Mathworks, Inc, version 2019b) using Welch's method 29 (FFT size 2048 points, 50% overlap and Hann windowing). Two tonal signals at different frequencies were played back: the 1 kHz signal contained several harmonics, some of which were as intense as the fundamental, whereas for the 14 kHz the harmonics had an intensity more than 40 dB below the fundamental (see supplementary material).Both signals lasted 500 ms, including 100 ms ramp-up and -down.Five stimulus levels were played back: 105, 115, 125, 135, and 145 dB re 1 μPa rms@1 m (measured in a 95% energy level window; see 30 for details).Playback signals were measured 1 m from the speaker using the same SoundTrap as described above, with a maximum received level variation of ± 6.9 dB around the otter's location at the feeder. The signals were played from a WAV file (sampling rate 48 kHz, 16 bits) in randomized sound pressure level order in Area A (Fig. 1) from an Electro-Voice UW-30W loudspeaker and a Sonar Product HS25 spherical transducer, attached to a vertical PVC pipe at a depth of 0.8 m and connected to a laptop computer with Adobe Audition (Adobe, Inc.).An underwater video camera (Divers pro fish-eye 10-021, LH-camera, Fredericia, Denmark) was mounted inside a PVC pipe just above the speaker facing the otters.The camera was connected via an Elgato video capture USB device (Elgato, Corsair Gaiming, Inc., Germany) to a laptop computer, which recorded both video and emitted acoustic signals (Fig. 1). One or two playback sessions (at 9 am and/or 1 pm) were made every experimental day, with 1-61 days between experimental days.During each experimental session, up to 7 sounds were played according to a randomized list mixing frequency and sound level.Up to 5 control trials were also randomly inserted into the trial list for each session.Control trials were identical to exposure trials but without any sound being played back.Experimental sessions were performed outside the otters' normal feeding sessions (at 11 am and 2 pm), using 200-300 g fish of their daily ration (800 g for the female and 900-1000 g for the male) for each experimental session.To minimise the risk of habituation, in each experimental session there were never more than 3 consecutive trials in a row with the same signal, and never more than 3 consecutive control trials.During the entire experimental period, the otters were exposed to each sound level 10-18 times for each frequency, with up to 10 exposures per otter for each frequency and sound level.The signals were played back at random intervals, with at least 1 min.between trials. The experimenter was located on the roof of the building approximately 5 m above Area A (Fig. 1) to minimise visual distraction of the otters.Before each session, a feeder with two netted tubes containing 2-4 pieces of common roach (Rutilus rutilus) were lowered from the roof to a depth of 0.8 m, at a distance of 1 m from the speaker.The feeder was first introduced without any sound present for 2 days.Experimental sessions began once the otters associated the feeder with food.The experimenter manually controlled the transmission of the signals, the video recording, and lowering and filling of the feeder, while observing the otters on a laptop via the underwater camera. The experiments were made under a permit from the Danish Animal Experiments Inspectorate (Ministry of Food, Agriculture and Fisheries of Denmark, permit nr.2023-15-0201-01608).The experiment was within the frame of the permission to keep the rehabilitated animals under human care.The study is reported in accordance with ARRIVE guidelines (www.arriv eguid elines.org). Recording animal behaviour All trials started at a randomized time (range 16 s-5:28 min) after the feeding station was in place in front of the speaker and underwater camera, to prevent the otters from predicting the timing of the signal.Playback was first initiated when the otter's head faced the feeder in an attempt to take a piece of fish.Data were collected separately for each individual otter present at the feeder during each trial via the video recordings.Behavioural responses were graded by two independent observers watching the recorded videos and using Response Scores (RS) from 0 to 3 (defined in Table 1, with examples given in Fig. 2).A behavioural response was defined by the otter orienting its head and/or body away from the feeder from playback start until 1 s after playback (Fig. 2).The scores of the two observers were not significantly different (Kruskal-Wallis test, df = 1, n = 191, p = 0.6), so the average observer score was used in the analysis.A response was defined as any score above a non-response, which was defined as no change in head and/or body position and denoted RS0.Control trials followed the same analysis protocol as stimulus trials. Dive duration was measured from the moment the otter's head submerged beneath the water until it resurfaced.If the otter went out of the field of view before resurfacing was observed, the dive duration for that instance was not recorded.The number of dives were counted for each dive performed, from the first submersion until a fish was retrieved and the animal returned to the surface.It included dives where no fish was retrieved.The duration of fish extraction from the feeder was measured from the moment the otter first interacted with the mouth of the feeder until the second a fish was successfully extracted, including time they spent at the surface after unsuccessful attempts and did not include the time spent consuming the fish.Otters never swallow fish under water 31 but bring them to the surface to consume them.After eating a piece of fish, the otter returned to the feeder, and a new recording of dive duration, number of dives and time to get the fish began.The session stopped when the feeder was emptied.All data points were included in the analysis. The study aimed to investigate the behavioural response, dive duration, number of dives, and the time to get fish out of the feeder, and how each of these response variables were affected by sound level, frequency, animal ID, number of trials, and exposure.The following statistical models were used to analyse the data in R (v. 3.5.2,2018): for the behavioural response (binary: response/no response), a logistic regression model was fitted using the generalised linear model, glm function with logit link.To analyse dive duration (time in seconds), number of dives and the time to extract fish, normality was assessed using the Shapiro-Wilk test from the stats package 32 .As the data were not normally distributed, a gamma regression model was fitted using the glm function with frequency, sound level, animal ID, trial number, and the interaction between animal ID and the before mentioned predictors.Dive duration of otters and the time spent on extracting fish were analysed using a generalized linear model with gamma distribution and log link function.The number of dives was fitted Poisson and negative binomial regression models, and a likelihood ratio test was performed to compare the two models, which showed that the negative binomial model was the best fit (Chisq = 8.5, df = 7, p = 0.29).We used stepwise regression for all model selections using the step function from the stats package 32 .This involves removing one predictor variable at a time until all remaining predictor variables were significant at a level of 0.05.After the final model was selected, we used diagnostic plots to check the assumptions of the model, including normality of residuals and homoscedasticity.All graphs were made using ggplot2 package in R 33 . Table 1.Definitions of response scores RS0-RS3.RS1-RS3 were used for RESPONSE, and RS0 for NO RESPONSE when defining the binary variables generalized linear model. Result We performed 18 playback sessions over 11 days, resulting in a total of 57 playback and 49 control trials.Each session lasted from 23 min to 1 h 24 min.The otters used on average 2.1 min.to collect and eat each fish before returning to the feeder, ready for a new trial.In total, there were 10-18 trials for each frequency and intensity combination, with either one or both otters present at the feeder.During the trials, the underwater ambient noise levels at the site of the feeder varied with less than 5 dB in the frequency range of sound stimulus (1 and 14 kHz, respectively; see supplementary material).When the circulation pump was turned off during sessions and on between sessions, the noise level increased by about 40 dB in the enclosure in the 1-14 kHz frequency range when the pump was turned on (see supplementary material). The behavioural responses of both animals were investigated using the binomial logistic regression model (glm, binomial, n = 190; Table 2).The intensity of acoustic stimuli had a significant impact on the behavioural response exhibited by the otters (p < 0.001).The odds ratio for sound level can be calculated as exp(0.028)= 1.03, which indicates that for a one-unit increase in sound level (dB), the odds of observing a behavioural response increase by a factor of 1.03, holding all other predictor variables constant.The difference in individual response was not statistically significant (p = 0.06).The binomial regression model had the lowest AIC value = 178.2, and the model's goodness of fit was measured using the null and residual deviance values (264.8 and 172.2, respectively).They indicate how well the model fits the data.A lower deviance value indicates a better fit.In this case, the residual deviance is lower than the null deviance, which suggests that the model explains more of the variation in the data than would be expected by chance.Therefore, we can conclude that the model is statistically significant and provides a good fit to the data.The dispersion parameter for the binomial distribution was estimated to be 1.This parameter measures the degree of overdispersion or underdispersion in the data.A value of 1 indicates that the data are neither overdispersed nor underdispersed but are homogeneously distributed.In other words, the behavioural responses of the observed individual otters are consistent with each other. According to the generalized linear model with a gamma distribution and log link function, the duration of dives was found to be influenced by sound level and individual (glm, gamma log, n = 462, p < 0.05; Table 2, Fig. 3).The analysis revealed that an increase in sound level was associated with a decrease in dive duration, with each unit increase resulting in a reduction of approximately 1 s (p = 0.01).Additionally, the analysis showed www.nature.com/scientificreports/ a notable effect on individual on dive duration, with the log of the response time expected to increase by 1.6 s for the male otter compared to the female (p < 0.001).The negative binomial regression model had the lowest AIC value = 2930.9,and both null and residual deviance values (185.1 and 175.1, respectively) confirmed the statistical significance of the model (p < 0.05).The dispersion parameter for the negative binomial distribution was estimated to be 0.3, which indicates that the variance of the duration of dives is proportional to the mean duration of dives.The investigation into the number of dives revealed significant influences from sound and the number of trials, with differences between the two animals (glm, negative binomial, n = 462, p < 0.005; Table 2).Specifically, an increase in sound level was associated with a decrease in the number of dives, with each unit increase in sound level resulting in a reduction of approximately one dive (p < 0.001; Table 2; Fig. 4).The number of trials positively correlated with an increase in the number of dives by one (p < 0.001).Notably, the male otter exhibited a higher dive frequency in general, recording 1.6 more dives than the female (p < 0.001), but reduced his diving frequency by 1.0 more compared to the female with increasing sound level (p < 0.001). The negative binomial regression model had the lowest AIC value = 2255.6,and both null and residual deviance values (658.1 and 448.7, respectively) confirmed the statistical significance of the model (p < 0.005).The dispersion parameter for the negative binomial distribution was estimated to be 6.3, underscoring the heterogeneity in dive behaviours between the two otters. The time the otters spent extracting a fish from the feeder was significantly affected by sound level, individual and number of trials (glm, gamma log, n = 165, Fig. 5).The analysis showed that an increase in sound level was associated with an increase in the time to get fish, with each sound level unit increase resulting in an increase of approximately one second (p < 0.001).There was a notable effect of individual on the time to get fish, with the log of the response time expected to increase by 2.5 times for the male otter compared to the female (p = 0.003).The number of trials significantly reduced the time for the male by one second (p = 0.02) while it increased one second for the female, though it was not statistically significant (p = 0.08). The gamma log-linear regression model with the lowest AIC value = 1480.2,had a null deviance of 195.7 and the residual deviance of 140.2 provided a statistically significant model (p < 0.05).The model further revealed a dispersion parameter of 1.2, highlighting the variability in the response variable.Overall, these findings underscore the nuanced interplay between sound level, individual characteristics, and trial dynamics in shaping the temporal aspects of otter behaviour when exposed to sound during fish extraction from a feeding station. Discussion This study investigates the underwater auditory capabilities and behavioural response of Eurasian otters to frequencies produced by AHDs.Both the male and female otter exhibited the ability to perceive 1 and 14 kHz sounds underwater, responding distinctly to stimuli of varying intensities.The behavioural responses showed a progressive increase as sound intensity increased from 105 to 145 dB re 1 μPa, with minimal responses during control trials, indicating otters' sensitivity to presented underwater sounds (Fig. 3).Our findings suggest that increased sound levels led to a reduction in both the duration and the number of dives by approximately one second and one dive, respectively.Furthermore, the time taken by the otters to extract fish from the underwater feeder increased by one second per elevated sound level, indicating a disruptive effect on otters' foraging behaviour, making them spend more time at the surface. Individual differences played a significant role, with the male otter exhibiting 1.3 s longer dive duration and a higher baseline dive frequency, recording 1.6 more dives than the female.Interestingly, the male otter reduced his diving frequency by one dive more compared to the female with increasing sound levels, suggesting a nuanced interplay between individual, sound exposure, and foraging behaviour.The male otter also took 2.5 times longer than the female to extract fish, but as trial numbers increased, the extraction time decreased.This could imply potential differences in caution or the ease of extracting fish between the two individuals, where the male was initially more cautious or faced greater challenges in extracting fish.The observed reduction in extraction time over trials suggests that the male adapted more quickly to the experimental setup compared to the female.This indicates potential adaptability differences between the individuals within the experimental setup. The experiments were conducted in shallow water, and the otters only had to dive to 0.8 m to reach the feeding station.Nolet et al. 34 suggested that otters tune their dive times to an expected success rate at the corresponding depth.The increase in the number of dives with the number of trials may be attributed to the animals adapting their diving behaviour as they learned how to retrieve fish from the feeder, ultimately reducing the time needed to extract fish.The reduction in the number of dives could likely be due to annoyance caused by playback sounds, emphasising the complex interplay between acoustic stimuli and otter behaviour.The experimental animals were raised by humans and exposed to high levels of noise from visitors in the public aquarium and water pumps.The minor behavioural responses observed at the lowest received sound levels may be explained by the sounds www.nature.com/scientificreports/being less annoying due to other acoustic disturbances.Also, the otters may have been highly motivated to stay near the feeder for obtaining food, overruling any urge to move away from the sound.These results collectively underscore the multifaceted nature of factors influencing diving behaviour, shedding light on the interplay between acoustic stimuli, trial dynamics, and inherent individual differences in otter behaviour.Overall, our study provides new insights into the effects of sound exposure on otters' behaviour and highlights the importance of considering individual differences and trial number when analysing otter behaviour. As both otters responded at all sound levels played here, we conclude that the otters' response threshold was below 105 dB re 1 µPa both at 1 and 14 kHz.There were a few responses in the control experiments, which could be caused by other disturbances from e.g., staff, visitors, or other anthropogenic noises.The observed head movements in the clear behavioural responses (RS2) were always directed towards the underwater loudspeaker, and the startle responses (RS3) were always directed away from it.This observation may suggest that the Eurasian otter has directional underwater hearing abilities.Another explanation could be that they learned to associate the speaker with noise, and that the directional displacement response was caused by visual cues instead of acoustic ones (e.g., observing the direction to the feeder). The extrapolation of our study findings to wild otter populations introduces a notable degree of uncertainty, particularly due to the unique circumstances of the two individuals used in our experiments.These otters have been in human care and on public display since their early pup stages, experiencing continuous exposure to human presence and associated noise within the controlled exhibit environment.As a result, their reactions to acoustic stimuli may not necessarily mirror those of otters in the wild, being more habituated to anthropogenic sounds. For otters to receive similar sound levels in the wild, the source intensity must be increased as the range to the source will be larger than in the experiments made here.At ranges between tens to hundreds of meters from the source, the source intensity (measured at 1 m) must be increased by at least 20-40 dB, depending on local sound propagation features.Spreading and absorption loss may both be higher and lower than what can be predicted from standard acoustic propagation model, so in situ measurements of sound spreading may be needed.In addition, background noise levels may be either higher or lower than the levels measured in the otter setup used for experiments here. The use of controlled sound exposures demonstrate that otters are acoustically vigilant and respond behaviourally to sound.The animals showed strong responses to relatively weak signals of both 1 kHz and 14 kHz signals, and their behavioural responses were significantly related to sound level, but not to sound frequency.Implementing AHDs at 1 kHz to deter seals would therefore not solve the problem of not affecting the otters in their natural habitat, as suggested to be the case for the harbour porpoise 17 .Comparing the in-air hearing with other animals, it appears that the Eurasian otter has inferior low frequency hearing 15 compared to other mustelids, such as the sea otter 16 , least weasel (Mustela nivalis) 35 , ferret (Mustela putorius) 36 , as well as pinnipeds, such as the harbour seal and northern fur seal (Callorhinus ursinus) 37,38 .As the otter forages under water and spends a large proportion of life under water, it is reasonable to assume that their hearing is adapted to this environment as indicated by our data. The findings here will allow managers to make more informed decisions by incorporating species-specific information into risk assessments.We conclude that an acoustic harassment device has a high potential of affecting the diving and foraging success of otters, as shown for both individuals with a significant decrease in dive duration and increase in the time necessary to get fish out of the feeder.The highest received sound level used during these experiments was 145 dB re 1 µPa.In comparison, effective commercial AHDs/seal scarers have a sound level of 190-200 dB re 1 µPa [39][40][41][42][43][44] .Using commercially available AHD devices in rivers will obviously incur much stronger reactions from otters, probably including displacement and habitat exclusion by restricting access to the sea and movements between rivers.The behavioural and dive responses at 1 kHz were significant and similar to the 14 kHz signals and can therefore not be recommended as an alternative.Research concerning sound reception in Eurasian otters is still needed to measure specific underwater hearing sensitivity (audiograms).In order to better describe the acoustic biology of this species, more comparative studies are needed of auditory anatomy, neurophysiology and detailed characteristics of acoustic communication in their natural environment. Conclusion The otters in this study showed clear behavioural response to both 1 and 14 kHz underwater sounds at 105-145 dB re 1 µPa.Otter sensitivity to sounds should be taken into consideration before installing AHDs to reduce presence of seals in areas with breeding and foraging otters or Natura 2000 areas where otters are under protection. Figure 1 . Figure 1.Drawing of otter enclosure in AQUA Aquarium and Wildlife Park, Denmark, illustrating land, logs, water (blue), the two otters at the end of the feeder and the setup in Area A. The letters refer to the areas with various water depths where background noise was measured.Top right: the setup on the roof above Area A, with the laptop connected to the blue underwater speaker visible in the water.The underwater camera mounted above the speaker. changes in head orientation RS1 Minor behavioural response Withdrawal from the feeder RS2 Clear behavioural response Withdrawal from the feeder, head movement towards the speaker, and/or change in direction RS3 Startle response Leaving feeder and/or surfacing NO RESPONSE RS0 N o changes in head orientation RESPONSE RS1-RS3 Withdrawal from the feeder, head movement towards the speaker, and/or other change in head direction, leaving and/or surfacing Figure 2 . Figure 2. Video frame sequences of a trial with no response (RS0) for 14 kHz at 115 dB re 1 µPa, minor response (RS1) for 14 kHz at 115 dB re 1 µPa clear response (RS2) for 14 kHz at 125 dB re 1 µPa, and a startle response (RS3) to 14 kHz at 135 dB re 1 µPa.The sequences show the otter 1 s before until 1 s after playback.The feeder was placed at 0.8 m depth facing the speaker at a distance of 1 m. Figure 4 . Figure 4. Dive duration (s) with no sound (control, left), 1 kHz (middle) and 14 kHz (right) playback exposures at 105-135 dB re 1 µPa.The average number of dives is shown a stippled lines for the female (pink) and the male (green) otter. Figure 5 . Figure5.Time (s) spent getting the fish out of the feeder with no sound (left), for 1 kHz (middle) and 14 kHz tones (right) for both female (purple) and male otter (green) divided by the source level 105-145 dB re 1 µPa.The overall average time for each animal to get the fish for the two frequencies is illustrated with the dashed lines (pink for female and green for male).	v3-fos-license
2016-03-22T00:56:01.885Z	2014-07-31T00:00:00.000	20783557	{ "extfieldsofstudy": [ "Chemistry", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1420-3049/19/8/11316/pdf", "pdf_hash": "ef648398ce5d36e57d2b6e939da8d1845a33ca0e", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45343", "s2fieldsofstudy": [ "Chemistry" ], "sha1": "2f0ed83287692df3fa367198fc8b6ee2dcd5e30b", "year": 2014 }	pes2o/s2orc	Redox-Dependent Conformational Switching of Diphenylacetylenes Herein we describe the design and synthesis of a redox-dependent single-molecule switch. Appending a ferrocene unit to a diphenylacetylene scaffold gives a redox-sensitive handle, which undergoes reversible one-electron oxidation, as demonstrated by cyclic voltammetry analysis. 1H-NMR spectroscopy of the partially oxidized switch and control compounds suggests that oxidation to the ferrocenium cation induces a change in hydrogen bonding interactions that results in a conformational switch. Introduction Conformational dynamism in the face of changing cellular redox conditions is essential to the survival of living organisms. Protein structure, and thus activity, is often regulated by redox-dependent disulfide bonds [1,2] and metal coordination [3,4]. These natural mechanisms of control inspired us to explore transforming the H-bonded diphenylacetylene (DPA) scaffold into a redox-dependent switch. OPEN ACCESS Much work has been carried out to characterize synthetic systems that can change conformation in response to oxidation state. The most well known of these involve the use of redox active catenanes, [5][6][7][8] metal coordination compounds, [9][10][11][12][13][14][15] rotaxanes, [11,12,14,[16][17][18][19][20][21][22] π-systems, [23][24][25][26][27][28] and crowded alkenes [29]. These systems are finding increasing application in solid-state electronic devices. Of particular note are the [2]-rotaxane systems in which a H-bond donating macrocycle shuttles between a strong H-bond accepting succinamide station and a weaker naphthalimide H-bond acceptor [30]. It has been demonstrated that reduction of the naphthalimide group to the radical anion increases its H-bond acceptor strength relative to the succinamide. This change causes the H-bond network to reconfigure, leading to a conformational switch. Analogously it should be possible to design a conformational switch in which a H-bond network reconfigures due to a redox-dependent modulation of a H-bond donor. Previous work has shown that the conformational equilibrium in H-bonded DPA's can be controlled by increasing the H-bond donation strength of one NH relative to the other [31,32]. This H-bond strength is readily adjusted through the conjugation of electron-donating or -withdrawing groups to the amide NH. By this same principle, conjugating a ferrocene (Fc) to the H-bond network should mediate H-bond donor strength in a redox-dependent fashion ( Figure 1). Although used extensively in the field of sensors, [33][34][35] ferrocene has received little attention as an actuator of redox-dependent conformational switching [36]. Fc is slightly electron-donating [37,38] as compared to a phenyl group, which suggests that the H-bonded equilibrium will be biased away from the ferrocenyl amide (FcA) in the neutral state ("reduced", Figure 1). However, ferrocene is also known to undergo a reversible single electron oxidation to the ferrocenium cation [39]. Studies on 1-ferrocenylcarboxamide systems show that the oxidation of Fc withdraws electron density from the amide NH bond, creating a stronger H-bond donor [40][41][42][43][44][45][46]. Thus, it is hypothesized that oxidation to Fe(III) should induce a switch in the H-bonded equilibrium toward the FcA ("oxidized", Figure 1). Synthesis In order to test this hypothesis we synthesized diphenylacetylene 3 via an amide bond formation between aniline 1 [31] and ferrocenoyl chloride 2 (Scheme 1). Solid-State Analysis of Conformation Insight concerning the conformation of diphenylacetylene 3 was obtained from single crystal X-ray diffraction following recrystallization from 1:1 methanol/chloroform ( Figure 2, see also Section 3.4.). The H-bond acceptor is bound to the benzamide NH with an N•••OC distance of 3.1 Å. The steric clash between the methyl ester and the phenyl ring causes the ring to rotate 35° out of the amide plane. The solid-state data confirm the predicted conformation, which is presumably favoured due to the electron-donating character of the Fc group reducing the H-bond donor capability of the corresponding amide, as well as the steric demands of the large sandwich complex. Determining the Solution Phase Conformational Bias As for the anion [47] and pH-dependent [32] switches described previously, the conformational bias can be determined by comparison of the 1 H-NMR spectrum of 3 with a set of control compounds. para-Substituted benzoic ester 5, which is incapable of intramolecular H-bonding, was synthesized via amide coupling of aniline 4 [31] with ferrocenoyl chloride (Scheme 2). The second control molecule, 12, was prepared from 3-nitroaniline via an adaptation of a literature route: [31] iodination of 3-nitroaniline, [48] followed by amidation, nitro-group reduction, Sonogashira coupling and coupling with ferrocenoyl chloride, afforded the desired bis(amide). The first of these, 5, estimates the chemical shift of the FcA NH in the absence of H-bonding by positioning the H-bond acceptor para to the alkyne linkage (0% control). The second control, 12, estimates the chemical shift of the FcA NH when it is completely H-bonded (100% control). The spectra of these controls are compared with parent compound 3 in Figure 3. Using Equation (1), the conformational bias is calculated to be 14.3%, which equates to a ratio of ~1.4:1 toward the benzamide NH, in agreement with the preference demonstrated in the solid state (vide supra): Characterizing the Redox Properties of Switch 3 with Cyclic Voltammetry The redox properties of the ferrocene-containing compounds 3, 5 and 12 were examined by cyclic voltammetry (Figure 4, Figure 4b) All three compounds exhibit one-electron oxidation waves characteristic of the Fe(II)/Fe(III) couple. The peak potential separations (ΔE p = 130-170 mV) are greater than the 59 mV expected for a reversible one-electron process; however, these numbers are on the order of the ferrocene internal reference (ΔE p, Fc = 165 mV at 100 mV·s −1 ), and the ratio of peak currents (I a /I c ) are all close to unity, indicating that these are chemically reversible processes. Positive currents correspond to oxidation. Potentials were measured relative to the ferrocenium/ferrocene redox couple as an internal standard, and are reported relative to the NHE; (b) Derived electrochemical parameters (see Experimental Section 3.1. for further details). Characterizing the Redox Properties of Switch 3 by Chemical Oxidation Having shown that putative switch 3 undergoes reversible oxidation, we attempted to use the 1 H-NMR conformational assay described above to analyze [3] [36,49]. They report that, while partial oxidation causes the Fc signals to broaden and shift downfield, other resonances belonging to protons further from the FcA group remain sharp. In order to determine the conformational ratio of a partially oxidized 3, conditions must be identified that bring about the partial oxidation of this Fc DPA system. Figure 5a shows the NMR spectra of 3 after exposure to various chemical oxidants. The extent of oxidation was qualitatively assessed by examining the cyclopentadienyl peaks, 5.1-4.0 ppm, which are known to broaden and shift downfield upon oxidation. After two hours only copper(II) chloride and silver tetrafluoroborate caused the expected change in the signals of 3. Additionally, copper(II) chloride caused some broadening of other signals in the aromatic region but the NH resonances are both identifiable. It is plausible this broadening is due to one or more of the oxidation states of copper acting as a Lewis acid with acceptor lone pairs of the switch compound. Silver tetrafluoroborate caused the loss of all spectral definition, most likely due to more complete oxidation ( Figure 5b). The presence of a ferrocenium group was also confirmed by the appearance of the characteristic peak at 640 nm in the absorbance spectrum ( Figure 5c) [50]. Paramagnetic NMR Characterization of the Solution Phase Conformation With a procedure to effect the partial oxidation of 3 in hand, we next began to investigate the influence of oxidation on the spectral features. Figure 6a compares the downfield NMR spectrum of [3] + with that of neutral 3. The FcA NH is broadened, and shifted downfield by 0.05 ppm while the benzamide NH migrates 0.14 ppm upfield. The singlet corresponding to the methyl ester was also affected, broadening and shifting downfield by ~0.6 ppm. To understand the meaning of these changes, control compound 5 was used to examine the effect of oxidation in the limiting case of 0% H-bond interaction. This compound was treated with copper(II) chloride for two hours and, after the presence of the ferrocenium was confirmed by UV-vis, its 1 H-NMR spectrum was acquired. This spectrum indicated that FcA NH and benzamide NH signals of [5] + shift upfield by 0.11 and 0.03 ppm respectively, relative to 5 (Figure 6b). The methyl ester resonance was unaffected by oxidation, presumably due to its isolation para to the alkyne. Lastly, 12 was oxidized with copper(II) chloride; the partially oxidized spectra of 12 and [12] + are shown in Figure 6c. In this case the FcA NH broadens but is not shifted, while the benzamide NH is not identifiable. Furthermore, the methyl ester peak was broadened to such an extent that it was not observable in the partially oxidized spectrum. Attempts to monitor IR stretching frequencies of the carbonyl, or NH, regions to provide conformational data was not possible due to poor resolution of overlapping signals. Taken together, these observations suggest that the H-bonded equilibrium is changing as the Fc group becomes oxidized. The benzamide NH resonance, for example, shifts upfield both in the absence and presence of the intramolecular H-bond (Figure 6a,b), but the signal migrates further when the H-bond is engaged ([3] + ). This increased upfield shift can be explained by a switch of the H-bond acceptor away from the benzamide NH, which causes this proton to be more shielded. The Fc + amide NH shifts upfield in the absence of H-bonding upon oxidation (Figure 6b), but shifts downfield when the H-bonded equilibrium is engaged (Figure 6a). A switch of the H-bond acceptor towards the Fc + amide NH, causing this proton to become deshielded, would explain the downfield shift in the spectrum of [3] + . To estimate the effect of partial oxidation on the conformational equilibrium, the spectra of [3] + , [5] + and [12] + are directly compared (Figure 6d) using the same method as in the neutral case. This comparison suggests that the equilibrium is biased toward the Fc + amide NH by 15.6%, or ~1.4:1 in favour of the Fc + amide (compared with 1.4:1 in favour of the benzamide prior to oxidation) under these partially oxidized conditions. Since the 1 H-NMR resonances corresponding to the parent unoxidized species are not evident it is probable that there is fast exchange between the unoxidized and oxidized forms, leading to average peak positions. Whilst this makes precise quantification of the bias more difficult to establish, the overall trends hold thus providing a semi-quantitative measurement. General Methods Dichloromethane, tetrahydrofuran, and N,N-dimethylformamide were dried using an Innovative Technology SPS-400 dry solvent system. Anhydrous methanol, ethanol, isopropanol, and dimethyl sulfoxide were purchased from Sigma-Aldrich and used directly from their SureSeal TM bottles. All reactions were performed under an atmosphere of dry nitrogen in oven-or flame-dried glassware and were monitored by thin-layer chromatography (TLC) using silica gel (visualized by UV light). Aqueous solutions were saturated unless otherwise stated. 1 H and 13 C NMR spectra were recorded on 400 or 500 MHz Bruker or 500 MHz Varian instruments. Chemical shifts (δ) are reported in parts per million after reference to residual isotopic solvent. Spectra measured in CDCl 3 were referenced to 7.27 and 77.16 ppm for 1 H and 13 C. Spectra measured in CD 2 Cl 2 were calibrated to 5.32 ( 1 H) and 53.52 ( 13 C) ppm. Coupling constants (J) are reported in Hertz (Hz). Proton assignments were pre-formed using MestReNova "multiplet reporter script", and were edited by hand. High-resolution mass spectra were measured on a 9.4 T Bruker Qe FT-ICR MS and values are the average of three measurements. All chemical drying was pre-formed with sodium sulfate unless otherwise stated. Cyclic voltammetry was performed with an EG&G Princeton Applied Research Model 273A potentiostat/galvanostat using platinum disc (1.6 mm diameter) working electrode, a platinum counter electrode, and a silver wire pseudo-reference electrode in a conventional three-electrode cell. Anhydrous dichloromethane was used as the solvent. The supporting electrolyte was 0.10 M tetrabutylammonium hexafluorophosphate, and bubbling with nitrogen deoxygenated the solution. Polishing with alumina slurry, followed by solvent rinses, cleaned the platinum disc working-electrode. The concentration of the electroactive compound was 2.0 mM. The potential of the pseudo-reference electrode was determined using the ferrocenium/ferrocene redox couple as an internal standard (with E 1/2 taken as 0.690 V vs. NHE in dichloromethane). For aqueous solutions, the supporting electrolyte was 0.1 M sodium sulfate. All reported voltammograms were recorded at a 100 mV·s −1 scan rate. All potentials listed in this manuscript are referenced to the normal hydrogen electrode (NHE) unless otherwise stated. Ferrocenoyl chloride (2). Oxalyl chloride (0.058 mL, 0.66 mmol) was added dropwise over 1 min to a solution of ferrocene monocarboxylic acid (0.0777 g, 0.338 mmol) in dichloromethane (3.0 mL) and N,N-dimethylformamide (1-3 drops). The mixture stirred for 30 min and concentrated with a stream of nitrogen gas, and then subjected to vacuum for 10 min. The resultant crude oil was used without further purification. (3). Pyridine (0.011 mL, 0.14 mmol) was added dropwise to a solution of amine 1 (0.0500 g, 0.135 mmol) and 4-dimethylaminopyridine (ca. 1 mg) in dichloromethane (4.5 mL). A solution of ferrocenoyl chloride 2 (0.9 mL, 0.3 M, 0.27 mmol) was added dropwise over 3 min and the mixture stirred for 18 h. Following dilution with dichloromethane and washing with 2 N hydrochloric acid, sodium hydrogen carbonate, and brine, the organic layers were dried over magnesium sulfate and concentrated in vacuo. (5). Pyridine (0.011 mL, 0.14 mmol) was added dropwise to a solution of amine 4 (0.0500 g, 0.135 mmol) and 4-dimethylaminopyridine (ca. 1 mg) in dichloromethane (4.5 mL). A solution of ferrocenoyl chloride 2 (0.9 mL, 0.3 M, 0.27 mmol) was added dropwise over 3 min and the mixture stirred for 18 h. Following dilution with dichloromethane and washing with 2 N hydrochloric acid, sodium hydrogen carbonate, and brine, the organic layers were dried over magnesium sulfate and concentrated in vacuo. The residue was purified by column chromatography on silica gel (10). Tin(II) chloride dihydrate (0.153 g, 0.68 mmol) was added to a solution of nitro aromatic 9 (0.050 g, 0.136 mmol) in ethyl acetate (5 mL). The mixture was stirred for 18 h and diluted with ethyl acetate and poured into sodium hydrogen carbonate. The mixture was filtered over Celite ® and the filtrate was extracted using ethyl acetate, dried over magnesium sulfate, and concentrated in vacuo. The residue was purified by column chromatography on silica gel (2:1 hexanes/ethyl acetate) to give the title compound 10 (0.031 g, 68%) as a white solid; R f 0.5 (11). A solution of 10 (0.031 g, 0.0917 mmol) and methyl 2-alkynylbenzoate 6 (0.0176 g, 0.110 mmol) in N,N-dimethylformamide (1.0 mL) and triethylamine (1.0 mL) was degassed by nitrogen stream for 10 min. Bis(triphenylphosphine)palladium(II) dichloride (0.0038 g, 0.0055 mmol) and copper(I) iodide (0.0017 g, 0.0092 mmol) were added and the mixture was heated to 70 °C. The reaction was allowed to stir for 2 h then diluted with ethyl acetate and washed with water. The organic layer was dried and concentrated in vacuo. The product was purified by column chromatography on silica gel Chemical Oxidation Procedure Copper(II) chloride (0.0025 g, 0.019 mmol) was added to a solution of ferrocenyl compound (3, 5, or 12) in dichloromethane (3.87 mL, 4.0 mM, 0.016 mmol), and the mixture stirred for 2 h, giving a dark greenish solution. The solution was filtered, transferred to a cuvette, and the UV-vis absorbance spectrum obtained to determine the presence of a ferrocenium group. The solution was concentrated and re-dissolved in 3.87 mL of CDCl 3 for NMR characterization. Single Crystal X-ray Diffraction Crystallographic data (excluding structure factors) have been deposited with the Cambridge Crystallographic Data Centre (CCDC: 871793) and copies of these data can be obtained free of charge via the web [52] The crystal-to-detector distance was 127.40 mm. Readout was performed in the 0.100 mm pixel mode. Data Reduction Of the 17,074 reflections that were collected, 4,399 were unique (R int = 0.0563). The linear absorption coefficient, μ, for Cu-Kα radiation is 49.717 cm −1 . An empirical absorption correction was applied which resulted in transmission factors ranging from 0.140 to 0.204. The data were corrected for Lorentz and polarization effects. Structure Solution and Refinement The structure was solved by direct methods [53] and expanded using Fourier techniques. The non-hydrogen atoms were refined anisotropically. Hydrogen atoms were refined using the riding model. The final cycle of full-matrix least-squares refinement on F 2 was based on 4383 observed reflections and 371 variable parameters and converged (largest parameter shift was 0.00 times its esd) with unweighted and weighted agreement factors of: R1 = ∑ \|\|Fo\| − \|Fc\|\|/∑ \|Fo\| = 0.0504 (2) wR2 = [∑ (w (Fo 2 -Fc 2 ) 2 )/∑ w(Fo 2 ) 2 ] 1/2 = 0.1294 The standard deviation of an observation of unit weight was 1.10. Unit weights were used. The maximum and minimum peaks on the final difference Fourier map corresponded to 0.58 and −0.86 e − /Å 3 , respectively. Neutral atom scattering factors were taken from Cromer and Waber [54]. Anomalous dispersion effects were included in Fcalc; [55] the values for ∆f' and ∆f" were those of Creagh and McAuley [56]. The values for the mass attenuation coefficients are those of Creagh and Hubbell [57]. All calculations were performed using the CrystalStructure [58] crystallographic software package except for refinement, which was performed using SHELXL-97 [59]. Conclusions The importance of redox-dependent conformational switching in Nature inspired the development of an analogous synthetic molecular switch. Previous studies have demonstrated redox switching by modulation of H-bond acceptor strength, but there are few examples in which this is mediated by tuning of H-bond donor ability. Conjugation of a ferrocenylcarboxamide to a diphenylacetylene scaffold provided a model system for redox-dependent conformational switching. X-ray crystallography established that the electron-donating character of the Fc group causes hydrogen bonding to the benzamide to be preferred in the solid state, and 1 H-NMR analysis confirmed this behavior in solution. Upon partial oxidation the compound undergoes a switching of conformational ratio from 1.4:1 in favour of the benzamide to 1.4:1 in favour of the ferrocenyl amide; however, due to paramagnetism of the Fe(III) species, 1 H-NMR analysis of the fully oxidized species was not possible. Cyclic voltammetry also supports the predicted conformational change upon oxidation, by showing that the Fe(II)/Fe(III) redox potentials decrease in order of the expected increase in hydrogen bonding to the ferrocenyl amides from the 0% control to the 100% control. Future work will investigate the use of alternative spectroscopic methods to assay diphenylacetylene switching under redox-mediated conditions, including the exploration of IR marker bands and colorimetric methods.	v3-fos-license
2021-10-23T15:24:51.774Z	2021-10-20T00:00:00.000	239523363	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2071-1050/13/21/11609/pdf", "pdf_hash": "364a1f296e83b20bfbc70087bdc84ba5f61eb7ca", "pdf_src": "MergedPDFExtraction", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45344", "s2fieldsofstudy": [ "Environmental Science", "Education" ], "sha1": "ea7d38c4592f3d1f52b271a7ae8dbabee55c6960", "year": 2021 }	pes2o/s2orc	The Learning Space as Support to Sustainable Development: A Revision of Uses and Design Processes : In the last decade, there has been an increasing interest in the role of space in the learning process. However, there is limited research about how different Learning Spaces (LS) can lead to Sustainable Development (SD). Therefore, this paper presents a systematic literature review aimed to identify how physical, virtual, and hybrid LS have been designed and used to support SD. From an initial sample of 204 articles between 2009 and 2021 found in the Scopus database, 33 were included after inclusion criteria were applied. Findings show a wide variety of focus in the uses of LS (e.g., promote education quality, accessibility, or environmental sustainability). In general, the design process of LS implies a top-bottom approach, where students remain as passive actors. Nevertheless, it has been identiﬁed a growing interest in codesign processes that promote broader participation and bottom-top perspectives. This study contributes to orientate the understanding of the concept of LS, and looks towards inspiring new teaching and learning practices. Introduction During the COVID-19 pandemic, education sector has been disrupted. Around 94% of worldwide students were affected by required mandatory close-downs of schools and other learning spaces [1]. During this period, homes became the main space where physical and digital elements converged to support education [2]. Through this crisis, in the public debate has been observed a rush for trying to get back to normal [3]. Contrarily, critical voices alert about the risk of returning as before (if it is even possible). Therefore, it is proposed to use this moment as an opportunity to re-orientate education systems towards Sustainable Development (SD) [4]. Three dimensions must be balanced and integrated to archive SD: social, environmental, and economic [5]. In this process, education plays a significant role in supporting the transition. However, the institutional activities directed to SD, have been limited to integrating sustainability concepts in their curriculum and offering specialty courses for careers in related fields. These activities are helpful but not enough to promote behavior and attitudinal changes [6]. Then, it is proposed to create interactive and learner-centered spaces [7]. This requires holistic approaches to address the different aspects of learning environments, refered to the diverse locations, contexts, and cultures in which students learn [8,9]. The research of learning environments initially was focused on psychological and social factors. Lately, there has been a growing focus on the role of spatial and physical elements involved in the learning process. This area is conceived as the study of Learning Spaces (LS) [10]. Empirical evidence shows that LS have a significant effect on learning outcomes. For example, it has been studied how the morphological composition of a physical environment can support six learning theories (behaviorism, cognitivism, constructivism, experiential, humanistic, and social-situational) [11]. Nevertheless, studies that explore how learning spaces are designed and used to archive specific learning outcomes are limited [12]. There is not a simple answer about how to design them, but there is an increasing interest in establishing principles for it [13]. Traditionally, LS refers to school-built spaces [14]. Regardless, it has been recognized that all areas occupied by humans at every scale can become LS [15]. Then LS may include outside-of-school locations, outdoor places, and work settings expanded with learning features [16]. Additionally, concepts as digital learning environments have been growing along with the integration of education with information and communication technologies (ICT). By recognizing the fast-changing context of the twenty century, advanced technologies have been proposed to enhance the learning experience. Examples include the internet of things, artificial intelligence, big data, augmented/virtual reality, and blockchain [11]. Then, it is considered that learning in 2021 occurs in multiple physical and digital spaces, at different times, and following a diversity of means and methods [17]. Some literature reviews education for SD, focusing on analyzing the main themes in this field [18], the conceptualization and operationalization of concepts [19], and the evaluation process of educational buildings and learning environments [10]. However, it has been identified that studies focused on the design and use of LS to support and promote SD are limited. Therefore, this work presents a systematic review of literature to identify case studies towards response the following research questions: RQ1. How have LS been used to support SD? RQ2. How LS that supports SD have been designed? This paper starts by offering a description of the main concepts of the review in Section 1.1.1. Then, the methodology used based on the PRISMA 2020 checklist is presented in Section 2. Main findings and insights that provide a general overview of cases where LS have supported SD are presented in Section 3. Reflections, recommendations, and further research topics are discussed in Section 4. Finally, Section 5 offers the conclusions of the authors. Among the different perspectives that seek to study and describe the phenomenon of learning, one of them considers that the key factor for its study is to observe it in context. Situated learning proposes that this phenomenon manifests itself as the result of the interactions of the student and a vast constellation of factors that transcend individual cognitive processes. One such factor is the place where it occurs [20]. These places can be studied from a socio-material perspective and described as learning environments, where space is one of its constituent elements [21]. The term space can refer to a time-lapse or to where matter is located and contained [22]. In this work we focus on the second one. From a physical perspective, space is where matter is located and contained. It can be described as a relatively objective three-dimensional extension of reality, defined by the contours of natural or architectural structures and locations [23]. In an abstract form, the matter is substituted by substantive items. This approach is helpful to describe the digital, cyber and/or virtual space phenomenon that is seen as an alternative and not as a copy of the physical space [24,25]. Access to digital spaces relies on a physical object (e.g., digital device screen) powered by information and communication technologies (ICT). Therefore, the study of digital spaces may include physical artifacts [26]. Education and SD The base foundations of the idea of sustainability and sustainable development can be identified since the first civilizations, gaining particular relevance in the XX century [27]. This is reflected in the actions of academics, activists, and international organizations as the United Nations towards promoting it globally. Although its consecution is considered a wicked problem, there have been efforts to establish a common ground for its study and oriented practice [28]. One of them is the development of the 2030 agenda, which includes 17 focus areas and 169 specific targets. This document offers a general strategy and proposes actions to tackle the global challenges to benefit current and future generations [29]. For this study, it is considered that this plan is helpful to operationalize the complexity of the SD concept. For SD, education is considered a goal and a mean at the same time. Therefore, it is regarded as one of the critical factors for SD [30]. The 2030 Agenda proposed by the United Nations presents 17 Sustainable Development Goals (SDGs). Objective number 4, called Quality Education, seeks to promote access to quality education inclusively, focusing on increasing the number of students at different educational levels [31]. However, not all education is considered to support SD. For this reason, the Education for Sustainable Development (ESD) approach has been proposed. In this case, the objective is empowering people to make informed and responsible decisions, where people are part of the process of seeking sustainability. For this, the development and assessment of competencies have been proposed (e.g., anticipation, collaboration, systemic thinking, critical thinking) [32]. Design for SD In order to archive SD, there is a need to transform strategies into actions [33]. This implies a conscious take decision process to develop experiments and interventions to materialize creativity) [34]. Modern design looks to specialize in applying knowledge from all specialized fields, combining technologies with human efforts to transform the world [35]. Social design and transitions design has been focusing on offering methods and tools to the consecution of sustainable development [36]. These perspectives are characterized by collaborative and holistic approaches to changing current into desirable situations by developing products, services, processes, systems and spaces [37]. In the educational context, the LS design specializes in the systematic analysis, planning, development, implementation, and evaluation of settings where learning takes place [17]. Therefore, it is considered relevant to identify formal and not-design processes in this field that promotes SD. Methods This literature review followed the recommendations proposed in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist. The keywords selected and used to create search strings were: learning space, sustainable, sustainability, design, process, and framework. The search included scientific articles, conference papers, and academic book chapters in English, which was conducted among the title, abstract, and keywords. The search was conducted on 1 May 2021, limiting to publications between 2009 and 2021 found in the Scopus electronic database. These years were selected by using the date that 4G networks started to operate as the time reference point. This event has been considered relevant into the diffusion of the web 2.0 or social web. After excluding duplicates, the initial sample resulted in 209 works. Then a screening and coding process was conducted to categorize the documents. By reading abstracts and full texts, a final sample was selected, excluding papers that do not contain explicitly concepts related to: • SD Goals (e.g., quality education); • ESD competencies (e.g., problem-solving); • Physical/digital/hybrid spaces. Inductive qualitative content analysis was used to code and categorize the items. Information about the characteristics of LS, SD focus, and design process was extracted from each work. The final sample resulted in thirty-three articles included in the review. The PRISMA flowchart that was followed in this study is presented in Figure 1. Records identified in database n = 282 Records after dulicates were removed n = 209 Duplicates removed n = 73 Full text articles after screening n = 117 Studies removed after applying inclusion criteria n = 84 Studies includes in the review n = 33 Findings Final sample included 33 articles (80%), 5 book chapters (16.7%), 2 reviews (6.7%) and 2 conference papers (6.7%). The disciplinary focus of the documents included education (42.4%), environmental sciences (33.0%), social sciences (15.2%), engineering (6.1%), multidisciplinary (3.0%). From a bibliometric perspective, it is observed that after 2017 the number of studies that include the LS and SD concepts has grown substantially. However, as Figure 2 shows, the number of studies included in this review remained relatively low for the screened sample. This is attributed to vague descriptions of concepts as space and sustainability in some of the studies reviewed. Physical Space In the revision it has been identified a growing interest in the relations between characteristics of physic space and learning ( Table 1). The majority of the works found were published between 2019 and 2021. Thirteen studies were selected in this category, representing 38% of the final sample. Table 1 shows the main findings. The broader topic in this area is related to how the characteristics of the buildings can influence the learning experience. Acoustic comfort is considered as a factor to facilitate innovation that helps in progress towards SDGs. A series of recommendations in order to improve the acoustic conditions of LS are mentioned. These ranged from special furniture, the use of impact-absorbing materials in interiors, and the design of new geometries of space [38]. Under a similar approach, academic performance and physical space perception of classrooms have been correlated. There, ventilation was found as a significant influence on academic performance, encouraging further research in the relation of space and learning from a holistic approach [39]. Another study explores open-air buildings and their relation with thermal and visual comfort, health, wellbeing, and energy consumption. There, good and bad practices are proposed to reduce energy and maintenance costs, prevent health disorders, and supporting the learning process [40]. In order to study the relationship between space with math and arts education, an Indoor Physical Environment Perception scale is used to assess classroom characteristics and classifying them into three categories: workspace comfort, natural environment, and building environment [41]. The importance of the physical, material, and aesthetics conditions of schools is emphasized, concluding that the design of learning spaces can support or be a barrier to the learning process. In other cases, spaces may be incoherent with sustainability practices, recommending further research to explore the daily aesthetics of learning spaces [42]. The design of schools is another recurrent topic. There, the importance of collaboration between multiple stakeholders is highlighted, especially to create a sustainable campus model [43]. In this area, it is recommended to include students with disabilities in the design process related to different elements of the campus. This group finds physical barriers in infrastructure and space as one of the main difficulties in their learning process [44]. Two cases shows specific learning spaces created with a focus to promote ESD and SDGs. First, a study case focused on industrial engineering education. There, in a wood workshop, the student studied and applied lean manufacturing concepts by simulating different role positions (e.g., operator, supervisor, inspector, manager). This led to the development of knowledge and competencies required in the job market [45]. Second, a proposal was found to design-build a tiny house to generate an experiential learning space. This context served to learn about issues related to energy, housing, and the environment. The proposal includes the active participation of students from structural analysis, sustainable design, architecture, and energy utilization courses [46]. An interesting research area was found in the use of spaces to promote the development of community and networks. For example, the exploration of the everyday activity of commensality (coming together around a table to eat) as a form to promote lifelong and intergenerational learning opportunities. Bring food to share sessions were held in classrooms where the students could do life with one another. In this context, the use of smartphones was questioned and discussed as an obstruction to social connectivity [47]. Under a similar approach, the staffroom is considered as a professional learning space, describing how these can serve as a hub to develop communities that promote SD [48]. Finally, some cases that describe the use of outdoor and out-of-school spaces to develop learning processes were found. For example, in order to teach science topics (chemistry, biology, physics, and math), a learning experience that involved going outside the classrooms and find water sources in the surroundings of schools was developed. There, students explored ideas about water treatment and the role of homeowners in the communities. Results showed a positive enjoyment and engagement of students [49]. Additionally, it was found how some universities promote regional sustainable transitions by conducting activities in spaces outside their campuses. The activities included a range of methods that involved expert talks, hands-on work, excursions, interviews, city walks, label rallies, visits, surveys, mappings, and games. The spaces that hosted these activities included stores, gardens, agriculture fields, newspaper offices, parks, start-ups, community housing projects, libraries, and festivals. The activities were received positively by the students, teachers, and regional partners. However, it implied additional administrative and organizational difficulties, ranging from insurances, entrance fees, transport, extra payment for teachers, and questions from authorities [50]. Digital Learning Space This alternative to physical space offers the possibility to navigate between multiple scales of spaces that promotes a variety of sustainability focus (Table 2). However, it was found that the access barriers to these spaces are a general concern in this research. It is considered that in order to access a digital LS, there is a need for a physical artifact, being the first element of that space. Following this perspective, a study case shows how Ipads were used to promote the problem-solving skills of students. Despite its popularity in schools, professors have found the Apple ecosystem locks as a barrier to implementing new teaching strategies. Storage capacity was an issue that caused significant problems when software upgrades, apps, and student work start to fill up. Additionally, a lack of technical knowledge was a fundamental challenge to solve in developing this educational program [51]. Access to hardware does not guarantee access to digital spaces. A study exploring the challenges of offering quality education to students with visual impairments during the COVID-19 pandemic found that the most crucial factor for students was access to internet data, which in most cases implied a cost for the students [52]. A solution to this challenge is proposed in the use of Facebook as a digital LS for indigenous students, instead of learning management systems traditionally used by universities to provide content (e.g., Moodle, Blackboard). Additionally to the familiarity with the interface and the social activity of Facebook, an important factor of the positive feedback was the fact this platform is part of their mobile plan [53]. Then, the range and characteristics of digital spaces available are a recurrent topic in this category. To support collaboration within and between communities to accelerate sustainable community development efforts, a specific digital LS was designed. This online platform offers meeting rooms, online boards, a forum, and a private library [54]. A similar approach was found in a digital LS designed for self-directed and team-based learning challenges. Here, the intention was to get an unobtrusive data collection of users in order to improve educational programs [55]. From a meso level, the concept of virtual university was identified. There, it is recognized the utility to study online courses from a learning environment perspective, referring to putting learning as the primary purpose and direction of it. However, it is proposed to explore the concept of virtual space going beyond simple online courses. This means that students could move, explore, experiment, have serendipity moments, make choices about this space, and then turn it into a place with cultural significance (e.g., protests, commerce, celebrations). As a result, this could facilitate lifelong learning and problem-solving thinking [56]. Finally, from a macro perspective, a research work shows how to promote the incorporation of ICT in education in a country. This case study presents how ICT-based public services have been created to support life-long learning. The objective of this strategy is to cover 100% of schools by digital campuses. This intends that anyone can learn anywhere at any time. By the end of June of 2021, a National Education Resource Public Service Platform in China had connected to 65 online platforms, having 12 million access to online learning courses, and reaching 3.37 million active users [57]. Hybrid Space This work identified cases where the LS is proposed as a combination of physical and digital elements but where their distinctions are recognized (Table 3). Therefore these LS are considered hybrid spaces [58]. During the pandemic period, challenges related to the digital divide have been evident. In this context, it is considered that to keep offering quality education, there is a need to integrate multiple physical and digital elements to create LS that respond to local contexts [59]. In the new normal, traditional teaching and learning have been a challenge, especially where the attendance to the physical classroom is voluntary. In this context, one of the main complaints of teachers is that, in order to support students that attend physical and digital spaces, they now have to teach the same course twice. In this context, an online-mergeoffline learning mode is presented. There, teachers and some students are physically in a classroom. Meanwhile, the rest of the students access remotely to the classroom by the use of interactive screens, cameras, tablets, and sound equipment. The pilot implementation of this model resulted in positive perceptions from teachers and students [60]. This is an example of how hybrid spaces can be understood. In this context, physical and digital LS merge, but it is possible to differentiate them for its study. A case study shows how to provide tools for embedding sustainability in teaching activities outside the pandemic context. As a part of a postgraduate certificate in academic practice, a module of ESD was developed and implemented in two iterations. The first one was conducted just in a classroom using a traditional higher education pedagogy. Focusing on implementing ESD philosophical principles, the second one included study activities in a digital space. Then, there were campus tours involving stairwells, boiler rooms, recycling/waste facilities, kitchens, coffee shops, and parks. Participants showed positive feedback regarding satisfaction, engagement, and perceived contribution to professional development [61]. Another case recognizes the difference between physical and digital elements. There was found an exploration in the use of iPads against the use of laptops to improve the performance of students inside classrooms. There, the public and private workspace concepts are presented, associated with personal and collaborative work. Results show that tablets are more helpful to support student-to-student interactions, improve communication, and triggering co-ideation processes [62]. Finally, it was found the use of the concept of enhanced physical space for reviewing the use of Bluetooth beacons. This technology can be used from attendance monitoring, energy-saving detection systems, offer specific information that promotes recycling activities, and stimulate learning by Augmented reality systems [63]. Design Process In general, the case studies found do not mention the design process of the LS presented. However, it was found that generally, it was followed a top-down approach. Under this perspective, initially, governments and education organizations are responsible for the design process. After formulating strategies, it relies on the teacher and its practice to use or find new uses of space. On a micro-scale, it was identified the classroom layouts as an active element in the learning process. These are usually organized by the teacher, anticipating the needs of the students. However, the difference between students implies that there is a constant negotiation between teachers and students. This reflects the importance of including the students and other actors in the design process, complementing top-down strategies with bottom-up initiatives [64]. This perspective was found in the discussion of the results of an initiative to promote innovative teaching and learning practices, observing that new bottom-up initiatives are constantly emerging. However, these rely on the support of top-down strategies to achieve success [65]. Following this perspective, a process where schools transformed traditional classroom arrangements into flexible learning spaces is presented. From interviews, discussions, brainstorming, and rapid prototyping, different actors were considered to envision a new LS. In this process, the integration of students, teachers, and the community was regarded as the key factor to success. Therefore, it is recommended to use collaborative end-user-centered perspectives instead of top-down approaches [66]. Under the approach to broaden the inclusion of different categories of actors, a Designbased research framework was used to develop sustainable online learning spaces for children with diabetes. In this process, the participation of young people (aged between 11-13), parents, and clinicians were considered fundamental for the project [67]. Additionally, there is an interesting case where a physical transition to a new university building was the catalyst for a reimagining of the teaching practice. This was followed by the design and implementation of a professional development program that looked towards next-generation learning spaces. The design process involved understanding the learning space design and creating the vision of an ideal learning environment from the faculty stakeholders. This vision resulted from a consultation period, where the faculty perceptions of room layouts were an essential element of the process. Additionally, a replica of a learning space was used to facilitate the modeling of teaching practices. Their outcomes assessment of the behavioral attitudinal change in the faculty members showed an increase in the work and collaborative learning, where the layouts of classrooms supported the teaching and learning practice [68]. Insights obtained from a LS codesign project proposes 4 phases: codesign activities, student exhibition and feedback, professional design evaluation, and final satisfaction evaluation survey [69]. A codesign and research-through design approach are also used to design a board game focused on promoting energy topics discussion [70]. A relevant insight found in this case study is that the ambiguity and inconclusiveness of the rules were helpful to evoke group discussions and trigger learning experiences. After a visit to Finland schools, some professors engaged in collaborative practices to seeding new learning experiences. By applying design thinking principles in the development of an outdoor learning experience, the voice of the most dissatisfied group of students was used to get insights about how to motivate students learning and development [50]. Codesign process that involves universities, teachers, and regional partners can be a key factor to the success of the course. This implies extra time and effort. Therefore, it is recommended to address this challenge by the development of networks between science and society. Here, empathy and trust are presented as crucial for teaching cooperation [51]. From a research and diagnosis perspective, the use of an Activity Centered Analysis and Design framework is proposed. In this work, an online survey was conducted to collect quantitative and qualitative data about technologies, material artifacts, and physical spaces of schools. Conclusion refers to the importance of the study of the physical, conceptual, and social structures to support educators towards transitions to new teaching practices that respond to contemporary times [71]. Conceptual Analysis This review found that the use of LS in some cases refers to a broader perspective that does not designates physical, hybrid, or digital spaces. For example, the use of inbetween LS for sustainability learning to describe transdisciplinary research and learning activities [72]. Another example is the use of the concept of LS to describe specific moments where students and staff engage in learning activities [73]. A similar use is identified conferences, forums, workshops, and similar events are proposed as LS to promote climate change learning. In these cases, it looks that the LS concept is used as a synonym of learning environment [74]. The use of LS as an abstract object is common. Under this perspective, the physical space often is not described, ignored, or simply forgotten. This is a sign of the lack of understanding about how space can affect the learning experience. An interesting comment that reflects this emerges from a participant of a Fellowship program described as a transformative LS. The experience was described as mentally and physically stretching. However, the study does not mention any characteristics of the physical conditions of the space where the intensive mental work was healed [47]. Despite the multiple possible interpretations of the term LS, it seems to be useful to establish conceptual differences to describe and understand the learning phenomenon. Therefore, this work proposes understanding the LS as part of a learning environment and three main categories for its study: physical, hybrid, and digital. A substantial part of the literature reviewed shows a constant intention to improve the quality of education and promote sustainable practices. However, in a vast number of works, there is no description of how the concept of sustainability is interpreted or how to assess the impact of interventions on it. Therefore, in some cases, it remains a vague objective or just as a buzzword. Thus, it is considered that the SDGs targets and ESD competencies are useful to operationalize the complexity of the sustainability concept and to describe its relationship with the educational phenomena. The Use of Physical, Digital and Hybrid LS In physical LS, the layout configuration plays an essential role in the learning process. In a large room with a heterogeneous sound quality, access to front rows can be a differential factor during lectures. In some cases, professors have meaningful interactions with students in front rows. Something similar occurred in the digital space. It has been observed that the physical size of the display may have a significant impact on how digital space is accessed and perceived. There exist major differences in the physical space between the screen of a smartphone, a laptop, a monitor, a digital blackboard, or even bigger screens. Therefore, challenges associated with physical LS could emerge in digital setups. For example, in videoconference (e.g., Zoom, Google Classroom) and board platforms (e.g., Trello, Padlet), the size of the screen limits the number of active cameras or boards that can be seen at the same time. Consequently, it is recommended to include the physical characteristics of the devices in the description of digital LS. In hybrid and digital spaces, hardware requirements go beyond the budget of students. This can be supported by laboratories and programs to reduce the cost of hardware. Additionally, private software and platforms may limit access. Then, it seems that there is a need for open-source solutions and guarantees for free access to the internet to promote education and SD. Design and Policies Opportunities The characteristics of physical LS affect the learning experience, impacting the quality of education and the development of competencies for SD. Therefore, their design is a concern for the education practice and research. In schools where students use an only classroom for an extended period, the space can be constantly adapted to different goals. However, classrooms are usually used for diverse courses that do not always share their learning objectives and pedagogies in higher education. This means that classrooms will maintain the same layout ignoring the needs of the specific needs of the courses. A simple solution is using mobile furniture, but offering flexible spaces and allowing the use of multiple spaces could trigger innovative teaching practices. The physical space should be designed collectively. In the process, there are many examples of spaces that can be used as referents to generate ideas about how to generate new spaces, from campus to furniture. In contrast, the existence of digital space is relatively new. Additionally, the lack of technical knowledge could be a barrier to generate new ideas about how digital space may be designed and used. This can impact the teaching and learning process, affecting the development of ESD competencies. It was found that new teaching practices promote experiential learning. Disciplines that usually have hands-on activities (e.g., engineering, architecture, medicine) present opportunities to use multiple spaces out-of-classrooms (e.g., workshops, laboratories, the campus, and the city). Additionally, these disciplines usually integrate SDGs and ESD concepts in their curriculum. However, there are few cases of the use of spaces to promote SD in fields such as law, finances, or management. This is reflected in the disciplines represented in the sample. Therefore, it is considered that there is an opportunity to develop new teaching for those branches of knowledge. An insight obtained in this research is the profound interest in generating a positive impact on the students. Nevertheless, there is a recurrent call for supporting the professors in this process. There are comments of students who considered that teachers are not always prepared to offer new teaching practices involving technology [65]. This implies the need for more training and professor development. There is a significant opportunity to create spaces for professors learning. This is a relevant topic for research because professors usually have the primary responsibility for establishing how the LS are used.In some cases, professors prefer to consult colleagues instead of access to development programs based on traditional pedagogies. Additionally, there is a need for education on SD and ESD competencies to be integrated into the teaching practices. Finally, there is a recurring concern about the resources required for activities out-of-classrooms, being a significant barrier to the development of new teaching practices. Conclusions The disruption caused by the COVID-19 pandemic has created massive pressure in the education sector. Therefore, it has been necessary to imagine teaching practices to respond to this crisis. In this process, different spaces not designed for the learning process have been used, offering evidence of the fundamental role of LS in the learning experience. However, there is still a vague use of the concept. In this concern, this article distinguish three types of LS: physical, digital, and hybrid. This can be useful for researchers to establish a common ground for describing learning setups. This work shows how LS are designed and used to support the achievement of SDGs and development of ESD competencies. The majority of the works reviewed came from before pandemic. Findings suggest that promoting SD is not the main focus of design and use of LS. However, sustainability has increasing importance in public debate for times to come. Therefore, the case studies reviewed give a general overview of possibilities in using different spaces to support SD. New collaboration practices may emerge by changing classroom layouts, using devices like tablets, and using new digital platforms. The use of out-of-classroom spaces could offer experiential learning and trigger innovative teaching practices. By including the voice of users, experts, and other stakeholders in the design process of LS, flexibility and accessibility can be improved. In a dynamic phenomenon as education, further and continuous research is necessary. Therefore, a typology of LS would be helpful for its description and assessment. Additionally, we call for new frameworks that provide a broader perspective of the relation between learning and space, including long life and informal learning.	v3-fos-license
2023-04-15T15:19:03.628Z	2023-04-13T00:00:00.000	258142988	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://doi.org/10.1080/14036096.2023.2200784", "pdf_hash": "8bf6e3d6e04054939b636477704a258d3c6261a7", "pdf_src": "TaylorAndFrancis", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45346", "s2fieldsofstudy": [ "Sociology", "Geography" ], "sha1": "9cb2f8258638ea3f49115cb0c7dffc2d62392bd3", "year": 2023 }	pes2o/s2orc	Understanding the Housing Pathways and Migration Plans of Young Talents in Metropolises–A Case Study of Shenzhen ABSTRACT In the context of skyrocketing house prices and fierce competition for talents between cities, this study explores the housing pathways of young talents and their future migration plans in Shenzhen, China. Using the housing pathways approach and Bourdieu’s theory of practice with three concepts, this study uncovers how structural factors and the often-overlooked agency factors together influence the formation of different housing pathways. Drawing on 18 semi-structured interviews with young talents, four different housing pathways were identified: staying at parents’ home, private renting to owning, talented renting, and progressive private renting. We found that the interaction of habitus and different forms of capital shapes different housing pathways. In addition, young talents following different housing pathways have various future migration plans. This paper sheds new light on the use of the housing pathways approach and Bourdieu’s theory of practice in providing a nuanced understanding of housing and migration behaviour. Introduction In the knowledge-based economy, cities all over the world are struggling to attract and retain young talents 1 to stimulate cities' social and economic development (Florida 2006). The housing career in the city is regarded as a major determinant of whether young talents stay or leave the city (Aner 2016;Cui, Geertman, and Hooimeijer 2015;Dainov and Sauka 2010;Teixeira and Drolet 2018). Some young talents successfully settled in cities following the upward housing ladder such as "rent-to-own" whereas others left the city due to twisted housing ladder routes, such as "snake", "slide down", or "move backwards" (Bobek, Pembroke, and Wickham 2020;Lennartz, Arundel, and Ronald 2016). In these cases, the housing situation did not improve or even deteriorated over time (Dainov and Sauka 2010). The twisted housing ladder routes are particularly relevant in the case of young migrants in metropolitan areas with high house prices like Amsterdam (Hochstenbach and Boterman 2015), Beijing (Wang, Li, and Deng 2017), Helsinki (Eskelä 2018), Klang valley (Hamzah and Zyed 2020), and San Francisco (Kober 2021), etc. In China, for example, an emerging social phenomenon called "Escape from Beijing, Shanghai, Guangzhou, and Shenzhen" indicates that the housing pressure in these firsttier cities 2 is so high that a number of young talents are choosing not to go to or leave these cities Xu, Wang, and Nygaard 2022). According to a survey among 2,000 respondents (previously) working or studying in first-tier cities, 71% have left or are considering leaving this city, with 64% of them blaming it on high house prices (China Youth Daily 2017). The above phenomenon reflects two important social problems. From the perspective of metropolitan cities, the loss of young talents is to some extent reducing the innovation and economic growth of cities (Florida 2005;Wong and Yip 1999). From the perspective of young talents, the precarious housing situation in metropolitan cities is equally alarming. In recent years, the shortage of housing supply, rising house prices and rents, increased educational costs, the tight mortgage market, unstable labour market conditions, riskier employment prospects, etc. have exacerbated the housing difficulties of young people all over the world (Hochstenbach and Boterman 2015;McKee 2012). Therefore, examining the housing situation of young talents in metropolitan cities, exploring what factors contribute to differences in their housing situations, and obtaining insight into their future migration plans are necessary and important. The results could help metropolitan cities to attract and retain young talents and to provide suggestions to find tools to improve young talents' housing situations. Much research has already been done into the housing situation of young people in metropolitan cities. Despite the wealth of literature in this area, there remain two research gaps. First, many existing studies are cross-sectional which tend to examine young people's static housing choices at a certain point in time, such as home-buying (Lennartz, Arundel, and Ronald 2014) and living in shared housing (Maalsen 2019). The cross-sectional approach neglects the changes in the housing situation over time (Coolen, Boelhouwer, and van Driel 2002). The housing conditions attained in the past can strongly influence the present housing situation (Bolt and van Kempen 2002). For example, people who previously had very negative experiences with private rental housing may not choose to continue renting in the future. The dynamic housing pathways 3 approach overcomes this problem by examining the housing situation over a period of time instead of only once at a specific point in time. The second gap concerns the lack of research into agency factors 4 . A lot of effort has been exerted on exploring the role of structural factors 4 in explaining differences in the housing situation of young people, including the changing housing market, the housing provision regime, the stability of the labour market, the institutional constraints, the intergenerational transfer, and other social-economic factors. For example, Boelhouwer (2020) studied the impact of the housing market on social inequalities and pointed out that the younger generation benefits less from mortgage interest tax relief because of changes in the Dutch housing mortgage policy since 2015. Maroto and Severson (2019) suggested that labour precarity influences young adults in Canada to enter the housing market. The findings of empirical research by Wang, Li, and Deng (2017) revealed that institutional factors, such as the household registration (hukou) system, still play a significant role in accessing social welfare (urban public housing) for skilled young workers in Beijing, China. Druta and Ronald (2017) explored the housing trajectories of young adults in the UK. They found that homeownership is an "ideal gift" from parents, which smoothens the housing pathway for young adults. The research of Lennartz and Helbrecht (2018) in Germany and Deng, Hoekstra, and Elsinga (2020) in China also show that family financial support becomes increasingly important in shaping youngsters' housing pathways. Xian and Forrest (2020) highlighted the need to focus on the impacts of social-economic factors, such as educational level and the specific local context on forming young people's housing choices. Despite these and other studies, relatively little attention has been paid to agency factors, such as the personal abilities of how young talents think and react to the structural factors that shape their housing pathways. Studies like Clapham (2005), Clapham et al. (2014), and Balampanidis (2020) have demonstrated that agency factors might play an important role in the formation of people's housing pathways. To fill in the gaps, i.e. the lack of research into the longitudinal housing situation and the exclusion of agency factors, the current study aims to obtain insights into the housing pathways and future migration plans of young talents and to explore differences in these housing pathways by considering both structural and agency factors. In this study, we conducted a qualitative retrospective study by exploring and analysing the past and present housing narratives of young talents in Shenzhen, China. During the past four decades, Shenzhen has transformed from a cluster of rural villages and townships of about 300,000 people into a metropolitan city with a population of more than 17 million, of which over 70% are migrants (Statistics Bureau of Shenzhen Municipality 2021). In addition, Shenzhen, as an international metropolitan city, has an average population age of 33 years. The proportion of talents reached 44.5% in 2018 (Liang 2020), which represents a typical gathering place for young talents. The diversity of the housing market in Shenzhen also provides a good arena for analysing the different housing situations of young talents (Li et al. 2021). Therefore, as a city with a large number of skilled young migrants, Shenzhen seems to serve as a good case to investigate the housing issues and migration plans of young talents. Specifically, we focus on addressing the following research questions: (1) What are the housing pathways of young talents in Shenzhen and how do they differ from each other? (2) What structural and agency factors are capable of explaining these differences? (3) What are the future migration plans of young talents following different types of housing pathways in Shenzhen? The housing pathways approach (Clapham 2005) and Bourdieu's theory of practice with concepts of "habitus", "field", and "capital" (Bourdieu 1984(Bourdieu , 1986 have been used as a research framework and theoretical basis for this study, respectively. The housing pathways approach allows us to study the housing practices of young talents over a period of time and to identify their different housing pathways. The housing pathways approach is defined by Clapham (2005) as a research framework used to frame thoughts, rather than a theory. While the approach emphasizes that both structural and agency factors influence housing outcomes, it does not reveal how these factors interact in doing so. Therefore, to explain what causes the differences in housing pathways of young talents in metropolitan cities, we turned to Bourdieu's theory of practice in which people are considered to be active individuals who use different "capital" to generate practice in a certain "field" according to their "habitus" (Bourdieu 1984(Bourdieu , 1986. Applied to the field of housing, a young talent's outcome of habitus, such as a need to buy a house, interacts with the acquired level of capital (economic, social, and cultural capital) to generate housing-related practices and outcomes towards the housing field, i.e. the owneroccupied sector. In summary, the housing pathways approach sheds light on the various patterns regarding the previous and current housing situations of young talents. Bourdieu's theory can be used as a solid theoretical framework to explain the differences between these housing patterns. Compared with previous research using the same theory (Hochstenbach and Boterman 2015), the current study is novel in introducing and examining the role of "habitus" in forming young talents' housing pathways. In addition, the current study links the housing pathways of young talents to their future migration plan, which extends the framework of the housing pathways approach. Housing Pathways Approach The concept of housing pathways is often used interchangeably with "housing career" and "housing trajectory" and is defined as patterns of interaction (practices) concerning house and home, over time and space (Clapham 2002). Rather than simply providing descriptions of the individual's or household's housing experiences, the housing pathways approach was developed by Clapham as a structure for analysis in the field of housing studies (Clapham 2002(Clapham , 2005Clapham et al. 2014). The housing pathways approach was adopted for this study as it provides a framework for housing research that is 1) dynamic, since it provides a deeper insight and understanding of the housing experience than just static location knowledge, 2) flexible, since it allows for the study of a shorter period of time 3) taking the individual (household) as the creative subject and 4) considering the impact of both structural and agency factors on housing outcomes. Figure 1 illustrates the analysis framework of this study by using the housing pathways approach. A housing pathway is linked to many other areas of life and runs alongside employment pathways (Clapham 2005). For example, the housing pathway is influenced by changes in household structure relating to marriage, the birth of children, or divorce. In addition, drawing from Giddens's structuration theory (Giddens 1984), the housing pathways approach highlights both structural and agency factors as influencing factors providing opportunities or imposing constraints on the individual's (household's) dynamic housing experience (Opit, Witten, and Kearns 2019). Along the housing pathway, individuals and households make housing choices among the opportunities open to them under the constraints (Clapham 2005). In recent literature, the housing pathways approach has been used as a tool to unfold various housing phenomena, such as residential stability (Meeus and De Decker 2015), homelessness (Fitzpatrick, Bramley, and Johnsen 2013), and housing of the elderly (Bates et al. 2020), young people (Hamzah and Zyed 2020;Hochstenbach and Boterman 2015), and highly-skilled migrants (Balampanidis 2020;Eskelä 2018). Habitus, Capital, and Field In the course of research attempting to interpret social phenomena and mechanisms such as class inequality and practical logic of everyday life, the French sociologist Bourdieu developed a set of theories and concepts (Power 1999). Bourdieu's theory emphasizes the interaction of structure and agency. His key theoretical claims are based on the three core concepts of "habitus", "capital", and "field" (Bourdieu 1984). The concept of habitus forms the core of the theoretical framework of Bourdieu. In the words of Bourdieu: The conditionings associated with a particular class of conditions of existence produce habitus, systems of durable, transposable dispositions, structured structures predisposed to function as structuring structures . . . . . . (Bourdieu 1990, 53). . . . . . . it is a socialized body, a structured body, a body which has incorporated the immanent structures of a world or of a particular sector of that world -a field -and which structures the perception of that world as well as action in that world (Bourdieu 1998, 81). In other words, habitus refers to a durable, transposable system of dispositions that are ways of being, observing, acting, and thinking, a system of scheme or schema or structures of perceptions, concepts, actions, distinctions, and principles for generating and organizing practices and representations (Bourdieu 1990, 53). Durable means that habitus is a relatively stable disposition acquired from past life experiences, and transposable means that habitus also adapts to changes in experience and environment. For example, childhood housing experiences shape habitus that affects present housing choices, and present housing experiences shape future housing habitus. Habitus is also "a way of describing the embodiment of social structures and history in individuals -it is a set of dispositions, internal to the individual, that both reflects external social structures and shapes how the individual perceives the world and acts in it" (Power 1999). From the above descriptions, it can be seen that habitus is an abstract, complex and multifaceted concept. In our view, habitus can be considered as a mediator between social structures, e.g. family and cultural norms, that influence the individual and the individual's perception of the world and action, which are the outcome of habitus (see Figure 2). Since habitus is a set of dispositions, which are difficult to describe, measure, and analyse, we have focused on the outcome of habitus in the current study. Specifically, the outcome of habitus consists of at least two components when analysing people's housing pathways. Firstly, the individuals' perception of the world can be operationalized into the attitude towards different housing tenure, such as private renting and owning. The attitude towards housing tenure has been proven to be one the of most important determinants of people's tenure choice (Lennartz 2013;. Secondly, individuals' actions can be operationalized into people's strategies for their housing choice. Such strategy can be influenced by habitus and attitudes towards different types of housing. For example, people who are from higher-income families or who always lived in an owner-occupied dwelling might have a habitus that shapes a relatively positive attitude towards an owner-occupied dwelling and they may use strategies, such as saving up or getting a mortgage, to buy a house. According to Bourdieu (1986), "capital is accumulated labour (in its materialized form or its incorporated, embodied form) which, when appropriated on a private, i.e. exclusive, basis by agents or groups of agents, enables them to appropriate social energy in the form of reified or living labour". Capital often presents itself in three forms: economic, cultural, and social capital. Economic capital can be directly and immediately converted into money and may be institutionalized as property rights. Cultural capital, in some situations, can be transformed into economic capital and can be institutionalized as educational qualifications. Social capital, accrued from social networks, can, in some cases, be transformed into economic capital and can be institutionalized in the form of titles of nobility (Bourdieu 1986). In addition to the traditional definition of capital above, other forms of capital also exist. In the housing field, factors like "holding a Shenzhen hukou", "being a young talent", "the willingness to enter the informal housing market" and "being employed by certain companies" are all resources that people can use to access specific housing opportunities. For example, young talents who are employed by a state-owned company have the opportunity to apply for talented rental housing. Drawing on the work of Boterman (2012) and Hochstenbach and Boterman (2015), examples of the economic, cultural, social capital, and other forms of capital -applied to the housing market -are shown in Table 1. The habitus and capital held by the actor must have a field to deploy. A field is a bounded social space that has its own principles. Bourdieu likens a field to a game that has specific rules and stakes, which includes players who occupy dominant positions according to their habitus and capital (Bourdieu 1984). The fields analysed by Bourdieu include the field of law, the field of art, the field of science, and the field of cultural production, to name just a few. In the current paper, the field refers to the field of housing. From the above analysis of the concepts of habitus, capital, and field, it is clear that these three concepts incorporate elements of both agency and structure. Bourdieu claimed that practice -what one does in everyday life -is the result of the relationship between an actor's habitus, different forms of capital, and the field of the action (Power 1999). The Shenzhen Context Whereas we explore the housing pathways of young talents in Shenzhen, some information about the Shenzhen housing market situation is needed. Since being designated as one of China's five Special Economic Zones in the 1980s, Shenzhen has experienced phenomenal population growth (from 0.31 million in 1979 to 17.63 million in 2020), which has contributed to a booming housing market. The average annual house price growth in Shenzhen was 11% over the last decade and its average house price is now the world's fifth most expensive (CBRE 2020). In contrast, the presence of poorly built, cheap, and available rental housing in urban villages has kept the average rent in Shenzhen relatively low, which in turn has contributed significantly to the development of the private rental sector (PRS) (Li et al. 2021;Zhou 2019). According to a recent report, 77% of the population lives in the private rental sector in Shenzhen (China Construction News 2022). The details of the different tenure types in Shenzhen are depicted in Table 2. There are three primary ways of entering the owner-occupied sector. First, to purchase commercial housing. Commercial housing used to be supplied without restrictions. However, since the Purchase Restriction Policy in 2010, entry access is restricted not only by economic capital but also by institutional restrictions including local household registration (hukou) and the years of participating in local social insurance (Gong and MacLachlan 2020). It is important to point out that the Shenzhen Talent Introduction Policy allows highly educated and highly skilled migrants to readily obtain a Shenzhen hukou (HRSSASM 2016). Second, government policy-supported housing can be purchased. Since 2010, the government has mainly supplied Affordable Commercial Residential Housing (ACRH), which refers to housing with limited size, sales price, and transferability period. There are restrictions on buying ACRH, based on the number of years of having obtained a formal Shenzhen hukou and paying local social insurance (Gong and MacLachlan 2020; Li and Shamsuddin 2022). As a result, younger people may rarely enter homeownership through ACRH. The third way is to enter homeownership in an informal way, such as purchasing a small property rights housing (SPRH). 5 By drastically cutting development costs, including land transfer fees and various taxes, the SPRH is priced at 40-60% below regular commercial house prices (He et al. 2019;Wang, Sun, and Li 2014). Therefore, those who have insufficient economic capital and are willing to take risks might choose to buy SPRH The private rented sector can be divided into three sub-sectors according to the variation in housing conditions, neighbourhood environment, and landlord service: urban village rental housing, commercial rental housing, and Long-term Rental Apartment (LTRA) (Li et al. (2021). All housing in these three sub-sectors is supplied by the market and there are no institutional restrictions on who can rent them. Urban village rental housing is informal housing built on collective land. Most urban villages are densely populated and have inadequate lighting and poor infrastructure. The housing conditions can be described as overcrowded, in the lack of basic facilities such as indoor toilets and kitchens (Wu 2016). Commercial rental housing is generally located in a gated community that provides a host of social, commercial, and recreational services which is rented out by the private owners of commercial housing and condominiums (Li et al. 2021;Wu 2012). LTRAs are defined as dwellings rented out and managed by professional companies with a tenancy period that is often longer than one year. LTRAs are usually decorated, furnished, equipped with appliances, ready for immediate move-in, with professional livein managers, and with rents typically 15% to 30% higher than comparable spaces nearby (Li et al. 2021;Zheng 2018). There are two main types of social rental housing in Shenzhen after 2010. The first one is Public Rental Housing (PRH), which refers to housing provided by the government with a limited size and rental price. The PRH is offered to low-and middle-income households or single residents with housing difficulties (HCBSM 2014). The rent of PRH is usually 30% of the market rent. The initial requirements for applying for PRH are to obtain a Shenzhen hukou and to participate in social insurance for 3 years. Those who meet the criteria need to apply first and then they will be put on a waiting list for suitable housing (Gong and MacLachlan 2020). The second one is Talented rental housing (TRH), which is provided for highly educated and skilled workers. The rent for TRH is usually 60% of the market rent. The current supply pattern of TRH requires the key enterprises 6 to apply to the government first. Then the government scrutinizes the application and supplies the TRH to qualified enterprises. The qualified enterprises then set their own allocation criteria accordingly to rent the TRH to their talents 7 (Gong and MacLachlan 2020; Wang and Pan 2019). Thus, the access of young talents to talent housing is directly related to their job opportunities. There are also other forms of residential housing, such as housing provided by the employer. The employer's housing is either self-built or rented from the market and then supplied to the employees. This is mainly the case in factories and service industries that provide shared accommodation for low-income migrants (Huang and Tao 2015). Or, stateowned enterprises such as hospitals, universities, and other institutions provide specialized housing for talents to attract them (Huang and Tao 2015). The latter applies to the young talents included in the current study. Thus, access to company-provided housing is also mainly related to job opportunities. Methodology The data of the current study draws from online semi-structured interviews with 18 young talents 1 , sometimes accompanied by their family members (if have), who work and live in Shenzhen. The interviews were conducted in November and December 2021 using social apps such as WeChat video and Tencent Meeting. University students were not included because all students can live in the special dormitories that Chinese universities provide. A combination method of snowball sampling and purposive sampling was adopted to reach the potential interviewees. The interviewees were recruited via personal networks and recommendations from interviewees. Different types of suitable interviewees were purposively selected to ensure a good mix in the resulting sample, based on gender, place of birth, educational background, marital status, occupation, and length of residence in Shenzhen. The characteristics of all respondents are shown in Table 3. Our interview was a retrospective investigation of young talents' housing pathways after entering employment in Shenzhen and lasted between 30 minutes and 2 hours. The interviewees were asked to reflect in detail on their housing pathways, from the first dwelling after entering employment in Shenzhen to the current dwelling at the time of the interview. More specifically, respondents were asked to provide information on: (1) detailed characteristics of each dwelling including location, tenure type, size, cost (rent or monthly payment), number of residents, travel convenience, neighbourhood amenities, etc.; (2) factors that influenced their choice for each dwelling; (3) perceptions/attitudes concerning each dwelling; (4) reasons for each move; (5) changes in life events that were related to housing, e.g. changing jobs, falling in love, and having children; (6) housing visions/plans for the next 5-10 years; (7) whether they intended to continue to work and live in Shenzhen in the next 5-10 years. All the interviews were audio-recorded with the respondents' oral consent and transcribed in Chinese. The names used in the current paper are pseudonyms and any personal information disclosed has been removed. The Atlas. ti program was used to code the transcripts using both inductive and deductive coding techniques (Fereday and Muir-Cochrane 2006). The coding system was first formulated according to the theoretical frameworks of both the housing pathways approach and the theory of Bourdieu and then developed to reflect the themes that emerged during the interviews. For example, we coded various housing situations according to the scheme provided in Table 2 and coded economic factors that influenced housing choices as economic capital. A coding scheme of various forms of habitus was developed in the course of the interviews. Finally, the built-in Networks function of Atlas. ti was used to construct relationships between the codes to identify, analyse and classify different housing pathways. For instance, the connections between the various housing situations (fields) of a respondent were constructed as a network resulting in the housing pathway of the respondent (see Figure 3). Adding codes to the type of capital, the various forms of habitus, and specific life events to this network further contributed to disentangling the various factors leading to differences in housing pathways. All codes and networks were written in English and quotations were translated from Chinese to English. Findings A summary of the 18 housing pathways is depicted in Figure 3. Important housing and other basic features are included in the Figure. Building on the work of Clapham et al. (2014) and Hochstenbach and Boterman (2015), we ultimately identified four different types of housing pathways based on the analysis of the codes and networks outlined in the current interviews. The four are housing pathways "staying at parents' home (SPH)", "private renting to owning in Shenzhen (PRTO)", "(private renting to) talented renting (PRTTR)", and "progressive private renting (PPR)". These four housing pathways are distinct from each other mainly in the past and present predominant tenure (housing fields) occupied by young talents as well as the number and type of moves. A summary of the various housing pathways is shown in Table 4. We also found that the formation of different housing pathways can be explained by the interaction of the various habitus and the various economic, social, and cultural capital (see Table 4). The next parts describe how habitus, capital, and the housing fields interact to shape different housing pathways. Housing Field, Habitus, and Capital Four respondents, who are Shenzhen locals, are following the SPH pathway, due to the fact that their parents are homeowners in Shenzhen. This group of respondents chooses to continue living with their parents after entering the labour market. For example, Linda has been working in Shenzhen for four years and has always been living in her parents' house. The possession of a Shenzhen hukou is the basic capital for respondents who follow this pathway. The capital allows them to receive many local welfare benefits such as the direct eligibility to buy a house and the possibility of public rental housing. In the case of Linda, she can buy a house when she wishes. We also found that respondents who follow this pathway have some economic capital of their own, primarily derived from their parents. Because of the intimate relationship between parents and children in China, it is common for children to still live with their parents even after they have started working (Li and Shin 2013;Or 2017). Respondents tend to obtain satisfactory housing conditions through mutual support with their parents. On the other hand, parents get the accompaniment and care of their children. These respondents have always lived in an owner-occupied dwelling, and, consequently, their habitus makes them more positive towards owner-occupied housing. Living with their parents also helped them to save money and get financial support from their parents to buy a house. In addition, some respondents were influenced by their parents' experience of successfully purchasing a house in Shenzhen and developed habitus that shaped their beliefs that homeownership is a good investment and that it is easy to afford a house by themselves in the long run. Respondents might use the economic capital that they can As locals, the cultural capital of knowing the housing market and system in Shenzhen is another typical feature of young talents who follow this housing pathway. For example, some respondents know how to find both satisfying and cheap housing, as well as which policies to use to obtain maximum benefits. I plan to buy a second-hand house in Longgang which is relatively cheap. [. . .. . .]I plan to buy it after I get married. Because as far as I know, a married couple can get a loan of 900,000 yuan, almost twice that of a single. So it will be more cost-effective to buy after marriage. (Jo, male, 26, bachelor, Shenzhen local) Frequency of Moving and Future Migration Young talents who follow the SPH pathways have seldom moved. The main reasons for moving in the future are getting married or having children. As for future migration plans, all respondents following this housing pathway stated that they will stay in Shenzhen. Housing Field, Habitus, and Capital On the PRTO pathway, respondents first enter the private rental market and (will) finally acquire homeownership. The respondents more often lived in commercial rental housing or LTRA instead of urban village rental housing. Several respondents mentioned that they could not accept the poor conditions of urban village rental housing. Young talents who follow the PRTO pathway either have usable economic capital or social capital or Shenzhen hukou. For example, the financial support of parents, which is economic capital, helps some respondents to buy a house. Lea and her husband's housing experience is an example of the PRTO pathway. They are both migrants who chose to work in Shenzhen after graduation. At first, they lived in an old commercial rental house. A year later, they made up their minds to buy a house. My husband and I are both migrants. I believe that if we want to settle down in Shenzhen, we must own a house there. I wouldn't get married if we didn't own a house. [. . .. . .] My husband's parents paid the down payment and we decorated the house with the money we saved from work. It's not a big house, but it's enough for us to live in. (Lea,female,29,bachelor,migrant) This example illustrates how the habitus shaped the belief about owning a house and how it influenced the way in which they view housing. It also shows how they employed the economic capital that is available to them, i.e. their savings and their parents' financial support, to achieve their goal, namely buying a house. The case of Sherry and her husband demonstrates how social capital and the readiness to enter the "grey market" of informal housing were used to obtain homeownership in Shenzhen. Sherry and her husband, who are both migrants, started working together in a community hospital in Shenzhen after graduation. At first, they chose to live in the doctors' collective accommodation of the hospital to save money. In their third year in Shenzhen, they bought their first house, a small property rights housing (SPRH). Two years later, they bought a second house, which is a normal commercial house. They used the word "anxiety" to describe their feelings in Shenzhen before they bought their first house. The atmosphere that has been created in our social networks before we went to Shenzhen and when we were already there was that the house prices in Shenzhen are constantly rising. We could apply for public rental housing, however, I got to know that there was a very long waiting list and those houses were in remote locations. We believed that we had to buy a house right away or it would become increasingly difficult to afford. We were very anxious. (Sherry,female,32,master,migrant) Their determination to buy a house in Shenzhen was reinforced by the signals they received from social networks. Their utilization of information and financial resources from social networks and their readiness to accept informal housing enabled them to enter homeownership successfully. Regarding their buying experience, they said: From chatting with some of our colleagues, we learned that it would be cheaper to buy a SPRH. Considering the limitation of our financial resources, we chose to take the risk of buying a SPRH. With our own savings (36%) and the money borrowed from relatives and friends (64%), we had just enough money to pay for a SPRH. (Sherry) In Sherry's case, the interaction of habitus and social capital can be summarized as follows: first, the anxious atmosphere in their social networks reinforced their belief to quickly buy a house. Then the information and financial resources received from social networks and the readiness to accept informal housing enabled them to buy. Frequency of Moving and Future Migration The results of the interviews showed that most respondents on this pathway rarely move, partly because they want to save money by reducing the cost of moving, and partly because they entered homeownership and no longer need to move. The main reasons to move are getting married or childbirth. Respondents stated that they could live alone in a low-quality dwelling but wanted to provide better living conditions for their partners and children. In China, educational resources are usually better for home-owners than for renters (Feng and Lu 2013). The future education of children is therefore one of the reasons why respondents move into homeownership. As for the future migration plans, those who were already homeowners in Shenzhen explicitly stated that they would stay in Shenzhen, as they had already settled down. However, the relationship between the wish to stay in Shenzhen and the wish to buy a house is not one-directional. Guffin (male,31,PhD,migrant) and David (male,25,master,migrant), for example, want to stay in Shenzhen because of the very nature of their work. Both of them are prepared to buy a house in Shenzhen. However, the latest Shenzhen 2020 policy requires migrants to have a Shenzhen hukou and to have paid social insurance for at least 3 years before they can purchase commercial housing. Housing Field, Habitus, and Capital Young talents following the PRTTR pathway enter talented rental housing or companyprovided rental housing (hereafter referred to together as talented housing) directly or after experiencing a period of private rental housing. Consistent with the discussion in Section 3 (the Shenzhen context), the results show that many respondents are employed in the public sector such as government departments, state-owned enterprises, and universities. Otherwise, they are employed by key private enterprises such as high-tech enterprises. The opportunity to live in talented housing was identified as a special kind of capital. In general, as long as young talents do not change their jobs, they can stay in talented housing for 6 years at a lower rent (Wang and Pan 2019). Therefore, the tenancy period for talented housing is relatively stable. According to the results of the interviews, the outcome of the habitus of the respondents on this pathway can be summarized as having a positive attitude towards talented renting and being opportunity-driven. They take advantage of cultural capital and/or social capital and try to obtain satisfying work to follow the PRTTR through the housing opportunities offered by their employers. Cultural capital in this regard refers primarily to the education received by young talents (Bourdieu 1986). The case of Yanni is an example of a highly educated individual taking the PRTTR pathway. Yanni completed her PhD at a leading university abroad. She joined a university in Shenzhen through an introduction scheme for talents. She first rented in the private rental market as there was a limited number of talented housing units at the university that required queuing. After six month-waiting, she moved into the university arranged talented housing. The rent for talented housing is only 60% of the market rent. The opportunity to rent such housing is very satisfying. (Yanni,female,32,PhD,migrant) The case of Lince is a good demonstration of how social capital and being employed by a certain company can be used to follow the PRTTR pathway. During his previous years of working in Shenzhen, he moved through the private rental market. Then after he got his present job, he and his family rented talented housing that was supplied by his present company. The information he and his company obtained from social networks (social capital), made it possible for him to enter the PRTTR pathway. Speaking about how he managed to live in the talented housing, he said: My current company specializes in consulting services for overseas talents, so we are very familiar with the talent policy in Shenzhen. [. . . . . .] Probably in late 2018, the company got information about the pending allocation of talented housing by the Shenzhen government in advance and immediately applied to the government, which ended up with about 11 sets of talented housing. As a senior employee of the company, I also applied to the company for talented housing immediately and was allowed to rent. (Lince,male,33,master,migrant) Frequency of Moving and Future Migration Some respondents almost directly entered the talented housing whereas others had to move repeatedly before they got this opportunity. As mentioned above, a change of job is the most important reason for them to move in the future, but marriage or childbirth are also reasons. Our findings showed that some respondents clearly expressed a preference to stay in Shenzhen, while others said that they were uncertain about their future migration plans. The main reasons for the uncertainty are related to work and housing affordability. As Bamboo mentioned, Many people come to Shenzhen to earn money. After working in Shenzhen for several years, they may go back to their home city or the capital city of their home province to settle down. They won't and can't stay in Shenzhen for a long time. At least I think this is the case for most of them. [. . . . . .] I come from Changsha, Hunan Province. [. . . . . .] An important point is that houses in Changsha are much more cost-effective (than in Shenzhen). I would definitely be willing to move there if it was possible to find a satisfying job (in Changsha). Shenzhen is a wonderful city, but if you don't have a place to live, your well-being will be reduced. [. . . . . .] But as I just started to work a year ago, I'm not too sure yet. (Bamboo,male,25,master,migrant) Housing Field, Habitus, and Capital Respondents who followed the PPR pathway are all migrants and frequently moved within the private rental sector. In general, the housing quality generally becomes better after each move to a new dwelling or stays the same. Most of them rented a house in urban villages in the early stage of their stay in Shenzhen. Later they moved into commercial rental housing and LTRA, if available. Some respondents own a house in the peripheral cities of Shenzhen or their hometown due to marriage or childbirth. However, they also rent in Shenzhen due to the inconvenience of the long commuting time to their other house. Those who follow the PPR pathway lack all kinds of capital, especially economic capital, compared to young talents who follow other pathways. This group of respondents has a mixed habitus. Some respondents had deep-rooted beliefs about homeownership shaped by their habitus due to social norms or the family environment in which they were raised. Their strategy was to buy a house in a city with lower prices in the periphery of Shenzhen because of the lack of economic capital. Other respondents' habitus resulted in a positive attitude towards private renting, which was seen as cheap and flexible. These respondents said they did not want to be "kidnapped" by their mortgage. The interaction of their habitus and lack of economic capital makes them stay trapped in the repetitive private rental sector. This pathway is well illustrated by Paul, a start-up entrepreneur who has been living in Shenzhen for almost 7 years. Upon graduating from university, he chose to move to work in Shenzhen. By the time of the interview, he has rented a total of five different houses, of which the first four were in urban villages. The first house was only 10 m 2 with one small single room and a tiny bathroom, no sunlight, damp and noisy, where he lived for three years. Later, he moved to a second similar house of 13 m 2 due to his change of workplace. After a year, he switched to the third house of about 18 m 2 for another year because he thought the second house was too small, too dark, and he couldn't stand the noisy surroundings anymore. Because he fell in love and planned to live with his girlfriend, which required more space for two people, they moved into the fourth one-bedroom house of about 25 m 2 . Over a year later, their child was born, and they moved into a relatively remote, old rental commercial housing of 80 m 2 . He said: I would be fine with renting all the time if the rent was always affordable. [. . .. . .] If you buy a house in Shenzhen, not only yourself but your parents and the whole family (financial resources) are kidnapped by the house. This is not the life that I want, and I do not like too much pressure. [. . .. . .] I'm still wandering about my future housing. (Paul,male,30,technician,migrant) It can be seen that Paul had a chequered housing experience in Shenzhen. This is partly due to the constraints of economic capital, but also due to his own habitus and choices, such as his openness and acceptance of continuous private renting. In addition, our research found that the ability of respondents to buy a house is influenced by the volume of their economic capital, but the conditions of rental housing are more determined by their own preferences. For example, Jann clearly expresses an economic motive. Jann still lives in an urban village rental housing. Regarding buying a house in Shenzhen, he comments: However, Lehi has decided otherwise. After experiencing several bad rental rooms in urban villages, she is now living in an expensive rental apartment. When asked why she chose the current more expensive rental apartment, I moved to an expensive apartment, not because of the increase in my salary or because I became rich but because I was unhappy with my job. So I wanted to live in a better apartment to increase my happiness. (Lehi,female,32,bachelor,migrant) Frequency of Moving and Future Migration Compared to other pathways, our results showed that respondents on this pathway moved more frequently. The reasons for moving vary and could be either because of a wish to move or because they had to move. The main reasons for active moves are changes in the workplace and the desire to improve living conditions. Important reasons for forced moves are the landlord selling or redecorating the house and arbitrary rent increases. Regarding future migration plans, some respondents on this pathway clearly indicated their intention to leave Shenzhen in the future. Others were not sure about their future migration plans. When asked about the reasons for leaving Shenzhen, the issue of housing affordability was repeatedly mentioned, with respondents stating the following: Discussion The main aim of the current study is to identify and understand the housing pathways of young talents in Shenzhen, as well as the structural and agent factors that underlie these housing pathways. Using the housing pathways approach and Bourdieu's theory of practice, three research questions were addressed. Housing Pathways of Young Talents in Shenzhen Regarding the first research question, four different types of housing pathways for young talents were identified: staying at parents' home (SPH), private renting to owning in Shenzhen (PRTO), (private renting to) talented renting (PRTTR), and progressive private renting (PPR). These four housing pathways are distinct from each other mainly in the past and present predominant tenure (housing fields) occupied by young talents as well as the number and type of moves (see Table 4). These results are consistent with the conclusion of Ford, Rugg, and Burrows (2002) that the degree to which pathways are tenure-specific varies. The first housing pathway (SPH)resembles the "stay at home to own" housing pathways identified by Clapham et al. (2014). However, unlike Clapham et al. (2014), i.e. young people perceive this pathway as a shameful, ad hoc strategy to save for future homeownership, we found that our respondents did not feel ashamed of living with their parents. Instead, both young talents and their parents believe that living together is a better way to take care of each other. This finding can easily be explained by the difference in cultural background, where living with parents is a form of filial respect in China (Yang 2021). The second housing pathway (PRTO) is in line with the traditional linear housing pathway of "rent to own" found in studies by Ford, Rugg, and Burrows (2002), Hamzah and Zyed (2020), etc. The third housing pathway (PRTTR) might be more specific for Shenzhen as it has gradually emerged since the enactment of the Talent Housing Project in 2010. The pathway is different from other studies that refer to young people living in social housing (Hochstenbach and Boterman 2015) or council housing (Clapham et al. 2014). Young talents who follow the PRTTR pathway are not low-income or those who have experienced a long waiting-list. Instead, they meet the conditions to be employed in one of the key enterprises that are supported by the Shenzhen government. The fourth housing pathway (PPR) seems comparable to the "progressive chaotic housing pathway" as outlined by Hochstenbach and Boterman (2015). Young talents in our study who followed this pathway remained in private renting, although most of them gradually improved their housing conditions. Note that our study did not provide indications for chaotic or homeless pathways as identified in previous studies (Clapham et al. 2014), probably because our study was conducted with young talents rather than young people in general. Young talents may have higher incomes or better access to resources than young people in general. What Structural and Agency Factors Explain Differences in Housing Pathways? Concerning the second research question, our results show that the formation of different housing pathways can be explained by the interaction of habitus and its outcome and various types of economic, social, cultural, and other types of capital within the housing fields (also see Table 4). The results are generally in agreement with Hochstenbach and Boterman (2015) and Boterman (2012), who distinguished three different housing pathways, i.e. linear, progressive chaotic, and reproductive chaotic, based on the combination of different search behaviour and different types of capital. The current study explains the role of habitus, which was mentioned only briefly by Hochstenbach and Boterman (2015), in more detail. However, we found that the concept of habitus is abstract, complicated, and difficult to describe. For this reason, we operationalized habitus into its outcomes, that is, the attitude towards housing tenure and the strategy of housing choice. We found that habitus shaped deep-rooted beliefs about homeownership in the first and second types of housing pathways (PRTO and SPH). This is in line with the previous research by Rowlands and Gurney (2000), who found that young people in the UK perceived homeownership as a social norm in British society and the solution to basic housing needs. In China, the meaning given to homeownership is even greater, as it is also considered to be a symbol of status and a prerequisite for marriage (Hu and Wang 2020). Besides from that, these respondents also believed that buying a house is a good investment, based on their parents' prior successful home-buying experiences. Furthermore, in both pathways, economic capital is used to acquire housing, such as borrowing money to buy a house. The two pathways differ with regard to the cultural and social capital. In SPH pathways, the cultural capital is used, which is knowledge of the local housing market. The dominant capital in PRTO pathways is the social capital and the readiness to enter the informal housing market. These differences might be explained by the fact that the respondents following the SPH pathway are Shenzhen locals whereas those following the PRTO pathway are mainly migrants. Young talents following the third housing pathway (PRTTR) rely more on cultural capital (educational attainment) and social capital (advanced information from social networks) and the capital of being employed by certain companies to access talented rental housing. They are characterized as having a positive attitude towards talented renting and being opportunity driven. They may select particular companies because of their housing opportunities, or they may be offered housing opportunities by selecting particular companies. The last housing pathway (PPR) is formed by a deep-rooted belief in homeownership, a positive attitude towards private renting and the lack of all kinds of capital, especially the economic capital. The combination of these beliefs and the lack of economic capital led to differences in housing outcomes. All respondents following this pathway stayed in private rentals in Shenzhen, but some of them also bought a (relatively cheap) house in the surrounding cities of Shenzhen. As for the reasons reported by the respondents for the lack of economic capital, it was partly because of the non-advantageous economic background of their parents. Along with the findings of Deng, Hoekstra, and Elsinga (2020) and Or (2017), our results highlight the important role of intergenerational transmission in young people's housing experiences. The Future Migration Plans of Young Talents As for the final research questions, we found a clear association between future migration plans and the various types of housing pathways. This confirms the conclusion of previous studies that the housing career influences migration behaviour (Aner 2016;Cui, Geertman, and Hooimeijer 2015;Dainov and Sauka 2010;Teixeira and Drolet 2018). Those who follow the SPH and PRTO housing pathway indicated that they will stay in Shenzhen for the next 5-10 years. We found that staying in Shenzhen is related to homeownership. This is inconsistent with the findings of Eskelä (2018), who stated that homeownership is not an influencing factor for skilled migrants to stay in Helsinki. The difference might be explained by the Chinese concept of putting down roots. In China, you don't have a home without a house (Xie and Chen 2018). In addition, we also found that some respondents living in stable rental housing had plans to stay in Shenzhen. This suggests that the ongoing Talent Housing Project may have some positive effects on talent retention. In contrast, some of the respondents following the PPR pathway have articulated their intention to leave Shenzhen in the future. Conclusion The present paper offers a refined comprehension of the housing and migration behaviour of young talents in Shenzhen. Based on the analysis of 18 semi-structured interviews, four distinct housing pathways were identified: 1) staying at parents' home, 2) private renting to owning, 3) talented renting, and 4) progressive private renting. Different housing pathways can be accounted for by the way diverse habitus and its outcome interact with various forms of economic, social, cultural, and other types of capital in the housing field. A clear association between future migration plans and types of housing pathways was found. For example, young talents who have undergone unstable housing situations are more likely to leave Shenzhen. The present paper provides a novel perspective on the application of the housing pathways approach and Bourdieu's theory of practice. We found housing pathways approach to be particularly useful for investigating peoples' housing history/trajectory because it provides a framework of thinking to guide the design of interview questions as well as the collection and analysis of data. Besides, we believe that the graphic housing pathways approach gives the researcher a clear perception and an intuitive understanding of the housing history/trajectory experienced by people. However, the housing pathways approach is not a theory but a research framework. We then used Bourdieu's three concepts of "habitus", "capital", and "field" as a further development to the housing pathways approach to explain how different housing pathways are shaped by the interaction of different structural and agency factors. The use of Bourdieu's theory helps to improve the depth of the research. Moreover, we present a more operationalized use of the concept of "habitus" that has only been mentioned by other authors but has not been used much in housing studies. Based on the definition of habitus, we found that habitus produces certain outcomes that can be summarized and described. We often cannot capture the specific habitus in housing analysis but we can summarize and describe the different outcomes of different habitus. We believe that this is an alternative way of analysing habitus. Although the study provides detailed information on the housing pathways of young talents in Shenzhen based on solid theoretical approaches, a number of important limitations need to be considered. First, the researchers used a convenience sample, meaning that the interviewees were recruited via personal networks and recommendations from interviewees. There might be a possible selection bias in the final sample. For example, the housing pathways of all our interviewees were eventually progressive. Young talents with regressive or chaotic housing pathways may have been out of our reach, may have refused participation, or may have left Shenzhen. Further research might investigate the housing pathways of young talents in Shenzhen using a more representative sample and might find additional types of housing pathways, among which are regressive and chaotic pathways as well. Second, with a small sample size of 18 interviewees in Shenzhen, caution must be applied, as the findings might not be representative of Shenzhen and might not be transferable to other cities and countries. Nevertheless, the theory and approach adopted in this paper can be employed by scholars studying this city or other cities in the future. The third limitation concerns the qualitative nature of the current study. Our results are more detailed in information and exploratory. However, a large quantitative study of the housing pathways of young talents is needed to obtain more accurate results regarding the variety of housing pathways and the share of young talents following each housing pathway. The findings of the current study have a number of important implications for policymakers in Shenzhen. Consistent with the implications statement in the introduction section, the policy implications will be discussed in two aspects: from the viewpoint of the city and that of its residents. The first aspect concerns the implication of how the city might attract and retain young talents. The findings of our research suggest that homeownership is a critical factor for young talents to stay in Shenzhen. It is not realistic to guarantee homeownership to all young talents in a city like Shenzhen with high house prices. However, reducing the barriers to homeownership for young talents, such as the hukou and social insurance requirements may help in attracting and retaining young talents. In addition, further improvements to the process and regulations for the purchase of Affordable Commercial Residential Housing (ACRH, Anjuxing Shangpinfang in Chinese) would also be beneficial in increasing the homeownership rate of young talents. We also found that young talents following the talented housing pathway are less likely to leave Shenzhen in the future. Increasing the supply of talented rental housing is therefore strongly recommended. The second aspect of implications concerns the improvement of the housing situation of young talents in Shenzhen. Our results show that young talents following the fourth housing pathway (progressive private renting) experienced frequent moves in the private rental sector, mainly in urban village rental housing. Their housing problems are well worth being looked at. Urban village rental housing is an important part of the initial housing transition for young talents due to its low rent. Policymakers could work on improving the living environment of rental housing in urban villages, such as the improvement of lighting in rental dwellings and the infrastructure in urban villages. Notes 1. There is no universal definition of talents. In academic terms, the definition of talent is often defined by educational attainment, vocational skills, and the creativity of the work. Young talents in this study are defined as people aged between 20 and 35, who have a bachelor's degree or above, or have a national vocational qualification certificate, or are professionals or managers of the companies. 2. It is recognized and a common practice to classify China's mainland cities into "tiers". According to the National Bureau of Statistics, four first-tier cities are Beijing, Shanghai, Guangzhou, and Shenzhen. There are 31 second-tier cities, which are mostly provincial capital cities (e.g. Wuhan) or sub-provincial cities (e.g. Qingdao). 3. The concept of housing pathway is often used interchangeably with housing career and housing trajectory. Simply put, it is the housing routes(outcomes) that an individual or a household takes over time. 4. According to Clapham (2012), structural factors refer to global, national, social and institutional related factors that can constrain or provide opportunities on individual decisionmaking such as housing policies, housing market, and social-demographic variables. In contrast, the central subject of the agency factor is the individual. Agency factors mean the ability and power of individual to take action to make things happen such as personal, attitudes, preferences and strategies. 5. Small property rights housing (SPRH) means an informal housing market taking place in the urban villages at the suburban zones of major cities (Liu, Wong, and Liu 2012). SPRH is built on collective land and Chinese law does not allow real estate development on collective land. Its legal validity is highly controversial and the purchase and sale of it in the housing market is not protected by law. In sum, SPRH is informal and characterized by a lack of public services, instability of tenure, and the violation of construction regulations. 6. Key enterprises have different requirements for each district in Shenzhen, for example, Luohu District 2021 recognizes 351 leading enterprises and recommended companies in the industry in the district as key enterprises that can apply for talent housing. 7. A few very outstanding talents and high-level talents designated by the Shenzhen government can apply for special talented housing in their personal names. As the percentage of such cases is low and it is difficult for the young talents we studied to be recognized as above talents, we will not mention it in detail. 8. In order to attract and retain talents to enhance urban international competitiveness and promote sustainable and healthy economic and social development, the Shenzhen government has included talents in the scope of housing security since 2010 with the launch of the Talent Housing Project (Rencai Anju Gongcheng/Banfa in Chinese). Talents are those with a bachelor's degree or above, or a national technician qualification of grade 2 or above, and those stipulated to be included in the shortage catalogue. TH is planned to be available for rent or sale, but as of the time of the survey (13 May 2022), what the authors have learned that TH in Shenzhen is still mainly for rent. Policies and regulations related to the sale of TH are being formulated. Therefore, TH in this study refers to Talented rental housing only. 9. According to the results of the seventh national census in Shenzhen 2021, the proportions of locals and non-locals are 71% and 29% respectively. Disclosure statement No potential conflict of interest was reported by the authors. Funding The work was supported by the China Scholarship Council [No. 202107720049]	v3-fos-license
2018-09-24T15:18:28.442Z	2018-09-21T00:00:00.000	52312463	{ "extfieldsofstudy": [ "Medicine", "Biology" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0204241&type=printable", "pdf_hash": "45333f707010dbbbed57970271cc0171eb04dbb2", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45349", "s2fieldsofstudy": [ "Biology", "Agricultural and Food Sciences" ], "sha1": "45333f707010dbbbed57970271cc0171eb04dbb2", "year": 2018 }	pes2o/s2orc	MYB1 transcription factor is a candidate responsible for red root skin in radish (Raphanus sativus L.) Root skin color is one of the economically important traits in radish (Raphanus sativus), and the pigmentation in red skin varieties is largely attributable to anthocyanin accumulation. Pelargonidin was found as a major anthocyanin pigment accumulated in the sub-epidermal layer of red radish roots. In the 20 F2 population generated from the F1 with red root skins, root skins with red and white colors segregated in a 3:1 ratio. Additionally, a test cross between a red F3 individual and a white skin individual gave rise to 1:1 segregation of red and white, indicating that the root skin color of radish is determined by a single locus and red color is dominant over white. We performed association mapping for root skin color using SNPs obtained from RNA-seq analysis. Segregation analysis on the 152 F3 test-cross population revealed an RsMyb1 transcription factor as a candidate gene to determine root skin color. A PCR marker based on the polymorphism within 2 kb of RsMyb1 was developed and tested on 12 and 152 individuals from F2 and F3 test cross populations, respectively, and red and white root skin colors were completely distinguished corresponding to the genotypes. Expression levels of RsMyb1 in red or purple root cultivars were significantly higher than in white root cultivars. These findings suggest that RsMyb1 is a crucial determinant for anthocyanin biosynthesis in radish roots, and the molecular marker developed in this study will be useful for marker-assisted selection for red skin individuals at early seedling stages. Introduction Radish (Raphanus sativus L., 2n = 18) belongs to the Brassicaceae family, within the tribe Brassiceae. Radish is a root vegetable that is grown and consumed worldwide. Its seedlings, leaves, and siliques are also used as fresh vegetables. Extensive breeding efforts have produced radish cultivars with diverse morphology, varying in root size, shape, color, and growing season, etc. Among them, the exterior root color, which varies from white to black, pink to red and purple, PLOS and yellow, is an interesting phenotypic character of R. sativus. Due to its agronomic importance and accessibility, the exterior root color has been observed and studied for a long time. The major red pigment accumulated in radish root skins is a pelargonidin-or cyanidin-type anthocyanin that is found within the cyanoplasts in the epidermis and in several outer layers of the cortex [1,2]. Besides major substances as color pigments, anthocyanins are well known for their health benefits, such as antioxidant activities and anti-obesity effects, which could add value to radish [3]. Previous genetic analyses showed that a single or multiple loci may control root skin color. As reviewed by Yarnell [4], several reports have demonstrated that red is dominant over white, with a genetic difference associated with the R locus [2,5]. Mostly, a cross between red and white produce red, purple, and white root skin progenies, with the ratio of 1:2:1. A cross between the radish accessions with specific R locus alleles gave rise to progenies with a segregation ratio of 3:1 for red and white [2], albeit some allelic combinations produced progenies with complicated segregation ratios with variable root skin colors [2,5]. For other root skin colors such as yellow and black, multiple loci have been proposed to be involved [4], but none of these genetic models to determine root skin color is yet supported by strong molecular evidence. Transcriptional regulation of anthocyanin biosynthesis and its accumulation have been relatively well characterized in many plant species including maize, petunia, and Arabidopsis. Several transcription factors are known to regulate anthocyanin biosynthesis in many plant species, in which three types of transcription factors Myb, bHLH, and the WD-repeat proteins play important regulatory roles. For example, C, R, and Pl genes have been cloned and characterized for anthocyanin biosynthesis in maize [6]. PAP1, PAP2, myb113 and myb114 are identified as functional homologs of maize counterparts in Arabidopsis, which are also involved in anthocyanin pigmentation [7]. The homologous genes in other species have been identified based on the sequence similarity, and have been shown to perform a similar function. These transcription factors are also interchangeable among species demonstrated by transgenic studies. In Raphanus, however, the genes controlling root color have not been determined due in part to limited genetic resources. Several studies have been conducted for genetics of root skin color in radish but neither the genes nor molecular mechanisms have been demonstrated. Recently, three radish genomes were published from different research groups [8][9][10], facilitating molecular studies of gene functions associated with specific traits as well as genome-based systemic breeding of radish in the near future. In this study, we report the simple color inheritance in radish, and the molecular genetic analysis allowed us to identify the putative R locus gene. In addition, a molecular marker developed from this study will contribute to markerassisted molecular breeding associated with anthocyanin pigmentation in radish. Plant materials A Chinese commercial F 1 cultivar R. sativus cv. Lian Yan No. 1 (Danong Seed Co., Liaonin Sheng, China) with the red skin color was used as a primary material for genetic analysis. The F 1 was self-pollinated to produce 20 F 2 population and further self-pollinated to generate F 3 population. The single F 2 descendant 35 F 3 population, which segregated for root skin color, was subjected to RNA-seq. Each F 3 individual was self-pollinated and the resulting 104 F 4 plants were planted to confirm the original F 3 genotype. F 3 plants were also test-crossed with a recessive homozygote to produce the mapping population (152 individuals, Rr × rr). Plants were grown in a 10-cm diameter and 6-cm height pot in a growth room at 24˚C under 16-hour day condition. For phenotyping, over 10 plants were grown in the field (Suwon, Korea) with conventional culture methods [11]. Currently available domestic commercial radish cultivars were analyzed for their root colors and RsMyb1 expression levels at 4 weeks after planting: KN Red King, KN Ruby King, KN Ruby Ball, and KN Bravo from Kwonnong Seed Co. (Cheongju, Korea), Bordeaux and Jeongwon (Syngenta Korea) for red or purple skin radish cultivars, and Geumbong (Kwonnong), Chungwoon (Syngenta Korea), and Jinjudaepyung (Jinju, a landrace) as white root skin cultivars. Microscopy Hand-sectioned samples were visualized under a stereomicroscope (SteREO Discovery. V12, Carl Zeiss, Germany) equipped with the digital camera (AxioCam MRc, Carl Zeiss, Germany). Sections were stained with toluidine blue O, if necessary, to visualize the cell shapes. HPLC analysis Anthocyanidin profiles of red and white root skins were investigated by HPLC analysis. Root skin was peeled from three homozygous red and white root individuals from F 4 and pooled, respectively. After grinding the root skin tissues with liquid nitrogen, the red pigment was extracted with a solvent mixture containing acetone (40%, v/v), methanol (40%, v/v), and formic acid (0.1%, v/v). A VDS C-18 column (4.6 × 250 mm, 5 μm, VDS Optilab, Germany) was used with a 10 μL injection volume and 0.8 mL min -1 flow rate. Identification of individual anthocyanidins was performed with HPLC and LC-MS as previously described [12]. RNA-seq analysis RNA-seq was performed with 31 separately barcoded libraries (8 white and 23 red root skin individuals), which were generated from the total RNA of young leaf tissues. The library construction and sequencing were performed as previously described [13]. On average, 1.5 Gb of raw sequence reads from each library was obtained. Low-quality sequences (< Q20) as well as barcodes and adaptor sequences were trimmed out using Trim Galore v.0.3.7 with the default parameters but '-gzip-paired'. The purified reads were subjected to the SNP analysis. The raw sequence data have been deposited in the NCBI/EBI/DDBJ Short RNA Archive (PRJNA347524). SNP calling and association mapping The published radish genome (RadishV1.2.2, radish-genome.org) was adopted as a matrix for mapping of sequence reads [8]. Each purified read sequences were aligned against the radish genome by running STAR v2.4.2a 2-pass alignment [14]. Duplicated reads from individual alignments were marked using Picard Tool (v2.4.1) MarkDuplicates and omitted in the following analyses. The Genome Analysis Toolkit (GATK; v3.5-0) RealignerTargetCreator and IndelRealigner protocols were used for global realignment of reads around indels [15]. SNP calling utilized GATK HaplotypeCaller and GenotypeGVCFs with default settings. The filtering settings to validate the SNPs were as follows: FS > 30, DP > 2000, and QD < 2.0. SNP calling was repeated twice with the GATK BaseRecalibrator protocol to reduce ambiguous SNPs. Finally, a total of 208,469 biallelic SNPs were retrieved from the raw, unfiltered 532,923 SNPs. Before association analysis, ungenotyped markers were imputed using Beagle v4.1 [16]. The association analysis to identify the region linked to the root skin color phenotype was performed with FaST-LMM v2.02, running on the Hardy-Weinberg estimate [17]. Marker development and segregation analysis PCR primers to differently amplify specific fragments present in the candidate region of chromosome R7 were designed according to the intergenic region sequences and used to detect polymorphism. Polymorphic PCR bands were used to detect recombination in the F 2 and F 3 test-crossed populations (Rr × rr). The primers of the PCR markers are listed in Table 1. PCR amplicons were visualized with EtBr on a 2% agarose gel. Quantitative real-time PCR (qRT-PCR) analysis Four red and four white individuals from F 3 test-crossed population (Rr × rr), were used for qRT-PCR. Root skins were peeled and frozen with liquid nitrogen. RNA was extracted with RNeasy Plant mini kit (Qiagen, Germany) and treated with On Column DNase (Qiagen, Germany), according to the manufacturer's protocol. cDNA was synthesized from 1 μg RNA, with Superscript III reverse transcriptase (Invitrogen, USA). qRT-PCR was performed using the Rotor-Gene Q 2plex HRM platform and QuantiFast SYBR Green PCR kit (Qiagen, Germany). For testing red and white root skin cultivars, whole roots including the hypocotyl, which were grown in a growth room, were used for RNA extraction and further qRT-PCR. Primers used for qRT-PCR are listed in Table 2. Root skin color is determined by a single dominant gene The radish plants were grown both in a growth room and in a field. The hypocotyl and root began radial expansion from 4 weeks after planting, and almost at the same time, root skin color became visible (Fig 1a). The population in this study had two different types of hypocotyls, red and white (Fig 1b). Red and white root skins were clearly distinguished under both growing conditions (Fig 1a and 1c). However, root skin color and hypocotyl color were not the same in many individuals, suggesting that pigmentation of root skin and hypocotyl is regulated by different processes. We found that red pigments were accumulated in 2-4 layers of subepidermal cells as observed under a stereo microscope (Fig 1e and 1f). The red pigment was identified as anthocyanins as previously reported [18,19]. Major anthocyanins were analyzed by HPLC with known standards (Fig 2), and all major peaks present in the red root skin extract were further confirmed by LC-MS (S1 Fig). Two anthocyanidins were detected as major pigments, in which pelargonidin was the most abundant comprising 98% of total anthocyanidins (S2 Fig). A small amount (2%) of cyanidin was detected in red root skin extracts. The F 1 Chinese commercial cultivar was self-fertilized and the exterior root color was found to segregate 3:1 for red and white in the F 2 population (Table 3). Self-pollinated progenies of white root skin radish all displayed white colors, whereas the descendants from red showed either all red or a 3:1 segregation ratio for red and white. The 3:1 segregation ratio was consistently observed in the F 2 , F 3 and F 4 segregation populations, in a total of 159 plants (P > 0.492, Table 3). Moreover, when F 3 heterozygous red root skin individuals (Rr) were testcrossed with white root skin (rr) individuals, the descendant showed a 1:1 (red: white) segregation ratio (n = 152, P = 1). From this analysis, we concluded that a single dominant gene controls the red exterior root pigmentation in these populations. Colors of other vegetative tissues are associated with root skin color Accumulation of anthocyanins was observed in most of the tissues including the hypocotyl, petiole, stem, and petal (Fig 1a-1d). Plants with a red root skin tend to display red colors in the hypocotyl and the petiole but not in the petal (Table 4). In order to determine the relationship of coloration in different tissues, the colors of the other parts of radish plants were observed and categorized into red and white. The petiole color was sometimes difficult to determine because medium reddish hues were observed. In that instance, the color of the adaxial side of the petiole was used as a color indicator. In total, 16.6% and 9.9% F 3 test-cross progenies displayed color differences between root skin and hypocotyl colors, and between root skin and petiole colors, respectively; and 18.4% showed discrepancy between the hypocotyl and petiole colors (Table 4). Two possible explanations could be proposed. First, two corresponding genes are closely linked. Second, the pigmentation of vegetative tissues including root skin, hypocotyl, and petiole, is regulated by the same gene. However, 47.6% of individuals had different root skin and petal colors (Table 4), suggesting different genetic regulations are involved in reproductive tissue coloration. The mismatch between the root exterior and hypocotyl and root skin color became evident within 4 weeks after planting, indicating the need for a molecular marker that can select red skin root at an earlier stage. Association mapping by RNA-seq identifies the R locus A single F 2 descended F 3 population, which segregated for root skin color, was subjected to association studies. RNA was extracted from young leaf tissues from 23 red (RR or Rr) and 8 white (rr) F 3 radish individuals and used for RNA-seq. We used leaf tissues for RNA-seq primarily to identify SNPs between different root color individuals without damage to the root system. By detecting SNPs from the RNA-seq data, we successfully identified a group of SNPs that cosegregated with the root skin color, in which the first group was likely to have SNPs specific to dominant (RR) or heterozygous (Rr) alleles for red, whereas the second group was predicted to have SNPs specific to the recessive (rr) alleles for white. A total of 208,469 SNPs were scattered throughout the entire chromosomes of radish genome (Fig 3a). Notably, a single major peak associated with root skin color was identified on radish chromosome R7 (Fig 3b), and thus, it is highly plausible that this region (9.11-11.78 Mb) contains the R locus responsible for root color determination. A total of 339 genes involved in various functional categories are annotated in this region. As described below, we further verified their association with the root color phenotype by performing cosegregation analysis with corresponding molecular markers, and assessed their possible contribution to root skin color in relation to the putative functions of these genes. RsMyb1 is located in the region of chromosome R7 In an attempt to identify the putative R locus gene, we first confirmed the association mapping results by developing molecular markers specific to 9.11-11.78 Mb region of chromosome R7. A total of 32 PCR markers were developed based upon bialleleic SNPs and indels, and seven of them were found polymorphic (Fig 3c and Table 1). Five markers were co-dominant amplifying allele-specific fragments, which greatly improved the quality of recombination analysis. Recombination frequencies between the putative R locus and the loci detected by PCR markers were analyzed for 12 F 2 individuals. The putative region was narrowed down to 8.47-12.37 Mb estimated by observed recombination frequency (Fig 3c). Remarkably, R7-9.36 completely cosegregated with the root skin color phenotype in F 2 . Further analysis revealed that R7-9.36 had a perfect cosegregation with the root skin color in 152 individuals of F 3 test cross population (Rr x rr) (Fig 3c). This indicates that the R locus is tightly linked to R-7.36, and therefore, nearby coding genes are the most likely candidate for the R locus gene determining the radish root color. Most notable among several coding genes in this region is a MYB1 transcription factor (called RsMyb1 hereafter), whose contribution to anthocyanin biosynthesis has been reported in a few plant species [6,7]. RsMyb1 is upregulated in red root skin Although we successfully identified SNPs associated with root skin colors from the RNA-seq analysis, the current transcriptome data are not relevant to the expression profiles explaining the root skin color phenotype because sequencing was conducted with leaf tissues. In order to assess the contribution of the coding genes present in region 8.47-12.37 Mb of chromosome R7 to root skin color, we have carefully chosen candidate genes according to the following criteria: first, differential expression levels between red and white root radishes (S3 Fig); second, root-specific expression reported in the Radish Genome Database (radish-genome.org); and third, regulatory functions for anthocyanin biosynthesis reported in other plant species. Eight genes were selected and subjected to further analysis to verify their contribution to root skin color as a putative R locus gene ( Table 2). Their expression levels were determined by qRT-PCR (Fig 4a), and three of them were shown to have significant differential expressions between red and white root tissues. A putative RNA-binding protein (R7_9141477-9142640) was highly expressed in white roots compared to red roots, whereas both MYB114-like transcription factor (Rs388430, called RsMyb1 hereafter) and isoflavone reductase (Rs386960, called RsIFR1 hereafter) were expressed at higher levels in red root tissues than in white ones (Fig 4a). Phylogeny analysis revealed that RsMyb1 was clustered together with the R2R3 motif transcription factors such as MYB113, MYB114, PAP1 and PAP2, all of which are known to be involved in the regulation of anthocyanin biosynthesis in other plant species (S4 Fig) [7]. Further expression analysis on red and white root radish cultivars also revealed that RsMyb1 was expressed at significantly higher levels in red root cultivars than in white root cultivars (Fig 4b). This strongly suggests that RsMyb1 expression is specific to red radish roots and determines the external root color of radish by regulating anthocyanin biosynthesis. RsMyb1 was cloned from red and white root F 3 individuals and their sequences were compared each other. RsMyb1 genes isolated from red and white root radishes (called RsMyb1-R and RsMyb1-W, respectively) are predicted to encode the polypeptides with the same size (249 amino acids) but with differences in amino acid composition at 18 positions (S5 Fig). However, these amino acid differences are unlikely to cause differential activities between RsMyb1-R and RsMyb1-W considering the conserved structure of RsMyb1 compared to other MYB transcription factors. Moreover, RsMyb1-R is more similar to RsMyb1 isolated from white root WK10039 cultivar, and RsMyb1-W is similar to RsMyb1 of red root Bordeaux (S5 Fig). Comparison of the nucleotide sequences outside the RsMyb1 coding region also revealed 13 indels and 14 SNPs between RsMyb1-R and RsMyb1-W alleles (data not shown), suggesting that these differences in cis-element might be responsible for differential expression of RsMyb1 in red and white root radishes, although we cannot completely rule out the possibility that polymorphism in RsMyb1 gene itself is responsible for differential pigmentation in red and white radish roots. Tissue-specific expression analysis revealed that RsMyb1 is expressed in all seedling stage tissues such as leaves, hypocotyls and radicle roots but expressed at significantly lower levels than in developing roots (S6 Fig). These observations are consistent with the expression database showing that RsMyb1 is predominantly expressed in roots during 3-9 weeks after planting with maximum expression at 3 weeks (radish-genome.org). Despite the relatively low expression levels in all seedling tissues, RsMyb1 is generally expressed at higher levels in most tissues of red root radish than in white radish (S6 Fig). Moreover, the expression level of RsMyb1 in the root skins of red root cultivars was found to be 158-518 times higher than in white roots of Jinju, which displayed the highest RsMyb1 expression among white root radish cultivars (Fig 4b). RsIFR1 was also found to be expressed mostly in roots but its expression level did not correlate with the root skin color (S7 Fig), suggesting that RsIFR1 is unlikely to determine the pigmentation of radish roots. All these findings strongly suggest that RsMyb1 is a strong candidate for the R locus gene determining the root skin color of radish. Discussion Root skin color is an interesting phenotype in radish, varying from white, yellow, red, and purple to black among others. Conventional genetic studies have reported possible modes of inheritance of radish root skin color [4]. In 1923, Frost reported that a cross between red and white root radishes produced the F 1 plants with purple or violet-pink roots, suggesting not only that purple color was dominant over red and white colors but also that white radish cultivars carried a purple gene since all the F 1 plants were purple or violet [5]. This study represented the recessive "red" gene by r, and its dominant "wild-type" (purple) allelomorph by R [5]. Consistently, it was also reported that a cross between the Long Red (red) and Icicle (white) cultivars produced the F 1 plants with entirely violet to violet-purplish roots, and subsequently the F 2 plants with a segregation ratio of 1:2:1 for red, violet, and white roots [2]. Similar results were obtained from another cross between Early Red (red) and Early White (white) [2]. These observations suggested that the root exterior color might be controlled by a single locus involving multiple alleles, the combination of which determined the pigmentation in the root exterior [2,5]. However, a later study found the genetics of radish root coloration more complicated with multiple factors involved [20]. This study proposed that radish root color is determined by the three factors R, B, and Y, by which the probable genotypes and root color phenotypes are assumed as follows: RRBByy, purple; RRbbyy, red; rrBByy, white; rrBBYY or rrBbYY, yellow; and rrBBY b Y b , black. In the cross between red and white root cultivars, the F 1 was always purple, and the F 2 showed a segregation of 9 purple: 3 red: 4 white. This indicates that the dominant R is responsible for red pigmentation, whereas B results in white by itself, but in conjunction with R, produces purple pigments [20]. In our study, the Chinese commercial F 1 cultivar Lian Yan No. 1 was used as a starting material, and a cross between homozygous red and white root individuals always produced red-colored roots. Its progeny displayed a segregation ratio of 3:1 for red and white, and a testcross of red-colored heterozygous individual with white one produced 1 red: 1 white. These observations strongly suggest that root pigmentation, in this case, is primarily governed by a Twelve F 2 individuals were genotyped with PCR makers, which were designated and named after their chromosomal positions. The R7-9.36 Mb marker was further applied to 152 individuals of F 3 test-cross population. The numbers of recombinants/chromosome tested are presented in parentheses (c). M, size marker; R, red; W, white root skin individuals. https://doi.org/10.1371/journal.pone.0204241.g003 Radish root skin color determined by MYB1 transcription factor single genetic factor that distinguishes between red and white colors, in which red color is completely dominant over white. When fitted to the Tatebe model, it is very likely that Lian Yan No. 1 used in this study is heterozygous for R (Rr) but homozygous recessive for both B and Y (bbyy), since no colors other than red and white were observed in any cross. According to the previous literature, the radish rook skin color appears to be largely determined by the genetic factors [1,2,5]. Contribution of environmental factors such as sunlight and abiotic stress are also important for anthocyanin biosynthesis, but the skin pigmentation in radish root is mostly qualitative even though some variations exist for the degree of coloration in this study. We identified a total of 208,469 SNPs from RNA-seq and performed association mapping for root skin color in 23 red and 8 white root individuals. A genomic region located on radish chromosome R7 was revealed to be highly associated with the root skin color phenotype, and we successfully identified a panel of candidate genes in the corresponding scaffold. Most notable among them is RsMyb1, encoding an R2R3-MYB transcription factor which is also known as an important regulator of anthocyanin biosynthesis in many plant species [21][22][23]. Recently, RsMyb1 was isolated from a cDNA library of purple root radish cultivar 'Bordeaux', and shown to ectopically upregulate anthocyanin biosynthesis when expressed in Arabidopsis and tobacco [24]. This suggests that RsMyb1, when activated in the root skin, is a key factor that regulates the anthocyanin biosynthesis pathway, eventually determining the root skin color. Utilizing the SNPs and indels identified in the scaffold spanning 9.11-11.78 Mb on chromosome R7, we developed several PCR markers (Table 1), and the following linkage analysis showed that the R7-9.36 marker was completely cosegregated with the root skin color phenotype in 12 F 2 and 159 F 3 testcross individuals (Fig 3). Considering the positions of this marker and RsMyb1 and their tight linkage to red/white root color phenotype, it is highly convincing that RsMyb1 is the R locus gene that determines the root skin color in radish. We found that RsMyb1 was highly upregulated (>100 fold than in white root cultivars) in all red or purple root radish cultivars examined (Fig 4), suggesting that RsMyb1 a common key factor for root skin color in other radish cultivars as well. However, while upregulation of RsMyb1 critical for anthocyanin biosynthesis in radish roots, pigmentation in other parts appears to be regulated by different mechanisms. For instance, F 2 and F 3 progenies of Lian Yan No. 1 showed no strong correlation between root coloration and that of other tissues. In addition, even white root cultivars sometimes developed red colors in hypocotyl, leaves, and flowers, displaying little correlation with root colors. Moreover, both red and white root radish cultivars have similar RsMyb1 coding sequences but their expression levels significantly differ, indicating that a difference in tissue-specific anthocyanin pigmentation is largely determined at the transcription level due probably to different cis element structures, which requires further investigation. HPLC analysis revealed that red radish root skins synthesized a large amount of pelargonidin (~98% of total anthocyanidin), with very little cyanidin (<2% of total anthocyanidin). Such ratio of pelargonidin and cyanidin was also reported for Chinese red root cultivar Hong Feng No. 1, which has a skin color and root shape similar to those of Lian Yan No. 1 [25]. (R7_10214323_10216096(-), isoflavone reductase homolog P3-like), Rs386430 (R7_10493232_10495210(-), mybrelated protein 306-like), Rs386290 (R7_10584636_10586808(-), transcription repressor MYB6-like), R7_11797046_11798366(-) (transcription factor MYB114-like). Ã p < 0.05, ÃÃ p < 0.01, ÃÃÃ p < 0.005. Bars indicate means ± standard deviation from three biological replicates. (b) Expression level of Rs388430 (RsMyb1) was analyzed with qRT-PCR in six red/purple and three white root radish cultivars as indicated. Rs388430 expression was consistently higher in red or purple root cultivars than in white root cultivars. Error bars indicate means ± standard deviation from the three biological replicates. https://doi.org/10.1371/journal.pone.0204241.g004 Radish root skin color determined by MYB1 transcription factor Every tissue we observed displayed red or pink color rather than purple. Notably, purple root radish cultivars Benikanmi and Bordeaux were reported to accumulate prominently more cyanidin in the root than red ones such as Man Tang Hong and Hong Feng No. 1 which have red root skins with high levels of pelargonidin [19,25,26]. This suggests that the ratio of pelargonidin and cyanidin affects the visualization of root color, with pelargonidin and cyanidin being responsible for red and purple colors, respectively. It was proposed that the H locus is involved in radish root color determination distinguishing between red and purple (or violet) [1]. Pelargonidin and cyanidin differ only for the presence (cyanidin) or absence (pelargonidin) of a hydroxyl group at 3' of ring B of flavonoid. It is notable that flavonoid 3' hydroxylase (F3'H) catalyzes 3' hydroxylation of dihydrokaempferol, a precursor of pelargonidin, into dihydroquercetin leading to the biosynthesis of cyanidin (S2 Fig). Therefore, the gene encoding F3'H will be likely a putative H locus gene that distinguishes between red and purple pigmentation in radish roots. Finally, this study demonstrates the feasibility of association mapping using RNA-seq data to identify the candidate gene responsible for a trait of interest. Recently available radish genome information will further facilitate the isolation of the genes involved in many interesting traits, and the molecular markers developed in this study will be highly useful for genomeassisted breeding of radish cultivars with high anthocyanin pigmentation.	v3-fos-license
2022-09-09T16:21:41.318Z	2022-09-07T00:00:00.000	252155156	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://link.springer.com/content/pdf/10.1007/s00187-022-00346-7.pdf", "pdf_hash": "9c985639f70ad95d27af134cc812b99f1dab8aa4", "pdf_src": "Springer", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45350", "s2fieldsofstudy": [ "Business" ], "sha1": "07f9316da321949acae10b5c983046f7eb2a9261", "year": 2022 }	pes2o/s2orc	Moderating effect of carbon accounting systems on strategy and carbon performance: a CDP analysis Carbon emissions bring significant risks and opportunities, and organisations have responded by adopting different strategies and environmental control systems, such as carbon accounting systems (CASs). However, it remains unclear whether a CAS can help reduce emissions, and what role is played by a CAS in the relationship between carbon strategy and carbon performance. Therefore, this paper analyses the strategy-accounting-performance nexus by drawing on 1672 firm-year observations of firms participating in the CDP in 2014 and 2015. The results suggest that the quality of a CAS is influenced by strategic choices; with a proactive carbon strategy being associated with a higher quality CAS. Further, proactive strategies and CASs are found to be associated with carbon savings and emissions reduction. The results indicate a moderating role of CASs on the strategy-performance relationship, with carbon strategy enabling higher carbon savings and lower emissions intensity in the presence of a high-quality CAS. Our findings suggest that formulation of carbon strategies and establishment of carbon measures can drive effective carbon mitigation. Introduction Climate change is a global issue. Climate emergencies have been declared by more than a thousand cities and local governments (ICEF, 2020). With global recognition of climate emergencies, and the need for rapid carbon mitigation, organisations are under increasing pressure to demonstrate how their climate/sustainability efforts are aligned with organisational strategies, and reflected in management control systems (Ghosh et al., 2019;Harris et al., 2019). Among different sustainability control systems, carbon accounting has received heightened scholarly interest. Carbon accounting enables the quantification of organisations' carbon footprint and carbon-related activities, and the use of this quantified information in organisational decision-making (Hartmann et al., 2013;Stechemesser & Guenther, 2012). Prior literature distinguishes between (reporting-driven) carbon accounting and (internally performance-driven) carbon control, whereby the research that draws upon the CDP is about carbon accounting, and the research that discusses accounting for eco-efficiency, cost savings and material flow accounting, is concerned with carbon control. Despite their differences, carbon accounting and carbon control are intertwined, because the information generated for carbon reporting may also affect carbon management, and vice-versa (Qian & Schaltegger, 2017;Qian et al., 2018). Using this reasoning, we try to understand the link between strategy, carbon accounting and carbon performance by drawing upon externally reported carbon information in the CDP. By doing so, we respond to calls for an exploration of how businesses internalise legitimacy pressures and demands for creating real improvements (Qian & Schaltegger, 2017), and to the need for more empirical studies on sustainability and carbon management accounting (Crutzen et al., 2017;Harris et al., 2019;Hartmann et al., 2013). Generally, the relevant literature recognises that the effective implementation of a sustainability strategy requires comprehensive environmental management control systems (EMCS) that ensure the integration of sustainability into core businesses, and push organizations towards a sustainable future (Epstein, 1996;Epstein & Wisner, 2005;Gond et al., 2012). Furthermore, previous studies have investigated the design and use of environmental controls, their drivers, and their impacts on organizational performance (Adams et al., 2007;Burritt et al., 2011;Henri & Journeault, 2010). 1 However, few studies have examined how sustainability strategy influences carbon control design, via a large dataset (Harris et al., 2019). Likewise, there is a very limited empirical evidence regarding the impact of carbon accounting on carbon mitigation. Additionally, prior literature often refers to carbon accounting in a limited sense, i.e., the quantification of the carbon footprint and the use of such information in decision making, and does not extend to other controls that also help fulfil carbon management objectives. Therefore, following Simons' (1991) concept of 'levers of control', we argue that CASs as a subset of EMCS, encompass carbonfocused environmental controls (eco-controls) that include "formalised procedures and systems" that "maintain or alter patterns in environmental activity" (Henri & Journeault, 2010, p. 64). Hence, CASs include procedures and systems, such as targets and budgets, strategic planning, and reporting systems, working as a package of control (Malmi & Brown, 2008), with the aim of achieving the carbon management objectives of organisations. Further, the control implications, for a firm adopting a proactive climate change strategy, versus another firm adopting a reactive strategy, are unclear in extant literature. As argued by Harris et al. (2019), future sustainability studies need to examine a larger set of controls as driven by, and implicated in, different strategic orientations. With firms being able to adopt different climate change strategies (Boiral, 2006;Kolk et al., 2008;Weinhofer & Hoffmann, 2010), it is critical to understand which carbon strategy is the most (or least) effective in reducing carbon emissions. Overall, this means that the strategic and EMCS mechanisms, through which firms drive carbon mitigation, are poorly understood. Therefore, this paper aims to address the three-way relationship between carbon strategy-accounting-performance by utilising an international sample of 1672 firmyear responses to the CDP in 2014 and 2015. The CDP is widely perceived as providing the largest and most comprehensive database of voluntary reporting of carbon-related performance and activities of large firms around the world Matsumura et al., 2014). This paper brings the focus on carbon performance by adopting a resource-based view, whereby CASs and proactive carbon strategies can help develop key resources and capabilities that enable organisations to improve their performance and competitive advantage (Eisenhardt & Martin, 2000;Teece et al., 1997). The proactive carbon strategies include strategic integration, reduction initiatives, policy engagement, value chain engagement, and carbon credit origination (Jeswani et al., 2008;Kolk & Pinkse, 2005;Weinhofer & Hoffman, 2010). Eight CAS components are examined: strategic planning, targets, carbon budget, financial performance measurement, non-financial performance measurement, project management method, incentives, and reporting. Considering eight components allows us to assess the implications of a CAS for carbon performance as a package (e.g., Henri & Journeault, 2010;. In this paper, carbon performance is proxied by annual carbon savings and emissions intensity of firms in a given year (Luo & Tang, 2016;. The findings of this study make three contributions to the literature. Firstly, consistent with a resource-based view, and extending prior empirical studies that focus on the carbon disclosure-performance relationship (Clarkson et al., 2015;Kolk et al., 2008;Luo & Tang, 2016;Qian & Schaltegger, 2017;Qian et al., 2018;Schiemann & Sakhel, 2018;, our findings suggest that high quality CASs are linked to higher annual carbon savings and lower emissions intensity. Our findings diverge from other studies that question the contribution of CASs, in that carbon accounting on its own is insufficient to achieve carbon reduction (Jackson & Kaesehage, 2020), and that the emissions data may be arranged to present a situation consistent with expectations (Lippert, 2015). Rather than just carbon measurement 1 3 or reporting, we argue that a comprehensive and formal CAS, that captures different components, can have a meaningful effect on carbon mitigation. Secondly, this study responds to calls in the literature to determine which strategies improve carbon performance (Qian & Schaltegger, 2017). We provide crosscountry evidence of the varying implications of different strategies on performance. Based on prior literature, this study takes into account five different types of proactive strategies. We find that a proactive carbon strategy is effective in improving carbon savings and reducing emissions intensity. This confirms prior literature regarding the positive impact of proactive strategies on environmental performance (Hart, 1995;Kolk & Hoffmann, 2007;Kolk et al., 2008) and argues that this impact applies equally to a climate change context. Thirdly, this research suggests a moderating role for CASs in the enactment of strategies for carbon performance. Prior empirical studies reveal a positive impact of CASs on carbon performance, but they do not consider the influence of carbon strategy. This study demonstrates direct effects of carbon strategy on carbon performance, as well as indirect effects as moderated by a CAS. The results are consistent with a resource based view (RBV), that a CAS performs a moderating role on the relationship between carbon strategy and carbon performance, highlighting that a high quality CAS enables strategy to have a stronger impact on carbon performance. The interaction between CASs and carbon strategy has an overall positive relationship with carbon savings and a negative relationship with emissions intensity. Consistent with studies on the CAS-strategy relationship (Bui & de Villiers, 2017;Bui & Fowler, 2019;Ghosh et al., 2019;) we argue that the adoption of both a CAS and strategy rather than each individually, is beneficial for annual carbon savings. This effect is the most pronounced among polluting firms, possibly because the combined effect may drive larger scale operational, behavioural and strategic changes which, in turn, result in significant carbon mitigation. The rest of the paper proceeds as follows. In Sect. 2, we review the prior literature in order to understand the relationships between carbon strategy, CASs and carbon performance. Section 3 discusses the theory underlying the link between carbon strategy, CASs, and carbon performance, and proposes the hypotheses. In Sect. 4 we explain the research design, and in Sect. 5 we discuss the results. Section 6 summarizes the paper and provides contributions, limitations and practical implications. Literature review This section reviews the literature on the relationship between carbon strategy and CASs, and between CASs and carbon performance. The carbon accounting system (CAS) and carbon strategy Despite the extensive literature on management control and strategy (e.g., Ferreira & Otley, 2009;Simons, 1990), the insights into how companies design or use management control to support sustainability strategies are only just emerging (Crutzen & Herzig, 2013;Henri & Journeault, 2010). EMCS can be designed to support sustainability strategies (Crutzen et al., 2017;Figge et al., 2002;Hansen & Schaltegger, 2016), but the role played by EMCS in various sustainability strategies is not well understood (Crutzen & Herzig, 2013;Ghost et al., 2019). With the global urgency of climate change, carbon emissions pose significant risks and provide (improvement) opportunities to organisations (Bebbington & Larrinaga-González, 2008;Cadez & Czerny, 2016) and the focus on carbon emssions has facilitated the growth of a plethora of private standards to guide carbon accounting and disclosure, such as the Greenhouse Gas (GHG) Protocol (Sundin & Ranganathan, 2002) and other standards (Green, 2010). Corollary there is an increasing integration of carbon accounting into corporate strategic management and EMCS (Engels, 2009;Hopwood, 2009;Luo & Tang, 2016). Carbon-focused management accounting can facilitate cross-departmental communication and increase the efficiency and effectiveness of information processing (Burritt et al., 2011;Kumarasiri & Jubb, 2016). Similarly, performance measurement systems enable organisations to maintain transparent accounts of carbon emissions, and identify reduction potential (Schaltegger & Csutora, 2012;Schaltegger & Zvezdov, 2015). Appropriate design and use of CASs can help develop key organisational capabilities and implement a proactive carbon strategy that helps achieve a competitive advantage (Bui & de Villiers, 2017;Henri & Journeault, 2010;Menguc et al., 2010). The carbon strategy and accounting link has been examined in different sectors, such as the automotive industry (Lee, 2012), forestry (Ellison et al., 2011) and agriculture (Huang & Mi, 2011), as well as in a cross-sectional context (Bui et al., 2020). While these studies suggest that carbon accounting is useful in mitigating carbon emissions and implementing carbon strategies, it is unclear which form of carbon accounting is useful, and for what type of carbon strategy. A few studies have looked at the organisational strategic responses in a climatechange-sensitive business environment (Bui & de Villiers, 2017;Cadez & Czerny, 2016;Kolk et al., 2008;Weinhofer & Hoffmann, 2010). For instance, Cadez and Czerny (2016) propose three strategic priorities, ranging from internal carbon reduction (i.e., combustion emissions reduction, process or product emissions reduction) to external carbon reduction (mainly through supply chains), and carbon compensation. Similarly, Weinhofer and Hoffman (2010) classify climate change strategies as focusing on either CO 2 compensation, CO 2 reduction, or carbon independence. Further, firms can adjust their climate change strategy from stable to reactive, anticipatory, proactive, or creative, hinging on the degree of uncertainty of regulatory requirements (Bui & de Villiers, 2017). However, it remains unclear how different climate change strategies can influence CASs. 2 CASs and carbon performance This section discusses the use of CASs in carbon management, as well as some limitations of CASs. Prior studies have examined the relationship between carbon emissions and disclosure and financial performance using proxies such as market value of equity (Matsumura et al., 2014;Saka & Oshika, 2014), cost of equity and cost of debt (Li et al., 2014). These studies support the view that carbon accounting and reporting enhance firms' financial performance. Another stream of literature examines CASs from a managerial perspective, and emphasises the impact of using CAS information on carbon performance, albeit with inconclusive evidence. Henri and Journeault (2010) provide empirical evidence for the positive impacts of eco-controls on various aspects of environmental performance, such as reduction in material costs, increased productivity, better relationships with stakeholders, or overall company reputation. Wijethilake et al. (2018) study of 175 manufacturing firms in Sri Lanka and find that EMCS moderates the relationship between environmental innovation strategy and organisational performance. In addition, studies have utilized CDP data to understand the association between carbon accounting and performance (Clarkson et al., 2015;Kolk et al., 2008;Luo & Tang, 2016;Qian & Schaltegger, 2017;Qian et al., 2018;Schiemann & Sakhel, 2018;. Tang and Luo's (2014) study of 45 Australian firms indicates that a firm can mitigate its carbon footprint through a high-quality carbon management system. Qian and Schaltegger (2017) analyse Global 500 companies, and find that change in carbon disclosure levels is associated positively with subsequent change in carbon performance. However, they do not examine components of carbon accounting and controls but, rather, the extent of disclosure. Differently, Qian et al., (2018) draw on the Corporate Sustainability Barometer (CSB) and CDP database of 114 large companies across the US, Germany, Australia and Japan, and find that the application of environmental accounting has a significant positive impact on both corporate carbon management and disclosure quality. Ott and Endrikat (2022), using CDP database of S&P 500, find that financial carbon-related incentives are associated with superior carbon performance, while non-financial incentives are not. This indicates the differing impacts of incentive design on carbon performance. Summing up, CDP based studies have not explored the impact of a comprehensive CAS on carbon performance. Hypothesis development This study adopts the lens of the natural resources-based view (RBV) as proposed by Hart (1995) to develop hypotheses. The RBV suggests that firms maintain their competitive advantage by utilising and nurturing resources that are not easily imitated by competitors. The RBV conceptualizes firms as bundles of resources heterogeneously distributed across firms, and suggests that these resource differences persist over time (Amit & Schoemaker, 1993;Wernerfelt, 1984). Resources must satisfy the key criteria of being valuable, rare, inimitable, and non-substitutable, if they are to lead to the achievement of sustainable competitive advantage (Barney, 2001). Resources enable the implementation of value-creating strategies via elements, such as physical assets, human resources, organizational assets, and competencies (Eisenhardt & Martin, 2000;Teece et al., 1997). Capabilities enable the interaction between these resources and their effective deployment. They are defined as "The firm's processes that use resources -specifically, the processes to integrate, reconfigure, gain and release resources-to match, and even create, market change" (Eisenhardt & Martin, 2000, p.1107. Prior research has suggested innovation, organizational learning, market orientation and entrepreneurship are among the primary capabilities needed to reach competitive advantage (Bhuian et al., 2005;Henri, 2006). Carbon strategy and the CAS Consistent with an RBV, the design and practice of EMCS should be tailored to the corporate strategic intent, in order to optimise organisational performance (Chenhall, 2003;Henri, 2006) and gain competitive capabilities (Henri, 2006;Widener, 2007). While sustainability strategy provides high-level direction and policy with regards to environmental issues, EMCS provide the specific tools for coordinating and aligning the resources and processes needed to turn such strategy into actual performance outcomes (Ghost et al., 2019;Lee, 2012). A proactive environmental strategy requires certain capabilities to allow the collective deployment of multiple resources, ensuring that they work in sync to improve organisational performance. For example, a pollution prevention strategy would require the monitoring and management of environmental impacts, over and beyond minimum regulatory requirements (Hart, 1995). This is achieved via an EMCS using performance measures and monitoring systems. Proactive environmental strategies require the provision of physical and monetary information regarding the ecological cost of organisational product, process, or activities (Adams & Frost, 2008). Indeed, long-term-oriented physical and monetised carbon accounts are used extensively when firms adopt creative or proactive strategies, as opposed to other strategies (Bui & Fowler, 2019). Consequently, changes in carbon strategies require modification of CASs to support the new strategic intents and objectives (Bui & de Villiers, 2017). A proactive environmental strategy would also require a good reporting system, so environmental issues are forwarded to relevant managers, and elevated to senior levels if they present significant risks to the organisation. Interactive controls allow top management's focused attention and intervention with regard to environmental issues of strategic importance (Simons, 1991). Furthermore, a proactive strategy involves a high organisational commitment to managing environmental performance (Hart, 1995). This inevitably requires some form of formal environmental target, and associated budgets and processes to ensure performance is monitored and corrected against the target (Bui et al., 2020). Accordingly, a more proactive climate change strategy is associated with more formal environmental controls (Pondeville et al., 2013), in order to affect various organisational decisions (Christ & Burritt, 2013). Therefore, we hypothesise the following. Hypothesis 1 A more proactive carbon strategy is associated with a higher quality carbon accounting system. CAS and carbon performance In accordance with RBV, we argue that CASs incorporate processes that ultimately provide a source of competitive advantage. Prior studies have documented the direct performance benefits of environmental controls, such as quality enhancement, cost savings, more accurate product pricing, and retention of skilled personnel (Dunk, 2007;Gunarathne & Lee, 2015). Indirect benefits include organisational learning, continuous innovation, stakeholder integration, and shared vision and goal congruence capabilities (Adams et al., 2007;Journeault, 2016). There are different types of controls, such as diagnostic controls and interactive controls. Diagnostic controls, such as managerial incentives, carbon targets and investment modelling, motivate organisational members to align their behaviour with carbon management objectives and, hence, lead to performance improvement. Extant literature provides conflicting evidence regarding the relationship between incentives and performance, from no relationship , to a negative relationship found between monetary incentives and carbon mitigation (Ioannou et al., 2016). In contrast, setting carbon targets enables firms to monitor emissions, set benchmarks for performance assessment, and control negative deviations from pre-determined targets (Adams et al., 2007;. Prior studies indicate that more difficult targets are more likely than less difficult ones to be accomplished, thus, supporting the impact of target setting on performance (Ioannou, et al., 2016;Larrinaga-González et al., 2001). The use of carbon measures in investment modelling can provide a platform for discussion and dialogue, and for encouraging innovation within the organisation (Bui & Fowler, 2019). Reporting systems and strategic planning often serve as interactive controls in carbon management (Bui et al., 2020;Simon, 1995). Reporting systems, notably more frequent communications on risk management and strategy from lower-to topmanagement levels, allow the detection of risks before they become real problems and threaten the achievement of organisational objectives (Simons, 1995;Van der Stede, 2001). Differently, a strategic planning process ensures the review of current strategies, evaluation of the risks and opportunities, and the formulation of new strategies. Prior studies have found that the board of directors plays a critical role in monitoring and reporting carbon information and ensuring climate change accountability to firm stakeholders (Ben-Amar & McIlkenny, 2015; Prado-Lorenzo & Garcia-Sanchez, 2010). Hence, frequent reporting of carbon risk information to the board will feed into the strategic planning process and the development of climate change strategies. Further, active scrutiny by the board is likely to result in intensive monitoring at lower management levels and the promotion organisational learning on carbon issues (Bui, 2011;Bui & de Villiers, 2017). Through these different processes and resulting capabilities, we expect that comprehensive CASs with diagnostic and interactive controls, that embed climate change issues, will lead to stronger carbon performances. The following hypothesis is, thus, formed: Hypothesis 2 A higher quality carbon accounting system is associated with stronger carbon performance. Carbon strategy and its effect on carbon performance (both directly and moderated by the CAS) Though it has been theoretically implied, the relationship between proactive carbon strategy and carbon performance has not been adequately investigated. For example, Clarkson et al. (2011) document a positive relationship between environmental and financial performance, where environmental performance is driven by a proactive environmental strategy. Using a S&P 500 sample, Moussa et al. (2020) report a positive link between carbon strategy and carbon performance and a mediating role for carbon strategy on the relationship between board environmental orientation and carbon performance. Prior literature based on the RBV has argued that a proactive environmental strategy can provide a source of competitive advantage. For example, a proactive strategy can improve environmental performance through investing in end-of-pipe pollution treatment or prevention, developing greener products, or pursuing sustainable development through low-impact technologies (Hart, 1995). Environmental proactivity can result in capabilities such as stakeholder integration, organizational learning, and continuous improvement (Sharma & Vredenburg, 1998). Firms can innovate on their own, or in collaboration with stakeholders and industry partners (Kolk & Hoffmann, 2007;Kolk et al., 2008) and, hence, they can enhance potential carbon savings or innovation outcomes. A proactive strategy also emphasises organisational changes, such as behaviour shifts towards more sustainable resource consumption and, thus, reduce negative environmental impacts (Aragón-Correa & Sharma, 2003). A carbon strategy that takes climate change issues seriously, also encourages risk-taking and entrepreneurship. This is because effective carbon mitigation goes beyond energy efficiency and requires technological transformation, which does not occur without significant investment with high risk, while the returns are realised only in the long term. Accordingly, a proactive climate change strategy results in capabilities that will lead to stronger carbon performance. Hence, we formulate the following hypothesis: Hypothesis 3a A more proactive carbon strategy is associated with stronger carbon performance. Further, we argue that this relationship is also moderated by the quality of the CAS. The RBV suggests that a sustainable competitive advantage relies on organisational "organizing", i.e., the ability to exploit the rare, valuable, or non-imitable capability or resources of an organisation (Barney, 2001). CASs help to organize resources and, hence, to implement carbon strategy through various mechanisms (Crutzen & Herzig, 2013). For instance, targets facilitate efficient resource allocation into areas that can result in the highest carbon reduction (Ioannou & Serafeim, 2012). Similarly, incentive systems can reinforce manager and staff motivation towards achieving carbon plans and initiatives as part of the carbon strategy (Bui et al., 2020). The integration of carbon indicators into investment modelling enables the reorientation of organisational resources towards lower carbon technologies and, hence, the achievement of a lower-carbon business strategy (Bui & Fowler, 2019). Further, the reporting of carbon information to the board allows top management monitoring of carbon performance, and timely action to correct deviations against the planned strategy (Moussa et al., 2020). Overall, by facilitating strategy implementation, CASs allow the mobilisation of organisational financial and non-financial resources in alignment with carbon strategies, leading to better carbon performance. Hence, in accordance with the RBV, we argue that a CAS strengthens the relationship between a proactive carbon strategy and carbon performance. Thus, the following hypothesis is formed: Hypothesis 3b A higher quality carbon accounting system moderates positively the relationship between a proactive carbon strategy and carbon performance. Sample selection This study utilises the information obtained from the CDP 2014-2015 database, in conjunction with firms' financial information obtained from the Thomson Reuters DataStream. Information captured in CDP is faithfully represented and reliable, as CDP questionnaires and scoring methodology are well-constructed, leaving little opportunity for managers to provide misleading information (Depoers et al., 2016). The disclosures to CDP, according to some studies , are indicative of the underlying carbon performance. Furthermore, the CDP database is regarded as the largest source of primary climate change information (Andrew & Cortese, 2011;Matsumura et al., 2014) and, therefore, is able to cover various aspects of corporates' climate change activities. We choose 2015 and 2014 as our years of investigation, owing to the consistency in the structures and content of the CDP questionnaires, 3 and the inclusion of two years allows us to control for change over time. After omission of observations with missing dependent and independent variables and zero emissions, and winzorizing of financial variables, we arrive at final sample of 1672 observations, as shown in Table 1. 3 Moderating effect of carbon accounting systems on strategy… Regression models To test the hypotheses, the following regression models are employed: The main variables of interest are CAS (carbon accounting system), ΣCP (carbon performance i.e., CARSAV and INTENS), PROACT (carbon strategy) and CASPROACT (interaction term between carbon accounting system and carbon strategy). Model 1 analyses the interplay between carbon strategy and the CAS (H 1 ), while model 2 examines the relationship between the quality of the CAS and (1) These examined relationships are portrayed in Fig. 1. The measurement of CAS This study adopts the scoring methodology recommended by CDP, with minor modifications to assess the quality of the CAS (detailed scoring methodology is in Appendix 1 Fig. 1 The key relationships in the study 4 Due to data limitations and the structures of CDP questionnaires, we are unable to discern the use of informal controls in carbon management. 5 Gond et al. (2012) also suggest a hybrid measurement system (such as the balanced scorecard). However, we are not able to construct the measure for this MCS, as firms do not disclose their sustainability balanced scorecard in their CDP responses. and Reporting. Accordingly, we use strategic planning to ascertain whether carbon issues are integrated into the strategic planning process. As absolute targets are often seen as potential inhibitors of future economic performance and, hence, more difficult to achieve (Ellerman & Wing, 2003;Sue Wing et al., 2006), absolute targets are awarded higher points than intensity targets. Budgets are captured to denote the existence of a carbon budget or fund. The financial measurement system captures the use of financial measures in carbon management, specifically, to determine (i) whether there is an internal price of carbon, and (ii) whether monetary savings are calculated from carbon reduction initiatives. Next, the non-financial measurement system represents the use of non-financial indicators of carbon management. Project management methods check for the adoption of a formal financial-related method (e.g., IRR or NPV) used to drive investments in carbon projects. Incentive captures the existence of some form of evaluation and reward system for carbon mitigation, either financial, non-financial, or both. Finally, based on Burritt et al. (2011) and Simons (1995), we develop the measure: Reporting; to represent the interactive control, i.e., whether carbon information is reported and monitored by the board. Accordingly, we are able to collect information about eight specific components of the CAS from the CDP questionnaire. This approach is also driven by, and is consistent with, the literature on EMCS and sustainability control systems. These components comprise, arguably, one of the most comprehensive indices in the literature focusing on carbon controls. The CAS is a composite index measure ranging in value from 0 to 13, computed by adding up the scores of the eight components. The measurement of carbon performance This paper employs two direct measures of carbon performance. Luo and Tang (2016) and Tang and Luo (2014) adopt a relative measure of carbon performance, an index based on four criteria: carbon intensity decline compared to the previous year, carbon intensity lower than the sector's median, at least one of the firm's targets being achieved, and carbon savings realised from at least one of the firm's emissions reduction initiatives. While this captures the likelihood of an improvement in carbon performance, a relative measurement does not capture the extent of improvement. Therefore, we capture the actual carbon performance via two direct measures: (i) CARSAV, the amount of estimated annual carbon savings achieved, computed by the natural logarithm of estimated annual CO 2 savings (metric tonnes CO 2 ) achieved from various initiatives implemented during the reporting year, 6 and (ii) INTENS, emissions intensity as computed by totalling scopes 1 and 2 and scaling by revenues. CARSAV captures both past carbon savings and on-going savings, hence, providing some perspective on the future carbon performance, while INTENS captures the current reporting year's emission level. Both measures are derived from the CDP questionnaire databases, consistent with the approach used prior studies (Chapple et al., 2013;Jung et al., 2018;Luo and Tang, 2014;Safiullah et al., 2021). 7 The measurement of carbon strategy Consistent with prior literature and the CDP questionnaire, proactive strategies represent a more proactive stance designed to reduce and offset emissions, or influence the policy-making process (Weinhofer & Hoffman, 2010). Hence, we measure proactive strategy as comprising strategic integration (Lee, 2012), innovation, and cooperation within or beyond the supply chain (Weinhofer & Hoffman, 2010), and political action to influence policy makers on climate change issues (Kolk & Pinkse, 2005;Jones & Levy, 2007). Strategic integration is represented by the integration of carbon issues into strategic processes (STRINT), and its score ranges from 0 to 3. Innovation is proxied by reduction initiatives (REDINI), political action by policy engagement (POLENG) and credit origination (CREORI), which are dummy variables taking the value of 1 should firms participate in any reduction initiatives, have a clear and consistent engagement process with policy matters, and originate their own carbon credits externally, respectively. Cooperation with supply chain partners is proxied by value chain integration (VALCHA) and it score ranges from 0 to 2. Proactive strategy (PROACT ) (ranging from 0 to 8) is measured as the sum of strategy integration, reduction initiatives, policy engagement, value chain integration, and credit origination. In additional analysis, we check to which extent reactive strategies are associated with performance benefits, as prior research suggests that proactive strategies are more likely to result in performance benefits than reactive strategies (Hart, 1995). Reactive strategies (REACT ) are those that focus on compensation strategies, and comprise ETS participation and credit purchasing. Table 2 summarises the expected signs of the coefficients based on the hypotheses. 1 3 Moderating effect of carbon accounting systems on strategy… Control variables SIZE is measured by the natural logarithm of total revenues, which has been found to influence carbon strategy, disclosure/control, and performance significantly (Alrazi et al., 2016;Chapple et al., 2013;Gallego-Álvareza et al., 2015;Journeault, 2016). ROA is measured by net income to total assets, as poor profitability may be one factor that limits firms' ability to embrace higher quality carbon accounting systems (Uchida & Ferraro, 2007), and TOBINSQ is calculated to control for corporate management capability, as more innovative firms tend to invest in greener products and low-carbon technologies (Clarkson et al., 2015;Daske et al., 2008). Finally, NEW is measured by age of the assets of the company. At the country level, several factors may drive corporate carbon-related strategy and accounting systems. First, developing countries may prioritise economic development (LNGDP) over environmental protection (Galeotti, 2007). Second, firms operating in code law (LAW) jurisdictions may adopt high quality carbon accounting systems, because such adoption can enable stronger firm-level corporate governance, to offset the weakness in the investor protection mechanism . Third, firms in countries with an ETS are subject to more regulatory pressures and, hence, are likely to adopt high quality carbon accounting systems . Hence, three variables, LNGDP, LAW and ETS are measured and controlled for. The details regarding the measurement of variables are defined in detail in Appendix 1. 1 3 Moderating effect of carbon accounting systems on strategy… Table 5 (Panel A) presents descriptive statistics of all our dependent and explanatory variables. We find that CARSAV has mean (median) values of 9.90 (9.86). INTENS has mean (median) values of 0.03 (0.00) with a standard deviation 0.10. The mean (median) value of CAS is 6.95 (7.00), out of a maximum possible 13 points. This shows that most firms do not adopt extensive carbon accounting systems. Furthermore, PROACT has a mean (median) value of 5.1300 (6.000) out of the maximum possible value of 8, indicating most firms adopt a variety of proactive strategic responses. The summary statistics for the control variables are also shown in Table 5 (Panel A). Table 5, 6, Panel B, presents the Pearson correlation coefficients. It demonstrates positive correlations between CAS and PROACT . CAS is also correlated with CARSAV, but not with INTENS. PROACT have positive correlations with CARSAV and CAS, and a negative correlation with INTENS. Further, we computed Variance Inflation Factors (VIFs 2.71) when estimating our regression models to test for signs of multi-collinearity between the explanatory variables. Thus, multi-collinearity is not a problem in our study (Hair et al., 2006). PROACT is associated with CAS positively and significantly, indicating that more extensive carbon strategy requires a higher quality carbon accounting system (Coff. = 0.6319, p < 0.01). 9 Overall, H 1 is supported. Consistent with an RBV, a CAS provides the organizing capability that enables a proactive carbon strategy to be implemented effectively. A proactive strategy requires that an organisation adopts reduction targets, performance measures and regular carbon reporting, and integrates carbon measures into investment decisions in order to realise a proactive strategy aiming at carbon mitigation. Accordingly, a high quality CAS is linked to higher annual carbon savings and a reduction in firms' emissions intensity. The relationship between the CAS and carbon performance (H 2 ) These findings are aligned with Journeault et al. (2016) and Luo and Tang (2016), who also found a positive association between environmental/carbon controls and carbon performance. 10 This is consistent with the RBV-focused literature that suggests better environmental (carbon) accounting brings about improved performance. Table 7 Model 3 shows the results of direct relationships between proactive carbon strategy and carbon performance. PROACT is associated positively with both CAR-SAV (Coff. = 0.0777, p < 0.01) and INTENS (Coff. = − 0.0055, p < 0.01). Overall, this confirms H 3a that firms with more proactive carbon strategies achieve higher carbon savings and lower carbon emissions. While aligning with a RBV perspective on the performance effects of a proactive strategy, our results are inconsistent with prior studies that question the usefulness of carbon management initiatives from a carbon reduction perspective (Damert et al., 2017;Doda et al., 2016). These studies, suggest that firms may do the talking before the walking, suggesting a gap between the talk (i.e., disclosure) and the impact of the actions (emissions reduction). Doda et al. (2016) suggest that firms might have already exploited the potential for emissions reduction before reporting. However, these studies use data prior to 2013 and do not capture proactive carbon strategy directly. Damert et al. (2017) include compensation strategies in the strategy index, while Doda et al. (2016) capture measurement and disclosure practices in the carbon management initiatives. Our study, by differentiating between proactive and reactive strategies, supports the notion that when firms pursue a proactive strategy in carbon management, their carbon performance is improved. The relationship between carbon strategy and carbon performance, both direct and moderated by the CAS (H 3a and H 3b ) The indirect relationship between strategy and performance is tested via the moderating effect between strategy and CAS in Table 7 model 4, where both CAS (the moderating variable) and PROACT (the independent variable) and the interaction term are included. Accordingly, CAS and PROACT are no longer associated significantly with carbon performance, but the interaction term PROACTCAS is significantly associated with both measurements of performance. Hence, the presence of high-quality CAS strengthens the effect of proactive strategy on carbon savings (CARSAV, Coff. = 0.0158, p < 0.01). However, the negative association of the interaction term and emission intensity (INTENS, Coff. = − 0.0019, p < 0.05) indicates that the effect of the combined presence of strategy and CAS is less than the sum of the individual effects on performance. In other words, the association between carbon strategy and emission intensity is lessened when firms adopt a higher quality CAS. Overall, the signs of the interaction terms are consistent with H3 b , that the combination of proactive strategy and high quality CAS is associated with higher carbon savings and lower emission intensity. Accounting for both direct and indirect impacts, the results indicate that carbon strategy has an overall positive relationship with carbon performance, both directly and in the presence of CAS. In other words, CAS moderates the relationship between carbon strategy and carbon performance. Table 7 indicates that firms with higher quality CASs tend to be bigger in size, and to operate in countries with a code law system. Further, based on Table 7 model (4), which controls for both CAS and carbon strategy, firms with more annual carbon savings (CARSAV) tend to be bigger in size and operate in countries without an ETS, while firms with higher emissions intensity (INTENS) operate in countries with higher economic development and a common law system. This confirms the role played by a voluntary context (no ETS regulation) in encouraging firms to adopt carbon mitigation initiatives and achieve carbon savings, while a code law system is more conducive to lower emissions intensity. Additional analysis Prior research also indicates that carbon strategies can be reactive or proactive (Jones & Levy, 2007;Kolk & Pinkse, 2005;Weinhofer & Hoffmann, 2010). Proactive strategies are more likely to result in performance benefits than reactive strategies (Hart, 1995). Table 8 model 1 show that REACT is associated with CAS positively and significantly, indicating that more extensive carbon strategy requires a higher quality carbon accounting system (Coff. = 0.3725, p < 0.01). 11 Table 8 model 1 indicates that a high-quality CAS is needed, whether firms follow a proactive or a reactive carbon strategy. Two explanations are possible here. First, it is established by existing research that strategy (regardless of being proactive or reactive) influences accounting systems (Arjaliès & Mundy, 2013;Langfield-Smith, 2005). Second, some form of CAS is needed to account for carbon-related activities, even though those activities involve credit purchase or emissions trading (reactive strategies). Further, results in Table 8 model 3 show that reactive strategy is linked to carbon savings (REACT, Coff. = 0.1722, p < 0.01) but has no relationship with emission intensity (REACT, Coff. = 0.0019, p > 0.1). No significant interaction terms in model (4) also suggest an absence of a moderating effect of CAS on reactive strategy-performance relationship. Overall, this confirms the lack of a clear association between reactive strategy and performance, partially explaining why earlier studies have not found a relationship between a composite strategy and carbon performance (Damert et al., 2017;Doda et al., 2016). Table 9 model 5, 12 when we include REACT and PROACT and their interaction terms with CAS in the same regression, the results hold that only proactive strategy has an indirect relationship with both measures of carbon performance via the moderating impact of CAS (CARSAV Coff. = 0.0630, P < 0.01;INTENS, Coff. = − 0.1898, p < 0.05). However, no such indirect relationship exists when firms adopt a reactive strategy. For robustness tests, we also adopt alternative measurements for carbon performance, when scaled by total assets, consistent with Chapple et al. (2013), Jung et al. (2018), Luo and Tang (2014) and Safiullah et al. (2021), and common shares outstanding, consistent with He et al. (2021). The results reported in Table 10 model 4 are qualitatively similar to our main results, confirming the existence of the moderating effect of CAS on the strategy-performance relationship. We further divide the sample into carbon intensive and carbon non-intensive firms based on the emitting nature of the industry in which a firm operates (Safiullah et al., 2021). According to CDP, we identify carbon intensive firms as high carbon emission or energy consuming industries (energy, utilities and materials sectors are defined as the most carbon-intensive firms). Results in Table 11 model 4 suggest the interaction term is associated with carbon savings among carbon intensive firms (Coff. = 0.0262, p < 0.05), while the association with emission intensity is observed only among carbon non-intensive firms (Coff. = − 0.0019, p < 0.05). Hence, CAS has the ability to strengthen the impact of strategy on carbon savings in polluting firms, whereas its presence in less polluting firms may reduce the effect of strategy. This is possibly due to the already low level of emission intensity, such that the adoption of more extensive CAS may not enable significantly more reduction in emission levels. Our main analyses focus on the relationship from carbon strategy to CAS, and from carbon strategy and CAS to carbon performance. However, it is possible that two-way relationships may exist. Specifically, carbon accounting systems may also have an impact on carbon strategy. We run a lagged model for CAS and PROACT 13 and in Table 12 model 6 show that CAS is associated with PROACT positively (Coff. = 0.2991, p < 0.01), indicating that a higher quality CAS may support proactive carbon strategy in the following year. Thus, arguably, measuring carbon emissions might result in increased emissions awareness among employees and managers who, in turn, might change organisational operations and strategies over time. Hence, introducing a good CAS may lead to increased awareness, and support the move to a more proactive strategy. 12 ΣCP = 0 + 1 CAS + 2 PROACT + 3 REACT + 4 CAS * PROACT + 5 CAS * REACT Control it + FE it + it (5) 13 PROACT it (2015) = α 0 + α 1 CAS it (2014) + ∑Control it + FE it + ε it. We also run lagged models to test for the association between carbon performance (as independent variable) and carbon strategy and CAS. 14 Results show that firms with higher carbon savings may support proactive carbon strategies (Coff. = 0.1686, p < 0.1, Model 7) and a higher quality CASs (Coff. = 0.3016, p < 0.01, Model 9) in the following year. Similarly, firms with lower emission intensities adopt more proactive carbon strategies (Coff. = − 0.2.0163, p < 0.05, Model 8) and higher quality CASs (Coff. = − 1.9934, p < 0.05, Model 10). Hence, emissions reduction in the previous year provides the motivation for firms to move to proactive carbon management, in the form of more extensive strategies, or higher quality CASs. Table 9 Carbon strategy, carbon accounting system and carbon performance: both proactive and reactive strategies Table 9 reports the OLS regressions results of testing the relationship between carbon strategy, carbon accounting system and carbon performance. The dependent variable is carbon performance takes two alternative measures: CARSAV (natural logarithm of estimated annual CO2e savings i.e., metric tonnes CO2e); and INTENS (computed by totalling scope 1 and 2 scaled by revenues). All variable definitions are in Appendix 1 *, , and * represent statistical significance at the 1%, 5%, and 10% levels, respectively (two-tailed tests) CARSAV INTENS Coefficient ( Combining this with our main results, we make two complementary arguments. On the one hand, CASs and carbon strategies appear to incentivise emissions mitigation and carbon savings. The presence of a high quality CAS moderates the relationship between carbon strategy and carbon performance, enabling a stronger effect of proactive strategy on achieving annual carbon savings, whilst lessening the impact of strategy with regards to lowering emission intensity. On the other hand, improved carbon performance incentivises firms to adopt more extensive carbon strategies and Table 10 Carbon strategy, carbon accounting system and carbon performance: alternative measures of carbon performance Table 10 reports the OLS regressions results of testing the relationship between carbon strategy, carbon accounting system and carbon performance. In Panel A the dependent variable is carbon performance takes two alternative measures: CARSAV (total estimated annual CO2 e savings scaled by total assets); and INTENS (computed by totalling scope 1 and 2 scaled by total assets). In Panel B the dependent variable is carbon performance takes two alternative measures: CARSAV (total estimated annual CO2 e savings scaled by common sharesoutstanding); and INTENS (computed by totalling scope 1 and 2 scaled by common shares outstanding). All variable definitions are in Appendix 1 *, , and * represent statistical significance at the 1%, 5%, and 10% levels, respectively (two-tailed tests) Variable ( high-quality CASs. This can be driven by competitiveness and concern to differentiate in the marketplace via extensive carbon management. Finally, in order to ensure that uneven country representation in our study does not drive the results, we re-estimate the models by i) excluding USA firm-year observations; ii) excluding the top 3 countries, being USA, UK and Japan. The results reported (Model 4) in Table 13 are similar to the results reported in Table 7, in terms of both the sign and statistical significance on the test variables of interest. We, thus, conclude that excluding the top countries does not drive/affect the results. Furthermore, Bose et al., (2021) suggest that investor protection (INV_PRO) can increase carbon regulatory oversight and hence this can effect firms' incentives to manage their carbon performance. Hence, we add an additional country variable, being investor protection Table 11 Carbon strategy, carbon accounting system and carbon performance: Carbon intensive vs nonintensive firms Table 11 reports the OLS regressions results of testing the relationship between carbon strategy, carbon accounting system and carbon performance. In Panel A and B the dependent variable is carbon performance takes two alternative measures: CARSAV (natural logarithm of estimated annual CO2e savings i.e., metric tonnes CO2e); and INTENS(computed by totalling scope 1 and 2 scaled by revenues). All variable definitions are in Appendix 1 *, , and * represent statistical significance at the 1%, 5%, and 10% levels, respectively (two-tailed tests) Variable ( Table 12 Carbon strategy, carbon accounting system and carbon performance: lagged analysis All variable definitions are in Appendix 1 *, , and * represent statistical significance at the 1%, 5%, and 10% levels, respectively (two-tailed tests) Variable(s) Table 13 also suggest that firms in countries with higher investor protection are more likely to achieve carbon savings and lower emissions intensity. Endogeneity In most business studies, endogeneity is a major issue owing to omitted variables, simultaneity, and the correlation between the explanatory variables and the error term in a regression model (Li, 2016). Endogeneity leads to inconsistent and biased estimates of the explanatory variables. Li (2016) demonstrates that the GMM has the greatest correction effect on the bias, followed by instrumental variables, fixed effect models, lagged dependent variables, and the addition of more control variables. Accordingly, we re-estimated our most comprehensive model (Model 4) using the dynamic GMM as developed by Blundell and Bond (1998) and applied by others (e.g., Al-Najjar & Belghitar 2011;El Ghoul et al., 2011;Eliwa et al., 2021). The results in Table 14 show that the variable PROACTCAS is positive and significant for CARSAV (Coff. = 0.0069, p < 0.05) and negative and significant for INTENS (Coff. = − 0.0019, p < 0.01). In Table 14, the results for the control variables are broadly consistent with the main results. Overall, this suggests that the endogeneity issues are not likely to influence our main findings. Conclusions Carbon emissions bring risks and opportunities to organisations (Bebbington & Larrinaga-González, 2008;Cadez & Czerny, 2016;Bui and Villiers, 2017), and organisations adopt different strategies and environmental control systems, such as CASs. However, existing research has provided limited insights into the influence of carbon strategy and CASs in improving carbon performance (that is, reducing carbon emissions and increasing carbon savings). This paper analyses the three-way relationship between strategy-accounting-performance in the context of climate change issues, by drawing upon the CDP database for 2014 and 2015. In doing so, it provides three contributions to the literature. Firstly, a CAS is useful in achieving carbon savings and reducing emissions intensity and, hence, plays a positive role in the fight against climate change, at least at the corporate level. Furthermore, a high quality CAS is associated with both proactive and reactive strategies and, hence, supports the significant role played by CASs in implementing different strategies and initiatives undertaken by corporations. Different from prior studies that are limited to one or several countries, or examine only a few components of carbon accounting Wijethilake et al., 2016;Qian et al., 2018) we provide cross-country evidence of the association between CASs and carbon performance, using a comprehensive index of carbon accounting and an international dataset that spans 30 countries. We argue that in order to motivate high carbon performance, a high-quality comprehensive CAS needs to be properly designed and used. Our comprehensive CAS includes components such as Moderating effect of carbon accounting systems on strategy… Table 13 Carbon strategy, carbon accounting system and carbon performance: Excluding USA, excluding USA, UK and Japan, additional country-level variable, investor protection strategic planning, financial and non-financial performance measures, targets, budgets, project management methods, incentive systems and reporting. Such a comprehensive CAS will provide a basis for best practices in carbon management to be developed and disseminated. Secondly, the paper highlights the positive relationships between proactive carbon strategies and carbon performance, via enhancing carbon savings and lowering emission intensity. Whilst prior studies either imply (Clarkson et al., 2011) or examine a single country context (Moussa et al., 2020), we contribute empirical evidence in an international context of the role played by proactive carbon strategies. A carbon strategy helps develop and nurture the unique resources and capabilities that, in turn, improve carbon performance. This applies to proactive strategies that encompass strategic integration, reduction initiatives, policy engagement, value chain engagement and carbon credit origination. However, when firms adopt reactive strategies such as emissions trading and credit purchase, the association applies Table 14 Carbon strategy, carbon accounting system and carbon performance: the endogeneity tests (GMM) Table 11 reports the GMM results of testing the relationship between carbon strategy, carbon accounting system and carbon performance. The dependent variable is carbon performance takes two alternative measures: CARSAV (natural logarithm of estimated annual CO2e savings i.e., metric tonnes CO2e); and INTENS (computed by totalling scope 1 and 2 scaled by revenues). All variable definitions are in Appendix 1 , , and * represent statistical significance at the 1%, 5%, and 10% levels, respectively (two-tailed tests) only to annual carbon savings and not to current year's emission intensity. Hence, we argue that proactive carbon strategy provides a better driver for both past and future carbon performance. Thirdly this study is arguably the first to provide empirical evidence regarding the moderating impact of CASs on the strategy-performance relationship. Most prior empirical studies have found a positive association between CASs and performance, but have not considered strategy as a driver. We found that a higher quality CAS helps proactive strategy to have more pronounced impact on carbon savings among polluting firms. Hence, we tentatively argue that there is more value to be gained for polluting firms to improve the quality of their CAS, as this will allow proactive strategy to achieve more annual carbon savings. This study has four main implications for practice. First, it provides insights to managers and practitioners into the significance of a high quality CAS in pursuing a strategy. It is argued that no matter what strategy a firm pursues, a high-quality CAS is essential for its effective implementation as CAS is needed to account for and manage carbon related activities. Second, a high quality CAS also contributes to the improvement of carbon performance. This will encompass a suite of carbon measures, for example, targets, budgets, incentives, strategic planning and project management methods. In other words, the more formalised a CAS, the more likely it is that firms will achieve a stronger carbon performance. Third, proactive strategies should be pursued to achieve ongoing carbon savings and lower emission intensity. Policymakers wishing to promote carbon mitigation will need to focus on schemes or mechanisms that encourage firms to undertake proactive strategies, including strategic integration, reduction initiatives, and credit origination, rather than to participate in emissions trading or credit purchase activities, which may not have an impact on emission intensity levels. Fourth, the moderating role of CASs indicates that firms that wish to achieve performance enhancement should consider establishing an appropriate CAS, so that when used in combination with a proactive strategy, higher performance outcomes result, compared with those potentially achieved under a proactive strategy alone. This study is subject to some limitations. Firstly, we focus on disclosure-derived carbon accounting mechanisms and, hence, we cannot make assertions regarding internally derived carbon accounting; for instance, those that are not reported in the CDP, or not reported accurately. Secondly, there might be reservations regarding the accuracy of the emissions data voluntarily disclosed by firms. 15 Thirdly, we examine only those firms that responded to CDP within a limited timeframe (from 2013 to 2015). Given that reducing carbon emissions may require investments (e.g., in renewable energy to replace fossil fuel burning), a lag over several years has to be considered. A longitudinal study would therefore be needed to analyse whether or not CASs help to improve performance. Fourthly, our use of a disclosure-based database limits the insights into internal strategies and operations of organisational carbon management. We are also unable to discern the presence and use of informal controls, such as peer pressures or culture, towards carbon management objectives. Independent surveys or case studies into both responding and non-responding firms may provide interesting comparative and in-depth insights, especially regarding the process of carbon accounting and strategy. To provide more comprehensive understanding of the three-way interaction between strategy-accounting-performance in achieving the carbon management objectives of organisations, future research can address these limitations through a wider inclusion of time periods, firms, and variables.	v3-fos-license
2023-09-06T06:42:41.770Z	2023-09-04T00:00:00.000	261530976	{ "extfieldsofstudy": [ "Mathematics" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://academic.oup.com/imrn/article-pdf/2024/7/6209/57129324/rnae014.pdf", "pdf_hash": "e8af06a71b2d91ecbdd0097a3840a5b3a7aaf2ff", "pdf_src": "ArXiv", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45351", "s2fieldsofstudy": [ "Mathematics" ], "sha1": "e8af06a71b2d91ecbdd0097a3840a5b3a7aaf2ff", "year": 2023 }	pes2o/s2orc	A non-symmetric Kesten criterion and ratio limit theorem for random walks on amenable groups We consider random walks on countable groups. A celebrated result of Kesten says that the spectral radius of a symmetric walk (whose support generates the group as a semigroup) is equal to one if and only if the group is amenable. We give an analogue of this result for finitely supported walks which are not symmetric. We also conclude a ratio limit theorem for amenable groups. Introduction Let G be a finitely generated countable group.Let µ be a probability measure on G, i.e. a function µ : G → R + such that g∈G µ(g) = 1.Let S µ := {g ∈ G : µ(g) > 0}, the support of µ.We say that µ is non-degenerate if S µ generates G as a semigroup.(We do not require S µ to be finite.) Let \| • \| be a word metric on G associated to some finite generating set.(We do not assume any connection between this set and S µ .)We say that µ has finite first moment if g∈G \|g\|µ(g) < ∞ and that µ has finite exponential moment of order c > 0 if g∈G e c\|g\| µ(g) < ∞. We will be interested in the spectral radius of this random walk, defined by λ(G, µ) := lim sup n→∞ (µ * n (e)) 1/n .Clearly, 0 ≤ λ(G, µ) ≤ 1.A celebrated theorem of Kesten (which does not even require G to be finitely generated) says that if µ is symmetric then λ(G, µ) = 1 if and only if G is amenable [8].(We recall that G is amenable if and only if it admits a Banach mean, i.e. a linear functional M : ℓ ∞ (G) → R such that M (1) = 1, inf g∈G f (g) ≤ M (f ) ≤ sup g∈G f (g), and M (f g ) = M (f ), where f g (x) = f (gx).See the papers of Følner [5] and Day [2] for further discussion.)The aim of this note is to generalise Kesten's criterion to the non-symmetric case. To state our generalisation, we need to consider the abelianisation of G. Since G is finitely generated, this has a finite rank k ≥ 0. Let G ab = G/[G, G] denote the abelianisation of G and let G ab T denote the torsion subgroup of G ab .Now set G = G ab /G ab T ∼ = Z k , for some k ≥ 0, (the torsion-free part of the abelianisation) and let π : G → G be the natural projection homomorphism.Write μ = π * (µ), i.e. μ(m) = g∈G π(g)=m µ(g). The special case where µ has finite support originally appeared in Dougall-Sharp [4], where it is written in the language of subshifts of finite type and Gibbs measures. A probability measure µ (with finite first moment) is said to be centred if for each homomorphism χ : G → R, we have g∈G χ(g)µ(g) = 0. Any such homomorphism factors through G so it is easy to see that µ is centred if and only if either k = 0 or In particular, µ is centred if and only if μ is centred. If, in addition, µ has a finite exponential moment of some order then we have the following result. Corollary 1.2.Let G be a finitely group and let µ be a non-degenerate probability measure on G. Provided µ has a finite exponential moment of some order, we have λ(G, µ) = 1 if and only if G is amenable and µ is centred. Theorem 1.1 allows us to prove a ratio limit theorem for amenable groups with an explicit limit.To avoid any parity issues, it is convenient to restrict to aperiodic walks.We say that (G, µ) is aperiodic if there exists n 0 ≥ 1 such that µ * n (e) > 0 for all n ≥ n 0 .Theorem 1.3 (Ratio limit theorem).Suppose that G is a finitely generated amenable and that µ is a non-degenerate probability measure on G. Assume in addition that (G, µ) is aperiodic.Then, for each g ∈ G, where ξ ∈ R k is the unique value for which λ(G, µ) = ϕ µ (ξ). Remark 1.4.One should compare this with a theorem of Avez [1] that says that if G is amenable and µ is symmetric, non-degenerate and aperiodic then lim n→∞ µ * n (g)/µ * n (e) = 1, for all g ∈ G. Remark 1.5.It should be noted that there is no a priori mechanism to guarantee that the ratios do indeed have a limit.However, notice that if one has the ratio limits for all g ∈ G, for some ξ then G is necessarily amenable.We give the short demonstration.From the hypothesis we have for any s ∈ G, µ * n (s −1 ) µ * n (e) = e ⟨ξ,π(s)⟩ . we then have In particular ϕ µ (ξ) < ∞.We proceed with the proof assuming that ξ = 0 and deduce the general case after.Now using that µ * n (e) we see that (1.1) with ϕ µ (0) = 1 implies that lim sup n→∞ (µ * n (e)) 1/n = 1.This contradicts Day's [2] generalisation of Kesten's criterion to the random walk operator spectral radius -a consequence of which is that, for any non-degenerate probability, we have that if G is non-amenable then lim sup n→∞ (µ * n (e)) 1/n < 1. For the general case ξ ̸ = 0, we have already shown that ϕ µ (ξ) < ∞, and so μ(g) = e ⟨ξ,π(g)⟩ ϕ µ (ξ) µ(g) is a well-defined probability measure on G with ratio limits equal to one, and we again conclude that G is amenable. Let us now outline the contents of the rest of the paper.In Section 2, we recall results of Stone on random walks on Z k that are essential to the formulation of our results, and the rather general results of Gerl.In Section 3, we give a proof of Corollary 1.2 assuming Theorem 1.1.We prove Theorem 1.1 in Sections 4 and 5. Theorem 1.3 is proved in Section 6, as a consequence of equidistribution results for countable state shifts. □ We now state a ratio limit theorem due to Stone. Then, for each m ∈ Z k , Ratio limit theorems are intimately related to the existence of harmonic functions.Given a random walk (G, µ), we define the random walk operator P µ : This may be written as a convolution (Some authors define f to be λ-harmonic if µ * f = λf .)If we write h ξ (m) = e −⟨ξ,m⟩ for the limit in Proposition 2.3 then we see that the function ȟξ (m ) is the smallest value for which there is a λ-harmonic function. One may ask about ratio limit theorems on more general groups than Z k .Following earlier work by Avez [1] and Gerl [6], a rather general ratio limit theorem was proved by Gerl [7], where it is obtained as a corollary of the following limit theorem.A detailed account and discussion may be found in the recent note by Woess [18].Proposition 2.4 (Gerl's fundamental theorem [7]).Suppose that µ is a nondegenerate probability measure on G such that (G, µ) is aperiodic.Then we have Gerl used this to prove the following conditional ratio limit theorem. Proposition 2.5 (Gerl's ratio limit theorem [7]).Suppose that µ is a non-degenerate probability measure on G such that (G, µ) is aperiodic.Suppose there is a set for some subsequence for all g ∈ G. In particular, if we have uniqueness of a λ(G, µ)-harmonic function for (G, µ) then we know the ratio limit theorem holds.The advantage of our Theorem 1.3, for amenable groups, is that we don't consider arbitrary harmonic functions instead we directly work with functions coming from the abelianisation. Proof of Corollary 1.2 In this section we prove Corollary 1.2, assuming Theorem 1.1.We note that ϕ µ = ϕ μ, so we can use Lemma 2.1. Since μ is non-degenerate, ϕ µ (v) is strictly convex on the set where it is finite.The hypothesis of a finite exponential moment implies that ϕ µ (v) is finite and differentiable in some neighbourhood of 0 ∈ R k .Therefore, ϕ µ (v) has its unique minimum at v = 0 if and only if ∇ϕ µ (0) = 0. Suppose that λ(G, μ) = 1; then, by Lemma 2.1, ϕ µ has its minimum at 0 and so i.e. µ is centred.On the other hand, if λ(G, μ) < 1 then, again by Lemma 2.1, the unique minimum of ϕ µ is not at 0 and so In this section we show that if λ(G, µ) = λ(G, μ) then G is amenable.(In Kesten's original theorem, this was the harder direction but in our case it is the easier implication.)Noting that ϕ µ = ϕ μ, recall from Section 2 that there is a unique We define a new probability measure µ ξ on G (analogous to the measure ω ξ on Z k in the proof of Corollary 2.2) by Proof of Theorem 1.1 ( ⇐= ).Assume that G is non-amenable.By Theorem 1 of Day [2] (see also Theorem 5.4 of Stadlbauer [13]), we see that the probability measure µ ξ satisfies lim sup Unpicking the definitions, µ * n ξ (e) = ϕ µ (ξ) −n µ * n (e).Hence lim sup n→∞ (µ * n (e)) 1/n < ϕ µ (ξ).□ Proof of Theorem 1.1 ( =⇒ ) In this section we will show the harder implication that if G is amenable then λ(G, μ) ≤ λ(G, µ), and hence that λ(G, μ) = λ(G, µ).We remark that the proof given here is significantly easier and more direct than the one we gave in [4].Following that proof would introduce a family of measures, indexed by g ∈ G, on the space S N µ × G, each describing the paths that visit S N µ × {g}.These measures (which are also analysed in [14]) are not required here. Let us begin by emphasising the following: though we know that μ has a λ(G, μ)harmonic function it plays no role in this part of proof !The first element of the proof is the following proposition.Subsequently, the rest of the section will be devoted to showing that show its hypothesis is satisfied with λ = λ(G, µ). Proposition 5.1.If there is a homomorphism h : G → R >0 , the multiplicative group of positive real numbers, such that, for all n ∈ N, Proof.Suppose such a homomorphism h exists.Since h is positive so we can throw away the terms where −π(g) ̸ = 0 and obtain Hence, for any δ < 1, This says that λ(G, μ) ≤ λδ −1 .Since we can take δ arbitrarily close to 1 we are done.□ We view λ(G, µ) as a convergence parameter for the series This formulation is reminiscent of the Poincaré series used in the construction of Patterson-Sullivan measures on the limit sets of Kleinian groups and more general limit sets and, indeed, we employ ideas from this theory The most relevant reference here is Roblin [9], which covers the basic tools of Patterson-Sullivan theory and the amenability "trick" we will use in the proof of Proposition 5.6.The series ζ(t) does not necessarily diverge at t = λ(G, µ) but one can modify the series, in a controlled way, to guarantee divergence at this critical parameter.The following appears as Lemma 3.2 in Denker and Urbanski [3] (see also [14]). n=1 be the sequence given by Lemma 5.2 for the series with a n = µ * n (e).We prefer to use c n = 1/b n , a decreasing sequence.Note that we have lim n→∞ c n−r /c n = 1 for all r ∈ N. We will work with a modified series ζ e c (t) defined by and also, for each g ∈ G, the series Proof.We begin by observing that, for every g, h ∈ G, we have and we may choose k ≥ 1 for which µ * k (gh −1 ) > 0. This gives us the inequality Since the numbers c n are positive and, for a fixed k, lim n→∞ c n /c n+k = 1, we have inf n c n /c n+k > 0. Hence . Since g, h are arbitrary the lemma follows.□ By the previous lemma and a standard diagonal argument, there exists a sequence t n → λ(G, µ) for which the following limits are well-defined for all g ∈ G.A crucial observation is the following. Lemma 5.4.For any r ∈ N, we have Proof.Fix r ∈ N and let ϵ > 0. Since lim n→∞ c n−r /c n = 1, we can choose n 0 such that 1 − ϵ ≤ c n−r /c n ≤ 1 + ϵ, for all n > n 0 .We will also use that Setting using that the terms in the series are non-negative.This gives and, since ϵ is arbitrary, For the lower bound, we have This gives λ r H(g) ≥ s∈G µ * r (s)H(s −1 g). □ Lemma 5.4 gives us the following corollary. Corollary 5.5.For any fixed γ ∈ G, we have Proof.Given γ ∈ G, we can find x 1 , . . ., x k ∈ S µ , for some k ≥ 1, such that and so sup g∈G H(γg)/H(g) is finite.Now we put γg on the right hand side and choose y 1 , . . ., y ℓ ∈ S µ such that and so inf g∈G H(γg)/H(g) is positive.□ We are now ready to use the amenability of G.We use the existence of a Banach mean on ℓ ∞ (G) to "average over g" in the last lemma. Proposition 5.6. There is a homomorphism Proof.Since any homomorphism h : G → R >0 factors through G, it suffices to show that there is a homomorphism h : G → R >0 such that, for all n ∈ N, we have (5.1) Let M be a Banach mean on ℓ ∞ (G).Jensen's inequality says that if φ is convex then (This is more familiar when the linear functional is integration with respect to a probability measure.One can check that we only need monotonicity, finite additivity, and the unit normalisation M (1) = 1.)We apply this to the function g → (H(γg)/H(g)) in ℓ ∞ (G).(Note we have used Corollary 5.5 to know that g → log(H(γg)/H(g)) is in ℓ ∞ (G).)Thus we obtain We will show that h satisfies (5.1), recalling that M is only finitely additive.Let {g k } ∞ k=1 be an enumeration of G and, for N ≥ 1, let G N = {g 1 , . . ., g N }.Lemma 5.4 gives us that, for any n ≥ 1 and any N ≥ 1, we have Taking the supremum over N gives (5.1).It remains to show that h is a homomorphism.Notice that using translation invariance of M .The conclusion follows.□ Combining Proposition 5.1 and Proposition 5.6 shows that if G is amenable then λ(G, μ) ≤ λ(G, µ). Equidistribution and proof of the ratio limit theorem In this section we use Theorem 1.1 to prove a ratio limit theorem for amenable groups, Theorem 1.3.Our arguments will also give a new proof of Proposition 2.4 in this setting.Our approach is based on weighted equidistribution results for countable state shift spaces.Suppose that G is amenable, that µ is non-degenerate and that (G, µ) is aperiodic.We let λ denote the common value given by Theorem 1.1.As above, µ ξ (g) = λ −1 h(g)µ(g), where h(g) = e ⟨ξ,π(g)⟩ . We consider the sequence space Σ = S N µ and let σ : Σ → Σ be the shift map: . ., n}.We give Σ the topology generated by cylinder sets (which are both open and closed). We denote by ν ξ the Bernoulli measure on Σ given by s=(s1,...,sn)∈Λn where we use the notation s ∞ ∈ Σ to mean the one-sided infinite concatenation of s = (s 1 , . . ., s n ) and δ s∞ denotes the Dirac measure at this point.We remark that we also have m n = 1 µ * n (e) s=(s1,...,sn)∈Λn µ(s 1 ) • • • µ(s n )δ s∞ but we do not use this formula.We will need to explicitly evaluate the measures m n on cylinder sets.Lemma 6.1.For a cylinder set [u 1 , . . ., u k ] we have that, for n > k, , Proof.This is a straightforward calculation.For n > k, Proof.We know that lim sup n→∞ (µ * n ξ (e)) in particular the limit exists and is equal to the limsup.□ In order to show the required convergence for the m n , we introduce some ideas and terminology from thermodynamic formalism and large deviation theory.A function φ : Σ → R is called locally Hölder continuous if (6.1) sup for some C > 0 and 0 < θ < 1, for all n ≥ 1.For a locally Hölder continuous function φ : Σ → R, we define the Gurevič pressure P G (φ) by (The original definition given by Sarig in [10] is somewhat different, and only requires, (6.1) to hold for n ≥ 2, but, by Corollary 1 of [11], the above formula gives the Gurevič pressure in our setting.)We now fix ), so that, in particular, P G (φ) = 0. Let χ be the indicator function of some cylinder.We can easily calculate from the definition that, for t ∈ R, P G (φ + tχ) ≤ max{0, \|t\|} for all t ∈ R. Hence, by Corollary 4 of [11], t → P G (φ + tχ) is real analytic for t ∈ R and, by Theorems 6.12 and 6.5 of [12], (6.2) dP (φ + tχ) dt t=0 = χ dν ξ . (The same discussion remains true if χ is replaced with any bounded locally Hölder function but indicator functions of cylinders are sufficient for our purposes.)For s ∈ S n ν , let τ s,n denote the orbital measure Following, Theorem 7.4 of [12], we have the following large deviations bound.Proposition 6.3.Given ϵ > 0, there exists C > 0 and η(ϵ) > 0 such that Proof.The proof is standard but we include it for completeness.We consider s ∈ S n µ such that χ dτ s,n > χ dν ξ + ϵ and χ dτ s,n < χ dν ξ − ϵ separately.For t > 0, we have Using P G (φ) = 0 and (6.2), we see that, for sufficiently small t > 0, we have Similarly, for t < 0, lim sup and, for sufficiently small t < 0, this upper bound is negative.Combing these estimates gives the result.□ Since Λ n ⊂ S n µ , we have the following immediate corollary.Corollary 6.4.For ϵ > 0, we have We can now establish the ratio limit theorem for amenable groups.as n → ∞. □	v3-fos-license
2020-11-05T09:10:36.670Z	2020-10-31T00:00:00.000	226250301	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2072-6694/12/11/3213/pdf", "pdf_hash": "b70c9bf6b4ccf90dfd9d437d761bde4ea14a2ec1", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45355", "s2fieldsofstudy": [ "Medicine", "Biology" ], "sha1": "40a54d73703391c0cd176d8b28e6f9b70867522c", "year": 2020 }	pes2o/s2orc	The Prognostic Role of Baseline Metabolic Tumor Burden and Systemic Inflammation Biomarkers in Metastatic Castration-Resistant Prostate Cancer Patients Treated with Radium-223: A Proof of Concept Study Simple Summary Radium-223 is an alpha-emitting radioisotope that selectively binds to increased bone turnover areas, such as metastatic sites, acting as a bone-seeking calcium mimetic drug. Its therapeutic function in metastatic castration-resistant prostate cancer patients relies on its capability to prolong overall survival, improve quality of life, and delay the first skeletal-related event. However, in the last few years, many studies showed that the survival benefit in the real-life patients might be lower than that initially reported, probably due to a suboptimal selection of patients with poorer prognostic clinical characteristics. In this scenario, it has emerged the urgent need for the identification of reliable biomarkers able to potentially identify patients most likely to benefit from Radium-223 since baseline. With this aim, this preliminary study is the first to combine the prognostic power of baseline FDG-PET/CT and systemic inflammation indexes in a cohort of metastatic castration-resistant prostate cancer patients undergoing Radium-223 administration. Abstract Over the last years has emerged the urgent need for the identification of reliable prognostic biomarkers able to potentially identify metastatic castration-resistant prostate cancer (mCRPC) patients most likely to benefit from Radium-223 (Ra-223) since baseline. In the present monocentric retrospective study, we analyzed the prognostic power of systemic inflammation biomarkers and 18F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography (FDG-PET)-derived parameters and their potential interplay in this clinical setting. The following baseline laboratory parameters were collected in 59 mCRPC patients treated with Ra-223: neutrophil-to-lymphocyte ratio (NLR), derived NLR (dNLR), lymphocyte-to-monocyte ratio (LMR), platelets-to-lymphocyte ratio (PLR), and systemic inflammation index (SII), while maximum Standardized Uptake Value, Metabolic Tumor Volume (MTV), and Total Lesion Glycolysis (TLG) were calculated in the 48 of them submitted to baseline FDG-PET. At the univariate analysis, NLR, dNLR, MTV, and TLG were able to predict the overall survival (OS). However, only NLR and MTV were independent predictors of OS at the multivariate analysis. Additionally, the occurrence of both increased NLR and MTV at baseline identified mCRPC patients at higher risk for lower long-term survival after treatment with Ra-223. In conclusion, the degree of systemic inflammation, the quantification of the metabolically active tumor burden and their combination might represent potentially valuable tools for identifying mCRPC patients who are most likely to benefit from Ra-223. However, further studies are needed to reproduce these findings in larger settings. Introduction Bone metastases affect more than 90% of patients with metastatic castration-resistant prostate cancer (mCRPC) patients and 20-50% of them develop skeletal-related events (SREs), which represent the main cause of impaired quality of life and death [1]. Radium-223 (Ra-223) is an alpha-emitting radionuclide which selectively binds to areas of increased bone turnover, such as metastatic sites, acting as a bone-seeking calcium mimetic drug [2]. Its short-range high-energy emission induces breaks in double-strand DNA filaments and targets tumor cell death [3]. The phase III Alpharadin in Symptomatic Prostate Cancer Patients (ALSYMPCA) trial investigated Ra-223 compared to a placebo in mCRPC patients with symptomatic bone metastases, limited lymph node metastases (<3 cm), and no visceral metastases [4]. In the ALSYMPCA trial, Ra-223 was demonstrated to prolong overall survival (OS), improve quality of life, and delay the first SRE [4]. According to these results, Ra-223 was subsequently approved by the Food and Drug Administration (FDA) for mCRPC patients. However, in the last few years, many retrospective studies showed that the survival benefit in real-life patients might be lower than that reported in the ALSYMPCA trial, probably due to a suboptimal selection of patients with poorer prognostic clinical characteristics [5][6][7]. Furthermore, in 2018 the European Medicine Agency (EMA) restricted the use of Ra-223 to patients with more than six osteoblastic lesions in progression after at least two prior lines of systemic therapies for mCRPC or ineligible for any available systemic mCRPC treatment [8]. However, the later timing of Ra-223 administration in the patients' clinical history might further negatively affect OS [9]. In this scenario, there is an urgent need to identify reliable biomarkers potentially able to improve the selection of patients most likely to benefit from Ra-223 treatment. Peripheral blood inflammatory parameters, such as neutrophils-to-lymphocytes (NLR), have been shown to significantly correlate with survival outcomes and therapeutic response in various cancers, including mCRPC, as potential cancer inflammation-associated markers [14][15][16][17]. These biomarkers are currently of great interest for their ready and easy accessibility in the clinical practice and their transversal role in many types of tumors and cancer treatments [15]. However, few studies have investigated their prognostic role in patients treated with Ra-223 [5,18,19]. On the other hand, we recently showed that baseline 18 F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography (FDG-PET) could stratify OS in a cohort of mCRPC patients who are candidates for Ra-223 [20]. However, the comparison and the potential interplay between peripheral blood inflammatory biomarkers and metabolic FDG-PET-derived parameters still need to be investigated in this clinical setting. Therefore, in the present proof of concept study, we analyzed the prognostic power of baseline inflammatory and functional imaging biomarkers and their potential interplay to identify mCRPC patients most likely to benefit from Ra-223. Study Population We performed a retrospective monocentric analysis of all consecutive mCRPC patients treated with Ra-223 from September 2016 to February 2020 at the IRCCS Ospedale Policlinico San Martino of Genova, Italy. The retrospective analysis was conducted under the Declaration of Helsinki, Good Clinical Practice, and local ethical and legal regulations. According to our standard procedure, all patients signed a written informed consent form, encompassing the use of anonymized data for retrospective research purposes before each imaging procedure and each Ra-223 administration. CRPC was defined as a serum testosterone level of <50 ng/dL following surgical or pharmaceutical castration. All patients fulfilled the inclusion criteria for Ra-223 therapy and were treated according to the standard Ra-223 regimen encompassing six intravenous administrations every four weeks (55 KBq/kg) [21]. According to the established guidelines for patient selection, a bone marrow reserve fulfilling the hematologic criteria necessary to administer Ra-223 was verified [21]. During the four weeks preceding the first Ra-223 administration, each patient underwent a contrast-enhanced CT and bone scan to select mCRPC patients with symptomatic bone metastases in the absence of visceral involvement (with the exception for lymph nodes with maximum diameter < 3 cm). In the same time interval, recruited patients were submitted to FDG-PET for prognostic purposes in agreement with the emerging prognostic role of this tool in patients with mCRPC, and in accordance with the national guidelines by the Italian Association of Medical Oncology (AIOM) [22,23]. According to our standard procedure, complete blood cell count, serum chemistry, prostate-specific antigen (PSA), alkaline phosphatase (ALP), and lactate dehydrogenase (LDH) were assessed at baseline and on the same day of each Ra-223 administration. During Ra-223 administration, patients continued androgen deprivation therapy and received the best standard of care, including antiresorptive agents (bisphosphonates/denosumab) and antalgic therapy [21]. The concomitant treatment with Abiraterone and Enzalutamide was not allowed [21]. Imaging Procedures and Images Analyses FDG-PET was performed according to the European Association of Nuclear Medicine (EANM) Guidelines [24]. PET/CT scans were performed using a 16-slices PET/CT hybrid system (Hirez-Biograph 16, Siemens Medical Solutions, USA). FDG-PET images were interpreted in consensus by three expert nuclear medicine physicians (M.B.; M.I.D.; A.M.) blinded to contrast-enhanced CT and bone scan results. From the attenuation corrected FDG-PET images, the maximum standardized uptake value (SUVmax) of the hottest bone lesion was obtained in the transaxial view. Further, a volume of interest was drawn using an SUV-based Cancers 2020, 12, 3213 4 of 16 automated contouring program (Syngo Siemens workstation, Siemens Medical Solutions, USA) with an isocounter threshold based on 40% of the SUVmax, as previously recommended [25]. Total Metabolic Tumor Volume (MTV) was obtained by the sum of all skeletal and extra-skeletal lesions. Total Lesion Glycolysis (TLG) was calculated as the sum of the product of MTV of each lesion, and the SUVmean value, which, in turn, was automatically calculated within each single MTV. Aiming to analyze the interobserver variation, a second expert PET reader (S.M.) measured MTV and TLG independently from the first group of observers. Statistical Analysis Descriptive analyses were conducted using percentages for binary variables and means/medians for continuous variables, reporting their dispersion values. Non-parametric Bland-Altman plots were used to evaluate median bias ad limits of agreement (2.5% and 97.5% percentiles) for MTV and TLG values measured by the two groups of PET readers [26]. A linear regression was applied to verify the eventual occurrence of proportional biases. To assess the association of parameters, biomarkers, and clinical characteristics (independent variables) with OS, univariable and multivariable Cox regression model were used. All parameters, biomarkers, and clinical characteristics, with a p-value < 0.10 at univariable analysis were selected for the multivariable analysis. Only those with a p-value < 0.10 were maintained in the final multivariable model. OS was calculated from the start of treatment to death from any cause, censored at last follow-up for patients who were alive, and was estimated by mean of the Kaplan-Meier (KM) approach. The multivariable model was performed both on the cohort with complete cases for inflammatory and FDG-PET parameters and for the whole cohort of patients after multiple imputations of missing values for FDG-PET parameters (MTV, TLG). Multiple imputation was performed using an iterative multivariate method based on chained equations. Eleven imputations were performed. To report KM curves of OS, continuous parameters were binarized. To find the best cut-off value, the Youden index from the ROC curve for survival data at 24 months was used. Hazard-ratios (HR) for Cox regression models were reported together with a 95% confidence interval (CI) and p-value. Due to a highly skewed distribution, for MTV and TLG, the log-transformed values were used in the analyses for a better interpretation. Gleason Score at diagnosis was categorized into three classes for clinical interpretation, as previously described [27]. Similarly, bone scan lesions were categorized in <6, 6-20, and >20, as performed in a sub-analysis of the ALSYMPCA study [4]. To assess the ability of each continuous parameter to discriminate between dead and censored patients, the Harrell's c-index for censored data was calculated for all inflammatory and FDG-PET parameters. Further, a calibration plot to assess the accuracy of the prediction for an individual patient according to single parameters and multivariate model was realized. A joint test was performed to investigate the overall evidence for linear miscalibration. Patients' and Treatment Characteristics Fifty-nine mCRPC patients treated with Ra-223 were included in the analysis. Of these, 48 (81.4%) had all complete data on both inflammatory biomarkers and FDG-PET parameters. The patient, tumor, and treatment characteristics are summarized in Table 1. All patients had a histological diagnosis of prostate cancer with a median Gleason score (GS) of 8 (range 5-10) with a GS ≥ 8 in 51% of patients; 47% of patients had metastatic disease at diagnosis. All patients underwent CT and bone scans at baseline, while 81% of the enrolled patient underwent FDG-PET at the same timepoint. CT scan revealed the occurrence of lymph node metastases in 27% of patients, while bone scan showed the presence of <6, 6-20, and >20 bone lesions in 16%, 40%, and 44% of patients, respectively. Baseline peripheral blood results were available in 100% of patients. The median number of Ra-223 cycles was 5 (range 1-6), 78% and 39% of patients completed three and all six cycles, respectively. Eight patients (13%) are currently still in treatment with Ra-223 at the time of data analysis. Thirty-five patients (59%) received chemotherapy before Ra-223, of which 19 patients (32%) received docetaxel only and 16 patients (27%) both docetaxel and cabazitaxel. Characteristics of patients with complete cases were similar to those of the whole cohort. Interobserver Agreement between PET Readers There was good agreement between the two groups of observers for measuring MTV and TLG. Bland-Altman plots showed a median difference of 0.26 (limits of agreements: −159.67-160. 19) and −15.35 (limits of agreements: −397.07-366.37), respectively (see also Figure S1). No proportional biases were observed (p = ns for both). Systemic Inflammation Indexes and FDG-Derived Parameters in the Prediction of OS. All patients included in the study were assessable for survival analysis and were followed-up for a median of 10 months. The median OS (mOS) was 11.6 months (95% CI: 9.1-14.7) in the whole cohort and 10.2 (95% CI: 7.1-14.7) in 48 patients with complete data. OS was 78.8% (CI: 65.6-87.3) in the whole cohort and 73.4% (58.4-84.2) in the reduced sample at 6 months, while it was 46.5% (CI: 32.4-59.4) in the whole cohort and 46.2% (CI: 30.6-60.5) in the reduced sample, respectively. Figure 1 shows the Kaplan-Meier survival function of the study cohort with all complete data (n = 48). Results from Kaplan-Meier analyses and univariable Cox regression analyses are reported in Figures 2 and 3 and in Table 2, respectively. Univariate and multivariate analyses were conducted on the complete cases set (n = 48). A further multivariate analysis was performed considering all sample (n = 59) after multiple imputation of missing data for FDG-PET parameters. Interobserver Agreement between PET Readers There was good agreement between the two groups of observers for measuring MTV and TLG. Bland-Altman plots showed a median difference of 0.26 (limits of agreements: −159.67-160. 19) and −15.35 (limits of agreements: −397.07-366.37), respectively (see also Figure S1). No proportional biases were observed (p = ns for both). Systemic Inflammation Indexes and FDG-Derived Parameters in the Prediction of OS. All patients included in the study were assessable for survival analysis and were followed-up for a median of 10 months. The median OS (mOS) was 11.6 months (95% CI: 9.1-14.7) in the whole cohort and 10.2 (95% CI: 7.1-14.7) in 48 patients with complete data. OS was 78.8% (CI: 65.6-87.3) in the whole cohort and 73.4% (58.4-84.2) in the reduced sample at 6 months, while it was 46.5% (CI: 32.4-59.4) in the whole cohort and 46.2% (CI: 30.6-60.5) in the reduced sample, respectively. Figure 1 shows the Kaplan-Meier survival function of the study cohort with all complete data (n = 48). Results from Kaplan-Meier analyses and univariable Cox regression analyses are reported in Figures 2,3 and in Table 2, respectively. Univariate and multivariate analyses were conducted on the complete cases set (n = 48). A further multivariate analysis was performed considering all sample (n = 59) after multiple imputation of missing data for FDG-PET parameters. Lower ECOG PS and ALP, as well as the absence of pathological lymph nodes, were associated with higher OS. Among systemic inflammation indexes, only NLR and dNLR reached the statistical significance at the univariate analysis (Table 2, Figure 2). In both cases, higher OS was observed for lower values of these systemic inflammation parameters. Similarly, lower MTV and TLG correlated with an increased OS (Table 2, Figure 3). Low to moderate c-index, testing the discriminative ability, was observed for almost all inflammation parameters while MTV and TLG showed a higher c-index of 0.75. Both these parameters also seemed to have good accuracy in prediction ( Figure 4) as also confirmed by a not significant result of the test for miscalibration (p-value = 0.49 for MTV and 0.64 for TLG). Low to moderate c-index, testing the discriminative ability, was observed for almost all inflammation parameters while MTV and TLG showed a higher c-index of 0.75. Both these parameters also seemed to have good accuracy in prediction ( Figure 4) as also confirmed by a not significant result of the test for miscalibration (p-value = 0.49 for MTV and 0.64 for TLG). The same parameters with prognostic value at the univariate analysis (apart from ALP, dNLR, and TLG) remained independently associated at the multivariate analysis for OS. The Harrell's C- The same parameters with prognostic value at the univariate analysis (apart from ALP, dNLR, and TLG) remained independently associated at the multivariate analysis for OS. The Harrell's C-index for this multivariable model was 0.81 in the reduced cohort (without imputation of missing value) and 0.79 in the whole cohort where FDG-PET parameters were imputed for the missing 11 patients. The Combination of Systemic Inflammation Indexes and FDG-Derived Parameters in the Prediction of OS Baseline NLR and MTV, both independently associated with OS in the Cox proportional hazard analyses, were thus combined. This allowed us to define an immune-metabolic-prognostic index (IMPI), as previously described by Castello et al. [28]. The combination of the above-mentioned parameters allowed us to identify three groups with different risk as it follows: low risk (neither NLR Figure 5. The prognostic power of IMPI was confirmed including this score in a multivariable model incorporating ECOG-PS, and lymph node metastases (p = 0.001). Discussion Ra-223 is one of the therapeutic options for mCRPC patients and was approved after the survival benefit observed in the phase III ALSYMPCA trial (mOS 15 versus 11 months) [4]. Nonetheless, in real-life experience, lower survival outcomes (mOS ranging from 8 to 13 months) compared to the results of the registration trial were observed [5]. This weaker survival benefit might be at least partially related to the suboptimal patients' selection process, which was further complicated by the restriction of the use of Ra-223 started in 2018 [29]. In fact, mCRPC candidates to Ra-223 therapy have weaker clinical characteristics compared to those included in clinical trials. Moreover, among the treatment options for mCRPC patients, the right collocation of Ra-223 treatment is not wellestablished as no comparative and sequential clinical trials are currently available [19]. There is, therefore, an unmet need to identify baseline clinical, biochemical, or imaging biomarkers able to improve the prognostic stratification of patients undergoing Ra-223. In recent years, many studies tried to identify biomarkers to better select patients most likely to benefit from Ra-223 and, therefore, to optimize treatment strategies. These are important to gain as much as possible in efficacy with few side effects, to improve survival outcomes of mCRPC and healthcare costs. Widely studied clinical variables with prognostic value in Ra-223 patients included ECOG PS, previous lines of therapy and prior chemotherapy [6,7,12,30]. Furthermore, the number of Ra-223 administered cycles was associated with OS [30,31]. Among laboratory variables, baseline PSA, LDH, and, especially, ALP (as an indirect index of disease burden) and hemoglobin (as an index of bone marrow reserve) have been shown to provide relevant prognostic insights in these patients [6,7,11,12,30,32]. Similar to previous studies, in our patient's cohort we observed a lower mOS compared to that Discussion Ra-223 is one of the therapeutic options for mCRPC patients and was approved after the survival benefit observed in the phase III ALSYMPCA trial (mOS 15 versus 11 months) [4]. Nonetheless, in real-life experience, lower survival outcomes (mOS ranging from 8 to 13 months) compared to the results of the registration trial were observed [5]. This weaker survival benefit might be at least partially related to the suboptimal patients' selection process, which was further complicated by the restriction of the use of Ra-223 started in 2018 [29]. In fact, mCRPC candidates to Ra-223 therapy have weaker clinical characteristics compared to those included in clinical trials. Moreover, among the treatment options for mCRPC patients, the right collocation of Ra-223 treatment is not well-established as no comparative and sequential clinical trials are currently available [19]. There is, therefore, an unmet need to identify baseline clinical, biochemical, or imaging biomarkers able to improve the prognostic stratification of patients undergoing Ra-223. In recent years, many studies tried to identify biomarkers to better select patients most likely to benefit from Ra-223 and, therefore, to optimize treatment strategies. These are important to gain as much as possible in efficacy with few side effects, to improve survival outcomes of mCRPC and healthcare costs. Widely studied clinical variables with prognostic value in Ra-223 patients included ECOG PS, previous lines of therapy and prior chemotherapy [6,7,12,30]. Furthermore, the number of Ra-223 administered cycles was associated with OS [30,31]. Among laboratory variables, baseline PSA, LDH, and, especially, ALP (as an indirect index of disease burden) and hemoglobin (as an index of bone marrow reserve) have been shown to provide relevant prognostic insights in these patients [6,7,11,12,30,32]. Similar to previous studies, in our patient's cohort we observed a lower mOS compared to that reported in the ALSYMPCA trial [4]. Moreover, the independent prognostic value of baseline clinical variables such as the ECOG PS, the presence of lymph node metastases and ALP levels was confirmed. Unlike previous studies, we extended our analysis to systemic inflammatory biomarkers as well as to FDG-PET-derived parameters. Tumor microenvironment and systemic inflammation are known to influence therapeutic response and clinical outcomes [33,34]. Hence, many inflammatory biomarkers are currently under investigation as tools to predict the therapeutic effect or prognosis in different types of advanced cancer [35]. NLR is the most studied, and it is widely established that higher levels of NLR predict poor OS regardless of the tumor type, stage and treatment [14,36]. Other types of inflammatory biomarkers were also assessed for their prognostic role in cancer [37][38][39][40]. Among them, NLR, PLR and SII have been shown to be prognostic in mCRCP patients treated with both chemotherapy or new-generation hormonal agents [16,17,[41][42][43][44][45]. However, few data on peripheral blood biomarkers are available in mCRPC patients treated with Ra-223. In the last years, few real-world studies showed a reliable, independent disease-related prognostic power of NLR in this setting, mainly related to the prediction of OS and, to a lesser extent, of PFS [5,19,46]. To the best of our knowledge, the present study represents the first analysis of different types of inflammatory biomarkers as prognostic factors in mCRPC patients undergoing Ra-223. However, while both NLR and dNRL were characterized by a prognostic power at the univariate analysis, only a lower baseline NLR was independently associated with longer OS. On the other hand, FDG-PET-derived parameters displaying the extent (MTV) and the intensity (TLG) of the metabolically active disease burden were predictors of OS. This result reproduces our previous study, which, however, was conducted on a smaller patient sample [20]. Furthermore, in the present study, the multivariate Cox regression analysis showed that the prognostic power of MTV was independent of the one provided by the degree of systemic inflammation. On the pathophysiological ground, obtained results might improve the comprehension of the still poorly defined molecular mechanisms underlying FDG accumulation in advanced CRPC patients. Indeed, while FDG-PET is not useful in naïve prostate cancer as it shows low FDG-avidity, CRPC patients are characterized by higher levels of FDG uptake, particularly in chemotherapy-refractory patients [47]. The progressive increased FDG avidity in the later stages of CRPC might be explained (at least theoretically) by two different mechanisms. On one side, the overexpression of GLUT-1 in cell membranes and the enhanced Warburg effect characterizing PC cells in advanced stages might justify this phenomenon [48,49]. On the other hand, emerging data support the role of local inflammation and, in particular, of FDG-avid macrophage and lymphocyte recruitment in the tumor microenvironment, as tumor-promoting factors driving PC from the hormone-sensitive stage to refractivity [50,51]. The observed independence between the prognostic power of FDG-PET imaging and systemic inflammation indexes (which roughly measure the degree of tumor inflammation) might imply the prevalence of the former rather than the latter pathway as the underlying mechanism of FDG uptake in mCRPC. In this scenario, the detection of FDG-avid de-differentiated (eventually low osteoblastic) metastatic disease might mirror greater tumor aggressiveness, possibly predicting the lower Ra-223 accumulation and the consequent low-response rate, regardless of the systemic inflammation state (Figure 6) [20,52]. The same considerations also allow us to interpret the observed capability of the composite of pretreatment NLR and MTV (termed IMPI) to stratify the OS. Indeed, patients at the higher IMPI risk class might be characterized by the occurrence of both tumor microenvironment inflammation and de-differentiation, leading to a worst long-term survival and Ra-223 treatment outcome. Our study has several limitations. First, a major limitation is represented by its retrospective and monocentric nature and the consequent low number of patients analyzed. Indeed, the observed prognostic inadequacy of the systemic inflammatory indexes other than NLR (and its derived counterpart) might be related to the present study's low statistical power. On the other hand, in a subgroup of 11 patients, baseline FDG-PET was not performed. We tried to overcome this limitation through the imputation of missing values. However, the current data should be considered preliminary, while a better-defined comparison between these biomarkers needs to be assessed in a larger multicentric setting. According to this, we are currently planning a retrospective multicenter study to evaluate further the prognostic and predictive value of peripheral blood biomarkers and FDG-PET in Ra-223-treated patients. In this setting, also the immune-metabolic-prognostic index will require further validation. However, despite the sample of our patients was small, it was highly representative of the population enrolled in the ALSYMCA trial [4] as it can be observed comparing the clinical characteristics of the two studies. Similarly, the prognostic role of well-known prognostic factors (ECOG, ALP, lymph node metastases) was confirmed anyway, as further proof of its representativeness. Lastly, the monocentric nature of the analyses may also represent one of the strengths of this study. In fact, all the enrolled patients were submitted to FDG imaging using the same PET/CT scanner, avoiding the possible influence of the inter-scanner variability on PET results, possibly hampering SUVmax, MTV, and TLG reproducibility [53]. Besides the dimension of the patient's cohort, the few clinical collected variables might also represent a potential limitation of the present study. According to this, due to the intrinsic limitations of retrospective data collection, the steroid use by each patient before Ra-223 administration was not recorded and considered as a possible confounding factor. Indeed, corticosteroids might have increased baseline NLR in some patients, introducing potential bias in the data interpretation. However, Lorente et al. previously reported an independent association between baseline NLR and OS in a cohort of mCRPC patients undergoing second-line chemotherapy, regardless of the corticosteroid use at baseline [54]. Moreover, the use of corticosteroids for palliation of symptoms in advanced mCRPC with boneinvolvement is well established in clinical practice, and this characteristic of the study cohort highly reflects the setting of the real-world. Larger multicentric studies are required to disclose the robustness of NLR concerning steroid administration in mCRPC patients undergoing Ra-223. Our study has several limitations. First, a major limitation is represented by its retrospective and monocentric nature and the consequent low number of patients analyzed. Indeed, the observed prognostic inadequacy of the systemic inflammatory indexes other than NLR (and its derived counterpart) might be related to the present study's low statistical power. On the other hand, in a subgroup of 11 patients, baseline FDG-PET was not performed. We tried to overcome this limitation through the imputation of missing values. However, the current data should be considered preliminary, while a better-defined comparison between these biomarkers needs to be assessed in a larger multicentric setting. According to this, we are currently planning a retrospective multicenter study to evaluate further the prognostic and predictive value of peripheral blood biomarkers and FDG-PET in Ra-223-treated patients. In this setting, also the immune-metabolic-prognostic index will require further validation. However, despite the sample of our patients was small, it was highly representative of the population enrolled in the ALSYMCA trial [4] as it can be observed comparing the clinical characteristics of the two studies. Similarly, the prognostic role of well-known prognostic factors (ECOG, ALP, lymph node metastases) was confirmed anyway, as further proof of its representativeness. Lastly, the monocentric nature of the analyses may also represent one of the strengths of this study. In fact, all the enrolled patients were submitted to FDG imaging using the same PET/CT scanner, avoiding the possible influence of the inter-scanner variability on PET results, possibly hampering SUVmax, MTV, and TLG reproducibility [53]. Besides the dimension of the patient's cohort, the few clinical collected variables might also represent a potential limitation of the present study. According to this, due to the intrinsic limitations of retrospective data collection, the steroid use by each patient before Ra-223 administration was not recorded and considered as a possible confounding factor. Indeed, corticosteroids might have increased baseline NLR in some patients, introducing potential bias in the data interpretation. However, Lorente et al. previously reported an independent association between baseline NLR and OS in a cohort of mCRPC patients undergoing second-line chemotherapy, regardless of the corticosteroid use at baseline [54]. Moreover, the use of corticosteroids for palliation of symptoms in advanced mCRPC with bone-involvement is well established in clinical practice, and this characteristic of the study cohort highly reflects the setting of the real-world. Larger multicentric studies are required to disclose the robustness of NLR concerning steroid administration in mCRPC patients undergoing Ra-223. Conclusions The degree of systemic inflammation and the quantification of the metabolically active tumor burden through FDG-PET imaging provide independent prognostic insights in mCRPC patients undergoing Ra-223. The combination of these biomarkers might represent a potentially valuable tool for identifying mCRPC patients who are most likely to benefit from Ra-223 since baseline. Larger studies are needed to further evaluate this hypothesis and, eventually, to confirm these preliminary results.	v3-fos-license
2016-05-04T20:20:58.661Z	2015-05-01T00:00:00.000	6160367	{ "extfieldsofstudy": [ "Computer Science", "Medicine", "Biology" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1004248&type=printable", "pdf_hash": "0f9ac03fb42836d3cb8684233160ca75ab8b3ec7", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45359", "s2fieldsofstudy": [ "Biology", "Computer Science" ], "sha1": "0f9ac03fb42836d3cb8684233160ca75ab8b3ec7", "year": 2015 }	pes2o/s2orc	Predicting Peptide-Mediated Interactions on a Genome-Wide Scale We describe a method to predict protein-protein interactions (PPIs) formed between structured domains and short peptide motifs. We take an integrative approach based on consensus patterns of known motifs in databases, structures of domain-motif complexes from the PDB and various sources of non-structural evidence. We combine this set of clues using a Bayesian classifier that reports the likelihood of an interaction and obtain significantly improved prediction performance when compared to individual sources of evidence and to previously reported algorithms. Our Bayesian approach was integrated into PrePPI, a structure-based PPI prediction method that, so far, has been limited to interactions formed between two structured domains. Around 80,000 new domain-motif mediated interactions were predicted, thus enhancing PrePPI’s coverage of the human protein interactome. Author Summary Complexes formed between a structured domain on one protein and an unstructured peptide on another are ubiquitous. However, they are often quite difficult to detect experimentally. The development of computational approaches to predict domain-motif interactions is therefore an important goal. We report a method to predict domain-motif interactions using a Bayesian approach to integrate evidence from a variety of sources, including threedimensional structural and non-structural information. The method was applied to the entire human proteome and showed significant improvement over existing methods. The method was incorporated into PrePPI, a computational pipeline for the prediction of protein-protein interactions that relies heavily on structural information. Approximately 80,000 new interactions were detected. The new PrePPI database provides easy access to about 400,000 human protein-protein interactions and should thus constitute a valuable resource in a variety of biological applications including the characterization of molecular interaction networks and, more generally, in the study of interactions mediated by proteins in families that may not be extensively studied experimentally. Introduction Mapping the human protein interactome has important implications for understanding basic biology and human disease at the molecular level [1]. High-throughput (HT) experimental techniques such as yeast two-hybrid and tandem affinity purification have been developed and applied to discover protein-protein interactions (PPIs) in multiple organisms on a genomewide scale [2]. However, these approaches have inherent limitations, leading to a substantial false positive rate [2,3] with many interactions likely undiscovered due to high rates of false negatives [2,4,5]. The development of reliable computational approaches to identify PPIs is therefore an important alternative to HT experimental techniques [6,7]. Computational predictions of PPIs are based on criteria such as sequence orthology [8], similarity in evolutionary history [9], genomic context [10], and literature curation [11]. Predictions based on detailed structural modeling of PPIs have also been developed [12] and recent approaches [13] that combine low resolution structural modeling with non-structural information have begun to expand the applicability of structure to a genome-wide scale. Interactions determined by HT experiments and computationally have been deposited in databases such as STRING [14] and PrePPI [13]. An important class of PPIs involves interactions between a short peptide motif of one partner, and a structured peptide recognition domain (PRD) from another [15][16][17][18]. Discoveries of new domain-motif interactions present unique challenges compared to domain-domain mediated interactions [16,19]. For a few major PRD families such as PDZ and SH3 domains, HT experimental techniques [19][20][21] such as phage display have been used to derive binding preferences which can subsequently be used to scan a genome to identify proteins likely to bind a given PRD. Computational modeling has also been used to predict domain-motif interactions [22][23][24][25][26][27]. In these studies, models for domain-motif complexes are built and evaluated with either physical or statistics-based scoring functions. Despite much progress, the experimental or computational effort involved significantly limits the scope of these studies to a few PRD families so that methods that enable predictions for a larger number of PRD families are needed. Databases such as the eukaryotic linear motif resource [28] (ELM) provide consensus sequence patterns for peptide motifs binding to many different PRD families, and methods such as iELM have been developed to make new predictions based on such information [29]. However, these patterns are often derived from a limited amount of data (e.g. from a few known binding sequences), which necessarily limit their coverage and accuracy. Surveys of available structures of protein-peptide complexes in the PDB have also identified candidate binding motifs [30] as well as generic structural characteristics for binding interfaces [31], but overall structural information has not been widely used in predicting new interactions except for a few PRD families. In this study we report a computational framework to predict interactions mediated by domain-motif interfaces. The method uses a Bayesian approach to integrate knowledge from the ELM database, domain-peptide structures from the PDB, and non-structural information. We have incorporated the method into PrePPI [13] and found that the addition of domain-motif predictions improves its performance in PPI detection. The new version of PrePPI now contains 400,000 PPI predictions. Results Predicting peptide-mediated PPIs Fig 1 outlines the combination of strategies we use to predict PPIs mediated by domain-motif interfaces. The first approach we tried is based on ELM [28], a manually curated database containing more than 200 classes of PRD/motif pairs. In an ELM class a PRD is represented by its Pfam family and a motif is represented by a consensus sequence pattern derived from peptides known to bind to that family. For each ELM class, we identified all pairs of human proteins containing the corresponding PRD/motifs and calculated an interaction likelihood ratio (LR) for each. The LRs were calculated, using a Bayesian approach, as the percentage of pairs of proteins having the PRD/motif match in a true positive set of known human PPIs divided by the same percentage of a true negative set of 1.6 million pairs of proteins that do not interact (see Methods). In this calculation we also considered whether the motifs are located in predicted disordered regions and whether their sequences are conserved evolutionarily (see Methods for details). Sequence conservation and disorder have been shown to be associated with functional motifs [32,33]. The performance of the PRD/motif predictor in rediscovering known human-human interactions was assessed using 5-fold cross-validation on the true positive and negative sets (see Methods) and compared with the iELM method developed by Weatheritt et al. [29] (Fig 2). iELM is also based on information from the ELM database and uses features such as sequence conservation and disorder incorporated into a support vector machine (SVM). In addition, instead of using Pfam to identify PRDs, Weatheritt et al. constructed their own Hidden Markov Models (HMMs) for each ELM class. As can be seen in Fig 2, PRD/motif performs better than iELM in the lower false positive region but the reverse is true in the higher false positive region. Similar results were obtained when using a precision-recall curve to evaluate performances, with PRD/motif having higher precision in the lower but not the higher recall region (S1 Fig). For two query proteins QA and QB, if QA has a peptide recognition domain DA and QB has a motif MB from the same ELM class, a likelihood for a putative interaction between QA and QB was calculated (see Methods) based on the identity of the ELM class, predicted disorder of MB, and the sequence conservation of MB and combined with likelihoods from other non-structural (NS) evidence including gene co-expression, gene ontology (GO) similarity and phylogenetic profile (PP) similarity. (B) Predictions made using information from PepX (method Struct). For two query proteins QA and QB, a putative interaction between DA and MB is suggested using a template complex structure from PepX. A likelihood for the interaction is calculated based on the structural similarity between DA and the template PRD component, the sequence similarity between MB and the template peptide motif, disorder prediction, and sequence conservation of MB. Again this likelihood was combined with non-structural evidence to obtain a final score. In what follows, we chose to use PRD/motif when extracting information from the ELM database based on its better performance in the lower false positive region. Despite its broad scope, certain domain-motif interactions, especially those not belonging to well-studied families, may not be included in the ELM database. To expand our coverage, we used experimentally determined complexes taken from the PepX database [34] as templates to model domain-motif interactions (Fig 1). PepX contains high-resolution structures of proteinpeptide complexes in the PDB whose peptide motif length ranges from 5 to 35 amino acids. Structural models for individual human proteins or their subdomains were retrieved from the PDB if available or from one of two homology model databases, ModBase [35] and SkyBase [36]. More than 10,000 human proteins have at least some part of their sequences covered by a structural model [13]. An interaction model for a pair of proteins was constructed if one protein contained a PRD that was structurally similar to a PRD in a given template in PepX and the other protein contained a short sequence motif with sequence similarity (based on BLOSUM62 scores [37], see Methods) to the motif component of the template. We only considered motifs whose BLOSUM scores ranked among the top 0.05% from the entire human proteome to retain a manageable number of candidate peptide motifs. We again used a Bayesian approach to estimate the likelihood of an interaction given the properties of the model. Sources of evidence integrated into our Bayesian scheme include the sequence similarity score between the candidate motif and the motif in the template, the structural similarity score between the candidate PRD and the PRD in the template, whether the candidate motif is located in predicted disordered regions, and whether sequences around the candidate motif are conserved evolutionarily (see Methods). The resulting predictor, referred to as Struct for Structural information alone (based on the PepX database), performed worse than PRD/motif in our cross validation test. However, a predictor that combines both (PRD/ motif+Struct) performs better than PRD/motif alone, showing that structural evidence is adding value to the predictions (Fig 2). We combined PRD/motif-based and Struct-based LRs with non-structural (NS) evidence that has previously been used to infer PPIs [38]. Specifically, for each pair of proteins, we considered their co-expression level, their gene ontology (GO) functional similarity, and their phylogenetic profile similarity. Derivation of LR scores for these sources of evidence was described previously [13,38] and the values obtained in our previous study [13] were directly used in the current one. A final score was obtained by multiplying the LR for the predicted domain-motif interface with the LR from non-structural evidence. The resulting integrative predictor, PRD/ motif+Struct+NS, was then compared to the method based only on NS evidence in rediscovering known human PPIs (Fig 2). The NS-based method outperforms PRD/motif and Struct, which is not surprising given that NS is not limited to peptide-mediated interactions. However, PRD/motif+Struct+NS offers a significant improvement over NS alone (Fig 2). Furthermore, the combination of methods dramatically increases the number of predicted interactions with LR score > 600 [13,38], referred to as "strong predictions" in this study. This LR value corresponds to a posterior probability of 0.5 that two proteins interact when assuming a prior odds of 1 in every 600 protein pairs interact. Details of the derivation can be found in Jansen et al. [38]. Using information from PRD/motif, Struct or NS alone led to 1,515; 0; and 15,376 strong predictions, respectively. In contrast, a total of 125,624 predictions were made when combining the three sources of evidence, representing 110,248 new predictions as compared to NS alone. This significant amplification highlights the value of combining independent clues. Notably, a total of 55 true positives can be detected before encountering the first false positive. To obtain further validation of our approach, we compared our strong predictions to the 257 known human domain-motif interactions found in the ELM database. A total of 75 known interactions were included in our strong predictions (123 when using a LR cutoff of 100), while using only evidence from PRD/motif, Struct or NS alone recovered only 6; 0; and 6 interactions, respectively. Furthermore, when using the combined sources of evidence, the LR scores for more than half of the interactions (40 out of 75) ranked among the top 20% of all strong predictions. These 75 interactions were not dominated by a particular ELM class as they spanned 33 out of the 57 classes that represent the 257 known human interactions. We also examined overlap of our predictions with 160 human domain-motif complexes in PepX. The overlap is only 42 for strong predictions but increases to 86 when using a lower LR cutoff at 100. Incorporating peptide-mediated PPI prediction methods into PrePPI As summarized above, we have previously developed PrePPI, a computational PPI prediction method that performs comparably to experimental HT approaches. PrePPI combines structural evidence with non-structural evidence using a Bayesian framework but currently lacks the capability to predict domain-motif mediated PPIs [13], To add this ability we compared the structure-based LR from the original PrePPI and the LR for domain-motif interfaces obtained from PRD/motif+Struct (see Methods). The larger of the two was chosen and multiplied with the LR obtained from NS evidence to generate a final LR for the interaction. As shown in Fig 3, the addition of evidence based on domain-motif interactions (PrePPI_PRD/motif+Struct) results in improvement in performance when compared to the original PrePPI (PrePPI_orig). In this comparison, we use the same true positive set described above but a larger true negative set of non-interacting pairs of proteins which was used in the original PrePPI (performance is nearly identical for both true negative sets, S2 Fig). PrePPI_PRD/motif+Struct yielded 78,898 additional strong predictions compared to Pre-PPI_orig. Although more than 40% of the predictions come from the 5 most prevalent PRD families (including SH2, SH3, 14-3-3, nuclear receptors and AGC kinase docking motif), over 130 ELM classes and 150 clusters of PepX template structures contributed to our results. Together with the original 317,814 interactions reported from PrePPI_orig, the new PrePPI which includes domain-motif mediated interactions contains a total of 396,712 predicted human PPIs. Discussion In this study we developed methods to predict PPIs mediated by domain-motif interfaces using both expert knowledge of domain-motif interactions in the ELM database and structures of domain-motif complexes in the PDB. Although there is some overlap between predictions made with the ELM and the structure-based approach (PRD/motif and Struct), differences between them likely led to the observed improvements when two strategies were combined. For example, the Bcl-2 families have multiple structural representatives in the PDB but are not included in ELM. Moreover, the sequence similarity scoring approach of PrePPI_Struct allows the identification of motifs outside of the consensus provided by ELM. For example, the motif sequences for several SH3 and nuclear box receptor complexes in PepX could not be described by consensus patterns from any of their corresponding classes in ELM. Overall, among our new strong predictions, 13,988 are made using motifs that cannot be described by consensus patterns from the corresponding ELM class. On the other hand, the use of consensus patterns as in ELM (and hence PrePPI_PRD/motif) can be effective in capturing the variability of motifs for large, well-studied families even when no structural information is available. In terms of finding PRDs, the structured-based method in PrePPI_Struct applies a filtering criterion to ensure that the candidate PRD aligns well structurally to the template PRD at the binding interface, which is not accounted for by the sequence-based Pfam definition in PrePPI_PRD/motif. As in the original PrePPI, combining sources of evidence that on their own provide only weak clues has a major impact on overall performance. For example, the consensus sequence patterns used in the PrePPI_PRD/motif approach can be promiscuous as can the use of sequence similarity in PrePPI_Struct, potentially leading to reduced prediction specificity. This can be especially problematic for interactions between candidate PRDs and motifs that interact via similar interfaces. For example, for the structural modeling component in PrePPI_Struct, it is possible that modeled interfaces for many different pairs of proteins share the same sets of clues if they are derived from the same template structure. In this case, prediction specificity would come from non-structural evidence. Moreover, prediction coverage based on the individual sources of evidence can be low as shown in Results, which highlights the importance of combining different sources of orthogonal information implicit in the Bayesian approach. It is widely appreciated that HT approaches including yeast-two hybrid and tandem affinity purification have limitations in detecting PPIs mediated by protein-peptide interfaces. Apart from issues such as their transient nature and high K d, they frequently depend on cellular conditions, many of which will never be sampled in an HT experiment [16,19], potentially resulting in very high false negative rates. Indeed, high-throughput screens focusing on individual PRDs often identify a surprisingly large number of binding partners [19]. Furthermore, it has recently been suggested [17] that the number of putative peptide motifs in the human proteome to be more than a million. The number was based on estimating the extent of disordered regions in the human proteome and the tendencies of these regions to be involved in binding. Motifs that undergo post-translational modification were also included in the estimate, based on their prevalence among a set of well-studied proteins [17]. Although there are certainly false positives in computational predictions, these issues highlight the importance of developing methods such as that described here that can be applied on a genome wide scale and are insensitive to such experimental difficulties. The large number of predictions we make provide hypotheses that can be further refined and tested by more in-depth experimental/computational studies. In addition, the integrative nature of our framework should also help provide the biological context for predicted interactions, further contributing to our understanding of this still largely unexplored portion of the human interactome. Human proteins A total of 20,318 unique human protein sequences were downloaded from UniProt [39] and constituted the human proteome in this study. Identification of domains and motifs for PrePPI_PRD/motif As of January 2014, a total of 203 ELM classes of motifs that shared similar sequence features and targeted by the same kind of domain were annotated in the ELM database. For each class, a consensus pattern for the motifs and the name of the Pfam family for the interacting PRD were retrieved from the database. Hidden Markov Models (HMMs) for each Pfam family were downloaded from the Pfam [40] website, and the hmmscan utility from the HMMER suite [41] was used to identify domains within each human protein using default cutoffs defined in the downloaded HMM files. Candidate motifs satisfying the consensus pattern were identified using an in-house Perl script. Identification of domains and motifs for PrePPI_Struct We obtained domain-motif structures from the PepX database [34] (multimers interacting with a single peptide were excluded). A PRD in PepX was used as a template to model a domain-motif interaction for a given human protein if it is structurally similar to the model for that protein as defined by a protein structural distance (PSD) less than 0.65 calculated with the program Ska [42]. An additional requirement is that in the structural alignment at least 75% of interfacial residues for the template PRD must align to surface residues on the structural model for the protein. Interfacial residues for the template PRD were defined as those with at least one atom located within 4.5 Å of the template peptide motif in the complex structure. Surface residues for the structural models of human proteins were identified using the program SUR-Face [43], with an accessible surface area cutoff of 10 Å 2 . PSD scores between candidate domains and the template PRDs were grouped into two bins, [0-0.3] and [0.3-0.65], for the Bayesian classifier. For a peptide motif in a given template that is x residues long, new potential binding motifs were identified by scanning a x-residue window across the whole human proteome. A sequence similarity score between the sliding window and the template motif was calculated using the BLOSUM62 scoring matrix [37]. Sequence motifs whose scores ranked among the top 0.05% among all such sliding windows were kept as candidate motifs. A cutoff based on percentage but not absolute BLOSUM62 scores enables comparison of motifs across different templates, which can vary greatly in length. For the Bayesian classifier, sequence similarity scores between candidate motifs and the template motifs were grouped into 4 bins: (1) scores within the top 0.0001%, (2) scores between the top 0.0001% and the top 0.001%, (3) scores between the top 0.001% and the top 0.01%, and (4) scores between the top 0.01% and the top 0.05%. Prediction of disorder The program IUPred [44] was used to predict if a candidate motif is likely located in a disordered region. A disorder score (ranging from 0 to 1) for each individual residue in the human proteome was obtained by running IUPred on all human protein sequences. For each motif, a score was then obtained by averaging the disorder score for each individual residue in the motif. For the Bayesian scoring, a binary classification of candidate motifs was used: a candidate motif is disordered if the averaged score is larger than 0.5, which is the cutoff recommended by IUPred. Calculation of sequence conservation scores The program GOPHER [45] was used to search for orthologs among the UniProt database for every human protein. Only orthologs belonging to species of the subphylum vertebrata were considered, as including orthologs from a larger range of species (e.g. metazoan) does not significantly impact performance. A multiple sequence alignment of the orthologs was then generated using the program Muscle [46]. A conservation score for each residue in the human protein was calculated as the information content for the corresponding column in the alignment. The score was multiplied by the percentage of non-gap residues in the column. A residue was determined to be conserved locally if its conservation score was higher than the average of such scores for its surrounding residues [47] (up to 31-residue upstream and downstream). For the Bayesian scoring, a binary classification of candidate motifs was used: a motif was classified as locally conserved if all residues in the candidate motif were locally conserved. Training set and the naïve Bayes classifier A naïve Bayes classifier was used to integrate different sources of evidence into a likelihood ratio (LR) for an interaction between two proteins. The datasets for training the classifier consist of a true positive set that includes 7,409 interactions compiled from a set of 5 databases [48][49][50][51][52] and supported by at least two publications, and a true negative set that contains 206,361,949 interactions not supported by any publication [13]. While one can assume that a non-reported interaction is likely to be non-interacting, the negative set will necessarily contain undiscovered true interactions which are just the ones we would like to detect. The reported FPR at a given LR (which assumes every prediction in the true negative set is wrong) can therefore be viewed as an upper bound on the true value. As constructing a reliable set of non-interacting proteins remains difficult, we proceeded with this simple definition. For Fig 2, in order to compare to iELM, we used a small set of 1.6 million pairs of proteins randomly sampled from the larger negative set, for which iELM scores were available. Results from the larger set were shown in Fig 3 (performance for both sets was nearly identical). For each property (referred to as a "clue"), c i , of an interaction between protein x and y, the conditional probability that one will observe the clue given that the interaction is in the true positive set or the true negative set is calculated as P(c i \|I xy,TP ) and P(c i \|I xy,TN ). The probability P(c i \|I xy,TP ) is calculated as P(c i \|I xy,TP ) = n/N TP , where n is simply the number of interactions in the true positive set with the clue c i , and N TP is the total number of interactions in the true positive set. P(c i \|I xy,TN ) is calculated analogously for the true negative set. A LR value can be calculated by dividing these two probabilities, LR(c i ) = P(c i \|I xy,TP ) / P(c i \|I xy,TN ), to reflect how strongly the clue c i is indicative of a true interaction. For the PRD/motif method based on the ELM database, a total of four clues were used for the domain-motif component: a) whether a domain-motif match from the same ELM class is present (LR(match)); b) the identity of the matching ELM class (LR(class)); c) whether the motif is located in a predicted disordered region (LR(diso)); d) whether the motif is conserved locally in sequence relative to its surrounding regions (LR(consv)). The latter three clues can be assumed to be independent of one another, but they all depend on the first clue being true. Their LR values were therefore normalized by the LR for the first clue, and the final LR for the domain-motif interface is therefore: For the Struct method based on the PepX database, a total of five clues were used for the domain-motif component: a) whether a domain-motif match from the same template structure is present (LR(match)); b) The PSD score between the candidate domain and the PRD component in the template (LR(PSD)); c) the sequence similarity score between the candidate motif and the motif component in the template (LR(SIM)); d) whether the motif is located in a predicted disordered region (LR(diso)); e) whether the motif is conserved locally in sequence relative to its surrounding regions (LR(consv)). As above, LRs for the latter four clues were normalized by LR(match) and the final LR for the domain-motif interface is: LRðDMIÞ ¼ LRðmatchÞ Á ðLRðPSDÞ=LRðmatchÞÞ Á ðLRðSIMÞ=LRðmatchÞÞ ÁðLRðdisoÞ=LRðmatchÞÞ Á ðLRðconsvÞ=LRðmatchÞÞ The LR for the domain-motif interface was then multiplied with LRs for non-structural evidence to obtain a final LR for the interaction. The LR values used in this study are provided as a supplemental table (S1 Table). LR scores for non-structural evidence based on co-expression, similarity in gene ontology, and similarity in phylogentic profiling calculated for the original PrePPI were used in this study [13]. Precision-recall curves were generated using the program AUCCalculator [53]. Evaluating iELM The iELM scores for the positive set and the randomly generated smaller negative set were kindly provided by Weatheritt et al. Incremental cutoffs of raw iELM scores were used to calculate the TPR and FPRs. If iELM makes multiple PRD/motif predictions for a single pair of protein, only the prediction with the highest score was considered in testing. Availability Predictions have been incorporated into the PrePPI database and can be downloaded for individual query proteins (https://honiglab.c2b2.columbia.edu/PrePPI/). New predictions are also provided as a supplement (S2 Table).	v3-fos-license
2021-06-11T13:21:16.300Z	2021-06-07T00:00:00.000	235396716	{ "extfieldsofstudy": [ "Medicine", "Biology" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://academic.oup.com/g3journal/article-pdf/11/11/jkab269/41298029/jkab269.pdf", "pdf_hash": "7870b1149538df528a30f272775af276127559a1", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45360", "s2fieldsofstudy": [ "Agricultural And Food Sciences" ], "sha1": "0b84581d0e63d65588fe360735eb19867372d329", "year": 2021 }	pes2o/s2orc	Genome-wide association mapping of Pyrenophora teres f. maculata and Pyrenophora teres f. teres resistance loci utilizing natural Turkish wild and landrace barley populations Abstract Unimproved landraces and wild relatives of crops are sources of genetic diversity that were lost post domestication in modern breeding programs. To tap into this rich resource, genome-wide association studies in large plant genomes have enabled the rapid genetic characterization of desired traits from natural landrace and wild populations. Wild barley (Hordeum spontaneum), the progenitor of domesticated barley (Hordeum vulgare), is dispersed across Asia and North Africa, and has co-evolved with the ascomycetous fungal pathogens Pyrenophora teres f. teres and P. teres f. maculata, the causal agents of the diseases net form of net blotch and spot form of net blotch, respectively. Thus, these wild and local adapted barley landraces from the region of origin of both the host and pathogen represent a diverse gene pool to identify new sources of resistance, due to millions of years of co-evolution. The barley—P. teres pathosystem is governed by complex genetic interactions with dominant, recessive, and incomplete resistances and susceptibilities, with many isolate-specific interactions. Here, we provide the first genome-wide association study of wild and landrace barley from the Fertile Crescent for resistance to both forms of P. teres. A total of 14 loci, four against P. teres f. maculata and 10 against P. teres f. teres, were identified in both wild and landrace populations, showing that both are genetic reservoirs for novel sources of resistance. We also highlight the importance of using multiple algorithms to both identify and validate additional loci. Introduction Net blotch caused by the ascomycetous fungal pathogen Pyrenophora teres (anamorph: Drechslera teres) is an economically important disease of barley worldwide. P. teres f. teres (Ptt) incites the net form of net blotch (NFNB) and P. teres f. maculata (Ptm) incites the spot form of net blotch (SFNB) (Smedegå rd-Petersen 1971). Both forms are responsible for large crop losses that typically range between 10% and 40% when susceptible cultivars are grown, however, under conducive environmental conditions losses can reach 100% (Piening and Kaufmann 1969;Mathre 1982;Moya et al. 2018). At least one form of the disease has been reported from all barley growing regions and, in many regions, both forms are present with annual fluctuation in predominance. This presents challenges to breeders, as both Ptt and Ptm interact with host resistance/susceptibility genes differentially, thus are considered distinct and treated as different diseases when breeding for resistance. However, as further characterization of resistant/susceptibility loci continues, overlaps in host-pathogen genetic interactions in both pathosystems are becoming more prevalent. Both Ptt and Ptm occur as genetically distinct populations and can be separated in the field based on lesion morphology. Although these two forms can be hybridized under laboratory conditions (Campbell and Crous 2003), hybridization under field conditions is extremely rare (Campbell et al. 2002;McLean et al. 2014;Akhavan et al. 2015;C¸elik O guz et al. 2018;Poudel et al. 2019). However, both forms of P. teres undergo form specific sexual as well as asexual reproduction (Karakaya et al. 2004;Serenius et al. 2005;Akhavan et al. 2015;C¸elik O guz et al. 2018, 2019bPoudel et al. 2019). The complex nature of this reproduction system poses serious evolutionary risks for resistance breeding as populations contain diverse effector repertoires, that in different combinations, can rapidly overcome deployed resistances (McDonald and Linde 2002). Yet, the use of resistant barley cultivars is the most environmentally friendly and economically feasible method of NFNB and SFNB control (Robinson and Jalli 1996;Afanasenko et al. 2009). Wild barleys and barley landraces are important sources of resistance against diverse biotic and abiotic stresses (Allard and Bradshaw 1964;Ceccarelli 1996;Yitbarek et al. 1998;Ellis et al. 2000;Jakob et al. 2014;Karakaya et al. 2016Karakaya et al. , 2020. Wild barley (Hordeum spontaneum) is known as the progenitor of modern-day barley (Hordeum vulgare) and grows naturally in the Fertile Crescent, regions of south and southeastern Turkey, North Africa, and Southwest Asia (Harlan and Zohary 1966;Nevo 1992;Zohary and Hopf 2000). Barley landraces gave rise to modern barley varieties (Thomas et al. 1998) as they were subjected to natural and artificial selection for the last 10,000 years (Ceccarelli and Grando 2000). Within the barley center of origin, Turkey is located at the ancestral hub of barley diversification regions including the Mediterranean, Horn of Africa, and the Tibetan Plateau (Muñoz-Amatriaín et al. 2014;Poets et al. 2015), with diverse barley landraces still used by Turkish farmers (Helbaek 1969;Kü n 1996;Pourkheirandish and Komatsuda 2007;Ergü n et al. 2017). Turkey is also at the center of origin of the P. teres pathogens, Ptt and Ptm. Barley lines can show differential responses to either or both forms of net blotch due to distinct yet complex genetic interactions with each form. Many barley genotypes may be resistant to the majority of isolates of one form yet susceptible to the alternate form of most isolates (Bockelman et al. 1983;Grewal et al. 2012). Thus, when breeding for resistance, the two forms of net blotch are treated as separate diseases (Liu et al. 2011;Usta et al. 2014;Yazıcı et al. 2015). In barley, genetic resistance to P. teres was first reported in Geschele (1928). Since both forms of net blotch were not described at that time, it was assumed that the net form was used. In the first studies to genetically characterize barley-P. teres interactions, Tifang was the resistant parent in the Tifang Â Atlas cross (Schaller 1955). Mode and Schaller (1958) identified the resistance genes Pt 1 , Pt 2 , and Pt 3 segregating in this cross. Bockelman et al. (1977) revised the naming of Pt 1 , Pt 2 , and Pt 3 resistance genes and described the Rpt1 (Pt 1 and Pt 2 ), Rpt2 (novel), and Rpt3 (Pt 3 ) loci on chromosomes 3H, 1H, and 2H, respectively. Further studies identified Rpt4 on chromosome 7H (Williams et al. 1999(Williams et al. , 2003, Rpt5 on chromosome 6H, Rpt6 on chromosome 5H (Manninen et al. 2006), Rpt7 on chromosome 4HL, and Rpt8 on chromosome 4HS (Franckowiak and Platz 2013), however, there are over 340 QTLs previously identified (Clare et al. 2020). Manninen et al. (2006) reclassified the locus Pt a as Rpt5 on chromosome 6H, and subsequently, three genes/alleles have been characterized at the locus (Rpt5.f, Spt1.k, Spt1.r) as dominant resistance or susceptibility genes (Franckowiak and Platz 2013;Richards et al. 2016). In multiple barley-Ptt genetic interaction studies, it has been shown that the Rpt5 locus is the most important resistance/susceptibility locus in this system. This complex locus putatively contains multiple resistance as well as susceptibility genes that have been characterized in diverse barley-P. teres interactions from around the world (Clare et al. 2020). Because the Rpt5 locus also shows dominant susceptibility in certain barley lines, additional alleles were designated Susceptibility to P. teres 1 (Spt1) by Richards et al., (2016). Furthermore, high-resolution genetic mapping and positional cloning efforts have identified Rpt5 and Spt1 candidate genes and functional validation are underway ). Resistance to Ptm originally appeared to be less complicated when compared to Ptt due to the presence of three major loci. These three loci were identified as Rpt4 on chromosome 7H (Williams et al. 1999(Williams et al. , 2003, Rpt6 on chromosome 5HS (Manninen et al. 2006), and Rpt8 on chromosome 4HS (Friesen et al. 2006;Franckowiak and Platz 2013). To date, over 140 QTLs have been reported to be implicated in the Ptm reaction, which have been collapsed into 36 unique loci, five of which are specific to Ptm and the rest showing some degree of overlap with known Ptt loci (Clare et al. 2020). These five unique loci that are specific to the Ptm interaction are SFNB-3H-78.53 on chromosome 2H (Burlakoti et al. 2017), QRptm-4H-120-125 on 4H (Tamang et al. 2019), QRptts-5H-106.00 on 5H (Adhikari et al. 2019), QRptm7-3 on 7H (Wang et al. 2015), and QRptm7-6/QRptm-7H-119-137 on 7H (Wang et al. 2015;Tamang et al. 2019). Considering that all currently designated resistance/susceptibility loci except for Rpt2 (only implicated in the Ptt interaction) have now been implicated in both Ptm and Ptt interactions (Clare et al. 2020), it is with some caution that it can be concluded that host-pathogen genetic interactions with the two forms and barley should be considered distinct. Thus, for both forms, with the exception of Rpt6, numerous researchers have described synonyms of all loci (Clare et al. 2020). Multiple genome-wide association mapping studies (GWAS) have investigated NFNB resistance in barley Wonneberger et al. 2017b;Amezrou et al. 2018;Adhikari et al. 2019;Daba et al. 2019;Novakazi et al. 2019;Rozanova et al. 2019;Adhikari et al. 2020). A large proportion of the resistance markers associated with NFNB resistance have been localized to the centromeric region of barley chromosome 6H . In GWAS for SFNB resistance, 29 (Wang et al. 2015), 27 , 11 (Burlakoti et al. 2017), and 1 (Daba et al. 2019) unique genomic loci were identified. Four important QTLs (QRptm7-4, QRptm7-6, QRptm7-7, and QRptm7-8) were mapped into a region covering the Rpt4 locus on chromosome 7HS (Wang et al. 2015). Burlakoti et al. (2017) identified a new and important QTL on chromosome 2HS that was predominately found in 6-rowed barley lines as compared to 2-rowed. Daba et al. (2019) also defined a new QTL on chromosome 6H associated with Ptm susceptibility in a large number of genotypes, which is a common mechanism in inverse gene-for-gene interactions with this pathogen. Vatter et al. (2017) performed nested association mapping for NFNB and described further interactions at the important Rpt5/ Spt1 locus. In barley, the newest approach to identify marker trait associations (MTAs) with Ptt resistance is exome QTL-seq. This approach identified a large number of MTAs on chromosomes 3H and 6H when analyzing resistant and susceptible bulks (Hisano et al. 2017). The resistance status of wild barley genotypes and barley landraces to P. teres has been reported by several research groups worldwide (Jana and Bailey, 1995;Lakew et al., 1995;Legge et al., 1996;Sato and Takeda, 1997;Fetch et al., 2003;Silvar et al., 2010;Endresen et al., 2011;Neupane et al., 2015;, 2019a, 2019c. However, molecular mapping studies of P. teres resistance in wild and landrace barleys have been limited (Yun et al. 2005;Vatter et al. 2017;Adhikari et al. 2019;Gyawali et al. 2019Gyawali et al. , 2020. In this study, four novel loci representing resistance to NFNB were mapped in Turkish wild barley and landraces. This study highlights the importance of surveying wild and unimproved barley lines for sources of resistance that may have been lost during domestication and modern breeding. These studies, focused on diversity in the barley primary germplasm pool, will provide new sources of resistance and associated markers to aid in deploying robust resistances against NFNB and SFNB. (2019a). Briefly, a total of 5-10 seeds were planted in 7 cm diameter plastic pots containing sterile soil, sand, and organic matter mixtures (60, 20, 20; v/v/v, respectively), depending on the number of seeds of each wild and landrace barley. The pots were kept in greenhouse conditions at 18-23 6 1 C night/day with 14 h/10 h light/dark regime before and after inoculation. Three virulent isolates of Ptm (GPS263, 13-179, and 13-167) and three virulent isolates of Ptt (GPS18, UHK77, and 13-130) were used in phenotyping studies. The inoculum was prepared from 10-day single spore cultures grown on potato dextrose agar kept at 16-23 6 2 C night/ day with a 10 h/14 h dark/light period. For the preparation of the inoculum, the mycelia were scraped from Petri dishes using a painting brush, washing with water, and filtered by cheesecloth. The inoculum concentration was adjusted to 15-20 Â 10 4 mycelial fragments/ml. One drop of Tween V R 20 was added to each 100 ml inoculum. Inoculation was carried out at the two to threeleaf stages by spraying inoculum over the barley lines with a hand spray until runoff. Following inoculation, plants were covered with nylon in transparent boxes with lids for 76 h. High humidity was maintained for a further 48 h with the nylon uncovered and ventilated. After 7 days, seedlings were evaluated for disease severity using the net and SFNB scales described by Tekauz (1985). (Ruff et al. 2020) barley SNP markers were used to genotype 295 barley accessions. Marker primers were divided into six and three total primer pools for Panels 1 and 2 for PCR amplification, respectively. PCR amplification and barcoding reactions were performed as described by Sharma Poudel et al. (2018) and Ruff et al. (2020). Briefly, nine primary PCR amplification reactions were performed per sample. Following amplification, equal volumes of primary PCR products were pooled into 96-well plates with each well containing all amplified markers for a given DNA sample. Next, barcoding PCR reactions were performed with a universal barcoding reverse primer and unique forward barcoding primers for each sample. Following barcoding reactions, samples were pooled and purified. A final PCR reaction was performed using sequencing primers to ensure the barcoding reaction was successful. Samples taken before and after the final amplification were run on agarose gel to verify the appropriate product size and amplification. Quantification of the barcoded libraries were performed using the Qubit dsDNA HS assay kit (Life Technologies, Carlsbad, CA, USA). Enrichments were carried out using the Ion OneTouch TM 2 System (Panel 1) and the Ion PI TM Hi-Q Sequencing 200 kit on the Ion Chef (Panel 2). Finally, samples were sequenced on the Ion Torrent Personal Genome Machine TM (Panel 1) and the Ion Proton TM (Panel 2) Systems using two Ion 318 TM chips and 3 Ion PI TM chips, respectively, following the manufacturer's standard protocols. The resulting marker panels were collapsed to eliminate duplicated markers from each panel. Imputation, filtering, and linkage disequilibrium Due to the heterozygosity present in the natural population, heterozygous calls (5.23%) were included in the analysis. Accessions and markers with more than 30% missing data were removed from analysis, resulting in 282 barley accessions and 530 markers. Missing data were imputed using LinkImpute, which uses a linkage disequilibrium k-nearest neighbor imputation (LD-kNNi) method (Money et al. 2015) in Trait Analysis by aSSociation, Evolution, and Linkage (TASSEL) 5.2.60 (Bradbury et al. 2007). Markers with a minor allele frequency of <0.05 were included in the analysis but were treated with caution based on best practice from the Genomic Association and Prediction Integrated Tool (GAPIT) 3.0 user manual. Linkage disequilibrium was calculated in TASSEL using a window size of 50 markers and an R 2 threshold of 0.8 resulting in 522 markers. Population structure, kinship matrices, and model algorithms Population structure was accounted for using STRUCTURE analysis and principle component analysis. A total of 522 markers were used for analysis of population structure. The software STRUCTURE v2.3.4 (Pritchard et al. 2000) was used to estimate population structure of the barley panel to create a population structure matrix (Q) to be used as a covariate. To determine the optimal number of subpopulations, an admixture ancestry model was used with a burnin of 10,000, followed by 25,000 Monte Carlo Markov Chain (MCMC) replications for k ¼ 1 to k ¼ 10 with ten iterations. STRUCTURE HARVESTER (Earl and vonHoldt 2012) was used to identify the optimal number of subpopulations using the Dk method (Evanno et al. 2005). The optimal k value was subsequently used to run a new STRUCTURE analysis using a burn-in of 100,000 followed by 100,000 MCMC replications. An individual was deemed to be part of a population if the membership probability was >0.8 . Individuals that did not achieve a value of 0.8 were deemed to have admixture ancestry. The final Q matrix was used as a fixed covariate in association models. Principle component analysis was conducted in R 3.6.3 using GAPIT 3.0 (Wang and Zhang 2020) with default settings. Principle components explaining at least 25% (PC1) and 50% (PC5) were used for further analysis. A naïve model using only genotypic (Supplementary File S1 and S2) and phenotypic data (Supplementary File S3) and an additional three fixed-effect models accounting for population structure [Q (Supplementary File S4), PC1, and PC5] were all performed using the GLM method. For initial discovery of the most appropriate method to account for random effect in the model, a kinship matrix (K) was constructed using the EMMA (Kang et al. 2008), Loiselle (Loiselle et al. 1995), and VanRaden (VanRaden 2008) algorithms in GAPIT 3.0 (Wang and Zhang 2020) with the MLM model (Yu et al. 2006). Based on these results, the EMMA derived K matrix (Supplementary File S5) was identified as the most powerful and used for subsequent analysis of mixed models for all isolates that included CMLM (Zhang et al. 2010), ECMLM (Li et al. 2014), and MLMM (Segura et al. 2012). Lastly, SUPER (Wang et al. 2014), FarmCPU (Liu et al. 2016), and BLINK (Huang et al. 2019) algorithms that reconstruct the kinship matrix were used for a total of nine random effect models. Due to the similarity of results, the MLM and MLMM methods were not used in further analysis. Mixed models included combinations to account for population structure (Q, PC1, and PC5), kinship (EMMA K), and algorithm methods (CMLM, ECMLM, SUPER, FarmCPU, and BLINK) for a total of 15 mixed models per isolate. The mean-squared deviation (MSD) was calculated for each model (Supplementary File S6, Mamidi et al. 2011), however visual inspection of QQ plots were performed to ensure the model was a good fit. This method was employed as models with the lowest MSD model often had highly correlated observed and expected -log 10 (p) values yielding zero significant markers. A Bonferroni correction was calculated at an a level of 0.01 and 0.05 for a -log10(p) threshold of 4.72 and 4.02, respectively. Final Manhattan and QQ plots were generated with R 3.6.3 package CMplot 3.5.1 (https://github.com/YinLiLin/R-CMplot). QTL identification Absolute marker positions were extracted for significant MTAs from the Morex reference genome Mascher et al. 2017;Monat et al. 2019) and compared to collapsed P. teres loci (Clare et al. 2020). Significant markers were declared distinct from previously identified loci, i.e., novel, if the nearest neighboring marker that was closer to previously reported loci was not significant or if the gap to currently delimited locus exceeded 10 Mbp in physical distance when no closer marker was present. Phenotypic analysis Wild barley was found to be statistically (Wilcoxon rank sum test) more resistant to Ptm isolate 13-179 and Ptt isolates GPS18 and UHK77, whereas landrace barley was shown to be statistically more resistant to Ptm isolate GPS263 and Ptt isolate 13-130 ( Figure 1). There was no significant difference between landrace and wild barley to Ptm isolate 13-167. Despite UHK77 being statistically different between landrace and wild barley, no significant MTAs were found. Marker panel analysis Using a 30% missing data threshold, a total of 282 of the 295 barley accessions phenotyped were adequately genotyped. Using a similarity of individual matrix, all remaining individuals were unique. Similarly, 530 of the 598 collapsed markers were deemed to have sufficient coverage across barley accessions using a 30% missing data threshold and were used in the subsequent analysis of the six P. teres isolates. To eliminate markers in linkage disequilibrium, an LD R 2 threshold of 0.8 and sliding window of 50 markers was used, resulting in 522 markers for use in the final analysis. Population structure and linkage disequilibrium STRUCTURE analysis identified an optimal k value ¼ 2, with 89 and 118 individuals in subpopulations one and two (Supplementary Figure S1). Subpopulation one consisted of wild accessions, whereas subpopulation two consisted of landraces. A total of 75 barley accessions comprised of both landrace and wild barley had population membership probabilities of less than 0.8 and were deemed to have an admixture ancestry. The first five principle components accounted for 29. 36, 9.4, 5.3, 4.0, and 3.7%, respectively, in the principle component analysis. Principle components were selected that accounted for at least 25% (PC1) and 50% (PC5) when eigenvalues were plotted on a cumulative scale. Association mapping analysis A total of 24 models were tested on each of the six P. teres isolates consisting of three Ptm and three Ptt isolates. The Ptm isolate 13-167 and Ptt isolate UHK77 contained no significant markers across all models tested. The Ptm isolates GPS263 and 13-179 contained one and three significant markers, respectively. The Ptt isolates GPS18 and 13-130 both contained five significant markers. Ptm isolate GPS263 Only one significant MTA was identified with Ptm isolate GPS263 on chromosomes 5H based on the second version of the cv. Morex reference genome (Monat et al. 2019). The SNP marker 12_20350 located on chromosome 5H at physical position 446449782 was identified at the -log10(p) threshold ¼ 4.72 in the K BLINK . Marker 12_20350 is embedded within the collapsed NBP_QRptt5-1 locus (Wonneberger et al. 2017a;Clare et al. 2020) and was not previously shown to be associated with Ptm interactions. Ptm isolate 13-179 The three significant MTAs identified with Ptm isolate 13-179 were located on chromosomes 3H, 4H, and 5H based on the second version of the Morex reference genome ( Clare et al. 2020), was identified at the -log10(p) threshold ¼ 4.02 in the K FarmCPU and PC1þK FarmCPU models. The marker 11_10510 is located on chromosome 4H at position 603258307 and was identified at the -log10(p) threshold ¼ 4.72 in the K BLINK and K SUPER models and at the -log10(p) threshold ¼ 4.02 in the PC5 GLM , QþK SUPER , and PC1þK SUPER models. In addition, the 11_10510 marker almost met the significance threshold using the K FarmCPU model. The 11_10510 marker is embedded within the Rpt8 locus (Friesen et al. 2006;Tamang et al. 2015;Richards et al. 2017;Vatter et al. 2017;Daba et al. 2019;Clare et al. 2020). Lastly, the marker SCRI_RS_160332 is located on chromosome 5H at position 474799503, $3.0 Mbp distal to the Qrptts-5HL.1 locus . SCRI_RS_160332 was identified with the K FarmCPU model at the -log10(p) threshold ¼ 4.02. Enrichment analysis Enrichment of either resistance or susceptibility alleles at each MTA were calculated for landraces and wild barley (Figure 2, Supplementary Figure S2). For Ptm isolate GPS263 and Ptt isolate 13-130, the landraces that were more resistant than the wild barley showed enrichment for the majority of the resistance alleles or a depletion of susceptibility alleles. For Ptm isolate 13-179 and Ptt isolate GPS18, the opposite was observed as the wild barley showed more resistance and enrichment for the majority of the resistance alleles or depletion of susceptibility alleles. Discussion Both NFNB and SFNB are worldwide threats to barley production and recent evidence shows that both Ptm and Ptt have evolved to infect and threaten wheat production as well (Tó th et al. 2008;Mikhailova et al. 2010;Perelló et al. 2019). Wild barley and landraces from the origin of cereal domestication represent a rich reservoir of net blotch resistance that could be integrated into elite varieties to help mitigate the threat. Analysis of the phenotypic responses of Turkish wild barley and landraces to regional Ptm and Ptt isolates showed evidence of landraces under selective pressures by the pathogen during domestication compared with wild barley as seen by the more compact distribution of the phenotypic scores (Tables 1 and 2, Figure 1). This demonstrates that wild barley harbors additional diversity for net blotch resistance that is not present in the landraces and could be exploited for barley variety development. This is corroborated by enrichment analysis that shows enrichment of the resistant haplotype for marker 11_20866 and 11_20754 in the wild barley lines (Figure 2, Supplementary Figure S2). However, the opposite is true for other loci such as the 11_20968 haplotype near the QRpt-3H.1 locus, that is located approximately 20 Mbp distal of the domestication gene non-brittle rachis 1, btr1 on the chromosome 3HS (Komatsuda et al. 2004;Wang et al. 2019). None of the wild barley accessions analyzed contained the resistance haplotype but it is enriched within landrace accessions (Figure 2, Supplementary Figure S2). The complete lack of the "resistance" marker 11_20968 haplotype in the wild barley could be explained by removal of the "susceptible" haplotype through a selective sweep within close proximity to the btr1 region during domestication. These results show the importance of surveying both landraces and wild barley accessions since important resistance and/or susceptibility loci that interact in the barley-P. teres pathosystem may have been lost or gained through domestication. We have found loci that are unique to wild barley or landraces indicating the importance of analyses of the entire primary barley germplasm pool to identify new sources of resistance for future breeding efforts. To date, only a handful of studies have utilized wild or landrace barley to map resistance loci using biparental populations (Metcalfe et al. 1970;Bockelman et al. 1977;Manninen et al. 2000Manninen et al. , 2006Williams et al. 2003;Koladia et al. 2017) (Figure 4). However, the two isolates for which no significant MTAs were identified may be due to the fact that low marker density was utilized in these analyses. Four of the MTAs were potentially novel and two that mapped to previously identified Ptt resistance loci that had not been reported to be involved in Ptm interactions. Additionally, while the remaining MTAs may not be novel, they may represent important alleles that could be incorporated into breeding programs. Thus, the association mapping identified an abundance of net blotch resistance/susceptibility loci within wild and landrace barley from the center of origin. In this study, barley chromosomes 3H and 7H contained the most MTAs with three, followed by two MTAs on chromosomes 1H, 5H, and 6H and one MTA on chromosomes 2H and 4H (Figure 4, Table 3). Of the 14 MTAs identified, four were identified against the Ptm isolates GPS263 and 13-179. The remaining ten MTAs were identified against Ptt isolates. When selecting the appropriate GWAS algorithm, the BLINK algorithm identified five MTAs, whereas the FarmCPU algorithm identified eleven MTAs, of which only two MTAs overlapped in both BLINK and FarmCPU algorithms. The SUPER algorithm identified one MTA using Ptm isolate 13-179 but this was also identified by the BLINK algorithm. Investigating model selection, the kinship (K) model identified 14 MTA, whereas the mixed kinship and population structure (Q þ K) model only identified one MTA, which was also identified by the K model and therefore no unique MTA. Although we would encourage higher marker saturation in future studies to confirm locus novelty, we would suggest continued best practice of testing multiple models (naïve, K, Q, Q þ K), along with the addition of including all modern association mapping algorithms based on the differing sets of MTAs identified using BLINK and FarmCPU algorithms. The 14 MTAs identified in this study were compared to the collapsed loci of Clare et al. (2020) and a similar strategy of determining novel loci was dependent on the significance of the nearest neighbor marker and previous incorporation into a locus. Using this strategy, we identified four potentially novel loci (QRpt-1H.1, QRpt-3H.1, QRpt-3H.3, QRpt-6H.1) in the barley-Ptt interaction, and two novel loci in the barley-Ptm interaction (QRpt-5H.1 and QRpt-5H.2 corresponding to NBP_QRptt5-1 and Qrptts-5HL.1, respectively) that had been previously described in the barley-Ptt interaction. The novel loci were detected on barley chromosomes 1H, 3H, and 6H. The QRpt-1H.1 and QRpt-6H.1 MTAs were associated in the interaction with Ptt isolate GPS18. QRpt-1H.1 is The data are represented as the mean phenotypic score of each barley class to the respective isolate. Manhattan and QQ plots for two models that show significant markers for more than one of the Pyrenophora teres f. maculata isolates GPS263 and 13-179, and P. teres f. teres isolates GPS18 and 13-130. Two loci that were previously implicated in Ptt resistance were also identified as novel MTAs for Ptm resistance in this study. The first locus, QRpt-5H.1, is located 3.0 Mbp distal to Qrptts-5HL.1 ) with marker SCRI_RS_160332. Additionally, since markers covering the Qrptts-5HL.1 locus were not included in either panel, and due to the close proximity of QRpt-5H.1 Qrptts-5HL.1, we believe that they are the same locus. The second locus, QRpt-5H.2, is embedded within the NBP_QRptt5-1 locus (Wonneberger et al. 2017a). The remaining loci are all embedded within previously identified loci (Table 3). Barley is predominately grown as a feed crop worldwide (IBGSC 2012), however in the United States where corn and soybeans are subsidized and used as feed crops, barley has been outcompeted and acreage has significantly dropped. This has pushed feed barley into less than optimal agricultural land due to its adaptability and hardiness . Quality malting barley demands premium prices because of its use in the multi-billion dollar added value brewing and distilling industry and is now the major class considered in breeding efforts in the US and Europe. However, in some regions of the world where traditional farming practices are still utilized barley is considered an important food crop (Helbaek 1969;Pourkheirandish and Komatsuda 2007;Gec¸it 2016; Ergü n et al. 2017) . Recent studies predicted that the effects of climate change will include higher temperatures, altered precipitation patterns, and higher disease pressure (Dawson et al. 2015), which could result in world malt barley shortages to supply the brewing and distilling industries (Xie et al. 2018). These predictions are beginning to be seen with the stagnating yields experienced in southern Europe (Dawson et al. 2015). Thus, barley breeding must maximize its potential in terms of quality and yield on the land it is currently afforded to sustain the demands for malting. One pillar of support for improving barley would be the introgression of predomestication resistance loci that are absent in current breeding programs to prevent substantial losses to net blotch, an important disease effecting barley production across the globe. Here, we report on the identification of novel loci from Turkish wild barley and landraces that could be introgressed into elite barley varieties. Data availability Isolates are available upon request. The authors ensure that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables. Supplemental data has been submitted to figshare: https://doi.org/10.25387/g3. 14725311. File SF1 contains genotyping data. File SF2 contains marker positions. File SF3 contains phenotyping data. Files SF4 and SF5 contain population structure and EMMA kindship matrix, respectively. File SF6 contains the mean-squared deviations of each association mapping model tested. Author contributions A.C.O., A.K., and R.S.B. conceived the study. A.C.O. and A.K. carried out phenotyping and DNA extractions. K.E. and D.S. carried out sequencing for genotyping data. S.C. and R.S.P. performed genotyping and analysis. S.C. wrote the manuscript with contributions from A.C.O., K.E., R.S.P. D.S., A.K., and R.B. All authors approved the final manuscript.	v3-fos-license
2018-03-14T13:23:58.573Z	2018-03-13T00:00:00.000	3878832	{ "extfieldsofstudy": [ "Computer Science", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://bmcbioinformatics.biomedcentral.com/track/pdf/10.1186/s12859-018-2107-4", "pdf_hash": "ac9d3caf419e8ed34794ab908a906e9d35eb98f9", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45362", "s2fieldsofstudy": [ "Computer Science" ], "sha1": "d3b1355ad8f2ec41ea11b53296d1b92e18fdb8b9", "year": 2018 }	pes2o/s2orc	Watchdog – a workflow management system for the distributed analysis of large-scale experimental data Background The development of high-throughput experimental technologies, such as next-generation sequencing, have led to new challenges for handling, analyzing and integrating the resulting large and diverse datasets. Bioinformatical analysis of these data commonly requires a number of mutually dependent steps applied to numerous samples for multiple conditions and replicates. To support these analyses, a number of workflow management systems (WMSs) have been developed to allow automated execution of corresponding analysis workflows. Major advantages of WMSs are the easy reproducibility of results as well as the reusability of workflows or their components. Results In this article, we present Watchdog, a WMS for the automated analysis of large-scale experimental data. Main features include straightforward processing of replicate data, support for distributed computer systems, customizable error detection and manual intervention into workflow execution. Watchdog is implemented in Java and thus platform-independent and allows easy sharing of workflows and corresponding program modules. It provides a graphical user interface (GUI) for workflow construction using pre-defined modules as well as a helper script for creating new module definitions. Execution of workflows is possible using either the GUI or a command-line interface and a web-interface is provided for monitoring the execution status and intervening in case of errors. To illustrate its potentials on a real-life example, a comprehensive workflow and modules for the analysis of RNA-seq experiments were implemented and are provided with the software in addition to simple test examples. Conclusions Watchdog is a powerful and flexible WMS for the analysis of large-scale high-throughput experiments. We believe it will greatly benefit both users with and without programming skills who want to develop and apply bioinformatical workflows with reasonable overhead. The software, example workflows and a comprehensive documentation are freely available at www.bio.ifi.lmu.de/watchdog. Electronic supplementary material The online version of this article (10.1186/s12859-018-2107-4) contains supplementary material, which is available to authorized users. Background The development of high-throughput experimental methods, in particular next-generation-sequencing (NGS), now allows large-scale measurements of thousands of properties of biological systems in parallel. For example, modern sequencing platforms now allow simultaneously quantifying the expression of all human protein-coding genes and non-coding RNAs (RNA-seq [1]), active translation high-throughput techniques for several conditions or in time-courses in multiple replicates (e.g. [5][6][7]). Analysis of such multi-omics datasets is quite complex and requires a lot of mutually dependent steps. As a consequence, large parts of the analysis often have to be repeated due to modifications of initial analysis steps. Furthermore, errors e.g. due to aborted program runs or improperly set parameters at intermediate steps have consequences for all downstream analyses and thus have to be monitored. Since each analysis consists of a set of smaller tasks (e.g read quality control, mapping against the genome, counting of reads for gene features), it can usually be represented in a structured way as a workflow. Automated execution of such workflows is made possible by workflow management systems (WMSs), which have a number of advantages. First, a workflow documents the steps performed during the analysis and ensures reproducibility. Second, the analysis can be executed in an unsupervised and parallelized manner for different conditions and replicates. Third, workflows may be reused for similar studies or shared between scientists. Finally, depending on the specific WMS, users with limited programming skills or experience with the particular analysis tools applied within the workflow may more or less easily apply complicated analyses on their own data. On the downside, the use of a WMS usually requires some initial training and some overhead for the definition of workflows. Moreover, the WMS implementation itself might restrict which analyses can be implemented as workflows in the system. Nevertheless, the advantages of WMSs generally outweigh the disadvantages for larger analyses. In recent years, several WMS have been developed that address different target groups or fields of research or differ in the implemented set of features. The most well-known example, Galaxy, was initially developed to enable experimentalists without programming experience to perform genomic data analyses in the web browser [8]. Other commonly used WMSs are KNIME [9], an open-source data analysis platform which allows programmers to extend its basic functionality by adding new Java programs, and Snakemake [10], a python-based WMS. Snakemake allows definition of tasks based on rules and automatically infers dependencies between tasks by matching filenames. A more detailed comparison of these WMSs is given in the Results section. In this article, we present Watchdog, a WMS designed to support bioinformaticians in the analysis of large high-throughput datasets with several conditions and replicates. Watchdog offers straightforward processing of replicate data and easy outsourcing of resource-intensive tasks on distributed computer systems. Additionally, Watchdog provides a sophisticated error detection system that can be customized by the user and allows manual intervention. Individual analysis tasks are encapsulated within so-called modules that can be easily shared between developers. Although Watchdog is implemented in Java, there is no restriction on which programs can be included as modules. In principle, Watchdog can be deployed on any operating system. Furthermore, to reduce the overhead for workflow design, a GUI is provided, which also enables users without programming experience to construct and run workflows using pre-defined modules. As a case study on how Watchdog can be applied, modules for read quality checks, read mapping, gene expression quantification and differential gene expression analysis were implemented and a workflow for analyzing differential gene expression in RNA-seq data was created. Watchdog, including documentation, implemented modules as well as the RNAseq analysis workflow and smaller test workflows can be obtained at www.bio.ifi.lmu.de/watchdog. Overview of Watchdog The core features of Watchdog and their relationships are outlined in Fig. 1 and briefly described in the following. More details and additional features not mentioned in this overview are described in subsequent sections, Additional files 1, 2 and 3 and in the manual available at www.bio.ifi.lmu.de/watchdog. Modules Modules encapsulate re-usable components that perform individual tasks, e.g. mapping of RNA-seq data, counting reads for gene features or visualizing results of downstream analyses. Each module is declared in an XSD file containing the command to execute and the names and valid ranges of parameters. In addition to the XSD file, a module can contain scripts or compiled binaries required by the module and a test script running on example data. Module developers are completely flexible in the implementation of individual modules. They can use the programming language of their choice, include binaries with their modules or automatically deploy required software using Conda (https:// conda.io/), Docker (https://www.docker.com/, an example module using a Docker image for Bowtie 2 [11] is included with Watchdog) or similar tools. Furthermore, Watchdog provides a helper bash script to generate the XSD definition file for new modules and (if required) a skeleton bash script that only needs to be extended by the program call. Essentially, any program that can be run from the command-line can be used in a module and several program calls can be combined in the same module using e.g. an additional bash script. In principle, a module could Fig. 1 Overview of Watchdog. a Modules are defined in an XSD format that describes the command to be executed and valid parameters. All modules together represent the software library that can be used in workflows and can be extended by defining new modules. b A workflow is defined in an XML format and consists of tasks that depend on each other. Among others, the XML format allows setting environment variables, defining different executors in the settings part of the workflow and processing replicate data in a straightforward way. c Watchdog parses the workflow, creates the corresponding tasks, executes them and verifies whether execution of each task terminated successfully or not. d Email notification (optional) and log files combined with either the GUI or a simple web-interface allow monitoring the execution of the workflow and intervening if necessary, e.g. by restarting tasks with modified parameters even contain a whole pipeline, such as Maker-P [12], but this would run counter the purpose of a WMS. Here, it would make more sense to separate the individual steps of the pipeline into different modules and then implement the pipeline as a Watchdog workflow. Finally, Watchdog is not limited to bioinformatics analyses, but can be also used for workflows from other domains. Workflows Workflows are defined in XML and specify a sequence of tasks to be executed, the values of their input parameters and dependencies between them. An example for a simple workflow is given in Fig. 2. Among other features that are described later, it is possible to define constants, environment variables and execution hosts in a dedicated settings element at the beginning of the workflow, redirect the standard error and standard output for individual tasks or define how detailed the user is informed on the execution status of tasks. The advantage of XML is that it is widely used in many contexts. Thus, a large fraction of potential Watchdog users should already be familiar with its syntax and only need to learn the Watchdog XML schema. Furthermore, numerous XML editors are available, including plugins for the widely used integrated development environment (IDE) Eclipse [13], which allow XML syntax checking and document structure highlighting. Finally, a number of software libraries for programmatically loading or writing XML are also available (e.g. Xerces for Java, C++ and Perl (http://xerces.apache.org/), ElementTree in Python). In addition, Watchdog also provides an intuitive GUI (denoted workflow designer) that can be used to design a workflow, export the corresponding XML file afterwards and run the workflow in the GUI. Watchdog The core element of Watchdog that executes the workflow was implemented in Java and therefore is, in principle, platform-independent. Individual modules, however, may depend on the particular platform used. For instance, if a module uses programs only available for particular operating systems (e.g. Linux, macOS, Windows), it can only be used for this particular system. As a first step, Watchdog validates the XML format of the input workflow and parses the XML file. Based on the XML file, an initial set of dependency-free tasks, i.e. Simple workflow in XML format. This example shows a simple Watchdog workflow executing a 30 second sleep task. A constant named WAIT_TIME is defined within the settings environment (line 5). Email notification of the user is enabled using the optional mail attribute of the tasks environment (line 8). Here, a task of type sleepTask with id 1 and name sleep is defined (lines 9-13). Either id or name can be used to refer to this task in dependency declarations of other tasks. Within the parameter environment of the sleepTask, values are assigned to required parameters (lines 10-12), which were specified in the XSD file of this particular module. In this case, the parameter wait is set to the value stored in the constant WAIT_TIME (line 11) tasks that do not depend on any other tasks, is generated and added to the WMS scheduler to execute them. Subsequently, the scheduler continuously identifies tasks for which dependencies have been resolved, i.e. all preceding tasks the task depends on have been executed successfully, and schedules them for execution. Once a task is completed, Watchdog verifies that the task finished successfully. In this case, the task generator and scheduler are informed since dependencies of other tasks might have become resolved. In case of an error, the user is informed via email (optional) and the task is added to the scheduler again but is blocked for execution until the user releases the block or modifies its parameters. Alternatively, the user may decide to skip the task or mark the error as resolved. User interfaces Watchdog provides both a command-line version as well as a GUI that can be used to execute workflows and to keep track of their processing. Moreover, a web-interface is provided to GUI and command-line users that displays the status of all tasks in a table-based form and allows monitoring and interacting with the execution of tasks by releasing scheduled tasks, changing parameters after a failed task execution and more (see Fig. 3). The link to the web-interface is either printed to standard output or sent to the user by email if they enabled email notification. In the latter case, the user will also be notified per email about execution failure (always) or success (optional). Finally, the command-line interface also allows resuming a workflow at any task or limiting the execution of the workflow to a subset of tasks using the -start (start execution at specified task), -stop (stop execution after specified task), -include (include this task in execution) and -exclude (exclude this task for execution) options. In the following more details are provided on principles and possibilities of workflow design in Watchdog and defining custom modules. The GUI is described in detail in Additional file 1. Process blocks for creating subtasks Analysis of high-throughput data often requires performing the same analysis steps in parallel for a number of samples representing different conditions or biological or technical replicates. To support these types of analyses, Watchdog uses so-called process blocks to automatically process tasks that differ only in values of parameters, e.g. short read alignment for all FASTQ files in a directory. For this purpose, process blocks define a set of instances, each of which contain one or more variables. For each instance, one subtask is created and subtask placeholders in the task definition are replaced with the variable values of the instance. For the example in which a task is executed for all FASTQ-files in a directory, each instance holds one variable containing the absolute file path of the file. The number of subtasks corresponds to the number of FASTQ-files in the directory. Currently four different types of process blocks are supported by Watchdog: process sequences, process folders, process tables and process input (Fig. 4). In case of process sequences (Fig. 4a) and process folders (Fig. 4b), instances only hold a single variable. Process sequences are comparable to for-loops as they generate instances containing numerical values (integer or floating-point numbers) with a fixed difference between two consecutive numbers (default: 1). Instances generated by process folders contain the absolute path to files and are generated based on a parent folder and a filename pattern. Process tables (Fig. 4c) and process input ( Fig. 4d) blocks can generate instances with multiple variables. Instances generated by a process table are based on the content of a tab-separated file. The rows of the table define individual instances and the columns the variables for each instance. In case of process input blocks, variables and instances Fig. 3 Web-interface of Watchdog. Each line of the table provides information on the status of a task or subtask. The drop-down menu at the end of each line allows to perform specific actions depending on the status of the task. The menu is shown for subtask 1-2, which could not be executed successfully. To generate this screenshot the example workflow depicted in Fig. 6 was processed, which compresses all log-files stored in directory /tmp/. Since the number of simultaneously running subtasks was set to at most 2 for this task, subtask 1-5 is put on hold until subtasks 1-3 and 1-4 have finished or the user manually releases the resource restriction Types of process blocks. With the help of process blocks, multiple tasks that differ only in the parameter values can be created without defining all of them separately. Four different types of process blocks are implemented that fall into two general classes. Instances of the first class contain only a single variable, either (a) a value from a numerical sequence (process sequence) or (b) a path to files (process folder). In (a), subtasks are created based on an integer sequence starting at 5 and ending at 7 with an increment of 1. In (b), a subtask is created for each sh-file in the folder /etc/. Instances of the second type can contain multiple variables, either (c) instances derived from tables (process table) or (d) instances based on return values returned by previous tasks this task depends on (process input). In (c), a table with two columns named name and type and two rows is used as input for the process table. This results in two subtasks for this task, one for each row. The process input block in (d) depends on a task with id 1, which itself had two subtasks. Hence, this task returns two instances, each containing the variables file and fCount obtained from its return variables are derived from return values of preceding tasks the task depends on. Figure 5 shows an example how process blocks can be defined and Fig. 6 shows how they can be used for creation of subtasks. In Additional file 2, a detailed description with examples is provided on how to use process blocks for the analysis of data sets with several replicates or conditions. Furthermore, Watchdog provides a plugin system that allows users with programming skills to implement novel types of process blocks without having to change the original Watchdog code (see Additional file 3). Dependencies By default, all tasks specified in a Watchdog workflow are independent of each other and are executed in a non- Definition of process blocks. In this example, two process blocks are defined within the processBlock environment (lines 2-5). In line 3, a process sequence named num is defined consisting of three instances (1, 5 and 9). In line 4, a process folder selecting all log-files in the /tmp/ directory is defined deterministic order. Alternatively, dependencies on either task or subtask level (details in the next paragraphs) can be defined using the id or name attribute of a task (see Fig. 7). Dependency definitions impose a partial order on tasks, meaning that tasks depending on other tasks will only be executed after those other tasks have finished successfully. Tasks without dependencies or resolved dependencies will still be executed in a non-deterministic order. Although explicit dependency definition adds a small manual overhead compared to automatic identification based on in-and output filenames as in Snakemake, it also provides more flexibility as dependencies can be defined that are not obvious from filenames. For instance, analysis of sequencing data usually involves quality control of sequencing reads, e.g. with FastQC [14], before mapping of reads, and users might want to investigate the results of quality control before proceeding to read mapping. However, output files of quality control are not an input to read mapping and thus this dependency could not be identified automatically. To provide more time to manually validate results of some intermediate steps, Watchdog allows adding checkpoints after individual tasks. After completion of a task with checkpoint, all dependent tasks are put on hold until the checkpoint is released. All checkpoints in Fig. 6 Usage of process blocks. The process block logFiles defined in Fig. 5 is used to generate several subtasks (line 1). These subtasks create compressed versions of the log-files stored in /tmp/. In this case, at most two subtasks are allowed to run simultaneously. Additional file 2 describes how process block variables can be accessed. Here, the placeholder {} is replaced by the variable values stored in the process block, i.e. the complete file paths, and [1] is replaced with the file names (without the '.log' file-ending) (lines 3-4) Fig. 7 Definition of dependencies. The task defined in this example creates subtasks using the process block logFiles from Fig. 5 (line 1) with both task and subtask dependencies. A task dependency on the task sleep defined in Fig. 2 is indicated in line 3. In addition, subtask dependencies to the task with id 2 defined in Fig. 6 are indicated in line 4. In this case, each subtask depends on the subtask of task 2 which was created using the same instance defined by the process block logFiles, i.e. the same file path a workflow can be deactivated upon workflow execution with the -disableCheckpoint flag of the Watchdog command-line version. Task dependencies A task B can depend on one or more other tasks A 1 to A n , which means that execution of task B is put on hold until tasks A 1 to A n have finished successfully. If some of the dependencies A 1 to A n use process blocks to create subtasks, task B is put on hold until all subtasks are finished successfully. Figure 8a illustrates the described behavior on a small example in which task B depends on three other tasks. a) b) Fig. 8 Types of dependencies. Dependencies can either be defined on (a) task or (b) subtask level. a Task B depends on tasks A 1 , A 2 and A 3 . Task A 2 uses a process block to create the three subtasks A 2−1 , A 2−2 and A 2−3 . Task B will be executed when A 1 , A 2 (including all subtasks) and A 3 have finished successfully. b Tasks A and B create subtasks using a process block. For example, task A might decompress files stored in a folder (by using a process folder) and task B might extract data from the decompressed files afterwards (by using a process input block). Here, subtask B x of B only depends on the subtask A x of A based on whose return values it is created Subtask dependencies If a subtask B x of a task B only depends on a particular subtask A x of A instead of all subtasks of A, the definition of subtask dependencies in the workflow allows executing B x as soon as A x has finished successfully (but not necessarily other subtasks of A). This is illustrated in Fig. 8b and can be explained easily for the most simple case when the process block used for task B is a process input block containing the return values of subtasks of A. In that case, a subtask B x depends only on the subtask A x of A that returned the instance resulting in the creation of B x . The use of subtask dependencies is particularly helpful if subtasks of A need different amounts of time to finish or cannot all be executed at the same time due to resource restrictions, such as a limited amount of CPUs or memory available. In this case, B x can be executed as soon as A x has finished but before all other subtasks of A have finished. An example application would be the conversion of SAM files resulting from read mapping (task A) to BAM files (task B). Parallel and distributed task execution By default all tasks are executed one after the other on the host running Watchdog (see Fig. 9a,b). In principle, however, tasks that are independent of each other or individual subtasks of a task can be executed in parallel. Watchdog implements three different types of executors that facilitate parallel execution of tasks: (i) local executor (Fig. 9c), (ii) remote executor (Fig. 9d) and (iii) cluster executor (Fig. 9e). All executors allow multi-threaded execution of tasks. In cases (i) and (ii) Watchdog uses multiple threads for parallel execution of tasks while in case (iii) the cluster master is utilized to distribute tasks on the cluster. Before execution or after completion or failure of tasks, files or directories can be created, deleted or copied to/from remote file systems (e.g. the file system of a remote or cluster executor) using so-called task actions. By default, Watchdog supports virtual file systems based on the protocols File, HTTP, HTTPS, FTP, FTPS and SFTP as well as the main memory (RAM). However, any file system with an implementation of the FileProvider interface from the Commons Virtual File System project of the Apache Software Foundation (http:// commons.apache.org/proper/commons-vfs/) can also be used (see manual). Executors and their resource limitations are declared in the settings element at the beginning of the workflow (see Fig. 10) and assigned to tasks based on their names. Within each workflow, an arbitrary number of executors of different types can be defined and any of these can be assigned to individual tasks. For instance, memory-intensive tasks might be executed on a dedicated high-memory computer using a remote executor while other tasks spawning many subtasks are distributed using Fig. 9 Parallel and distributed task execution. Three different types of executors are implemented in Watchdog: (i) execution on the local host that runs Watchdog, (ii) remote execution via SSH or (iii) cluster execution using DRMAA or the Slurm Workload Manager. a In this example, the four subtasks 1a, 1b, 2a and 2b are created by Watchdog based on tasks 1 and 2 using process blocks. Task 2a depends on 1a, and 2b on 1b. All tasks are assumed to require the same runtime. b By default, one task is executed after the other on the host running Watchdog. c Watchdog also allows parallel execution in all three execution modes (local, remote and cluster execution). d For remote execution, Watchdog establishes a SSH connection to pre-defined execution hosts and randomly distributes the tasks that should be executed to these execution hosts. e For cluster execution, the DRMAA or Slurm master receives tasks to execute and redirects them to its execution hosts. Watchdog has no influence on which execution host is used for task execution because the tasks are distributed by the internal DRMAA or Slurm scheduler. f During slave mode (supported for remote and cluster execution), tasks or subtasks that depend on each other are scheduled on the same execution host, which allows using the local disk space of the host for storage of files that are needed only temporarily but by different tasks a cluster executor and non-resource-intensive tasks are run using a local executor. Here, the number of simultaneously running (sub)tasks can be restricted on task (see Fig. 6) or executor level (see Fig. 10), e.g. to not occupy the whole cluster with many long-running tasks. Provided the name of a particular executor remains the same, everything else can be modified about this executor without having to change the tasks part of the workflow. This includes not only resource limitations or the maximum number of running tasks but even the type of executor, for instance when moving the workflow to a different system. Every host that accepts secure shell connections (SSH) can be used as a remote executor (see Fig. 9d). In this case, a passphrase-protected private key for user authentication must be provided. For cluster execution, any grid computing infrastructures that implement the Distributed Resource Management Application API (DRMAA) can be utilized (see Fig. 9e). By default, Watchdog uses the Sun Grid Engine (SGE) but other systems that provide a DRMAA Java binding can also be used. Furthermore, Watchdog provides a plugin system that allows users with programming skills to add new executor types without having to change the original Watchdog code. This plugin system is explained in detail in Additional file 3 and was used to additionally implement an executor for computing clusters or supercomputers running the Slurm Workload Manager (https://slurm.schedmd.com/). The plugin system can also be used to provide support for cloud computing services that do not allow SSH. Support for the Message Passing Interface (MPI) is not explicitly modeled in Watchdog, but MPI can be used by individual modules if it is supported by the selected executor. Finally, to allow storage of potentially large temporary files on the local hard disk of cluster execution hosts and sharing of these files between tasks, Watchdog also implements a so-called slave mode (see Fig. 9f). In slave mode, the scheduler ensures that tasks or subtasks depending on each other are processed on the same host allowing them to share temporary files on the local file system. For this purpose, a new slave is first started on an execution host, which establishes a network connection to the master (i.e. the host running Watchdog) and then receives tasks from the master for processing. Error detection and handling During execution of workflows, a number of errors can occur resulting either in aborted program runs or incorrect output. To identify such errors, Watchdog implements a sophisticated error checking system that allows flexible extension by the user. For this purpose, Watchdog first checks the exit code of the executed module. By definition an exit code of zero indicates that the called command was executed successfully. However, some tools return zero as exit code regardless of whether the command succeeded or failed. Thus, the exit code alone is not a reliable indicator whether the command was executed successfully. Furthermore, a command can technically succeed without the desired result being obtained. For instance, the mapping rate for RNA-seq data may be very low due to wrong parameter choices or low quality of reads. To handle such cases, the user has the option to implement custom success and error checkers in Java that are executed by Watchdog after a task is finished. Two steps must be performed to use custom checkers: implementation in Java and invocation in the XML workflow (see Fig. 11 for an example and the manual for details). Once the task is finished, the checkers are evaluated in the same order as they were added in the XML workflow. In cases in which both success and error were detected by Fig. 11 Invocation of a custom error checker. The example illustrates how a custom error checker implemented in class CErr located in directory /home/ can be added to a task (line 3). In line 4 and 5, two arguments of type string and integer are forwarded to the constructor of the error checker different checkers, the task will be treated as failed. When an error is detected, the user is informed via email notification (if enabled, otherwise the information is printed to standard output), including the name of the execution host, the executed command, the returned exit code and the detected errors. Information on failure or success is also available via the web-interface, which then allows to perform several actions: (i) modify the parameter values for the task and restart it, (ii) simply restart the task, (iii) ignore the failure of the task or (iv) manually mark the task as successfully resolved. In case of (iii), (sub)tasks that depend on that task will not be executed, but other (sub)tasks will continue to be scheduled and executed. To continue with the processing of tasks depending on the failed task, option (iv) can be used. In this case, values of return parameters of the failed task can be entered manually via the web-interface. Option (i) is useful if a task was executed with inappropriate parameter values and avoids having to restart the workflow at this point and potentially repeating tasks that are defined later in the workflow but are not dependent on the failed task. As Watchdog aims to execute all tasks without (unresolved) dependencies as soon as executors and resource limitations allow, these other tasks might already be running or even be finished. Option (ii) is helpful if a (sub)task fails due to some temporary technical problem in the system, a bug in a program used in the corresponding module or missing software. The user can then restart the (sub)task as soon as the technical problem or the bug is resolved or the software has been installed without having to restart the other successfully finished or still running (sub)tasks. Here, the XSD definition of a module cannot be changed during a workflow run as XSD files are loaded at the beginning of workflow execution, but the underlying program itself can be modified as long as the way it is called remains the same. Option (iii) allows to finish an analysis for most samples of a larger set even if individual samples could not be successfully processed, e.g. due to corrupt data. Finally, option (iv) is useful if custom error checkers detect a problem with the results, but the user nevertheless wants to finish the analysis. Defining custom modules Watchdog is shipped with 20 predefined modules, but the central idea of the module concept is that every developer can define their own modules, use them in connection with Watchdog or share them with other users. Each module consists of a folder containing the XSD module definition file and optional scripts, binaries and test scripts. It should be noted here that while the complete encapsulation of tasks within modules is advantageous for larger tasks consisting of several steps or including additional checks on in-or output, the required module creation adds some burden if only a quick command is to be executed, such as a file conversion or creation of a simple plot. However, to reduce the resulting overhead for module creation, a helper bash script is available for unix-based systems that interactively leads the developer through the creation of the XSD definition file. For this purpose, the script asks which parameters and flags to add. In addition, optional return parameters can be specified that are required if the module should be used as process input block. If the command should not be called directly because additional functions (e.g. checks for existence of input and output files and availability of programs) should be executed before or after the invocation of the command, the helper script can generate a skeleton bash script that has to be only edited by the developer to include the program and additional function calls. Please note that modules shipped with Watchdog were created with the helper script, thus XSD files and large fractions of bash scripts were created automatically with relatively little manual overhead. Once the XSD file for a module is created, the module can be used in a workflow. By default, Watchdog assumes that modules are located in a directory named modules/ in the installation directory of Watchdog. However, the user can define additional module folders at the beginning of the workflow. Example workflows For testing and getting to know the potentials of Watchdog by first-time users, two longer example workflows are provided with the software, which are documented extensively within the XML file (contained in the examples sub-directory of the Watchdog installation directory after configuring the examples, see manual for details). All example workflows can also be loaded into the GUI in order to get familiar with its usage (see Additional file 1). In order to provide workflows that can be used for practically relevant problems, 20 modules were developed that are shipped together with Watchdog. In addition, several smaller example workflows are provided, each demonstrating one particular feature of Watchdog. They are explained in detail in the manual. The next paragraphs describe the two longer example workflows and the corresponding test dataset. Test dataset A small test dataset consisting of RNA-seq reads is included in the Watchdog examples directory. It is a subset of a recently published time-series dataset on HSV-1 lytic infection of a human cell line [5]. For this purpose, reads mapping to chromosome 21 were extracted for both an uninfected sample and a sample obtained after eight hours of infection. Both samples in total contain about 308,000 reads. Workflow 1 -Basic information extraction This workflow represents a simple example for testing Watchdog and uses modules encapsulating the programs gzip, grep and join, which are usually installed on unixbased systems by default. Processing of the workflow requires about 50MB of storage and less than one minute on a modern desktop computer. As a first step, gzipped FASTQ files are decompressed. Afterwards, read headers and read sequences are extracted into separate files. To demonstrate the ability of Watchdog to restrict the number of simultaneously running jobs, the sequence extraction tasks are limited to one simultaneous run, while the header extraction tasks are run in parallel (at most 4 simultaneously). Once the extraction tasks are finished, the resulting files from each sample are compressed and merged. Workflow 2 -Differential gene expression This workflow illustrates Watchdog's potentials for running a more complex and practically relevant analysis. It implements a workflow for differential gene expression analysis of RNA-seq data and uses a number of external software programs for this purpose. Thus, although XSD files for corresponding modules are provided by Watchdog, the underlying software tools have to be installed and paths to binaries added to the environment before running this workflow. The individual modules contain dependency checks for the required software that will trigger an error if some of them are missing. Software required by modules used in the workflow include FastQC [14], ContextMap 2 [15], BWA [16], samtools [17], featureCounts [18], RSeQC [19], R [20], DEseq [21], DEseq2 [22], limma [23], and edgeR [24]. The workflow can be restricted to just the initial analysis steps using the -start and -stop options of the Watchdog command-line version and individual analyses steps can be in-or excluded using the -include and -exclude options. Thus, parts of this workflow can be tested without having to install all programs. Please also note that the workflow was tested on Linux and may not immediately work on macOS due to differences in pre-installed software. Before executing the workflow a few constants have to be set, which are marked as TODO in the comments of the XML file. Processing of the workflow requires about 300MB of storage and a few minutes on a modern desktop computer. The first step is again decompression of gzipped FASTQ files. Afterwards, quality assessment is performed for each replicate using FastQC, which generates various quality reports for raw sequencing data. Subsequently, the reads are mapped to chromosome 21 of the human genome using ContextMap 2. After read mapping is completed, the resulting SAM files are converted to BAM files and BAM files are indexed using modules based on samtools. Afterwards, reads are summarized to read counts per gene using featureCounts. As methods for differential gene expression detection may require replicates, pseudoreplicates are generated by running featureCounts twice with different parameters. This was done in order to provide a simple example that can be executed as fast as possible and should not be applied when real data is analyzed. In parallel, quality reports on the read mapping results are generated using RSeQC. Finally, limma, edgeR, DEseq and DEseq2 are applied on the gene count table in order to detect differentially expressed genes. All four programs are run as part of one module, DETest, which also combines result tables of the different methods. Several of the provided modules also generate figures using R. Comparison with other WMSs Most WMSs can be grouped into two types based on how much programming skills are required in order to create a workflow. If a well-engineered GUI or web interface is provided, users with basic computer skills should be able to create their own workflows. However, GUIs can also restrict the user as some features may not be accessible. Hence, a second group of WMSs addresses users with more advanced programming skills and knowledge of WMS-specific programming or scripting languages. As a comprehensive comparison of all available WMS is outside the scope of this article, two commonly used representatives of each group were selected and compared with Watchdog. Figure 12 lists features of each WMS, which are grouped into the categories setup, workflow design, workflow execution and integration of new tools. As representative WMSs Galaxy [8], KNIME [9], Snakemake [10], and Nextflow [25] were chosen. In the following paragraphs, the selected WMSs are discussed. Because all four WMSs as well as Watchdog allow nonprogrammers to execute predefined workflows, this property is not further discussed. Furthermore, an analysis of the computational overhead of Watchdog and Snakemake showed that the computational overhead of using either WMS (and likely any other) is negligible compared to the actual runtime of the executed tasks (see Additional file 4). Galaxy The most well-known WMS for bioinformatic analyses is Galaxy [8]. It was initially developed to enable experimentalists without programming experience to perform genomic data analyses in the web browser. Users can upload their own data to a Galaxy server, select and combine available analysis tools from a menu and configure them using web forms. To automatically perform the same workflow on several samples in a larger data set, so-called collections can be used. In addition to computer resources, Galaxy provides a web-platform for sharing tools, datasets and complete workflows. Moreover, users can set up private Galaxy servers. In order to integrate a new tool, an XML-file has to be created that specifies the input and output parameters. Optionally, test cases and the expected output of a test case can be defined. Once the XML-file has been prepared, Galaxy must be made aware of the new tool and be re-started. If public Galaxy servers should be used, all input data must be uploaded to the public Galaxy servers. This is especially problematic for users with only low-bandwidth internet access who want to analyze large high-throughput datasets but cannot set up their own server. In summary, Galaxy is a good choice for users with little programming experience who want to analyze data using a comfortable GUI, might not have access to enough computer resources for analysis of large high-throughput data otherwise, appreciate the availability of a lot of predefined tools and workflows and do not mind the manual overhead. KNIME The Konstanz Information Miner, abbreviated as KNIME [9], is an open-source data analysis platform implemented in Java and based on the IDE Eclipse [13]. It allows programmers to extend its basic functionality by adding so-called nodes. In order to create a new node, at least three interfaces must be implemented in Java: (i) a model class that contains the data structure of the node and provides its functionality, (ii) view classes that visualize the results once the node was executed and (iii) a dialog class used to visualize the parameters of the node and to allow the user to change them. One disadvantage for node developers is that the design of the dialog is labor-intensive, in particular for nodes that accept a lot of parameters. Another shortcoming of KNIME is that only Java code can be executed using the built-in functionality. Hence, wrapper classes have to be implemented in Java if a node requires external binaries or scripts. Furthermore, KNIME does not support distributed execution in its free version. However, two extensions can be bought that allow either workflow execution on the SGE or on a dedicated server. Hence, the free version of KNIME is not suitable for the analysis of large high-throughput data. However, KNIME can be used by people without programming skills for the analysis of smaller datasets using predefined nodes, especially, if a GUI is required that can be used to interactively inspect and visualize the results of the analysis. Snakemake A workflow processed by Snakemake [10] is defined as a set of rules. These rules must be specified in Snakemake's Fig. 12 Comparison of Watchdog with other WMSs. Comparison was performed using features grouped into the categories setup, workflow design, workflow execution and integration. Workflow is abbreviated as workf. in this table. Integration refers to the integration of new data analysis tools into the particular WMS. Footnotes: 1 six non-free extensions are available; 2 since version 2.4.8, rules can also explicitly refer to the output of other rules; 3 explanation: includes a way to automatically run a predefined workflow for a variable number of replicates based on filename patterns; 4 have to be created manually in the web-interface from uploaded files; 5 explanation: finished steps of the workflow can return variables that are used by subsequent steps as input; 6 can only return the names of output files; 7 other supported executors: Watchdog: new executors can be added with the plugin system, Galaxy: PBS/Torque, Open Grid Engine, Univa Grid Engine, Platform LSF, HTCondor, Slurm, Galaxy Pulsar, Snakemake: can also use cluster engines with access to a common file system and a submit command that accepts shell scripts as first argument, Nextflow: SGE, LSF, Slurm, PBS/Torque, NQSII, HTCondor, Ignite; 8 non-free extensions for SGE or dedicated server support are available; 9 custom executors for cloud computing services can be created using the plugin system; 10 Watchdog: HTTP/S, FTP/S and SFTP by default, can be extended to any remote file system with an implementation of the FileProvider interface from the Commons Virtual File System project, Galaxy: Object Store plugins for S3, Azure, iRODS, Snakemake: S3, GS, SFTP, HTTP, FTP, Dropbox, XRootD, NCBI, WebDAV, GFAL, GridFTP. Nextflow: HTTP/S, FTP, S3; 11 a hard-coded error checker triggered on keywords 'exception' and 'error' in standard output and error is provided; 12 depends on the node implementation and left to developer; 13 explanation: usage of local storage during distributed execution in order to avoid unnecessary load on the shared storage system; 14 direct integration of python code is possible; 15 own scripting language available; 16 explanation: describes the concept used to separate workflow definition and functionality (e.g. Watchdog's modules) in order to allow easy re-use of functionality; 17 modules can include binaries in the module directory or automatically deploy required software using Conda, Singularity, Docker or similar tools available on the used system own language in a text file named Snakefile. Similar to GNU Make, which was developed to resolve complex dependencies between source files, each rule describes how output files can be generated from input files using shell commands, external scripts or native python code. At the beginning of workflow execution, Snakemake automatically infers the rule execution order and dependencies based on the names of the input and output files for each rule. From version 2.4.8 on, dependencies can also be declared by explicitly referring to the output of rules defined further above. Workflows can be applied automatically to a variable number of samples using wildcards, i.e. filename patterns on present files. In Snakemake, there is no clear separation between the tool library and workflow definition as the command used to generate output files is defined in the rule definition itself. Starting with version 3.5.5, Snakemake introduced re-usable wrapper scripts e.g. around command-line tools. In addition, it provides the possibility to include either individual rules or complete workflows as sub-workflows. Thus, Snakemake now allows both encapsulation of integrated tools as well as quickly adding commands directly into the workflow. By default, no new jobs are scheduled in Snakemake as soon as one error is detected based on the exit code of the executed command. Accordingly, the processing of the complete workflow is halted until the user fixes the problem. This is of particular disadvantage if time-consuming tasks are applied on many replicates in parallel and one error for one replicate prevents execution of tasks for other replicates. While this default mode can be overridden by the -keep-going flag, this flag has to be set when starting execution of the workflow and applies globally independent of which particular parts of the workflow caused the error. In addition, the option -restart-times allows automatically restarting jobs after failure for a predefined number of times and each rule can specify how resource constraints are adapted in case of restarts. However, this option is only useful in case of random failure or failure due to insufficient resources. If errors result from incorrect program calls or inappropriate parameter values, restarting the task will only result in the same error again. Finally, Snakemake is the only one of the compared WMSs that does not provide return variables that can be used as parameters in later steps. In summary, Snakemake is a much improved version of GNU Make. Programmers will be able to create and execute own workflows using Snakemake once they learned the syntax and semantic of the Snakemake workflow definition language. However, as Snakemake does not offer a GUI or editor for workflow design, most experimentalists without programming skills will not be able to create their own workflows. Nextflow The idea behind the WMS Nextflow [25] is to use pipes to transfer information from one task to subsequent tasks. In Unix, pipes act as shared data streams between two processes whereby one process writes data to a stream and another reads that data in the same order as it was written. In Nextflow, different tasks communicate through channels, which are equivalent to pipes, by using them as input and output. A workflow consists of several tasks, which are denoted as processes and are defined using Nextflow's own language. The commands that are executed by processes can be either bash commands or defined in Nextflow's own scripting language. Nextflow also provides the possibility to apply a task on a set of input files that follow a specific filename pattern using a channel that is filled with the filenames at runtime. By default, all running processes are killed by Nextflow if a single process causes an error. This is particularly inconvenient if tasks with long runtimes are processed (e.g. transcriptome assembly based on RNA-seq reads). However, alternative error strategies can be defined for each task before workflow execution, which allow to either wait for the completion of scheduled tasks, ignore execution errors for this process or resubmit the process. In the latter case, computing resources can also be adjusted dynamically. In Nextflow, there is no encapsulation of integrated tools at all since the commands to execute are defined in the file containing the workflow. While this is advantageous for quickly executing simple tasks, reusing tasks in the same or other workflows requires code duplication. Furthermore, Nextflow also does not offer a GUI for workflow design, which makes it hard for beginners to create their own workflows as they must be written in Nextflow's own very comprehensive programming language. Conclusion In this article, we present the WMS Watchdog, which was developed to support the automated and distributed analysis of large-scale experimental data, in particular next-generation sequencing data. The core features of Watchdog include straightforward processing of replicate data, support for and flexible combination of distributed computing or remote executors and customizable error detection that allows automated identification of technical and content-related failure as well as manual user intervention. Due to the wide use of XML, most potential users of Watchdog will already be familiar with the syntax used in Watchdog and only need to learn the semantic. This is in contrast to other WMSs that use their own syntax. Furthermore, Watchdog's powerful GUI also allows nonprogrammers to construct workflows using predefined modules. Moreover, module developers are completely free in which software or programming language they use in their modules. Here, the modular design of the tool library provides an easy way for sharing modules by simply sharing the module folder. In summary, Watchdog combines advantages of existing WMSs and provides a number of novel useful features for more flexible and convenient execution and control of workflows. Thus, we believe that it will benefit both experienced bioinformaticians as well experimentalists with no or limited programming skills for the analysis of large-scale experimental data.	v3-fos-license
2021-02-14T06:16:18.353Z	2021-02-01T00:00:00.000	231911382	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2076-2607/9/2/350/pdf", "pdf_hash": "a7c931a9cef23f6c62fbc647d2f468a8644d1518", "pdf_src": "MergedPDFExtraction", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45363", "s2fieldsofstudy": [ "Environmental Science", "Biology" ], "sha1": "8397cdc409de74385d45b76cd2a528eb0cebe403", "year": 2021 }	pes2o/s2orc	Streptococcal Infections in Marine Mammals Marine mammals are sentinels for the marine ecosystem and threatened by numerous factors including infectious diseases. One of the most frequently isolated bacteria are beta-hemolytic streptococci. However, knowledge on ecology and epidemiology of streptococcal species in marine mammals is very limited. This review summarizes published reports on streptococcal species, which have been detected in marine mammals. Furthermore, we discuss streptococcal transmission between and adaptation to their marine mammalian hosts. We conclude that streptococci colonize and/or infect marine mammals very frequently, but in many cases, streptococci isolated from marine mammals have not been further identified. How these bacteria disseminate and adapt to their specific niches can only be speculated due to the lack of respective research. Considering the relevance of pathogenic streptococci for marine mammals as part of the marine ecosystem, it seems that they have been neglected and should receive scientific interest in the future. Introduction The marine ecosystem is challenged by numerous factors such as anthropogenic pollution including wastewater [1,2], plastic [3], chemical [4,5], and noise pollution [6], fisheries and prey depletion [7,8], offshore-construction [4], shipping [9,10], harmful algal blooms [11,12], climate change, and acidification [13][14][15][16][17]. Marine mammals are considered sentinels for the marine ecosystem [18][19][20] and, thus, the assessment of their health status should be of global interest and importance. Cumulative effects caused by different anthropogenic activities (e.g., pollution of the environment with PCBs, PBDEs, microplastic) can have suppressive effects on the immune system of marine mammals and this might result in a higher susceptibility for infectious diseases [21][22][23][24]. There are a number of studies indicating a higher prevalence of or risk for infectious disease in correlation with those (anthropogenic induced) threatens. For instance, in a recent study, microplastic was found in each of 50 tested animals belonging to 10 different marine mammal species. Furthermore, animals that died due to infectious diseases showed a slightly higher amount of microplastic in the intestines [3]. Sanderson and Alexander [17] found that climate related factors such as sea surface temperature have a significant effect on the occurrence of infectious disease-induced mass mortality events. In another study, the risk for infectious disease mortality increased by 2% for each 1 mg/kg increase of polychlorinated biphenyls (PCBs) in the blubber of harbor porpoises [25]. Infectious diseases are one of the most frequent causes of death in marine mammals [26-33] and beta-hemolytic streptococci are frequently isolated and associated with disease [26,30,31,[34][35][36][37]. The genus Streptococcus belongs to the family Streptococcaceae, the order Lactobacillales, the class Bacilli, and the phylum Firmicutes. Streptococci are non-motile, Gram-positive, catalase-negative, non-spore forming, and chemo-organotrophic with a Streptococcal Findings in Marine Mammals To the best of our knowledge, 10 streptococcal species were isolated and identified more than once from 23 species of Pinnipedia and Cetaceae worldwide (Figure 1, Supplementary Table S1). Streptococcus agalactiae S. agalactiae, also known as a bovine and human pathogen [59][60][61], was isolated from a wound and navel infection of grey seals (Halichoerus grypus) in North Rona in 1985 and 1986 [62] and from lesions of fight wounds, pneumonia, lymphadenitis as well as from lung and spleen samples of Antarctic fur seals (Arctocephalus gazella) from 1984-1987 on Bird Island, South Georgia [63]. Later, it was isolated from lesions and visceral organs (liver and lung) of a captive common bottlenose dolphin (Tursiops truncatus) that suffered from fatal necrotizing fasciitis and myositis [64]. One year later, the isolation of S. agalactiae from epaxial muscles of a wild stranded bottlenose dolphin was reported [65]. This strain caused 90% mortalities in tilapia in experimental infections and showed high similarity with strains associated with mullet kill in the concurrent Kuwait Bay. A mullet was found in the stomach of the dead dolphin, which might have served as a possible way of transmission. A study of human S. agalactiae strains from fish, seals, a dolphin, and a frog indicated zoonotic and anthroponotic hazard by causing severe disease in fish and compromising food security [66]. Between 2012 and 2014, S. agalactiae was isolated from a stranded grey seal on the British coast with ocular pathology [67]. In the Waikiki Aquarium, Honolulu, Haiwaii, S. agalactiae was isolated from two male healthy Hawaiian monk seals (M. schauinslandi) as part of their aerobic bacterial flora in the nasal cavity [68]. S. agalactiae is also known as serious fish pathogen [69][70][71]. In Brazil, high virulent strains were isolated from diseased Nile tilapia and transmission occurred by direct contact or through water [70]. Infection experiments confirmed the disease and revealed low LD 50 for Nile tilapia. However, isolates from cattle did not cause any clinical signs in Nile tilapia and channel catfish indicating host specification and adaptation [72]. Human and bovine strains of S. agalactiae were able to cause disease in Nile tilapia, although there was no genetic relatedness of strains from fish, bovine, and human origin [73]. This suggests that the ability to cross host-specific barrier is not necessarily reflected by genetic linkage. Virulence gene profiling of S. agalctiae isolated from diseased tilapia in Thailand revealed a positive correlation of virulence genes content and pathogenicity [74]. Virulence genes for adhesion, invasion, and immune evasion were identified. Another study demonstrated that there were fish-specific genes or loci that were associated with disease in fish, while strains missing these regions were not able to cause morbidity in tilapia [75]. In addition, these fish-specific genes were mainly clustered in regions with signatures of mobile genetic elements and one fish-specific gene was found in the region, where the virulence genes rib or bca are in the human strain indicating genetic adaptation to the fish host. Streptococcus bovis S. bovis has been isolated from the gastrointestinal tract and feces of cattle, sheep, goats [76], and dogs [77]. It has also been identified as human pathogen associated with endocarditis [78], meningitis [79], septic arthritis [80], bacteremia, and gastrointestinal disease [81]. Virulence factors associated with S. bovis infection were, for instance, extracellular proteins [82] and antigens [83]. S. bovis was detected in fur seals with pneumonia that was characterized by extensive polymorphonuclear infiltrations and necrosis or very widespread abscess formation and, frequently, by additionally fibrinous exudative pleurisy [63]. Together with S. phocae and S. canis it was also isolated from dead herpesvirus-positive harbor seal pups at the Smith Island, Washington [32]. A monk seal pup (Monachus monachus) found on the island Deserta Grande, Portugal died due to septicemia and S. bovis was isolated and considered as a potential causative agent [84]. In 2006, S. bovis isolation (together with S. equisimilis/mitis) from free-ranging bottlenose dolphins that were captured, sampled, and released in coastal Gulf of Mexico and Atlantic ocean waters was reported [85]. Streptococcus canis S. canis was first isolated from cows with mastitis and from dogs with different pathological findings [86], but can also cause infections in humans [56,57,87]. It has also been isolated from minks [88], feral cats [89] and cats [90]. Its virulence and pathogenicity were confirmed by the presence of virulence genes such as for fibronectin-binding protein, M proteins, protective antigen, and streptolysin [91][92][93][94]. The M protein of S. canis has also a high-affinity immunoglobulin G binding activity, which is not species-specific and facilitate S. canis to interact with different hosts [95]. A beta-hemolytic Streptococcus sp., biochemically similar to S. canis, was cultured from pyogranulomatus lesion of the laryngeal cartilages and epiglottis of an adult harbor seal (Phoca vitulina) [27]. S. canis was also isolated from peritoneal effusion of a captive California sea lion (Zalophus californianus) of the US Navy's marine mammal program [96] and from a California sea lion with bilateral corneal ulceration of the London Zoo, UK [97]. During an increased mortality among South American fur seal (Arctocephalus australis) pups on Guafo Island, Chile, South America, S. canis together with S. marimammalium were isolated and associated with moderate to marked, multifocal, mucopurulent bronchopneumonia [98]. In August 1994, S. canis (and also S. phocae, S. equi subsp. zooepidemicus) was isolated from spleen, liver, and kidney of Cape fur seals (Arctocephalus pusillus pusillus) at Cape Cross, Namibia that suffered from respiratory infections and abortions associated with starvation [99]. Seven cases of stranded harbor porpoises (Phocoena phocoena) in England and Wales between 1990 and 1996 had a S. canis septicemia, which was isolated from lungs with pulmonary abscesses and enlarged pulmonary associated lymph nodes [100]. S. canis (together with S. phocae) was cultured from blood and lung samples of a dead, stranded Northern fur seal (Callorhinus ursinus) with necrotizing and fibrinous pneumonia infiltrated by band neutrophils with intraluminal abscess of bronchi at the coast of Niigata, Japan [46]. In 2005, the isolation of S. canis from two dead harbor seal pups on Smith Island, Washington was reported [32,101]. One died from omphalophlebitis and the other from omphalitis with subsequent peritonitis. S. canis was also isolated from the oral cavity of a male, healthy Hawaiian monk seal of the Waikiki Aquarium, Honolulu, Hawaii [68]. Streptococcus dysgalactiae S. dysgalactiae subsp. equisimilis, previously known as S. equisimilis [102] and found in humans [103] and different animals such as dogs, cows, pigs, and horses [104], was isolated from Antarctic fur seal pups with septicemia and rhinitis in South Georgia, UK between 1979-1982 [105] and 1986 from a grey seal cow on North Rona, Scotland [62]. In the years 1988 and 1989, an increased number of harbor porpoise carcasses was observed in North and Baltic Seas [36]. Thirty-five isolates of beta-hemolytic streptococci were classified in Lancefield group L and identified as S. dysgalactiae subsp. dysgalactiae. In 1997, S. dysgalactiae and S. dysgalactiae subsp. equisimilis isolates were found in a dead, wild monk seal pup in association with a septicemia on the island Deserta Grande, Portugal [84]. Three isolates identified as S. dysgalactiae subsp. dysgalactiae were obtained from phocid seals (harbor and grey seals) stranded in the North and Baltic Seas between 1995 and 2000 [106]. Between 2005-2011, pathologic and microbiological findings of a southern right whale (Eubalaena australis) calf from Brazil indicated that beta-hemolytic S. dysgalactiae septicemia was responsible for the death of the animal [107]. Streptococcus equi S. equi causes infections in horses [108] and was associated with canine infectious respiratory disease [109]. A systemic infection with S. equi in a horse handler has also been reported [110]. Further studies confirmed the zoonotic potential of S. equi [53,111]. In November 1978, a female North Atlantic pilot whale (Globicephala melaena) suffering from bronchopneumonia with lesions stranded on Metis Beach, Canada and S. equi (no further identification) was isolated from lung parenchyma, pharynx, and pericardial fluid [35]. A study from 1980 reported the isolation of S. equi subsp. zooepidemicus (previously S. zooepidemicus) from grey seals associated with mild, purulent pneumonia [49]. In 1994, it was isolated from the conjunctiva and trachea of two adult female Cape fur seals that had septicemic S. phocae in Namibia [99]. A total of 32 beta-hemolytic streptococcal isolates, collected during the phocine distemper outbreak in 2002 from 28 different harbor seals of the German North Sea, were identified as S. equi subsp. zooepidemicus [112]. Later, the same scientists isolated the same or at least very close related strains of S. equi subsp. zooepidemicus from grey seals and other harbor seals [113]. A retrospective study on 42 dead bottlenose dolphins from the US Navy Marine Mammal Program during 1980-2010 demonstrated an association of the isolation of S. equi subsp. zooepidemicus with pneumonia [114]. 16S rRNA sequences for S. equi (and S. phocae) were found in blow samples collected from four wild healthy Indo-Pacific bottlenose dolphins (T. aduncus) in Shark Bay (SB), Western Australia, in 2012 [115]. Streptococcus halichoeri S. halichoeri, characterized as non-hemolytic and classified in Lancefield group B, was first isolated from dead grey seals in Iverness and Cornwall, UK [50] and few years later, in 2012, also from the kidney of a stranded, female Stellar sea lion (Eumetopias jubatus) in South Korea [116]. Also, in 2012, a severe case of a human infection with S. halichoeri was reported [117]. The patient had no contact to seals, but to fish, which could have been a possible transmission route. However, this was not tested. Another human infection was reported in 2018, where a man suffered from skin cellulitis due to S. halichoeri [118]. Shewmaker et al. [119] compared human and seal strains and concluded two subspecies S. halichoeri subsp. halichoeri for the seal isolates and S. halichoeri subsp. hominis for strains associated with human infections. The core genome of 20 S. halichoeri isolates from different hosts including dogs and minks contained 19 different streptococcal virulence factors, of which most were associated with adherence followed by proteases and toxins emphasizing its pathogenic potential [120,121]. Streptococcus iniae S. iniae was described as new species in 1976, when it was first isolated from a captive Amazon freshwater dolphin (Inia geoffrensis) suffering from a dermatologic syndrome called "golf ball disease" in the Steinhart Aquarium, San Francisco, USA [122]. Further isolates were obtained from captive freshwater dolphins housed at the Niagara Falls Aquarium in New York, USA two years later [123], and in 1983 from a captive Amazon River dolphin at the Pittsburgh Zoo, USA that also developed the "golf ball disease" [124]. In 2015, a common dolphin died due to bacterial septicemia at Beijing aquarium, China, where S. iniae was isolated from hilar lymph nodes and pancreas of the dolphin [125]. S. iniae is also a serious fish pathogen [58,126]. Virulence mechanisms include a capsule with antiphagocytic function [127], the cytotoxin ß-hemolysin streptolysin S [128], an extracellular nuclease and s secreted nucleotidase that play an important role in immune evasion [129], a polysaccharide deacetylase involved in adherence, invasion, lysozyme resistance and survival in fish blood [130], and M-like protein [131]. Comparative genomics revealed genetic differences between strains from different hosts including I. geoffrensis and identified two plasticity zones that reflect adaptation to specific host environments [132]. Furthermore, the dolphin isolates differed from the fish and human isolates in lacking a capsule, forming denser and thicker biofilms, increased ability to withstand oxidative stress and were genetically highly divergent to the other isolates [133]. In addition, there were conserved mutation rates and mismatch/oxidized-guanine repair systems within phylogenetic clades, but significant differences between major phylogenetic lineages. Mutators might facilitate adaptation to novel hosts including immune escape. This indicates that S. iniae has the genetic repertoire to adapt very well to many different hosts. Streptococcus marimammalium S. marimammalium was first isolated from the lung of a dead harbor seal and a dead grey seal in Iverness, Scotland [51]. In 2007/2008, it was also isolated (together with S. agalactiae and S. canis) from nasal and oral swabs of two healthy Hawaiian monk seals from the Waikiki Aquarium, Honolulu, Hawaii [68]. In 2016, it was also isolated from South American Fur Seal Pups with moderate to marked, multifocal, mucopurulent bronchopneumonia on Guafo Island, Chile, South America [98]. To our knowledge, nothing is known about virulence factors and pathogenicity of S. marimammalium. Streptococcus mitis S. mitis is mainly known as member of the human oral cavity [134,135] and as opportunistic pathogen causing endocarditis and bloodstream infections in neutropenic and immunocompromised patients [136][137][138]. It is closely related to the human pathogen S. pneumoniae and its genome contain virulence genes involved in colonization and adherence, which might also be important for commensals to interact with host cells [139]. However, genes for hyaluronidase and capsular genes were absent. S. mitis was isolated in 1985 from a blowhole swab of a captive, healthy white whale (Delphinapterus leucas) 139 days after captivity at the Mystic Marinelife Aquarium Connecticut, USA [140]. In 1985, it was isolated from lesions of a grey seal with peritonitis in North Rona, Scotland [62] and between 2012-2014 from clinically normal eyes of two grey seals stranded on the British coast [67]. These findings suggest that S. mitis might also be a commensal in some marine mammals. The commensalism and pathogenesis of S. mitis is reviewed by Mitchel, 2011 [141]. Streptococcus phocae S. phocae was first isolated and identified from lung, liver, spleen, and kidney samples of harbor seals suffering from pneumonia with areas of consolidation, purulent exudate in the bronchi, interlobular edema, and emphysema during a phocine distemper virus outbreak in northwestern Europe [34]. Later, S. phocae was also isolated from diseased Atlantic salmon [142,143], stranded southern sea otters [144], and as gut commensal of Indian white shrimp [145]. Two subspecies are described, S. phocae subsp. salmonis for isolates from Atlantic salmon and S. phocae subsp. phocae for isolates from seals [146]. In August 1994, beta-hemolytic streptococci with high similarity to S. phocae were isolated from spleen, liver, and kidney of Cape fur seals at Cape Cross, Namibia that suffered from respiratory infections and abortions [99]. A total of 69 S. phocae isolates were obtained from harbor and grey seals from the North and the Baltic Sea investigated between 1995 and 2000 [106]. A study on phocid seals (harbor and grey seals) that were older than 19 months from the North Sea of Schleswig-Holstein, Germany reported two S. phocae isolates from intestines of phocid seals with intestinal displacements [147]. During diagnostic evaluation by the Animal Health Center, Abbotsford, British Columbia, Canada S. phocae was isolated from harbor seals with an increase of prevalence since 2000, ringed seal (P. hispida) pups from arctic Canada and two stranded harbor porpoises from Washington State [48]. In spring and summer 2000, more than 10,000 Caspian seals (Pusa caspica) were found dead with canine distemper virus infection as primary diagnosis [47]. The investigated animals suffered from broncho-interstitial pneumonia, lymphocytic necrosis and depletion in lymphoid organs, and the presence of typical intracytoplasmic inclusion bodies in multiple epithelia. S. phocae was isolated from three of eight animals. Between 2001 and 2003, vaginal and preputial swabs of California Sea Lions were collected for investigations of genital bacterial infections and urogenital carcinoma [37]. S. phocae was isolated from three specimen of cancer and three specimens of non-cancer animals stranded along the central and northern California coast. In November 2007, a short-beaked common dolphin (Delphinus delphis) stranded at La Graciosa, Canary Islands [45]. Diagnostic evaluation revealed bacterial septicemia, fibrino-necrotizing to pyogranulomatous dermatitis and panniculitis, embolic pneumonia, neutrophilic and lymphoplasmacytic meningo-choroiditis, random neutrophilic hepatitis, lymphoplasmacytic myocarditis and epicarditis, necrotizing adrenalitis, suppurative endometritis, and multicentric reactive lymphadenopathy cutaneous purulent nodules in the tail fluke, vegetative mitral valve endocarditis, and presumed postpartum pyometra. S. phocae could be cultured from lung, brain, and adrenal gland tissue. Morbillivirus was detected in the epithelium of the choroid plexus of the fourth ventricle. In November 2009, a female spotted seal (Phoca largha) stranded at Kotzebue Sound, Alaska and was diagnosed with pyometra [148]. S. phocae was isolated from the purulent discharge in uterine contents. Three Navy bottlenose dolphins (T. truncatus) developed in the time between 2009 and 2010 a strangles-like syndrome associated with S. phocae, which was isolated after the animals showed clinical signs such as inflammatory hemogram, neutrophilic leukocytosis, and unilateral cervical lymphadenopathy [149]. [115]. In February 2014, S. phocae was isolated from a carcass of a subadult male northern seal at the coast of Niigata, Japan that suffered from necrotizing and fibrinous pneumonia with diffuse abscesses of all lung lobes and massive necrosis of kidney and liver [46]. Between 2010 to 2017 the health of captive and stranded Alaskan ice seals were investigated and S. phocae isolates were obtained from blood, abscess, and lymph node samples from ringed seals [152]. Harbor seals stranded at the coast of San Juan County, Washington, USA between 2002 to 2018 were examined and from one adult female animal a fatal septicemia caused by S. phocae was reported [153]. While the presence of an antiphagocytic capsule and virulence of S. phocae subsp. salmonis to Atlantic salmon has been demonstrated in infectivity experiments [142,154,155], whole genome analyses of S. phocae subsp. phocae identified typical streptococcal virulence factors such as fibronectin-binding proteins, the toxin streptolysin S and genes encoding for a capsule [156]. Invasion of fish and mammalian cell lines by S. phocae subsp. phocae has also been shown and confirmed its pathogenic potential [154]. However, S. phocae subsp. phocae also seems to be a commensal of the oral cavity of grey seals as revealed by microbiome analyses and 16S rRNA sequencing. A transmission of S. phocae to harbor porpoises via bites is also indicated [157] and S. phocae might be an opportunistic pathogen, at least for seals. Streptococcus viridans Group In very few studies, streptococci isolated from marine mammals were identified as members of the S. viridans group (viridans streptococci), which includes streptococci that are usually alpha-hemolytic and inhabit the oral cavity, intestinal, and vaginal tract [158][159][160]. This group is very heterogeneous and includes species such as S. anginosus, S. mitis, S. sanguinis, S. mutans, and S. salivarius, which can also cause endocarditis [161], bacteremia [162], and respiratory infections [163]. Viridans streptococci were isolated from superficial abscesses, wounds, ocular and urethral discharges, and umbilici of live and from lung and liver samples of dead elephant seals, California sea lions and harbor seals stranded between January 1994 and December 1995 along the California Coast [164]. Viridans streptococci were isolated in mixed cultures with Arcanobacterium phocae from California sea lions, harbor seals, Northern elephant seals, sea otter and common dolphin stranded along the central California coast between 1994 and 2000 [165]. In Beluga whales that stranded at Cook Inlet (Alaska, USA) between 1998 and 2013 an isolate was identified as member of the S. viridans group [166]. Also, viridans streptococci were isolated from gastric fluid samples of free-ranging bottlenose dolphins from the southeastern United States during a catch and release health assessment between 2003 to 2005 [167]. One-Time Only Detections of Streptococcal Species from Marine Mammals In studies described above, streptococcal species have been isolated and identified at least twice or more. In the following, reports on one-time only descriptions of streptococcal species are summarized. S. uberis was found in dead free-ranging male Antarctic fur seals with pneumonia and extensive polymorphonuclear infiltrations and necrosis or very widespread abscess formation and frequently there was an associated fibrinous exudative pleurisy [63]. S. oralis was isolated and identified by API strips from three swabs taken from healthy eyes of free-ranging grey seals stranded on the British coast between November 2012 and February 2014 [67]. In a metagenome dataset of blood, muscle, and fecal samples of a living stranded sperm whale (Physeter catodon) S. anginosus, S. pneumoniae, and S. suis were detected in blood and fecal samples, but not in the muscles [168]. The animal died 79 h after rescue. S. intermedius was detected in blow samples of free-ranging and presumably healthy grey whales from Magdalena Bay and the Gulf of California by polymerase chain reaction [169]. Streptococcal Infections in Marine Mammals: Virulence and Mechanims of Pathogenicity Streptococci are a phylogenetically diverse group and, hence, their virulence and infection mechanisms differ intensely. A general idea of streptococcal infection in (marine) mammals is displayed in Figure 2. One important requirement for a successful infection is the resistance of pathogenic streptococci to host phagocytosis. The major antiphagocytic factors are the polysaccharide capsule, which is also the basis for serotyping [127,[170][171][172][173] and the streptococcal M protein of pyogenic streptococci [43,174]. The most critical phase in infection is the adhesion of streptococci to host cells. This is enabled by pili and/or adhesins such as fibronectin-and collagen-binding proteins and sortases [175,176]. It has also been demonstrated that streptococcal species are able to invade host cells [154,177,178] and produce toxins such as streptolysin S and streptolysin O [179,180]. There are numerous other virulence factors, for instance, streptokinases, secretory proteins that interacts with host plasminogen [181], or peptidoglycan deacetylases that protects the bacteria from host lysozymes [130,182]. To investigate the molecular mechanisms of streptococcal infection in marine mammals, primary cell cultures of marine mammals are required that can be infected and evaluated. Many streptococci occur as opportunistic pathogens or as secondary infection [45, [183][184][185][186][187]. This might also be the case for marine mammals. S. agalactiae, S. canis and S. marimammalium were found in the nasal cavity of two captured, healthy Hawaiian monk seals (M. schauinslandi) without any clinical signs [68]. A short-beaked common dolphin (Delphinus delphis) was coinfected by Streptococcus phocae and cetacean morbillivirus indicating S. phocae as secondary infection [45]. 16S rRNA sequencing of blow samples collected from four wild and healthy Indo-Pacific bottlenose dolphins (Tursiops aduncus) in Shark Bay, Western Australia identified S. phocae, S. equi and Streptococcus Group D as members of the blow microbiome without clinical signs [115]. The development of an infectious disease depends on the successful infection of a pathogen and on its virulence. This also includes the ability to evade host immune defense. Streptococci have evolved many mechanisms of immune invasion. For instance, the poreforming toxin streptolysin O induces caspase-dependent macrophage apoptosis [188] and the gene sets for streptolysin S, which is responsible for the beta-hemolysis, was also found in S. phocae subsp. phocae [156]. The human pathogenic S. pyogenes can recruit and colonize collagen type IV via surface-bound fibronectin, and the collagen fibers protect the bacterial cells from opsonizing antibodies [189]. Proteins that interact with collagen were also found in other streptococcal species including S. phocae [190]. Group A streptococci (e.g., S. pyogenes) can bind red blood cells by S protein for immune evasion [191]. In S. phocae subsp. phocae, an immunoglobulin G degrading enzyme, called ideP, was identified, which cleaves IgG of seals and thus, contributes to immune evasion [192]. Hence, streptococci that infect marine mammals requires specific adaptation to their immune system, as only immune evasion guarantee virulence and infection. The immune system of marine mammals is similar to other (terrestrial) mammals regarding general mechanisms [193]. In grey seals, the main type of immunoglobulin was IgG with two subclasses [194]. Furthermore, the authors discussed whether the susceptibility for bacterial infections including streptococcal infections in grey seal pups is related to the observed low values of IgG in pup serum in comparison to the relatively high values in the colostrum. Immunoglobulin classes homologous to human IgG, IgM, and IgA were identified also in dolphins and sea lions [195] indicating that streptococci face similar conditions and molecules when jumping from terrestrial to marine mammals. One difference in streptococcal infection between marine and terrestrial mammals could be the fact that plasminogen from marine mammals could be activated by human plasma including urokinase, but not by streptokinase vaccine [196,197]. Moreover, it is known that the genes for streptokinase of group A streptococci showed a high heterogeneity even in strains with the same serotype indicating immunological and chemical differences [198]. We did not find any information on streptokinases in streptococci of marine mammals and, hence, it can only be speculated if they just lost the gene or if there a genetic variations reflecting host adaptability. Adaptation of Streptococci to Marine Mammals Many streptococcal species found in marine mammals are also present in terrestrial mammals, although there are some physiological differences. That raises the question, to what extend and how these streptococci adapt to their specific marine host environments. Since this has not been studied yet in any detail here, we can only speculate and discuss it. The body temperature of marine mammals is comparable to that of terrestrial mammals [199][200][201][202][203], thus, it does not require special adaptations. However, gas physiology and rapid change in hydrostatic pressure during diving might be a challenge for the bacteria. The hydrostatic pressure of seawater increases about 0.1 atmosphere per meter of depth. It has been shown that hydrostatic pressure can influence the growth and viability of marine and terrestrial bacteria negatively [204,205]. At least the closely related Lactococcus lactis (previously S. lactis) was able to survive 500 atmosphere representing around 5000 m depth, while many other bacteria including species of Bacillus, Clostridium, and Staphylococcus died at pressure of 400 or less atmospheres [205]. In conclusion, streptococcal species and other bacteria might not have a problem with the hydrostatic pressure that increases during diving of marine mammals. However, bacteria of marine mammals, depending on which body part they inhabit (e.g., skin, intestines, respiratory tract) and if they were acquired from the marine environment, have to challenge the salinity of seawater. Seawater has an average salinity of 3.5%, but salt tolerance of streptococcal species is usually tested at 6.5% NaCl and 40% bile salts, respectively, [34,142,206,207]. The salt tolerance of S. iniae was tested at 2.0%, 4.0%, and 6.5% NaCl and growth was observed for the first two conditions [122]. In addition, few S. thermophilus strains and S. uberis grow at 4.0% NaCl, but not at 6.5% [208,209]. The fish pathogen S. parauberis is known to persist in seawater, probably by switching to dormancy as a survival strategy [210]. Therefore, it is also possible for other streptococcal species to persist and maybe even grow in seawater. Streptococci infecting marine mammals often use the respiratory tract as port of entry for colonization infection. Thus, it is likely that they have adapted to this special host niche. Notably, the respiratory tract differs between terrestrial and marine mammals, e.g., in anatomy, immune response, gas physiology (O 2 /CO 2 exchange), humidity, and chemical composition of the mucus. For instance, it has been shown that lung surfactant of pinnipeds has higher amounts of anti-adhesive components compared to terrestrial mammals, which probably supports alveolar opening after collapse during diving [211]. To our knowledge, there are no specific studies on genetic adaptation of streptococcal species to their marine mammalian host. However, whole genome analyses of human and fish/frog S. agalactiae strains revealed genetic adaptation to fish host by gene reduction and different gene expression such as of virulence associated genes [212]. A comparative genomic study of S. dysgalactiae species suggest that changes in gene content, selection of orthologous protein-coding loci and operon promoters involving mobile elements enables streptococci to adapt to changing environments and new hosts [213]. S. phocae subsp. phocae showed host specificity by the immunoglobulin G degrading enzyme ideP that solely cleaves IgG of grey and harbor seals, but not from harbor porpoises or non-marine mammals indicating functional adaptation [192]. S. halichoeri is assumed to have marine origin, although it has also been isolated from dog, bluefox, finnraccoon, and mink, as it has a great number of adhesins and salt tolerance proteins [120]. Lateral gene transfer between different streptococcal species is discussed as potential way of host adaptation [214]. For instance, lateral gene transfer was observed between S. canis and human S. urinalis and bovine S. agalctiae and S, dysgalactiae subsp. dysgalactiae mediated by variety of mobile genetic elements. This might also be true for marine mammals, where streptococci adapted to the marine mammalian host such as S. phocae might exchange genetic elements with acquired streptococcal species from fish or other sources. Further studies are required to understand the adaptation and interaction of streptococcal species and their marine mammal host as there is a huge scientific gap. Epidemiology and Possible Transmission Routes of Streptococci Species in Marine Mammals It is not known how marine mammals acquire or get infected with streptococci. Studies on the ecology and the environment could provide some insights about possible transmission routes and the epidemiology of streptococcal infections in marine mammals. In Figure 3, we give an overview about possible transmission routes and sources for streptococcal species. With many social species among marine mammals, transmission of pathogens are density-or frequency-dependent according to their social behavior [215,216]. The diet, e.g., fish could serve as reservoir for (pathogenic) streptococcal species and consequently as source of infection for marine mammals, which is also indicated by the study of Evans et al. [65], where the same S. agalactiae strain was found in both a dead dolphin and its diet, a mullet which was found in its stomach. S. iniae and S. agalactiae were also isolated from wild fish indicating fish as potential source for streptococci [217,218]. In addition, S. halichoeri is supposed to be transmitted to mink and Finnish dogs via fed fish [120]. S. phocae is also a well-known pathogen of Atlantic salmon [143]. However, the fish isolates differ from isolates of marine mammals, which was also the reason to suggest two subspecies of S. phocae [146]. The genetic differences could be a result of adaptation from fish to the mammalian host. Pathogenic streptococci and other bacteria could be introduced into seawater by shipping traffic including ballast water [219,220], human activities including recreational activity [221][222][223][224][225] and waste (water) [66,[226][227][228][229] and from the terrestrial environment via rivers and storm water [230,231], wind transport [232,233] or animals, e.g., semi-aquatic mammals, such as pinnipeds, from which streptococci are frequently isolated. For instance, human pathogenic S. agalactiae strains were identified in grey seals indicating that sea mammals were exposed to human pathogens via human effluents that contaminate coastal surface waters [66]. S. phocae has been found in the oral cavity of grey seals and in bite wounds of harbor porpoises probably caused by grey seals suggesting an interspecies transmission [157]. Seabirds can also shed pathogenic organisms into seawater [228,234,235] and streptococci were also detected in the gastrointestinal tract of seabirds [236]. If these transmission routes are real, streptococcal contamination and infections might increase with increasing human activities including growing cities in coastal regions and higher rates of shipping traffic, e.g., an increase of pollution and a higher likelihood of ship strikes or fisheries interactions. In addition, with higher temperatures due to climate change the persistence of pathogens in seawater is probably enhanced and, thus, there is a higher risk of infection [17,215]. The mortality of tilapias due to S. agalactiae infection was increased in higher water temperature [237]. This could also be the case for marine mammals, but there are no studies yet. In addition, habitat loss can enhance transmission by leading to higher population densities with higher contact rates. To our knowledge streptococci are no or not abundant members of the natural marine microbiome [238]. This raises the question, where they come from. Nevertheless, streptococci that are introduced by the routes mentioned above might cause dermal disease or enter marine mammals through open wounds and other traumata. Díaz-Delgado et al. [45] proposed that the S. phocae infection in a short-beaked common dolphin occurred through cutaneous penetration after a skin traumata, as the dolphin showed cutaneous disease with firm, raised, and occasionally ulcerated purulent subcutaneous nodules along the ventral, dorsal, and cranial edge of the caudal fluke, bilaterally, and more prominently at its insertion with the peduncle. However, studies on the distribution and presence of streptococci in healthy animals including the screening of the environment are necessary to investigate possible transmission routes or confirm these streptococcal species as commensals or members of the normal microbiome. Conclusions and Outlook Taken together, this is an overview about streptococcal species that were identified in marine mammals. Streptococcal species play an important role in the health of marine mammals all over the world. However, while beta-hemolytic streptococci are frequently isolated from marine mammals, only relatively few isolates were further identified (to the species level). There are further marine mammals from which (beta-hemolytic) streptococci have been isolated, but not identified to the species level, such as the blue whale (Balaenoptera musculus) [239], the gray whale (Eschrichtius robustus) [239], the sperm whale (Physeter macrocephalus) [240], the killer whale (Orcinus orca) [241], and the pacific walrus (Odobenus rosmarus divergens) [242]. This scientific gap makes it difficult to evaluate the diversity, distribution, and epidemiology of streptococcal species among marine mammals and needs to be filled. In addition, while there are lineages of streptococcal species that are quite host-specific, there are others that seem to have a more groad host spectrum and are easily transmissible between different hosts. For instance, in addition to marine mammals, S. canis can be found in dogs and cows [86], S. halichoeri in humans [119], and S. phocae in Atlantic salmon [143] and shrimps [145]. For some of these species, subspecies were defined based on their host-related differences such as S. phocae subsp. phocae and S. phocae subsp. salmonis [146] or S. halichoeri subsp. halichoeri and S. halichoeri subsp. hominis [119]. Furthermore, some of the streptococcal species found in marine mammals are major fish pathogens such as S. agalactiae [243], S. iniae [244], and S. phocae [143] and even zoonotic infections are possible, but the lack of data does not allow clear risk assessments. S. iniae caused infections in humans that handled live or freshly killed fish [58,245]. S. canis infections in humans are summarized by Galpérine et al. [56] and an endocarditis due to S. canis has been reported by Ansallem et al. [246]. S. equi subsp. zooepidemicus is also known to cause zoonotic disease in humans [53,247]. Hence, people working with marine mammals should also be aware of the zoonotic potential of streptococcal species. It is also not fully understood how the different streptococcal species are involved in diseases of marine mammals, although they are frequently isolated from sick or dead animals. With increasing chemical pollution and other anthropogenic activities in the marine ecosystem, the health of marine mammals is threatened and thus, they are more susceptible to infectious diseases. Hence, more research is needed on the epidemiology and pathogenic potential of streptococcal species in marine mammals. In conclusion, streptococcal species are isolated from many different marine mammal species world-wide. Further investigations on the role of the different streptococci species on the health status of marine mammals is urgently needed as streptococci are found with high prevalence in diseased marine mammals. This also underlines the need of additional information on the zoonotic potential of streptococci species found in marine mammals.	v3-fos-license
2016-05-12T22:15:10.714Z	2014-08-05T00:00:00.000	13319471	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://ojrd.biomedcentral.com/track/pdf/10.1186/s13023-014-0122-8", "pdf_hash": "2cf5039b3d2c1cb1f56fdb1ac86a10f5b411f39a", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45364", "s2fieldsofstudy": [ "Medicine" ], "sha1": "25b0686c2739bcb715d9d3334ad9fceead1e1ee5", "year": 2014 }	pes2o/s2orc	Systemic therapy of Cushing’s syndrome Cushing’s disease (CD) in a stricter sense derives from pathologic adrenocorticotropic hormone (ACTH) secretion usually triggered by micro- or macroadenoma of the pituitary gland. It is, thus, a form of secondary hypercortisolism. In contrast, Cushing’s syndrome (CS) describes the complexity of clinical consequences triggered by excessive cortisol blood levels over extended periods of time irrespective of their origin. CS is a rare disease according to the European orphan regulation affecting not more than 5/10,000 persons in Europe. CD most commonly affects adults aged 20–50 years with a marked female preponderance (1:5 ratio of male vs. female). Patient presentation and clinical symptoms substantially vary depending on duration and plasma levels of cortisol. In 80% of cases CS is ACTH-dependent and in 20% of cases it is ACTH-independent, respectively. Endogenous CS usually is a result of a pituitary tumor. Clinical manifestation of CS, apart from corticotropin-releasing hormone (CRH-), ACTH-, and cortisol-producing (malign and benign) tumors may also be by exogenous glucocorticoid intake. Diagnosis of hypercortisolism (irrespective of its origin) comprises the following: Complete blood count including serum electrolytes, blood sugar etc., urinary free cortisol (UFC) from 24 h-urine sampling and circadian profile of plasma cortisol, plasma ACTH, dehydroepiandrosterone, testosterone itself, and urine steroid profile, Low-Dose-Dexamethasone-Test, High-Dose-Dexamethasone-Test, after endocrine diagnostic tests: magnetic resonance imaging (MRI), ultra-sound, computer tomography (CT) and other localization diagnostics. First-line therapy is trans-sphenoidal surgery (TSS) of the pituitary adenoma (in case of ACTH-producing tumors). In patients not amenable for surgery radiotherapy remains an option. Pharmacological therapy applies when these two options are not amenable or refused. In cases when pharmacological therapy becomes necessary, Pasireotide should be used in first-line in CD. CS patients are at an overall 4-fold higher mortality rate than age- and gender-matched subjects in the general population. The following article describes the most prominent substances used for clinical management of CS and gives a systematic overview of safety profiles, pharmacokinetic (PK)-parameters, and regulatory framework. Introduction Both, Cushing's disease (CD) and Cushing's syndrome (CS), are rare diseases characterized by high cortisol blood levels and impairment of circadian oscillation [1]. CD derives from pathologic adrenocorticotropic hormone (ACTH) secretion usually triggered by micro-or macroadenoma of the pituitary gland [2]. It is, thus, a form of secondary hypercortisolism. In contrast, CS describes the complexity of clinical consequences triggered by excessive cortisol blood levels over extended periods of time irrespective of their origin [3]. Clinical manifestation of CS, apart from corticotropin-releasing hormone (CRH-), ACTH-, and cortisol-producing (malign and benign) tumors may also be by exogenous glucocorticoid intake ( Figure 1B). Patient presentation and clinical symptoms substantially vary depending on duration and plasma levels of cortisol. In 80% of cases CS is ACTH-dependent and in 20% of cases is ACTH-independent, respectively [4]. Clinical differentiation with respect to the onset of a certain treatment algorithm is by ACTH dependency and independency, respectively: ACTH-dependent ➢ Pituitary adenoma (CD in strict sense)~70% [5] ➢ Ectopic secretion of ACTH by non-pituitary tumors~15% (i.e. neuroendocrine tumors such as small-cell lung cancer (SCLC), carcinoid tumors, and medullary carcinoma of the thyroid) [6] ➢ Ectopic secretion of CRH by non-hypothalamic tumors causing pituitary hypersecretion of ACTH < 1% [7] ➢ Iatrogenic or factitious CS due to administration of exogenous ACTH < 1% ACTH-independent ➢ Adrenocortical adenomas and carcinomas~20% [8] ➢ Primary pigmented nodular adrenocortical disease < 1% [9] ➢ Bilateral ACTH-independent adrenal hyperplasia < 1% [10] However, systemic treatment options are limited and clinical evidence of these options is scarce (with Pasireotide as the exception from the rule). The regulatory status of pharmacological treatment options are reflected in Table 1. Chemical structures of medicines used in-label and offlabel are depicted in Figure 2. According to the European Community Register [11] (accessed on March 12 th , 2014) the medicinal products were granted an orphan designation in the context of CS. Etiology and symptoms CS is a rare disease according to the European orphan regulation [12] affecting not more than 5/10,000 persons in Europe. CS is a heterogeneous disorder that arises from multiple causes and has a broad spectrum of eventually fatal co-morbidities such as diabetes and hypertension. ERCUSYN is the synonym for The European Registry on Cushing's Syndrome. In a startling publication from 2011 Valassi et al. [13] describe the baseline demographic and clinical characteristics from: ➢ infections due to generalized immune suppression ➢ striae ➢ vascular fragility ➢ hypokalemia ➢ osteoporosis and eventually fractures ➢ muscle weakness Thus, adverse events are indistinguishable from longterm (sometimes unavoidable) glucocorticoid therapy. The metabolic consequences of cortisol excess include: ➢ weight gain ➢ central obesity ➢ skin atrophy ➢ glucose intolerance entailed by diabetes and insulin resistance ➢ dyslipidemia ➢ hypertension and ➢ clotting disorders (eventually even hypercoagulability [14]. Especially weight gain, diabetes, and hypertension are hazardous for patients due to two reasons: 1) They are (apart from infections) responsible for a 50% mortality within the first 5 years after diagnosis mainly due to cardiovascular events [15]. 2) These symptoms usually are diagnosed as metabolic syndrome, thereby delaying time to diagnosis [16]. According to current data time to diagnosis after the first symptoms is 6.0 years in mean [17]. In this regard, un-specificity of symptoms is highly problematic alike the slow-progressive nature of the disease [18]. However, months or even years of un-diagnosed CS may cause irreversible organ damage or be fatal due to late-stage diagnosis of malignant disease (i.e. ectopic ACTH producing SCLC). Recovery from the a.m. co-morbidities occurs, but may be delayed or incomplete. The duration of chronic hypercortisolism may determine reversibility of the co-morbidities associated with CD [19]. In addition, CS side-effects lead to an increased cardiovascular risk, thus, rendering even timely diagnosed CS a life-threatening disease, which is also due to infections: 71.4% of the deaths from CS can be attributed to cardiovascular causes or infection [20]. Standardized mortality ratios (SMR) in Cushing patients throughout the Ntali-study were statistically significantly elevated in the overall analysis (SMR 9.3; 95% CI, 6.2 -13.4, p < 0.001), as well as in all subgroups of patients under investigation. Patients with ectopic CS had the worst outcome with a probability of only 77.6% to survive the next 5 years. A systematic analysis of mortality studies in patients with CS as a consequence of an adrenal adenoma was undertaken by Graversen and colleagues in 2012 [21]. CD patients with persistent disease after initial surgery had a SMR of 3.73 (95% CI: 2.31 -6.01), whereas mortality of CD patients with initial remission did not differ significantly from the general population (SMR: 1.23 (95% CI: 0.51 -2.97)). This study confirms the excess mortality in patients with CD if remission after initial surgery is not achieved. Epidemiology of Cushing's syndrome as a rare disease Approximately 1% of the population use exogenous steroids, of which 70% experience one or more adverse events [22]. In contrast, a national register in Denmark reported an annual incidence for endogenous CS of two cases per million people [23]. Endogenous CS is usually (in 70% of cases) a result of a pituitary tumor. In general, CS is rare: the reported incidence of endogenous CS worldwide ranges from 0.7 -2.4 cases per million per year [24]. CD most commonly affects adults aged 20 -50 years with a marked female preponderance (1:5 ratio of male vs. female). CS patients are at an overall 4-fold higher mortality rate than age-and gender-matched subjects in the general population [25]. When diagnosed the prevalence of comorbidities has been reported as follows: 58 -85% of patients have hypertension, 32 -41% obesity, 20 -47% diabetes mellitus, 50 -81% major depression, 31 -50% osteoporosis, and 38 -71% dyslipidemia [19]. Lines of treatment Depending on the cause of these excessive cortisol concentrations in the patient's blood, different treatment modalities are possible in routine patient care. Therapy of choice is trans-sphenoidal surgery (TSS) of the pituitary adenoma (in case of ACTH-producing tumors). In patients not amenable for surgery radiotherapy remains an option. Thus, primary lines of therapy in CD (i.e. not medication associated CS) consists of TSS and pituitary irradiation. TSS is currently the most frequently recommended treatment except for patients who are poor surgical candidates, have invasive tumors, or who refuse surgery. The surgical effectiveness varies depending on expertise in pituitary surgery and the size and extension of the anatomic mass. Hypopituitarism is common after TSS with a range between 13 and 81% [3]. Early remission in terms of cortisol blood level normalization is achieved regularly ( Figure 1A). Patients who do not achieve normalization with surgery require additional treatment, usually with radiotherapy and/or medication. If damage to the surrounding normal pituitary tissue occurs during surgery, the patient may require lifelong pituitary hormone replacement. However, an examination of remission and recurrence rates in long-term follow-up studies reveals that potentially up to 40% to 50% of patients could require additional treatment [26]. Thus, systemic (medicinal) treatment of CS may be more important than traditionally thought ( Figure 1B). Pituitary irradiation is usually reserved for patients who have tumor remaining after surgery, for patients who are poor candidates for surgery, and for patients who do not respond adequately to surgery and/or medication. The main disadvantages of radiotherapy are that i) normalization of ACTH secretion may take extended periods of time (eventually years) to occur demanding for medication while success of radiation is awaited, and ii) that patients may develop generalized anterior pituitary insufficiency [27]. Pharmacological therapy applies when these two options are not amenable or refused. In cases when pharmacological therapy becomes necessary, Pasireotide should be used in first-line in ACTH-dependent CS. However, systemic pharmacotherapy of CS often showssometimes dose-limitingside effects. A systematic overview of safety profiles is given in Table 2. The source of data in Table 2 is a systematic recherché in section 4.8 of currently (March 2014) approved SmPCs of licensed medicines. In the following sections drugs used in CS treatment ( Figure 2) are referred to in alphabetical order (as in all Tables and Figures). Molecular interaction of the respective drug with cortisol-biosynthesis is depicted in Figure 3, whereas PK parameters of the respective substances are listed in Table 3. Aminoglutethimide Aminoglutethimide from its molecular mechanism is a non-selective, non-steroidal aromatase inhibitor. Due to its non-selective mechanism of action it also inhibits endogenous cortisol synthesis. Schteingart et al. back in 1966 showed that Aminoglutethimide can alleviate clinical symptoms of CS in a patient suffering from metastatic adrenal cancer [37]. However, the same consortium of authors found one year later that its effect can be overruled by excessive secretion of ACTH [38]. Following a report of three patients receiving Aminoglutethimide in combination with Metyrapone, these substances were also given concomitantly [39]. However, Aminoglutethimide is marketed no more. Cabergoline Two types of receptors are widely expressed at the surface of pituitary adenoma cells: the somatostatin receptor subtype 5 (SSTR-5) and the dopamine receptor subtype 2 (D 2 ). As mentioned above, SSTR-5 is occupied by Pasireotide prevalently compared to other somatostatins. Dopamine agonists (e.g. bromocriptine, quinagolide, cabergoline) bind to D 2 receptors in the pituitary gland. In CD corticotroph adenomas mainly express D 2 receptors (and SSTR-5). Agonists at these receptors inhibit ACTH-release in cell culture of corticotroph adenomas. In those in vitro model systems compounds that target SSTR-5 (like Pasireotide) or D 2 (cabergoline) have shown efficacy in subsets of patients in the clinical setting. Combination therapy by administration of both types of compounds yielded promising results. These clinical reports of enhanced efficacy of combination therapy with SSTR-and dopamine agonist treatment (originally initiated for pituitary adenoma therapy in suppressing growth hormone hypersecretion) lead to the novel concept of somatostatin-dopamine chimeric molecules, e.g. BIM-23A760 [40]. Preliminary in vitro results suggest that the affinity to D 2 receptors is crucial for inhibition of prolactin gene expression and growth hormone secretion. However, the impact of these substances on CS in patients remains to be elucidated. In addition, a recent study showed that D 2 receptors are also expressed in the majority of ectopic ACTH-producing syndrome (EAS) cases and that cabergoline may decrease cortisol levels in certain subsets of these patients [41]. Etomidate Etomidate (just like Ketoconazole) from the chemical perspective is an imidazole derivative. It was developed in the sixties as parenteral hypnotic drug [42]. Mechanistically, Etomidate is a centrally acting γ-aminobutyric acid type A (GABA-A) receptor agonist [43]. Low serum cortisol levels resulting in hypoadrenalism was a randomly discovered side effect of Etomidate [44]. Etomidate (like Metyrapone) inhibits the mitochondrial cytochrome P450 (CYP)-dependent adrenal enzyme 11-β-hydroxylase that catalyzes the production of cortisol from deoxycortisol, thereby lowering cortisol blood levels within hours [45]. In a recent systematic review, Preda et al. came to the conclusion that Etomidate might be of clinical benefit in last-line therapy under intensive care condition. In the outpatient endocrinology setting for patients with severe hypercortisolism there might be a role for Etomidate [30]. Etomidate usually is thought to be efficacious only in hypnotic doses, but this appears not to be the case: in the Preda publication, non-hypnotic doses of only 0.3 mg/kg were active in lowering cortisol levels. Glitazones Glitazones (Thiazolidinedions) like Rosiglitazone or Pioglitazone are anti-diabetic insulin sensitizing drugs mediating their effects via activation of the transcription factor peroxisome proliferator-activated receptor γ (PPAR γ). Activation of PPARγ results in effects on adipogenesis, carbohydrate and lipid metabolism, inflammation processes, and cell proliferation [46]. Additionally, PPARγ agonists have been shown to inhibit the growth of several tumor cells derived from lung, breast, prostate, and colon cancer tissues [47][48][49][50]. Immunocytochemical PPARγ expression could be shown in autopsy-derived human pituitary tissue. PPARγ expression was primarily restricted and co-localized with ACTH [51]. However, in pituitary tumors, including ACTH secreting tumors, PPARγ expression is increased as compared to the normal pituitary [52]. In rat and human corticotroph adenoma cell lines Rosiglitazone decreased tumor cell growth, increased apoptosis, and lowered ACTH secretion by inhibiting mRNA expression of its precursor-protein pro-opiomelanocortin (POMC). PPARγ has furthermore been shown to act as tumor suppressor gene in animals [53,54]. It was hypothesized that glitazones might be useful in treating corticotroph pituitary adenomas by inhibiting ACTH synthesis and secretion and tumor growth. Subsequently, clinical studies investigating the role of glitazones in the treatment of CD were conducted: In one study, 14 patients with active CD (7 untreated and 7 after unsuccessful surgery) were treated with 8 -16 mg of Rosiglitazone for 1 -7 months [55]. In six patients, 24hour urinary free cortisol (UFC) was significantly lowered and two of them showed clinical improvement at 7-month follow-up. However, after 10 month cortisol levels increased and clinical signs relapsed [56]. Pre-operative Rosiglitazone treatment (8 mg/day) of two patients with pituitarydependent CS lowered cortisol levels in the 24-hour urine and lead to clinical improvement [57]. In a study of 10 patients, four prior to surgery, four following relapse after surgery, and two immediately after failed surgery treated with 4 -16 mg of Rosiglitazone for 1 -8 months, no consistent reductions in UFC levels were found. Only 3 of 10 patients had normalized urinary cortisol levels up to 8 months [58]. Side effects reported included edema, weight gain, somnolence, and increased hirsutism. Although most studies used Rosiglitazone, one study with Pioglitazone is available [59]. In none of the five patients with CD treated with 45 mg Pioglitazone for 30 days any UFC responses were observed. Taken together, glitazone treatment failed to reproduce the effects seen in vitro and in animal studies. Only Rosiglitazone could be shown to be effective at least in a small subset of patients. However, Rosiglitazone is currently suspended in the European Union due to an increased cardiovascular risk. [32] Abbreviations: AUC, area under the curve; b.i.d., twice daily; C max , maximal plasma concentration; q.d., every day; t max , time after administration when C max is reached; 1 500 mg; 2 750 mg administration; 3 n = 12, dose = 0.5 mg; 4 in-vitro study with concentrations of 0.1 -10 ng/ml; 5 pediatric population; 6 measured in dogs, 50 mg/kg, tablets in food; 7 200 mg/m 2 ; 8 also available as long-acting release formulation in the strengths 20, 40, 60 mg; 9 AUC 0-inf ; 10 AUC 0-24 h : 600 μg q.d. (n = 11); 11 30 mg single dose in 24 healthy volunteers; 12 AUC 0-24 h ; source of information: SmPCs of the corresponding medicines unless otherwise indicated (if not licensed for treatment of CS, then in other indication; i.e. Ketoconazole, Etomidate, and Temozolomide). If not marketed any more (Aminoglutethimide) the last approved SmPC was used. Ketoconazole Ketoconazole originally was developed as an anti-fungal medicine. In fact, it was one of the first orally bioavailable systemic anti-mycotic drugs but is also used locally. Chemically it belongs to the first-generation imidazole antifungal drugs (in contrast to triazols). Due to its hepatotoxic properties, oral formulations of Ketoconazole are no longer marketed as anti-fungal medicines as recommended by the European Medicines Agency (EMA) after the Committee for Medicinal Products for Human Use (CHMP) concluded that the risk of liver injury is greater than the benefits in treating fungal infections [60]. In fungi it mechanistically acts as an ergosterole synthesis inhibitor by blocking various CYP-dependent enzymes, thereby explaining both, its hepatotoxic side-effects and its efficacy in Cushing's treatment: Steroidogenesis in the zona fasciculata of the adrenal cortex mainly depends on CYP enzymes [61]. At lower doses, C17-20-lyase and 17-α-hydroxylase are inhibited in both the testicular and adrenal glands. In contrast, cholesterol side-chain cleavage enzymes and 11-β-hydroxylase are inhibited only at higher doses. However, a negative benefit-risk-ratio for one indication (anti-fungal) where alternatives exist does not imperatively mean this also accounts for other indications (Cushing). EMAs safety concern was due to severe hepatic injury but the incidence was only 1 in 15000 cases [62], thus, benefit risk might be different in a disease with high mortality and less alternatives compared to the anti-fungal indication. Currently, Ketoconazole is used off-label in CS, however, marketing authorization applications (MAAs) are underway to change this unsatisfactory status. Thus, future will tell whether Ketoconazole has a positive benefit-risk ratio in CS and currently no final judgment can be made. Ketoconazole may have clinical benefit in CS patients with hyperandrogenic symptoms as it also inhibits steroidogenesis of androgens (for instance in the testes). The occurrence of hirsutism with Metyrapone treatment in women may lead a clinician to choose Ketoconazole in this case or may lead to a combination therapy of Metyrapone and Ketoconazole. Some patients even may require combination of both to achieve control of CS. The dose of Ketoconazole used as monotherapy in patients with CD ranges from 200 mg to 1200 mg per day [63]. CT) and other localization diagnostics This diagnosis flow-scheme also is reflected in current international text books [64] and for instance in the clinical practice guidelines of the endocrine society [65]: "After excluding exogenous glucocorticoid use, we recommend testing for Cushing's syndrome in patients with multiple and progressive features compatible with the syndrome, particularly those with a high discriminatory value, and patients with adrenal incidentaloma. We recommend initial use of one test with high diagnostic accuracy (urine cortisol, late night salivary cortisol, 1 mg overnight or 2 mg 48-h dexamethasone suppression test). We recommend that patients with an abnormal result see an endocrinologist and undergo a second test, either one of the above or, in some cases, a serum midnight cortisol or dexamethasone-CRH test." Metyrapone, thus, is an excellent example for national traditions determining clinical management of rare dis-eases. In CS treatment usually is a case-by-case decision because of the heterogeneity of the patients' population especially in regard to their co-morbidities. Consequently, no (national or international) medicinal guidelines or treatment algorithms of high evidence for management of CS exist. A consensus statement on ACTH-dependent CS published in 2008 by Biller et al. emphasized clinical management of CS as a multi-disciplinary approach on an individualized basis [66]. Despite its efficacy in lowering cortisol blood levels, Metyrapone is not constantly effective and therefore does not cover the entire CS-population. Clinical and/or biochemical improvements can be achieved in 60 to 80% of patients and are usually associated with improvements in symptoms. In general, Metyrapone appeared to be well tolerated, but may sometimes lead to an accumulation of androgens in women during long-term treatment (hirsutism). Mifepristone Mifepristone (experimental name RU-486) originally was developed as the so-called "abortion pill". Chemically it is an all-trans steroid substituted at positions 11 and 17. Mechanistically, it is an antagonist at the progesterone receptor (PR), thereby explaining its acute abortive effect. It also shows antagonistic activity at the glucocorticoid receptor (GR) with a higher affinity than Dexamethasone. Thus, unlike other agents, Mifepristone does not decrease cortisol synthesis but directly antagonizes its effects. Therefore, it is intuitive that patients dosed with Mifepristone can hardly be diagnosed by serum cortisol measurements as they remain unchanged. Mifepristone is not approved in the indication CS. Side effects include decreased plasma potassium levels with a potential for heart conduction abnormalities, vaginal bleeding, and endometrial thickening (not surprising for a progesterone antagonist). Notions on Mifepristone Apart from the fact that everyone knows it as an abortion pill, it must not be forgotten that Mifepristone originally was developed as an anti-glucocorticoid. Soon afterwards, it was found to block PRs intrinsic activity. Due to its pronounced cortisol-blocking effect, new generation antiprogestins were developed aiming at reducing antiglucocorticoid activity while maintaining anti-progesterone activity (i.e. ORG-31710, CDB-2914, CDB-4124). However, only Mifepristone has been licensed by US Food and Drug Administration (FDA) to terminate early pregnancy (working as an antiprogestin) or ameliorating the hyperglycemia in CS-patients [67]. In contrast, other PR antagonists like ulipristal and proellex are currently investigated for their potential to alleviate symptoms in endometriosis and uterine fibroids, thereby approaching indications like other hormone ablative medicines (i.e. GnRH analogues) [68]. Interestingly, Tieszen et al. investigated the potential of PR antagonists, to inhibit the growth of cells from endocrine related cancers (i.e. ovarian-, breast-, and prostate cancer cells), expressing different sets of hormone receptors. The authors found, that all cancer cells were inhibited in growth irrespective of their PR status, as there are: MCF-7 breast cancer cells carrying PR, MDA-MB-231 breast cancer cells with no PR expression, PR negative and androgen receptor positive LNCaP prostate cancer cells, and PR negative androgen receptor positive PC3 prostate cancer cells are all inhibited by mifepristone with similar potency [69]. However, all cell lines under investigation in this study express GR and derivatives of mifepristone with lower GR affinity are less effective. Therefore, it can be concluded that GR action might contribute to anti-tumor effects of PR antagonists. Thus, PRs may not be required for the inhibition of cancer growth triggered by Mifepristone. In addition, it has been shown by another group, that Mifepristone blocked the growth of estrogen receptor negative and PR negative MDA-MB-231 breast cancer cells [70]. Mitotane Mitotane is another unusual drug used in certain very rare cases of CS. Mitotane (Lysodren®) is authorized only for symptomatic treatment of advanced (very rare) adrenal cortical carcinoma (ACC). Chemically it is an orthoderivative of the long-term known insecticide DDT (dichloro-diphenyl-trichloroethane), the bis-trans-form of the molecule. Its biochemical mechanism is not yet fully understood, but inhibition of side-chain cleavage of cholesterol (a decisive step in endogenous steroidogenesis) seems to play a role as well as blockade of 3β-hydroxysteroid-deshydrogenase [71]. Like its mother substance DDT Mitotane also accumulates in adipose tissue, which might then be entailed with longer half-life. Especially in Cushing's patients this is of particular concern, as these patients usually present with higher fat mass compared to age-adjusted matched population. Therefore, it usually takes weeks until complete efficacy is reached. In the current SmPC it is even mentioned that 3 -5 months can be assumed until the intended plasma level of 14 -20 mg/L is reached. This window should be reached (titrated) by monitoring patients' blood levels. Thus, Mitotane fulfills the requirements of a Narrow Therapeutic Index Drug (NTID) according to the FDA definition. Mitotane has an orphan designation and, thus, the pivotal registration trial was performed in 177 patients only showing an increase in recurrence-free interval after radical surgery followed by Mitotane compared to surgery alone [72]. Grade 3 side effects were mainly neurologic (confusion, ataxia, vertigo) or biochemical (elevated γglutamyltransferase). Thus, Mitotane might be of clinical benefit in some patients, adjuvant to radical surgery. Pasireotide and somatostatin analogues Pasireotide mechanistically is a somatostatin-analogue (SSA). SSAs are the first-line medical treatment in GHsecreting adenomas (with the clinical manifestation of acromegaly). Currently two different molecular entities, octreotide (Sandostatin®) and lanreotide (Somatuline®), are marketed. Octreotide, which was licensed 25 years ago, is available as a short-acting subcutaneous formulation for twice-daily administration, and a long-acting (LAR) microsphere preparation administered by intra muscular injection every 4 weeks. Lanreotide is available in a microsphere formulation (sustained release) and a high-concentrate aqueous solution (Autogel). From clinical safety point of view SSA side effects include gallstones and gastrointestinal affection (diarrhea, nausea, abdominal pain). According to the route of administration injection site reactions (pain, swelling) have also been reported. Pasireotide (experimental name SOM230) itself is a cyclic hexapeptide and was licensed via the central route in Europe in 2012 as first-line pharmacological therapy option. The European Public Assessment Report (EPAR) is dated June 1 st 2012 [73]. Endogenous somatostatin is a peptide hormone widely distributed in the endocrine system. Somatostatin action is mediated through five different SSTR subtypes (SSTR-1-SSTR-5). SSTRs belong to the superfamily of tripartite G-protein coupled receptors. They are expressed over the whole body in various tissues, with cells from different tissues expressing different receptor subtypes at different densities. The physiological actions of somatostatin are numerous. It is an inhibitory protein of endocrine secretion of various organs, including the pituitary, pancreas, gastrointestinal tract, thyroid, kidney, and adrenal glands. Among others, it inhibits gallbladder contractility and bile flow, and stimulates gastrointestinal water and electrolyte absorption. Pasireotide has a slightly different SSTR binding profile than the firstgeneration SSAs Octreotide and Lanreotide, with high affinity to four of the five receptors (SSTR-1, −2, −3 and −5). Compared to Octreotide the binding affinity of Pasireotide is 30 -40 times greater for SSTR-1 and SSTR-5 and 5 times greater for SSTR-3, whereas the affinity for SSTR-2 is comparable. In general, medicines for the treatment of CS (i.e. Mifepristone, Ketoconazole) are not selective and initially were not developed for the treatment of CS. In contrast, Pasireotide exhibits at least a subtype-prevalence for subtype-5 of the SSTR in the pituitary gland. SSTR-5 prevalently is expressed at the surface of ACTH secreting cells, thus showing at least partial specificity in lowering excessive cortisol blood levels. Therefore, first-line pharmacological therapy is Pasireotide injection (usually while definitive therapy is awaited). In addition, the basis of the centralized approval for Pasireotide was a clinical development program consisting (only regarding patients) of 1. a 22-patient phase II study reviewed in [74] 2. a randomized-dose, double-blinded pivotal phase III study [75]. The study design was as follows: However, it should be noticed that Pasireotide is indicated only in a subgroup of CD-patients and shows satisfactory efficacy only in~30% of patients. Side effects with this agent are similar to those of other SSAs. Hyperglycemia-associated side-effects are of particular concern in CS patients and were reported in 73% of patients in the pivotal study. Temozolomide Temozolomide (Temodal®) is licensed for Grade III and IV glioblastoma multiforme concomitant to radiotherapy [76]. Temozolomide is an orally bioavailable, centrally active alkylating agent. Following a systematic review from 2012 up to now, 46 cases of adenohypophysial tumors are published in Medline that were treated with Temozolomide (30 adenomas, 16 carcinomas). 60% of the adenomas and 69% of carcinomas responded favorably to treatment [77]. Thus, in (rare) cases of highly aggressive growing macroademomas or pituitary carcinomas Temozolomide might be an option. Trilostane (only in veterinary use) Chemically Trilostane is a 2-cyano-4,5-epoxy-steroid. Trilostane reversibly suppresses adrenal function. From its molecular mechanism of action it inhibits cortisol biosynthesis by prevalently blocking 3-β-hydroxysteroidedehydrogenase. This, however, does not only block cortisol-and corticosteron-synthesis, but also biosynthesis of aldosterone. Thus, Trilostane un-specifically blocks both, glucocorticoids and mineralocorticoids. High doses are reported to even block gonadal steroidogenesis. However, Trilostane is not used as human medicine; its only indication is CS in dogs, irrespective whether CS derives from pituitary or adrenal cortisol-overproduction. For many years before Mitotane has been considered the medical treatment of choice for dogs with adrenaldependent hyperadrenocorticism [78]. Therapy of Cushing's syndrome in children First of all, in the pediatric population no final conclusion can be drawn on any of the medicines under investigation due to lack of pivotal evidence. This is best reflected by the fact that the EMA accepted a waiver for clinical data from pediatric patients for centrally approved Pasireotide (a so-called "PIP-Waiver"; PIP, pediatric investigation plan). Throughout the SmPCs it is mentioned that only limited evidence is available for children and, thus, the respective medicine is not recommended for use in children. However, in those drugs used off-label even this is based on other indications. There is only one exemption from the rule; for Metyrapone the following can be read in the SmPC: "In children the dosage should be 15 mg/kg bodyweight, with a minimum dose of 250 mg every 4 hours for 6 doses." However, even for Metyrapone evidence in children is very low (15 documented cases) [2,79,80]. Co-medication for symptomatic therapy Clinical management of co-morbidities of CS is a complex challenge. Post-surgery management of CS patients is tri-phasic. Ragnarsson and Johannsson summarized systemic therapy while awaiting success of surgery (or radiotherapy) [3]. Usually supportive therapy of CS co-morbidities is by individual therapy depending on clinical presentation of the patient. Certain hazardous (life-threatening) symptoms are treated on a case-by-case decision according to medicinal guidelines of the respective symptoms. Conclusion CS can arise from different pathological conditions ranging from benign pituitary adenomas to adrenal carcinomas or ectopic cortisol secretion. Thus, various drugs with different target structures and mechanisms of action are used in systemic therapy of CS comprising somatostatin-analogues, cortisol synthesis inhibitors, receptor antagonists and unusual substances like Temozolomide and Mitotane for certain very rare conditions. Some mechanisms of action are well known (Pasireotide, Ketoconazole) others are not fully understood (Mitotane).These drugs differ in terms of safety profile, route of administration, and indications considerably. In addition, safety profiles differ considerably and become hardly predictable, if combinations are needed due to insufficient clinical control of CS by only one substance. Many of these substances are used offlabel and constitute an ultima ratio approach after failure of preceding therapy options. Thus, clinical and/or at least epidemiological data will be needed to finally judge on safety and efficacy in regard of a debilitating underlying disease. Search strategy Recherché of clinical PK-parameter and special populations i) SmPCs as published on the heads of medicines agencies (HMA) homepage (http://mri.medagencies.org/Human/) were accessed in March and June 2014. Section 5.2 (Pharmacokinetics) of SmPCs systematically was searched for PK-parameters. ii) A search in PubMed/Medline was performed in March and June 2014 using the international non-proprietary name (INN) of the respective medicinal product combined with the terms "cushing" and "indication" as well as "cushing" and "pediatric". PubMed recherché was restricted by searching only in clinical trials. Safety profile assessment SmPCs of currently marketed medicinal products were accessed (if not licensed for treatment of CS, then in other indication; i.e. Ketoconazole, Etomidate, Temozolomide, and Pioglitazone). If not marketed any more (Aminoglutethimide) the last approved SmPC was used except for Trilostane, were veterinary SmPC of Swissmedic was used. Side-effects were categorized by frequency for preferred terms (PT) of the systems organ class (SOC) system (XX = common, very common; X = rare, uncommon; not known).	v3-fos-license
2018-04-03T00:40:11.222Z	2017-08-14T00:00:00.000	4478152	{ "extfieldsofstudy": [ "Biology", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/s41598-017-08680-w.pdf", "pdf_hash": "47bc77a931e31d5b30156cc473e0f8ac51438791", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45365", "s2fieldsofstudy": [ "Biology", "Medicine" ], "sha1": "47bc77a931e31d5b30156cc473e0f8ac51438791", "year": 2017 }	pes2o/s2orc	Krüppel-like factor 6 is a transcriptional activator of autophagy in acute liver injury Krüppel-like factor 6 (KLF6) is a transcription factor and tumor suppressor. We previously identified KLF6 as mediator of hepatocyte glucose and lipid homeostasis. The loss or reduction of KLF6 is linked to the progression of hepatocellular carcinoma, but its contribution to liver regeneration and repair in acute liver injury are lacking so far. Here we explore the role of KLF6 in acute liver injury models in mice, and in patients with acute liver failure (ALF). KLF6 was induced in hepatocytes in ALF, and in both acetaminophen (APAP)- and carbon tetrachloride (CCl4)-treated mice. In mice with hepatocyte-specific Klf6 knockout (DeltaKlf6), cell proliferation following partial hepatectomy (PHx) was increased compared to controls. Interestingly, key autophagic markers and mediators LC3-II, Atg7 and Beclin1 were reduced in DeltaKlf6 mice livers. Using luciferase assay and ChIP, KLF6 was established as a direct transcriptional activator of ATG7 and BECLIN1, but was dependent on the presence of p53. Here we show, that KLF6 expression is induced in ALF and in the regenerating liver, where it activates autophagy by transcriptional induction of ATG7 and BECLIN1 in a p53-dependent manner. These findings couple the activity of an important growth inhibitor in liver to the induction of autophagy in hepatocytes. contributes to hepatic insulin resistance and the progression of non-alcoholic fatty liver disease (NAFLD) to non-alcoholic steatohepatitis (NASH) and NASH-cirrhosis 4 . KLF6 also affects peroxisome proliferator activated receptor gamma (PPARγ)-signaling in NAFLD 3,5 . Besides their metabolic functions, PPARα and PPARγ regulate cell proliferation and apoptosis 6 . Furthermore, KLF6 has been identified as a tumor suppressor gene that is inactivated or downregulated in different cancers including prostate, colon and hepatocellular carcinomas 7,8 . Consistent with its inhibitory effect on cell proliferation, KLF6 transactivates genes controlling cell proliferation, including p21, E-Cadherin and pituary tumor-transforming gene 1 (PTTG1) [8][9][10][11][12][13][14] . Despite its clear growth regulatory activity in hepatic metabolism and cancer, there are no studies evaluating the role of KLF6 in liver regeneration and hepatocyte proliferation. Acute liver injury and acute liver failure (ALF) are rare but serious conditions leading to hepatocyte death that occur in a previously healthy organ. ALF is characterized by rapid induction of hepatocyte necro-apoptosis, leading to jaundice, hepatic encephalopathy and coagulopathy 15 . The underlying causes of ALF encompass autoimmune, viral, toxic or vascular diseases, with drug-induced liver injury and acetaminophen (APAP) poisoning as the most predominant etiologies in Western population 16,17 . Acetaminophen is a widely used analgesic and antipyretic drug. Intake of high doses can result in ALF that is characterized by a rapid loss of liver cells and hepatic function due to enhanced production of reactive oxygen species (ROS), causing cellular stress and induction of cell death [17][18][19] . Specific treatment (N-acetyl cysteine (NAC)) promotes liver regeneration by compensation of hepatic cell loss and induction of proliferation of remaining cells and by the activation and potential differentiation of quiescent progenitor cells 20,21 . Liver regeneration is governed by a delicate interplay of cytokines, chemokines and the activation of proliferative and anti-apoptotic signaling pathways. Recent studies have identified autophagy, a conserved mechanism to recycle cellular components in cell starvation, to play a role in hepatocellular regeneration in APAP-induced ALF by reduction of cellular stress [22][23][24] . In this study, we aimed to investigate the role of KLF6 in liver regeneration following acute hepatocellular injury and ALF, and identified autophagy-related genes to be transcriptionally regulated by KLF6. Results KLF6 is induced in hepatocytes during acute human liver injury. We compared KLF6-expression by immunohistochemistry between liver tissue from patients with ALF and without (morbidly obese patients who underwent bariatric surgery without NASH (NAS < 2) or fibrosis; for patients' demographical data see Supplementary Table S1). KLF6-expression was low in non-acute injury livers and localized primarily in the cytoplasm of cholangiocytes, with modest staining in the cytosol or nuclei of hepatocytes (Fig. 1A). In contrast, significantly higher nuclear KLF6-expression was detected in hepatocytes in liver tissue of ALF patients, while the bile duct regions showed low levels of KLF6 ( Fig. 1B; for H&E images of patients' liver tissue, please see Supplementary Figure S1, for quantification of nuclear KLF6 in hepatocytes see Supplementary Table S1). KLF6 attenuates liver regeneration and autophagy after partial hepatectomy in mice. We performed 70% partial hepatectomy (PHx) in C57Bl/6-mice as an established model of liver regeneration 25 . Animals were sacrificed 12 h, 24 h and 48 h after PHx and the remnant liver was analyzed. Expression of Klf6 was significantly upregulated in liver tissue following PHx in wildtype mice ( Fig. 2A) and, as observed in human ALF, was mostly detected in the nuclei of hepatocytes (Supplementary Figure S2A). Next, PHx was performed in mice with a hepatocyte specific Klf6-knockout (DeltaKlf6) compared to controls 2, 3 . Enhanced hepatocyte proliferation was observed in the absence of Klf6, as assessed by PCNA-staining ( Fig. 2B for quantification of KLF6 affects liver regeneration and expression levels of autophagy-related genes after partial hepatectomy (PHx). Klf6 expression levels were determined by qRT-PCR in mouse liver tissue before (pre-OP) and 12 h, 24 h or 48 h after 70% partial hepatectomy (PHx, n = 6/group) (A). Cell proliferation was assessed by quantification of PCNA positive cells in liver tissue of wildtype (wt) and DeltaKlf6 mice 72 h after PHx (B). Expression-levels of Atg7 (C) and Beclin1 (D) were measured by qRT-PCR in liver tissue of wt and DeltaKlf6 mice before and 12 h after PHx. Autophagy was assessed by LC3 Western blotting and quantified by densitometry of specific protein bands (E,F) in liver tissue of wt and DeltaKlf6 mice 72 h after PHx. Shown are representative Western blot images (E) and densitometric quantification of LC3-II-bands normalized to loading control beta-Actin (F); fold change versus control shown as mean ± SEM of n = 6 mice per group; full length Western blot images are given in Supplementary information). Represents p-value < 0.05 and *Indicates p-value < 0.01 as assessed by 2-way ANOVA comparing wt mice with DeltaKlf6 animals at the same time point after PHx. Figure S2B for immunohistochemical PCNA images), and by quantifying liver-to-body-weight-ratios (Table 1). In KLF6-over-expressing HepG2 cells, following transient transfection with a KLF6-expression vector (pcIneo-KLF6), proliferation appeared attenuated compared to empty control vector (pcIneo) transfected control cells as assessed by BrdU cell proliferation assay (Supplementary Figure S2C). PCNA positive cells and Supplementary Liver tissue from DeltaKlf6 mice 12 h after PHx was subjected to RNA-microarray analysis (Affymetrix GeneChip HT MG-430 PM), which revealed changes in expression of autophagy-related genes compared to wt controls (heatmap included in Supplementary Figure S2D). QRT-PCR confirmed significant reductions in Atg7 and Beclin1 expression in DeltaKlf6 livers before and 12 h after PHx compared to wt controls (Fig. 2C,D). Attenuation of LC3-II-expression after PHx assessed by Western blot was correlated with loss of Klf6 in DeltaKlf6 mice (Fig. 2E,F). Western blot revealed high levels of LC3-II in livers of wt mice 72 h post PHx, which was significantly attenuated in the absence of Klf6 in liver tissue of DeltaKlf6 animals. KLF6 induction parallels induction of autophagy in vivo and in cell culture. To investigate Klf6-expression in an established in vivo model of APAP-induced liver injury 26 we employed C57Bl/6-mice that received an intra-peritoneal injection of APAP (500 mg/kg bodyweight) or saline in controls (H&E images of liver tissue from APAP-or vehicle-treated mice are shown in Supplementary Figure S3A). The animals were sacrificed 8 h after injection and levels of Klf6 gene expression in liver tissue were assessed by qRT-PCR. APAP-injection resulted in significant liver damage as indicated by increased serum ALT-, AST-and GLDH-levels 8 h after treatment ( Table 2). Hepatic Klf6-expression was significantly increased 8 h after APAP-injection (Fig. 3A). Comparing vehicle treated C57Bl/6 mice with those receiving APAP injection; LC3-II-levels were significantly enhanced in murine liver tissue after APAP-injection (Fig. 3B, Supplementary Figure S3B for Western blot image). To evaluate Klf6 expression in another model of acute liver injury, we injected a single dose of CCl 4 to DeltaKlf6 and wildtype animals.In livers of animals sacrificed 48 h after receiving an acute dose of CCl 4 , in parallel to acute liver damage (see Table 2 for serum parameters of liver injury) Klf6-expression was as well significantly upregulated, compared to mice treated with corn oil alone (Fig. 3C). In this model of acute injury LC3-II levels were induced after CCl 4 injection, (Fig. 3D,E). To validate the in vivo observations, we quantified KLF6-expression in APAP treated cell culture models. Therefore, we treated HepaRG cells, which resemble the metabolic function of human hepatocytes, with different concentrations of APAP (5 mM, 10 mM and 20 mM) for 24 h to induce cellular stress and damage. Here, KLF6 was significantly upregulated after APAP treatment in a dose-dependent fashion (Supplementary Figure S3C). Similarly, in HepG2 cells, treatment with APAP for 24 h significantly induced KLF6-levels (Supplementary Figure S3D). We then quantified autophagy-induction in APAP-treated HepG2 cells and observed increased LC3-II and p62 levels compared to control cells (Supplementary Figure S3E,F,H and I). KLF6 induces autophagy and binds to promoter regions of BECLIN1 and ATG7. To verify the functional interaction between KLF6 and autophagy related targets, we transiently transfected HepG2 cells with an empty control vector (pcIneo) or a KLF6-expression vector (pcIneo-KLF6) in order to quantify autophagy induction in KLF6-over-expressing cells (Fig. 4). In parallel to KLF6-overexpression (Fig. 4A,B), LC3-II was increased and p62 levels were decreased in these cells (Fig. 4C,D). To assess autophagosome formation, we performed transmission electron microscopy with control vector transfected HepG2 (pcIneo) cells and pcIneo-KLF6 transfected HepG2 cells. In KLF6 over-expressing HepG2 cells (Fig. 5C), there were more autophagy-positive cells compared to pcIneo-transfected HepG2 cells ( Fig. 5A; for quantification of autophagy-positive cells see Supplementary Table S2). As a control for autophagy induction and autophagosome-formation we treated pcIneo-transfected HepG2 cells with 15 µM of rapamycin for 6 h (Fig. 5B). In addition, we performed Autophagy Tandem Sensor RFP-GFP-LC3B assay (Fig. 6), which confirmed increased formation of autophagosomes in KLF6-over-expressing HepG2 cells ( in LC3-II levels following chloroquine treatment comparing pcIneo and pcIneo-KLF6 transfected HepG2 cells (Fig. 4D). KLF6 belongs to the family of zinc finger proteins that regulate target genes and cellular pathways by binding to specific DNA motifs. We identified potential KLF6 binding motifs within the promoter regions of the autophagy related genes ATG7 and BECLIN1, and then confirmed transcriptional activation by luciferase reporter assays. To do so, we performed co-transfection with reporter plasmids in KLF6-over-expressing HepG2 cells and specific luciferase reporter plasmids carrying the promoter regions of ATG7 or BECLIN1. A background control was comprised of a commercially available random control vector containing a non-conserved, non-genic and non-repetitive fragment of equal length to the specific sequence upstream of the luciferase gene. As shown in Fig. 4E in HepG2 cells over-expressing KLF6, ATG7 promoter activity was significantly higher compared to control plasmid transfected cells. Thus, KLF6 transactivates ATG7 and therefore might influence the level of autophagy. Next, we performed chromatin immune precipitation assays (ChIP), which confirmed direct binding of KLF6 to the promoter regions of ATG7 and BECLIN1 (Fig. 4G). Interestingly, despite active binding of KLF6 to the BECLIN1 promoter, BECLIN1 luciferase activity was not altered by KLF6-overexpression in HepG2 cells (Fig. 4E). Predicted binding elements of KLF6 to promoter regions of ATG7 or BECLIN1 were assessed by using ChIP-seq data from KLF6-transfected HepG2 cells that were obtained from the NIH Encyclopedia of DNA Elements (ENCODE) database. This analysis clearly identified protein-DNA binding sides of KLF6 on regions encoding ATG7 and BECLIN1 (Supplementary Figure S2F+G). The p53-dependant transcriptional activation of ATG7 and BECLIN1 by KLF6 is independent of apoptosis. A direct interaction between KLF6 and p53 has previously been demonstrated in the context of IGF-1 regulation 27,28 . In contrast to KLF6, several direct and indirect interactions between autophagy and p53 have identified p53 as an important regulator of autophagy 29 . To investigate a potential interaction between KLF6 and p53 in the context of autophagy-induction, we used p53-deficient HepG2-303 cells to determine if KLF6 still leads to upregulation of autophagy-related genes in the absence of p53. To do so, we performed luciferase assays in KLF6-over-expressing HepG2-303 cells transfected with promoter-reporter constructs for ATG7 and BECLIN1. In contrast to p53-expressing HepG2 cells, the ATG7 promoter was not activated in KLF6-over-expressing HepG2-303 cells (Fig. 4F). Interestingly, the activation of BECLIN1 in HepG2-303 cells was enhanced in KLF6-over-expressing cells compared to control cells, pointing towards p53-dependent (ATG7) and p53-independent (BECLIN1) mechanisms, by which KLF6 regulates autophagy-related effector proteins. However, LC3-II-levels were obviously not changed in HepG2-303 cells treated with APAP or in HepG2-303 cells over-expressing KLF6 (Supplementary Figure S3G,J), implying that p53 is required to enhance LC3-II as a marker for increased autophagosome formation. To elucidate potential non-transcriptional effects of KLF6 on autophagy-induction, we further investigated its role in apoptosis-induction. Following cellular stress, autophagy can block apoptosis or caspase activation and promote survival by clearance of reactive oxygen species or damaged proteins. A switch from autophagy to apoptosis may occur, since autophagic and apoptotic molecules including BECLIN1 and BCL-2 interact directly 30,31 . Since p53 is also an activator of apoptosis and several mediators involved in autophagy-induction also contribute to Caspase-regulation/apoptosis-regulation, we measured expression levels of the apoptosis-related molecules BAX, BAD and BCL-2. Expression of these genes was not changed in KLF6-over-expressing cells compared to empty vector transfected HepG2 cells. However, as previously published, expression levels of P21 were reduced in KLF6-over-expressing HepG2 cells (Supplementary Table S3). Additionally, we performed a Proteome Profiler human Apoptosis Array to analyze the expression profiles of 35 apoptosis-related proteins using cell lysates from normal HepG2 cells (pcIneo) and KLF6-over-expressing HepG2 cells. This array did not highlight any differences between control vector transfected and KLF6-over-expressing HepG2 cells (Supplementary Figure S2E). Discussion KLF6 is a growth suppressor gene and the inactivation of KLF6 is associated with multiple human tumors 7,8,11 . Among several mechanisms of tumor suppression, KLF6 inhibits cell cycle progression and proliferation 32 . However, the behavior of KLF6 during liver regeneration following acute liver injury has not been assessed to date. With this study we establish that KLF6 is induced and translocated to the nucleus in hepatocytes among different models of acute liver injury. This activation is associated with enhanced hepatocyte proliferation in early liver regeneration. We further identify KLF6 as a transcriptional activator of ATG7 and BECLIN1, thereby establishing KLF6 as a novel mediator of autophagy. This novel function of KLF6 depends on the presence of p53, but appears to be independent of apoptosis. Healthy liver tissue has the ability to compensate for the loss of organ function in case of induced cell stress, acute injury or cell death. However, the excessive loss of functional liver tissue may lead to ALF. Following cell loss or death, activation of cell proliferation and regeneration, combined with attenuation of growth suppressor activity within remnant liver tissue restores liver cell mass. Downregulation of KLF6, a tumor suppressor gene that inhibits proliferation through induction of p21 and in synergy with p53 8, 10, 12 has been observed in primary liver tumors and is associated with a worse outcome in cancer 1,11,33 . Following PHx in mice, hepatocyte-specific deletion of Klf6 accelerates cell proliferation at early time points after resection. The later loss of growth induction in DeltaKlf6 mice suggests that mechanisms not related to hepatocellular Klf6 override its anti-proliferative effects as hepatocyte regeneration progresses. Furthermore, these observations might as well be confounded by Klf6-expression in non-parenchymal cells 34,35 . Here, cell proliferation was slightly reduced in in vitro experiments using KLF6-over-expressing HepG2 cells as shown by BrdU assay. Furthermore, KLF6-overexpression was accompanied with reduced expression levels of p21 in transiently transfected HepG2 cells. This transcription factor regulates cell cycle progression, DNA replication and repair by regulating the activity of different cyclin dependent kinases; its activation is controlled by the tumor suppressor protein p53 36,37 . Nonetheless, we observed a strong hepatocyte induction of KLF6 in models of acute liver injury and ALF patients, and an early proliferative advantage for hepatocyte-specific Klf6 knockout mice undergoing PHx. A marked reduction of autophagic vesicles in hepatocytes was first observed in 1979 by Pfeifer in rats undergoing PHx 38 . More recently, autophagy has been established as an essential mechanism required for liver regeneration after PHx, since in liver-specific Atg5 knockout mice liver regeneration and cell division are markedly impaired after PHx due to reduced ATP levels and decreased β-oxidation 39 . Here, utilizing a hepatocyte-specific Klf6 knockout model, we identified Klf6 as a transcriptional activator of the autophagy related genes Atg7 and Beclin1 in PHx and acute CCl 4 induced liver injury. Metabolism of APAP results in formation of NAPQI (N-acetyl-p-benzoquinone imine), which reacts with glutathione (GSH) to form GSH-adducts that can be secreted. In APAP overdose with progressive GSH-depletion NAPQI binds to cellular proteins and causes mitochondrial damages leading to cell death (mainly necrosis) and inflammation. In liver injury following APAP overdose, autophagy represses apoptosis, reduces cellular stress, inflammation and injury by removing damaged cells and organelles [22][23][24]31 . Ni et al. showed that SQSTM1/p62 plays an important role in reducing APAP protein adducts, while after shRNA-mediated p62-knockdown APAP protein adducts were increased in primary hepatocytes 23 . In aging mice, autophagy and hepatocellular apoptosis are induced, leading to impaired liver regeneration following PHx 40 . In a related study, autophagy played a critical role in liver regeneration and in the preservation of cellular quality, preventing hepatocytes from becoming fully senescent and hypertrophic. This effect was most likely mediated by p21 and stimulation of interleukins 39 . Interestingly, in a PHx model, mTOR inhibition severely impaired liver regeneration and increased autophagy rate. These effects were partly reversed by stimulation of the IL-6 and HGF pathways 41 . Our gene array data uncovered altered expression of autophagy-regulatory proteins in mice lacking hepatocyte Klf6 (DeltaKlf6 mice) following PHx. Accordingly, we documented the parallel induction of autophagy and KLF6 in several models of liver injury. In DeltaKlf6 mice, autophagy-induction was attenuated compared to controls and KLF6-over-expressing HepG2 cells showed increased LC3-II accumulation and formation of autophagosomes, while there was no evidence for increased autophagic flux in conditions of KLF6-over-expression as compared to control conditions. We then analyzed whether KLF6 functionally interacts with promoter regions of several autophagy-related genes, which contain conserved KLF6-binding motifs. ChIP assay analysis confirmed direct binding of KLF6 to promoter regions of ATG7 and BECLIN1. Interestingly, KLF6-mediated transcriptional activation of ATG7 is dependent on p53, since KLF6-overexpression activated the ATG7 promoter in HepG2, but not in p53 deficient Hep-G2-303 cells. Conversely, BECLIN1 transcriptional activation was induced by KLF6-overexpression under p53 deficient conditions, while KLF6 had no effect on BECLIN1 in HepG2 cells. A functional interaction between KLF6 and p53 has previously been described. Rubinstein et al. observed that a transcriptional effect of KLF6 on the IGF-1 receptor is dependent on the presence of p53 27 , and KLF6 itself is a transcriptional target of IGF1, which also requires p53 28 . KLF6 can also repress MDM2, which binds to the tumor suppressor p53 and thus accelerates its degradation in a mouse model of hepatocellular cancer 8 . Here, we observed a novel transcriptional activity of KLF6 by inducing two autophagy related genes (BECLIN1 in p53 deficient cells and ATG7 in the presence of p53) is switched, based on the presence or absence of p53. Beyond its role in autophagy, Beclin1 has also been described as a tumor suppressor gene in many cancer types and shares a BH3 domain with pro-apoptotic genes like Bid or Bad 42 . In our study expression levels of BAX, BID and BCL-2 were not changed in KLF6 over-expressing HepG2 cells. Furthermore, Beclin1 can alter p53 expression by regulating deubiquitination of p53 by USP10 43 . To date, no interaction between KLF6 and either BECLIN1 or ATG7 has been reported. Interestingly, the absence or presence of p53 determines a pro-tumorigenic or tumor-suppressing property of autophagy in a mouse model of pancreatic cancer 44 . Thus, KLF6 might serve as an important mediator in autophagy-induction but has no impact on apoptosis in the context of acute liver injury. Taken together, our findings establish that KLF6-expression is induced in models of acute liver injury and in patients with ALF. Here, we describe for the first time a direct transcriptional activation of autophagy-related genes by KLF6. This transcriptional activation depends on the presence (ATG7) or absence (BECLIN1) of p53. Thus, KLF6 drives autophagy-induction and autophagy-related cell death in acute liver injury. Material and Methods Cell culture. HepG2 cells were grown in DMEM-High-Glucose medium (Invitrogen, Calrsbad, CA, USA) with 10% of fetal bovine serum (FBS, Biochrom, Berlin, Germany, 1000 U/ml penicillin, 0.1 mg/ml streptomycin and 2 mM L-glutamine (PAA, Pasching, Austria). Cells were kept in an atmosphere with 5% CO 2 under 37 °C following standard protocols. In HepG2-303 cells p53 was stably knocked out and they were cultivated as described elsewhere 45 . Additional cell culture protocols are provided in Supplementary material. Transmission electron microscopy and Autophagy Tandem Sensor assay. HepG2 cells were transfected as described above with pcIneo or pcIneo-KLF6. For induction of autophagy and monitoring of autophagosome-formation, HepG2 cells were treated with 15 µM of Rapamycin for 6 h (Medchem Express, Monmouth Junctions, NJ USA). After incubation, cells were fixed for 2 h at room temperature using 2.5% glutaraldehyde in 0.1 M PB buffer (0.1 M Na 2 HPO 4 ; 0.1 M KH 2 PO 4 buffer). Cells were washed with PB buffer, removed from the cell culture dish; the cell pellet was postfixed in 2% osmium tetroxide, dehydrated in a graded series of alcohol and embedded in epoxy resin. Ultrathin sections were post-stained with uranyl acetate (1%) and lead citrate (0.4%). Sections were viewed in a Jeol TEM1400 Plus (Jeol, Tokyo, Japan). For visualization of autophagosomes in pcIneo or pcIneo-KLF6 transfected HepG2 cells we used the Premo ™ Autophagy Tandem Sensor RFP-GFP-LC3B Kit (Thermo Scientific/Life Technologies, Darmstadt, Germany) according to manufacturer's protocol. Fixed cells were viewed with a Leica SP8 confocal microscope (Leica Microsystems, Wetzlar, Germany). Chromatin immunoprecipitation assay. For Chromatin immunoprecipitation assay (ChIP) cells were cross-linked with a final concentration of 1% formaldehyde for 10 min at 37 °C, then washed and harvested in SDS lysis buffer (10% SDS; 0.5 M EDTA; 1 M Tris-HCl; containing proteinase inhibitor cocktail from Sigma-Aldrich, St. Louis, MO, USA) and sheared by sonication to fragment DNA. Samples were immunoprecipitated with 10 µg of anti-KLF6 antibody (polyclonal antibody KLF6 (R-173) or monoclonal antibody KLF6 (E-10) (Santa Cruz Biotechnologies, Dallas, TX, USA), anti-histone H3 antibody (Abcam, Cambridge, UK) or control IgG (Abcam) and protein-A/G agarose beads (Santa Cruz Biotechnologies). Following removal of cross-linked DNA/protein complexes by Proteinase K (Qiagen, Hilden, Germany) treatment, immunoprecipitated DNA was purified using QIAamp DNA Mini Kit (Qiagen) and used for PCR with ATG7 or BECLIN1 primers (Supplementary Table S3), encompassing the promoter region −200 bp to −400 bp upstream of transcriptional start site to amplify immunoprecipitated DNA, PCR products were visualized on an agarose gel. Animals and surgical procedures. Mice with a floxed Klf6 targeting vector (C57Bl/6;129Sv, Genentech, San Francicso, CA, USA) 46 were crossed with mice expressing Cre recombinase (Cre) under control of the albumin promoter (B6.Cg-Tg(Alb-cre) 21 Mgn/J; Jackson Labs, Bar Habor, ME, USA). After backcrossing, male offspring expressing Cre with two floxed Klf6 alleles were used as the experimental group ('DeltaKlf6'). Mice with two floxed alleles and no Cre expression were used as controls (wt). Temperature, humidity and light-dark cycle conditions were controlled; mice were allowed food and water ad libitum. Before surgical intervention animals were anesthetized, 70%PHx was performed by removing the left and median lobes of the liver 47 . Mice were sacrificed after 3 h, 12 h, 24 h, 48 h and 72 h following surgical intervention, respectively. Protocol for Affymetrix microarray analysis from liver tissue post PHx is given as Supplementary material. To induce APAP-induced liver injury C57Bl/6 mice received intraperitoneal injection of APAP (500 mg/kg bodyweight, Sigma-Aldrich) and were sacrificed 8 h after APAP-administration 26 . Carbon tetrachloride (CCl 4 , Sigma-Aldrich)-induced acute liver injury was achieved by intra-peritoneal injection of 2 µl/g bodyweight of CCl 4 or corn oil; animals were killed after 48 h. Experiments were conducted in three different facilities in accordance with relevant guidelines and regulations. Studies were approved by the Institutional Animal Care and Use Committee (IACUC) of Icahn School of Medicine at Mount Sinai (reference number LA09-00251), and the State authority for environment and animal welfare in Northrhine-Westfalia (LANUV, reference number 84-02.04-2013) for work conducted at the University of Duisburg-Essen and the RWTH Aachen. For baseline characteristics see Tables 1 and 2. Ethical considerations. All investigations in human material and the use of patient samples were approved by the Ethics Committee (Institutional Review Board) of the University Hospital Essen (reference numbers: 14-6066-BO and 09-4252) and the study protocol conformed to the ethical guidelines of the Declaration of Helsinki. Sample allocation in non-acute liver injury patients that underwent bariatric surgery was undertaken following patients' informed consent. As patient data and samples of the historic cohort of ALF patients were analyzed retrospectively from stored samples that were obtained for routine clinical use, informed consent from these subjects was explicitly not required according to the local ethics committee. Histopathology and sample handling. Liver tissue from mice or ALF patients (Supplementary Table S13) was stored in 4.5% formalin-solution, paraffin-embedded and sectioned. Stainings were performed using standard protocols; rabbit anti-KLF6 (R-173; Santa Cruz Biotechnology). Liver tissue for RNA and protein isolation was frozen in liquid nitrogen. Total RNA and protein from liver tissue were isolated by TRIzol ® extraction (Invitrogen), RNA was purified utilizing RNeasy Mini Kit (Qiagen). Protein lysates from cells were prepared using lysis buffer (50 mM Tris-HCl; 150 nM NaCl; 0.1% NP-40; 1% desoxycholic acid) containing complete mini EDTA-free protease inhibitor cocktail and phosphostop (Roche). Quantitative real time PCR. Reverse transcription was performed with the QuantiTect-RT kit (Qiagen) using 1 µg of total RNA. Specific mRNA expression levels were measured by quantitative realtime-PCR (qRT-PCR) performed on a CFX96 Touch Real-Time PCR Detection System (Bio-Rad) using QuantiTect SYBR Green Kit (Qiagen) in a volume of 15 μl including 2 μl of cDNA. Oligonucleotide sequences of used primers are shown in Supplementary Table S4. Melting curves were collected to ascertain specificity of PCR-products. Changes in mRNA-expression were calculated by the ΔΔ-Ct method and are presented as foldchange in relation to expression of a reference gene (HPRT or Sdha). Western blotting. For SDS-PAGE 30 µg of total protein were separated; immunoblotting was performed using standard procedures with the following primary antibodies: LC3 (Abcam), KLF6-R173 (Santa Cruz Biotechnologies), p62 (Enzo Lifesciences, Antwerpen, Belgium), β-Actin 13E5 and GAPDH 14C10 (Cell Signaling). After incubation with the appropriate horseradish peroxidase-conjugated secondary antibody, bound antibodies were visualized using ECL-Prime (GE Healthcare, Chalfont St. Giles, UK). Blotting images were generated using ChemiDoc System and Quantity One software (Bio-Rad) to quantify the densities of the bands.	v3-fos-license
2017-07-29T21:26:51.219Z	2017-06-01T00:00:00.000	4006333	{ "extfieldsofstudy": [ "Psychology", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2072-6643/9/6/606/pdf", "pdf_hash": "b7e1f5215cb0388862c2ae067572f378a608288f", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45368", "s2fieldsofstudy": [ "Environmental Science", "Medicine" ], "sha1": "63e77e6795845b41da018287a9419eb981a177d5", "year": 2017 }	pes2o/s2orc	Lifestyle Patterns and Weight Status in Spanish Adults: The ANIBES Study Limited knowledge is available on lifestyle patterns in Spanish adults. We investigated dietary patterns and possible meaningful clustering of physical activity, sedentary behavior, sleep time, and smoking in Spanish adults aged 18–64 years and their association with obesity. Analysis was based on a subsample (n = 1617) of the cross-sectional ANIBES study in Spain. We performed exploratory factor analysis and subsequent cluster analysis of dietary patterns, physical activity, sedentary behaviors, sleep time, and smoking. Logistic regression analysis was used to explore the association between the cluster solutions and obesity. Factor analysis identified four dietary patterns, “Traditional DP”, “Mediterranean DP”, “Snack DP” and “Dairy-sweet DP”. Dietary patterns, physical activity behaviors, sedentary behaviors, sleep time, and smoking in Spanish adults aggregated into three different clusters of lifestyle patterns: “Mixed diet-physically active-low sedentary lifestyle pattern”, “Not poor diet-low physical activity-low sedentary lifestyle pattern” and “Poor diet-low physical activity-sedentary lifestyle pattern”. A higher proportion of people aged 18–30 years was classified into the “Poor diet-low physical activity-sedentary lifestyle pattern”. The prevalence odds ratio for obesity in men in the “Mixed diet-physically active-low sedentary lifestyle pattern” was significantly lower compared to those in the “Poor diet-low physical activity-sedentary lifestyle pattern”. Those behavior patterns are helpful to identify specific issues in population subgroups and inform intervention strategies. The findings in this study underline the importance of designing and implementing interventions that address multiple health risk practices, considering lifestyle patterns and associated determinants. Introduction Overweight and obesity have progressively increased during the last decades and have become a major issue in public health, both in developed and developing economies across the five continents [1,2]. In Spain, recent data report that more than half of adults aged 18-64 years are classified as overweight or obese [3,4]. This is particularly worrying for the negative impact of this condition on health and quality of life [2,5,6]. In fact, the Global Burden of Disease project highlights that high body mass index (BMI) values were among the main risk factors driving most death and disability combined in the country in 2015 [7]. Overweight and obesity result from an imbalance between energy intake and expenditure. The role of diet in obesity is complex. Most research in this area has focused on specific foods and nutrients [8,9]. Nevertheless, the analysis of food patterns is particularly interesting, since foods are usually consumed in combinations and those may have synergistic, antagonistic, or moderating effects [10]. In addition, to date, it is acknowledged that joint interactions of multiple variables acting at different levels influence weight gain, such as lifestyles [11][12][13], sleep, including rhythm, duration or quality of sleep [14,15], eating behaviors [16], socioeconomic level, education, and other factors [4,11]. Data-driven methods explore the similarities between different food options in specific population groups [8]. Specifically, cluster analyses include several techniques aimed at grouping together individuals sharing a number of features or similar lifestyles. This classification allows for a better understanding of the influences of behaviors and lifestyles, as well as the potential cumulative effects of an unhealthy combination of those factors on the development of overweight and obesity [16]. Research of dietary patterns and the potential combination of those with other lifestyles can contribute to identifying effective strategies for the prevention of overweight and obesity among adults, as well as its social and health consequences [8]. Considering the above, the objectives of this paper are (a) to identify food patterns in the Spanish adult population; (b) to investigate if energy balance-related behaviors tend to assemble into meaningful patterns in Spanish adults; (c) to describe existing relationships between socio-demographical factors and different lifestyle patterns; and, finally; (d) to analyze the potential association of those correlates with excess body weight. Materials and Methods The data for this analysis were obtained from the ANIBES study, an observational cross-sectional study conducted on a random sample of the Spanish population aged 9-75 years. The aims and procedures used in ANIBES have been previously reported [17,18]. Briefly, the sample design of the ANIBES study was based on the Spanish population census 2012 according to sex, age, residence, and regional population size. A stratified multistep sampling procedure was used, with random selection of addresses in the municipalities and age and sex quotas for individuals within households. Interlocked quotas were established for age in the regions and size of habitat within the region. The sample selection procedure was based on random paths; 128 sampling points were selected. The final sample of the study consisted of 2009 individuals (1013 men, 50.4%, 996 women, 49.6%). In addition, a boost sample was recruited for the younger groups (9-12 years; 13-17 years, and 18-24 years) in order to ensure at least 200 individuals in each age group. Data for individuals aged 18-64 years were used for this analysis. That age range is meaningful from a risk prevention point of view, since it includes adults in the active working life span, thus, relevant when designing targeted preventive and health promotion strategies, considering different settings, such as workplaces, primary health care, and community-based interventions. In addition, the focus on the 18-64 year-old range was also relevant for comparison with previous studies conducted in Spain [19]. Data were collected between mid-September and mid-November 2013. The final protocol of the study was approved by the ethical research committee of the Community of Madrid, Spain. Lifestyle Factors Diet Dietary intake was assessed by means of a face-to-face 24-h diet recall interview assisted by a food picture atlas to estimate portion sizes. In addition participants completed a three-day food record aided by a tablet device (Samsung Galaxy Tab 2 7.0, Samsung Electronics, Suwon, South Korea), two consecutive weekdays and one weekend day, recording all foods and beverages consumed at home and away from home. Processing of food record inputs, coding and data cleaning has been extensively reported elsewhere [17,18]. Energy and nutrient intakes were calculated using new specifically-developed software, (VD-FEN 2.1 software -Dietary Evaluation Programme, Spanish Nutrition Foundation, Madrid, Spain) for the ANIBES study [20]. Consumption of all food and beverages was arranged into 16 food groups, 45 subgroups, and 754 food items, for in-depth analysis, based on the structure of the food composition database considering similarities in nutrient profile (Supplementary Materials Table S1). Physical Activity Physical activity data were collected by face-to-face interview using the validated International Physical Activity Questionnaire (IPAQ) [21]. Total minutes per week were computed for moderate to vigorous physical activity based on the IPAQ guidelines for data processing and analyses [22]. Data were cleaned and truncated based on IPAQ guidelines and previous research [23]. Additionally, the total minutes per week of commuting-related physical activity (walking, biking) were computed. IPAQ data was used for this analysis. Z-scores of minutes per week for each type of activity were calculated. Sleep Duration Sleep habits included the number of hours slept per night, on average, as reported by each individual. Smoking Smoking habits were assessed by different items in the protocol, including "During the past year, how many cigarettes per day did you smoke on average?" Body Measurements Anthropometric measurements were taken individually by trained interviewers, following international standard procedures previously tested in two pilot studies [24], reported in detail elsewhere [17,18]. Body mass index (BMI) was calculated as body weight in kilograms divided by the square of body height in meters. Overweight status was defined as BMI ≥ 25; obesity as BMI ≥ 30. Education The education levels were established in accordance with the Spanish educational system. After preliminary analysis of the distribution of the variable, categories were collapsed and recoded into a three-point scale, as follows: (1) low (less than seven years of education; primary school or less); (2) medium (7-12 years of education; lower to higher secondary education); and (3) high (13 years or more of education; higher vocational, college, and university studies). Geographical Area Geographical area in the country was collapsed into four different categories: north-northwest region; eastern region; central region and southern region. Data Cleaning Detailed data cleaning procedures have been previously described [17,18]. Participants were considered fully eligible after a verified quality check of the input from the tablet device of adequately completed three-day food records. After data cleaning stages, individuals remained in the database if they had successfully completed both face-to-face interviews during fieldwork and had measured weight, height, and waist circumference data. Of the initial sample of 1655 recruited individuals aged 18-64 years, 1617 individuals satisfied the inclusion criteria and had complete data for all of the variables included in this analysis. Data Analysis All statistical tests were performed using IBM SPSS Statistics for Windows, Version 22.0 (IBM Corp., Armonk, NY, USA). Descriptive statistics were computed for each variable. Dietary Patterns Exploratory factor analysis was performed to identify underlying dietary patterns, using the average food weight (g/day) consumed by each individual (three-day food record plus one-day 24-h recall) from 38 food groups as input variables. Food groups were used to further collapse dietary intake data in order to avoid missing data from non-consumers of episodically consumed foods. Z-scores for each food group were calculated to prevent the components being dominated by the foods that provide the highest amounts. Bartlett's test of sphericity and the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy were used to verify the appropriateness of factor analysis. Factors were also orthogonally rotated (the varimax option) to enhance the difference between loadings, which allowed for easier interpretability. Factors were retained based on the following criteria: factor eigenvalue > 1.20, identification of a break point in the scree plot, the proportion of variance explained, and factor interpretability [25]. The strength and direction of the associations between patterns and food groups were described through a rotated factor loading matrix. Food groups with factor loadings > 0.30 and communality > 0.20 were retained in the patterns identified. The factor score for each pattern was constructed by summing the observed intakes of the component food items weighted by the factor loading. A high factor score for a given pattern indicated high intake of the foods constituting that food factor, and a low score indicated low intake of those foods. Lifestyle Patterns To identify clusters with similar dietary patterns, physical activities, sedentary activities, sleeping habits and smoking a combination of hierarchical and non-hierarchical clustering analysis was used [26]. The variables used had different arithmetic scales; thus, Z-scores were calculated to standardize the dataset before clustering, to avoid a greater contribution to the distance of variables having larger ranges than variables with smaller ranges. Univariate and multivariate outliers (>3 SD) were removed. First, hierarchical cluster analysis was performed using Ward's method, based on squared Euclidian distances. Several possible cluster solutions were identified and compared to inform the next step, considering the coefficients and fusion level. A non-hierarchical k-means clustering procedure was used, specifying the number of clusters identified in the first step, using a random initial seed and 10 iterations in order to further refine the preliminary solution by optimizing the classification. The final cluster solution was selected based on interpretability and the percent of the study population in each cluster. Reliability and stability of the final cluster solution was tested by randomly taking a subsample (50%) of the total sample and repeating the analyses on this subsample. To check agreement, a kappa statistic was calculated between the cluster solutions of the subsample and that of the total sample. Pearson's chi-square tests were used to investigate the differences in cluster distribution by gender, age group, education level, geographical area, and BMI status. One-way ANOVA was used to compare physical activities, sedentary behaviors and sleep time across clusters stratified by gender and age group. General linear models were used to estimate multivariate means for food consumption and dietary pattern scores across clusters adjusted for age and energy intake. Binary logistic regression analysis was used to explore the prevalence odds ratios for obesity and overweight among lifestyle patterns. The models were adjusted for energy intake, sex, age, educational level, and geographical area. Statistical tests were two-tailed with a 5% level of significance. Sample Characteristics After exclusion of outliers and participants with incomplete data, 1617 subjects aged 18-64 years, 781 men and 836 women, were included in the analyses (Table S2). Some 38% of the sample was classified in the BMI range 25-29.9 and 21.6% had BMI values ≥ 30. There was no significant difference between men and women in age distribution, level of education, or geographical area. However, overweight and obesity rates were significantly higher in men than women. Dietary Patterns Dietary patterns were computed for the entire sample. Bartlett's test of sphericity and KMO = 0.591 supported the appropriateness of factor analysis. Four major factors were extracted, which explained 33.1% of the variance in the model. The first dietary pattern (DP) was labeled "Traditional DP" which had the highest loading on olive oil and vegetables, high scores on fish, meat, and fruit, and negative scores on pasta and so-called "pre-cooked" foods, which include food items such as croquettes and other processed foods usually prepared deep-fried for consumption. A DP labeled "Mediterranean DP" had high scores on water, fruit, yoghourt, fish, vegetables, cheese, and olive oil, and negative scores on meat and sugar sweetened beverages. A DP labeled "Snack DP" had high scores on bread, processed and cold meats, alcoholic beverages, salted snacks, cheese, and juices. Finally, "Dairy-sweet DP" had high scores on milk, sugar and sweets, cakes, pastry, and juices, and negative scores on alcoholic beverages. Overall, these patterns explained 33.07% of the variance ( Figure 1). Mean factor scores for "Traditional DP" and "Mediterranean DP" were significantly higher in the oldest age group (50-64 years), while "Dairy-sweet DP" factor scores were significantly higher in the younger ones (18-30 years). Men had significantly higher scores than women for "Snack DP" adjusted for age and energy intake; "Mediterranean DP" and "Dairy-sweet DP" had significantly higher scores among women. "Mediterranean DP" and "Traditional DP" factor scores were significantly higher in people with a higher educational level. Factor scores for the "Traditional DP" and "Dairy-sweet DP", adjusted for energy intake, age, and gender, were significantly higher in the north-northwest region, while "Mediterranean DP" and "Snack DP" factor scores were significantly higher in the Eastern Mediterranean region. Mean factor scores for "Traditional DP" and "Mediterranean DP" were significantly higher in the oldest age group (50-64 years), while "Dairy-sweet DP" factor scores were significantly higher in the younger ones (18-30 years). Men had significantly higher scores than women for "Snack DP" adjusted for age and energy intake; "Mediterranean DP" and "Dairy-sweet DP" had significantly higher scores among women. "Mediterranean DP" and "Traditional DP" factor scores were significantly higher in people with a higher educational level. Factor scores for the "Traditional DP" and "Dairy-sweet DP", adjusted for energy intake, age, and gender, were significantly higher in the north-northwest region, while "Mediterranean DP" and "Snack DP" factor scores were significantly higher in the Eastern Mediterranean region. Lifestyle Patterns Based on the four identified DPs, minutes per week of vigorous, moderate physical activity, walking, biking, sedentary time, sleep duration on weekdays, and smoking habits, the three-cluster solution was found to be adequate and meaningful regarding the different patterns. The kappa statistic (κ = 0.94) suggested good agreement. Lifestyle Patterns Based on the four identified DPs, minutes per week of vigorous, moderate physical activity, walking, biking, sedentary time, sleep duration on weekdays, and smoking habits, the three-cluster solution was found to be adequate and meaningful regarding the different patterns. The kappa statistic (κ = 0.94) suggested good agreement. Characteristics of the subjects classified in the lifestyle patterns identified are described in Table 1. The "Mixed diet-physically active-low sedentary lifestyle pattern" included 13% of the sample and a significantly higher proportion of men. The "Not poor diet-low physical activity-low sedentary lifestyle pattern" included 63.3% of the sample, a significantly higher proportion of women. The "Poor diet-low physical activity-sedentary lifestyle pattern" included 23.6% of the sample. Pearson's chi-square tests were used to investigate the differences in cluster distribution by gender, age group, education level, geographical area, and BMI status. A higher percentage of people across age groups were classified into the "Not poor diet-low physical activity-low sedentary lifestyle pattern". However, a significantly lower proportion of individuals in the older age group (50-64 years) was classified in the "Mixed diet-physically active-low sedentary lifestyle pattern" and a higher proportion of people aged 18-30 years were classified into the "Poor diet-low physical activity-sedentary lifestyle pattern". The highest proportion of people with a lower educational level was classified into the "Not poor diet-low physical activity-low sedentary lifestyle pattern". There was no difference in the clusters by geographical area. Prevalence rates of obesity were significantly higher among people allocated in the "Not poor diet-low physical activity-low sedentary lifestyle pattern" and overweight in the "Mixed diet-physically active-low sedentary lifestyle pattern", however, those differences were not significant in the stratified analysis by age and gender. Table 2 describes physical activity behaviors, sedentary, sleep time on weekdays, smoking behavior, and dietary pattern Z-scores in the lifestyle patterns. Vigorous physical activity, moderate physical activity, walking time, as well as the Z-scores of the Mediterranean DP were significantly higher in men and women classified in the "Mixed diet-physically active-low sedentary lifestyle pattern". Z-scores for "Snack DP" were also higher in men in that lifestyle pattern. Men and women in the "Poor diet-low physical activity-sedentary lifestyle pattern" had significantly higher scores on the "Dairy-sweet DP" and sedentary time. Consumption of selected food groups and beverages by lifestyle pattern in men and women is described in Table 3. Consumption of fruit, pasta, olive oil, water and alcoholic beverages, particularly wine and beer, was significantly higher in men and women included in the "Mixed diet-physically active-low sedentary lifestyle pattern". Consumption of milk, cakes and pastry, sugar and sweets was significantly higher in men and women classified in the "Poor diet-low physical activity-sedentary lifestyle pattern". Men in this lifestyle pattern showed significantly higher consumption of pre-cooked deep fried foods and high alcoholic content beverages. Women in this pattern had significantly higher consumption of savory snacks, juices and sugar sweetened soft drinks beverages. Prevalence of obesity was compared between lifestyle patterns, adjusting for gender, age, educational level, geographical area and energy intake ( Table 4). The prevalence odds ratio (POR) for obesity in men, 0.52 (IC 95% 0.29-0.92) allocated in the "Mixed diet-physically active-low sedentary lifestyle pattern" was significantly lower compared to those in the "Poor diet-low physical activity-sedentary lifestyle pattern". Discussion In this cross-sectional study conducted in a random sample of Spanish adult population aged 18-64 years, four dietary patterns were characterized by factor analysis and subsequently used to identify three different lifestyle patterns based on the DPs, physical activities, sedentary time, and sleeping and smoking habits, using cluster analyses: a "Mixed diet-physically active-low sedentary lifestyle pattern", a "Not poor diet-low physical activity-low sedentary lifestyle pattern" which included 63.3% of the sample, and a "Poor diet-low physical activity-sedentary lifestyle pattern". A higher proportion of people aged 18-30 years was classified into the "Poor diet-low physical activity-sedentary lifestyle pattern". The prevalence odds ratio for obesity in men allocated in the "Mixed diet-physically active-low sedentary lifestyle pattern" was significantly lower compared to those in the "Poor diet-low physical activity-sedentary lifestyle pattern". Analysis of dietary patterns has become an important approach in food consumption studies and nutritional epidemiology [11,27,28] Researchers report different methodological approaches and procedures to identify diet and lifestyle patterns, such as a priori defined patterns or data-driven, identified using principal components (PCA) [29], factor analysis [11,30], cluster analysis [31][32][33] and, more recently, hybrid methods [2]. Some pieces of research have identified lifestyle patterns associated to specific phenotypes and obesity in several population groups [34,35] and contributed to unravel the consequences that different diets may have on health [2,11,[35][36][37][38]. In this study factor analysis was used to identify dietary patterns that were later used in cluster analysis. In the ANIBES study four DPs were singled out in adults. Similar DPs have been reported [37]. In Spain, in the context of the DORICA study, a pooled analysis of population-based cross-sectional surveys conducted between 1990 and 2000, using cluster analysis of three different DPs were characterized. A "Meat rich DP" with higher consumption of meat, cereals, and potatoes, an "Unbalanced DP" with higher consumption of milk and low consumption of vegetables and cereals, and a "Mediterranean DP" with higher consumption of vegetables, fruit, fish, and olive oil. The "Unbalanced DP" was associated with higher alcohol consumption, inadequate fruit and vegetable consumption and poor physical activity. This pattern was associated to higher prevalence odds ratio for metabolic syndrome. Conversely, the "Mediterranean DP" was associated with better fat intake quality profile and fiber intake [19]. More recently other authors have identified two DPs using factor analysis in Spanish adults in the context of Food4Me Project. A pattern with higher consumption of fast and processed foods, potatoes, red and white meats, processed cereals, and low consumption of fruit and vegetables, directly associated with BMI. A second DP showed higher consumption of eggs, fish, legumes, nuts, low-calorie beverages, oil, vegetables, and white meat [8]. Despite cross-sectional studies not allowing for causal inference, food patterns with higher energy density and low fiber content have been associated with higher obesity prevalence rates [2]. In longitudinal studies, such as the Baltimore Study, dietary patterns high in reduced-fat dairy, high fiber grains and cereals, vegetables and fruits, and low in meats, sugar-sweetened soft drinks, refined grains, and high fat dairy products, were significantly associated with lower weight gain, prospectively [39]. An energy-dense, high-saturated fat and low-fiber dietary pattern with high loadings of fast foods and snacks and low loadings of fruits and vegetables was shown to increase body weight, waist circumference, blood pressure, serum insulin, and lipid profile during a 10-year follow-up in severely obese Swedish adults [40]. In line with this, in adult Canadians a strong and consistent relation between an energy-dense, high-fat, low-fiber density dietary pattern and the risk of obesity was observed, significant in different population subgroups [2]. In the ANIBES study, 45.6% of male and 27.5% of female Spanish adults did not meet the recommendation of 150 min/week of moderate PA. In addition, 56.2% of male and 74% of female adults did not meet the recommendations for vigorous PA [13]. In this study we analyzed if these physical activity behaviors combined in a meaningful way with other energy balance-related behaviors, DPs, and lifestyles. We also used this procedure to identify dietary and lifestyle patterns in Spanish children and adolescents and identified two patterns: an "Unhealthier lifestyle pattern" a combination of low physical activity and poorer diet, and a "Healthier lifestyle pattern" which combined high physical activity, low sedentary behavior, longer sleep duration, and healthier diet. Studies analyzing individual behaviors have found healthier eating habits, such as adequate fruit and vegetable consumption, to be more prevalent among women than in men [41], but adherence to physical activity recommendations were more prevalent in men than in women [42]. In line with the above, in this study "Snack DP" mean scores adjusted for age and energy intake were significantly higher in men than in women. Conversely, those of the "Mediterranean DP" were significantly higher in women. Men and women classified in the "Mixed diet-physically active-low sedentary lifestyle pattern" showed the highest mean scores for the "Mediterranean DP". However, men in that lifestyle pattern showed the highest mean scores for the "Snack DP", as well. The highest mean scores for the "Dairy sweet DP" were observed in men and women classified into the "Poor diet-low physical activity-sedentary lifestyle pattern". In this study a significantly higher proportion of men was classified into the "Mixed diet-physically active-low sedentary lifestyle pattern" (71.9%), while significantly more women were classified into the "Not poor diet-low physical activity-low sedentary lifestyle pattern" (58.5%). Men classified in the "Mixed diet-physically active-low sedentary lifestyle pattern" were less likely to be obese (POR 0.52; 95% IC 0.29-0.93), compared to those in the "Poor diet-low physical activity-sedentary lifestyle pattern". Cassidy et al. reported obese adults to be two to five times more likely to report an "unhealthy phenotype", a combination of low physical activity, high TV viewing, and poor sleep duration compared to normal weight adults in the UK Biobank [43]. A significantly higher proportion of women classified into the "Mixed diet-physically active-low sedentary lifestyle pattern" had a high educational level (35.6%). Women with a low educational level were more likely to be obese compared to those highly educated. Clustering and co-occurrence of multiple risk behaviors has been reported to be more likely in men and women with lower education or intermediate educational level compared with those with higher education [44]. As shown in this study, individuals may follow a variety of unhealthy lifestyle behaviors which combine to favor weight gain, or mixtures of healthy and unhealthy practices. The findings in this study underline the importance of designing and implementing interventions that address multiple health risk habits, considering lifestyle patterns, clustering of risk behaviors, and associated determinants. It is recognized that interventions which target more than one risk behavior have the potential for greater health benefits, enhance health promotion opportunities and can contribute to reduce health care costs. Such an approach allows for more adequate tailoring of interventions to participants' profiles and helps to address health inequalities, reinforcing policies to ensure interventions accessible to socially-disadvantaged groups. The ANIBES study has several strengths which include the careful design, protocol, and methodology used for dietary data collection and validated questionnaires used to collect information on physical activity which have shown good reliability and reproducibility. The study was conducted on a random population sample of the Spanish population aged 9-75 years. A limitation of this study is its cross-sectional design, which provides evidence for associations but not causal relationships. Measures of physical activity relied on self-reports and could be biased, although a careful multistep quality control procedure was implemented to minimize bias. In addition, we used factor analysis to identify DPs based on food consumption information collected by three-day food records and one 24-h recall, considering food groups to further collapse dietary intake data and avoid missing data from non-consumers of episodically consumed foods. Finally, both factor analysis and cluster analysis are procedures commonly used to identify DPs and analyze clustering of lifestyles. However, those procedures rely on several subjective decisions that may influence outcomes regarding the number and type of patterns and clusters identified. Conclusions Four dietary patterns were characterized in Spanish adults aged 18-64 years, one of them closer to the Mediterranean DP. These patterns were subsequently used to identify three different lifestyle patterns: a "Mixed diet-physically active-low sedentary lifestyle pattern", a "Not poor diet-low physical activity-low sedentary lifestyle pattern" which included 63.3% of the sample, and a "Poor diet-low physical activity-sedentary lifestyle pattern". A higher proportion of people aged 18-30 years was classified into the "Poor diet-low physical activity-sedentary lifestyle pattern". The prevalence odds ratio for obesity in men allocated in the "Mixed diet-physically active-low sedentary lifestyle pattern" was significantly lower compared to those in the "Poor diet-low physical activity-sedentary lifestyle pattern". Although prospective research on large population samples would be desirable to further analyze and track over time the lifestyle patterns and how they influence the development of overweight and obesity, these behavior patterns are helpful to identify specific issues in population subgroups and inform intervention strategies. Further research is needed on factors associated with lifestyle patterns to gain insight in the population subgroups at higher risk. The findings in this study underline the importance of designing and implementing interventions that address multiple health risk practices, considering lifestyle patterns and associated determinants. Supplementary Materials: The following are available online at www.mdpi.com/2072-6643/9/6/606/s1, Table S1: Food groups and subgroups in the ANIBES Study, Table S2: Characteristics of the sample analyzed, Figure S1: Final cluster center scores for lifestyle patterns identified in Spanish adults aged 18-64 years in the ANIBES study.	v3-fos-license
2022-05-19T15:10:37.974Z	2022-05-01T00:00:00.000	248880551	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2304-8158/11/10/1451/pdf?version=1652785252", "pdf_hash": "e88c04807c229fd8ac1ca0965665bf7aa9428c9e", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45370", "s2fieldsofstudy": [ "Biology" ], "sha1": "d23b96f8da93259b0bb73962e8b7a29ef49a3b85", "year": 2022 }	pes2o/s2orc	Lactiplantibacillus plantarum Y42 in Biofilm and Planktonic States Improves Intestinal Barrier Integrity and Modulates Gut Microbiota of Balb/c Mice In our previous study, Lactiplantibacillus plantarum Y42 showed some potential probiotic functions and the ability to form biofilm. The aim of this study was to compare the similarities and differences in the probiotic and physiological traits of L. plantarum Y42 in the biofilm and planktonic states. L. plantarum Y42 in the biofilm state was proven to have higher survival after passing through mimic gastrointestinal fluid, as well as excellent adhesion properties on the HT-29 cell monolayers, than those in the planktonic state. The expression of tight junction proteins (TJ proteins) of HT-29 cell monolayers treated by L. plantarum Y42 in the planktonic state increased, while similar changes were not observed in the HT-29 cells treated by the strain in the biofilm state. Furthermore, Balb/c mice were orally administered L. plantarum Y42 in the biofilm and planktonic states, respectively. Compared to the planktonic state, the oral administration of L. plantarum Y42 in the biofilm state significantly boosted IgA levels and improved the immunity of the mice. High-throughput sequencing showed that the diversity and structure of the intestinal flora of the mice were changed after the oral administration of L. plantarum Y42, including the up-regulated relative abundance of Lactobacillus in the intestinal tract of the mice, with no difference between the biofilm and planktonic states. Moreover, oral administration of L. plantarum Y42 in biofilm and planktonic states reduced the release of proinflammatory factors, to a certain extent, in the serum of the mice. The similarities and differences in the probiotic and physiological properties of L. plantarum Y42 in the biofilm and planktonic states can be contributed to the reasonable application of the strain. Introduction In recent years, probiotics have been widely used as a human food additive and are widely researched for its probiotic properties [1]. Several probiotics have been demonstrated to protect the gastrointestinal tract from pathogens [2], regulate the immune system [3], alleviate diarrhea [4], and reduce cholesterol [5]; additionally, they have shown much other activity targeted to human health. It is essential that probiotic strains pass through the gastrointestinal tract and maintain high survival rate, which is considered to be one of the necessary factors for probiotics to promote their physiological functions [6,7]. Some strains belonging to lactobacilli are considered as potential probiotics and reside in human and animal intestines, mainly in the biofilm mode [8,9]. Biofilm is an aggregate of cells that are usually embedded in a self-produced extracellular polymer matrix and adhere to the surface of an organism or non-organism [10]. Lactobacilli strains have the ability to form biofilm, and bacteria cells in biofilm state have higher resistance to poor growth conditions [11]. For example, Lactobacillus plantarum subsp. plantarum JCM 1149 in Biofilm Formation of L. plantarum Y42 The biofilm formation of L. plantarum Y42 was measured using the methods, with some modifications, described by Stepanovic [17]. The concentration of L. plantarum Y42 was adjusted to 2 × 10 6 CFU/mL with fresh MRS medium. A total of 100 µL of the bacteria suspension was then placed into an optically clear 96-well plastic plate (Costar 3599, Corning, NY, USA), with a negative control containing 100 µL MRS medium. The total biofilm was quantified using the crystal violet (CV) assay, after the microtiter plates were incubated for 24 h at 37 • C. First, the medium was removed, and each well was washed four times with 200 µL sterile phosphate buffer saline (PBS) to remove unattached cells. After that, 200 µL per well of 95% methanol was added to each well of the plate, in order to fix the cells for 15 min. The biofilm was stained with 200 µL of 2.0% crystal violet solution for 5 min after the methanol had evaporated and then washed with distilled water. Finally, 33% acetic acid was added for 30 min to dissolve the biofilm with the dye attached, and the optical density (OD) at 570 nm was determined by Multiskan GO 1510 (Thermo Fisher Scientific, MA, USA). According to Sun et al.'s report, biofilm formation capacity of strains were classified as follows: OD ≤ ODc non-adherent; ODc < OD ≤ 2 × ODc weakly adherent; 2 × ODc < OD ≤ 4 × ODc moderately adherent; 4 × ODc < OD strongly adherent [15]. The ODc value was the sum of the average value and three times of the standard deviation of the blank optical density. Biofilm Growth Curve and Observation by SEM L. plantarum Y42 with 2% inoculation (v/v) was inoculated into fresh MRS. The microtiter plates filled with 100 µL of L. plantarum Y42 bacteria suspension were incubated for 8,12,14,20, and 24 h at 37 • C, and biofilm formation was quantified using CV assay, as described above. The morphologies of L. plantarum Y42 cultured for 8 and 20 h on sterile glass coverslips were observed according to Deng's method, with some modifications [18]. Coverslips were immersed in each well of the 6-well polystyrene plates containing 2 mL suspension of mixed bacterial. After incubation at 37 • C for the desired time, the coverslips were aseptically removed with sterile forceps and rinsed with sterile PBS. The specimens were first fixed using 2.5% glutaraldehyde for 6-8 h, then dehydrated in a series of graded aqueous ethanol solutions and, finally, air dried at room temperature. After gold spraying, the samples were observed by SEM. Preparation of Planktonic and Biofilm Cells Planktonic and biofilm cells of L. plantarum Y42 were prepared according to the method reported by Sun et al. and Sadiq et al. [15,16]. According to the formation curve of biofilm of the strain, we selected the strains cultured for 8 h at 37 • C under static conditions as the planktonic state; correspondingly, biofilm cells were formed at 37 • C for 20 h under static conditions. After incubation, cells were harvested from MRS broth cultures by centrifugation (6000× g for 5 min at 4 • C). Finally, cell pellets were washed twice with phosphate-buffered saline (PBS), and the harvested pellets were resuspended in PBS at 10 9 CFU/mL for subsequent experiments. Tolerance to Artificial Gastrointestinal Conditions Evaluating the gastrointestinal tolerance capacity of L. plantarum Y42 in planktonic and biofilm states was highly important, as they may have different effects in humans. The gastrointestinal tolerance of strains was previously evaluated by Zhou et al. [19]. For the preparation of artificial simulated gastric juice, 0.3% pepsin (w/v) was dissolved in phosphate buffer solution; then, the pH was adjusted to 2.5. Additionally, the phosphate buffer solution containing 1.8% bile salt (w/v) and 0.3% trypsin (w/v) was prepared; then, the pH was adjusted to 8.0, which is known as artificial simulated intestinal juice. The 50 µL pre-cultured strains (1 × 10 9 CFU/mL), in the biofilm and planktonic states, were inoculated in 450 µL simulated gastric juice and cultured at 37 • C for 3 h; then, 50 µL culture medium was inoculated at 450 µL simulated intestinal fluid and cultured at 37 • C for 8 h. The survival number of bacteria in the simulated gastric juice and simulated intestinal fluid was determined via the plate counting method, using MRS agar medium. The survival rate of the strains was evaluated by the following equation: The human colon cancer cell line HT-29 was obtained from the Chinese Academy of Science (Shanghai, China). The HT-29 cells were cultivated in RPMI-1640 medium supplemented with 10% (v/v) (56 • C, 30 min) fetal calf serum (Sijiqing Co., Ltd., Hangzhou, China) at 37 • C, with 5% CO 2 and 95% air [20]. Then, cell viability was assessed using trypan-blue dye (0.2%) in PBS (pH 7.2), and the cell number was determined by hemocytometer. Adhesion Assay on HT-29 Cells of Biofilm and Planktonic Bacteria The adhesion assay was performed as described by Greppi, with minor modifications [21]. Briefly, HT-29 cells were seeded at 2.5 × 10 5 cells per well in 24-well microtiter plates and cultured above for 2-4 days. Then, the cells were washed twice with PBS, in order to remove antibiotics. L. plantarum Y42 in planktonic and biofilm states were resuspended in RPMI-1640 medium, without antibiotics, up to 1 × 10 8 CFU/mL, respectively. A total of 100 µL of the L. plantarum Y42 suspensions were transferred into the wells and incubated for 2 h at 37 • C. To remove unattached bacteria, the HT-29 cells were washed three times with PBS. Afterwards, 500 µL of Triton-X solution (5% in PBS) was added into the wells to digest HT-29 cells. Then, the mixtures in each well were homogenized. The mixture solution was continuously diluted, and the L. plantarum Y42 number attached to the HT-29 Animal and Experimental Design All animal experiments were carried out in accordance with the guidelines of the Experimental Animal Ethics Committee of Dalian Polytechnic University (SYXK2017-0005). Seven-week-old Balb/c female mice (~20 g) were used in the study. They were purchased from Liaoning Changsheng Biotechnology Company (Benxi, China) and allowed to acclimate for one week prior to experiment; they were provided with fresh water and commercial food. The mice were then randomly divided into three experimental groups, with six animals each group, which were expressed as the con, planktonic (oral administered with planktonic cells of L. plantarum Y42), and biofilm groups (oral administered with biofilm cells of L. plantarum Y42). Afterwards, the cells were washed twice in sterile physiological saline and suspended in sterile physiological saline solution to a final concentration of 10 9 CFU/mL. The planktonic and biofilm groups were orally administered to the mice once daily, respectively, at a dose of 0.2 mL (10 9 CFU), for a period of 14 d. The mice were deprived of food and water for 12 h, then anesthetized, and blood was collected from the retrobulbar plexus of mice on the 14th d. After centrifugation (2000× g, 10 min, 4 • C), a serum sample was collected. The colons were separated from the proximal rectum, close to its passage under the pelvisternum. 2.8. IgA, IgG, IgM, IL-6, TNF-α, and IL-10 Analysis Serum samples were obtained from blood of mice by centrifugation at 2010× g for 10 min at 4 • C and stored at −80 • C after freezing in liquid nitrogen. The concentrations of IgA, IgG, IL-6, TNF-α, and IL-10 in the serum of the mice were detected by ELISA kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China), according to the instructions of the manufacturer. Western Blotting Analysis The western blotting analysis was performed as described by Fu et al., with minor modifications [4]. Total proteins were obtained from mice colon tissue of the mice and HT-29 cells by protein extraction kit (Solarbio Life Science, Beijing, China), according to manufacturer's description. The concentration of protein was quantified using the BCA protein assay kit (Solarbio Life Science, Beijing, China). Next, the equivalent protein samples were subjected to electrophoresis using sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE), and electrophoresis was stopped just after the electrophoretic band of bromophenol blue ran out. After electrophoresis was completed, the proteins were transferred to polyvinylidene difluoride (PVDF) membranes, and the PVDF membranes were sealed with 5% skim milk powder for 1 h. The skim milk powder on the PVDF was washed using TBST (five times each time) for 10 min. Then, the PVDF membranes were incubated overnight at 4 • C with primary antibodies, including occludin, claudin-1, and ZO-1 (1:2000, Beyotime Institute of Biotechnology, Shanghai, China). After washing with TBST (5 times each time) for 10 min, the membranes were warmed with secondary antibody (1:2000, Beyotime Institute of Biotechnology, Shanghai, China). Finally, the specific bands were visualized with the ECL detection kit (Beyotime Institute of Biotechnology, Shanghai, China), and the bands were visualized for quantification using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Gut Microbiota Analysis Fresh feces of the mice were frozen with liquid nitrogen immediately after collection and kept at −80 • C. The genomic DNA of the sample was extracted by the cetyltrimethy- lammonium bromide method. Briefly, the samples were added to CTAB buffer, and the total DNA was extracted with chloroform: isoamyl alcohol (24:1), and the precipitation was washed twice with 75% ethanol and dissolved with ddH 2 O. Then, the purity and concentration of DNA were detected by 1% gel electrophoresis. All PCR reactions were carried out with 15 µL of Phusion ® High-Fidelity PCR Master Mix (New England Biolabs, Ipswich, MA, USA), 2 µM of forward and reverse primers, and about 10 ng template DNA. Mixture PCR products was purified with Qiagen gel extraction kit (Qiagen, Dusseldorf, Germany). Sequencing libraries were generated using TruSeq ® DNA PCR-Free Sample Preparation kit (Illumina, San Diego, CA, USA), following manufacturer's recommendations, and index codes were added. HTS was performed using an Illumina HiSeq platform (Novogene Bioinformation Science and Technology Co., Ltd., Beijing, China). Statistical Analysis Results were expressed as the mean ± SD. Statistical analysis were performed using GraphPad Prism software (version 8.2.1) and repeated-measures ANOVA, with the level of significance set at p < 0.05. Biofilm Formation of L. plantarum Y42 The biofilm formation of L. plantarum Y42 was determined by the CV assay. L. plantarum Y42 adhered on polystyrene and formed biofilm after 24 h growth, having a moderate capacity for biofilm formation (OD 570 mean value = 0.56). Sun et al. estimated the biofilm formation ability of 79 strains of lactic acid bacteria (LAB) using CV assay, with the OD 570nm values ranging from 0.412 ± 0.054 to 2.266 ± 0.057, and found that different strains had different biofilm-forming abilities, which may be due to the species specificity of the strains [15]. As shown in Figure 1A, biofilm formation could be affected by culture time. When culture time reached 8 h, the initial adhesion of L. plantarum Y42 on polystyrene started, but L. plantarum Y42 biofilm was not obvious. As the culture time extended to 20 h, the mature biofilm of L. plantarum Y42 on polystyrene was formed. During the period from 20 to 24 h, the total biofilm decreased, due to the mature biofilm falling off. Tolerance of Biofilm and Planktonic L. plantarum Y42 to Artificial Gastrointestinal Conditions To test the tolerance of the L. plantarum Y42 in the different states under the human gastrointestinal tract, the viability of the L. plantarum Y42 under the artificial gastrointestinal conditions was evaluated. As shown in Table 1, compared with initial cell numbers, under the acid stress of gastric fluid, the survival rates of L. plantarum Y42 in different states showed no significant difference. Then, L. plantarum Y42 in the biofilm state showed a significantly higher (p < 0.05) survival rate when exposed to simulated intestinal fluid than that in the planktonic state, indicating that L. plantarum Y42 in the biofilm state had greater resistance to gastrointestinal conditions. Table 1. Tolerance to artificial gastric and intestinal conditions of L. plantarum Y42 in the biofilm In addition, SEM was employed to observe the morphologies of the L. plantarum Y42 on glass coverslips at 8 and 20 h. As shown in Figure 1B, L. plantarum Y42 existed only as a single or paired bacterium at 8 h and was considered as a planktonic state. At 20 h, L. plantarum Y42 cells aggregated together and stuck to the coverslips, indicating the matured biofilm at 20 h. Therefore, L. plantarum Y42 cells cultured for 8 h were selected as a planktonic state, and L. plantarum Y42 cells cultured for 20 h were selected as a biofilm state. Tolerance of Biofilm and Planktonic L. plantarum Y42 to Artificial Gastrointestinal Conditions To test the tolerance of the L. plantarum Y42 in the different states under the human gastrointestinal tract, the viability of the L. plantarum Y42 under the artificial gastrointestinal conditions was evaluated. As shown in Table 1, compared with initial cell numbers, under the acid stress of gastric fluid, the survival rates of L. plantarum Y42 in different states showed no significant difference. Then, L. plantarum Y42 in the biofilm state showed a significantly higher (p < 0.05) survival rate when exposed to simulated intestinal fluid than that in the planktonic state, indicating that L. plantarum Y42 in the biofilm state had greater resistance to gastrointestinal conditions. The survival of bacteria in the gastrointestinal tract is important for the physiological functions of probiotics [22]. L. plantarum Y42 in the biofilm state showed a relatively stronger resistance to artificial gastrointestinal conditions. L. plantarum (No. 23941) in the biofilms showed excellent gastrointestinal resistance, as compared with planktonic cells [23], which was consistent with our present results. It was reported that the heat shock proteins and some amino acid biosynthetic pathways of L. plantarum J26 in the biofilm state were increased, which enhanced the ability of L. plantarum J26 to resist environmental stress [15]. In addition, extracellular matrix secreted by L. paraplantarum L-ZS9, such as polysaccharides and proteins, are important to withstand external pressure [24]. Adhesion Rates of Biofilm and Planktonic L. plantarum Y42 on HT-29 Cell Monolayers Adhesion on the intestinal epithelium of the host is an important profile for probiotics [2]. As shown in Figure 2, the adhesion rate of L. plantarum Y42 in the biofilm state on HT-29 cell monolayers was 1.1 times higher than that of the planktonic state. It was proven that L. plantarum DB200 in the biofilm state excreted more adhesion-related proteins than that of the planktonic state and showed the highest autoaggregation [14]. Effects of Biofilm and Planktonic L. plantarum Y42 on the TJ Proteins Expression of HT-29 Cell Monolayers To evaluate the effects of L. plantarum Y42 in the biofilm and planktonic states on barrier function of the HT-29 cell monolayers, the expression of TJ proteins (claudin-1, occludin, and ZO-1) was assessed by western blotting. As shown in Figure 3, compared Effects of Biofilm and Planktonic L. plantarum Y42 on the TJ Proteins Expression of HT-29 Cell Monolayers To evaluate the effects of L. plantarum Y42 in the biofilm and planktonic states on barrier function of the HT-29 cell monolayers, the expression of TJ proteins (claudin-1, occludin, and ZO-1) was assessed by western blotting. As shown in Figure 3, compared to the BF group, the relative expression of claudin-1 and ZO-1 of the HT-29 cell monolayers in the PL group was significantly up-regulated. Similarly, the expression of occludin was also up-regulated to certain levels in the PL groups, but was not significantly different to the BF group. Animal Experiments 3.5.1. Effects of Biofilm and Planktonic L. plantarum Y42 Administration on Body Weight of Balb/c Mice Furthermore, the L. plantarum Y42 in the biofilm and planktonic states was orally administered to Balb/c female mice, respectively. The mice did not experience any negative effects as a result of the strain oral treatment, and the change of body weight was also measured daily during L. plantarum Y42 oral administration. As shown in Figure 4, the weight gains of the Balb/c mice among the three groups showed no significant difference (p > 0.05). Each experimental group of mice gained normal weight, indicating that L. plantarum Y42 is safe for animal growth. TJ proteins play an important role in maintaining the intestinal barrier function, which are in dynamic change under the stimulation of external substances [25]. TJ proteins separate coelenterates from their internal environment and limit the transmission of pathogens, toxins, and allergens as a physical barrier. In the present study, L. plantarum Y42 in the planktonic state increased the expression of TJ proteins, and we concluded that planktonic L. plantarum Y42 enhanced the intestinal barrier function, compared to biofilm L. plantarum Y42. It was reported that there were significant differences in amino acid metabolic pathways of L. paraplantarum L-ZS9 in the biofilm and planktonic states, and the contents of arginine, proline, and L-valine in the biofilm cells were increased, while only L-glutamine was increased in planktonic cells [23]. Glutamine is an important nutrient in the metabolism of small intestines and maintains the intestinal mucosal barrier by increasing the height of the intestinal villi, reducing the permeability of the intestinal mucosa, and preventing bacterial migration [26,27]. Several studies have shown that the extracellular matrix of bacteria biofilm, such as extracellular polysaccharide, which limits the communication between the environment and bacteria cells interior and prevents the bacteria from binding to the cell receptor toll-like signaling (TLR) of the intestinal epithelial cells [28]. TLR signaling mediates the TJ proteins' changes in colon tissues after feeding of L. plantarum WCFS1 [29]. Compared to the planktonic L. plantarum Y42, L. plantarum Y42 in the biofilm state down-regulated the tight junction proteins expression. The explanation for this is unknown, but it might be due to biofilm L. plantarum Y42 impacting ZO-1 and occludin translocation in cells [30]. Furthermore, the L. plantarum Y42 in the biofilm and planktonic states was orally administered to Balb/c female mice, respectively. The mice did not experience any negative effects as a result of the strain oral treatment, and the change of body weight was also measured daily during L. plantarum Y42 oral administration. As shown in Figure 4, the weight gains of the Balb/c mice among the three groups showed no significant difference (p > 0.05). Each experimental group of mice gained normal weight, indicating that L. plantarum Y42 is safe for animal growth. control group; PL means planktonic group; BF means biofilm group. Animal Experiments 3.5.1. Effects of Biofilm and Planktonic L. plantarum Y42 Administration on Body of Balb/c Mice Furthermore, the L. plantarum Y42 in the biofilm and planktonic states w administered to Balb/c female mice, respectively. The mice did not experience a tive effects as a result of the strain oral treatment, and the change of body weight measured daily during L. plantarum Y42 oral administration. As shown in Figu weight gains of the Balb/c mice among the three groups showed no significant d (p > 0.05). Each experimental group of mice gained normal weight, indicatin plantarum Y42 is safe for animal growth. . Effects of L. plantarum Y42 in the planktonic (PL) and biofilm (BF) states on weight gain in mice. All data are presented as mean ± SD (n = 6 mice per group). Note: Con means control group; PL means planktonic group; BF means biofilm group. Effects of Biofilm and Planktonic L. plantarum Y42 Administration on Immunity of Balb/c Mice Probiotics could increase the serum immunoglobulin contents of the host [31]. In order to evaluate the effects of L. plantarum Y42 in the biofilm and planktonic states on the immunology in the serum of mice, the contents of IgA, IgG, and IgM were measured. As shown in Figure 5A-C, the concentration of IgA in BF group was significantly higher than PL group (p < 0.05). Additionally, the IgG and IgM concentrations in the three groups showed no significant differences (p > 0.05). These results indicated that the concentrations of immunoglobulin in both groups remained relatively normal, and the addition of probiotics moderated and prevented an excessive increase in immunoglobulin. During the experiment, the levels of IL-6, IL-10, and TNF-α in the three groups were similar, indicating that the L. plantarum Y42 in biofilm and planktonic states did not cause an inflammatory response in the healthy Balb/c mice. A previous study showed that P. pentosaceus LI05 significantly reduced the serum proinflammatory cytokine levels, which is not in line with our results [32]. However, research showed that excessive amounts of proinflammatory cytokines could cause intestinal tissue injury and damage the body's immune balance [33]. An appropriate amount of proinflammatory cytokines could regulate the immune response, resist or eliminate pathogen infection, promote the repair of damaged tissues, and cause tumor cell apoptosis [34]. pentosaceus LI05 significantly reduced the serum proinflammatory cytokine levels, which is not in line with our results [32]. However, research showed that excessive amounts of proinflammatory cytokines could cause intestinal tissue injury and damage the body's immune balance [33]. An appropriate amount of proinflammatory cytokines could regulate the immune response, resist or eliminate pathogen infection, promote the repair of damaged tissues, and cause tumor cell apoptosis [34]. Effects of Biofilm and Planktonic L. plantarum Y42 Administration on the TJ Proteins Expression of Balb/c Mice Furthermore, to further verify that effects of L. plantarum Y42 in the biofilm and planktonic states on intestinal barrier function, TJ proteins expression in the Balb/c mice were measured. We found that L. plantarum Y42 in the planktonic state significantly promoted the expression of the TJ proteins ZO-1, claudin-1, and occludin in the colon of the mice (Figure 6), which was consistent with the results in vitro. The effects of probiotic in different states on intestinal barrier function has rarely been studied. Thus, the specific causes of this phenomenon are not clear and will require additional research. As discussed above, the differences in amino acid metabolic pathways and biofilm matrix of probiotic strains in the biofilm and planktonic states may account for the phenomenon. Effects of Biofilm and Planktonic L. plantarum Y42 Administration on the TJ Proteins Expression of Balb/c Mice Furthermore, to further verify that effects of L. plantarum Y42 in the biofilm and planktonic states on intestinal barrier function, TJ proteins expression in the Balb/c mice were measured. We found that L. plantarum Y42 in the planktonic state significantly promoted the expression of the TJ proteins ZO-1, claudin-1, and occludin in the colon of the mice (Figure 6), which was consistent with the results in vitro. The effects of probiotic in different states on intestinal barrier function has rarely been studied. Thus, the specific causes of this phenomenon are not clear and will require additional research. As discussed above, the differences in amino acid metabolic pathways and biofilm matrix of probiotic strains in the biofilm and planktonic states may account for the phenomenon. The microbial composition of feces in the Balb/c mice was analyzed by 16S rDNA sequencing. Firstly, the overall structural changes of intestinal flora were profiled. As shown in Figure 7A, the goods coverage score of each group was more than 99%, indicating that the depth of the sequencing provided was adequate to the subsequent bioinformatics analysis. As shown in Figure 7B, the OTUs of the PL and BF groups, shared with Figure 6. Effect of L. plantarum Y42 in the planktonic and biofilm states administration on the TJ proteins expression in mice; : p < 0.05. All data are presented as mean ± SD. Note: Con means control group; PL means planktonic group; BF means biofilm group. Effects of Biofilm and Planktonic L. plantarum Y42 Administration on the Gut Microbiota of Balb/c Mice The microbial composition of feces in the Balb/c mice was analyzed by 16S rDNA sequencing. Firstly, the overall structural changes of intestinal flora were profiled. As shown in Figure 7A, the goods coverage score of each group was more than 99%, indicating that the depth of the sequencing provided was adequate to the subsequent bioinformatics analysis. As shown in Figure 7B, the OTUs of the PL and BF groups, shared with the Con group, were 1020 and 615, respectively, in which 593 OTUs were shared by all groups. To evaluate the dissimilarity and community composition of three groups, a PCoA map was established. As shown in Figure 7C, compared with the Con group, the mice with oral administration of L. plantarum Y42 in different states formed an obvious clustering phenomenon, indicating that L. plantarum Y42 in the biofilm and planktonic states altered the communities of intestinal flora in a characteristic direction. The richness and diversity of intestinal microbiota were assessed by α-diversity analysis, including the Shannon, Simpson's, Chao1, and ACE indices. As shown in Figure 8, the α -diversity (the Simpson's, Chao1, and ACE indices) of the intestinal flora of Balb/c mice in L. plantarum Y42 intervention groups showed no significant difference among the PL and BF groups. However, the result of Shannon index was significantly decreased in the BF groups, compared to the PL group, suggesting that L. plantarum Y42 in biofilm and planktonic states could change the gut flora structure of the Balb/c mice. The richness and diversity of intestinal microbiota were assessed by α-diversity analysis, including the Shannon, Simpson's, Chao1, and ACE indices. As shown in Figure 8, the α -diversity (the Simpson's, Chao1, and ACE indices) of the intestinal flora of Balb/c mice in L. plantarum Y42 intervention groups showed no significant difference among the PL and BF groups. However, the result of Shannon index was significantly decreased in the BF groups, compared to the PL group, suggesting that L. plantarum Y42 in biofilm and planktonic states could change the gut flora structure of the Balb/c mice. Next, the average bacterial compositional profiles were summarized at the phylum and genus levels, respectively, as shown in Figure 9. At the phylum level, the dominant phyla in three groups were Firmicutes, Bacteroidetes, Proteobacteria, and Actinomycetes. As shown in Figure 9C, the relative abundance of Firmicutes in the BF group was higher than in the PL group, while Bacteroidetes in the PL group was higher. At the genus level, compared with the Con group, Lactobacillus and Helicobacter increased in the PL and BF groups. Compared with the PL group, the relative abundance of lactobacilli in the biofilm state was higher, but not significantly different. Next, the average bacterial compositional profiles were summarized at the phylum and genus levels, respectively, as shown in Figure 9. At the phylum level, the dominant phyla in three groups were Firmicutes, Bacteroidetes, Proteobacteria, and Actinomycetes. As shown in Figure 9C, the relative abundance of Firmicutes in the BF group was higher than in the PL group, while Bacteroidetes in the PL group was higher. At the genus level, compared with the Con group, Lactobacillus and Helicobacter increased in the PL and BF groups. Compared with the PL group, the relative abundance of lactobacilli in the biofilm state was higher, but not significantly different. In order to further clarify the characteristic microorganisms of the intestinal flora in the Con and experimental groups, LEfSe analysis was carried out. According to Figure 10, there are three taxa in the Con group, five taxa in the BF group, and two taxa in the PL group. In the BF group, the characteristic microorganisms were identified as Bacilli (order level) and Lactobacillales-Lactobacillaceae-Lactobacillus (order-family-genus level). In this study, Lactobacillus increased in the mice intestinal tract of the BF group, indicating that the administration of L. plantarum Y42 in the biofilm state could increase the survival of lactobacilli in gastrointestinal tract, thus increasing its growth and adhesion. We included that the slow release of bacteria inside the biofilms and higher colonization efficiency in the intestinal tract may be the reasons why L. plantarum Y42 administration in the biofilm state makes lactobacilli increased in mice gut. Probiotics have a great influence on intestinal flora. Additionally, at the genus level, the relative abundance of Helicobacter after planktonic and biofilm L. plantarum Y42 administration increased. However, Helicobacter ganmani, which is the first anaerobic species of Helicobacter, decreased significantly after the intragastric administration of probiotics, which was inconsistent with our finding [35]. These results indicated that Y42 in different states could play different effects on regulating and controlling intestinal flora. In order to further clarify the characteristic microorganisms of the intestinal flora in the Con and experimental groups, LEfSe analysis was carried out. According to Figure 10, there are three taxa in the Con group, five taxa in the BF group, and two taxa in the PL group. In the BF group, the characteristic microorganisms were identified as Bacilli (order level) and Lactobacillales-Lactobacillaceae-Lactobacillus (order-family-genus level). In this study, Lactobacillus increased in the mice intestinal tract of the BF group, indicating that the administration of L. plantarum Y42 in the biofilm state could increase the survival of lactobacilli in gastrointestinal tract, thus increasing its growth and adhesion. We included that the slow release of bacteria inside the biofilms and higher colonization efficiency in the intestinal tract may be the reasons why L. plantarum Y42 administration in the biofilm state makes lactobacilli increased in mice gut. Probiotics have a great influence on intestinal flora. Additionally, at the genus level, the relative abundance of Helicobacter after planktonic and biofilm L. plantarum Y42 administration increased. However, Helicobacter ganmani, which is the first anaerobic species of Helicobacter, decreased significantly after the intragastric administration of probiotics, which was inconsistent with our finding [35]. These results indicated that Y42 in different states could play different effects on regulating and controlling intestinal flora. The functional profiling of the microbial communities of the mice was predicted using Tax4Fun in this study. As shown in Figure 11, to understand the underlying mechanisms of the differences in the PL and BF groups, from the function prediction of Tax4Fun, the differential functions of the bacterial genes in the three groups mainly included vari- Predicted Functional Genes in the Gut Microbiota of the Balb/c Mice The functional profiling of the microbial communities of the mice was predicted using Tax4Fun in this study. As shown in Figure 11, to understand the underlying mechanisms of the differences in the PL and BF groups, from the function prediction of Tax4Fun, the differential functions of the bacterial genes in the three groups mainly included various metabolisms and transport. Most notably, the main metabolic functions are mainly the alanine, aspartate, and glutamate metabolisms, as well as the galactose metabolisms in the BF group. Additionally, it was shown that there were obvious differences in the ABC transporters, quorum sensing, and two-component system of the functions of the bacterial genes between the PL and BF groups. Our results revealed that quorum sensing was significantly higher in the planktonic-treated group than in the biofilm-treated group. Quorum sensing is a cell-to-cell communication process that depends on the extracellular signal molecules secreted by bacteria, called autoinducers [36]. Through a sophisticated intercellular communication network, these signal molecules drive the changes in gene expression and coordinate collective activity [37]. We hypothesized that bacteria in different states would stimulate quorum sensing among the microflora, in order to maintain the microflora's relative stability. We need to further study the molecular mechanism regarding interactions between the microbial communities. To further determine the relationship between the intestinal microbiome and TJ proteins, their correlation was calculated, and then the relationship was visualized as shown in Figure 12. The Spearman correlation analysis revealed the relative expression of clau- Figure 11. Heatmap of Tax4Fun functional prediction analysis for differential bacteria among the three groups. Note: PL means planktonic group; BF means biofilm group. 3.5.6. Correlation Analysis between Occludin, Claudin-1, ZO-1, and Intestinal Microbiome Diversity of the Balb/c Mice To further determine the relationship between the intestinal microbiome and TJ proteins, their correlation was calculated, and then the relationship was visualized as shown in Figure 12. The Spearman correlation analysis revealed the relative expression of claudin-1 was negatively correlated with the abundance of Bacteroides and Staphylococcus. Bacteroides are predominant human colonic commensals, which are closely related to the mucosal surface; however, it was found that Bacteroides fragilis strains can invade intestinal tissue and cause damage [38,39]. ZO-1 protein's expression was negatively correlated with Enterorhabdus and Candidatus_Saccharimonas; however, it was positively coorelated with Roseburia and Ruminiclostridium. A previous study reported that, after severe burn injury, the relative abundance of Roseburia and Ruminiclostridium increased in mice, leading to the occurrence of intestinal barrier disruption, which was not consistent with our present results [40]. Notably, the relationships between bacterial compositions and the intestinal barrier were only mathematically predicted, and additional experimental confirmation is required. Conclusions The effects of L. plantarum Y42 in the planktonic and biofilm states on the intestinal barrier function and gut flora structure of Balb/c mice were studied. We found that L. plantarum Y42 in the biofilm state showed stronger resistance to mimic gastrointestinal fluid and higher adhesion rate on the HT-29 cell monolayers than the planktonic state. L. plantarum Y42 treatment in the planktonic state clearly increased TJ protein expression, thus maintaining the integrity of the intestinal epithelial barrier on the HT-29 cell monolayers, as well as in the colon of mice. In addition, the L. plantarum Y42 in the biofilm and planktonic states showed different effects on the intestinal flora in mice. High-throughput sequencing showed that L. plantarum Y42 in the planktonic state increased the intestinal flora diversity. L. plantarum Y42 increased the proportion of lactobacilli of the intestinal Figure 12. Heatmap of Spearman correlation analysis between the top 20 abundant genera of the intestinal microbiome and TJ proteins of mice. The Spearman correlation coefficient r ranges from −0.5 to 0.5; r < 0 is negative correlation; r > 0 is positive correlation; : p < 0.05. Note: O means occludin; C means claudin-1; Z means ZO-1. Conclusions The effects of L. plantarum Y42 in the planktonic and biofilm states on the intestinal barrier function and gut flora structure of Balb/c mice were studied. We found that L. plantarum Y42 in the biofilm state showed stronger resistance to mimic gastrointestinal fluid and higher adhesion rate on the HT-29 cell monolayers than the planktonic state. L. plantarum Y42 treatment in the planktonic state clearly increased TJ protein expression, thus maintaining the integrity of the intestinal epithelial barrier on the HT-29 cell monolayers, as well as in the colon of mice. In addition, the L. plantarum Y42 in the biofilm and planktonic states showed different effects on the intestinal flora in mice. High-throughput sequencing showed that L. plantarum Y42 in the planktonic state increased the intestinal flora diversity. L. plantarum Y42 increased the proportion of lactobacilli of the intestinal flora in mice, especially the biofilm state. In conclusion, although L. plantarum Y42 in the biofilm state could increase gastrointestinal resistance ability and adhesion rates on the HT-29 cell monolayers, L. plantarum Y42 in the planktonic state could promote the expression of TJ proteins and increase the diversity of the microflora. The data in this study give the similarities and differences in the probiotic and physiological properties of L. plantarum Y42 in the biofilm and planktonic states, which may provide a new idea for the research and development of probiotics. Informed Consent Statement: Not applicable. Data Availability Statement: The data presented in this study are available on request from the corresponding author.	v3-fos-license
2019-08-15T18:08:12.900Z	2019-06-24T00:00:00.000	199616913	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "http://j.skums.ac.ir/PDF/jskums-21-125.pdf", "pdf_hash": "28bbed2b163fcb95c85cf9efa54b30282018c044", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45374", "s2fieldsofstudy": [ "Medicine" ], "sha1": "542c896d78fc32a8ad61d0da881c85b1be47858d", "year": 2019 }	pes2o/s2orc	Effect of acute aerobic exercise in different times of day on iron status and hematological factors in professional football players Background and aims: Exercise time is one of the main challenges of athletes. The purpose of this study was to investigate the effect of acute aerobic exercise at different times of the day on iron status and hematological factors in professional football players. Methods: In this quasi-experimental study, 40 professional football players were randomly divided into morning exercise, evening exercise, morning control, and evening control groups. The experimental groups performed Bruce’s test in the morning and evening. To measure iron, ferritin, red blood cell (RBC), and hemoglobin (Hb), the blood samples were taken before, immediately after, as well as 24 and 72 hours after the Bruce test. Repeated-measure ANOVA and two-way ANOVA (grouptime) were used to analyze the obtained data. Results: The results showed a significant increase immediately after an acute aerobic exercise while demonstrating a significant decrease in RBC and Hb in morning and evening exercise groups 24 and 72 hours after the exercise compared to the control groups. In addition, a significant reduction immediately after the acute aerobic exercise, whereas a significant increase 24 and 72 hours after the exercise, was observed in the serum levels of iron and ferritin in the morning and evening exercise groups compared to the control groups. As regards the serum levels of iron, an immediate significant decrease, along with a significant increase 24 hours after the acute aerobic exercise was found between the morning and evening exercise groups (P ≤ 0.05). Conclusion: It seems that performing acute endurance training in the evening is safe in terms of reducing the serum levels of iron and developing the “exercise-induced anemia”. Finally, at least 24 hours is required for the recovery of hematological parameters after acute endurance training. Introduction Physiological and hematological changes in athletes after training and competition have always been of interest to the sports researchers. In addition, the intensity, duration, and the type of exercise are factors that affect the performance, along with physiological and hematological changes in athletes (1). Training time has always been a challenge for improving performance and reducing the inflammation for the athletes. It is recognized that the best sport performances happen in the evening that are related to the maximum body temperature (2). On the other hand, iron metabolism is considered as one of the most challenging subjects in sports physiology (3). Iron is a vital mineral that plays a key role in the transfer of oxygen through hemoglobin (Hb) to tissues, the storage of oxygen in the muscle by myoglobin, and numerous processes involved in the regeneration of oxidative adenosine triphosphate. Further, the amount of iron in the tissues is one of the factors that determines the capacity of exercise in the individuals and is closely related to athlete performance (4). Football is an intense endurance-based sport that has a positive relationship with the reduction of iron levels, ferritin, and red blood cell (RBC). Several studies have focused on the effects of intense and strength training on the iron status of the body. The term "exercise-induced anemia" is often used to describe that Hb levels are close to clinical anemia (12 and 13 g of Hb per 100 mL in men and women, respectively). It is believed that this trouble is due to the intense exercise (3,5,6). It is shown that the iron status and hematological indices of football players are undergoing significant changes as a result of exercise. Sporiš et al found that after a football match, the levels of iron and ferritin of under 21-year football players of the Croatian national team significantly reduced while there were no significant differences between the levels of RBC and Hb (6). Similarly, Silva et al observed a significant increase in the amount of RBC, Hb, and hematocrit after an intense anaerobic activity session (7). Jaksimovich et al showed that the 16-year-old football players of the Serbian national team had lower levels of 15 hematological indexes including RBC, white blood cell, Hb, ferritin, and iron during the tournament season and in the training sessions compared to non-athlete peers (8). It also seems that the duration, intensity, and the time of the activity and hence their effects on hematological parameters are important. Mohammad et al concluded that the iron surfaces, ferritin, and the RBC of professional Egyptian runners have an opposite relationship with the distance traveled by such runners (5). Based on the results of previous studies, iron seems to have a positive effect on sports activities and the performance of endurance activities such as football and delay fatigue (9,10). Obviously, RBC damage in the blood vessels of the leg is the reason for Hb and hematocrit decrease in athletes when they collide with the ground and damage the RBC in the capillary gastro-intestinal tract during exercise, especially long-term endurance exercise which may reduce the Hb concentration of the athletes (4). On the other hand, daytime rhythm regulates a large number of the cells, hormones, the cytokines of body, and the like, as well as the physiological functions of the cardiovascular system and blood. Therefore, more acute cardiovascular events occur in the morning (11). The philosophy of athletic training sessions in the morning and evening and at different times of the day is the reason for the importance of day-night rhythm. However, there is still no general agreement on the best exercise time. It is shown that blood pressure and heart beat rate demonstrate a day-night rhythm. In addition, the pattern of the day-night rhythm indicated the most severe cardiovascular events in the morning with regard to body temperature, blood pressure, blood fluidity, and fibrinogen changes (9,12). Contradictory results are available about the effect of exercise time on hematological indices. Based on the result of previous research, plasma fibrinogen levels increased significantly after the morning exercise, but only a slight increase of 2% was observed in the afternoon (13). Further, it was found that the implementation of a session of maximal aerobic exercise in the morning and evening could have an effect on hematological factors, but the exercise time was unaffected by the changes (1). Furthermore, it was reported that the performance of professional women swimmers in the 100-meter was influenced by the time of the day and the subjects recorded a better time in the evening (2). Considering the above-mentioned issues and the importance of sports activities, especially the intense activities in the performance of professional athletes, the present research aimed to find whether there is a significant difference in the iron status and hematological factors of professional football players immediately after and 24 and 72 hours after acute aerobic activity in the morning and evening. Methods This study is a quasi-experimental single-blinded research with a repeated measure design. A total of 40 football players from Khuzestan clubs participated in this study. A nonprobability convenience sampling technique was used to collect the subjects. Moreover, they were selected from the same clubs and based on the same training, the lack of supplementation of vitamin C, iron, and B-group vitamins, adult group with no injuries and footballing history of at least 5 years. Regarding physiological differences, all subjects were selected from among the players, while no goalkeepers were used. The players came to visit exercise physiology lab and the Faculty of Physical Education in the Shahid Chamran University of Ahvaz (Iran) in order to become familiar with its environment. After they signed the letter of satisfaction for participation, anthropometric measurements including height, weight, and body mass index by bio-electrical impedance, and VO2 max by Bruce test were conducted on a treadmill the Olympia 3/3, Javern South Korea (14). The subjects were assured that all their information would remain confidential. Additionally, the subjects were randomly assigned to four groups of morning exercises, morning control, evening exercise, and evening control (each containing =10 players). First, the football players were divided according to the defender, midfielder, or striker position and then randomly and equally from each post in each of the groups. The morning and evening exercise groups were performed the Bruce test respectively in the morning (8-10 am) and evening (16-18 pm). Blood samples were taken before, immediately after, as well as 24 and 72 hours after the Bruce test in order to measure the serum iron levels, serum ferritin, RBC, and Hb. Then, the samples were poured into the EDTA tubes (ethylene diamine tetra acetic acid) and transferred to the laboratory immediately after the completion of blood sampling. RBC and Hb in this study were measured by using an automatic counting machine (Mindray-BC 5300 Auto Hematology Analyzer). Next, the samples were centrifuged at a rate of 3500 rpm for 10 minutes at 4°C to isolate the serum, followed by the extraction of the serum in special microtubes and storage at a temperature of 70°C until further measurement. Serum iron and ferritin were measured by ELISA method using a microplate reader (Hiperion Company, Germany) and by special kits (serum iron kit: Greiner manufactured in Germany, as well as ferritin kit: Pioneer Iranian Medicine Company). A 24-hour food questionnaire was developed to provide nutritional information to the subjects in order to record the food that they consume during the three days before and after the main test. The questionnaires were then analyzed by Nutritionist 4 software (15). For statistical analysis, the mean and standard deviation were used as descriptive statistics. Then, Shapiro-Wilkes test and Leven test were used to evaluate the normality of the data and the equality of variances, respectively. In addition, oneway ANOVA was utilized to compare the dietary data of all groups. Further, repeated-measures and two-way ANOVA (grouptime) tests were applied for intra-group and between-group comparisons in each stage of the study. When the ANOVA test was significant, a post-hoc test was used for pairwise comparison. Data were analyzed by SPSS software, version 21. The significance level was considered to be P<0.05. Results The mean and standard deviation of physiological and anthropometric characteristics of the subjects are presented in Table 1. Furthermore, Table 2 shows the mean and standard deviation of the energy intake and nutrients status of the subjects in the four groups from 72 hours before and after the main test. Based on the results, there was no significant difference in the mean daily energy intake, the percentage of protein, carbohydrate, fat, fiber, vitamin C, vitamin B12, calcium, and iron among the four groups (P ≥ 0.05). Similarly, Table 3 represents the levels of RBC, Hb, and the serum levels of iron and ferritin in four groups before and immediately after 24 and 72 hours after acute aerobic exercise. The results of two-way ANOVA demonstrated a significant difference regarding the effect of the interaction of grouptime only on the serum levels of iron (P = 0.041). Moreover, the results of ANOVA and Bonferroni post-hoc tests showed a significant increase in the RBC and Hb in the morning exercise group compared to the morning control group (P = 0.001 and P = 0.009, respectively) and evening exercise group compared to the evening control group (P = 0.005 and P = 0.013, respectively). However, a significant decrease was observed in the serum levels of iron and ferritin in the morning training group compared to morning control (P = 0.008 and P = 0.011, respectively) and evening exercise groups compared to the evening control group (P = 0.003 and P = 0.001, respectively) immediately after the acute aerobic exercise. Additionally, there was a significant reduction immediately after, but a significant increase 24 hours after the acute aerobic exercise in the serum levels of iron in the morning exercise group compared to the evening exercise group (P = 0.039, P = 0.043, respectively). As a result, comparing immediately and 24 hours after acute aerobic exercise, there was a significant increase in the serum levels of iron and ferritin in the morning exercise group versus morning control group (P = 0.022 and P = 0.023, respectively) and the evening exercise group compared to the control evening group (P = 0.007 and P = 0.005, respectively). This increase in the serum levels of iron in the exercise group was significant in the morning exercise group (P = 0.039). In addition, comparing immediately and 72 hours after the acute aerobic exercise, a significant reduction was found in the RBC and Hb concentration in the morning exercise group compared to morning control group (P = 0.02 and P = 0.012, respectively), as well as the evening exercise group compared to the evening control group (P = 0.013 and P = 0.0018, respectively). However, the serum levels of iron and ferritin demonstrated a significant increase in the morning exercise group compared to the morning control group (P = 0.013 and P = 0.021, respectively), as well as the evening exercise group compared to the evening control group (P = 0.002 and P = 0.001, respectively). Discussion Iron metabolism responded to sports activity as a challenge in the field of sports physiology and yielded contradictory results. The reduction, increase, and the stability of the iron levels are reported as a result of exercise activity (3). Further, the iron status and hematological parameters as a result of exercise activity depend on several factors such as the amount of iron excretion (16), RBC hemolysis (17,18), intestinal absorption (19), andiron retention (20), along with the change and conversion of erythrocytes and Hb (21) or inflammation (22). A decrease in the levels of iron, ferritin, and RBC in intense exercise activity can be attributed to the hemolysis due to the stroke, which is related to the intensity of the exercise and the increase in temperature by 6%-11% excretion through sweating (3). Furthermore, the reduction of iron levels resulted from extreme sports activities is possibly due to a decline in the transfer of bivalent metal (Divalent metal transporter-1), Ferroportin 1, the carrier protein (Haem carrier protein-1), hephaestin, and ceruloplasmin (9,19). The results of this study showed that RBC and Hb levels in the morning and evening exercise groups were significantly higher than those in the control groups after the aerobic activity, while the iron and ferritin levels decreased significantly. Moreover, the reduction in the serum levels of the iron in the evening training group was significant compared to the morning training group. Regardless of the exercise time, a significant increase was observed in Hb and RBC in both training groups. Our research results are consistent with those of many studies on hematological profiles. Significant increases in Hb and RBC in teenage Serbian football players were reported in relation to non-athlete peers (8). Additionally, in line with the current study, Mohamed et al, in their study on female professional runners in Egypt (short distance runners (n=6), middle distance runners (n=5), and long distance runners (n=3)), found an inverse proportional relationship between iron and ferritin levels and the distance of running among elite runners so that the iron levels decrease by increasing the running distance (5). In addition, Sporiš et al showed that the levels of iron and ferritin of men football players in the under-21 Croatian team significantly decreased after a football match. They also reported a non-significant reduction in RBC and Hb Note. The results of repeated-measure ANOVA test for intra-group comparisons and one-way ANOVA for between-group comparisons at each stage of the test and two-way ANOVA (grouptime) for the interaction of time and group. The significance level was considered to be P < 0.05. a Significant difference between before exercise and immediately after the exercise. b Significant difference between 24 hours and immediately after the exercise. c Significant difference between 72 hours and immediately after the exercise. d Significant difference between morning exercise and morning control group. e Significant difference between evening exercise and evening control group. f Significant difference between the morning exercise group and evening exercise group. levels (6). Regarding the effect of exercise time, the result of a study revealed that a session of intense aerobic activity in the morning and evening increased the RBC, white blood cells, and Hb, but no difference was observed in this regard between the two times (1). On the other hand, fibrinogen levels are found to increase after the endurance and resistance training in the morning. Therefore, from the point of view of cardiovascular health, exercise is safer in the evening (3). Based on the available information, exercise activity causes erythrocytosis, which increases the concentration of erythrocytes to 25%. These changes are initially justified by the call for blood supply because the blood stores a large number of cells, but less plasma compared to the circulating blood (23). Iron is also used in the bone marrow, liver, and spleen, and ferritin is utilized to reconstruct Hb and RBC, to maintain high levels of RBC and Hb, and to transport the oxygen to the tissues (3). Therefore, given the intensity of exercise, the increase in Hb and RBC and the reduction in iron and ferritin in the present study are such that the increase in temperature, sweating, inflammation, decreased plasma volume, and the destruction of the blood cells can decrease the levels of hematological indices and are used to maintain the levels of Hb and RBC, iron, and ferritin in the liver and spleen (23). The results of this study further showed that the iron levels and serum ferritin in the morning and evening exercises group increased significantly after 24 hours of recovery following the exercise activity in comparison with the control group. Furthermore, the serum iron levels in the evening exercise group increased significantly compared to the morning exercise group. Conversely, 72 hours of recovery after the acute aerobic activity significantly decreased the levels of Hb and RBC while a significant increase was observed in serum iron and ferritin levels in the morning and evening exercise groups compared to those in the morning and evening control groups. There is not much research in the field of changes in hematology factors and iron indices during recovery. Accordingly, 2 hours of recovery after the acute aerobic exercise was reported to have no significant effect on Hb and RBC of the athlete men (1). However, no significant changes in fibrinogen levels and high sensitivity C-reactive protein were found in active men three hours after the endurance and resistance exercise (3). Therefore, it seems that more time is needed for recovery and changes in hematological factors after the exercise, especially severe physical activity and the levels of blood factors and plasma volume re suggested to return to their initial level after 48-72 hours of acute aerobic exercise (24). Sweating is argued to be one of the cases that reduce the iron levels during intense activities and excrete between 6% and 11% of body iron (25). Iron retention is regarded as another iron-depleting cause after the exercise, which is called "exercise-induced anemia". An increase in plasma volume due to intracellular fluid transfer to the vessel and the reduction of Hb concentration are believed to lead to such anemia. Predictably, the increase in plasma volume that occurs as a result of the training is initiated 48-72 hours after the exercise (24). In the present study, iron and ferritin levels increased after 24 and 72 hours of recovery. Based on the results of a study, an exercise in the morning caused a greater reduction in the serum levels compared to the evening exercise, which could be due to higher temperatures and consequently sweating (26). Contrarily, after 24 hours of recovery, a greater increase in the serum iron of the morning training group was observed compared to the evening exercise group, which is likely to be in line with other changes as a result of alterations in plasma volume (20). In our study, a decrease in Hb and RBC was observed 72 hours after the acute aerobic exercise. However, a significant increase was found in RBC and Hb levels during acute aerobic activity as well. At first, the mechanism of this increase and decrease probably demonstrates negative feedback during the intensive training, which is due to the need for more oxygen. Moreover, the synthesis of Hb and RBC increases while the synthesis of RBC cells decreases as the level of erythropoietin secretion decreases (27). Therefore, iron intake for the synthesis of Hb is considered as one of the reasons for a reduction in the serum levels of iron and ferritin. In the recovery period, the levels of Hb and RBC decrease and reach their initial levels with changes in plasma volume, the levels of erythropoietin, and the oxygen demand (9). Conclusion Regarding the results of this study, it seems that performing the exercise activity in the evening is important in terms of the reduction of serum iron levels and the occurrence of "exercise-induced anemia". On the other hand, according to the results of the present study, professional football players require at least 24-72 hours to recover after intensive endurance exercises in order to restore safe levels of hematological parameters.	v3-fos-license
2020-03-13T14:43:32.525Z	2020-03-13T00:00:00.000	212681194	{ "extfieldsofstudy": [ "Medicine", "Biology" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/s41467-020-15223-x.pdf", "pdf_hash": "a0a507e5c918340a718e0049b879070121a7735c", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45377", "s2fieldsofstudy": [ "Biology" ], "sha1": "c40931816db0f3c985d58b6879b9344d56970aa1", "year": 2020 }	pes2o/s2orc	Multi-level and lineage-specific interactomes of the Hox transcription factor Ubx contribute to its functional specificity Transcription factors (TFs) control cell fates by precisely orchestrating gene expression. However, how individual TFs promote transcriptional diversity remains unclear. Here, we use the Hox TF Ultrabithorax (Ubx) as a model to explore how a single TF specifies multiple cell types. Using proximity-dependent Biotin IDentification in Drosophila, we identify Ubx interactomes in three embryonic tissues. We find that Ubx interacts with largely non-overlapping sets of proteins with few having tissue-specific RNA expression. Instead most interactors are active in many cell types, controlling gene expression from chromatin regulation to the initiation of translation. Genetic interaction assays in vivo confirm that they act strictly lineage- and process-specific. Thus, functional specificity of Ubx seems to play out at several regulatory levels and to result from the controlled restriction of the interaction potential by the cellular environment. Thereby, it challenges long-standing assumptions such as differential RNA expression as determinant for protein complexes. Overall this is a massive paper that (i) established a new method for sensitive capture of protein partners in Drosophila and (ii) proposed a new model based on robust data for understanding Hox tissue-specificity in vivo. In particular, the role of few tissue-specific cofactors for making a link between Hox proteins and other general cofactors is a very appealing model that could likely apply to other major developmental regulators. In conclusion, both the experimental approach and outreached results make the work of Carnesecchi et al. well suitable for Nature Communication. Several major and minor points should however be considered by the authors before acceptance for publication. Major points: -Because of the huge amount of data it is important to make it more accessible. For example the information on how interactions were considered as biologically relevant (found in two replicates at least) should be provided in the main text or a main figure legend (for instance this information appears only in the M&M section). With this regard what is the proportion of interactions found only once? Are some cofactors found in three or even four replicates? If so, could these interactions be considered as more biologically relevant? What about the nlsBirA* construct: did this construct lead to more unique or more common interactions in replicates when compared to Ubx constructs? In sum, it will be interesting to know whether a ranking of interactions could be done from the BioID purification itself. -Provide a table with only the cofactors considered as biologically relevant for Ubx and Ubx51. -One key control protein is Ubx51, which does not bind DNA. Unfortunately, control experiments with Ubx51 are not really convincing. For example, in Supp. Fig.1c, there is clearly less Ubx51 than Ubx in the input, making the interpretation of Exd-coIP difficult. Not surprisingly, it seems that Ubx51 is systematically less biotinylated than Ubx from immunostaining in the embryo. How variable is it? Is Ubx51 systematically less biotinylated than Ubx? Along the same line, authors compared interaction properties of specific cofactors with Ubx and Ubx51. They concluded that CtBP and Tin preferentially interacted with Ubx, but the level of Ubx51 is less than Ubx in the input (Fig. 4a). -With regard to the validation of interactions in S2 cells, authors should provide a table recapitulating their observations. For example when considering Tin, how many times it was found in the replicates of Ubx and Ubx51, and how strong was the interaction when tested in S2 cells? This should be done systematically for the individual tested cofactors. All westerns could potentially be shown in Supplementary for better clarity in the main figures and Tables. In line with this kind of experiment, testing 2/3 negative cofactors from the BioID purification (never found or found only once) should be considered to further validate the relevance of the approach. -In the search for enriched binding sites in the mesoderm, the sequence provided for Exd is not convincing (quite degenerate). Why authors did not look more precisely at consensus Hox/Exd binding sites (it will make more sense with regard to the role of Exd)? Are there enriched Hox/Exd binding sites? -The molecular dissection of Ubx/Tin interaction with several truncated/mutated forms was performed with GST-pull down assays, which are DNA-binding independent. Given the DNA-binding dependency of this interaction, these assays should be repeated in the presence of DNA (classical EMSAs), with a probe derived from either the Dpp enhancer or the consensus Hox/Tin binding site. Minor points: -There is a repetition of "Intriguingly". Authors should avoid this term when not necessary (which is the case in several instances). -The last sentence of the summary is a bit too much over-extrapolated. I suggest to remove or to attenuate this conclusion. First, authors provided evidence that tissue-specific cofactors like Tin or Grh are important for mediating tissue-specific interactions with more ubiquitously expressed proteins. Second, their analysis rather showed that there are tissue-specific TF-enhancer interactions that correlate with tissue-specific expression (especially in the mesoderm). -Supp. Fig1b: the interaction with Med19 is not really convincing. Authors could focus on the interaction with Exd only (considering the previously mentioned caution with Ubx51). -The information on partners of Ubx based on their expression profile from previous work (Domsch et al.) should be provided more precisely, with values, in a supplementary table. This information is important given the central message of the paper. -It would be informative to have the heatmap of Ubx51 and get more easily accessible (digested) information about the common and specific interactions between Ubx and Ubx51 (as a supplementary table). For example, are DNA-binding interactions more lineage-restricted or not? - Fig. 8 is not really illustrative of the main message of the paper: the information of how Ubx could reach tissue-specificity with general cofactors is missing. It will be more useful to have a final figure focusing on this main message of the paper. Reviewer #2 (Remarks to the Author): The article entitled "Multi-level and cell type specificity of protein interactome assembly by a Hox Transcription factor" by Carnesecchi et al., aimed to characterize various interactomes of a nuclear transcription factor, Ubx, in multiple drosophila embryo cell types to elucidate the dependency of various protein interactions on Ubx's specific function within each specific cell type. The group utilized a cell-type directed in vivo proximity labeling BioID approach coupled with high resolution mass spectrometry to identify and quantitate these various Ubx interactomes and followed up on this discovery work with ample validation of interesting findings. Overall, this is a well controlled and well thought out experimental design which accounts for nonspecific interaction discovery data; in the sense of non-specific protein biotinylation (nucleus directed GFP-BioID as a background control) and functional specificity of Ubx (WT Ubx vs mutant that abolished binding to chromatin). There are, however, several significant concerns about thoroughness of the proteomics discovery data in both how it is presented in the manuscript/supplemental data and the transparency of the resulting protein identification and quantitation metrics. Comments; 1. The experimental detail included in the "Mass Spectrometry Preparation" sections in Materials and Methods needs to be drastically expanded. Specifically, very little detail is given on how the mass spectrometer was operated (scan settings, resolution, duty cycles, agc settings, etc.) to acquire the data. Unfortunately, in the field of proteomics there is often a lack of transparency on these details which could have profound consequences on experimental replication of these findings in another laboratory. 2. In the same section, please add solvent systems (mobile phase A/B composition; acetonitrile with 0.1% formic acid?) in the description of the LC separation. 3. The description of the analytical UPLC column (Acclaim PepMap RSLC) should not read "75 um x 2um" as its dimensions, but rather 75 um x 50 cm UPLC column packed with 2 um Acclaim PepMap RSLC particles. 4. Although the authors list the proteins identified, there are critical pieces of information not readily accessible to the reader (including supplemental data) including; A) number of unique peptides identified to each protein, B) consistency of expression across biological replicates (easily expressed as intragroup coefficient of variations), C) enrichment values vs background (as fold change or log2 fold change) or p-value. 5. No AUC or peak intensity quantitative data is presented in the manuscript (including supplemental data). It is nearly impossible to differentiate the interaction confidence of any of the presented interactors with the information presented. No pvalues/fold changes nor any type of technical vs biological variation data is presented. The authors did upload MaxQuant output files (zipped files @ 60Gb, much larger unzipped), however it would take significant knowledge of that program, expertise in reading this file structure, and additional software to be able to calculate those values from the raw data uploaded. This is beyond what the scientific community should be responsible to do. Quant values need to be presented in table form for at least those species of interest with associated p-values. 6. Please justify the criteria that proteins significantly enriched in only 2 of the 4 replicates was sufficient to be considered biologically relevant. Depending on that particular proteins intragroup technical+biological variation, this 50% criteria could easily be an artifact from aberrant signals. 7. Please describe the roll up method used for going from the peptide level measurements and areaunder-the-curve calculations to the protein level. This manuscript addresses the central issue of context-dependency in transcription factor function. This issue is particularly puzzling for the Hox homeotic genes, which act like GPS coordinate cues, instructing all cells in a segment regarding positional identity. This means that Hox genes act in a whole range of tissues/cells within their segments, governing a number of developmental outcomes, unique to each tissue/cell type, and unique to that region. How Hox genes can act in this contextdependent manner is largely unknown. The current study combines the BioID technology, the powerful molecular genetic tools available in Drosophila, and the in-depth knowledge of Hox function in the developing embryo, to identify protein interaction partners for the well-studied Hox factor Ubx. By cleverly using both a wild type and a DNA-binding Ubx construct, they can furthermore discriminate between protein interactions occurring on, versus off, the DNA. They also compare the protein interaction landscape for three tissues; mesoderm, neuroectoderm, and nervous system, and intriguingly find that the interaction network differs between the three tissues. They go on to focus on a subset of interaction partners, and confirm the identified protein interactions by other methods, as well as identify mutant phenotypes that are in line with the differential Ubx interactomes. These findings represent a major leap forward in our understanding of TF context-dependency, and will be of interest to a whole range of scientists interested in developmental biology, gene regulation and context-dependent gene function. However, the study is not without faults, and there are a number of things can could help improve it. Major issues: 1) The main weakness of the study pertains to Ubx function in the CNS. This refers both to the actual experiments conducted, and to the minimalistic description of Ubx function in the CNS. First, they only mention one role of Ubx; the control of PCD in subsets of motor neurons. However, they do not actually use this as readout. Instead, they assess motor axon pathfinding (Supplemental Figure 8), but do not describe the previously published phenotype, and only show low-resolution images that are difficult to interpret. Moreover, they claim that there is no motor axon phenotype in Ubx/Ubx, but a recent publication indeed revealed a motor axon projection defect in Ubx/Ubx mutants (Hessinger et al 2017; they already reference this paper, but for a different purpose). Against this backdrop, it is difficult to evaluate the phenotypes, or lack thereof, of the other genes studied. Second, other known roles for Ubx in the CNS are very similar to the known roles of Grh, and since they do find Grh as a partner, and go on to specifically study this particular interaction to some extent, these particular roles of Ubx (and Grh) would be worth emphasizing. Specifically, while Hox genes, based upon in situ hybridization, do express in the early neuroectoderm, Hox proteins are not initially expressed in NBs (PMIDs 28392108). This is due to the absence of Elav protein, which is only expressed in neurons and stabilizes Ubx mRNA (PMID 24803653). In fact, the lack of Elav protein in glia also explains the lack of Hox protein expression in glia (PMID 15537690), something that can be mitigated by ectopic expression of Elav (PMID 24803653). Moreover, Hox expression is also repressed in NBs by early NB factors (PMID 29112852). Relatedly, Grh is the last gene in the temporal gene cascade (reviewed in PMIDs 24400340, 23962839) and thus expressed late in NBs. Hence, Ubx and Grh become co-expressed late in NBs, as an effect of the aforementioned three regulatory mechanisms. In late NBs they regulate three biological events: the Type I->0 daughter proliferation switch (PMIDs 25073156, 28392108), NB cell cycle exit (PMIDs 9651493, 28392108, 20485487), and NB PCD (PMIDs 28392108, 20485487, 16049114). Predicted direct targets for the first two events are dap, CycE, E2f1 and stg, while the RHG family (rpr, hid, grm, sickle) are likely targets for the PCD. The finding of a physical interaction between Ubx and Grh is very interesting, but there is no functional follow-up regarding what this could mean inside the CNS, in spite of a substantial body of work regarding their similar roles in the CNS. 2) The finding that Ubx mostly interacts with ubiquitously or rather broadly expressed proteins raises the issue of sensitivity. Would Ubx-interacting partners that are expressed only by small sub-sets of cells actually be detected by BioID, and what is the limit? The cell culture system allows them to determine the sensitivity of the method, by mixing in different percentages of mBUbx/Exd coexpressing cells with mBUbx-only expressing cells (or untransfected control cells), and scoring for biotinylation of Exd by mass spectroscopy. Alternatively, they could drive mBUbx with a more restricted driver(s) and assess the BioID profile. 3) Figure 3a: They show the interaction network for Ubx partners in the mesoderm (twi-Gal4). It would be important to also show the same networks for sca-Gal4 and elav-Gal4. Figure 5: It would be important to include, in the supplement, gene lists of the genes that are: cobound by Ubx-Tin and by Ubx-Grh (5a); overlap and differ in Ubx-Tin versus Ubx-Grh binding (5d); overlap in transcriptome versus DNA-binding (5e). Minor issues: 5) It is a common misconception, but elav-Gal4 is not a pan-neuronal driver. While the Elav protein is only present in neurons, the elav-Gal4 driver drives robustly in neuroblasts, neurons and glia, starting from St11 and onward (PMID 17994541). So it is a pan-neural driver. 6) Page 13: I am surprised that only 18% of genes bound by Ubx-Tin in the mesoderm were affected in Ubx mesodermal mutants? Also, this statement does not refer to any specific figure/table. 7) Figure S2d; missing "myc" labels. 8) Please number the pages and rows. 9) Figure 3d-g' and 4b: I found it difficult to assess overlap. I suggest also showing single immunostained images. POINT-BY-POINT RESPONSE TO THE REVIEWERS: We would like to thank all the reviewers for their constructive criticism and important suggestions, which have helped to improve the manuscript significantly. We have included new data and have changed some aspects according to reviewer suggestions, which we describe in detail below. Reviewer #1: thought the study is well suited for Nature Communication, as it establishes a new method for sensitive capture of protein partners in Drosophila and proposes a new model based on robust data for understanding Hox tissue-specificity in vivo. In particular, the reviewer thought that the role of few tissue-specific cofactors for making a link between Hox proteins and other general cofactors is a very appealing model that could likely apply to other major developmental regulators. There were a few concerns raised by this reviewer, which we addressed in the following way: Because of the huge amount of data, it is important to make it more accessible. For example, the information on how interactions were considered as biologically relevant (found in two replicates at least) should be provided in the main text or a main figure legend (for instance this information appears only in the M&M section). With this regard what is the proportion of interactions found only once? Are some cofactors found in three or even four replicates? If so, could these interactions be considered as more biologically relevant? What about the nlsBirA construct: did this construct lead to more unique or more common interactions in replicates when compared to Ubx constructs? In sum, it will be interesting to know whether a ranking of interactions could be done from the BioID purification itself. We thank the reviewer for his interest in the method and to help us providing a more in depth and clearer view of the data. 1) The detail of the calculation was added in the legend of Figure 2 and the detail of the replicates added in the main text. 2) Of note, as developed in the response to reviewer #2, the Supplementary Tables 1, 2, 3 summarizing, for each tissue and replicate, the unique peptide number, LFQ value, the table of the calculation of log2-LFQ ratio and the summary of the confidence interval calculation (Supplementary Tables 4,5,6) are also now included in order to provide a better accessibility to the data processing and pipeline for reader from different fields of research. The lists of the proteins significantly enriched in all replicates are provided in Supplementary Tables 7, 8, 9. 3) The proportion of the proteins/partners found in common between replicates of each tissue are now provided in Supplementary Tables 7, 8, 9 as well as illustrated by Venn diagram for each tissue, WT and N51A Ubx proteins. In addition, we provide now the list and related Venn diagram in Supplementary Table 10 of the protein found in common between each tissue and identified at least in one replicate. This further re-enforces the specificity and unique signature of the tissue-specific interactome but also provide further information for future study on these interaction partners. 4) The reviewer also asked if the interaction found in more than 2 replicates might be more relevant, if it could be used for "ranking" the partners (Supplementary Tables 7,8,9). We do not believe that it is the case, as this ranking might be dependent on the inherent variation of protein biotinylation between embryos (that we normalized by pooling), and the intrinsic stochasticity of the MS-process. Another argument is that, interaction and ranking might not be always correlated with unique function. As an example, Tin was enriched in only 2/4 replicates. In contrast, Tin is a strong interactor of Ubx, as revealed by interaction studies (co-IP, GST-pull down) and highly relevant biologically as shown by genetic interaction and functional assay, confirming the difficulty to apply a ranking based on replicate fishing. That is why we chose to rely on a stringent selection based on replicate enrichment combined with the validation with other approach (biochemical and functional). Hypothesis might be proposed for the one enriched in more than 2 replicates, though taking into account that it could be biological variability instead of true partner-ranking. 5) The reviewer further mentioned the nls-BirA* construct. We are actually not sure what the reviewer is asking precisely. But maybe this explanation can clarify the situation. In this study, we did not use a nls-BirA* but the BirAnls-GFP fusion construct. We used this construct as a general control to discriminate between proteins being biotinylated due to Ubx (as proteins are brought in close proximity to BirA thanks to Ubx) or whether proteins are biotinylated because they come in close proximity to BirA* for reasons unrelated to Ubx, as all proteins are translated and nuclear proteins are imported into the nucleus etc. Thus, we use BirAnls-GFP as a general control to identify those that specifically interact with Ubx in the different tissues, due to its size similar to BirA-Ubx (more than the nls-BirA* alone). Provide a table with only the cofactors considered as biologically relevant for Ubx and Ubx51. We thank the reviewer for highlighting this essential dataset. The table was already provided in the word file, it is now added as excel file as Supplementary Table 12 for more transparency. One key control protein is Ubx51, which does not bind DNA. Unfortunately, control experiments with Ubx51 are not really convincing. For example, in Supp. Fig.1c, there is clearly less Ubx51 than Ubx in the input, making the interpretation of Exd-coIP difficult. Not surprisingly, it seems that Ubx51 is systematically less biotinylated than Ubx from immunostaining in the embryo. How variable is it? Is Ubx51 systematically less biotinylated than Ubx? Along the same line, authors compared interaction properties of specific cofactors with Ubx and Ubx51. They concluded that CtBP and Tin preferentially interacted with Ubx, but the level of Ubx51 is less than Ubx in the input (Fig. 4a). 1) We thank the reviewer for his/her careful reading and highlighting an important point. We would like to stress that the Figure 1c is a BioID (not a co-IP) and the AP fraction is containing biotinylated proteins. The level of proteins is thus due to biotinylation level, meaning the number and accessibility of the lysine residue. For example, GFP is less biotinylated itself, probably due to less lysine residues, less accessible. Thus, we did not compare the BirA* fused protein itself (in term of auto-biotinylation) as it will be different due to the intrinsic nature of the protein. In contrast, BirA* has the same activity, indicating that a close-proximity partner will be biotinylated at the same level if it is in close proximity with the same frequency (and distance). Thus, we can compare the efficiency of biotinylation of a same protein/partner with the different BirA fused proteins (and not the constructs themselves). 2) In line, we performed another BioID-control experiment with adjusted level of BirA-fused proteins. We also now provided the related quantification of Exd-enrichment (normalized to the expression level given by the input). The AP-ratio of flag-Exd with N51A/WT normalized to the basal expression flag-Exd is N51A:WT=0.3:1. Of note, we observed a slightly slower expression of UbxN51A compared to UbxWT (0.8:1) as the reviewer also noticed. However, the differential expression level of N51A and WT Ubx constructs cannot account for the difference of enrichment of Exd in the AP fraction N51A compared to WT (0.3:1). Figure 1e (BioID1 vs BioID2) confirmed a similar expression level between Ubx N51A and WT. 4) This slight differential expression between the constructs are also the reason why we chose to compare the calculation over the control of each of them WT/GFP and N51A/GFP and not WT/N51A, as it might insert a bias due to biotinylation level/protein level. We then compared the presence or absence of partners in the one found at least in two replicate in WT/GFP on a side and N51A/GFP on the other side. We chose not to apply a ranking that might not be relevant biologically taking all these bullet points into account. 5) In contrast to the BioID, the coIP experiments may give a better ranking, even though it is based not on frequency and distance but strength & stability of the interaction. In line, co-IP experiments shown in Supplementary Figure 6f, and now quantified in summary Supplementary Table 13, revealed a similar enrichment of Exd with Ubx WT or N51A. It strongly suggests that, more than the interaction-strength, the interaction frequency of Exd-UbxWT is higher than the one of Exd-UbxN51A, opening a large avenue for studying the dynamic of the Exd/Hox complex in the future. With regard to the validation of interactions in S2 cells, authors should provide a table recapitulating their observations. For example, when considering Tin, how many times it was found in the replicates of Ubx and Ubx51, and how strong was the interaction when tested in S2 cells? This should be done systematically for the individual tested cofactors. All westerns could potentially be shown in Supplementary for better clarity in the main figures and Tables. In line with this kind of experiment, testing 2/3 negative cofactors from the BioID purification (never found or found only once) should be considered to further validate the relevance of the approach. We thank the reviewer for her/his thoughtful advices for improving the clarity of the manuscript. As requested, we provided now a more comprehensive table in Supplementary Table 13 containing the different partners, positive and negative controls tested. We feel that the provided table clearly shows that we already tested a set of positive (Exd, Med19, M1BP) and negative (Ncm, Tubulin, Pc) control, whether in cells or at the endogenous level that validate the whole approach. Moreover, as suggested, coIP-related immunoblots are now grouped in the Supplementary Figure 6. In the search for enriched binding sites in the mesoderm, the sequence provided for Exd is not convincing (quite degenerate). Why authors did not look more precisely at consensus Hox/Exd binding sites (it will make more sense with regard to the role of Exd)? Are there enriched Hox/Exd binding sites? We deeply appreciate the reviewer's interest and careful reading of the different analyses performed in the study, from BioID to genomic analysis. Indeed, the Exd binding site provided is quite degenerate, as it is the result of an unbiased search of enriched motifs using MEME software. From MEME, as mentioned, the search is not biased. Thus, there is not a directed search specifically for Ubx/Exd and if the site is not present, it is just revealing that it is not enriched primarily. However, it also intrigued us; therefore, we searched for enrichment of Hox/Exd high affinity site using the Selex-seq matrix provided by the work of the lab of Richard Mann: dataset adj_p-value Meso-Ubx 3.87E-02 Ubx-Tin 1.00E+00 Neuro-Ubx 1.00E+00 Ubx-Grh 1.00E+00 The analysis revealed a significant enrichment of the Hox/Exd sites only in the mesodermspecific ChIP-seq of Ubx (and the p-value is rather low). None of the common Ubx-Tin, Ubx-Grh as well as neural-specific ChIP-seq of Ubx is enriched for this optimal Hox/Exd binding site. It opens a diversity of hypothesis concerning: 1, the requirement of Exd/Hox complex to regulate these target genes; 2, the biological requirement of this optimal site for driving Ubx transcriptional function in vivo. The molecular dissection of Ubx/Tin interaction with several truncated/mutated forms was performed with GST-pull down assays, which are DNA-binding independent. Given the DNAbinding dependency of this interaction, these assays should be repeated in the presence of DNA (classical EMSAs), with a probe derived from either the Dpp enhancer or the consensus Hox/Tin binding site. 1) We thank the reviewer for her/his interest to the molecular dissection of Ubx/Tin interaction. Importantly, we clarified that the interaction is not DNA dependent (interaction in vitro without DNA) but the functional cooperation is (Luciferase assay, Genetic interaction). The question of DNA-binding cooperation is indeed relevant and we now provide as Supplementary Figure 7e the EMSA on the dpp enhancer. This experiment re-enforces the fact that: 1), they both directly bind the dpp enhancer, in a range of 100bp, 2), they are able to bind both in complex illustrated by the supershift of the 'Ubx+Tin' band +V5/MBP antibody, (most probably too heavy to enter the gel), and independently, illustrated by the single binding. 2) Of note, the first data did not show a DNA-dependency of the interaction as GST-pull down experiments showed a direct interaction between Ubx and Tin without the need of DNA. In contrast, coIP and BioID experiment revealed that, in a more physiological context, the interaction is happening more often (BioID, frequency) and is probably more stable (coIP, N51AvsWT Ubx) on DNA (see now summary of coIP quantification, in Supplementary Table 13). This could explain why the results of the EMSA experiments are difficult to interpret in term of cooperation at the DNA-binding level. In order to decipher the requirement of DNAbinding abilities of both Tin and Ubx, we are now providing luciferase assay with Tin WT or N51A (Supplementary Figure 7b), showing that a functional DNA-binding domain of both Tin and Ubx is required for functional cooperation on the dpp enhancer (bound by ChIP for Ubx and Tin, also shown now by EMSA). To conclude, it strongly suggests that both interaction and functional DNA binding are required for proper transcriptional cooperation. 3) The question of the cooperation for DNA-binding/chromatin-loading might be explored in future, using more sensitive in vivo method such as Fluorescence after photobleaching to study the residence time on the chromatin and thereby the influence on each other DNAinteraction (strength, frequency). We already have preliminary data showing that, indeed, interaction (and not DNA-binding) is sufficient for enhancing Tin chromatin residency by Ubx (in Drosophila S2R+ cells). we are happy to share this result with the reviewer, but we feel that it is going beyond the scope of this manuscript and will only be presented here as personal communication. There is a repetition of "Intriguingly". Authors should avoid this term when not necessary (which is the case in several instances). The text has been modified accordingly. The last sentence of the summary is a bit too much over-extrapolated. I suggest to remove or to attenuate this conclusion. First, authors provided evidence that tissue-specific cofactors like Tin or Grh are important for mediating tissue-specific interactions with more ubiquitously expressed proteins. Second, their analysis rather showed that there are tissuespecific TF-enhancer interactions that correlate with tissue-specific expression (especially in the mesoderm). "Taken together, our work reveals that TF-interaction networks underlying cell type specification may be much more complex than previously thought and challenges two longstanding assumptions in developmental biology, namely the relevance of differential RNA expression as determinant for protein complexes and the focused view on TF-enhancer interactions for lineage specific gene expression." We feel that the sentence is already attenuated as we precised it "may […] challenge". As the abstract has been reduced do to editorial policy of Nature Communication, the sentence has been in addition shortened, which we hope will satisfy the reviewer. Supp. Fig1b: the interaction with Med19 is not really convincing. Authors could focus on the interaction with Exd only (considering the previously mentioned caution with Ubx51). We feel that the Med19 AP-fishing is worth it to keep. Though, the level is low, it is the validation of the method for capturing a partner expressed at the endogenous level, in contrast with Exd that is over-expressed. Exd fishing with BioID has been adjusted by new cleaner BioID experiment with BirA-GFP, -UbxN51A and -UbxWT and quantification. The information on partners of Ubx based on their expression profile from previous work (Domsch et al.) should be provided more precisely, with values, in a supplementary table. This information is important given the central message of the paper. The data has been now provided in Supplementary Table 11. It would be informative to have the heatmap of Ubx51 and get more easily accessible (digested) information about the common and specific interactions between Ubx and Ubx51 (as a supplementary table). For example, are DNA-binding interactions more lineagerestricted or not? The tables of the protein enriched in the UbxN51A, WT or common were already provided but we acknowledge that the word files were not clearly readable. We now provided the tables as Supplementary Tables 7 8, 9 and 10, for each tissue, and presenting the protein enriched in the different fraction (chromatin, nucleus, nucleoplasm) according to UbxWT and N51A partner-enrichment (Supplementary Table 11). We also provided within these files a short information concerning the function or the family of the protein/partners identified. We believe that it will be much better readable and the data more accessible this way. Fig. 8 is not really illustrative of the main message of the paper: the information of how Ubx could reach tissue-specificity with general cofactors is missing. It will be more useful to have a final figure focusing on this main message of the paper. We clarified the legend to make it closer to the point. Reviewer #2: highlighted that this study is a well-controlled and well thought out design. There were a few concerns raised by this reviewer, which we addressed in the following way: Major Points: The experimental detail included in the "Mass Spectrometry Preparation" sections in Materials and Methods needs to be drastically expanded. Specifically, very little detail is given on how the mass spectrometer was operated (scan settings, resolution, duty cycles, agc settings, etc.) to acquire the data. Unfortunately, in the field of proteomics there is often a lack of transparency on these details which could have profound consequences on experimental replication of these findings in another laboratory. We thank the reviewer to help us improving the clarity and transparency of the MSprocessing. The text has been modified accordingly in the Material and Methods. In the same section, please add solvent systems (mobile phase A/B composition; acetonitrile with 0.1% formic acid?) in the description of the LC separation. The text has been modified accordingly. The description of the analytical UPLC column (Acclaim PepMap RSLC) should not read "75 um x 2um" as its dimensions, but rather 75 um x 50 cm UPLC column packed with 2 um Acclaim PepMap RSLC particles. The text has been modified accordingly. Although the authors list the proteins identified, there are critical pieces of information not readily accessible to the reader (including supplemental data) including; A) number of unique peptides identified to each protein, B) consistency of expression across biological replicates (easily expressed as intragroup coefficient of variations), C) enrichment values vs background (as fold change or log2 fold change) or p-value. We acknowledge the reviewer that we did not provide a comprehensive table of the data. We now provide several new files, illustrating the different step of analysis as well as, as described for reviewer #1 answer, more detailed tables of the protein per replicate for each sample. The Supplementary Tables 1, 2 and 3 summarize, for each tissue and replicate, the unique peptide number, LFQ value. The table of the calculation of log2-LFQ ratio and the summary of the confidence interval calculation (Supplementary Tables 4, 5 and 6 and summarized lists in Supplementary Tables 7, 8, 9) are now included in order to provide a better accessibility of the data processing to any reader. Consistency across replicate is represented by Pearson correlation and heatmap (Figure 1 and Supplementary Figures 3, 4). No AUC or peak intensity quantitative data is presented in the manuscript (including supplemental data). It is nearly impossible to differentiate the interaction confidence of any of the presented interactors with the information presented. No pvalues/fold changes nor any type of technical vs biological variation data is presented. The authors did upload MaxQuant output files (zipped files @ 60Gb, much larger unzipped), however it would take significant knowledge of that program, expertise in reading this file structure, and additional software to be able to calculate those values from the raw data uploaded. This is beyond what the scientific community should be responsible to do. Quant values need to be presented in table form for at least those species of interest with associated p-values. As mentioned above, the Supplementary Tables 1, 2 Please justify the criteria that proteins significantly enriched in only 2 of the 4 replicates was sufficient to be considered biologically relevant. Depending on that particular proteins intragroup technical+biological variation, this 50% criteria could easily be an artifact from aberrant signals. This point has been stressed in the response of a question of reviewer #1, asking if the replicates could be used for "ranking" the partners. We believe that this ranking/fishing is dependent of the inherent variation of protein biotinylation between embryos, and the intrinsic stochasticity of the MS-process. Furthermore, ranking might not be always correlated with unique function, as exemplified by the strong partner Tin, enriched in 2/4 replicates. We feel that this answer is correlated with the question of reviewer #2 as it is part of the justification of our analysis and selection of the partners. In fact, the intrinsic variation of biotinylation in vivo, the purification and inherent stochastic parameter of the MS-process are the major driver of our analysis design. We chose to rely on a stringent selection based: 1. Statistically relevant enrichment ratio of log2-LFQ Ubx/GFP for each independent replicate (Affinity-purified sample), 2., followed by the selection of the one found significantly enriched at least in 2/4 replicates, 3., combined with the validation with other approach (biochemical and functional). The point 1. is, we believe, a standard set up of calculation using LFQ data. We chose next the selection of 2/4 replicates, that provide the balance between biological variability (we may lose interesting partners like Tin with a selection of 3/4) and false-positive that may appear if partners found at least in one replicate were selected. Further validation by other approaches (interaction and function, point 3.) are re-enforcing the suitability of our pipeline. Though taken carefully, the comparison of partner enriched at least in 1 replicate are now provided in Supplementary Table 10 and might be the primary support for future study. The description of the pipeline of analysis has also been extended in the Material and Methods. Please describe the roll up method used for going from the peptide level measurements and area-under-the-curve calculations to the protein level. Reviewer #3: thought that the findings of this study represent a major leap forward in our understanding of TF context-dependency, and will be of interest to a whole range of scientists interested in developmental biology, gene regulation and context-dependent gene function. There were a few concerns raised by this reviewer, which we addressed in the following way: Major Points: The main weakness of the study pertains to Ubx function in the CNS. This refers both to the actual experiments conducted, and to the minimalistic description of Ubx function in the CNS. First, they only mention one role of Ubx; the control of PCD in subsets of motor neurons. However, they do not actually use this as readout. Instead, they assess motor axon pathfinding (Supplemental Figure 8) (PMIDs 25073156, 28392108), NB cell cycle exit (PMIDs 9651493, 28392108, 20485487), and NB PCD (PMIDs 28392108, 20485487, 16049114 1) We thank the reviewer to help us improving the scientific content of the manuscript and acknowledge that the study was lacking more in-depth analysis on the neural system. We strongly thank the reviewer for her/his guidance and already stress that, due to editorial policies, we could not be able to cite all the proposed bibliography that was, undoubtedly, a solid base of the revision process. Concerning the Fasciclin2 staining, we acknowledged that Ubx mutant phenotype was not detailed in the legend and is now provided. As described in Hessinger et al (2017), the penetrance of the phenotype is rather low. In contrast, the homozygous double mutant of Brm and Ubx, for which we observed an absence of the axon projection in A1 segment, is constant. Moreover, the absence of phenotype associated with the double homozygous mutant of Ubx and the other partners can re-enforce the absence of genetic interaction in this lineage, even if both copy of the 2 partners are absent. To consolidate the crucial point of lineage-specific functional cooperation, we took advantage of the strong phenotype observed on neuroblast (NB) number in Ubx homozygous mutants, for which we observed and described now a functional cooperation between Ubx and Grh only in the neural system but not in the mesoderm (Supplementary Figure 9). Subsequently, we asked whether interaction partners identified in the mesoderm (Srp54, snRNPU1-70K, Brm, Tin), could affect the NBs number and if double heterozygous mutant embryos present a phenotype using the NBs number as read-out (Supplementary Figure 10). Importantly, this was not the case for most of them, with the exception of Brm, which we also identified by BioID of Ubx in the neural lineage. The new data is now provided in Supplementary Figure 10 and quantified in Supplementary 2) In line with the guidance of the reviewer on the role of Hox and Grh on neurogenesis, we analysed more in depth the genome wide binding profile and provided the result here as a personal communication. The cooperation between Ubx and Grh in regulating neuroblast number may not depend on the regulation of cell death, as apoptosis-related genes such as reaper, hid, grm and sickle are not commonly bound by Ubx and Grh (Supplementary Table 15). Several genes involved in cell cycle regulation (dap, e2f1 and stg) were co-bound, however, proliferation-related GO-terms were absent from the Ubx-Grh bound genes dataset (Figure 4a, Supplementary Tables 14-15). It is quite interesting to see that cell death seems to not be regulated by the cooperation between Ubx and Grh, while some genes involved in cell cycle regulation are bound by both TFs. Though, this is only correlating genome binding profile, beyond the scope of this manuscript, it provides new entry point for deciphering the functional cooperation if Ubx-Grh in future, and more generally the collaboration between Hox-Grh and other lineage-restricted TF in neurogenesis. 1) The reviewer raised an important point concerning the identification of rather ubiquitous proteins. However, we do think that the data are providing the confidence and strength in term of sensitivity. First, the comparison of BioID and tissue-specific transcriptome (Supplementary Table 11) revealed that some interactors such as the mesodermal partner Srp54, are expressed at the same RNA level in the mesoderm and nervous system, some like tfAP-2 (and Grh) are expressed at very low mRNA level (2-10RPKMs) but still detected. Another example is the neural-partner top1 which is expressed (at least at the RNA level) at higher level in the mesodermal-transcriptome than the neural one, confirming that the method is not limited by sensitivity but rather interaction specificity. Second, a perfect example is the fishing of Grh in the neural system, also expressed at very low mRNA level. As exhibited in Supplementary Figure 6g, showing the endogenous co-localisation of Ubx, Grh and elav, a rather small subset of cells express the 3 proteins at these stages, indicating that the elav-BioID captured rare co-occurences with respect to cell number but still occurring with high frequency within these cells. In contrast, the well-known Exd partner of Ubx, ubiquitously expressed was not captured by the BioID, re-enforcing the suitability of the method to capture sensitive (and more frequent or close-proximity) interactions. 2) Finally, this point is also now illustrated by comparing the BioID and coIP capture of Exd with Ubx WT and N51A. As described in the response to reviewer #1, Exd interaction strength or stability is similar for Ubx WT or N51A as indicated by co-IP. In contrast, BioID, based on interaction frequency and distance, revealed a stronger biotinylation and subsequent affinitypurification of Exd with Ubx WT than the mutant. It strongly suggests that frequency/closeproximity on DNA, more that interaction-strength, is an essential parameter for the functional Exd/Hox complex and further re-enforces the suitability of BioID for deciphering in vivo regulatory networks. Of note, Figure 5 in now Figure 4, according to reviewer #1 comments. The corresponding gene lists has been added in the following tables: Supplementary Figure 4 Minor Points: It is a common misconception, but elav-Gal4 is not a pan-neuronal driver. While the Elav protein is only present in neurons, the elav-Gal4 driver drives robustly in neuroblasts, neurons and glia, starting from St11 and onward (PMID 17994541). So it is a pan-neural driver. We thank the reviewer to help us improving the accuracy of the scientific vocabulary of the manuscript. The text has been modified has proposed for elav reference as a pan-neural driver and neural tissue. Page 13: I am surprised that only 18% of genes bound by Ubx-Tin in the mesoderm were affected in Ubx mesodermal mutants? Also, this statement does not refer to any specific figure/table. The related list of genes has been provided in Supplementary Table 16. Moreover, we acknowledge a mistake in the calculation as 74/367 are affected (so 20% and not 18%) by the Ubx mesodermal Knock Down (KD). This dataset is from Domsch et al, 2019 in which, we performed mesoderm-specific depletion of endogenous GFP-Ubx using the DegradFP system (degradation tissue-specific of the protein via proteasomal system and nanobodies targeting the GFP). In this publication, we hypothesized concerning the rather small amount of genes mis-regulated upon Ubx-KD. In fact, Ubx-KD is associated with an over-expression of AbdA and Antp that might compensate for Ubx absence on a set of target genes. Consequently, the number of target genes bound by Ubx-Tin and affected by Ubx-KD is limited intrinsically by the dataset. Figure S2d; missing "myc" labels. The missing tag "myc" has been added. Please number the pages and rows. The numbers have been provided	v3-fos-license
2021-12-16T06:23:27.095Z	2021-12-07T00:00:00.000	245143741	{ "extfieldsofstudy": [ "Computer Science", "Medicine", "Physics" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://direct.mit.edu/artl/article-pdf/doi/10.1162/artl_a_00351/1977023/artl_a_00351.pdf", "pdf_hash": "8e310d70ec0b6e96a67f7284f5e49547f501f979", "pdf_src": "Arxiv", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45380", "s2fieldsofstudy": [ "Computer Science" ], "sha1": "fea0506264f95bb66ff0d71e3abcf84f13681e56", "year": 2022 }	pes2o/s2orc	Monte Carlo Physarum Machine: Characteristics of Pattern Formation in Continuous Stochastic Transport Networks We present Monte Carlo Physarum Machine: a computational model suitable for reconstructing continuous transport networks from sparse 2D and 3D data. MCPM is a probabilistic generalization of Jones's 2010 agent-based model for simulating the growth of Physarum polycephalum slime mold. We compare MCPM to Jones's work on theoretical grounds, and describe a task-specific variant designed for reconstructing the large-scale distribution of gas and dark matter in the Universe known as the Cosmic web. To analyze the new model, we first explore MCPM's self-patterning behavior, showing a wide range of continuous network-like morphologies -- called"polyphorms"-- that the model produces from geometrically intuitive parameters. Applying MCPM to both simulated and observational cosmological datasets, we then evaluate its ability to produce consistent 3D density maps of the Cosmic web. Finally, we examine other possible tasks where MCPM could be useful, along with several examples of fitting to domain-specific data as proofs of concept. Introduction To understand a process, one must become it: the intelligence that evolved in nature therefore adapted itself to design pressures presented by various natural processes. To take an example that is central to this paper, the protist Physarum polycephalum 'slime mold' has been successfully used as an unconventional analog computer [Adamatzky2010, Adamatzky2016]. As a 'thinking machine' and a creative pattern finder [Adamatzky2013], Physarum can solve hard spatial problems via its physical growth --most prominently, finding optimal paths and transport networks connecting sets of data points where an underlying network structure is expected. The disadvantages, however, are significant: slow growth of the actual organism, difficulty of feeding it large inputs (e.g., more than a few dozen data points), and limited dimensionality. For these reasons, it is useful to instead use a virtual, simulated counterpart --even if not as complex as its biological template --as for many problems a simulation that qualitatively resembles Physarum's spatial behavior can be expressive enough [Jones2010, Jones2015]. The universe of problems involving interconnected network-like structures ever expands in the Information Age -mostly due to communication and transportation networks (including road and utility networks, printed circuits, socio-economic networks, and the Internet). The importance of epidemiological network modeling has never been as apparent as in 2020/2021. And, in parallel to man-made networks, we continue to discover new natural phenomena with significant interconnected patterns: from molecular structures, through neuronal networks, circulatory/capillary systems, fungal and rhizomatic networks, all the way up to the network of dark matter filaments and intergalactic gas known as the Cosmic web. All these phenomena share an important feature: in one way or another they are built on the notion of optimal transport. Our initial concern in this line of work has been with the last case: reconstructing the filamentary structures of the cosmic matter distribution in a consistent and verifiable way. To this end, we present Monte Carlo Physarum Machine (MCPM): a simulated Physarum Machine that algorithmically mimics the organism's growth. Relying on Physarum's ability to closely approximate optimal transport networks, we feed to its virtual counterpart sparse point data as 3D attractors. MCPM's agents navigate these data to find most probable paths between them; the superimposed aggregate of their trajectories then forms the candidate network structure. We capture the totality of these trajectories as a density field in 3D space, giving rise to continuous network structures which we will refer to as "polyphorms" (see the examples in Figure 1). To demonstrate the practicality of the model, we apply MCPM to galaxy and dark matter halo datasets, interpreting the resulting continuous networks as proxies for the knots and filaments of the Cosmic web. The first contribution of this paper is the full description of the MCPM model (Section 4). MCPM is a generalization of the seminal method described by Jones [2010], from which we draw much inspiration (Section 5). We expand Jones's work in several ways: • All decisions in MCPM are stochastic. This makes the model significantly more customizable, as prior domain-specific knowledge can be encoded in probability density functions sampled by the model's agents. • We extend Jones's model from 2D to 3D, building on the stochastic formulation. We achieve this by redesigning the sampling stencil to work with only a binary sensing decision, rather than performing dense directional sampling as in Jones's proposed 3D extension [Jones2015]. Consequently, this choice also opens up the future possibility for higher-dimensional application of the method. • We add a new data modality decoupled from the algorithm's signaling mechanism, to reconstruct the agent distribution. This modality, called trace, records the collective spatial history of agents' positions. After the model reaches equilibrium, the trace provides more accurate networks in conditions where the agent density by itself would be too low for getting robust reconstructions. The second contribution is the case study of applying MCPM to datasets whose geometric complexity is higher than that of optimal transport networks but still resemble them to a large degree, especially with regard to their topology. Here we study the fitting process and reconstruction results using a task-specific variant of MCPM (Section 6), which we have previously applied in the context of cosmological modeling and visualization [Burchett2020, Simha2020, Elek2021]. We demonstrate how the probabilistic form of MCPM benefits the task of reconstructing the Cosmic web, the structural features of which are best described by a density field, rather than a network in the strict sense (Section 7). We posit that many real-world datasets share this property and offer several trials to support this (Section 8). Physarum Machines: real and virtual Nature offers a wealth of inspirations for solving difficult computational problems, often referred to as metaheuristics. Natural selection itself can be framed as a constrained optimization process with selective pressures acting as design criteria. This is a fact that the tradition of evolutionary computation builds on [Goldberg1989, Hingston2008], as well as more directly biology-inspired optimization methods [Mirjalili2020]. Among the sources of inspirations, one organism clearly shines: Physarum polycephalum, commonly known as 'slime mold', has captured the attention of computational researchers over the past two decades (and half a century longer in biology [Bonner2009] The approach of using Physarum as an analog computer --leading to the moniker Physarum Machines --has been popular because of the organism's propensity to systematically explore its environment for food and shape itself into intricate networks to interconnect it. Food sources thus become straightforward proxies for input data, while different chemical and physical stimuli are available to further steer Physarum's growth [Adamatzky2010]. There are downsides too, however. Physarum grows slowly and is only suitable for small inputs (approximately up to dozens of distinct points The large-scale simulations also unanimously agree on the expected distribution of the dark matter and thus the IGM: after billions of years, the combined effect of gravitational attraction and dark-energetic repulsion has shaped the universe as a vast network ( Figure 2). This quasi-fractal structure [Scrimgeour2012], widely known as the Cosmic web, consists of knots (galaxies and their clusters) interconnected by a complex network of filaments and separated by vast cells of significantly under-dense space called voids. In addition, the entire structure is supported by a scaffolding of dark matter, according to theoretical predictions. These simulations themselves have been constructed to confirm the earlier findings of the Zeldovich framework [Zeldovich1970, Doroshkevich1970, Icke1973], which on theoretical grounds predicts the emergence of these very structures under the influence of the dominant force dynamics known in the Universe. The pertinent question is: how do we transfer the indirect knowledge about the Cosmic web to actually reconstructing its structure? By definition, the IGM filaments would consist of diffuse gas that generally does not emit light at levels detectable by modern telescope instrumentation, making them extremely challenging to observe directly. On top of that, even the galaxies, which form within filaments and nodes and serve as the only luminous tracers of the Cosmic web, are observable only to a degree. For instance, with increasing redshift (i.e., distance from Earth), galaxies (particularly fainter ones) are more difficult to detect and may not be captured by astronomical surveys. We are therefore dealing with inherently incomplete, heterogeneous data. . To our knowledge, no available method is able to recover a complete 3D density map that is inherently driven by the underlying filamentary structure of possibly incomplete (especially observational) data. Given that the Cosmic web is a heterogeneous multi-scale structure without an intrinsic topology, having an accurate estimate for its density distribution is bound to benefit many use cases, including our prior work: better understanding the transition from the galactic to intergalactic environments [Burchett2020], characterizing the intergalactic medium towards singular astronomical phenomena [Simha2020], and potentially shedding new light on the missing baryon problem itself. We approached this challenge from the perspective of finding a sound structural interpolator for the galaxy data, which on the scales where the Cosmic web exhibits distinct features (about 1 to 100 Megaparsecs) can be represented as a weighted set of points in 3D space. Optimally, the resulting model has to enable the following: • detection of distinct anisotropic structures at least several Megaparsecs away from the galaxies, but ideally providing consistent density estimates in the entire region of interest; • detection of geometric features across several orders of magnitude in terms of both spatial scale and density; • structural transfer, allowing a calibration of the model on dense simulated data before deploying it on sparse observed data. Per our findings, the presented biologically inspired swarm-based approach is able to cover all the above criteria. In this paper, we focus on the simulation and modeling perspective; more details towards the required astronomical and visualization tasks are provided in Elek et al. [2021]. Overview of Methodology In its core, MCPM is a dynamic computational model defined by a set of priors: a set of probability density functions and their parameters obtained from domain-specific knowledge and/or fitting to training data. Once configured, the model can be fitted to input data: a weighted set of points in 2D or 3D (Figure 3, left). The result of this fitting is a continuous geometric structure (Figure 3, middle) which we interpret as a transport network over the input data. Rather than a graph, this geometry is represented as a density field stored as a densely sampled regular lattice (Figure 3, right). Section 4 defines the core components of MCPM on an abstract level, i.e. without specifying the prior probability density functions. In Section 5, we outline the differences between MCPM and the method of Jones [2010] and show that it is its natural generalization. Following that in Section 6, we detail the specific version of MCPM that we designed for the task of finding a meaningful Cosmic web map from observational data provided by astronomers: either galaxies or dark matter halos. Monte Carlo Physarum Machine MCPM is a hybrid model --it has a discrete and a continuous component. The discrete component is an ensemble of particle-like agents that can freely navigate the simulation domain; these serve as a fragmented representation of the virtual organism. The continuous component is a 3D scalar lattice that represents the concentration of a marker that facilitates information exchange between the agents and the data. The model's behavior is based on a feedback loop between these two components, executed in two alternating steps: propagation and relaxation (see the attached pseudocode and diagrams in Fig. 4). (1) The propagation step is executed in parallel for each of the agents, which are the model's device for exploring the simulation domain. Each agent's state is represented by a position and a movement direction, which are stochastically updated ( Figure 4a-c) to navigate through the deposit field (referred to as 'trail' in [Jones2010], see Figure 4d, gray cells). The deposit field is stored as a 3D lattice of scalar values, representing the marker (in the biological context usually referred to as 'chemo-attractant') emitted by both the input data points as well as by the agents. The deposit effectively guides the agents towards the other agents as well as the data: the agents move with higher likelihood to the regions where deposit values are higher. In addition to the deposit, we also maintain a scalar trace field ( Figure 4d, green cells) which records the agents' equilibrium spatial density, but does not participate in their guiding (detailed explanation of trace in Section 5). propagation steps (Section 4). While individual agents follow a seemingly random set of paths (left), a superposition of the entire swarm reveals an underlying structure present in the data (right). We store the resulting 3D structure as a density field called 'trace'. Successive propagation steps build up the agents' trajectories: random walks that follow the structure of the deposit field, and are recorded in the trace field. While each individual agent draws a seemingly chaotic path ( Figure 5, left), a superposition of many agents averaged over a sufficient time window results in a smooth converged structure ( Figure 5, right). It is worth noting that the data points are also represented by agents, but of a special type: one that does not move, but merely emits deposit according to its weight ('mass'). We found this to be the most consistent way of handling the input data: the relative structural impact of the agents and the data now simply becomes a question of tuning the amount of deposit they each emit. (2) The relaxation step ensures that the simulation eventually reaches an equilibrium. To this end, the deposit field is spatially diffused by a small isotropic kernel, and attenuated: that is, each cell is multiplied by a value < 1. The trace field is also attenuated but does not diffuse in order to preserve the geometric features in the agents' distribution, which the trace represents (please refer to Section 5 for further discussion on the trace field). The simulation reaches its equilibrium when the amount of deposit and trace injected into the respective fields in the propagation step equals the amount removed by the attenuation in the relaxation step. With the parameters used in our experiments (Section 7), this usually takes hundreds of iterations. Probabilistic sampling The core of MCPM is the agent propagation defined in terms of three probability distributions, the configuration of which can be tuned in runtime. These are: • P dir for sampling the agents' directional decisions during the sensing and movement phases; defined on the unit sphere relatively to the current direction of propagation. • P dist for sampling the agents' distance decisions during the sensing and movement phases; defined in positive real numbers along the current propagating direction. • P mut for making the binary 'mutation' decision whether the agent should remain on its current course, or branch into the newly sampled direction according to P dir . By specifying these three distributions one defines a particular instance of MCPM --we discuss this further in the following Section 5. For our Cosmic web detection task, we detail our choices of these three distributions in Section 6. Now, it is good to take a step back and understand why such a model produces spatial networks and why these align with the input data. Figure 6 supports this discussion. The key is that the agents are navigated towards large 'pools' of deposit --this is ensured by defining P mut such that it preferentially selects those directions where higher deposit values were sensed. Agents are therefore attracted towards data in their vicinity (thus interconnecting them), as well as other nearby agents (thus reinforcing existing pathways). The shape characteristics of the network (curvature of the paths, size and acuity of the features, as well as connectivity patterns) are further dependent on the combined influence of P dir and P dist . In particular, P dir needs to be strongly forward-oriented, which ensures momentum preservation and prevents the agents from getting stuck in local pools of built-up deposit. Figure 6: Summary of all the data modalities that MCPM operates with. Left: individual agents (white particles) flowing among the data points (red particles). Middle: pools of deposit emitted by the data (gold) interconnected by the trace field (pink). Right: the resulting filamentary structure contained in the trace field, representing the transport network. In summary, MCPM solves tasks by acting as a structural interpolator for input data. When adequately configured and parametrized, it can find natural transport networks that connect neighboring points as a human observer would intuitively do --even in the topologically richer 3D space (Section 7). These networks are represented by the converged trace, i.e., by a continuous density field rather than an explicit graph. Relation to Max-PM Our main motivation for extending the 'Max-PM' model of Jones [2010] was the question of parametrization. While Max-PM is configurable enough to reproduce the morphology of Physarum Polycephalum and beyond, the agents respond to different deposit concentrations in always the same way: follow the direction of maximum deposit concentration. Hence the alias "Max-PM", PM standing for Physarum Machine. At any rate, this behavior --especially after our initial experiments in 3D --lead to overly condensed pathways and only moderately connected networks (more in Section 7). While captivating and resembling a 3D Physarum, the fits that Max-PM produced did not match our target cosmological data well. Another observation was the necessary increase of sampling effort: Max-PM in 2D uses 1+2 samples for directional sensing, our extension to 3D needs at minimum 1+8 directions (determined empirically). This increases the sampling effort considerably as well as the number of access operations to the deposit lattice. These considerations lead to us targeting the agents' directional sampling, specifically modifying it to use stochastic Monte Carlo sampling. In the remainder of this section we discuss our specific extensions of Max-PM in detail. Figure 7: A small segment of a transport network grown on the same data using Max-PM (left) and MCPM (right). Max-PM yields a clean, well defined structure, which however does not consistently cover all the input data, in contrast to the MCPM variant (32% data points missed by Max-PM, compared to 0.072% missed by MCPM). Stochastic sampling of agent trajectories The way in which Max-PM treats agent navigation is suitable for the original context: the Physarum agents use chemotaxis to move around, following the chemoattractant concentration gradient (represented by the deposit field). In ours and arguably many other scenarios --where the main concern is recovering a network structure that fits target data --we need a more complex behavior that can be configured based on available knowledge about the data distribution. The main concern is that in any given geometric configuration of point data a multitude of feasible pathways is available to meaningfully interconnect them. Figure 8: Simple configuration where the agents arriving from data point A need to split evenly between points B and C (a). In an actual reconstruction scenario, this corresponds to a bundle of agents splitting about evenly and branching out into two separate filaments (b,c) which afterwards merge with the flow of the other adjacent agents. As an example, consider the elementary configuration in Figure 8a: we want the resulting network to branch out when connecting the data point A with B and C, which means that the agent traveling from A has to choose steering towards B or C with roughly the same probability. This ensures that the aggregate distribution of all agents passing through this location is going to have a branched shape. In Figure 8b-c we see an actual such configuration: agents arriving from the bottom-left branch split in two 'streams' and fluently transition to the more complex region in the right. This behavior is represented by the mutation decision (Section 4) encoded by the discrete probability P mut --that is, whether the agent should remain on its current course, or branch out in the direction generated in the sensing step. To provide additional flexibility in representing paths with different scales and curvatures, we modified the agents' spatial decisions to behave probabilistically as well, rather than using constant steps and angles for the agents' movement [Jones2010]. This behavior is defined by the continuous probability density functions P dir (2D distribution on a sphere) and P dist (1D distribution on a half-line). We can now demonstrate how Max-PM can be framed as a special case of MCPM. • Define the mutation probability P mut as d 1 , where d 0 is the deposit value ahead of the agent and d 1 is the deposit in the mutated direction. Such definition effectively means that the larger of the deposits will always be selected for the agent to follow. • Define P dir and P dist using Dirac delta distributions with the angular and distance parameters as offsets for the peaks. In Section 6 we discuss the P dir , P dist and P mut that define the particular variant of MCPM we employed for reconstructing the Cosmic web structure. Extension from 2D to 3D Enabled by the notion of the mutation probability P mut we can now simplify the sampling stencil --that is, how many directions need to be sampled during the agent's sensing stage. We observed that in Max-PM the network complexity (connectivity) depends on the number of directional samples, and that this becomes even more pronounced in the topologically richer 3D space. Given that the directional navigation in MCPM is controlled by P dir and P mut , we can reduce the necessary number of samples to two: one in the forward direction, and one in the mutation direction. Any direction where P dir is nonzero can potentially be sampled, and even directions with low deposit have a chance to be selected (subject to the specific definition of P mut ). This has two advantages: the savings in directional sampling can be reinvested into increasing the resolution of the agent's trajectory, and extensions to higher than three dimensions are now possible without increasing the sampling effort. The fact that high-dimensional sampling and integration problems can be effectively solved with binary sampling stencils is well established in the Monte Carlo simulation community. Both direct [Kajiya1986, Veach1997, Kalos2008] and Markov-chain Monte Carlo [Metropolis1953, Hastings1970, Veach1997a] avoid the curse of dimensionality this way when constructing random walks. Moreover, this strategy can even be the most efficient one, as documented, for instance, by the ubiquity of path tracing methods for simulating light transport in the visual effects industry [Christensen2016, Fong2017]. Indeed, our method draws significant inspiration from these works. Aggregation of agent trajectories Finally --in addition to the deposit field --we define an additional data modality in MCPM: the trace field, or simply "trace". Just like the deposit, the trace is stored in a 3D lattice with the same resolution ( Figure 4d and Figure 6, middle). Trace is computed as a superposition of all locations visited by each agent within a certain time window --it is an equilibrium spatial distribution of agent trajectories. Please refer to the pseudocode in Section 4 for how the trace is constructed on the algorithmic level --formally, we can define it as a probability density P T over positions x in 3D space: where N is the number of agents in the cell adjacent to x, and n is a normalization constant. The expected value of N(x) is the Monte Carlo estimate calculated as a moving average, as a function of the attenuation parameter (cf. pseudocode for the relaxation step in Section 4). Rather than using the deposit to recover the reconstructed network (as in Jones's method [2010,2015]) we use the trace for this purpose. This has multiple advantages: 1. Resolution: in contrast with the deposit, the trace field is not subject to spatial diffusion --structural details are therefore preserved (Figure 9). 2. Specificity: recovering the network from the trace becomes much easier than from the deposit, as in the latter, the data term has by design a much higher impact than the agent term. 3. Interpretability: while the exact mathematical properties of the trace in relation to P mut , P dir and P dist are yet unknown, they are nevertheless in a direct relationship. One way to interpret the trace is the total probability of the agents' distribution marginalized over all possible geometric configurations that can occur in a given dataset. Another way of interpreting the trace is that it provides a continuous estimate of the geometric and topological structure outlined by the input data. This is related to the use of such a structure by Jones and Saeed [2007], referred to as 'trail' in their work. The application of trail here was to substitute for standard image-processing kernels, as used for denoising or contrast enhancement. Unlike their trail, the trace lives in a different frame of reference than the input data: the data are sparse and unordered points, while the trace is a function densely sampled in space and continuous. Other than that, these data structures are conceptually similar. To succinctly describe the trace in text, we establish the term "polyphorm". We will be using this term to refer to the particular continuous geometries represented by the trace field, as well as a general label of the concept of a continuous transport network, i.e., a network implicitly defined by the trace density field (as opposed to a network explicitly represented by a graph or other means). Figure 9: Comparing the distribution of data-emitted deposit (gold) and the reconstructed trace field (purple) in two sample configurations. While the deposit field is sparse and diffused, the trace is continuous and sharp, and thus serves as a better estimate of the transport network. Fully continuous representation Our last design choice is to fully decouple the representation of agents and the underlying discrete data structures. In Max-PM, only one agent can be present in each cell of the deposit field at any given time. This is suitable for simulating the actual organism, as each agent represents a small fragment of the body and therefore occupies a certain volume of space. In our case, this behavior is no longer desired: our agents represent an abstract spatial distribution with possibly several orders of magnitude in dynamic range. For instance, cosmic overdensities between 10 -3 and 10 2 are important in the astrophysical inquiry. We therefore allow agents to move freely without enforcing any limit on their number in each grid cell. This has additional benefits: it decreases the amount of necessary bookeeping, and decouples the reconstructed network structure from the data structures' resolution (as demonstrated in Section 7). MCPM for Cosmic Web Mapping In Sections 4 and 5 we described the components of MCPM and provided rationale behind our design. Now we specify the probabilities P dir , P dist and P mut which together define the variant of MCPM used in our previous work [Burchett2020, Simha2020, Elek2021] for the Cosmic web reconstruction task. We likewise use it in the experiments throughout Section 7 and most of the paper. Directional distribution P dir The main criterion for P dir is that the agents should maintain a certain momentum --otherwise, if allowed to turn too rapidly, they would invariably get stuck in local maxima of the deposit field. The simplest choice here is uniform distribution over the forward-facing cone of directions. This distribution is parametrized by a single scalar parameter: the opening angle of the cone. Routines for efficiently drawing samples from this distribution can be found for instance in [Watson1982]. Other viable choices of P dir include the spherical Gaussian, the Von Mises-Fisher distribution, the Henyey-Greenstein distribution [Henyey1941], and their truncated versions. When configuring P dir , the rule of thumb we used in our fitting confirms the findings of Jones [2010]: the sensing angle should be larger than the mutation (i.e. turning) angle. We typically use twice the value for sensing than for turning. The variation of these and other parameters is further explored in Section 7. Distance distribution P dist Since the Cosmic web is built up from filaments consisting mostly of hot gas, we use the Maxwell-Boltzmann distribution for sampling the agents' sensing and movement distances. This statistical distribution describes the velocity dispersion of particles in idealized gas and plasma [Mandl1988], which makes it suitable for our use case and potentially for modeling other physical systems. Efficient sampling from this distribution is described e.g. by Hernandez [2017]. Other good candidates for P dist include the exponential distribution (e.g. to model bacterial motion dynamics [Li2008]), log-normal (for particles under Brownian motion), Poisson, Gamma, and other distributions defined in R+. Convex combinations of multiple modes are also possible, if different distinct scales of features are expected in the modeled system. The weight of the tail in P dist determines the range of feature scales MCPM will be able to reconstruct: in the simplest case of constant sensing and movement distances, the model will be sensitive to only one characteristic feature size. When generating the sensing and movement distances for an agent, we use the same seed to sample P dist --this is to avoid the cases when the movement distance would exceed the sensing distance. Even more so than in the directional sampling, here we find that the sensing distance should be significantly larger than the movement distance: a good starting point is a difference of about an order of magnitude. Mutation probability P mut In contrast to distributions P dir and P dist , this is a binary probability that an agent deviates (branches away) from its current course. Its definition needs to ensure that agents predominantly steer towards higher concentration of the deposit. (This, in our experience, is the sufficient condition for the simulation to equilibrate in a configuration that follows the data.) If this were the opposite case, the agents would actually be repulsed by the data, as well as from each other. With reference to the pseudocode in Section 4 we define P mut as P mut (d 0 , d 1 , s) = d 1 s / (d 0 s + d 1 s ) where d 0 is the deposit in the forward direction, d 1 is the deposit in the mutated direction, and s >= 0 is the sampling exponent. We examine the impact of the sampling exponent (and other parameters) on the resulting network geometry in Section 7. An alternative definition of P mut could assume that the agents move in a participating medium with density proportional to the deposit, in which case the agent dynamics would be subject to the Eikonal equation and P mut could be derived from the Beer-Lambert law. This is not our case however, as collisions between the intergalactic medium particles do not have appreciable impact on the Cosmic web geometry. It is worth noting that by increasing the values of the sampling exponent we approach the limit case discussed in Section 5, i.e., MCPM becomes identical with Max-PM for s → ∞. Experiments Our open-source implementation of MCPM called Polyphorm (github.com/CreativeCodingLab/Polyphorm) is written in C++ and uses DirectX with GPU compute shaders for executing the parallel propagation and relaxation steps, as well as facilitating an interactive visualization. Our GPU implementation was developed on an NVIDIA TitanX, and is capable of fitting 10 million agents over 1024 3 deposit/trace grids at roughly 150-200 ms per model iteration, with an additional 30-100 ms needed for the visualization. Including the agents and rendering buffers, no more than 6 GB GPU memory is consumed at that resolution, using float16 precision to store both the deposit and trace grids. In all our experiments the model converges in fewer than 700 iterations (about 1-3 minutes of real time). It is thanks to the massive parallelism of modern GPUs, combined with the high locality of memory accesses in MCPM, that we are able to simulate 10M or more agents at interactive rates. This is essential for obtaining noise-free simulation results. Further performance data are available in Elek et al. [2021]. The performance scales close to linear with the number of voxels, and sub-linearly with the number of agents: 100M agents runs at 300 ms ceteris paribus, that is 10x more agents at only 2x the slowdown (compared to 10M agents). Using more agents has minimal impact on the model's spatial details, but it does increase effective resolution by proportionally reducing the Monte Carlo noise levels. All of the model's parameters (specific values are provided in the following subsections, with reference to the pseudocode in Section 4) can be modified at any point during the simulation, which allows for easy tuning of the resulting 3D polyphorm structures. Edits typically take several seconds to manifest, both visually and in the value of the loss function (in Section 7.3 we discuss the details of our employed loss function). We rely on cosmological datasets to conduct the practical part of the evaluation, in which we focus on fitting the model. . Here we will be concerned with the behavior of MCPM and the geometry generated by it. For the visualizations, we use direct volume rendering [Beyer2015] (for field data) and particle rendering (for the agents). Sometimes, in addition to projections of the full 3D data, we isolate slices and cross-sections where better insight is necessary. Full exposition of our visualization methodology is provided in Elek et al. [2021]. All the following experiments use the MCPM variant defined in Section 6, unless stated otherwise. 7.1 Self-patterning Even though we primarily care about the data-driven mode of operation, we start by looking at polyphorms produced by the model without any prior stimuli. In the following examples we work with a 540 3 simulation grid uniformly filled with 2M agents at the start of the simulation. In line with Jones's findings [Jones2010] we observe that the free patterns created by MCPM are stable in the qualitative sense. That means that for any given parametrization the generated features will be comparable, but rarely static: the generated polyphorms are in constant flux. In the experiment in Figure 10 we explore a number of configurations that lead to distinctive results. This experiment was conducted in a single session, changing one or multiple MCPM parameters at a time and waiting for the simulation to sufficiently converge after resetting the agents to random initial positions. In addition to these meta-stable patterns, a plethora of transient ones is accessible. These occur when the model's configuration changes significantly, as a sort of dynamic phase change. Figure 11 shows a few illustrative examples of serendipitous trips through the parameter space. Given that these forms are dependent on the momentary state of the model before the phase change, and are unique to both the original and modified parametrization, there is currently no way to enumerate them systematically. Even though all the parameters impact the polyphorm, the most significant determinants of shape are the Sensing angle, Sensing distance, and the respective movement parameters. These would end up to be the parameters we adjusted the most frequently during fitting. In the experiment in Figure 12 we therefore focus on exploring these fitting dimensions. Curiously, we observe correlation between the stability of the fit and how interconnected the network structure is: the polyphorms in the upper right part of the figure were much more persistent in time than those in the left. Another important design dimension is to consider different behaviors of the agents. In particular, many natural systems contain repulsion as another important shaping force. Encoding this type of behavior in MCPM simply means to reverse the sampling decision given by P mut (see pseudocode in Section 4). Consequently, the agents will navigate towards lower deposit concentrations, i.e., locations not occupied by other agents or data points. More complex repulsion behaviors could also be encoded through specialized redesigns of P dir and P mut . Figure 13 documents three MCPM runs in 1024 3 simulation grids and 10M agents, comparing the original attractive agent behavior, its repulsive inversion, and a 50/50 mix of the two behaviors. In line with the previous experiments, the attractive behavior leads to a network-like polyphorm that, after forming, remains relatively stable and slowly condenses to thicker and sparser structure. On the other hand, the repulsive behavior forms cellular patterns around the sites where they were initialized. These patterns are only transient however: as the agents continue repelling each other, the pattern breaks down and dissipates. Finally the mixed behavior --where each agent randomly decides to be attracted or repelled by deposit --leads to a structure that is a blend of both: sites form cells, neighboring sites merge together, and the resulting interconnected structure slowly diffuses. Figure 13 maps the temporal progression of each regime, and additionally shows a 3D view of the entire structure at a point when its features are most visible. Uniform stimuli We now start introducing data in homogeneous configurations. We show that MCPM produces meaningful networks over points with different uniform distributions in 3D: in a regular lattice (Figure 14), distributed randomly (Figure 15), and following a low-discrepancy 'blue noise' distribution [Ahmed2016] (Figure 16). In a regular lattice ( Figure 14) MCPM robustly finds straight connections between points and due to the multiscale movement of the agents (cf. Section 6) also the diagonals of the sub-boxes. The model can be tweaked to prefer the diagonals simply by increasing SD to about 70--80 vox. The fitting takes about 300 iterations of the model to converge (see the discontinuity in the energy plot in Figure 14, bottom, where we restarted the fitting process). Remarkably all the data points receive nearly identical density of the agents, as shown in the histogram. Randomly distributed points ( Figure 15) have a tendency to cluster, leading to a spurious reinforcement of some of the fitted network connections. As a result, fitting from uniformly initialized agents takes somewhat longer, about 500 iterations. The density histogram at data points is close to Gaussian, with a slight positive skew, and variance proportional to the density fluctuation of the data. In Figure 16, using a low-discrepancy Poisson-like distribution of the data [Ahmed2016], we obtain perhaps the most organic of homogeneous patterns. Since the low-discrepancy points emulate the distribution of cells in biological tissues (and similar such phenomena), the resulting fit is resembling a natural scaffolding, not unlike bone marrow (also cf. Section 8.3). The structure also bears resemblance to rhizomatic systems (such as tree roots and mycelial networks). The fitting takes about 500 iterations and results in a density distribution not unlike the random case, but with a slightly stronger positive skew. Tweaking the values of SE and then trying different distance sampling distributions would impact the acuity of the fitted network, and would expand or narrow the resulting distribution histogram. Fit energy (loss function) MCPM solves the problem of reconstructing a volumetric transport network over a set of discrete points. The network should include all input points, perhaps except for outliers. We also want the weight of the input points to influence the strength of the network connections, i.e., to attract more agents. The energy function we have used for fitting encapsulates all of the above considerations. To compute the energy I of the current MCPM fit, we compute the mean of the trace values summed over the input data point locations, normalized by the data points' weights: This design has one shortcoming: collapsed fits can occur, where all the agents are attracted to the nearest data point (Figure 17, right). Compared to a 'healthy' transport network (Figure 17, left) the collapsed fit lacks connections between neighbors, rather resulting in a disconnected set of 'blobs'. Since there is no established definition of connectedness in this context, we leave this aspect for visual evaluation and use the above defined energy function to only explore the fitting parameter space locally. The interactive visualization in Polyphorm and the ability to inspect the fit by slicing and different rendering modes was therefore essential to obtaining good reconstruction results [Elek2021]. Bolshoi-Planck data (simulated) Our first evaluation of an actual reconstruction scenario is based on the Bolshoi-Planck (B-P) cosmological simulation at redshift zero [Klypin2016] with dark matter halos extracted by the Rockstar method [Behroozi2012]. Halos are considered to be significant clusters of dark matter and are the formation sites of galaxies; the most massive halos are typically located at the intersections (knots) of filaments. Importantly, at the scales of our interest the halos are sufficiently represented by their 3D position and mass. The full dataset contains 16M halos, most of which have extremely low mass and are not significant in determining the overall filamentary structures. Therefore, we remove all halos with mass M halo < 5x10 10 M sun , leaving approximately 840k of the significant ones. These halos span a cubic volume approximately 170 Mpc on a side. Because dark matter is gravitationally coupled to conventional, baryonic matter (on large scales), it follows that their distributions will mostly have the same structure. Also, since the B-P dataset is simulated, it does not suffer from key limitations of observed data: namely, data points missing due to occlusions or other observational issues. This makes the B-P data an ideal fitting prior, and as such we use them to calibrate the MCPM model's parameters to further use with the observed Sloan Digital Sky Survey (SDSS) data (Section 7.5). Figure 18 shows the MCPM fit to B-P data. We obtain a detailed polyphorm with features on multiple scales: in line with theoretical predictions, we observe characteristic loops and filaments on scales from a single Mpc to about 100 Mpc. The quasi-fractal nature in the data is also revealed --we observe self-similar knots and voids (lacunae) at different scales. Fitting to the B-P data takes about 900 iterations and (similarly to the uniformly distributed point sets) results in a near-Gaussian density distribution at the halo locations, as seen in the histogram in Figure 18. The fits match the underlying data extremely well: more than 99.9% of the halos are covered by the reconstructed network, and as detailed in Burchett et al. [2020] we find a strong correlation between the MCPM trace and the ground-truth particle density of the B-P dataset. This was crucial to establish the mapping between the MCPM trace density and the Cosmic overdensity, to yield interpretable astronomical results. Further numerical evaluation of this correlation is provided in Section 7.10. The progression of the fitting and the formation of the obtained geometric features is documented in Figure 19. Sloan Digital Sky Survey data (observed) Having verified in Section 7.4 that MCPM is capable of fitting simulated structures arising from the current cosmological theory, we now move to the actual observed data. The target here is to reconstruct the Cosmic web spanning 37.6k spectroscopically observed galaxies extracted from the Sloan Digital Sky Survey (SDSS) catalog [Alam2015]. The selected galaxies are within the redshift range [0.0138, 0.0318], equivalent to the distances between 187 and 437 million light years from Earth, which on Cosmic scales is relatively close to home. Just as with the Bolshoi-Planck data, the galaxies at these scales are sufficiently represented as points in 3D space (derived from the celestial coordinates of each galaxy and its corresponding redshift-derived distance) with weights proportional to the galactic masses. After the spatial transformation of galaxy coordinates, the galaxies are contained within a volume with 280 Mpc in its largest dimension. The main challenge faced here is the incompleteness of the data: some galaxies elude observation either due to their low apparent brightness, or being too closely projected on the sky to other bright galaxies. Whatever the case, we need MCPM to be robust under these conditions. The most important precaution is to prevent the model from 'detecting' filaments where there are not sufficient observations to support them. To this end, we disable the agent-emitted deposit to make sure that all detected structures are due to actual data. Figure 20 shows the MCPM fit to the SDSS data. To obtain this result, we use the same model parameters as for fitting the Bolshoi-Planck data, thus facilitating a prior-informed feature transfer. In line with that, the fitted polyphorm ends up with structurally similar features --in spite of fitting to an order-of-magnitude fewer points, we observe filaments, knots and voids on the expected scales. The density distribution of the network likewise exhibits a clearly near-Gaussian profile, as well as similar convergence characteristics (Figure 21). The MCPM fit also reveals an artifact in the data: the so-called fingers of god, elongated structures pointing towards Earth (visible in the vertical 2D slices in Figure 20, right). These result from biased conversions of redshift to line-of-sight distance in the observations; compensating for them is therefore subject to corrections applied to the data itself --for more details please refer to the discussion in Burchett et al. [2020]. Currently we do not apply any such corrections. Figure 20: MCPM fit to the 37.6k galaxies in the SDSS catalog, simulated using deposit/trace grids with 1024x584x560 voxels, 10M agents, and otherwise identical parameters as in the Bolshoi-Planck case. The simulation domain was 310 Mpc across the largest dimension. A key validation of this fit is based on correlating the reconstructed filament network with spectroscopic sightlines from distant quasi-stellar objects ('quasars'). This is the main result in Burchett et al. [2020], in which we found a strong positive correlation between neutral hydrogen absorption signatures and the MCPM-derived density in low-to-medium density regimes, and a significant anti-correlation in the high-density regime. The latter can be attributed to the lack of neutral hydrogen in the close proximity of galaxies, as hydrodynamical shocks and other processes associated with galaxy formation ionizes the hydrogen gas and thus suppressing its ability to absorb the light from quasars. Crucially, what enabled the validation of the MCPM-generated map is the continuous nature of the fit. Rather than obtaining a binary polyphorm detecting the presence of agents, the probabilistic nature of MCPM yields a scalar trace field expressing the time-averaged density of the agents. We interpret this trace simultaneously as a density and a probability that the detected features are significant; as described above, we establish a monotonic mapping between the MCPM trace and the cosmic overdensity, which is a directly interpretable quantity in the astronomical context. At this point, in reference to the requirements formulated in Section 2, we can conclude that the model qualifies for the designated task of reconstructing the Cosmic web map from observed data: • MCPM naturally synthesizes anisotropic filamentary features, and, when configured correctly, finds a cohesive transport network over the input data. • MCPM outputs a continuous density map with features across multiple levels of scale and intensity. • MCPM can be easily pre-trained on prior data and transferred to target data through the optimized values of its (hyper)-parameters. In the remainder of this section, we perform additional experiments to probe the model for its specific behaviors and robustness under degradation of the input data. We choose the Bolshoi-Planck dataset for this evaluation due to its higher point density. Distribution analysis In analyzing the polyphorm patterns recovered by MCPM it is important to ask the question whether these are not arbitrary: just random, haphazard connections between neighboring points. A detailed look into the input data and the resulting fits shows that this is not the case: as analyzed in Figure 22, many individual data points come together to give rise to any single filament. This matches the prior domain knowledge: the average distance between neighboring galaxies is in the order of 100 kpc, while the Cosmic web structures occur roughly between 1 and 100 Mpc. Crucially, it is the galaxies (or dark matter halos) whose distribution is aligned with the nodes of the Cosmic web, not the other way around. Their locations therefore sample the much smoother high-level structure of the Cosmic web, and this ultimately is what MCPM traces. Figure 22: Zooming in a section of the Bolshoi-Planck dataset about 50 Mpc wide (middle) and the filaments reconstructed therein. The magnifications show the raw data: dark matter halos as single red pixels and the MCPM agents in white. We see that the halos sample the domain with a significantly higher frequency than the actual reconstructed filaments. The above is of course subject to the configuration of the model. Our fits to both B-P and SDSS data use the sensing distance of 2.5 Mpc (as calibrated by fitting to the former). Structures significantly smaller than this would not be captured in the fit: they are essentially invisible to the agents. This is what makes the domain-specific knowledge valuable --it constrains the fitting process and prevents us from obtaining nonexistent structures. Going one step further, in Figure 23 we zoom in on individual knots (intersections) of the reconstructed filaments. Here the visualization segments the trace field into three bins: low (blue), medium (green) and high (red) density values. The agents' distribution here agrees with the astronomical intuition: the cores of the filaments and the knots receive the highest trace densities, which then fall off towards the outskirts of the observed structures. Figure 23: Two examples of complex knot specimens found in the Bolshoi-Planck data. These are regions about 10 Mpc wide, corresponding to 70x50x50 voxels (each visualized slice is 10 voxels thick). We observe the knots and filament cores to have the highest density (red), with a gradual fall-off outwards (green to blue). Figure 24: Fitting to Bolshoi-Planck data with variable Sampling exponent (SE). With increasing SE we obtain sharper polyphorms with reduced connectivity and worse data coverage (left to right). In the extreme case of Max-PM, the network condenses into nearly discrete filaments with a poor coverage of the input dataset (see the 'null' field in the plots, represented by the leftmost gray bar in the density histograms). Mutation probability The agents' decision whether to branch out, and with what probability, is one of the key determinants of the shape and complexity of the generated polyphorms. Here we explore this aspect of the model. The variant of MCPM defined in Section 6 uses a probabilistic formulation for the mutation decision, controlled by a single scalar parameter: sampling exponent, SE for short. Increasing the value of SE leads to sharper and more concentrated filamentary structures (Figure 24), which also increases the energy of the fit. On the downside, the number of data points not covered by the fitted polyphorm increases, and for the limit case of infinite SE we miss more than 3% of data. The fitting process therefore involves finding a tradeoff between acuity of the structure and its coverage. In result, our fitting to the astronomical datasets uses moderate values of SE, usually between 2.2 and 4. In the right panel of Figure 24 we compare to Max-PM [Jones2010]. Here we used 8 directional samples to compute the maximum of the deposit for the agents to follow in every step. As discussed in Section 5, this increases the structural acuity even further, optimizing the network by erasing smaller filaments, rounding off loops etc. While a sensible behavior for Physarum, this behavior causes a dramatic reduction of the network connectivity. As a result, 32% of input data points are missed by Max-PM, making this a poor fit (in spite of the seemingly high fitting energy I, cf. Section 7.3). Section 7.10 offers a more detailed quantitative comparison between MCPM and Max-PM. Trace & deposit field resolution Here we examine the scaling of MCPM with respect to the resolution of the deposit and trace fields' representation. In Figure 25 we compare three different field grid resolutions (with proportionally scaled numbers of agents), keeping the remaining model parameters fixed at the nominal values. We observe that while the amount of reproduced detail decreases with resolution --and the amount of MC noise increases as the number of agents goes down --the overall structure of the polyphorm remains largely intact. This is a desirable outcome, as we want to avoid a tight coupling between the precision of the representation and the geometry of the reconstructed network. Dataset decimation The ability of the model to retain structure is important when it is trained on dense data (such as the Bolshoi-Planck halos) and then transferred to incomplete data from observations. In Figure 26 we look at the structure retention when reducing the number of dark matter halos in the BP dataset by thresholding the halo mass. By keeping only the heavier subset of halos we emulate the actual observations: the smaller, dimmer galaxies are less likely to be captured by SDSS and other such surveys. We see that even when decimating the input data by nearly an order of magnitude (from 870k halos down to about 100k), MCPM finds a good portion of the filaments --all of the larger ones and some smaller ones. This gives us a clue about how much structure might be missing when fitting to the SDSS data. In Burchett et al. [2020] we explain the process of matching the density of these two distinct datasets in more detail. Figure 26: Fitting to a variable number of Bolshoi-Planck halos, and otherwise using the standard parametrization for this dataset. While we observe a reduced number of detected filaments with fewer halos, the overall structure is preserved without major differences. Quantitative comparison between MCPM and Max-PM We conclude this section with a detailed numerical study of MCPM in comparison to Max-PM [Jones2010], complementing the preceding qualitative results. In this experiment we fit both models to the Bolshoi-Planck simulation data (Section 7.4). Focusing on the accuracy of the fit, we do not apply any thresholding and retain all 12M dark matter halos, with the ultimate aim to match the reference density field produced by the simulation. In contrast to the halo catalog, the reference field represents the actual density of the simulated matter tracers, binned in a 1024 3 grid. To measure the extent to which MCPM and Max-PM match the reference data, we repeat the same procedure used to obtain the cosmological overdensity mapping in Burchett et al. [2020]. We first fit to the full 12M-halo catalog using both MCPM and Max-PM in a 1024 3 simulation grid (which allows us to establish a one-to-one mapping to the reference grid without any interpolation). We then standardize both fits to make sure their means and variances match, thus obtaining comparable distributions. Finally, we compute 2D histograms binned over the fitted and the target density ranges, recording the mutual correlations between the corresponding voxels in 3D space. The results are summarized in Figure 27. We observe two key differences between MCPM and Max-PM. First, MCPM covers a significantly wider range of the reference density field, roughly by 5 orders of magnitude in the low end of the distribution. In practice, this means that MCPM can capture much fainter features of the target structure, and characterize smaller substructure within the Cosmic web. This mirrors our findings from Section 7.7, where we see a significant portion of the halos being missed by the Max-PM fit. Second, the MCPM fit has considerably better precision compared to Max-PM. This is apparent from the spread of the histograms, which we visualize by plotting the percentile bounds of the per-row conditional distributions. From these results, we conclude that MCPM is able to reconstruct an accurate proxy of the reference density field from very sparse data, i.e., using 12M weighted halo points compared to the 1B voxels in the reference density grid. Compared to Max-PM, MCPM improves the Cosmic web density reconstruction by several orders of magnitude, which in practical terms means extending the reconstruction from only the dense structures (knots and major filaments) all the way to the edges of cosmic voids. Discussion The presented MCPM model is suitable for revealing geometric structures in incomplete data, as demonstrated by mapping the Cosmic web filaments in SDSS galaxy data. It is especially effective if a training dataset is available to determine the prior parameters (as we did with fitting to Bolshoi-Planck dark matter halos) and if expert knowledge is available to determine the appropriate probability distributions (as we did by using the Maxwell-Boltzmann distribution for agent distance sampling). Fitting MCPM to data is a robust, reliable process. For any given parametrization the model eventually converges to the same outcome (minus the MC variance), regardless of the initial distribution of the agents. To our surprise, we have not observed any tendency of the model to get stuck in local optima. We now conclude the paper with an open-ended discussion about the current state of the model and what research directions we plan to explore in the future. Mapping the Cosmic web In hindsight, using a virtual physarum machine to reconstruct the Cosmic web seems like a very intuitive thing to do: MCPM has already provided a viable solution for several astronomical use cases [Burchett2020, Simha2020]. We continue to inquire into the implications of this possibility. This includes: • further ways to validate the model's performance in the presented tasks, as well as in new contexts; • continue the design pathway towards a more specific (accurate) variant of MCPM tailored for astronomy; • developing deeper mathematical foundations for these models; • topological analysis for graph extraction (directly applicable to catalog the Cosmic web filaments, but also in other tasks; see below). Figure 28: Experimental fit to a single page of the EAGLE dataset (about 1M data points representing quanta of gas, selected as the heaviest slice from the original 16M particles). MCPM here serves both as a reconstruction and a visualization method, providing a convenient estimate of the gas density. This applies also to using MCPM for examining and quantifying other astronomical datasets. One example is the EAGLE dataset [Schaye2015], which includes billions of simulated macro-particles (tracers) of the intergalactic medium but over a smaller high-resolution volume and including gas and the resulting complex hydrodynamical astrophysic. For a demonstration, in Figure 28 we fit to a single page (a sub-volume of a single redshift snapshot) of these data: a cubic region with roughly 4 Mpc across, i.e., two orders of magnitude smaller than the Bolshoi-Planck data. This is the smallest scale where the Cosmic web starts becoming apparent --on the edge of circumgalactic halos, which are well resolved in these data. The structures that MCPM finds here are intriguing, and further investigation involving observational data will show whether they are real and significant. Temporally varying data In this paper we have focused on static input data, and sought an equilibrium solution that best represents it. Of course, Physarum itself does not behave this way --it never truly halts its growth and remains in a constant state of flux during its plasmodium stage. This behavior happens in response to environmental changes around the organism. For MCPM, the above translates to handling dynamic, time-varying data. Cosmological inquiry routinely uses time-varying data, since the evolution of the universe can be approximately traced by matter halos (Section 7.4) that vary in position, mass and numbers. Beyond cosmology, such data are produced by models of population dynamics, traffic, epidemiology, neuronal networks and others. Figure 29: MCPM reconstruction of the brain connectome from a sparse representation (105 high-level nodes weighted by their mutual connectivity). The symmetry (or lack thereof) in the connectivity patterns is easily distinguishable, as well as the localization of the information 'flow'. As an illustrative example, we consider the problem of 3D edge bundling in complex network visualization [Holten2009, Bach2017, Ferreira2018]. We used MCPM to fit and visualize a human brain connectome [Xu2019], a set of 105 high-level nodes representing different sub-regions of the brain (so-called modes). To simplify this experiment ( Figure 29), we disregard the connectome edges and simply weigh each node by its overall connectivity strength. The result is a fuzzy activity map. In future, adding the edges as a directional guide for the agents could provide us with a more intricate spatial map with hierarchical grouping of connections and color differentiation between the modes. Currently, MCPM can be used for time-varying data if the temporal evolution of the dataset is matched with the model dynamics. For dynamic data with stored snapshots --such as the cosmological simulations --this can be achieved by fitting to a weighted combination of two adjacent snapshots, similar to keyframing in computer animation. In the more general case of dynamically evolving data (e.g., produced in real-time by a simulation), the sensing decisions of the agents would need to incorporate additional knowledge about the data, for instance by extrapolating the current state of the dataset forward in time. Biomimetic modeling Whether the polyphorm is a natural 3D generalization of the real-world Physarum's plasmodium is debatable. But undeniably the patterns generated by MCPM do have an organic quality to them. For instance, networks generated on low-discrepancy sampled point sets --even with spatially varying point density -resemble organic scaffoldings found in diatoms, corals, and bone marrow Figure 30. This could have applications in tissue engineering [Derby2012, Thrivikraman2019], either via 3D printing or other fabrication methods. 3D printing in general would benefit from organically inspired internal object scaffoldings, since the predominantly used periodic lattices do not lead to even stress distribution [Safonov2017]. MCPM could be applied in this context to generate aperiodic quasi-crystalline structures that mimic scaffoldings produced by topological optimization methods. Since MCPM is designed with the intent of being calibrated on reference data, we could find structurally sound parametrizations by fitting to topologically optimized scaffoldings, and then apply these efficiently during the preprocessing of each 3D print with sensitivity to their peculiar geometries. Beside scaffoldings, it would be very interesting to see whether other natural networks --like mycelium or even neuronal networks --could be modeled by specialized variants of MCPM. And similarly to the real organism, adding a repulsive gradient could lead to further and more expressive types of polyphorm geometries. Artificial networks Many kinds of constructed (abstract) networks are conveniently represented as graphs (social, logistic, dependency, decision, automaton, and other such networks). Some network optimization problems are also amenable to Physarum-based solutions [Sun2017]. Especially problems with a lot of inherent uncertainty like traffic routing and disease spreading could benefit from MCPM, thanks to its customizable probabilistic form. In Zhou et al. [2020] we investigate the application of MCPM to language embedding data. Language embeddings [Mikolov2013, Vaswani2017] are compressed high-dimensional representations produced by transformer neural nets. Their usefulness and predictive power is unmatched in natural language processing, as demonstrated by the recent GPT-3 model for instance [Brown2020]. Our work [Zhou2020] presents early evidence that embeddings could have an internal structure described by an optimal transport network, see the example in Figure 31. Figure 31: Auto-stereoscopic rendering of the MCPM fit to the Word2Vec language embedding data. All 300k word tokens contained in Word2Vec (representing the Wikipedia 2017 corpus) are projected from the original 512-dimensional embedding space down to 3D using UMAP (red clusters), and then reconstructed with MCPM (yellow-cyan colormap). The result reveals a complex polyphorm interconnecting the main central cluster, as well as a number of smaller, isolated clusters. The 3D structure can be seen by crossing eyes until the two sides merge. Given that MCPM can produce robust continuous networks up to three dimensions, we see a great potential in generalizing it to higher dimensions, which now becomes feasible thanks to the stochastic formulation and binary sampling stencil. Visualizing and automatically navigating high-dimensional latent spaces such as the language (and other kinds of) embeddings would be very beneficial, as insights into these increasingly popular compressed representations are still lacking. The self-organizing map is a topology-preserving dimensionality reduction algorithm: similar to UMAP, it projects high-dimensional training data to a low-dimensional space (typically 2D) while preserving neighborhood relationships. During the training process the map reorganizes itself by locally 'gravitating' towards clusters of data in the original high-dimensional space. The neural gas network operates similarly; however, its topology is not predetermined -instead, neural gas assumes the presence of a high-dimensional topological structure which it tries to adapt to. Compared to these methods, MCPM does not maintain explicit connectivity between the agents; instead, they flow freely in the domain and all information exchange is done through the deposit field. As a result, the topology of the fit is implicit in the continuous trace density estimate. Recovering the topology explicitly, e.g. as a graph would be valuable both in the context of the Cosmic web reconstruction (to provide a classification mechanism for the filaments and knots) but also outside of it; adapting the neural gas algorithm to work on density fields seems like a viable option here. The advantage MCPM offers over these methods is the continuous nature of the reconstructed network, permitting its interpretation as a density field [Burchett2020, Simha2020] or a throughput field [Zhou2020]. This is enabled by the core design of MCPM: the solution is not a fixed point in the search space, but rather, an average of the possible dynamic agent trajectories. MCPM as a complex dynamic system A noteworthy aspect of the model is that we often find the best-fitting polyphorms near tipping points of its dynamic behavior. This happens especially with regard to changing the scale-related parameters like the Sensing distance (SD). In a typical scenario we would start fitting with high values of SD, then gradually decrease it. That in turn decreases the size of reconstructed features, up to the point where the system reaches a supercritical state and 'dissolves' in Monte Carlo noise. That tipping point would usually be a local energy minimum. It would be interesting to further examine this aspect of the model more formally. Based on this, we speculate that MCPM could detect its tipping points, e.g., through a redesign of the loss function (Section 7.3). This could lead to a specialized physarum-based AI. Such an enhanced model would be able to locally modify its parametrization in order to fit to heterogeneous data. One possible path towards that would be to implement a spatial voting scheme where the agents would locally contribute information from their recent exploration of the domain and contribute their individual parameter estimates. Conclusion We have presented Monte Carlo Physarum Machine (MCPM), a stochastic computational model inspired by the growth and foraging behavior of Physarum polycephalum 'slime mold'. MCPM takes as its input a weighted point cloud in 2D or 3D, and produces a continuous transport network spanning the point data as an output. The output is represented by a scalar density field expressing the equilibrated spatial agent concentration, and we refer to this entity as a polyphorm to capture both its geometric quality as well as origin in a single term. The main contribution of this work with regard to its predecessor [Jones2010] is the probabilistic approach for modeling the agents' behavior. This makes the model configurable for a range of applications. The specific version of MCPM analyzed in this paper has been designed for the application in astronomy, aiming to make sense of large cosmological datasets and reconstruct the filamentary structure of the Cosmic web. Thanks to MCPM, we have made new discoveries relating to the structure of the intergalactic medium, which composes the Cosmic web [Burchett2020, Simha2020]. We will continue this line of inquiry and seek out other applications where MCPM can be of use, such as scientific visualization [Elek2021] and natural language processing [Zhou2020]. Our open-source implementation of MCPM is called Polyphorm (github.com/CreativeCodingLab/Polyphorm). Polyphorm runs primarily on the GPU, taking advantage of the inherent parallelizability of MCPM, and provides an interactive reconstruction and visualization experience. We invite the reader to use the software, import their own domain-specific data and enrich our understanding of where such a model could be meaningfully applied. There seems to be a degree of universality to MCPM --our experience so far has been that people with very different backgrounds have been able to identify with the generated polyphorm geometries in their specific ways. This is not surprising, since the agents approximately follow paths of least resistance. This energy-minimization property will likely be part of many natural or abstract systems. We believe that this work reinforces the possibility of Nature as an information processing entity --a computable Universe. The piece of evidence we contribute is that the small set of mathematically coherent rules presented here might apply to such different phenomena at vastly divergent scales.	v3-fos-license
2021-03-13T06:16:44.397Z	2021-03-11T00:00:00.000	232209735	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/s41598-021-85218-1.pdf", "pdf_hash": "16854a2b563715b0fbad40660d55d3db0ce6fa4f", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45383", "s2fieldsofstudy": [ "Biology" ], "sha1": "b623e93bb76d848bf1f16dc2759aea268f78fff4", "year": 2021 }	pes2o/s2orc	Pseudomonas fluorescens F113 type VI secretion systems mediate bacterial killing and adaption to the rhizosphere microbiome The genome of Pseudomonas fluorescens F113, a model rhizobacterium and a plant growth-promoting agent, encodes three putative type VI secretion systems (T6SSs); F1-, F2- and F3-T6SS. Bioinformatic analysis of the F113 T6SSs has revealed that they belong to group 3, group 1.1, and group 4a, respectively, similar to those previously described in Pseudomonas aeruginosa. In addition, in silico analyses allowed us to identify genes encoding a total of five orphan VgrG proteins and eight putative effectors (Tfe), some with their cognate immunity protein (Tfi) pairs. Genes encoding Tfe and Tfi are found in the proximity of P. fluorescens F113 vgrG, hcp, eagR and tap genes. RNA-Seq analyses in liquid culture and rhizosphere have revealed that F1- and F3-T6SS are expressed under all conditions, indicating that they are active systems, while F2-T6SS did not show any relevant expression under the tested conditions. The analysis of structural mutants in the three T6SSs has shown that the active F1- and F3-T6SSs are involved in interbacterial killing while F2 is not active in these conditions and its role is still unknown.. A rhizosphere colonization analysis of the double mutant affected in the F1- and F3-T6SS clusters showed that the double mutant was severely impaired in persistence in the rhizosphere microbiome, revealing the importance of these two systems for rhizosphere adaption. Results and discussion Distribution of T6SSs in the Pseudomonas fluorescens species complex. In silico analyses of the genomes of 134 strains belonging to the P. fluorescens species complex revealed that only 20% of them encode T6SS clusters (Table S3). The number of T6SS clusters in a single strain fluctuated from zero in P. fluorescens UK4 to three in P. fluorescens F113, and most strains contained two or three clusters (Table S3). Overall, we identified sixty-two complete T6SS gene clusters distributed mostly in three main phylogenetic clades. We referred to these three groups as 1.1, 3, and 4A ( Fig. 1) following the previous nomenclature 7, 21 . The distribution of the clusters in these three groups is the same as in P. aeruginosa 2,11 . No clusters were found corresponding to group 5, typical of Agrobacterium spp. and very few belonged to clusters 1.2 and 4B as in P. putida 38 . Each of these groups contains distinguishable genetic architecture and features (Fig. 2), as described in the next section. Pseudomonas fluorescens F113 contains three putative T6SSs. An extensive genomic analysis of the strain F113 using bioinformatics approaches (e.g., BLASTP, SMART, or Phyre-version 2) allowed us to identify a large number of T6SS-related ORFs (Tables S4-S7). Most of the genes are clustered in three genomic regions that we have named F1-, F2-and F3-T6SS (Fig. 2, Tables S4-S6). We identified a total of three hcp and eight vgrG genes,one of the hcp and six of the vgrG are orphan genes that were found scattered on the chromosome and not within the main T6SS clusters ( Fig. 2 and Table S7). Phylogenetic analysis revealed that similarly to P. aeruginosa PAO1, the F1-T6SS belongs to phylogenetic group 3, F2-T6SS to group 1.1 and the F3-T6SS to group 4A (Fig. 1). The large F1 cluster (44 Kb) contains a set of genes putatively encoding regulatory proteins previously described in P. aeruginosa: fha, tagF, pppA, ppkA, and tagRSTQ. Interestingly, it does not contain the gene that encodes the accessory component TagJ (Fig. 2) present in P. aeruginosa H1-T6SS 24,65 and conserved in other clusters from the same phylogenetic group (Group 3). In those clusters, tagJ is found located downstream hcp and upstream tssE, and it has been recently described as an accessory component that stabilizes the sheath from the baseplate 24 . Instead, in this location, P. fluorescens, contains the orthologues of tae4-tai4 from Salmonella typhimurium, which encode an EI pair (see below). A gene encoding a DUF1795 domain-containing protein is found downstream vgrG1a within the F1-cluster. The product of this gene is also known as an EagR chaperone for a downstream Rhs type effector first identified in Serratia marcescens 40 . Two Rhs-type effectors are encoded downstream vgrG1a and the eagR chaperone in F113 (Fig. 2) and are described in the following section. The F2-T6SS is a shorter cluster (around 29 Kb) with four genes putatively encoding regulatory proteins previously defined in P. aeruginosa 66 (stp, stk, fha2 and sfa2). All the core components are present within the cluster except for hcp2 (PSF113_1976) that is found in a different genomic region. A gene encoding a DUF4123 domain-containing protein chaperone 37,48 (Tap2) is found downstream vgrG2a followed by a hypothetical protein and a lipase, that are likely to correspond to an immunity-toxin pair ( Fig. 2 and Table S5). The F3-T6SS cluster, with an intermediate size of 35 Kb, does not contain genes encoding regulatory proteins ( Fig. 2 and Table S6), as opposed to P. aeruginosa, which contains sfa3, a gene encoding a putative sigma-54 transcriptional regulator enhancer. Interestingly, further down the T6SS cluster and three no-T6SS genes, we identified genes putatively encoding a Toxin Complex (TC) composed of one TcdB and two TccC subunits. The absence of the TcdA subunit, which forms an injection-like structure, has been observed in some strains 67 and it opens the possibility for this complex to be secreted through the T6SS. www.nature.com/scientificreports/ A total of eight vgrG genes are found in the chromosome of P. fluorescens F113. Three of these vgrG genes are within the F1-, F2-and F3-T6SS, whereas the other five genes are scattered over the chromosome (Table S7). A phylogenetic study of the eight VgrG proteins shows that VgrG1a and 1b clustered in the same phylogenetic group as P. aeruginosa VgrG1 proteins. On the other hand, P. fluorescens VgrG2a/b, VgrG4, and VGrG5a/b are closely related among them and cluster with P. aeruginosa VgrG4, 5 and 6 ( Figure S1). Identification of eight putative type 6 effectors in P. fluorescens F113. As stated before, genes encoding T6SS effector and their cognate immunity proteins are frequently found genetically associated to hcp and vgrG genes and in some cases to genes encoding T6SS chaperones or adaptors (i.e., EagR and Tap/Tec proteins) 15,36,37,48,68 . We have performed an in silico analysis that allowed us to identify a total of eight putative EI pairs in the proximity of F113 vgrG, hcp, eagR, and tap genes ( Fig. 2 and Tables S4-S7). These EI pairs have been named Tfe and Tfi standing for Type six F113 effector and immunity, respectively (Figs. 2, 3 and Tables S4-S7). www.nature.com/scientificreports/ Within the F1 cluster and downstream hcp1, we have identified the EI pair tfe1-tfi1. Tfe1 is orthologue to Salmonella typhimurium Tae4. Tae4 is a type VI amidase effector, which degrades the peptidoglycan of the cell wall and whose immunity protein is Tai4 43 . Furthermore, a gene encoding an EagR chaperone (eagR1) is found downstream vgrG1a in the F1-cluster and followed by two genes (tfe2 and tfe3) encoding Rhs effectors in a row. These effectors both have the same domain architecture: an N-terminal PAAR domain 69,70 , an intermediate www.nature.com/scientificreports/ conserved Rhs domain 71 confined by specific RVxxxxxxxxG and PxxxxDPxGL motifs, and a C-terminal region encoding the toxic domain. The toxic domain of Tfe2 is a putative RES domain, whose orthologue in P. putida is known to cause depletion of intracellular NAD + levels leading to inhibition of cell growth 72 (Fig. 3). In parallel, the C-terminal domain of Tfe3 carries a putative nuclease of the HNH/ENDO VII superfamily with a conserved WHH domain. tfe2 and tfe3 putative cognate immunity pairs, named tfi2 and tfi3, are found immediately downstream of the genes encoding their effectors and have 92 and 136 amino acids, respectively (Table S4 and Figs. 2, 3). Interestingly, tfi3 is a newly identified ORF in the P. fluorescens F113 genome and we have named the locus PSF113_5811.1 to indicate that the locus is located downstream PSF113_5811 (Table S4 and Fig. 2). Whereas Tfi2 immunity protein has no homologues or recognizable features, Tfi3 belongs to the Smi1/Knr4 family, which has been previously related to immunity proteins of polymorphic toxins 73 . Genes encoding putative effectors were also found within clusters F2 and F3, downstream vgrG2a and vgrG3 respectively and downstream orphan vgrG genes including vgrG1b, vgrG2b and vgrG5a (Fig. 2). The tfe4, tfe7, and tfe8 genes within the F2 and vgrG2b and vgrG5a operons, respectively, are genetically associated with genes encoding Tap chaperones (tap2, tap2b, and tap5a) (Fig. 2). Tfe4 is a lipase from class 3, putatively targeting eukaryotic and prokaryotic membranes by hydrolysing long-chain acyl-triglycerides from lipids into di-and monoglycerides, glycerol, and free fatty acids. Tfi4 is the cognate immunity protein of Tfe4 and it does not contain any known function or recognisable domain (Table S5). Tfe7 is a homologue of a ribosomal RNA large subunit methyltransferase D required for the full methylation of 23S ribosomal RNA. Interestingly, no gene encoding an immunity pair can be found linked to this effector gene, indicating that it might not be necessary for the producer strain. This effector could replace a functional methyltransferase and only be toxic in those strains where this enzyme is necessary for bacterial viability. A similar case has been described in P. aeruginosa PAO1, in which the effector Tse8 is not linked to an immunity pair and replaces a functional component of the transamidosome complex only present in some strains 74 . Tfe8 may represent a new type of effector identified for the first time in this work and contains a MatE domain. None of these three Tap-linked effectors contains a conserved N-terminal MIX motif considered a marker for T6SS effectors 75 and frequently found in Tap-associated effectors 38 . Gene tfe5 is located downstream vgrG3 and linked to a small PAAR-encoding gene. Tfe5 is a homologue of the TseF toxin from P. aeruginosa, an iron scavenging effector that interacts with PQS vesicles to help bacteria grow under very limiting conditions 76 . A small gene located between clpv3 and vgrG3 encodes a protein of 188 amino acids with no recognizable features but has a C-terminal domain of 50 amino acids that resembles a coronavirus RNA-binding domain according to a Phyre prediction (confidence: 80.5). The tfe6 is linked to the orphan VgrG1b cluster and has a similar genetic architecture to P. aeruginosa PAO1 VgrG1b cluster 53 . In PAO1 cluster, PA0095, PA0096, PA0097, PA0098, PA0099, PA0100, and PA0101 encode VgrG1b, an OB-fold, an immunoglobulin-like and a thiolase-like protein, a PAAR protein with a C-terminal cytotoxic domain, an immunity protein, and a heat repeat-containing protein respectively (Table S7). Pseudomonas fluorescens F113 vgrG1b cluster (PSF113_2885-PSF113_2890) only differs from the PAO1 one in the sequence of the toxic domain and the immunity pair, a common characteristic of this genetic island previously described by 53 . The specific function of Tfe6 and the mechanistic of the cognate immunity pair Tfi7 remains unknown. No putative effector-encoding genes were identified linked to hcp2, vgrG4, and vgr5b clusters and we hypothesised that they could be found somewhere else in the chromosome as orphan T6SS effectors as described before in other T6SS clusters, including Vibrio proteolyticus 77 among others. In summary, we identified eight putative T6SS effectors in the F113 genome. Three of them, Tfe2, Tfe3, and Tfe6, contain an N-terminal PAAR-domain (Figs. 2, 3) and are considered "specialised" effectors, whether the others that are not fused to any T6SS component are considered "cargo" effectors. F1-and F3-T6SSs genes are expressed in P. fluorescens F113. In order to determine the expression of genes encoding the components of the three T6SSs in P. fluorescens F113, we used a new dataset from an RNA-Seq study of F113 and derivatives grown under different culture conditions and in the alfalfa (Medicago sativa) rhizosphere. For this analysis, we chose the structural genes from each of the three systems (tssABCDEFGHI-JKLM). As shown in Fig. 4A, F1-and F3-T6SS cluster genes were significantly expressed under all tested conditions: growing in liquid cultures of Minimal Sucrose-Asparagine (SA) at exponential and stationary phases and colonising the alfalfa rhizosphere. On the contrary, F2-T6SS genes showed little, if any, activity under the same conditions. Expression of F1-T6SS genes was similar in all culture conditions regardless of the growth phase analysed, and these genes were also expressed during rhizosphere colonization, suggesting that the F1-T6SS is constitutively expressed in F113. The expression of F3-T6SS genes is lower than that of the F1-T6SS genes in all conditions. This situation is similar to P. putida, which contains three clusters named K1-, K2-and K3-T6SS, that do not belong to the same phylogenetic groups of P. fluorescens F113. In P. putida, the K1-T6SS is constitutively expressed under laboratory conditions and is active in vitro and in planta assays 38 ; however, no activity has been reported for K2-and K3-T6SS clusters to date. Conversely, in the case of P. aeruginosa that harbours the same groups as F113 and share regulatory components within the T6SS clusters, T6SS genes are not constitutively expressed and all of them are tightly regulated at transcriptional, post-transcriptional, and post-translational levels 11,[78][79][80] . In PAO1, the three systems have the common regulator Rsm 52 and the widely studied H1-T6SS is known to also be regulated by many other factors including RetS, GacA/S, TagF, PpkA/PppA, TagQRST, and FHA 2,33,34,[80][81][82] . To study the regulation of T6SSs during the process of rhizosphere colonization, we also analysed the T6SSs genes with differential expression in RNA-Seq data from the P. fluorescens F113 amrZ and fleQ mutants compared to the wild-type strain grown in the rhizosphere of alfalfa. AmrZ and FleQ are two transcription factors (TFs) crucial during competitive rhizosphere colonization in this bacterium [83][84][85][86][87] www.nature.com/scientificreports/ and it has been previously linked to the control of T6SS in P. aeruginosa 52 . FleQ belongs to the NtrC/NifA family and has been related to the regulation of T6SS and c-di-GMP metabolism in P. putida 89 . The differential gene expression analyses shown in Fig. 4B has revealed that in the rhizospheric environment, AmrZ functions as a negative transcriptional regulator of F1-and F3-T6SS, while FleQ acts as a positive regulator. This antagonist role of both transcriptional regulators fits with the proposed model for the AmrZ/FleQ hub, which has been proposed in F113 to act as an oscillator with opposing effects in gene expression, in order to integrate the bacterial responses to the environment 83 . In addition to the vrgG genes related to the F1-and F3-T6SS, most orphan vrgG genes are also negatively regulated by AmrZ under the tested conditions. However, FleQ can act both as a positive and negative regulator of certain vgrG genes. These results show that both AmrZ and FleQ regulate T6SSs in F113, as previously shown for AmrZ in P. aeruginosa 52 and FleQ in P. putida 89 . P. fluorescens F113 F1-and F3-T6SS are implicated in bacterial killing. T6SS is a critical element in the antibacterial activity of some strains due to the injection of T6SS toxins into competitor target cells 5,7,30,42 . Therefore, to analyse the role of T6SS in P. fluorescens F113 in inter-bacterial competition, we performed bacterial killing assays as previously described 38 . In these assays, P. fluorescens F113 and its isogenic mutants were used as predators, and Escherichia coli harbouring a pK18mobsacB plasmid, containing the lacZ gene that confers blue colour to the colony in the presence of X-gal, was used as prey. Bacteria were mixed in a 1:1 ratio (predator:prey), co-cultured for 5 h and a sample of each mix was grown on selective plates for 48 h. Additionally, serial dilutions of the different assays were plated on selective media for predator and prey quantification. The tssA genes of P. fluorescens F113 were selected as targets for the construction of insertional mutants. Mutants affected in each of the three systems were used as predators: tssA1 − , tssA2 − and tssA3 − for F1-, F2-and F3-T6SS, respectively. Additionally, a double mutant tssA1 − /tssA3 − was constructed and used as predator. In the bacterial competition assays (Fig. 5), the wild-type strain was able to kill E. coli cells efficiently as observed for the significant prey survival reduction. However, single and double mutants in the structural genes tssA1 and tssA3 (F1-and F3-T6SSs) were affected in E. coli killing (Fig. 5), showing a survival rate of the prey similar to the control without predator. These results indicate that both systems are functional and have bactericidal activity, as www.nature.com/scientificreports/ has been shown before for other T6SSs in pseudomonads 4,90,91 . By contrast, the tssA2 mutant (F2-T6SS) exhibited the same capacity to outcompete E. coli that the wild type strain, indicating that the F2-T6SS is not involved in the antibacterial activity of P. fluorescens F113 against E. coli under our experimental conditions. It is interesting to note that we have not observed expression of the genes encoding this system in any of the tested conditions, suggesting that the lack of function might be a consequence of lack of expression and therefore, the F2-T6SS could be functional under other yet-unknown conditions. Lack of expression and/or antibacterial activity has been observed in other pseudomonads in laboratory conditions, like P. aeruginosa wild type strain 2,52 and P. putida K2-and K3-T6SS 38 . Pseudomonas fluorescens F113 T6SSs are important for the adaption to the rhizosphere microbiome. It has been described that the T6SSs have a role in modulating and shaping the natural microbiota and, in the case of plant-associated bacteria, these weapons are of interest for bacterial persistence in the plant niche, reviewed in 92 . Also Vacheron et al., 2019, showed that T6SS of P. protegens contributed to the invasion of the gut microbiome of an insect. Therefore, we wanted to know whether F1 and F3 T6SS could play a role in F113 competence in the rhizosphere. We inoculated 7-day-old tomato plants growing in agricultural, non-sterile soil microcosms, with the wild type strain and the double mutant F1 − /F3 − strain. Bacteria from microcosms were isolated 2 weeks after inoculation and F113 derivatives were selected by using their rifampicin marker resistance. As shown in Fig. 6, when the inoculation is done with the double mutant strain, there is a significant (90%) reduction in the bacterial recovery after rhizosphere colonization compared with the wild-type strain. This result shows that the two tested T6SSs play a relevant role in invading, establishing and/or persisting in the tomato rhizosphere microbiome. The effect of these T6SSs in microbiome adaption is likely due to their role in the interbacterial competition that might confer F113 with the capacity to outcompete foes. Conclusions Three T6SS clusters (F1-, F2-and F3-T6SS) and eight effectors (Tfe1, 2, 3, 4, 5, 6, 7, 8) together with 6 immunity proteins (Tfi1, 2, 3, 4, 6 y 8) have been identified in P. fluorescens F113. At least two of these systems, F1-and F3-T6SS are functional in laboratory conditions and more importantly in the rhizosphere and possess bactericidal activity. The systems are crucial elements in the colonization of the rhizosphere, most likely providing by providing F113 with the capacity to fight competitors from the rhizosphere microbiome. coli growth for each competition experiment. Blue colour in bacterial patches on LB plates supplemented with X-gal and kanamycin indicates E. coli growth. Error bars indicate the mean ± s.d. of three biological replicates, and significance was calculated using ANOVA test (*P < 0.01); ns, not significant differences when compared to non-competing E. coli. Bioinformatic analyses. Pseudomonas gene sequences were obtained from the Pseudomonas Genome database 95 . BLASTP analyses were performed at the NCBI website 96 and amino acid sequence searches using SMART 97,98 . The Protein Homology/analogy Recognition Engine (Phyre 2 ) server was used to perform structural-base homology prediction 99 . The phylogenetic tree was constructed using MEGA X 100 . PSORTb v 3.0.2 software was used to predict subcellular location of proteins 101 , TMHMM server v.2.0 to predict transmembrane domains 102 , and SignalP to predict signal peptides 103 . The molecular biology tools included in Benchling platform were used to identify novel open reading frames (ORFs) (Benchling, Inc., https ://bench ling.com/acade mic) and the putative coding proteins were run using NCBI BLASTP tool to determine the degree of conservation. Construction of T6SS mutants. Insertional single mutants in PSF113_2422 (tssA3), 5797 (tssA1) and 5829 (tssA2) genes were constructed by electroporation of the plasmid vector pCR2.1-TOPO (Invitrogen) carrying an internal region of the corresponding gene (400 bp ca.), in the case of the double mutant tssA1 − /A3 the plasmid vector pG18mob2 was employed. Single and double recombinant mutants were selected by kanamycin (25 μg ml −1 ) and gentamycin (4 μg ml −1 ) resistance in SA medium and checked by PCR and Southern Blot. Primers used are listed in Supplementary Table S2. Bacterial RNA isolation from the rhizosphere. Rhizosphere colonization of alfalfa (Medicago sativa var. Resis) plants was performed essentially as described in 104 . Seven-day-old seedlings growing in Falcon tubes, using vermiculite as substrate, were inoculated with 1 mL (1 × 10 8 CFU) of P. fluorescens F113 or derivatives amrZand fleQmutants. RNA was extracted seven days post-inoculation. Aerial parts of alfalfa plants were removed, 4 mL of Phosphate-buffered saline (PBS) and 6 mL RNAlater 105 added and tubes were vortexed for 2 min to resuspend bacterial cells. The mix of the 32 preparations per sample was filtered through four layers of sterile muslin cloth in pyrex funnels and separated into six 50 mL tubes. The filtrate was centrifuge 1 min at 1000 rpm and 4 °C. The supernatant was transferred to a sterile tube and centrifuged at 10,000 rpm for 20 min and 4 °C. Supernatants were discarded and pelleted cells dried before liquid nitrogen freezing. RNA isolation was performed as indicated in 84 . Bioinformatic RNA-Seq data processing. After sequencing the RNA, the quality of the raw reads was checked using FastQC. Then, sequence reads were clipped and filtered using Trimmomatic v 0.35 106 to remove www.nature.com/scientificreports/ chaperones and low-quality nucleotides, defining a four nts sliding window with an average phred quality of 15 and 50 nts as minimum read length. High-quality reads were directly used for transcript-level quantification using Salmon software 107 , which performs a quasi-mapping to the new annotation of the P. fluorescens F113 CDSs (GenBank: NC_016830) and transcript quantification. Normalization of counts and differential gene expression was calculated with DESeq2 1.24.0 R package 108 . Differential gene expression comparisons were made setting a threshold for log 2 fold change (mutant/wild-type) of ≤ − 1/≥ 1 and a stringent p-adjusted value cutoff ≤ 0.001. RNA-Seq reads have been deposited in to the NCBI Sequence Read Archive database and it is available under the BioProject PRJNA419480: BioSamples accessions: F113 culture in exponential growth phase (SAMN17839758 and SAMN17839763), F113 culture in stationary growth phase (SAMN17839759 and SAMN17839764), F113 from the rhizosphere (SAMN17839757 and SAMN17839762), F113 amrZ mutant from the rhizosphere (SAMN17839760 and SAMN17839765) and F113 fleQ mutant from the rhizosphere (SAMN17839761 and SAMN17839766). Interbacterial competition assays. Competition assays to assess T6SSs were performed on solid media plates according to the previously reported protocol 51 . Pseudomonas fluorescens F113 or isogenic derivatives (predator) and E. coli DH5α containing pK18mobsacB (prey) were grown overnight in LB medium. Next morning, each culture was adjusted to OD 600 of 1.0. 100 μL of predator and 100 μL of prey strains were mixed and co-cultured for 5 h at 200 rpm and 28 °C. 20 μL of each culture was spotted onto LB-agar supplemented with 5-bromo-4-chloro-3-indolyl-D-galactopyranoside (X-gal) and kanamycin and incubated at 28 °C. Additionally, serial dilutions of the different assays were plated to quantify the number of CFUs in each of them. Three biologically independent experiments were performed. Rhizosphere colonization analysis. Tomato seeds (Rebelion F1, Vilmorin, France) were sterilized with 70 % ethanol and 5 % hypochlorite, washed and germinated on sterile 1.0 % agar plates for 24 h at 28 °C. Seedlings (one per tube) were then transferred into sterile 50 mL Falcon tubes containing 16 g of a mix 1:1 of agricultural soil and sterile sand (Merck KGaA, Darmstadt, Germany). Each tube was embedded with 3 mL of autoclaved water, and incubated in a controlled environment room (25 °C, 16 h light cycle). After seven days, each tube/plant was inoculated with 1 mL (1 × 10 6 CFU/mL) culture of either the P. fluorescens F113 wild type strain or the double mutant (tssA1 − /A3 − ) strain. Plants were grown for another two weeks, after which aerial parts were removed; 10 mL of saline solution (0.85 %) was added to each tube and vortexed thoroughly to resuspend the bacteria. After decantation of the soil and sand matrix, serial dilutions of the supernatant were plated onto SA plates supplemented with rifampicin (100 μg mL −1 ) and nystatin (50 μg mL −1 ). Assays were performed with six plants per condition tested and non-inoculated plants were used as a negative control. Statistical analyses. The normal distribution of data was checked with Shapiro-Wilk test. When distribution was normal, the data were analysed with one-way ANOVA test, where multiple pairwise-comparisons between strains were performed with Tukey HSD test. When normality was not met, data were analysed with the Kruskal-Wallis non-parametric test. All data were analysed with R package version 3.6.3 109 .	v3-fos-license
2018-04-03T00:41:41.100Z	2016-01-06T00:00:00.000	17859557	{ "extfieldsofstudy": [ "Biology", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0145812&type=printable", "pdf_hash": "f99151ab0ce1f8d920fec0fbc319b0ce707f031a", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45384", "s2fieldsofstudy": [ "Biology" ], "sha1": "f99151ab0ce1f8d920fec0fbc319b0ce707f031a", "year": 2016 }	pes2o/s2orc	Discovery of New Compounds Active against Plasmodium falciparum by High Throughput Screening of Microbial Natural Products Due to the low structural diversity within the set of antimalarial drugs currently available in the clinic and the increasing number of cases of resistance, there is an urgent need to find new compounds with novel modes of action to treat the disease. Microbial natural products are characterized by their large diversity provided in terms of the chemical complexity of the compounds and the novelty of structures. Microbial natural products extracts have been underexplored in the search for new antiparasitic drugs and even more so in the discovery of new antimalarials. Our objective was to find new druggable natural products with antimalarial properties from the MEDINA natural products collection, one of the largest natural product libraries harboring more than 130,000 microbial extracts. In this work, we describe the optimization process and the results of a phenotypic high throughput screen (HTS) based on measurements of Plasmodium lactate dehydrogenase. A subset of more than 20,000 extracts from the MEDINA microbial products collection has been explored, leading to the discovery of 3 new compounds with antimalarial activity. In addition, we report on the novel antiplasmodial activity of 4 previously described natural products. Introduction Malaria is widespread in tropical and subtropical regions, including parts of America, Asia and Africa. An estimated 3.2 billion people are at the risk of suffering malaria and from one-half to one million deaths were reported in 2014 (World Malaria Report. 2014). In 2014, 97 countries and territories had ongoing malaria transmission. Most deaths from malaria are caused by Plasmodium falciparum, one of the five species of human infectious malaria parasites. The increasing resistance of P. falciparum to the available drugs [1] and new efforts to eradicate malaria all drive the need to develop new, effective and affordable antimalarial agents. Despite the development of new technologies to study resistance acquisition [2][3][4] and our increasing understanding of P. falciparum biology, few new drug targets have been clinically validated. At present, there are only four classes of valid antimalarial compounds: quinine or other aminoquinolines, antifolate compounds, artemisinin derivatives, and the hydroxyl napthoquinone atovaquone. This lack of structural diversity denotes a need to explore other sources of structures, and natural products from microorganisms render a unique chemical space for this purpose. Natural products are one of the most important sources for new chemical scaffolds. They have been largely exploited in the discovery of new drugs, and around 60% of the drugs available nowadays derive directly or indirectly from natural products [5,6]. Many of the antibiotics or drugs in use such as camptothecin, lovastatin, maytansine, paclitaxel, reserpine and silibinin are all natural products. Some of the first-line malaria treatments currently used are isolated from plants, such as artemisin and quinine. On the other hand, microbial natural products have been underexplored in this field, although they offer great advantages for the potential discovery of novel bioactive products and the possibility of large-scale production. Unfortunately, to date, natural product libraries have not been extensively used in the search for new antimalarials in large-scale campaigns using high throughput screening (HTS) [7,8]. Drug discovery through HTS allows the large-scale testing of potentially active products, accelerating the identification of molecules for further development. There are several methods for detecting erythrocyte infection and drug susceptibility. However, not all of these assay formats are suitable for HTS due to several factors such as cost, safety, assay stability, equipment availability and quality of data produced. Frequently, methods for HTS technology are based on the measurement of DNA content in strains of malaria parasites using SYBR Green [9], GFP [10], and 4',6'-diamidino-2-phenylindole [11], or in a stably expressed cytoplasmic firefly luciferase parasite strain (3D7-luc) [12,13]. Nevertheless, since its description [14], the lactate dehydrogenase (LDH) assay has been increasingly used for Plasmodium growth determination, due to its robustness and specificity. PfLDH activity measurements, which are proportional to culture parasitaemia, provide specificity through the use of 3-acetylpyridine adenine dinucleotide (APAD) as cofactor, since the human homologue present in red blood cells carries out this reaction at a very slow rate in the presence of this cofactor instead of NADH. In the present work, we have screened more than 20,000 natural extracts from the MEDINA collection against P. falciparum using the assay based on LDH activity. This is the first time that this screening approach has been applied directly to the study of natural extracts from a high diversity of microorganisms. Using this methodology, we have identified 7 compounds with antimalarial activity. Three are new/novel structures of which two have been previously described as a result of this screening [15,16] while pepstatin K is reported herein for the first time. Four are known compounds whose antimalarial properties had not been previously reported. All these findings provide an encouraging starting point that supports a renovated interest in discovering and optimizing novel antimalarial compounds from microbial natural products. Materials and Methods No specific permissions were required for the collection of samples in the Vallibierca valley, Huesca, Spain because Spanish legislation does not regulate the access to soils in public areas (since it is neither a National Park nor a private owned land). We confirm that the studies involve only soil samples and these samples do not involve endangered or protected species. Preparation of P. falciparum cultures for assay The parasite P. falciparum 3D7 cultures were synchronized using 5% sorbitol as previously described [17] and 96 h later, the level of parasitaemia was determined by light microscopy counting of a minimum of 500 erythrocytes on a Giemsa-stained thin blood smear. Parasites were noted to be late-ring and early trophozoites. The stock culture was then diluted with complete medium and normal human erythrocytes to a starting 2% hematocrit and 0.25% parasitaemia in 25 μL of volume for 384-well plates (LDH assay) and 5% hematocrit and 0.5% parasitaemia in 90 μL for 96-well plates (fluorescent assay). Both were incubated at 37°C for 72 h and then frozen for at least 24 h. LDH assay conditions The extracts, fractions and pure compounds were evaluated in 384-well plates after 72 h of incubation. Each plate also included positive growth controls, where only medium was added, and negative growth controls with 100 nM of chloroquine. Plates were thawed at room temperature for at least 1 h. To evaluate LDH activity, 70 μL of freshly made reaction mix, containing 143 mM sodium l-lactate, 143 μM APAD, 178.75 μM NBT, 1 U/mL diaphorase, 0.7% Tween 20 and 100 mM Tris-HCl (pH 8.0) were dispensed into plates. Plates were shaken to ensure mixing and absorbance was measured at 650 nm after 10 min of incubation at room temperature. Absorbance was determined with a microplate reader VICTOR2 Wallac spectrofluorometer. The read time for each plate was 3 minutes. Plates were prepared at 3.5 minute intervals and read in a sequential order in the plate reader. This method gives a signal to noise ratio of 10 under the conditions used [18]. With the 384-well plates, integrity of erythrocytes and LDH activity can be inspected visually, allowing for the rapid detection of dispensing errors and interferences by extracts. Yellow wells contain active inhibitors where parasite growth has been abolished and the LDH reaction has been inhibited. The EC 50 for chloroquine obtained this way was 8.72 ± 0.29 nM (S1 Fig), which was similar to previously reported values [19][20][21]. SYBR Green assay conditions For the fluorescence assay, after 72 h of growth, 100 μL of SYBR Green I in lysis buffer (0.2 μL of SYBR Green I/mL of lysis buffer containing 20 mM Tris-HCl pH 7.5, 5 mM EDTA, 0.008% saponin and 0.08% Triton X-100) was added to each well, and the contents were mixed. After 2 h of incubation in the dark at room temperature, fluorescence was measured with a Spectra MAX GEMINI EM microplate reader (Molecular Devices) with excitation and emission wavelengths of 485 and 530 nm, respectively. Absorbance measurement, visual inspection and statistical analyses The inhibition percentage of each extract was determined by the equation: Where A neg is the optical density of the negative control at 650 nm and A pos is the optical density of the positive control at 650 nm. Data were analyzed using the Genedata Screener program, Condoseo module (Genedata AG, Switzerland). An extract was considered to have activity when the percentage of growth inhibition was higher than 70%. The Z' factor predicts the robustness of an assay by taking into account the mean and standard deviation of both positive and negative assay controls. The robust Z' factor (RZ' factor) is based on the Z' factor, but standard deviations and means are replaced by the robust standard deviations and medians, respectively. In all experiments performed in this work where a minimum of 700 384-well plates have been used, the RZ' factor obtained was between 0.7-0.8. Microbial extracts collection For the primary screening campaign, a subset of 20,000 microbial extracts from different modules of the MEDINA natural products collection (Table 1) was used. The microbial extracts were obtained from bacterial and fungal strains cultivated in different nutritional conditions and extracted with acetone (1:1) for 1 h in an orbital shaker. Extracts were then centrifuged at 1500 xg for 15 min and the supernatant concentrated to half the volume in the presence or not of a final concentration of 20% DMSO. Extracts were stored at -20°C in 96-well ABgene v-bottom plates until needed. Primary screening and dose-response experiments All the extracts were first screened against P. falciparum and those exhibiting over 70% growth inhibition were selected as actives and confirmed in the LDH and SYRB Green assays. The extracts selected from this stage were then tested in a five-point dose-response assay, using as the first titration point the dilution identified as active in the primary screen and then performing four subsequent 2-fold serial dilutions. Extracts that exhibited dose-responses indicative of good potency were selected for de-replication by tandem liquid chromatography mass spectrometry (LC-MS). LC-MS analyses were carried out as described previously [22]. Bioassay-guided extract fractionation Extracts with LC-MS profiles suggestive of containing active novel compounds were selected to confirm the activity from a 100 mL regrowth of the producing strain in the same production conditions [22,23]. Hits containing compounds of interest detected in the previous step were regrowth at 1 L scale. Extraction with acetone and a first chromatographic separation using SP-207ss brominated polystyrenic resin was performed as previously described [16]. Active fractions from this first chromatographic step were subjected to one or several steps of preparative and semi-preparative reversed phase HPLC on a Gilson GX-281 apparatus until a pure compound was obtained. The chromatographic column, solvent system and gradient conditions for each HPLC separation were selected depending on the particular compound of interest contained in each sample. Isolation of pepstatins A and K The compounds were isolated from 1 L growth of Kitasatospora mediocidica F-136,264. The producing strain was isolated from a soil sample collected under a Juniperus communis tree collected in the Vallibierca valley, Huesca, Spain. A seed culture of strain F-136,264 was prepared in a inoculum medium (soluble starch 20 g/L, dextrose 10 g/L, NZ amine EKC (Sigma) 5 g/L, Difco beef extract 3 g/L, Bacto peptone 5 g/L, yeast extract 5 g/L, and CaCO 3 1 g/L, adjusted to pH 7.0 with NaOH before addition of 1g/L CaCO 3 ), at 28°C with 220 rpm orbital shaking. A 5% (v/v) of the seed culture was used to inoculate each of the seven 500 mL flasks containing 150 mL of the production medium (corn dextrin 20 g/L, beta cyclo dextrin 10 g/L, tomato paste 20 g/L, Bacto yeast extract 10 g/L, CoCl 2 .6H 2 O 5 mg/L), and the flasks were incubated at 28°C for 7 days in a rotary shaker at 220 rpm and 70% humidity before harvesting. The 1 L culture was extracted with acetone (1 L) under continuous shaking at 220 rpm for 3 h. The mycelium was then separated by centrifugation and the supernatant (ca. 2 L) was concentrated to 1 L under a stream of nitrogen. This solution was loaded (with continuous 1:1 water dilution, discarding the flow-through) on a column packed with SP-207SS reversed phase resin (brominated styrenic polymer, 65 g) previously equilibrated with water. The column was further washed with water (1 L) and afterwards eluted at 8 mL/min on an automatic flash-chromatography system (CombiFlashRf, Teledyne Isco) using a linear gradient from 10% to 100% acetone in water (in 12.5 min) with a final 100% acetone step of 15 min, collecting 9 fractions of 20 mL. LC-MS analysis allowed the identification of pepstatin A [24] and a new member of this family of compounds that we designated as pepstatin K in the bioactive fraction. This fraction was further purified by reversed phase semipreparative HPLC (Agilent Zorbax SB-C8, 9.4 × 250 mm, 7 um; 3.6 mL/min, UV detection at 210 nm) with a linear gradient of water-CH 3 CN of 5% to 100% CH 3 CN over 37 min to yield pure pepstatin A (11 mg) and pepstatin K (8 mg). Optimization of assay in a 384-well format For the primary screening of a subset of the MEDINA collection, the PfLDH assay was used. This methodology has been previously reported to be adequate for a 384-well plate format in HTS [18], being a robust, sensitive, selective and reproducible assay. We adapted the method to our laboratory conditions and facilities. Firstly, we tried different hematocrits and parasitaemias including those described in the above-mentioned paper. However, while conditions such as 5% haematocrit and 0.5% parasitaemia gave rise to limited linearity of the assay, a 2% haematocrit and 0.25% parasitaemia provided a perfect correlation between growth and signal. In addition, we tested different concentrations of diaphorase ranging from 2.83 U/mL to 0.5 U/ mL, not observing differences in the slope of the curve. On the other hand, concentrations below 0.5 U/mL gave lower rates (Fig 1A) and, thus, the final concentration of diaphorase to be used in the screening was established at 1 U/mL. The SYBR Green protocol was used as a confirmatory counter-assay in a 96-well plate format [25]. This assay quantifies DNA (number of parasites) and thus allows to disregard extracts that interfere or are inhibitors in the LDH assay. Other fluorescent DNA dyes such as YOYO and the mixture of YOYO and SYBR Green, a combination previously used in HTS campaigns [26], were also tested (data not shown). MEDINA's extract collection comprises various modules, characterized by the diversity of the microbial strains represented (mainly bacteria, actinomycetes and fungi), the complexity of the natural product extracts (whole broth crude extracts, SPE extracts or fractions) and the different extraction procedures. We defined the sample dilutions to be applied to the different modules of the collection in the assay according to the percentage of positive hits obtained, to achieve a representative number of hits that could be adequately handled in terms of scale up capabilities. In order to adjust the most appropriate assay dilution of the original collection, we tested crude extracts from actinomycetes and from fungi at three different concentrations. The desired hit rate per 384-well plate was approximately 2.5% when applying a cut-off value for positives of 70% growth inhibition. In both cases, the hit rate decreased with increasing dilutions. Extracts from actinomycetes showed a higher hit rate than extracts from fungi and consequently the dilutions set for these extracts were 1/200 and 1/50, respectively (Fig 1B). Certain modules of the MEDINA extract collection contain 20% of DMSO, a solvent that inhibits Plasmodium growth at concentrations as low as 0.001%. To avoid side effects due to DMSO, aliquots of these extracts were previously evaporated, thus eliminating all the DMSO present, and re-dissolved in methanol, a solvent which is less aggressive against Plasmodium cultures and does not affect parasite growth even at 1%. Screening campaigns, identification of actives and LC-MS de-replication The MEDINA microbial natural product collection is composed of different modules. Initially, we focused our studies on a subset of 11,124 extracts from module A, containing both fungi and actinomycetes extracts in aqueous solution. A second subset of 8,560 extracts was selected from module B, containing fungal and bacterial extracts from previously selected strains known to produce antimicrobial activities. Extracts from this module contained 20% DMSO, and required a total evaporation followed by re-dissolving in methanol prior to the assay. Lastly, a third subset was selected from module C containing 560 extracts from bacteria of the classes Proteobacteria and Bacteroidetes in 20% DMSO. Since the assay dilution applied in this latter case was significantly high (1/3000), elimination of DMSO was not necessary (Table 1). Potential hits obtained in the primary screening were tested again in triplicate. We established that 70% of the hits obtained in the primary screen were confirmed to be active when assayed again in triplicate in an independent assay. The confirmed hits were then subjected to dose-response curves of at least 8 points obtained from serial dilutions of the extracts and using the PfLDH assay to establish potency. Extracts with a capacity to inhibit growth of at least 70% when diluted 8 times were then assayed in dose-response experiments with the secondary SYBR Green assay (Fig 2). While interference in fluorescent intensity assays in general by natural product extracts is well documented [27,28], here the SYBR Green assay is used as a counter screen and it is unlikely that extract interference is an issue in both absorbance and fluorescence based assays. Therefore, all the extracts corroborating the inhibition in this assay were considered as confirmed positive hits of the primary screening. As new antimalarial compounds were sought, and in order to avoid the growth of extracts with known toxic compounds, all the confirmed hits obtained in the primary screening were subjected to LC-MS analysis, which allows the detection and identification of the main components of the extract. The known compounds detected by LC-MS are shown in Fig 3. A database search was performed using an in-house developed application, which matches UV-LC-MS data of metabolites in the active extracts to UV-LC-MS data of known metabolites stored in our proprietary database obtained using the same LC-MS conditions [23]. The results are presented taking into account the source of the extract, actinomycetes or fungi. In the case of actinomycetes, antibiotics such as nigericins, oligomycins and actinomycins, among others, were identified. Nigericins are ionophores that catalyze the electroneutral exchange of K + for H + [29], whereas oligomycins and actinomycins act at the level of mitochondrial function and transcription, respectively [30,31]. Many of the known compounds identified are produced by species of the genus Streptomyces. A total of 11% (actinomycetes) and 25% (fungi) of the hits correspond to known compounds that were detected at low frequency in the positive extracts. In the case of fungi, several known compounds such as the leucinostatins were identified, which have been shown to exert their cytotoxic action by perturbing mitochondrial oxidative phosphorylation [32]. Up to 45% of the hits correspond to extracts in which no known compounds were identified using MEDINA's proprietary database. After de-replication of extracts with known compounds, 60 extracts of module A, 150 of module B and 21 of module C were chosen for further characterization. Nineteen extracts from actinomycetes and 28 from fungi from module B were chosen for further scale-up according to the potency and the taxonomy of the producing strains. In the case of module C, since the data of cytotoxicity in three different human cell lines were available, only those extracts with no previously described cytotoxic effect were subjected to small scale growth, thereby reducing the number of extracts from 21 to 5. Small scale growth for bioassay-guided fractionation After primary screening and hit selection, a critical step was the reproduction of the activity observed against P. falciparum in a new growth of the microorganism, assuring that the compound responsible for growth inhibition is present in the new extract. This confirmation was carried out in two phases, first a small scale growth (100 mL volume) was performed and, after confirmation of the bioactivity, a second large scale growth (1 L) was carried out to generate enough material for the isolation of the active molecule. After small scale growth, the extracts generated were assayed in dose-response experiments to confirm inhibition of Plasmodium growth of more than 70%. Once the activity was confirmed, the positive samples were subjected to semipreparative HPLC fractionation, and the resulting fractions (80 fractions per sample) were analyzed for antiplasmodial activity. Plates were assayed at different concentrations to delimit the fractions containing the highest percentage of growth inhibition and, thus, identifying those that contain higher amounts of the most potent bioactive components. Once defined, the presence of known compounds was again established by LC-MS and LC-HRMS. After growth and fraction analysis, 17 extracts were further confirmed as containing potentially novel compounds and therefore selected for scale-up growth and isolation of the bioactive components. Growth scale-up and bioassay-guided purification After verification of the antimalarial compound production in the small scale growth, the producing strain was cultivated in higher volumes (1 L), in order to provide enough sample for purifying the compounds responsible for the antimalarial activity. After this growth step, the extracts produced were first subjected to a low-resolution chromatographic step on SP-207ss resin generating 8-10 fractions, each of which was evaluated in a dose-response manner against P. falciparum. The most active fractions were then separated into several fractions by semipreparative or preparative HPLC, in a similar way to the procedure followed for the small scale regrowth. Fractions and subfractions were assayed for antiplasmodial activity and pure compounds responsible for the activity were re-purified from the most active fractions through semipreparative HPLC. Identification of pure compounds Although not initially identified, final purification allowed for the identification of four known bioactive fungal compounds: aselacin A, petasol, sporidesmin A and conglobatin. In addition, three new compounds were identified. One is pepstatin K, a new peptide belonging to the pepstatin family containing two units of the unusual γ-amino acid statin ((3S,4S)-4-amino-3hydroxy-6-methylheptanoic acid) produced by the Kitasatopora mediocidica F-136,264. On the other hand lasionectrin produced by Lasionectria sp. CF-176994, a naphtopyrone first discovered in this screening and which has been recently purified from fungal extracts [15] and a novel fungal betaine lipid MDN-0104 produced by Heterospora chenopolii CBS109836 were obtained [16]. We have previously reported the isolation and structures of these two latter compounds [15,16]. Additionally two new cyclic peptides and a new tetramic acid are presently under evaluation and will be published elsewhere. The structure and EC 50 obtained for each compound is indicated in Fig 4. Structural characterization of pepstatin K A pseudomolecular ion at m/z 720.4555 obtained by ESI-TOF analysis established a molecular formula of C 37 H 61 N 5 O 9 for pepstatin K (calc. for C 37 H 62 N 5 O 9 + 720.4542). Analysis of the 1D ( 1 H and 13 C) and 2D (COSY, HSQC and HMBC) NMR spectra of the compound (Table 2) revealed the presence in the molecule of several amino acid units, including two valine, one alanine and two statine residues. Additionally, several signals in the low field region of the 1 H NMR spectrum and the presence of six signals in the sp 2 region of the 13 C NMR spectrum accounted for the presence in the molecule of a phenyl containing residue that was eventually identified as 2-phenyletanoic acid (PE) based on correlations observed in the HMBC spectrum. The sequence of amino acid residues was partially established by analysis of key cross-peaks observed in the HMBC spectrum and MS/MS fragmentation (Fig 5). Thus, a correlation between the proton at δ H 4.31 and a carbon at δ C 173.8 was indicative of the partial sequence Discussion The screening described herein has proved to be a valuable and exploitable tool in the search for new antimalarials. Due to the growing need for new drugs to treat this disease, exploring new sources of chemical scaffolds could contribute to the identification of novel drugs with new modes of action. Chemical libraries created by iterative synthesis around a few families of compounds are not always the best option when the objective is to identify non-previously exploited novel scaffolds. While several high throughput screens against P. falciparum of vast compound collections have been previously published [18,26], to date no high throughput studies have been performed in the field of microbial natural products. One of the potential challenges of using natural products collections is the expertise required to address the complexity of isolating and elucidating the pure compound responsible for the activity from positive extract identified in a primary screen. Unfortunately this process sometimes involves the identification of previously described compounds as is the case of petasol [33] obtained in the present study. The application of this sensitive technology [18] combined with the use of semi-automated fractionation and miniaturized LC-MS, LC-HRMS and NMR analyses, has allowed for a fast and efficient identification of minor extract components at an early stage of the discovery process thus optimizing the identification of novel actives. Likewise, high throughput technology has permitted the analysis of hundreds of extracts and multiple fractions in a highly cost-effective and time-efficient manner. Another limitation for natural product exploitation is in some cases the low yield of the active compounds that can be overcome applying different scale-up strategies as well as traditional strain optimization of the production of distinct metabolites or heterologous expression in suitable hosts. While the identification of novel compounds is of major interest, known compounds should not be disregarded since in many cases their antimalarial activity has not been previously described and they could thus provide a starting point for the development of new antiplasmodial agents. Drug repurposing constitutes a productive avenue for the identification of new therapies. Examples are the use of antifungal imidazoles recently tested in clinical trials against Chagas disease [34] or the recently approved combination therapy of nifurtimox (an antichagasic agent) with eflornithine for the treatment of sleeping sickness [35]. In summary, the compounds identified in the present natural products screening provide valuable information for antimalarial drug discovery from microbial sources. Although the most druggable compounds identified exhibit EC 50 values in the micromolar range, these hits could be considered as good starting points for a lead optimization process. While the present report describes the first data obtained in the screening of approximately one-seventh of the extracts available, efforts are currently underway to complete the analysis of the wealth of information contained within the whole MEDINA natural product extract collection. Supporting Information S1 Appendix. 1 H NMR and 13 C NMR spectra for pepstatin K.	v3-fos-license
2020-03-12T10:15:14.273Z	2020-03-10T00:00:00.000	215554405	{ "extfieldsofstudy": [ "Medicine", "Geography" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://onlinelibrary.wiley.com/doi/pdfdirect/10.1002/ece3.6114", "pdf_hash": "dc7eebe370e3858cde8d09a69c5e0a461c5eeffa", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45385", "s2fieldsofstudy": [ "Environmental Science", "Biology" ], "sha1": "2be7380a94f04d929d22a6da69e032b38dc8a11d", "year": 2020 }	pes2o/s2orc	Reductions in the dietary niche of southern sea otters (Enhydra lutris nereis) from the Holocene to the Anthropocene Abstract The sea otter (Enhydra lutris) is a marine mammal hunted to near extinction during the 1800s. Despite their well‐known modern importance as a keystone species, we know little about historical sea otter ecology. Here, we characterize the ecological niche of ancient southern sea otters (E. lutris nereis) using δ13C analysis and δ15N analysis of bones recovered from archaeological sites spanning ~7,000 to 350 years before present (N = 112 individuals) at five regions along the coast of California. These data are compared with previously published data on modern animals (N = 165) and potential modern prey items. In addition, we analyze the δ15N of individual amino acids for 23 individuals to test for differences in sea otter trophic ecology through time. After correcting for tissue‐specific and temporal isotopic effects, we employ nonparametric statistics and Bayesian niche models to quantify differences among ancient and modern animals. We find ancient otters occupied a larger isotopic niche than nearly all modern localities; likely reflecting broader habitat and prey use in prefur trade populations. In addition, ancient sea otters at the most southerly sites occupied an isotopic niche that was more than twice as large as ancient otters from northerly regions. This likely reflects greater invertebrate prey diversity in southern California relative to northern California. Thus, we suggest the potential dietary niche of sea otters in southern California could be larger than in central and northern California. At two sites, Año Nuevo and Monterey Bay, ancient otters had significantly higher δ15N values than modern populations. Amino acid δ15N data indicated this resulted from shifting baseline isotope values, rather than a change in sea otter trophic ecology. Our results help in better understanding the contemporary ecological role of sea otters and exemplify the strength of combing zooarchaeological and biological information to provide baseline data for conservation efforts. \| INTRODUC TI ON The sea otter, Enhydra lutris (Figure 1), is a marine mammal found in coastal nearshore ecosystems along the North Pacific. Across much of their range-Alaska, British Columbia, Washington, Oregon, and central California (Bodkin, 2015)-sea otters are considered a "keystone species" (Paine, 1969) disproportionately influencing ecosystem structure and function through indirect effects of predation (Estes & Palmisano, 1974;Hughes et al., 2016). The control by sea otters of key invertebrate populations prevents the overgrazing of kelp forests, and clears seagrass of harmful epiphytes, allowing these primary producers to flourish and provide habitat for a range of diverse taxa (Estes & Palmisano, 1974;Hughes et al., 2016;Steneck et al., 2002). Sea otters thus help maintain resilient coastal ecosystems, increase nearshore productivity, and provide valuable ecosystem services (Estes & Palmisano, 1974;Hughes et al., 2016;Wilmers, Estes, Edwards, Laidre, & Konar, 2012). The current global sea otter population and range are severely reduced from historical levels, due to commercial fur trade activity in the 18th and 19th centuries (Riedman & Estes, 1990). Prior to the initiation of the North Pacific fur trade in the mid-1700s, the global sea otter population may have been 150,000-300,000 individuals in a more or less contiguous distribution from Russia to Baja California (Riedman & Estes, 1990). Over the next ~150 years, sea otters were hunted nearly to extinction; by the early 20th century only 1,000-2,000 were left (Riedman & Estes, 1990). In 1911, when the International Fur Seal Treaty afforded sea otters some protection small remnant populations persisted in Russia, southwest Alaska, Haida Gwaii, Prince William Sound, and central California. From the mid-1960s onwards, otters were periodically translocated from these remnant populations to southeastern Alaska, British Columbia, Washington, Oregon, and other California localities (Bodkin, 2015). Today, sea otter populations from eastern Russia to southeastern Alaska and British Columbia have recovered much of their prefur trade distribution (Riedman & Estes, 1990). In contrast, the coastal areas from Washington to southern California support relatively small and spatially isolated populations; this fragmented distribution reflects the failure of the Oregon translocation (Bodkin, 2015) and slow rate of recovery and natural range spread of the California population. Despite the conservation success represented by postfur trade sea otter population recovery, it remains unclear whether sea otters have been fully restored to their historical ecological niches, which we here define as the combination of habitat (e.g., kelp forest versus soft sediment), and prey species utilized. We propose that industrial-scale exploitation of the marine environment by humans (McCauley et al. 2015) may have led to a constriction of the ecological niche of modern sea otters relative to the past. This should be particularly true for the southern subspecies (Enhydra lutris nereis, Figure 1) which is currently found only in central California, an area with high human coastal densities and large-scale fisheries. Sea otter recovery in this region has been sluggish, averaging only ~2% annual population growth (Tinker & Hatfield, 2018). Reasons suggested include the following: (a) the linear and narrow coastal shelf of California that limits access to unoccupied habitats (Lafferty & Tinker, 2014), (b) conflicts between otters and macroinvertebrate fisheries (Carswell, Speckman, & Gill, 2015), and (c) novel threats such as infectious disease and mortality caused by white sharks (Bodkin, 2015;Tinker & Hatfield, 2018;Tinker, Hatfield, Harris, & Ames, 2016). Defining a preindustrial ecological baseline for southern sea otters will thus benefit conservation efforts by identifying critical resources or habitats to protect, as well as potential functional roles and species interactions. In sea otters, the isotopic niche is an established proxy for dietary niche, as otters consume a wide variety of macroinvertebrate prey fueled by two isotopically distinct sources of primary production: phytoplankton and macroalgae (Newsome et al., 2009;Page, Reed, Brzezinski, Melack, & Dugan, 2008). Consequently, bulk analysis of vibrissae or other tissues that record dietary inputs over long time scales provides an accurate and high-resolution proxy for actual dietary niche breadth and variation, at both individual and population levels (Elliott Smith, Newsome, Estes, & Tinker, 2015;Newsome et al., 2009Newsome et al., , 2015. Further, cutting-edge isotopic techniques for analyzing individual amino acids within proteinaceous tissues can identify whether spatiotemporal shifts in bulk tissue isotope values are due to trophic level or baseline ecosystem changes (Chikaraishi et al., 2014;Whiteman, Elliott Smith, Besser, & Newsome, 2019). We can thus characterize the ecological niche of southern sea otter populations before and after their near extirpation by using bulk and amino acid isotope analysis of ancient and modern sea otter tissues. Here, we use zooarchaeological collections and modern tissue samples collected from southern sea otters to (a) establish an ecological baseline for the species in California and (b) evaluate changes F I G U R E 1 Southern sea otter (Enhydra lutris nereis) consuming a striped shore crab (Pachygrapsus crassipes). Photograph taken in Moss Landing, California, by Joe Tomoleoni. Used with permission in their dietary niche over the past 7,000 years. We test whether the isotopic niche, measured from bulk tissue δ 13 C and δ 15 N values, occupied by sea otters has remained constant over time at five regions along the central and southern California coastline. We evaluate the extent that modern sea otters occupy the ancient isotopic space as a proxy for the recovery of their historical ecological niche. We also compare these data to the modern isotopic prey space. Finally, we use individual amino acid δ 15 N analysis to examine whether prey choice, environmental conditions (or both) have changed. Our results provide a framework for interpreting the contemporary ecological role of sea otters and for identifying potential areas of their historical niche that are underutilized and could be promoted with conservation efforts. \| Modern samples To characterize the modern isotopic niche, we used previously pub- Our data encompass nearly the entire contemporary range of southern sea otters (Tinker & Hatfield, 2018) and represent a mix of males and females (Elliott Smith et al., 2015). All sampled individuals were independent foragers (weaned immature otters to aged adults), thus excluding dependent pups. To quantify the potential isotopic niche available for modern sea otters, we used published isotopic data from Newsome et al. (2015) on the prey types most commonly consumed in Monterey Bay and Big Sur (20 species) and San Nicolas Island (11 species) to define a possible isotopic prey space. \| Archaeological samples To characterize the historical isotopic niche, we sampled 107 bones from 10 California archaeological sites in close proximity to where modern sea otters were captured. These included both mainland and island archaeological sites and covered the entirety of the modern southern sea otter range-from Point Año Nuevo to San Nicolas Island (Table 1, Figure 2). We also include published isotope data from CA-SLO-2 near San Luis Obispo (Jones et al., 2011). Sample sizes and estimated site ages based on AMS radiocarbon dates are presented in Table 1; details on the excavation and identification of faunal remains can be found in the references therein. Care was taken not to resample individuals by considering specimen context (e.g., unit/ level) and element. Within units/levels or for single component sites (e.g., SMA-238), specimens were considered unique if they exhibited distinct (>1.0‰) δ 13 C or δ 15 N values (Clark, Horstmann, & Misarti, 2017). Where possible we sampled only adult or subadult individuals. From each specimen, we removed ~100-mg of bone for stable isotope analysis using a Dremel tool. \| Comparative tissue dataset Comparison of modern and ancient sea otter samples necessitates isotope data from two distinct proteinaceous tissues: bone collagen and vibrissae keratin. These proteins can exhibit systematic isotopic differences commonly referred to as tissue-specific isotope discrimination (Vanderklift & Ponsard, 2003). To quantify and correct for this, we sampled bone collagen, muscle, liver, and vibrissae from 29 sea \| Sample selection for amino acid δ 15 N (AA δ 15 N) The isotopic analysis of individual amino acids (AA) within proteinaceous tissues is a cutting-edge technique in ecological studies. By breaking down whole protein complexes and applying fundamental biochemical principles, it is possible to disentangle shifts in animals' trophic ecology from ecosystem-level changes associated with changes in the baseline isotopic composition of food webs (Chikaraishi et al., 2014). The δ 15 N of certain "source" amino acids (e.g., phenylalanine and lysine) are not heavily modified by animals during assimilation and tissue synthesis due to their lack of participation in metabolic processes such as deamination. Consequently, source amino acids experience little isotopic alteration as they move through food webs, providing an indicator of the baseline δ 15 N composition. Conversely, "trophic" amino acids, such as glutamic acid and proline, are heavily involved in metabolic processes and exhibit consistent 15 N enrichment with each trophic step. Thus, the magnitude of the nitrogen isotope difference between source and trophic amino acids can be used as a proxy for the trophic level of an individual, whereas source amino acids can be used to infer baseline ecosystem δ 15 N composition (Chikaraishi et al., 2014;Whiteman et al., 2019). To examine whether spatiotemporal changes in bulk tissue isotope values of sea otters were due to baseline isotopic or dietary shifts, we selected a subsample of 4-5 individuals from each archaeological region (excluding CA-SLO-2) for AA δ 15 N analysis (Table 3, Appendix S5). In addition, we selected five modern Monterey Bay individuals from stranded otter bone collagen samples (Table 3, Appendix S5). We did not analyze modern sea otter vibrissae for AA δ 15 N analysis because to our knowledge no study has addressed AAspecific tissue discrimination. \| Isotopic analysis For both bulk analysis and amino acid analysis, we report all isotopic results as δ values: δ 13 C or δ 15 N = 1,000*[(R samp /R std ) − 1], where R samp and R std are the 13 C: 12 C or 15 N: 14 N ratios of the sample and standard, respectively. Prior to analysis, bone collagen subsamples were cleaned of sediment and then demineralized with 0.25 N hydrochloric acid for 15-72 hr at 5°C. Each sample was then lipid extracted with three sequential 24 hr soaks in 2:1 chloroform:methanol and lyophilized after a deionized water rinse. Muscle and liver samples were also lipid extracted, rinsed, and lyophilized. Sea otter vibrissae were cleaned with 2:1 chloroform:methanol to remove surface contaminants. For bone, muscle, and liver, 0.5-0.6 mg of each subsample was weighed into 3 × 5 mm tin capsules. Vibrissae were subsampled following Newsome et al. (2009). For AA δ 15 N analysis, ~5-10 mg of extracted collagen was chemically processed following established protocols (Whiteman et al., 2019). See Appendix S1 for details on all isotopic measurements and quality control. \| Data corrections and statistical analysis We corrected for both tissue-specific discrimination and temporal (Suess Effect; Cullen, Rosenthal, & Falkowski, 2001) isotopic shifts prior to analysis (Appendix S2). The resulting dataset (Dryad accession https://doi.org/10.5061/dryad.ttdz0 8ktj) violated a number of important assumptions of ANOVA. We thus tested for differences among modern and ancient sea otter δ 13 C and δ 15 N isotope values at each site using the nonparametric Cramér test (Baringhaus & Franz, 2004). We also employed Kruskal-Wallis, and pairwise Wilcoxon signed-rank comparisons with Bonferroni adjusted p-values to examine differences for each isotope system between ancient and modern otters; we report pairwise comparisons in Dryad accession https://doi.org/10.5061/ dryad.ttdz0 8ktj. We likewise compared δ 13 C and δ 15 N isotopic values of ancient otters across different archaeological sites and regions. F I G U R E 2 Locations of ancient and modern sea otter populations. Regions with archaeological southern sea otter remains represented by gray circles. Modern sea otter populations sampled for this study represented by black triangles (Elliott Smith et al., 2015). Modern range redrawn from Tinker and Hatfield (2018). Acronyms for regions are as follows: ANO, Año Nuevo; MBY, Monterey Bay; BSR, Big Sur Reserve; SLO, San Louis Obispo; SBC, Santa Barbara Channel; SMI, San Miguel Island; and SNI, San Nicolas Island. For a full list of sites and sample sizes, see Table 1 Our amino acid δ 15 N dataset did not exhibit deviations from normality, thus we used one-way ANOVA to evaluate for differences in source and trophic AA δ 15 N, as well as trophic-source offset among sites. We calculated pairwise trophic-source offsets, as well as average offsets following methods from Bradley et al. (2015), using glutamic acid (Glu) and hydroxyproline/proline (Hyp-Pro) as trophic AAs and phenylalanine (Phe) and lysine (Lys) as source AAs. To characterize isotopic/dietary niche space and variability of ancient and modern otter populations, we used Bayesian standard ellipse areas (SEA B ) (Stable Isotope Bayesian Ellipses in R; SIBER-Jackson, Inger, Parnell, & Bearhop, 2011). We ran the model with archaeological sites lumped within regions, and then with each site considered separately. In the latter case, we excluded the three samples from SMI-602, and all samples from MNT-831 which had poorly constrained age ranges. We also calculated SEA B for modern California prey items from Monterey Bay/Big Sur and San Nicolas Island. We calculated median SEA B and associated credibility intervals (SD) from 10,000 iterations of the model, as well as the proportion of SEA B iterations from one group larger than the SEA B of another group. Finally, we tested for the influence of time averaging on isotopic space (SEA B ) using a linear model of median SEA B versus time span occupied by each site (see Table 1 and Figure 5); modern TA B L E 1 Ancient and modern sea otter samples. For archaeological data, the age column represents years before present (yBP), based on calibrated AMS radiocarbon dates from the listed references. For modern samples, this represents the years of sea otter captures/ collections (AD). With the exception of CA-SLO-2, all archaeological samples presented were analyzed in the current study. Modern sea otter data come from Elliott Smith et al., 2015; we analyzed an additional seven modern individuals from the San Louis Obispo region samples were given a time span of 10 years. We performed all data corrections and statistical analyses using Program R (v.3.2.0). \| Bulk δ 13 C and δ 15 N values We \| Amino acid δ 15 N values We found regional and temporal differences among otter populations in both source and trophic amino acids (Table 3 Table 3 and Appendix S5. However, we found no statistical difference in the average offset between trophic and source amino acids between ancient and modern otters [F(1,20) = 0.29, p = .60]. very similar (Appendix S4). We also found differences in SEA B of ancient and modern sea otter populations relative to the potential prey space produced by analysis of modern prey items. All modern sea otter populations occupied <35% of the median potential prey space, in stark contrast to archaeological sites that ranged from 31% To evaluate the effect of time averaging, we compared the median SEA B of all ancient and modern sites and modern prey (excluding SMI-602 and MNT-831) to the estimated time span represented by each site (Table 1). A resulting linear model found no effect of time span on ellipse area size (R 2 = 0.06, p = .19; Figure 5), indicating differences in ellipse area between ancient and modern populations are driven by biological patterns and not statistical artifacts. \| Isotopic niche of ancient and modern sea otters We found striking differences in the isotopic niche space of ancient and modern sea otters in California, particularly in the southern portion of the range. Ancient otters from the five regions occupied a larger isotopic niche than nearly all modern localities ( Figure 4, Table 2, Appendix S3). However, at the three most southerly sites, ancient otters exhibited SEA B more than twice as large as modern counterparts ( Figure 4, Table 2, Appendix S3). We suspect this is due to the diversity of nearshore marine communities in southern California. The area near the Santa Barbara Channel is a major zoogeographic transition zone (Figure 2), with nearshore and island ecosystems south of Point Conception having a greater complexity of substrate, milder wave environments, and more variable upwelling regimes in comparison with the rest of California (Graham et al., 2008). As a result, these areas support a greater diversity of invertebrate and vertebrate taxa, and higher rates of endemism than sites in central and northern California (Graham et al., 2008;Seapy & Littler, 1980). This plethora of secondary prey species likely contributed to the greater dietary diversity of ancient sea otter populations in this region. In particular, we note the striking variability in δ 13 C values among ancient otters from San Miguel and San Nicolas Islands recently established sea otter populations at low density target the largest, most energy-rich prey, leading to low levels of dietary diversity and a small population-level isotopic niche space (Estes, Riedman, Staedler, Tinker, & Lyon, 2003;Tinker et al., 2008Tinker et al., , 2012. As densities increase, high-ranked resources become depleted, and individuals broaden their diet resulting in an increase in population-level dietary/isotopic niche breadth (Newsome et al., 2009;Tinker et al., 2008). This latter pattern can be seen in modern sea otters in the northerly sites of Monterey Bay and Big Sur, where otters are likely near carrying capacity (Tinker et al., 2019). Importantly, in these areas, the size of the ancient sea otter isotopic niche is similar to their modern counterparts, indicating populations in central California have largely recovered their historical dietary breadth ( Figure 4, Table 2, Appendix S3). In contrast, the isotopic data from archaeological remains of sea otters in southern California suggest that populations in this region could occupy a much larger dietary niche than currently observed. \| Latitudinal trends in modern and ancient sea otter isotopic values Among modern otters, there was a ~4‰ increase in mean δ 15 N with decreasing latitude (Figure 3). Some of this can be explained In contrast, high densities of otters in Monterey Bay and Big Sur F I G U R E 4 Isotopic niche space of ancient and modern sea otter populations. Left-side panels show δ 13 C and δ 15 N values of ancient (gray circles) and modern (black diamonds) sea otters. Ellipses represent 50% of the total amount of isotopic space occupied by each population. Right-side shows median Bayesian ellipse size ± credibility intervals for each population. All sea otter data were also compared to modern prey data from Newsome et al. (2015). From top to bottom, the ancient/modern comparisons are as follows: ancient Año Nuevo otters (ANO) with modern Monterey Bay otter and prey (MBY), ancient MBY with modern MBY and prey, ancient San Louis Obispo (SLO) with modern SLO and MBY prey, ancient San Miguel Island (SMI) with modern Santa Barbara Channel (SBC) and San Nicolas Island prey, and ancient San Nicolas Island (SNI) with modern SNI and prey. For a full list of sites and sample sizes, see Table 1 mean they include a greater amount of smaller, lower trophic level invertebrates in the diet (Tinker et al., 2008) and thus have lower δ 15 N values (Newsome et al., 2009). Among ancient otter populations, we found no overall trend with decreasing latitude; however, we did find differences in mean δ 15 N between ancient and modern otters at some sites. Ancient otters from Año Nuevo and Monterey Bay had significantly higher δ 15 N values than modern counterparts (Figure 3, Dryad acces- Holocene. Notably, we did not find a significant difference in the average offset between our trophic and source amino acid δ 15 N values between locales or time periods (Table 3), which indicates that the trophic level of sea otters has not significantly changed over the past 7,000 years (Chikaraishi et al., 2014;Whiteman et al., 2019). Instead, we found differences in the baseline δ 15 N values among regions with northerly ancient Año Nuevo (12.3 ± 1.1‰) otters exhibiting the highest mean (±SD) Phe (source) δ 15 N values, and southerly ancient San Miguel Island otters the lowest (9.8 ± 0.9‰; Table 3), a pattern opposite to that observed today (Vokhshoori & McCarthy, 2014). In addition, we noted a wide range in the mean (±SD) differences between otter trophic and source δ 15 N values between sites, with ancient San Nicolas Island otters having the lowest (8.3 ± 1.8‰), and modern Monterey Bay (11.7 ± 1.6‰) and ancient Año Nuevo (11.7 ± 1.8‰) the highest, offsets (Table 3). This variation likely reflects the trophic flexibility of otters as documented in the modern through high-resolution observational datasets (Estes et al., 2003;Tinker et al., 2008). \| Zooarchaeological data and modern southern sea otter ecology The analysis of ancient faunal remains from archaeological sites poses a number of quantitative and theoretical issues when comparing them to modern ecological datasets. Most notably, archaeological sites in even the best scenarios represent time-averaged assemblages on the order of hundreds to thousands of years, and so can never realize the high resolution of modern ecological sampling conducted over seasonal to decadal timescales (Rick & Lockwood, 2013). Here, our ancient sea otter samples are likely an aggregation of multiple generations. In addition, we cannot rule out the modifi- Second, when characterizing the historical ecological niche, incorporation of data across hundreds or even thousands of years provides a more representative view of a species ecology than a seasonal or multi-annual snapshot. Such an approach, which examines ecological patterns over evolutionarily relevant timescales, is a proven way of establishing conservation baselines and characterizing the plasticity of animals to long-term natural or anthropogenic environmental change (e.g., Jackson et al., 2001;Rick & Lockwood, 2013). Our results have implications for the conservation of southern sea otters, and for efforts to minimize negative interactions between otters and human populations. Our findings suggest sea otters have a much greater potential niche space in southern California than currently occupied and further, that this niche may be larger than that utilized by modern, high density, sea otter populations in central California (Tinker et al., 2008(Tinker et al., , 2019. Southern sea otters are currently expanding their range into southern California where they TA B L E 3 δ 15 N values of individual amino acids from a subset of ancient and modern sea otters. Presented here are two commonly reported "trophic" and "source" amino acids, respectively, glutamic acid (Glu), hydroxyproline-proline (Hyp-Pro), and phenylalanine (Phe), and lysine (Lys). Also reported is the mean (±SD) Glu-Phe offset for each locale and the average offset between these trophic and source amino acids as calculated following methods by Bradley et al. (2015). have not occurred for centuries; our data provide clues as to how their diets will broaden as they continue to establish south of Point Conception. However, the ancient dietary niche space of southern sea otters that we have characterized here was prior to the development of commercial fisheries. Over the past two centuries, intensive fishing has exploited several of the main macroinvertebrate species consumed by sea otters, including abalone, and red urchins (Dayton, Tegner, Edwards, & Riser, 1998;Leet, 2001). Consequently, the historical dietary niche occupied by sea otters may be unobtainable until conservation and management efforts restore higher densities of important invertebrate prey. Despite this, our dataset speaks to a long-term history of interactions between humans and sea otters. The coastal archaeological record (e.g., Jones et al., 2011;Misarti et al., 2009;Szpak et al., 2012) demonstrates that humans have been living with, and harvesting, sea otters and their prey items for at least 10,000 years. The isotopic data we present here show that despite this, ancient sea otters had an ecological niche equivalent to, or greater than modern populations, suggesting that they occurred at high density in the past despite being subjected to harvest pressure. Such insights may aid in developing species-or ecosystem-based management plans that promote long-term sustainable relationships between the competing needs of humans and top predators. CO N FLI C T O F I NTE R E S T None declared. F I G U R E 5 Relationship between median SEA B and time span. Time span for each archaeological site presented in Table 1; modern prey and otters were given a time span of 10 years	v3-fos-license
2021-09-27T18:06:30.415Z	2021-08-20T00:00:00.000	238734382	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://www.mdpi.com/2673-947X/1/2/7/pdf", "pdf_hash": "78770aff7dc8ec201d5036ca1a827539b6442171", "pdf_src": "ScienceParsePlus", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45386", "s2fieldsofstudy": [ "Medicine", "Biology" ], "sha1": "8b5a33d8460815a94b2af9812036097cacd009bc", "year": 2021 }	pes2o/s2orc	Clinical Failure and Emergence of Resistance during Third Generation Cephalosporin Therapy for Enterobacter spp. Infection: Is the Risk Overestimated? A Prospective Multicentric Study : Background: Clinical and microbiological guidelines recommend treating infections caused by Enterobacter spp. with cefepime or carbapenems. The main objective of this study was to assess the risk of clinical failure with third generation cephalosporin (3GC) therapy compared to other β -lactams for infections caused by Enterobacter spp. Our secondary objective was to evaluate the risk of emergence of resistance during therapy. Methods: We conducted a prospective observational study in seven French hospitals over an 18-month period including all patients with a pulmonary and/or bloodstream infection due to Enterobacter spp. susceptible to 3GC. Results: Seventy-four patients were included in our study. Among them, 26 (35%) received a 3GC as a ﬁrst-line treatment, and clinical improvements were observed for 13/21 (62%) of them. Four (5%) cases of emergence of 3GC resistance were observed during therapy including one in the 3GC group. 3GC therapy can be safely used as ﬁrst-line therapy especially for non-severe patients suffering from pulmonary or bloodstream infections due to Enterobacter spp. Conclusions: Emergence of 3GC resistance remains a rare event, and there is a lack of evidence of the beneﬁt of last-line antibiotics therapies. Introduction The Enterobacteriaceae family is a major cause of community and healthcare related infections. Several species as Enterobacter spp., Serratia marcescens, Citrobacter freundii, Providencia spp. and Morganella morganii possess natural chromosomic AmpC β-lactamase that hydrolyze benzylpenicillin, amoxicillin, ampicillin and cefazolin [1]. Other β-lactamins such as third generation cephalosporins (3GCs) may also be affected in the case of an overproduction of these enzymes. Due to a high risk of mutant selection, several recommendations [1,2] suggest avoiding use of 3GC to treat AmpC producing Enterobacteriaceae (AE). Previous studies reported a high risk of resistance reaching 19% in patients with infections related to Enterobacter spp. during 3GC antibiotic therapy [3,4]. Nevertheless, these studies suffer from methodological bias, as authors did not take into account several confounding factors involved in the selection of resistant strains such as the PK/PD parameters (minimal inhibitory concentration, MIC; source of infection; initial dosage and modalities of administration) and did not discriminate the risk of emergence of resistance according to species. Hence, a recent prospective study underlined a lower risk of resistance during 3GC therapy and observed different risks according to the involved species and site of infection. Indeed, among the different species possessing AmpC β-lactamase, Enterobacter spp. seems to have the highest risk of emergence of resistance under treatment compared to Morganella morganii and Serratia marcescens. Furthermore, source of infection seems to play a role in undrained biliary tract infection as a factor associated with the emergence of resistance [5]. The 3GCs remain the first-line recommended antibiotics as probabilistic treatment for infections related to Enterobacteriaceae, despite the fact that Enterobacter spp. are more and more frequently isolated, being ranked fourth and sixth among the bacteria isolated from patients hospitalized with pneumonia in the USA and Europe, respectively [6]. Despite antimicrobial guidelines discouraging the use of 3GC (EUCAST) when Enterobacter spp. is secondarily identified, the decision to switch the 3GC initially used to another β-lactamin such as cefepime, piperacillin, piperacillin-tazobactam or carbapenem varies according to infectious disease practitioners. Indeed, the link between therapeutic failure and the use of 3GC remains uncertain except for urinary tract infection, for which 3GC can be used safely because adequate antibiotic penetration can be achieved [5,7,8]. For all other sites of infection, most studies were retrospective and only partially investigated the outcome of patients treated with 3GC for AE related infection. We therefore prospectively assessed the risk of clinical failure with 3GC therapy compared to other β-lactams in two deep infections: pulmonary and bloodstream infections due to Enterobacter spp. Our secondary objective was to evaluate the risk of emergence of resistance during therapy. Design and Study Population We conducted a prospective observational study in 7 French hospitals over an 18-month period (from 1 January 2017 to 1 June 2018). All patients with a pulmonary or bloodstream infection, as defined below, due to Enterobacter spp. susceptible to 3GC were included and monitored until the time of discharge. All data were prospectively and anonymously collected. Characteristics of Patients At inclusion the following data were recorded: demographics characteristics, underlying disease (chronic cardiac failure, chronic pulmonary disease, chronic renal failure with renal clearance, diabetes mellitus, hematologic malignancies, solid cancer, inflammatory disease), immunosuppressive treatment (immunosuppressive drugs including tumor necrosis factor (TNF) antagonists, corticosteroid and chemotherapy drugs), Charlson Comorbidity Index [9] and prior use of antibiotics (during the last 3 months). Clinical and Physiological Parameters At time of infection, the following data were recorded: initial severity of infection measured with Pitt score (non-ICU patients) and SOFA score (ICU patients), source of infection, presence of central venous catheter (CVC), inflammatory parameters such as procalcitonin (PCT), C-reactive protein (CRP), total leukocyte count, uremia and creatininemia when available. At 72 h after the first dose of antibiotic treatment, clinical data and inflammatory parameters were recorded as: body temperature, heart and respiratory rate, PCT, CRP and persistence of the organism at the site of infection determined as positivity of microbiological samples. Emergence of resistance was defined as a positive blood culture or a positive sample at the same site of infection to Enterobacter spp. overproducing AmpC during treatment. All antibiotics prescription data, dosages and modalities of administration were collected from the beginning of the infection to the discharge. Definition Pulmonary infection was considered when: -Impairment of lung function and infiltration in the chest X-ray film was observed; -Fever ≥ 38.3 • C or leukocytosis ≥ 10,000/µL was observed. Pulmonary infection was related to Enterobacter spp. if isolated at a concentration of: -≥10 4 CFU/mL from bronchoalveolar lavage (BAL) sample; -≥10 3 CFU/mL from protected specimen brushing; -≥10 5 CFU/mL from tracheal aspiration. Bloodstream infection related to Enterobacter spp. was defined as a positive blood culture with sepsis signs according to Bone's criteria [10]. The primary source of bacteremia was determined according to CDC criteria [11]. Otherwise, it was defined as primary bacteremia with undetermined source of infection. Initial time of infection was defined as the day of the first sample collection date positive to Enterobacter spp., which corresponds to the day of first clinical symptoms for hospitalized patients. Clinical failure was defined as the persistence of clinical symptoms 72 h after the antibiotic treatment had been started and persistence of abnormal heart and respiratory rate. Biological failure was defined as PCT or CRP level > half of the initial dosage 72 h after antibiotic treatment. Renal clearance was calculated using the Cockcroft and Gault formula. Emergence of 3GC resistance was defined as a positive culture to Enterobacter spp. overproducing AmpC β-lactamase at the site of infection. Mortality was defined as the 30-day mortality after the infection. First-line treatment was defined as first antibiotic therapy received to treat the infection. Adequate therapy was defined as a standard parenteral dose of an antimicrobial to which the Enterobacter strain was susceptible in vitro or a standard oral dose of an antimicrobial with good bioavailability to which the Enterobacter strain was susceptible in vitro such as fluoroquinolones and sulfamethoxazole-trimethoprim. Definitive treatment was defined as the last antibiotic treatment received. Microbiological Methods Enterobacter spp. were identified on bacterial culture by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS, Bruker Daltonics, Billerica, MA, USA). Susceptibility testing was performed according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines [12]. When the isolate was resistant to any of the 3GCs tested (cefotaxime, ceftriaxone and ceftazidime), it was regarded as resistant to all broad spectrum cephalosporins. ESBL producing Enterobacter sp. were excluded. 3GC resistant Enterobacter spp. by AmpC β-lactamase overproduction was confirmed following EUCAST guidelines [12]. Statistics The descriptive statistics are based on means (±standard deviation) or medians (minimum-maximum) depending on the distribution of quantitative variables. The qualitative variables are described in terms of number (%). Univariate comparisons used standard statistical tests after verifying the distribution of variables with a degree of significance of 5%; 95% confidence intervals are provided for each estimate. Calculations were performed using the R and SAS software. Study Population During the study period, 79 cases were reported; among them, five (6%) were excluded because of lack of sufficient data, resulting in a final total of 74 infections related to Enterobacter spp. susceptible to 3GC (Figure 1). Antibiotic Therapy All patients (n = 74, 100%) patients received a β-lactam as first-line therapy and were considered as adequate in all cases according to antibiotic susceptibility tests. Descriptions of antibiotics therapies are listed Table 2. Among the 74 patients, 13 (17.6%) were treated with a combination therapy, including an aminoglycoside or a fluoroquinolone in 10 (13.5%) and three (4.1%) cases (Table 3), respectively. In 32 (43.2%) cases, first-line antibiotic therapy was switched to another βlactam with an average time of 2.2 days. Among them, 19 (59.4%) were switched before day 3 after initiating antibiotic therapy, and 13 (40.6%) were switched after day 3. Third Generation Cephalosporin Therapy 3GCs were initiated as first-line therapy for 26 (35.1%) patients with a mean duration of treatment of 5 (± 4.5) days. Characteristics of patients in 3GC therapy and characteristics of infections are described Table 3. First Antibiotic Therapy, Clinical Failure and Emergence of Resistance Patient outcomes are described in Table 4. Biological and clinical improvements were observed in 70.4% and 88.6% of cases, respectively. For all patients we observed a microbiological eradication defined by the reduction of the initially positive microbiological sample (majority of blood cultures). Four (5%) cases of emergence of 3GC resistance were observed during therapy. Characteristics of infections with emergence of 3GC resistance are described in Table 5. All the four patients with 3GC-resistant isolates were hospitalized in the ICU, and three out of four (75%) had a pulmonary infection. None of these patients received combination therapy. According to univariate analysis, piperacillin ± tazobactam as first-line therapy was the only variable associated with a lower clinical improvement and a higher risk of emergence of resistance, but the difference was not statistically significant (77% vs. 94% p = 0.08 and 13% vs. 2% p = 0.08, respectively). As shown by the univariate analysis, no factor was associated with better biological improvement. Among our population, overall 30-day mortality was 14%, and treatment with 3GC as first-line therapy was the only variable associated with a lower mortality (not statistically significant; 4% vs. 19%; p = 0.08). Discussion In the present prospective observational study, we found that 3GC therapy used as first-line therapy to treat pulmonary and/or BSI related to Enterobacter spp. was not associated with clinical failure and emergence of resistance. Moreover, we found a low rate of emergence of resistance during therapy with only four cases (5%) in our cohort, and this emergence of resistance was not associated with clinical failure. Several studies reported the risk of emergence of resistance during 3GC therapy, but none of them examined the clinical impact of this phenomenon as primary objective. Nevertheless, overall mortality was examined, but 3GC therapy or emergence of resistance during therapy was not associated with a higher mortality. A landmark prospective observational study underlined a high risk of emergence of resistance for AE infection treated with 3GC, but this was not associated with 14 days mortality (4/31 13% vs. 15/87 17%; p > 0.2) [3]. Similarly, in a prospective study including 732 AE infections, 218 patients received 3GC therapy, and Choï et al. found that emergence of resistance during 3GC therapy was not associated with mortality (1/11 9.1% vs. 2/207 1.0%; p = 0.144) [5]. In our study, 3GC was the most common antibiotic prescribed as initial therapy and, as previously described, was not associated with a higher mortality, but several confounding factors were not included in the analysis such as initial gravity, co-morbidities and site of infection. Interestingly, 3GC therapy was associated with a lower Pitt score and non-ICU patients (Table 3) suggesting that we can safely use 3GC therapy for non-severe patients. Half 3GC initial therapies were switched for another β-lactamin such as cefepime, while clinical improvement was observed in 92% of cases. Furthermore, in this study, none of the empirical treatments active against Enterobacter spp. were associated with better outcome as previously described [13]. Therefore, we can question the necessity of switching treatment for patients with a clinical improvement 72 h after 3GC therapy. Previous old studies [3,4] reported an emergence of resistance rate of 19% with several unreported limitations. In a more recent study, Choï et al. found an overall incidence for 3GC resistance during therapy of 5.0%, and among AE, Enterobacter spp. have the highest risk of emergence of 3GC resistance during therapy with 8.3%. In our study, we also found a low rate of emergence of resistance during therapy with only four cases (5%). Surprisingly, emergence of resistance occurred more frequently for patients treated with piperacillin ± tazobactam, but the difference, due to a low number of events, was not statistically significant. A previous exposure to piperacillin ± tazobactam has been reported to be a risk factor for isolating broad spectrum cephalosporin resistant Enterobacter spp. [14] but not emergence of 3GC resistance during therapy. We cannot consider the impact of tazobactam as an inducer in emergence of the 3GC resistance phenomenon. Firstly, tazobactam is only a weak inducer of AmpC beta-lactamase, in comparison to clavulanate [15]. Secondly, emergence of 3GC resistance is due to the selection of a constitutive 3GC resistant mutant among a susceptible population and not to antibiotic induction, a temporary in vitro phenomenon [16]. This emergence seems theoretically more frequent when piperacillin is used compared to 3GC in our study. This could be explained by the high basal MIC of Enterobacter spp. for piperacillin (1 µg/mL compared to 0.03 µg/mL cefotaxime and cefepime) [17]. Epidemiological cut-off values (ECOFFs) for Enterobacter cloacae reported by EUCAST are 8.0 µg/mL for piperacillin ± tazobactam, 0.5 µg/mL for cefotaxime and 0.125 µg/mL for cefepime. Therefore, our results suggest that piperacillin ± tazobactam should be prescribed cautiously for Enterobacter spp. infec-tions especially as we found the lowest clinical improvement with this treatment. Several studies have already warned of the necessity to optimize the administration and use a high dosage of piperacillin-tazobactam in order to reach pharmacological targets for ICU or severe-sepsis patients [18,19]. Emergence of 3GC resistance during therapy for AE infections has always been associated with the use of 3GC [3][4][5]20,21]. To our knowledge, this is the first prospective multi-centric study including infections due to Enterobacter spp. only susceptible to 3GC. Emergence of 3GC resistance is due to the selection of a constitutive 3GC resistant mutant among a susceptible population. We found that this is a rare event, and the choice of the molecule is not the key issue; other factors such as the initial severity, MIC, dosage, the inoculum size and the source of infection should be taken into consideration as Choi et al. suggested. One of the main limitations of our study is the small sample size of our cohort, as in others [3,20,21]. This is because we included only deep infections such as pulmonary and bloodstream infections and only strains susceptible to 3GC. We could not generalize our results for different infection sites because of the sample size; only univariate analysis could be performed. Moreover, our study was observational but not randomized. Conclusions Clinical failure related to emergence of 3GC resistant AE by AmpC overproduction is still a debated issue. In this prospective multicentric observational study, we found that 3GC therapy can be safely used as first-line therapy especially for non-severe patients suffering from pulmonary or bloodstream infections due to Enterobacter spp. Emergence of 3GC resistance remains a rare event, and there is a lack of evidence of the benefit of last-line antibiotics therapies.	v3-fos-license
2023-01-17T14:27:33.380Z	2022-01-25T00:00:00.000	255868490	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1186/s12935-022-02468-3", "pdf_hash": "8940bae41fabd4d48cabc695a499c85ce0dfa438", "pdf_src": "SpringerNature", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45387", "s2fieldsofstudy": [ "Medicine" ], "sha1": "8940bae41fabd4d48cabc695a499c85ce0dfa438", "year": 2022 }	pes2o/s2orc	Green tea-derived theabrownin induces cellular senescence and apoptosis of hepatocellular carcinoma through p53 signaling activation and bypassed JNK signaling suppression Theabrownin (TB) is a bioactive component of tea and has been reported to exert effects against many human cancers, but its efficacy and mechanism on hepatocellular carcinoma (HCC) with different p53 genotypes remains unclarified. MTT assay, DAPI staining, flow cytometry and SA-β-gal staining were applied to evaluate the effects of TB on HCC cells. Quantitative real time PCR (qPCR) and Western blot (WB) were conducted to explore the molecular mechanism of TB. A xenograft model of zebrafish was established to evaluate the anti-tumor effect of TB. MTT assays showed that TB significantly inhibited the proliferation of SK-Hep-1, HepG2, and Huh7 cells in a dose-dependent manner, of which SK-Hep-1 was the most sensitive one with the lowest IC50 values. The animal data showed that TB remarkably suppressed SK-Hep-1 tumor growth in xenograft model of zebrafish. The cellular data showed TB's pro-apoptotic and pro-senescent effect on SK-Hep-1 cells. The molecular results revealed the mechanism of TB that p53 signaling pathway (p-ATM, p-ATR, γ-H2AX, p-Chk2, and p-p53) was activated with up-regulation of downstream senescent genes (P16, P21, IL-6 and IL-8) as well as apoptotic genes (Bim, Bax and PUMA) and proteins (Bax, c-Casp9 and c-PARP). The p53-mediated mechanism was verified by using p53-siRNA. Moreover, by using JNK-siRNA, we found JNK as a bypass regulator in TB's mechanism. To sum up, TB exerted tumor-inhibitory, pro-senescent and pro-apoptotic effects on SK-Hep-1 cells through ATM-Chk2-p53 signaling axis in accompany with JNK bypass regulation. This is the first report on the pro-senescent effect and multi-target (p53 and JNK) mechanism of TB on HCC cells, providing new insights into the underlying mechanisms of TB's anti-HCC efficacy. Introduction Liver cancer is a major contributor to the world's cancer burden, with more than 800,000 new cases and 700,000 deaths each year [1]. Hepatocellular carcinoma (HCC), a principal histologic type of liver cancer, represents more than 75% of primary liver cancers [2]. Due to the aging and population growth, the global incidence of HCC increased by 75% between 1990 and 2015, with the highest incident, mortality and years of life lost in east Asia [3]. The worldwide risk factor of HCC is heterogeneous. Hepatitis B virus (HBV) is the leading cause of incident cases of HCC in Africa and East Asia, while alcoholic liver disease (ALD) and hepatitis C (HCV) are the most common risk factor for HCC in the USA [4]. Prognosis of HCC is poor all around the world, resulting in a rough equivalent of incidence and mortality rates [5]. For early-stage HCC, radiofrequency ablation or surgical resection remains the main treatment. However, up to 75% of patients undergoing surgery experience recurrence within 5 years [6]. Over the last decade, targeted therapy (sorafenib) has become the major systemic strategy which can significantly improve the overall survival for patients with unresectable HCC [7]. However, the adverse effects of sorafenib, such as diarrhea, hypertension, and hand-foot skin reaction (HFSR), as well as its low bioavailability limit its clinical application [8,9]. Therefore, there is an urgent need to explore potential drug candidates for HCC. Alike many carcinomas, HCC has multiple genomic mutations. The prevalent mutations locate at TERT promoters, such as TP53, CTNNB1, AXIN1 and CDKN2A [10,11]. In most cases of HCC (> 90%), telomerase activation, relating to TERT promoter mutations, is necessary for malignant transformation and tumor progression [12,13]. Of these, CTNNB1 mutations frequently activate the Wnt/β-catenin pathway, particularly in patients with HBV uninfection and well-differentiated tumors (11-37% of HCC cases) [14,15]. By contrast, inactivation of p53 caused by TP53 mutations particularly appears in cases related to HBV infection and aflatoxin B1 exposure [16][17][18]. As a tumor suppressor, TP53 encodes p53 transcriptional factor to prevent tumor development through permanently suppression of cell proliferation by cell cycle arrest and facilitation of cell death by apoptosis [19]. However, TP53 mutation or deletion occurs in nearly a half of human cancers, while tumors carrying wild-type TP53 usually get rid of the p53 defense mechanism via interaction with negative regulators, such as MDM2 and MDM4 [20,21]. Thus, reactivation of p53 becomes a potential strategy for cancer treatment [22][23][24][25] For instance, a p53-MDM2 inhibitor, RG7388, activates p53 signaling pathway by selectively blocking p53-MDM2 binding, exhibiting encouraging anti-cancer efficacy in several different clinical trials [26,27]. Green tea, derived from leaves of Camellia sinensis, was originally used as medicine in ancient China. Meanwhile, it was one of the most prevalent beverages worldwide for centuries. In recent decades, green tea's health benefits have been extensively studied, including antiinflammation, cardiovascular-protection, anti-obesity, anti-cancer, and hepato-protection [28][29][30][31][32]. A large prospective cohort study on 164,681 adult Chinese men has concluded that 10 g or more green tea consumption per day decreased the mortality from cancers, indicating an anti-cancer efficacy of green tea [33]. As the main pigment of tea, theabrownin (TB) is a heterogeneous polymer (3-50 kDa) with high water solubility and possesses regulatory effects in improving metabolism of glucose and serum lipids [34][35][36]. Our previous studies have discovered TB's pro-apoptotic effects on human carcinoma cells and sarcoma cells both through a p53-mediated mechanism [37,38]. To date, the efficacy of TB on human HCC cells with different p53 genotypes remains unclear. To fill this gap, this study adopted HCC cell lines, including SK-Hep-1 (p53-WT), HepG2 (p53-WT) and Huh7 (p53-mut), to evaluate TBʹs effects and the p53-related mechanism. Cell culture The human HCC SK-Hep-1, HepG2, and Huh7 cell lines were purchased from Cell Bank of Chinese Academy of Sciences, China. SK-Hep-1 cells were cultured in MEM α containing 10% FBS at 37 °C and 5% CO 2 . Cells were passaged two or three times per week. HepG2 and Huh7 cells were cultured in DMEM containing 10% FBS at 37 °C and 5% CO 2 . Cells were passaged two or three times per week. Zebrafish Wild-type AB strain of zebrafish was purchased from the China Zebrafish Resource Center, Institute of Hydrobiology, China Academy of Science (Wuhan, China) and accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (SYXK 2012-0171). Larval zebrafish (2 dpf, days post fertilization) were obtained by natural pair-mating and raised at 28 °C under a 10/14 h dark/light cycle. Cell viability assay and morphological observations Cells viability was determined by MTT assays. Briefly, cells (8 × 10 3 cells/well) were seeded in 96-well plates overnight, followed by exposure to various concentrations of TB for 24 or 48 h. All wells were added ten microliters of MTT reagent (5.0 mg/ml) and inoculated for 4 h. Afterwards, DMSO was added to dissolve the formazan crystals. Optical density (OD) values at 490 nm were detected by a microplate reader (Bio-Rad, USA). Inhibitory rate (%) = [1 − (OD TB /OD control )] × 100. The 50% inhibitory concentrations (IC 50 ) of 24 and 48 h were calculated by SPSS Software. And the IC 25 , IC 50 , and IC 75 were chosen as the low, medium and high doses of TB for further use. Animal experiment in zebrafish To determine the dose range of TB, larval zebrafish (3 dpf ) were treated with various concentrations of TB (0, 6.25, 12.5, 25, 50, 100, 250, 500, 1000, and 2000 µg/ml) for 24 h, followed by the observation under a stereoscopic microscope. According to the preliminary studies, no observed adverse effect level (NOAEL) of TB was estimated. Each larval zebrafish (2 dpf ) was microinjected 200 SK-Hep-1 cells labeled with CM-Dil (red fluorescence) into the yolk sac as previously described [37]. After 24 h, 30 larval/group were selected under fluorescence microscope and randomly cultured in 6-well plates. The model group, Cis-platin (15 µg/ml) group, and TB (1.7, 5.6, and 16.7 µg/ml) group were set up. Zebrafish of each group were observed under fluorescence microscope after 24 h treatment. The fluorescence intensity (FI) of SK-Hep-1 xenograft tumor was calculated by Image pro plus 6.0 software. The inhibitory rate (IF) was calculated as: DAPI staining The apoptosis of TB-treated SK-Hep-1 cells was detected using DAPI staining. Briefly, SK-Hep-1 cells were treated with TB at 50, 100, and 150 µg/ml for 24 h respectively. Cells were fixed with 4% paraformaldehyde for 15 min and then stained with DAPI (Invitrogen, USA) for 4 min in dark. The apoptotic nuclei of cells (bright blue nuclei and condensed chromatin with apoptotic bodies) were observed under a fluorescence microscope. Flow cytometry The apoptosis of TB-treated SK-Hep-1 cells was also detected using an Annexin-V/PI apoptosis kit. Briefly, TB-treated cells were collected and then labeled with FITC Annexin V and PI according to the manufacturer's protocol. Fluorescence intensity of the cells was measured by a flow cytometry (BD Biosciences, USA). The analysis was replicated thrice and the values of upper right and lower right quadrants of the flow cytometric dot plot were summed to calculate the apoptosis rate (%) for each TB treatment. To quantify the percentage of SA-β-gal-positive cells, five digital images were randomly captured by a microscope and the positive cells from each group were counted. Quantitative real time PCR (qPCR) assay As previously described [38], total RNA was extracted using TRIzol reagent and quantified by NanoDrop2000 spectrophotometer (Thermo, USA). Then the RNA was reverse transcribed to cDNA using Primescript RT master mix (TaKaRa, Japan). According to the instructions, mRNA quantity was determined by RT-qPCR system (Applied Biosystems, USA) with SYBR Premix Ex Taq II (Tli RnaseH Plus). The relative mRNA expressions were analyzed by 2 −ΔΔCT method and β-Actin was used as the reference gene. Primer sequences are shown in Table 1. Statistical analysis The data analyses were performed using SPSS statistics software and were expressed as the mean ± standard deviation (SD). Statistical significance among different groups were examined using one-way ANOVA followed by Fisher's least significant difference (LSD) comparison. p < 0.05 was considered statistically significant. Anti-proliferative effect of TB Anti-proliferative effect of TB on HCC cell lines was determined by cell viability assays and cell morphological observation. TB inhibited proliferation of SK-Hep-1, HepG2 and Huh7 cells from 50 to 600 µg/ml (Fig. 1a). Given that the inhibitory effect on SK-Hep-1 cells was the strongest, SK-Hep-1 cells were selected for further assays. As depicted in Fig. 1b in vitro assays. Microscopic analyses showed that TB treatment decreased the number of SK-Hep-1 cells and increased the number of the round cells and shrunken cells (Fig. 1c). Anti-tumor effect of TB in vivo The zebrafish mortality and adverse events caused by TB were shown in Fig. 2a. TB obviously induced the adverse events (abnormal roll over) of zebrafish from 25 to 500 µg/ml. And the first fish death was observed at 500 µg/ml of TB, and all fishes were dead at 1000 µg/ml. Based on regression curve, TB's NOAEL was calculated as 16.7 µg/ml. And 1.7, 5.6 and 16.7 µg/ml were selected as low, medium and high doses for in vivo experiment. As depicted in Fig. 2b, the HCC xenograft model was established successfully in zebrafish and TB dose-dependently suppressed SK-Hep-1 tumor growth. The inhibitory rates of TB at 1.7, 5.6 and 16.7 µg/ml were 26.15%, 32.78% and 56.30%, respectively. Cis-platin, one of the first-line of anticancer drugs, was applied as a positive control group. As depicted in Fig. 2c, high dose of TB (16.7 µg/ml) exerted comparable tumor-inhibitory effect to Cis-platin (15 µg/ml) in the zebrafish model within a short period. Pro-apoptotic effect of TB DAPI staining was conducted to assess TB-induced apoptosis of SK-Hep-1 cells. As depicted in Fig. 3a, the untreated cells had rounded nuclei with normal blue color, whereas the TB-treated cells had bright blue nuclei and condensed chromatin with apoptotic bodies (indicated by arrows). Similarly, Annexin-V/PI staining also showed significant TB-induced apoptosis on SK-Hep-1 cells (Fig. 3b). The total apoptotic rates (early and late) of TB at 50, 100 and 150 µg/ml were 10.40%, 34.38% and 66.04%, respectively. These results suggested that TB dose-dependently induced apoptosis on SK-Hep-1 cells. Pro-senescent effect of TB SA-β-gal assay was conducted to evaluate the senescence induced by TB on SK-Hep-1 cells. As shown in Fig. 4, the number of SA-β-Gal-positive cells was obviously increased with TB treatment from 50 to 100 µg/ml, indicating that TB dose-dependently induced senescence on SK-Hep-1. Molecular actions of TB The relative mRNA expression of targeted genes in TBtreated SK-Hep-1 cells was determined by qPCR assay. As depicted in Fig. 5, the expression of P53 was dosedependently up-regulated by TB at the transcriptional level. TB also increased the expression of down-stream senescent genes (P16, P21, IL-6 and IL-8) and apoptotic genes (Bax, Bim and PUMA). The protein expression of targeted molecules in TBtreated SK-Hep-1 cells was determined by WB. A strong phosphorylation of ATM, ATR, Chk2, and p53 was detected in TB-treated SK-Hep-1 cells (Fig. 6), indicating the activation of p53 signaling pathway. From low to high doses, TB obviously increased the expression of downstream apoptotic markers (γ-H2AX, c-PARP, c-Casp9 and Bax ) with down-regulation of anti-apoptotic protein Bcl-2. The regulatory effects of TB on these genes and protein were dose-dependent. Verification of p53-mediated mechanism of TB p53-siRNA was applied to determine whether p53 signaling mediates the senescence and apoptosis in TB-treated SK-Hep-1 cells. The transfection efficacy was confirmed by qPCR and WB assays ( Fig. 7c and d). Cell viability assay and DAPI staining showed that p53-siRNA significantly counteracted the TB-mediated anti-proliferative and pro-apoptotic effect on SK-Hep-1 cells (Fig. 7a and b). qPCR assays revealed that p53-siRNA significantly suppressed the expression of senescent genes (P21, GADD45α and IL-6) and antagonized the regulation of TB on these senescent genes and apoptotic genes (Bax and Bim) (Fig. 7c). WB assays revealed that p53-siRNA notably decreased the expression of p21 and Bax, increased the expression of Bcl-2, and counteracted the regulation of TB on these proteins. Our previous studies have shown that TB induced apoptosis on p53-mut HCC cells via activation of JNK signaling [39]. Thus, JNK-siRNA was applied to verify the involvement of JNK in the action mechanism of TB. As shown in Fig. 7e, TB obviously up-regulated the expression of JNK1 and JNK2 and JNK-siRNA substantially reversed the regulation of TB on these genes as well as on the senescent (GADD45α and IL-8) and apoptotic genes (Bax and PUMA). The above results indicated that p53 signaling and JNK participated in the TB-induced cellular senescence and apoptosis of SK-Hep-1 cells. Discussion TB is known to improve the lipid metabolism and reduce the cholesterol level of liver, indicating its beneficial effect on liver [34]. Our study revealed that TB exhibited significant inhibitory effects on HCC cell lines (SK-Hep-1, HepG2, and Huh7), in which the effect on p53-WT SK-Hep-1 cells was the strongest. HepG2 cells are well-differentiated, which can well defend cellular stresses. In comparison, SK-Hep-1 cells were poorly differentiated with more sensitivity to cellular stresses, and easily suffered from severer DNA strand breaks [40]. Thus, SK-Hep-1 cells might be more sensitive to the treatment of TB, which explained why the effect of TB on HepG2 cells was milder than SK-Hep-1. Accordingly, we conducted the present study to determine the anti-HCC efficacy and explore the mechanism of TB on SK-Hep-1 cell line. Our findings demonstrated that TB significantly suppressed SK-Hep-1 tumor growth in xenograft zebrafish. Zebrafish, an important model for cancer research, overcomes the drawbacks of murine xenograft models as follow: (1) no immune rejection at larval stage; (2) transparent body enabling live imaging of tumor growth; and (3) lower ethical impact when used at the larval stage [41]. Surprisingly, we found that Cis-platin showed slightly weaker inhibitory effect than high dose of TB. This maybe because oral administration of Cisplatin through cultured water was performed instead of the routine intravenous injection, which might reduce its bio-availability. Interestingly, the reported in vitro IC 50 of Cis-platin (5.13 ± 0.09 µg/ml) was much lower than that of TB [42], suggesting a difference of in vivo mechanism (See figure on next page.) Fig. 7 a Apoptotic morphology of SK-Hep-1 cells with p53-siRNA and TB (100 µg/ml) treatment by DAPI staining. Scale bar: 50 μm. Values are presented as mean ± SD (n = 5). *p < 0.01 vs. siNC group, ## p < 0.01 vs. siNC group plus TB treatment. b Inhibitory rate of TB (100 µg/ml) with nontargeting control siRNA or p53-siRNA treatments on SK-Hep-1 cells at 24 h. Values were presented as the mean ± SD (n = 5). ## p < 0.01 vs. siNC group plus TB treatment. c Relative mRNA expression of SK-Hep-1 cells with p53-siRNA and TB (100 µg/ml) treatments for 24 h. Values are presented as mean ± SD (n = 3). p < 0.05 and *p < 0.01 vs. siNC group, # p < 0.05 and ## p < 0.01 vs. siNC group plus TB treatment. d Relative protein expression of SK-Hep-1 cells with p53-siRNA and TB (100 µg/ml) treatments for 24 h. Values are presented as mean ± SD (n = 3). p < 0.05 and *p < 0.01 vs. siNC group, ## p < 0.01 vs. siNC group plus TB treatment. e Relative mRNA expression of SK-Hep-1 cells with JNK-siRNA and TB (100 µg/ml) treatments for 24 h. Values are presented as mean ± SD (n = 3). p < 0.05 and **p < 0.01 vs. siNC group, # p < 0.05 and ## p < 0.01 vs. siNC group plus TB treatment. siNC, nontargeting control siRNA-treated group; sip53, p53-siRNA-treated group; siJNK, JNK-siRNA-treated group Xu et al. Cancer Cell International (2022) between TB and Cis-platin. The effective dose range of TB (from 1.7 to 16.7 µg/ml) in zebrafish can be estimated as 0.08 to 0.80 mg/kg in human by dose conversion, and dosage of green tea in human was roughly 1.13 to 13.17 mg/kg, which are very low and suitable for clinical application [43,44]. In vitro results showed that TB induced cellular senescence and apoptosis of SK-Hep-1 cells through activation of ATM-Chk2-p53 cascade with bypass regulation of JNK. The innovation points of this study are as follows: (1) demonstration of the inhibitory efficacy of TB on p53-WT HCC cells (SK-Hep-1); (2) clarification of the action mechanism of TB through cell apoptosis and cellular senescence; and (3) determination of TBʹs molecular mechanism via ATM-Chk2-p53 signaling pathway with JNK bypass regulation. Cellular senescence is a cell state characterized by permanent cell-cycle arrest with widespread changes in chromatin organization and gene expression [45,46]. p53 and pRb are critical transcriptional regulators in cellular senescence. p21 is one of the most important targets of p53 transcriptional activity in senescent process, whereas p16 is a positive upstream regulator of pRB [47]. A vital feature of senescent cells is the secretion of senescent associated secretory phenotype (SASP), such as proinflammatory cytokines and chemokines, growth factors, etc. [48,49]. Except for p53, p16, and p21, IL6 and IL8 are also the central components of SASP and act as important markers of cellular senescence [50,51]. Our data showed that TB significantly induced senescent phenotype (SA-β-gal positive) of SK-Hep-1 cells with up-regulation of all the above senescent markers (P53, P16, P21, IL-6 and IL-8) (Figs. 4 and 5), indicating that TB induced cellular senescence to suppress SK-Hep-1 cells through p53-related mechanism. Some chemotherapeutics possess anti-cancer efficacy via inducing cellular senescence, such as Palbociclib, a specific CDK4/6 inhibitor, which was approved in 2015 for clinical treatment of advanced breast cancer [52]. ATM and ATR are initiating kinases of DNA damage response (DDR) cascade, while γ-H2AX serves as a sensitive biomarker for DNA damage during DNA doublestrand breaks (DSBs) [53,54]. In response to the DNA DSBs, ATM phosphorylates itself at Ser1981 to activate extensive substrates to mediate cell cycle checkpoint control, DNA repair or apoptosis [55]. The serine/threonine kinase Chk2 is another component of DDR, which requires ATM-activated phosphorylation at several residues including Thr68 [56]. Activated Chk2 phosphorylates p53, enhancing its stability and activity to induce apoptosis through Bcl-2 and caspase-dependent manners [57,58]. Therefore, TB induced apoptosis of SK-Hep-1 cells with DNA DSBs through ATM-Chk2-p53 signaling pathway. Subsequently, the downstream apoptotic proteins (Bax, c-Casp9 and c-PARP) and genes (PUMA, Bim and Bax) were activated, resulting in the mitochondrial pathway of apoptosis (Fig. 8). Casp9 is the initiating caspase associated with the mitochondrial apoptosis [59]. Once activated, Casp9 cleaves and activates downstream effectors to cleave PARP, which promotes cellular disassembly and serves as a hallmark of apoptosis [60,61]. Previous report has shown that p53 signaling plays a critical role in modulating cellular responses to DNA damage, leading to irreversible cellular senescence and apoptosis [62]. This study verified the p53 signalingmediated pro-apoptotic and pro-senescent mechanism of TB by using p53-siRNA (Fig. 7). In our previous study, TB induced apoptosis in p53-mut Huh7 cells via activation of JNK signaling pathway, indicating that JNK might also participate into the action mechanism of TB on HCC cells [39]. Correspondingly, we revealed that JNK was a bypass regulator involved in the mechanism of TB (Fig. 7e), suggesting a multi-target mechanism of TB on both ATM-Chk2-p53 and JNK signaling (Fig. 8). Alike most natural products, TBʹs in vitro effects on cancer cells are milder than chemotherapeutics, with higher IC 50 s. However, this study demonstrated that TB had considerable efficacy against HCC cell mass in xenograft zebrafish and the in vivo efficacy was even better than Cis-platin. Our finding was in consistent with previous reports about the in vivo anti-cancer study of TB [37,39], which suggested that TB might not only induce cancer cell apoptosis in a direct way, but also suppress in vivo cancer cell mass in an indirect way (e.g., anti-angiogenesis and immunoregulation) [63][64][65]. Further studies are needed to explore the in vivo mechanism of TB.	v3-fos-license
2017-08-03T02:32:48.225Z	2014-02-11T00:00:00.000	20127922	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GREEN", "oa_url": "https://europepmc.org/articles/pmc4125753?pdf=render", "pdf_hash": "1be1850663f9a00e0bc7ae0efc7f2df5b68b5cfd", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45388", "s2fieldsofstudy": [ "Medicine" ], "sha1": "be11c8411b7438d1da20203ab903c8286f27f499", "year": 2014 }	pes2o/s2orc	Requirement for prevention of periodontitis in patients with metabolic syndrome There is growing evidence that suggests a relationship among periodontitis and the risk factors for cardiovascular disease, including obesity, dyslipidemia, diabetes and hypertension; their association is known as metabolic syndrome (MetS). Periodontal disease was proposed as a part of MetS (1). It was found that women with more components of MetS had a significantly higher odds ratio for greater probing depth and clinical attachment loss (2). There is growing evidence that suggests a relationship among periodontitis and the risk factors for cardiovascular disease, including obesity, dyslipidemia, diabetes and hypertension; their association is known as metabolic syndrome (MetS). Periodontal disease was proposed as a part of MetS (1). It was found that women with more components of MetS had a significantly higher odds ratio for greater probing depth and clinical attachment loss (2). Objective of the study was to compare periodontal condition and gingiva microcirculation in patients with different MetS components and healthy volunteers without periodontal lesions. Study design 40 patients with MetS (aged 34-64, 38% males) were equally divided into two groups depending on the number of MetS components. In the first group were 20 patients with central obesity and 2>MetS factors (hypertension/ diabetes/ dyslipidemia/ microalbuminuria); in the second group were 20 obese patients with one MetS factor. The control group included 15 healthy volunteers (age 19-23 years; 50% males) without periodontal lesions. Methods Periodontal condition of all patients was accessed using Simplified Oral Hygiene Index (OHI-S), Silness & Loe plaque index (PI), sulcus bleeding index (SBI), papillarymarginal-alveolar index (PMA), Russell periodontal index (RPI); panoramic X-ray was taken. Periodontal blood flow was measured by ultrasonic Doppler device "Minimax-Doppler-K" ("Minimax", Saint-Petersburg) with making indirect cold test. Statistical analysis was performed using SPSS package. Results Patients with MetS had poor oral hygiene (OHI-S 1 >2,3; OHI-S 2 >1,9). In 1 st group severe periodontitis was diagnosed in 56% of patients while in 2 nd group it was found only in 25% of patients. There was no significant difference in SBI and PMA between patients with MetS, however the difference was significant between patients with MetS and control group (p 1,3; 2,3 <0,001). RPI was also higher in 1 st group than in 2 nd group (p 1, 2 <0,05). Periodontal pockets were deeper in 1 st group than in 2 nd group (7,3±1,9 mm vs. 5,6±0,9 mm, p 1,2 <0,05). Patients with 2>MetS factors had a linear velocity value (V, mm/s) 26% lower and a volume velocity value (Q, mm3/sec) 35% lower than did patients of 2 nd group. In 1 st group in comparison with control group reduction of V was 49% and reduction of Q was 57% while the difference between 2 nd group and controls was 32% in V and 39% in Q. The normal vascular response of gingiva vessels to cold test was found only in 63% controls. 100% of patients with 2>MetS factors developed the atypical vascular response to cold test while in 2 nd group it was determined in 69% of patients. The weakened vascular response was detected in 31% of patients in the 2 nd group and 27% of controls. Conclusions Patients with more MetS components develop more pronounced periodontal destruction and gingiva microvascular dysfunction than patients with less number of MetS components. Patients with MetS have reduced blood flow velocity values (V>32%; Q>39%) in comparison with healthy controls. Patients with central obesity and even one MetS factor are recommended to have complex dental investigation and supportive periodontal treatment.	v3-fos-license
2022-12-03T14:09:45.270Z	2018-02-28T00:00:00.000	254170266	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://link.springer.com/content/pdf/10.1007/s10182-018-0322-y.pdf", "pdf_hash": "d294c7c6309639690b76e5e3e6a233b208104842", "pdf_src": "SpringerNature", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45389", "s2fieldsofstudy": [ "Mathematics" ], "sha1": "d294c7c6309639690b76e5e3e6a233b208104842", "year": 2018 }	pes2o/s2orc	Study of the bivariate survival data using frailty models based on Lévy processes Frailty models allow us to take into account the non-observable inhomogeneity of individual hazard functions. Although models with time-independent frailty have been intensively studied over the last decades and a wide range of applications in survival analysis have been found, the studies based on the models with time-dependent frailty are relatively rare. In this paper, we formulate and prove two propositions related to the identifiability of the bivariate survival models with frailty given by a nonnegative bivariate Lévy process. We discuss parametric and semiparametric procedures for estimating unknown parameters and baseline hazard functions. Numerical experiments with simulated and real data illustrate these procedures. The statements of the propositions can be easily extended to the multivariate case. Introduction In survival analysis, frailty is usually defined as a non-observable momentan risk of failure and is included in survival models in the form of an unknown nonnegative random variable or random process characterizing non-homogeneity of population with respect to hazard function. Usually, frailty enters the model multiplicatively to the hazard function and allows us to take into account the correlations between failure times. Observed covariates can also be included in the multivariate survival models in the form of Cox-like regression. Identifiability and other properties of the univariate and multivariate survival models with time-independent frailty have been studied in some depth, and these models now are popular and widely used in survival studies and in the search for genes influencing longevity. Several models with time-dependent frailty have also been suggested. Woodbury and Manton (1997) introduced a stochastic process model of human mortality and ageing. They considered hazard as a quadratic function of stochastically changing unobserved frailty. The model was further extended by Yashin and Manton (1997) to consider a partially observed frailty process. Modifications of this model were successfully applied to the analyses of longitudinal data with informative censoring. Ideas on further development and applications of these results to studying ageing and longevity are summarized by Yashin et al. (2012). Gjessing et al. (2003) considered an approach based on the proportional hazard model with time-dependent frailty given by the formula Here, X (t) is a nonnegative Lévy process (subordinator) with Laplace exponent (c), R(t) = t 0 r (u)du defines the time transformation of subordinator for a nonnegative rate function r (t), and the nonnegative weight function a(u, s) determines the extent to which the previous behavior of transformed subordinator influences the hazard function at time t. The authors derived the expressions for the population survival and hazard functions in a general case: where λ(t) is the baseline hazard function and b(u, t) = t u λ(s)a (u, s−u)ds. Gjessing et al. (2003) showed also that under some conditions, quasi-stationary distributions of survivors can arise. This implies the constant limiting population hazard rate, in spite of the increase of the baseline hazard function. Ata and Özel (2012) considered the proportional hazard model with time-dependent frailty given by the discrete compound Poisson process and applied this model to the earthquake data and to traffic accidents data from Turkey. The aim of this paper is to study the problem of identifiability for bivariate survival models with/without observed covariates and with time-dependent frailty when this frailty is given by a Lévy process (or Lévy processes). In addition, we demonstrate how these models can be used for longevity datasets based on simulated data. In Sect. 2, we give the definitions of the univariate survival model under mixed proportional hazard specification and the bivariate correlated frailty model. We then discuss the definitions of the uni-and bivariate survival model with time-dependent frailty given by the nonnegative Lévy processes. At the end of this section, we discuss known findings related to the identifiability of survival models with time-independent frailty, formulate two new propositions about the identifiability of the bivariate survival models with time-dependent frailties, and give the EM algorithm for estimating unknown baseline hazard functions and parameters for the correlated bivariate model with time-dependent frailty. In Sect. 3, we discuss the results of estimations of the baseline hazard functions and unknown parameters (including the Cox-regression coefficients and parameters characterizing the frailty process) in experiments with simulated and real data for parametric and semiparametric approaches. The real-world example was based on the data from the Danish Twin Registry. Conclusions and outlook are presented in Sect. 4. Survival analysis under a frailty setting Under mixed proportional hazard specification (Abbring and van den Berg 2003), the hazard rates of the failure times for two related individuals depend multiplicatively on the respective baseline hazards λ j (t), regressor functions χ j (u j ) with observed vector of covariates u j , and unobserved nonnegative random variable (frailty) Z j , characterizing the heterogeneity in the population with respect to hazard λ j The function χ j (u j ) is frequently specified as χ j (u j ) = exp(β * j u j ) (the Coxregression term) for some transposed vector parameter β * j , j = 1, 2. The univariate population survivals S j (t\|χ j (u j )) = P(T j > t j ) for random times of death T j are with cumulative hazard function j (t) = t 0 λ j (τ )dτ and Laplace transform L(c) = Ee −cZ . In the bivariate correlated frailty model proposed by Yashin et al. (1995), individual frailties (Z 1 , Z 2 ) were constructed under assumptions for independent nonnegative random variables Y 0 , Y 1 and Y 2 and some nonnegative constants m 0 , m 1 and m 2 . Given frailties (Z 1 , Z 2 ), life spans for both individuals were assumed to be conditionally independent. Since the scale factor common to all subjects in the population can be absorbed into the baseline hazard functions λ j (.), j = 1, 2, we can put EZ 1 = EZ 2 = 1. The heterogeneity in the population can be characterized by the variance of frailties VarZ 1 = σ 2 1 , VarZ 2 = σ 2 2 . The correlation between frailties Corr(Z 1 , Z 2 ) we will denote by ρ. The assumption that the individual frailty is determined at birth and does not change with age seems to be too strong and unrealistic. To make the approach more flexible, we can weaken this assumption and suppose that the frailty is a random process. Similarly to Gjessing et al. (2003), we shall consider the frailty process Z = Z (t) : t 0 defined by a nonnegative Lévy process. In accordance with Lévy-Khinchin formula, such a process can be characterized by its Laplace transform with Laplace exponent (c), while c is the argument of Laplace transform. Note that Examples of Lévy processes include the standard compound Poisson process, the compound Poisson process with general jump distribution, gamma processes, stable processes, and PVF (power variance function) processes (Gjessing et al. 2003). In this paper, we shall consider the nonnegative Lévy processes (subordinators) with the Laplace exponent given by nonnegative drift d and the jump measure ν with support (0, ∞) satisfying ∞ 0 min(1, x)ν(dx) < ∞. The Laplace exponent is an increasing and concave function. The Laplace exponent and the jump measure for the gamma process are given by the formulas with the shape ht and the scale parameter γ for the gamma-distributed random variable Z (t). This corresponds to the values of htγ −1 and htγ −2 for mean and variance of Z (t), respectively. To avoid non-identifiability of the model, we shall standardize the frailty distribution and put h = γ . In this case, EZ (t) = t, VarZ (t) = tγ −1 for t 0. In "Appendix 1," we can find the formulas for calculating univariate population survivals in this case for a constant and an exponential (Gompertz) baseline hazard functions. We will denote the Laplace exponent for the univariate frailty processes Z j (t), j = 1, 2 by j (.). Let (Z 1 (t), Z 2 (t)) be a bivariate Lévy process with nonnegative components and the Laplace exponent given by where d, c ∈ R 2 + , < x, y > denotes the dot product of vectors x, y ∈ R 2 , the Lévy measure ν(A) for any Borel set A ∈ R 2 + is the expected number of jumps on the time interval [0,1], whose sizes belong to A, and the following integrability conditions for Lévy measure are satisfied: Note that biv (c 1 , 0) = 1 (c 1 ) and biv (0, c 2 ) = 2 (c 2 ) for marginal Lévy processes. Model identifiability The identifiability of the univariate model with unspecified functional form of frailty distribution and baseline hazard has been studied by Elbers and Ridder (1982). This model is identifiable given information on T for finite EZ and is not identifiable when frailty has an infinite mean. Identifiability of the correlated frailty models using data on the pair (T 1 , T 2 ) was proved by Honoré (1993) under the assumption of finite means of Z 1 and Z 2 . Yashin and Iachine (1999) proved the identifiability of the correlated frailty model without observed covariates assuming that Z 1 and Z 2 are gamma distributed. Abbring and van den Berg (2003) studied the identifiability of the mixed proportional hazards competing risks model. We adopt this method to investigate the identifiability of the mixed bivariate survival model for time-dependent correlated frailties. Then, the mixed bivariate survival model for time-dependent correlated frailties is identified from the bivariate distribution of the failure times (T 1 , T 2 \|u 1 , u 2 ). Differentiating S j (t\|χ j ) in t, dividing then by χ j , and setting formally χ j → 0, we get the following equations: Since the expression on the right hand of (9) is observed, we find the hazard function in the form The constant j (0) can be found from the equation j (t * ) = 1, j = 1, 2 (Assumption 2). If individual frailties (Z 1 , Z 2 ) are constructed under assumptions given by for some positive α, we can weaken the conditions of Proposition 1 and prove the identifiability of the model in the absence of observed covariates. Proposition 2 Let the following assumptions be satisfied. Assumption 2. Baseline hazard functions for some real number b > 0 and i = 0, 1, 2. Then, the mixed bivariate survival model for time-dependent correlated frailties is identified from the bivariate distribution of the failure times (T 1 , T 2 ). The proof of Proposition 2 is given in "Appendix 3." Model validation In this section, we will assume that χ(u) = exp(β * u). To validate the regression model, we need to estimate the vectors of Cox-regression parameters β 1 and β 2 , vector parameter defining frailty ζ [in the case of correlated frailty model either (σ 2 1 , σ 2 2 , ρ) for time-independent gamma-frailty or (γ −1 1 , γ −1 2 , ) for gamma-frailty process given by the Laplace exponent (6)], and the baseline hazard function λ 1 (t) and λ 2 (t). If the baseline hazard functions follow a parametric form such as the Gompertz or the Weibull function with vector parameter ξ , we can use the classic maximum likelihood method to estimate all unknown parameters. The log-likelihood in this case is given by Here, θ = (β, ζ, ξ) is the vector of unknown parameters, β = (β 1 , β 2 ) stands for Cox-regression coefficients, ζ for frailty parameters, and ξ for parameters defining the baseline hazard functions λ j (t), j = 1, 2. The data include information about life spans (t i1 , t i2 ), observed covariates (u i1 , u i2 ), and censoring (δ i1 , δ i2 ) for a twin pair i, i = 1, . . . , n. The estimate of the vector parameter θ we can find by maximizing the log-likelihood function (13). If the form of the baseline hazard functions is not specified, the estimates can be obtained by the EM algorithm. This algorithm combines the maximum partial likelihood estimator of the vector parameter (β, ζ ) with the Breslow estimator (Breslow 1972) of the cumulative baseline hazard function (t) = ( 1 (t), 2 (t)). The EM algorithm is an iterative procedure with two steps-E (expectation) and M (maximization) on each iteration. It works as follows: Let f (z i1 , z i2 \|t i1 , t i2 , ζ ) be the density function of the random variable (Z i1 (t i1 ), Z i2 (t i2 )) given parameter vector ζ , i = 1, . . . , n. Denote the estimates of (t), ζ , and β on the l th iteration byˆ l (t), ζ l , andβ l , respectively. Similar to Gorfine and Hsu (2011), we define the failure counting process N i j = δ i j Ind(T i j t) and the at-risk process X i j = Ind(T i j t), where T i j is the random time to the failure of twin j in twin pair i, i = 1, . . . , n, j = 1, 2. Define random processes where the symbol "ˆ" means replacing the unknown frailty Z i j (t) with its conditional expectation given the observed data and the current estimates of (t), ζ and β. The convergence of the EM algorithm and the properties of parameter estimates have been discussed elsewhere (Zeng and Lin 2007). For the correlated frailty model with time-independent gamma-distributed frailty, the calculation of expression (16) has been discussed in detail by Iachine (1995). The calculation of this expression in the case of the gamma-frailty process for the same model can be found in "Appendix 4." In the parametric approach, the choice of the appropriate baseline hazard function plays an important role. The Gompertz function does not guarantee the good fit of the marginal survival function for real longevity data. The following gamma parameterization of the univariate survival function gives better results by fitting the real data in the model with time-independent frailty is the pseudo-baseline cumulative hazard, t 30, s 2 ,ã,b > 0, and σ 2 > 0 is the variance of the time-independent frailty. Given parameters s 2 , σ 2 ,ã, andb, it is not difficult to find the true baseline cumulative hazard (t) in the form In the case of the time-dependent frailty given by a Lévy process with Laplace exponent (c), we need to consider the following analog of Eq. (17): Unfortunately, Eq. (18) does not have a closed-form solution with respect to (t). In experiments with real data, we will find the approximative solution to (18) such that the function dyn appr (t) is a non-decreasing, nonnegative, piecewise-constant function satisfying the following conditions: dyn appr (0) = 0, for times-to-event t i , i = 1, . . . , 2n, sorted in non-decreasing order, 0 can be calculated recurrently for i = 1, 2, . . . , 2n using a simple bisectional procedure. Note that the function dyn appr (t) converges pointwise to the solution of (19) as n → ∞ and the distance between neighboring moments t i tends to zero. To compare two approaches, we assume that in the case of the time-independent frailty, the cumulative hazard is also a non-decreasing, nonnegative, piecewiseconstant function stat appr (t) satisfying the following conditions: stat 3 Results Experiments with simulated data In this subsection, we will discuss the results of the consistency test for the correlated frailty models with time-dependent and time-independent frailties (11). It was assumed that α = 1, VarZ 1 = VarZ 2 = σ 2 , and Corr(Z 1 , Z 2 ) = ρ in the case of the time-independent frailty or VarZ 1 (1) = VarZ 2 (1) = γ −1 and Corr(Z 1 (1), Z 2 (1)) = in the case of the time-dependent frailty, baseline hazard functions λ j (t) followed Gompertz (exponential) form λ j (t) = a exp(bt), and an observed covariate u influenced longevity so that the conditional hazard function was defined by μ j (t j \|Z j , u j ) = Z j exp(βu j )λ j (t), j = 1, 2. The covariates were randomly generated from the uniform distribution on the interval [0,1] and were independent for the individuals. The (true) values for data generating are given in Tables 1 and 2 and have been chosen so that the mean and the standard deviation of the generated times-to-event were equal to approximately 75 and 12 years, respectively. The bivariate times-to-event have been generated using formula (4) with σ 1 = σ 2 = σ in the case of the time-independent frailty and using formula (25) given in "Appendix 2" in the case of time-dependent frailty. In both cases, it was assumed that 1 (t) = 2 (t) = (a/b)(exp(bt) − 1) and χ 1 (u) = χ 2 (u) = exp(βu). We have not truncated or censored the generated data. We estimated unknown parameters and cumulative hazard functions using the classic maximum likelihood estimator (parametric method) and the EM algorithm (semiparametric method). In all cases, we simulated 100 datasets for 500 twin pairs. Table 1 shows the results of simulation study for the time-independent gammafrailty model without truncation and censoring. Empirical means and standard deviations of estimates were calculated using the classic maximum likelihood method and the EM algorithm, respectively. Table 2 shows the results of the simulation study for the time-dependent gammafrailty model without truncation and censoring. Empirical means and standard deviations of estimates were also calculated using the classic maximum likelihood method and the EM algorithm, respectively. Analysis of estimates in both tables does not indicate the presence of any bias, and estimates calculated using the classic maximum likelihood estimator are generally more efficient. Furthermore, the estimates of the Cox-regression coefficients and parameters characterizing the frailty distribution are closer to true vales than those for the EM algorithm. One can see in Fig. 1 that in both cases (the time-dependent and the time-independent frailty), the empirical mean log baseline cumulative hazard trajectory calculated using the EM algorithm fits the true log baseline cumulative hazard trajectory very well. Experiments with real data For experiments with real data, we used the datasets from the Danish Twin Registry (DTR). This registry was created in the 1950s. It is one of the oldest populationbased registries in the world and contains information about twins born in Denmark since 1870 and who survived to age 6. Multiple births were manually ascertained in birth registers from all 2200 parishes in Denmark. As soon as a twin was traced, a questionnaire was mailed to the twin, to her/his partner or to their closest relatives if neither of the twin partners were alive. The zygosity of twins was assessed on the basis of questions about phenotypic similarities. The reliability of the zygosity diagnosis was validated by comparing laboratory methods based on the blood, serum, and enzyme group determination. In general, the misclassification rates were less than 5%. Other information includes the data on sex, birth, cause of death, health, and lifestyle. An important feature of the Danish twin survival data is their right censoring and left truncation. In our study, we used the longevity data on the like-sex twins with known zygosity born between 1870 and 1900 and who survived until age 30. This non-censored data include 470 male monozygotic (MZ) twin pairs, 475 female MZ twin pairs, 780 male dizygotic (DZ) twin pairs, and 835 male DZ twin pairs. Further details on the Danish Twin Registry can be found in Hauge (1981). Since the EM algorithm suffers from its slow convergence, we have estimated unknown parameters for the time-independent and the time-dependent frailty models using the classic maximum likelihood method. Table 3 shows these estimates, the logarithm of the maximum value of the likelihood function, and the value of the AKAIKE Information Criterion (AIC) for the data from the Danish Twin Registry described above. The estimates of parameter s 2 were very close to zero in all experiments with real data. We have put this parameter equal to zero to avoid the efficiency loss of estimates. The AIC values for the model with time-dependent frailty are slightly smaller than the ones for the model with time-independent frailty. That is, the model with time-dependent frailty is slightly more informative than the one with time-independent frailty. Figures 2 and 3 show estimated and empirical bivariate probability density functions for the time-independent and the time-dependent frailty models. Note that the shapes of the estimated bivariate probability density functions are very similar. Discussion Frailty models are a powerful tool for studying non-observable inhomogeneity in a population related to time-to-failure (e.g., death or disease). Models with timeindependent frailty have been intensively studied over the last decades and have found a wide range of applications in survival analysis and in searching for genes influencing longevity. However, the studies based on the models with time-dependent frailty are scarce. In this paper, we have attempted to improve the knowledge in this area and to study some properties of multivariate survival models with time-dependent frailty components. Proposition 1 we have formulated and proved for the bivariate case. It is not difficult to generalize this result and to prove the identifiability of the frailty model with observed covariates for any number J of related individuals equal or greater than 1 if the time-dependent frailty is a multivariate Lévy process. Similarly, we can generalize Proposition 2 for the case of J 2. However, the number of frailty components in the multivariate analog of the decomposition (11) will be equal to 2 J − 1. The shared frailty model where all individuals in a family or cluster share the same non-observable risk of failure does not meet this problem. In experiments with simulated data, we tested for consistency and used parametric and semiparametric approaches. In the parametric approach, we assumed that the parametric form of the baseline hazard functions is known and follows the Gompertz form. All unknown parameters characterizing frailty distribution, baseline hazard function, and Cox-regression parameters were estimated directly by maximizing the likelihood function. In the semiparametric approach, we used the EM algorithm and estimated the Cox-regression parameters and the parameters of frailty distribution by maximizing the partial likelihood function. The cumulative baseline hazard function was estimated using the Breslow estimator regarding this function as infinite-dimensional parameters. The EM algorithm suffers from its slow convergence. Moreover, in the semiparametric approach, the number of calculations increases with the number of individuals much more rapidly than in the parametric one. It leads to the drastic slowing of the convergence of the EM algorithm and increases substantially the time of Experiments with real data show that the proposed method and the method with time-independent frailty produce similar shapes of the estimated bivariate probability density functions. The baseline cumulative hazard functions have been chosen so that the estimated marginal survival functions guarantee the best fit to the empirical ones according to . A large degree of similarity of the estimated bivariate density functions for the models with time-dependent and time-independent frailties in the range of ages 30-100 years guarantees the similar bivariate fit. The difference between the two approaches can involve the shape of the baseline hazard functions and the asymptotic behavior of the bivariate probability density functions. The models with time-dependent and time-independent frailties are not nested. Therefore, we cannot compare them using the likelihood ratio test. For this purpose, the AKAIKE Information Criterion can be used. In accordance with this criterion, the model with time-dependent frailty is slightly more informative compared to the one with timeindependent frailty for the data we considered. Gorfine and Hsu (2011) studied the robustness of the multivariate survival models with frailty components against the violations of the model assumptions. It was found that unnecessary modeling of the dependency between the frailty variates can lead to some efficiency loss of parameter estimates. Misspecification of the frailty distribution can introduce a bias in estimates. Misspecification of the baseline hazard functions can lead to severe bias of all estimates if we use the parametric maximum likelihood estimator, where the baseline hazard functions follow the parametric form. The nonparametric maximum likelihood estimator does not suffer from this drawback. Note that in experiments with real data, we have used a flexible parametrization of the baseline cumulative hazard functions given by formulas (19)-(20). This parameterization does not presume any knowledge about the form of the baseline hazard function. It is sufficient to have a good approximation of the marginal survival function. An extension of the present study can include the investigation of identifiability of the survival models with competing risks and time-dependent frailty components. The piecewise-constant approximation of the cumulative hazard function has been used in experiments with real data [formulas (19)-(20)]. Other approximative functions such as piecewise linear or piecewise exponential can be used to improve the bivariate goodness-of-fit. Further, numerical experiments with real data are needed to understand whether the proposed method improves the goodness-of fit on the method with time-independent frailty. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Appendix 1: Univariate survival function for frailty model based on a gamma process 0} is a gamma process with Laplace transform given by the Lévy-Khinchin formula where G (c) is the Laplace exponent of the Lévy process Z , the argument for Laplace transform c 0 and Here, ht and γ are the shape and the scale parameters, respectively, for gammadistributed random variable Z (t). Since Z (t) is a.s. increasing in t, we can consider it as subordinator. It is easy to check that where superscript "(l)" stands for lth-order argument derivative. In nonhomogeneous population, the process Z (individual for each member of population) can characterize the individual risk of mortality (frailty). In accordance with the definition of the frailty model under mixed proportional hazard specification, we have for covariate vector u, parameter vector β, and baseline hazard λ(t). In Gjessing et al. (2003) (Theorem 1), it was shown that the population survival in this case can be calculated For constant λ(t) = λ 0 , it holds that (τ, t; u) = λ 0 e β * u (t − τ ), Using the following formula for indefinite integral we get for the baseline hazard given by the Gompertz function λ(t) = a exp(bt) Here, f (t, u) = (aχ(u)/(bγ (1 + aχ(u)e bt /(bγ )) and Li 2 (z) is the dilogarithm function defined by This function is an analytical extension of the infinite series If the baseline hazard function is given by the Gompertz function, we can use formula (22) to calculate the bivariate population survival. Appendix 3: The proof of Proposition 2 To prove Proposition 2, we need to check several auxiliary statements. Lemma 1 Let c 1 , c 2 0. Then, Proof Assume that c 1 c 2 0 and set c = c 1 − c 2 . Then, It is easy to check that Therefore, Last result holds also in the case of c 2 > c 1 0. Lemma 2 Let (c) be a Laplace exponent given by (5), and (t) is an increasing continuous function defined in R + . Then, Proof The statement of lemma follows directly from formula (5) and monotone increase of the function (.). The bivariate survival function for correlated frailty processes given by (11) is where i (.), i = 0, 1, 2, is the Laplace exponent for the frailty process Y i (t) and Using formula (28) and Assumptions 2-3 of Proposition 2, it can be shown that Note that expressions on the left hand of (29) are observable, g 1 (t) = g 0 (t, 0) and g 2 (t) = g 0 (0, t). Let L 1 be a set of all continuous increasing functions (t) on R + satisfying (0) = 0 and (t) → ∞ as t → ∞. Given Laplace exponent 0 , consider the map A 1 defined by Lemma 3 A 1 is a bijective map from L 1 to image A 1 (L 1 ) of L 1 under A 1 . Proof It is easy to see that (A 1 )(t) is a continuous monotone increasing function with (A 1 )(0) = 0 and (A 1 )(t) → ∞ as t → ∞ for ∈ L 1 . That is, A 1 (L 1 ) ⊂ L 1 . For any 0 < T < ∞, the set L T 1 of all continuous monotone increasing functions on [0, T ] satisfying (0) = 0 is a complete metric space with usual supremum distance Taking into account that (t) is differentiable in t a.e. and 0 (0) = 0, we get that and, therefore, In accordance with the mean value theorem applied to the integral on the right hand of (31), there exists τ ∈ [0, τ ] such that Since ∞ 0 xν(dx) is finite, the function on the left hand of (33) is integrable on [0, t], t < ∞. Assume that two solutions a , b of (32) satisfy a (t) = b (t) on the interval [0, t 0 ], t 0 0. We shall show now that these solutions satisfy this property on the interval [t 0 , t 0 + t ε ] for some t ε > 0 too. Let the map B 1 be given by Taking into account g (τ ) 0 and inequalities (26) and (27), we get that for all t ∈ [0, t 0 + t ε ] and some positive q < 1. It means that Since this map is also surjective, it is bijective Define the set L 2 of all functions 12 (t 1 , t 2 , α) such that for 1 , 2 ∈ L 1 , and α > 0. Given Laplace exponent 0 , consider the map A 2 defined by Proof Since (t) = 12 (t, t) ∈ L 1 , 1 (t) = 12 (t, 0) ∈ L 1 , and α 2 (t) = 12 (0, t) ∈ L 1 for t ∈ [0, ∞), in accordance with Lemma 3, these functions can be uniquely defined by their images g 0 (t, t) = (A 2 1,2 )(t, t) = (A 1 )(t), g 0 (t, 0) = (A 2 12 )(t, 0) = (A 1 1 )(t), and g 0 (0, t) = (A 2 (α 12 ))(0, t) = (A 1 (α 2 ))(t), respectively. Assume now that t 2 > t 1 and the solution 12 (t 1 , t) to (34) is uniquely defined for 0 t t 0 with t 0 t 1 . Define the map B 2 for t > t 0 by Similarly to the proof of Lemma 3, it can be shown that the solution to is also uniquely defined on the interval [t 0 , t 0 + t ε ] for some t ε > 0 and, therefore, is uniquely defined for all t t 1 . The same result can be checked in the case of t 1 > t 2 . For the sake of contradiction, assume now that there are functions a 0 (. , and real positive numbers α a and α b such that holds for all t 1 , t 2 0. In accordance with Lemmas 3-4, the maps A a i and A b i are invertible on their images of L i , i = 1, 2. Moreover, Note that i 1 (t 1 ) = i 12 (t 1 , 0) and α i i we get from (36) Now we will prove that the action of the map T ba 2 (resp. T ab 2 ) on the function a 12 (resp. b 12 ) calculated in the point (t 1 , t 2 ) depends only on the value of b 12 (t 1 , t 2 ) (resp. a 12 (t 1 , t 2 )). Lemma 5 Given the maps a 0 (.), b 0 (.), a 1 (.), b 1 (.), a 2 (.), b 2 (.), and real positive numbers α a and α b , there exist uniquely defined, monotone increasing, continuous functions f ab : R 1 + → R 1 + and f ba : Proof Put α a 1 = α b 1 = 1. From (33) and Lemma 3, it follows that given the functions are strictly monotone increasing, continuous, and uniquely defined. It holds that . Therefore, the maps f ba , are also strictly monotone increasing and continuous. Since for some continuous monotone increasing functions f ba : R 1 + → R 1 + and f ab : R 1 + → R 1 + . Substituting in (39) t 2 = 0 or t 1 = 0, we get that f ab 1 = f ab 2 = f ab and f ba Proof of Proposition 2 Note firstly that in accordance with Assumption 4, the functions i (c), i = 0, 1, 2, are analytic ones on Re(c) > −b. Since f ba (.) and f ab (.) are defined on R 1 + , additive and continuous, they are linear and it holds that f ab (x) = C ab 1 x + C ab 2 and f ba (x) = C ba 1 x + C ba 2 for some constant C ab 1 , C ab 2 , C ba 1 , and C ba 2 . As f ab (0) = f ba (0) = 0, we have that C ab 2 = C ba 2 = 0. Moreover, C ab 1 = C ba 1 = 1 and α a = α b because a i (t * ) = b i (t * ) = 1 for some t * > 0 and i = 1, 2. From here, it follows that f ab (.) and f ba (.) are identity transformations and a i (t) = b i (t) = i (t) for all t 0, i = 1, 2. It holds also that (A a 1 i )(t) = (A b 1 i )(t) for all t 0, i = 1, 2. To complete the proof of Proposition 2, we need to show that a 0 (x) = b 0 (x) for any x ∈ R 1 + . Indeed, the maps a 0 and b 0 are real analytic functions on R 1 + and, therefore, ( 1 (t) − 1 (τ )) n dτ for small values of 1 (t). This holds iff ( a 0 ) (n) (0) = ( b 0 ) (n) (0) for all nonnegative integer n and means that a 0 (x) = b 0 (x) = 0 (x) in some neighborhood of 0. Since 0 (x) is a real analytic function on R 1 + , this function is uniquely defined in this area. Similarly, it can be proved that the functions i (x), i = 1, 2, are also uniquely defined in R 1 + .	v3-fos-license
2018-04-03T05:12:31.781Z	2015-05-07T00:00:00.000	4527686	{ "extfieldsofstudy": [ "Medicine", "Chemistry" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://pubs.rsc.org/en/content/articlepdf/2015/cc/c5cc01914a", "pdf_hash": "53f84665c460d461d8a43b29687315e0fa4adb4b", "pdf_src": "ScienceParsePlus", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45391", "s2fieldsofstudy": [ "Chemistry" ], "sha1": "09d21b644eba13befb0ab8164cd1a05982a9d58d", "year": 2015 }	pes2o/s2orc	Photocatalytic multiphase micro-droplet reactors based on complex coacervation. We describe the synthesis and characterisation of novel photocatalytic multiphase micro-droplet reactors comprising TiO 2 nanosheets dispersed in poly(diallyldimethylammonium) chloride and adenosine 5 0 -triphosphate or poly(ethylene glycol) 4-nonylphenyl 3-sulfopropyl complex coacervates. We demonstrate significant variations in the degree of equilibrium partitioning of small molecule dyes into the coacervate droplet systems and exploit this behaviour to successfully conduct selective photocatalytic dye degradation. Coacervate micro-droplets are formed spontaneously by liquidliquid phase separation from mixtures of oppositely charged polyelectrolytes in aqueous solution. 1 Recent studies have demonstrated that complex coacervates can also be prepared using highly charged small molecules such as adenosine triphosphate (ATP) in combination with cationic peptides, polypeptides or polymers to produce molecularly crowded membrane-free protocells. [2][3][4] The droplets are typically tens or hundreds of micrometres in size, highly enriched in peptides and nucleotides, stable across a large pH and temperature range, and capable of sequestering a wide range of low and high molecular weight components including enzymes and their substrates. As a consequence, increased reaction rates and yields have been reported for enzyme cascade reactions confined within the coacervate phase, 2,5 suggesting that coacervate droplets present an attractive option as multiphase micro-droplet reactors. In this paper, we explore the possibility of using coacervate micro-droplets for the design and construction of new waterdispersible heterogeneous photocatalytic chemical microsystems. We demonstrate that preformed titania nanosheets (TiO 2 NS) can be spontaneously sequestered at high concentrations within the chemically enriched coacervate micro-droplets. The TiO 2 NS are prepared by HF etching, 6,7 and display predominately catalytically active {001} facets, 8,9 which increases the quantum efficiency of these photo-resistant, photo-oxidative nanomaterials. We show that a range of organic dye molecules can be sequestered by the coacervate micro-droplets, and demonstrate that charge or hydrophobic complementarity is responsible for the observed differences in partitioning. Moreover, we exploit these differences to perform selective photocatalytic dye degradation within coacervate droplets containing a mixture of organic dyes. Positively charged TiO 2 NS (zeta potential = +26 mV) were synthesized as described previously (ESI, † Methods). 7 Aqueous turbid dispersions of photocatalytically active coacervate microdroplets were prepared at pH 7-8 by mixing aqueous solutions of ATP or the anionic polymer surfactant, poly(ethylene glycol) 4-nonylphenyl 3-sulfopropyl ether (KPSE, Fig. S1, ESI †) with a poly(diallyldimethylammonium) chloride (PDDA, M w 100-200 kDa, Fig. S1, ESI †) solution containing a suspension of TiO 2 NS. Optical microscopy images showed discrete spherical micro-droplets that were ca. 50 and 5 mm in diameter for the TiO 2 NS/PDDA/ATP and TiO 2 NS/PDDA/KPSE coacervates, respectively, and which in both cases contained high-contrast material that was attributed to the presence of sequestered TiO 2 NS aggregates ( Fig. 1). UV-vis spectroscopic analysis at 325 nm of the centrifuged coacervate phase and aqueous supernatant gave equilibrium partition coefficients (K = [TiO 2 NS] in /[TiO 2 NS] out ) of 6150 and 60 for the PDDA/ATP and PDDA/KPSE systems, respectively, consistent with high levels of uptake of the inorganic nanosheets in the molecularly crowded coacervate phases. We attributed the lower K value associated with the TiO 2 NS/PDDA/KPSE system to the reduced polarity of the polymer surfactant compared with the ATP constituent. To assess the potential of the TiO 2 NS-containing complex coacervates as photocatalytic multiphase micro-droplet reactors, we first determined whether water-soluble organic molecules could be spontaneously sequestered into the membrane-free micro-compartments. A range of dyes, including Methylene Blue (MB, cationic), Rhodamine B (RhB, zwitterionic), Brilliant Red X-3B (X3B, anionic), and Sulforhodamine B (SRhB, net negatively charged) (ESI, † Fig. S2), were added to the coacervate suspensions at pH 7-8 and the K values determined. Significantly, in all but one case (RhB in PDDA/ATP), the dye molecules strongly partitioned into the molecularly crowded coacervate phases, but the uptake selectivity differed considerably between the PDDA/ ATP and PDDA/KPSE systems. In the former, the following order of K values was observed; X-3B (234) 4 SRhB (35) 4 MB (28) 4 RhB (0.55), whilst partitioning in the PDDA/KPSE coacervate followed the sequence; RhB (38) 4 SRhB (24) 4 X-3B (12) 4 MB (6). These differences were primarily attributed to charge or hydrophobic matching between the host and guest components. In this regard, simulations of the pH-dependent log D partition coefficients 10 for each dye molecule over a pH range of 1 to 12 gave constant values between pH 7 and 8 for all the dyes (ESI, † Methods and Fig. S3), and showed decreasing hydrophobic character in the order of RhB 4 MB 4 SRhB Z X-3B. This sequence was inversely correlated to the values of K for the PDDA/ATP coacervates, indicating that polar and charge interactions were the main driving force for sequestration in this system at pH 7-8. Significantly, increased values of K correlated with those dye molecules containing the highest number of negative charges (X-3B and SRhB) (ESI, † Table S2), suggesting that high levels of sequestration were facilitated by partial displacement of ATP anions associated with electrostatic binding of the guest molecules to PDDA. In contrast, higher K values in the PDDA/KPSE coacervates were associated with the more hydrophobic dye molecules. Given that a range of dye molecules could be successfully taken up by the complex coacervates, we exploited these host-guest ensembles as membrane-free multiphase micro-droplet reactors. The photocatalytic activity of TiO 2 NS/PDDA/ATP and TiO 2 NS/ PDDA/KPSE micro-droplets was assessed by exposing suspensions of the coacervates and sequestered dye molecules to UV radiation and comparing the rates of dye molecule degradation to control experiments involving dispersions of TiO 2 NS in water, or coacervate micro-droplets without TiO 2 NS. All samples containing TiO 2 NS showed an exponential decay in dye concentration over periods of between 10 and 30 min ( Fig. 2 and Fig. S4 and S5, ESI †), but the decay constants varied considerably (ESI, † Table S1). For example, the decay constants (l) determined for the photoinduced degradation of MB in TiO 2 NS/water, TiO 2 NS/PDDA/KPSE or TiO 2 NS/PDDA/ATP were 0.5, 0.1 and 0.03 min À1 respectively. Interestingly, photocatalysis in the TiO 2 NS/PDDA/KPSE or TiO 2 NS/PDDA/ATP micro-droplets was reduced compared with analogous reactions undertaken in the presence of TiO 2 NS dispersed in water. We attributed this to a reduction in UV intensity due to increased light scattering in the droplet phases, or an aggregation-induced reduction of the accessible surface area of the TiO 2 NS catalyst within the coacervate matrix, or both. For example, destabilization of the TiO 2 NS/ PDDA/KPSE droplets by addition of aqueous NaCl such that the turbid dispersions became clear solutions was associated with an increase in the l values for all the dyes (Fig. 2, ESI, † Table S1 and Fig. S4 and S5). Similarly, the catalytic efficiency values were higher in the TiO 2 NS/PDDA/KPSE micro-droplets compared with the bulk coacervate over all TiO 2 NS concentrations studied (ESI, † Fig. S6 and Table S3), confirming that photocatalysis was sensitive to optical transparency. In contrast, minimal changes in l were observed when the TiO 2 NS/PDDA/ ATP coacervates were disassembled (ESI, † Table S1), suggesting that interactions between the components of the coacervate droplet and sequestered TiO 2 NS, rather than a loss of incident light intensity due to scattering effects, were responsible for the reduced photocatalytic efficiency in this system. This was consistent with DLS results, which that showed that extensive aggregation of TiO 2 NS occurred in the presence of ATP (ESI, † Table S4). In general, TiO 2 NS/PDDA/KPSE micro-droplets were more photocatalytically active than their TiO 2 NS/PDDA/ATP counterparts, suggesting that the presence of ATP inhibited dye photodegradation possibly by passivation of the charged {00À1} surface of the nanosheets via phosphate-mediated ATP-Ti(IV) interactions, 11 or shielding of the dye molecules via p-p interactions with the adenine group of ATP, 4,12 or both. The latter appeared to be prominent for SRhB, which did not undergo any significant degradation in the TiO 2 NS/PDDA/ATP micro-droplets, or when these droplets were disassembled in the presence of NaCl (Fig. 2b), or in control experiments involving irradiated samples of TiO 2 NS dispersed in an aqueous solution of ATP (ESI, † Fig. S7). The ability to selectively sequester and degrade different dyes using the TiO 2 NS/PDDA/ATP micro-droplets was exploited by performing simultaneous dye degradation experiments on MB and RhB with K values of 28 and 0.55 respectively. The dyes were co-sequestered into the coacervate droplets and exposed to UV light. Simultaneous photo-degradation was monitored by changes in the visible excitation absorption bands at 668 nm (MB) and 555 nm (RhB), and the exponential decay constants compared with values obtained for mixtures of the dyes in aqueous TiO 2 NS dispersions with or without PDDA present. Both of the aqueous controls exhibited time-dependent reductions in the 668 and 555 nm absorption bands, signifying simultaneous degradation of MB and RhB (ESI, † Fig. S8 and S9) with associated l MB /l RhB ratios of 2.3 AE 0.2 (without PPDA) and 0.6 AE 0.1 (with PDDA), which were in reasonable agreement with the same ratio (1.5 AE 0.1) obtained from the individual aqueous dye degradation experiments (ESI, † Table S1). In contrast, analogous experiments performed over two hours using the TiO 2 NS/PDDA/ATP microdroplet dispersion showed only a small reduction in intensity of the 555 nm absorption band along with almost complete loss of the 668 nm absorption feature to give a l MB /l RhB ratio of 18 AE 4 (Fig. 3). Thus, it was possible to achieve selective degradation of MB in the presence of RhB by spatial confinement of the dye molecules specifically within the TiO 2 NS/PDDA/ATP micro-droplets. In conclusion, this study has shown that photocatalytically active titania nanosheets can be sequestered at high concentrations into coacervate droplets, and used as multiphase micro-droplet reactors. The nanoparticle-containing droplets spontaneously sequester a range of organic dyes with various equilibrium partition constants that are dependent on the chemical functionalities of the guest molecules. We demonstrate photocatalytic dye degradation specifically within the micro-droplets, and utilise differences in partitioning to perform selective photocatalytic degradation in a binary dye system. Our results suggest that the rational integration of catalytic nanoscale objects within membrane-free compartments capable of selective small molecule uptake and storage can be used as a modular approach to the development of hybrid biphasic nano/micromaterials with tuneable properties and reactivity that could have diverse applications in detoxification and bioremediation processes.	v3-fos-license
2021-05-10T00:03:18.815Z	2021-02-02T00:00:00.000	234020812	{ "extfieldsofstudy": [ "Business" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.e3s-conferences.org/articles/e3sconf/pdf/2021/10/e3sconf_icies2020_00049.pdf", "pdf_hash": "8a5a47fed5dc1dd3faa6be29b6905758d7d60d09", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45392", "s2fieldsofstudy": [ "Environmental Science", "Economics" ], "sha1": "bedc60a0ff3232321fe79c62b5a4c19d5fc08ab8", "year": 2021 }	pes2o/s2orc	Optimization of costs for innovations of industrial enterprises Western Ukraine in ensuring sustainable environmental development . The structural transformations taking place in the world economy are due to the transition from the «industrial economy» to the «knowledge economy», which is characterized by the dominance of intellectual capital and innovation. It is determined that an important condition for sustainable economic development of the country and increasing competitiveness is the effective implementation of innovation. The purpose of the study is to determine the optimal cost structure for innovation and its impact on sales revenue. Within this framework, attention is focused on the use of elements of economic and mathematical modeling. The scientific novelty is to develop a model that will substantiate the relationship between the studied indicators of innovation costs, will allow to predict the amount of income from sales and ensure the achievement of its optimal value. It is established that the proposed multiple linear regression quite clearly describes the initial data and can be used for prediction. The practical significance of the obtained results lies in the applied orientation of the approaches, the use of which improves the management of innovation costs and increases the level of reliability of economic information of industrial enterprises of Ukraine and as a result will improve the environmental situation. The study will assess the impact of economic entities on the environment and preserve the natural resource potential in order to achieve sustainable economic development. Introduction In the context of globalization, changes in the trajectory of economic progress lead to the acceleration of innovative development, the transition to a strategy based on knowledge and intelligence. For the estimation of innovative development of country for today important are the international rating such as, Global Innovation Index, Bloomberg Innovation Index, Global Competitiveness Index, Innovation Union Scoreboard. According to the Global Innovation Index [1], in 2019 Ukraine took 47th place in the ranking among 129 countries, in 2018 -43rd place among 126 countries, while in 2017 -50th place among 127 countries, so during the last three years there is no steady progress.However, compared to previous periods, Ukraine has established itself in the top 50 innovative countries in the world, surpassing such post-Soviet countries as Moldova, Georgia, Kazakhstan, Azerbaijan, Belarus, Kyrgyzstan, Tajikistan.Despite this, Ukraine still lags behind innovation-oriented countries in the ranking.Ukraine traditionally demonstrates the highest indicators of innovation in the field of education, which accordingly affects the country's overall innovation index.According to the indicator «Total expenditures on R&D, % of GDP» in 2019, Ukraine ranked 67th in the ranking (0,4% of GDP).The value of this indicator remains low compared to other countries, especially innovation-oriented, which negatively affects the innovative development of the country. In the structure of the index of global competitiveness [2] an important place is occupied by the sub-index «Innovation», which reflects the innovation component of the country.According to this sub-index, in 2019 Ukraine ranked 60th out of 141 countries and lagged behind some post-Soviet countries, including Estonia (34th place), Lithuania (42nd place) and Latvia (54th place). In general, Ukraine's position in most rankings that analyze the level of innovative development, its potential and technological capabilities are not satisfactory for this stage of development.That is why the tasks of approaches to the analysis of innovation activity become especially important. ICIES'2020 «innovation», «innovation process» J.A. Schumpeter combined with the theory of long-term cyclic oscillations -the theory of «long waves» M.D. Kondratiev (1892Kondratiev ( -1931)).In the 80's of the last century, based on the theory of J.A. Schumpeter, the famous German scientist-economist G.O. Mensh [4] in the book «Technological stalemate: innovation overcomes depression» concludes about the high concentration of basic innovations that overcome the «technological stalemate» and mark the beginning of a new trend in the economy -improving its key performance.G.O. Mensh explains the unevenness of innovation activity by the peculiarities of the functioning of a market economy. At the present stage of economic development, many domestic scientists have dealt with the problems of analyzing the costs of innovation.Thus, Lisovska L.S., Karyy O.I. [5] investigated the factors for joint innovation between enterprises and proposed to group them according to certain characteristics.According to the authors, the use of these features will contribute to a better understanding of the principles and mechanisms of innovative cooperation.Tsenov M.O.[6] proposed a model of economic growth based on the innovative development of the enterprise, which consists in calculating the main indicators of the model and conducting a comparative analysis of indicators with real indicators of economic activity.The team of authors S. Bondarenko, L. Verbivska, N. Dobrianska, G. Iefimova, V. Pavlova, O. Mamrotska [7] studied the issues of managing the costs of innovation to ensure economic security.Within the framework of this methodology the principle of optimal management is presented.In the context of economic security I. Maslak, N. Ye.Grishko, O.O.Hlazunova, K.O.Vorobiova [8] also worked out approaches to cost management of innovative activities of mining enterprises.Fedulova I.V. [9] considered certain types of costs for innovation and their impact on the volume of sold innovative products and considered conceptual approaches to establishing their optimal structure based on performance.Olefirenko O.M. [10] presented an economic and mathematical model for optimizing sales costs of innovative and active machinebuilding enterprise.However, without diminishing the significant contribution of researchers, some problematic issues of the methodology of analysis and optimization of innovation costs remain unresolved. Model and Data Analysis One of the most important aspects of enterprise development is the optimization of costs or achieving a level that will provide the required financial result under certain conditions.To do this, use the elements of economic and mathematical modeling. The mathematical model of optimizing the cost of innovation assumes that all variables, parameters, constraints and the objective function of the model are quantifiable.If variables X = ( where, i  -is a system of problem constraints that describes the necessary conditions for reaching the extremum. Dependencies (1), ( 2) are an optimization economicmathematical model, in the construction of which it should be remembered that it must adequately describe the real economic processes occurring in the system, and take into account only significant phenomena in the process.The system of constraints or task (2) describes the necessary conditions for reaching the extremum.Schematically, the stages of the process of building a model for optimizing the cost of innovation are shown in Fig. 1.The study will be conducted using the method of multiple linear regression to build a model of the relationship between the volume (income) from sales and cost elements for innovation of domestic industrial enterprises. Optimization of innovation costs for the company is to achieve such volumes with given constraints, under which the income from sales will be maximum. When forming the system of constraints of the model, it was important that the growth rate of sales revenue should exceed the growth rate of innovation costs.Thus, the system of conditions of the problem has the form: Q -income (amount) of sales in the current and base periods, respectively, UAH As a factor of the cost of innovation, we propose to use the types of costs in the accounts [11].In this case, the model of optimization of innovation costs by the criterion of income maximization will look like: Q(x)=a 0 +a 1 x 1 + a 2 x 2 +a 3 x 3 +a 4 x 4 +a 5 x 5 max  where, x 1 -costs of innovations in the acquisition (creation) of new or improvement (modernization, reconstruction) of existing technologies, equipment, machinery, etc., thousand UAH; x 2 -costs of innovations in the acquisition (creation) and implementation of copyright, intellectual property (inventions, utility models, industrial designs, etc.), thousand UAH.; x 3 -costs of innovation as part of the costs of preparation and development of new products and production, thousand UAH.; x 4 -costs of innovation as part of overhead costs, thousand UAH.; x 5 -costs of innovation as part of other operating costs, thousand UAH.; Q(x) -income from sales of products, thousand UAH. Results and Discussion Using the initial data (Table 1) obtained from statistical collections; we determine the parameters of the model. We make sure that the system of constraints of the model is calculated by calculating the growth rate (Table 2 [12 -15]. ICIES'2020 As can be seen from table. 3 the growth rate of sales revenue exceeds the growth rate of innovation costs for all parameter values x ( ), that is, it can be argued that the requirements of the system of restrictions are met. To establish and determine the factors of influence, we will build a correlation matrix based on data for 2018. To build an econometric model, we determine the main coefficients of multifactor regression (Table 3). The equation of the dependence of income from sales of products from these factors x 1x 5 will look like: F(x)=39,2338-0,0007x 1 +0,0115x 2 -0,0017x 3 -0,0011x 4 +0,0022x 5 Let's construct the schedule of initial and result data, proceeding from the found equation of dependence (fig.2).As can be seen from Fig. 3 initial values of income from sales of products very well describe the results.Thus, we can conclude that the constructed multiple linear regression quite clearly describes the original data and it can be used to predict F(x).The coefficient of determination is 0,995 (see table. Conclusion One of the ways to reach a compromise between economic development and the ecological state of the environment is the implementation of the concept of sustainable development, which provides for further economic development taking into account the need to preserve the environment.The scale and nature of nature management have led to the fact that at the present stage a critical element in the system of economic relations are innovation costs, and opportunities for sustainable Thus, the use of economic-mathematical model in planning the costs of innovation allows you to describe the relationships between the initial variables of the model, assess the impact of costs on performance, find the best solution for the value of these costs, and contributes to effective management. In the course of the research the regularities of the influence of the level of total expenses on innovation activity on the indicators of financial results of industrial enterprises activity were revealed and the interdependence of these indicators was explained with the help of the formed formulas.The obtained results allow to evaluate the indicators of innovative development of industrial enterprises of Western Ukraine.Further application of the model allows to make the following analysis of innovation processes and forecasts of enterprise development depending on the direction of investments in innovation. , 2 y that can not be controlled, and the condition of the system under study is determined by m constraints, the mathematical model can be written: find a point Y = ( 1 y , …, n y ), in which the extremum (minimum or maximum) of the objective function is reached f (X,C): Fig. 1 . Fig. 1.Stages of building a model for optimizing the cost of innovation Source: Developed by the author Selection and analysis of the result indicator Y Determination of factors influencing the result indicator Y Construction of a statistical model of the dependence of the resultant indicator Y on factors X Analysis of the reliability of the constructed model Construction of a model for optimizing the cost of innovation by the criterion of maximizing revenue from sales Consideration of the system of restrictions ICIES'2020 of expenses for innovations of the enterprise in the previous period; p Q , 0 p 4), which indicates a close relationship between the resultant and factor traits (variation of the resultant trait by 99,5% due to certain factors and only 0.05% -other factors that are not taken into account To determine the adequacy of the model, we find the value of the Fcriterion R F = 44,8372.Tabular value tab F (0,05;7;5)=4,3026.Because R F > tab F (44,8372>4,3026), then the coefficient of determination is statistically significant. development largely depend on the efficiency of its use and reproduction.That is why the study of environmental and economic issues in Ukraine is extremely relevant. Table 1 . ) Initial data for calculations of optimization of expenses for innovations on criterion of maximization of the income of the industrial enterprises Western Ukraine in 2018 Table 3 . The results of calculating the coefficients of multifactor regression Schedule of initial and result data of income from sales of products	v3-fos-license
2021-04-05T07:07:57.090Z	2021-04-11T00:00:00.000	233017172	{ "extfieldsofstudy": [ "Computer Science" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://turcomat.org/index.php/turkbilmat/article/download/1609/1362", "pdf_hash": "b1c35dd3d59994e186619860ac261fbce858e755", "pdf_src": "MergedPDFExtraction", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45395", "s2fieldsofstudy": [ "Computer Science" ], "sha1": "82a0900141c6f86bb685d30e1da85ac6c9b25b23", "year": 2021 }	pes2o/s2orc	Performance Comparison for Spam Detection in Social Media Using Deep Learning Algorithms Social media applications like Twitter, Instagram, Facebook have helped people to connect to each other. This has been eased due to high-speed internet. However, this has invited various spam messages through tweets or Facebook. The sole purpose of such messages is aggregation or exploitation of personal data in terms of finances or medical records, political benefit’s or community violence. This makes spam detection an extreme value-added service. We tend to recommend a 1D CNN algorithmic technique and compare results with variants of CNN and with boosting algorithms. The model is braced with linguistics data in the illustration of the words with the assistance of knowledge-bases such as Word2vec and fast ext. This improves the end to end performance, by providing higher linguistics vector illustration of input testing words. Projected Experimental results show the efficiency of the projected approach from the point of view of accuracy, F1-score and response time. Introduction Recently with the dawn of the social media platforms, people are enabled to grow and to communicate efficiently. This opportunity has been threatened through spams, malicious links, and malware [1]. This makes Spam detection extremely required value-added service for any social platform. Initially the spam messages used to get detected by manual process or by simple filter rules for commons properties. Automation in spam detection advance of basic machine-learning algorithms which do not produce spam detection models. Initially, spam started spreading with email spams. Additionally, the SMS is a price effective technique used for converting individual messages to the vector form. Possible purchasers, encompass a higher rate of response as compared to email spam. In conjunction with emails and SMS, social networking platforms like Twitter, Facebook and an instant traveler like WhatsApp, etc., also are tributary to a greater portion of spam on the network. It will be a complex activity of spam detection without any filter at the receiving node. One of the initial classifiers is rule-based, having a lot of formally written principles. These classifiers were used to get deployed to an ample space of purchasers. It comprises a set of pre-defined rules that are applied to associate degree of incoming messages and these messages were labelled as spam if their check score exceeds the threshold value. Even after the spam is detected, the success of these ways is restricted and needs to be combined with different machine learning methods so as to give fairly sensible results. Naive Bayes, Radom forests etc., are few of the standard classifiers. These classifiers are complicated, thanks to feature extracting options from the text, which helps to identify the pattern of spam and ham messages [2]. Most frequently used feature extracting models are bag of words models with token frequency as a common factor. Convolutional Neural network is the deep learning algorithm that addresses the accurate classification of the text messages as spam or ham. Literature Survey Gauri Jain et al. [1] had proposed an architecture focused on short spam content on SMP like Twitter. This is in contrast with earlier long spams emails detections and deletion. Using basic configurations. of CNN algorithms, the outcomes showed that the proposed model proved to be efficient with the use of Twitter and SMS text datasets. Thayakorn Dangkesee et al. [3] has proposed a model that was used for spam detection by the victimization of spam word lists using a billboard URL-based security tool. Naive Bayes algorithm has been used to analyses the data using data types such as all data and specific data. It has boosted the performance of the spam detector than usual. One can show their methods fulfills the experimental result. Rutuja Katpatal [4] has formed an additional input training dataset to classify unlabeled tweets using another dataset. Author has proposed a scheme that adjusts training data sets. Dropping too old samples after a specific time has helped to eliminate unusual information saving space. Ms. Sayali Kamble et al. [5] has exhibited the plan of ongoing vector space denotation of words. Evaluation of a novel AI-based way to deal with specialization Social spam detection. Their general research goal for consequently shifting and recognizing spammers who point social destinations was to discover methodology ie.SAND, to find compelling devices. Guanjun Lin et al. [6] has detected the nine mainstream algorithms that were compared to understand the most suitable algorithm. The stability of each algorithm had been studied thoroughly. It had indicated the variation of the training time according to CPU core. Tingmin Wu, et al. [7] has proposed a system of twitter spam detection. While the paper had addressed the then-existing challenges like low speed and feature extraction difficulties thoroughly; the paper written had experimentally proven the comparisons between achieved results and existing accuracies through the indication of graphs . Proposed Methodology The System consists of a model that is trained on Convolutional Neural Network rule. On general terms, CNN is most well-liked for image classification. The variation of CNN 1D has been used for spam text information classification. CNN contains various types of layers that are typically improved in terms of accelerating accuracy during and after the implementation. Here, in the projected system, CNN will work as a classifier to investigate whether or not, the text statement is spam. The model permits to form associate unattended learning as well as supervised mode of learning rule for getting vector representations for input texts. Models like fast texts makes us of neural network for word embeddings The proposed framework comprises of Word2Vec and fast text advancements along with Convolutional Neural Network (CNN) to complete the model. The proposed framework will also be comprised of varieties of CNN models in terms of the number of filters and convolutional layers of the CNN algorithm. These can be used for performance comparisons between different variations of CNN. The research will work around CNN varieties and layers. Architecture 1. The Proposed system has the advantage of multiple platforms for computer files for model development. 2. The System has been developed with quite one algorithmic program, so Prediction guarantees are inflated. 3. Live updates area unit involved in prediction so its area unit typically used for live recommendation. 4. The proposed system varies in the filter and convolutional layer combination. 5. The proposed system will compare the results of the CNN algorithms with that of the boosting algorithms and hence will try to achieve the maximum accuracy with better precision. 6. We are also trying to train the module through the vernacular language data sets like Hindi or Marathi. 7. Fig.3.1 shows an overview of the proposed system architecture. 8. Out of the complete training data set, 80% of the data will be used for the training the model under supervised learning technique, and the remaining 20% of the data set will be used as a test set to generate the accuracy and response time calculation tests. Algorithms Fast text and Word2Vec algorithms are used for implementation. Scope of fast text: The scope of fast text is to work as a runtime library for classification and vector conversion of input texts. It is used for processing the number of tasks. It is based on the C++ platform. Fast text allows the end-user or the developer to coach supervised and unsupervised classification of input texts. Generally, it trains the models of a Skip-gram model or continuous bag of words (CBOW). It makes use of function such as Soft Max or negative sampling loss functions. Huge number of linguistic embeddings can be trained with the use of fast text. Scope of Word2Vec: Word2Vec may be a shallow, two-layered neural network which is trained to reconstruct linguistic contexts of words. Every unique word gets assigned a vector notation in vector space. This conversion is taken place across the given classes. Words which are share the common class context are placed in close proximity in the vector space. Word2vec thus has proved to be efficient models in the word embeddings techniques [8]. Like fast text model, the word2vec models are majorly applicable to CBOW or Skip gram models [9]. 1D CNN algorithm The process of feature learning has been used in 1D CNN algorithm. The algorithm maps the extracted features of the input text sequence.1D algorithm is especially useful in datasets which have segments of fixed length. 1D CNN Algorithm The Algorithm of a 1D-CNN is formed through the following important steps: where, is bias ℎ neuron bias at layer , −1 is ℎ neuron output of layer l-1 , −1 is the ℎ kernel of layer −1 . 1 (.,.) is used to perform 'in-valid' 1D convolution without zero-padding. Figure 2. indicates the overview of the CNN based structure model. This window contains the main modules of the final model, which are summary tab, prediction using word2vec tab, XG boost tab, ADA boost tab, gradient boost tab which in turn at the backend runs the CNN algorithms 2. Figure 3. represents the window where, the input words are converted to vector forms and then are fed to the CNN algorithm for the classification using filters and convolutional layers. An authenticated user needs to give input in the form of text and then the model will give output. In this case, the model will process the input text into vector form and then will feed as the input to CNN algorithm and based on the labeled learning, CNN will classify the sentence into spam or ham 3. Figure 4. shows the results of various CNN variations (filter size and iterations or epoch) in terms of Training accuracy, validation accuracy, precision, and F1 score. Figure 7 represents the performance comparison chart in terms of % accuracy, F1 score and precision. Conclusion After research implementation of the proposed project and idea, the system reflects the conclusion that the CNN-LSTM variant model (III) along with Word2Vec performs efficiently in terms of % accuracy, F1 score, and precision [ Figure 7]. The % accuracy has been up to 96% on average. As indicated above, we have conducted several rounds of training and validation on the CNN algorithm and on machine learning algorithms such as the ADA boost and the XG boost. In comparison, it has been concluded that the Word2Vec model takes around 8 to 12 seconds to load and respond, as shown in the graphical representation in the results and discussion section. However, with CNN-LSTM fast text model takes much more time to load and respond. Fast text takes around 1 to 2 minutes. Both are relying on the earlier stage of model training. 1D Convolutional Neural Network algorithm is used for the training model. The number of iterations of the execution has been based on the epoch. As a part of future work, of the proposed model, multilingual twitter dataset like Hindi, Marathi can be used to arrive at the efficient performance delivery. After deriving accuracy, later on, these individual results will be performance compared with the results of the ADA/XG boosting algorithm. Changed spams can be added in the input datasets to reduce the spam drifts. [7].	v3-fos-license
2016-05-04T20:20:58.661Z	2014-11-06T00:00:00.000	17875160	{ "extfieldsofstudy": [ "Biology", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1016/j.dib.2014.10.003", "pdf_hash": "0fd59007a2b9097aad5d5fe62cb15b6770dd8746", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45397", "s2fieldsofstudy": [ "Biology", "Medicine" ], "sha1": "0fd59007a2b9097aad5d5fe62cb15b6770dd8746", "year": 2014 }	pes2o/s2orc	Data in support of peptidomic analysis of spermatozoa during epididymal maturation The final differentiation of the male germ cell occurs in the epididymal duct where the spermatozoa develop the ability to be motile and fertilize an ovum. Understanding of these biological processes is the key to understanding and controlling male fertility. Comparative studies between several epididymal maturation states could be an informative approach to finding sperm modifications related to maturation and fertility. Here we show the data from differential peptidomic/proteomic analyses on spermatozoa isolated from 4 epididymal regions (immature to mature stage) using a profiling approach based on MALDI-TOF mass spectrometry and, combined to top-down MS in order to identify peptidoforms and proteoforms. By this way, 172m/z peaks ranging between 2 and 20 kDa were found to be modified during maturation of sperm. A total of 62m/z were identified corresponding to 32 different molecular species. The interpretation of these data can be found in the research article published by Labas and colleagues in the Journal of Proteomics in 2014 [1]. modified during maturation of sperm. A total of 62m/z were identified corresponding to 32 different molecular species. The interpretation of these data can be found in the research article published by Labas Value of the data First description of peptidome/degradome of epididymal spermatozoa from boar (Sus scrofa). Molecular phenotypes distinguishing degrees of maturation of spermatozoa during the epididymal transit. First description of protease activities involved in maturation of epididymal spermatozoa. Data Mean peak values obtained from intact cells (IC), detergent-soluble extracts (SD) and detergentinsoluble extracts (ID) associated with immature (IC2, SD2, ID2) and mature epididymal spermatozoa (IC9, SD9, ID9) are shown in Supplementary Table 1. Only 172m/z peak values for which the foldchanges were 42 and which presented at least one significant difference between epididymal samples (p o0.05) are reported in this table. Some of them are identified. The variation index was calculated as the product of the absolute difference between the maximum (max) and minimum (min) peak intensity values multiplied by the fold-change (fold) between the two extreme values. The variations are expressed as linear decrease (LD) or linear increase (LI) when all the mean values for the four epididymal samples were significantly different from each other (p o0.05), expressed as increase (I) or decrease (D) when at least one of the mean values was not significantly different from the others, and intermediate (inter.) when at least one value was different. In the first column, m/z values in green correspond to m/z peaks observed only in IC analysis (a total of 135m/z). Rows with color in IC, SD and ID columns correspond to specific peaks for the sample preparation. Table 2. This table shows the raw file name associated to the NCBInr accession number, the gene name, the protein description, the location, the characterized post-translational modifications, the number of b ions, y ions and total ions, the delta mass (Da and ppm) between the ProSight theoretical mass M and the mass M observed by nano-ESI-FTMS, the mass M (Da) observed by nano-ESI-FTMS, the ProSight theoretical mass M (Da), the ProSight PDE score, the E-value identification probability, the p score, the precursor m/z and mass type with corresponding charge state (z), the fragmentation mode, the database used for identification, the signal/noise, the ProSight search type mode, the precursor mass type, the precursor mass tolerance (Da or ppm), the fragment mass type, the fragment mass tolerance (ppm), the delta M mode and disulfide activation/deactivation, the minimum of matching fragment between observed and theoretical fragmentation mass spectra, the include modified forms activation/ deactivation, the sequence, the calculated theoretical average and monoisotopic mass [M þH] þ (Da) and the mass [MþH] þ (Da) observed previously by MALDI-MS. Materials and methods: organ sampling and sperm preparation The epididymides were collected from four one-year-old adult boars. The luminal contents of the tubules of four epididymal regions (E2, E4, E6 and E9) were collected by micro perfusion [2]. Spermatozoa were isolated by centrifugation. The pellets were washed once with PBS (140 mM NaCl, 15 mM KCl, 7 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , pH 7.4), centrifuged and resuspended in PBS. To remove protein contamination, the spermatozoa were centrifuged on 40% Percoll in PBS. The pellets were washed again with PBS and resuspended in 20 mM Tris-HCl, pH 6.8, and 260 mM sucrose (Trissucrose buffer (TS)) ( Fig. 1 Whole and fractionated cell preparations For whole cell analysis, about 2 Â 10 6 spermatozoa were spotted for each epididymal region onto the MALDI sample probe and immediately mixed with a sinapinic acid matrix solution. MALDI-TOF profiling of whole sperm cells and sub-cellular fractions All samples were analyzed by a MALDI-TOF mass spectrometer (Waters Corporation, Micromass Ltd., Manchester, UK) operating in positive linear mode as previously described [1]. Spectral profiles were collected in the 2000-20,000m/z mass range. Data processing was performed using MassLynx ™ 4.0 software. To increase mass accuracy, internal calibration was performed. Thus, the major unknown 6797m/z constant was used as "lock mass" for all spectra. Intact spermatozoa and corresponding detergent-soluble and -insoluble extracts obtained from 4 epididymal regions (E2, E4, E6 and E9) of both epididymes from 4 animals were analyzed by MALDI-TOF MS, with 8 replicates for each of the three sample preparations. A total of 768 spectra were generated in this study. The spectra were analyzed by Progenesis MALDI ™ 1.2 software (NonLinear Dynamics) as previously described [1]. After alignment of all spectra, a total of 253m/z peaks were detected (Supplementary Table 1). A total of 135m/z molecular species were characterized by Intact Cell MALDI-MS and 118m/z were newly observed by MALDI MS from SD and ID fractions. Quantitative analysis linked to the sperm maturation process In order to characterize peak differences between epididymal spermatozoa, the intensity of each normalized peak was subjected to one way analysis of variance with Progenesis (factor epididymal regions) and to three-way analysis of variance with R software (factors being animals, epididymis, epididymal regions, the replicate interactions being the residuals), as previously described [1]. All peak signals with a differential fold-change greater than two average values of normalized intensity and a p value o 0.05 were selected (Supplementary Table 1). Thus 89m/z peaks were retained for intact cells (IC), 112m/z peaks for detergent-soluble extracts (SD) and 59m/ z peaks for detergent-insoluble extracts (ID) for a total of 172 unique m/z peaks (Supplementary Table 1). Top-down mass spectrometry Identification of peptidoforms and proteoforms (endogenous species) was achieved by acquiring nano-ESI tandem high resolution mass spectrometry (MS and MS/MS). Molecular species were previously extracted with 1% and 5% formic acid from IC (region 2 or 9) or from ID fraction (region 9), desalted and concentrated using ZipTip C4 (Millipore Corporation, Billerica, MA). Eluted peptides and proteins were directly analyzed using a LTQ Orbitrap Velos mass spectrometer (Thermo Fisher Scientific, Germany) operating in positive mode, as previously described (1). Data were acquired using Xcalibur software v2.1 (Thermo Fisher Scientific, San Jose, CA). All analyses were performed manually using a high-high strategy, meaning that a MS spectrum in the 400-2000m/z mass range was followed by a MS/MS spectrum obtained by High energy Collisional Dissociation (HCD). Thus 412m/z corresponding to 217 non-redundant molecular ions were selected to induce HCD fragmentation. Identification and structural characterization were performed using ProSight PC software 2.0 (Thermo Scientific, San Jose). Raw data files were processed by THRASH (signal/noise: 2-3), and data were compared to a simple annotated "Sus scrofa" house database generated from NCBInr using Proteome Discoverer (Thermo Fisher Scientific). Automated searches were performed using the "Absolute Mass and Biomarker" search options. The mass tolerances were set at 5 ppm for the monoisotopic precursors, 5 Da for the average precursor and 15 ppm for fragment ions mass tolerance. Disulfide modifications and N-terminal post-translation modifications (acetylation and initial methionine cleavage) were activated. Post-translational modifications such as phosphorylation and disulfide bridges were confirmed using the manual Single Protein mode. Proposed sequences with E-value o1 Â 10 À 2 were considered positively identified with a minimum of 10 matching fragment ions. The data were deposited with the ProteomeXchange Consortium (http://proteome central.proteomexchange.org) via the PRIDE partner repository [3,4] with the dataset identifier PXD001303. A total of 62m/z were identified and attributed to 32 different molecular species corresponding to three intact and whole proteins and 58 peptides from 29 proteins (Supplementary Table 2). Forty-five of these peptides presented tryptic or semi-tryptic cleavages, suggesting protease activities by trypsin-like or kallicrein enzymes (Supplementary Table 2). Gene symbols were mapped for peptidoforms and proteoforms identified by top-down MS, and analyzed using the online PANTHER classification system (database version 9.0; http://www. pantherdb.org/) [5] (Fig. 2). Go terms from the Biological Process and the Molecular Function domains were considered. The background dataset for the analysis was the Homo sapiens and Sus scrofa genomes. Conflict of interest The authors declare that they have no conflicts of interest. Fig. 2. Classification of identified molecular species based on biological process and molecular function using PANTHER classification system.	v3-fos-license
2018-12-16T14:02:42.144Z	2018-12-14T00:00:00.000	54591161	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0209015&type=printable", "pdf_hash": "ae61ad042984b478add8d54060f0e8ebf4161972", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45399", "s2fieldsofstudy": [ "Medicine", "Engineering" ], "sha1": "b0f2c56a6068d07756d71b4f72ee8e0ddc616d79", "year": 2018 }	pes2o/s2orc	Identification of knee gait waveform pattern alterations in individuals with patellofemoral pain using fast Fourier transform Patellofemoral pain (PFP) is one of the most common overuse injuries of the knee. Previous research has found that individuals with PFP exhibit differences in peak hip kinematics; however, differences in peak knee kinematics, where the pain originates, are difficult to elucidate. To better understand the mechanism behind PFP, we sought to characterize differences in knee gait kinematic waveform patterns in individuals with PFP compared to healthy individuals using fast Fourier transform (FFT). Sixteen control and sixteen individuals with PFP participated in a fast walk protocol. FFT was used to decompose the sagittal, frontal and transverse plane knee gait waveforms into sinusoidal signals. A two-way ANOVA and Bonferroni post hoc analysis compared group, limb and interaction effects on sagittal, frontal and transverse amplitude, frequency and phase components between control and PFP individuals gait waveforms. Differences in frequency and phase values were found in the sagittal and frontal plane knee waveforms between the control and PFP groups. The signal-to-noise ratio also reported significant differences between the PFP and control limbs in the sagittal (p<0.01) and frontal planes (p = 0.04). The findings indicate that differences in gait patterns in the individuals with PFP were not the result of amplitude differences, but differences attributed to temporal changes in gait patterns detected by the frequency and phase metrics. These changes suggest that individuals with PFP adopted a more deliberate, stiffer gait and exhibit altered joint coordination. And the FFT technique could serve as a fast, quantifiable tool for clinicians to detect PFP. Introduction Patellofemoral pain (PFP) is one of the most common overuse running injuries in active populations; such as runners and the military, and is known to affect women at a disproportionately higher rate than men [1][2][3]. PFP often becomes chronic with up to 50-90% of individuals reporting continued symptoms at [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19] year follow ups [4][5][6]. Alterations in joint mechanics in subjects with PFP have most commonly been observed at the hip based on analysis of peak kinematics during tasks such as walking and running; despite PFP pain originating at the knee [7,8]. The difficulty in detecting differences at the knee suggests that alternate techniques could be explored to capture changes in knee gait mechanics in individuals with PFP. Previous research found that individuals suffering from PFP are characterized by delayed quadriceps muscle activation but not deficits in muscle strength [9][10][11][12]. This altered muscle activation could explain why differences in peak knee angle are not always present but differences in the timing of peak knee angles may be. Fast Fourier Transform (FFT) is a technique that can assess both magnitude and temporal characteristics in waveforms and has successfully identified critical temporal metrics from gait data in individuals with neurological disorders [13,14]. Here FFT was used to identify both magnitude and temporal differences in knee kinematics in control individuals and those suffering from PFP during fast walking. FFT is an established method that has not been extensively explored in individuals with PFP [13][14][15][16][17][18]. Tinley et al. (2002) [18] used Fourier analysis to develop a gait index to classify abnormal gait in children from sagittal plane hip, knee and ankle kinematic waveforms. Schneider et al. (1983) [19] used it to delineate differences in individual's knee joint disease from their ground reaction force (GRF) gait patterns. Both studies focused on detecting magnitude differences in gait patterns. Winfree et al. (2015) [20], however, were able to extract temporal differences in stride patterns in individuals with neurological impairments. The ability to assess gait quality and detect joint degeneration from gait patterns highlights the sensitivity of the FFT method and suggests that FFT could identify differences in knee kinematics that may be present in individuals with PFP. Gait variability is often used to identify the presence of a pathological condition [21][22][23][24]. Cunningham et al. (2014) [21] found that individuals with PFP exhibited greater variability at specific intervals during the gait cycle, which they suggested represented a deliberate act to alter loading on the knee. Conversely, similar studies have reported reduced variability in individuals with PFP when assessing coupling angle variability [22,23]. This study used signal-tonoise ratio to quantify knee kinematic variability during fast walking. Given that each participant performed the same walking task, they each will exhibit similar knee kinematic "signals." Therefore, differences in amplitude, frequency and phase obtained from the FFT analysis will reflect the signal "noise." While SNR (signal-to-noise ratio) is typically used for signal and image filtering, we will explore its ability to denote gait variability in individuals with PFP [25,26]. The objective of this paper is to implement FFT to detect changes in sagittal, frontal, and transverse knee gait patterns during fast walking from the amplitude, frequency and phase components in controls and individuals with PFP. Additionally, SNR will be used to assess gait pattern variability between the two populations. We hypothesize that individuals suffering from PFP will not show any differences in gait amplitude but will exhibit larger frequency and phase values which would reflect delayed peak knee angles during stance. We also hypothesize that the PFP individuals will exhibit greater knee gait kinematic variability based on SNR than their control counterparts. Differences observed in these metrics could serve as gait biomarkers for the early identification of PFP. Instrumented gait analysis The University of Kentucky Institutional Review Board granted permission to conduct this research. The participants ranged from 18 to 35 years old and provided written informed consent required by the institutional review board. Sixteen control (mean (SD) age: 21.1 (3.9) yrs; height: 1.7 (0.1) m; mass: 64.5 (11.8) kg) and sixteen PFP (age: 22.4 (4.0) yrs; height: 1.7 (0.1) m; mass: 71.4 (19.5) kg) participants performed a fast walking protocol. A power analysis indicated that at least of 15 participants per group were needed to detected a moderate effect (0.75) to achieve adequate statistical power (α = 0.05; 1-β = 0.80). PFP was verified in the participants by a licensed physical therapist. The physical therapist confirmed the individual had PFP if they reported a minimum pain of 3 out of 10 in their patellofemoral joint on the numerical pain rating scale when running and they had to have pain during either the patellar compression or retropatellar/peripatellar palpation test. Physical examination eliminated ligamentous, tendon and internal derangement. Participants were excluded if they had any previous patellar dislocation and/or instability. All control participants could not have had a previous knee surgery and were injury free for the past six months. The control group was age, weight and height matched to the PFP group. An established fifty-six reflective marker design was used to place anatomical and tracking markers on the participants [8,24]. Anatomical markers were positioned on the C7 vertebrae, acromion processes, sternum located on the trunk, the iliac spines, iliac crests, greater trochanters, femoral epicondyles, tibial plateaus, malleoli, and the first and fifth metatarsal heads on the lower extremities. Four sets of four rigid cluster markers were placed on the thighs and shanks and they served as the tracking markers during the walking trials. The clusters are rigid shells with four markers on them that were tightly wrapped on the distal end of the shank and thigh of the individual. This approach of including clusters was used to minimize the effect of soft tissue artifact and Manal et al. (2003) [27] previously showed this had good agreement with the results using bone pins. For this study, each participant walked on a Bertec instrumented treadmill (Bertec Corp, Columbus, Ohio; New Balance, Brighton, MA) with embedded force plates while wearing WR662 running sneakers. Prior to the data collection, each participant also performed a five-minute walking warm-up to become acclimated to the treadmill. To determine the fast walk speed, participants walked on the treadmill as the speed was repeatedly increased by 0.1m/s. When the speed increased to the point the participants could no longer walk but had to run, the speed was decreased by 0.1m/s and set as the fast walk speed. Participants performed a fast walking protocol because at slower speeds individuals may be able to compensate for the pain. However, because faster speeds require greater muscle activation, any differences in motor control should be magnified [28,29]. Additionally, running has completely different task demands and is not well tolerated by all individuals with PFP. Ten consecutive strides were collected during the fast walk for each individual. Marker trajectories were recorded via a 10-camera motion capture system (Motion Analysis Corp, Santa Rosa, USA) and sampled at 200 Hz. Visual 3D (C-motion, Germantown, MD, USA) was used to filter the marker data, compute the functional hip joint centers and perform inverse kinematics for the lower extremities. The knee joint center was obtained by computing the midpoint of the femoral markers [27]. The marker data was low-pass filtered at 8Hz using a 4 th order Butterworth filter and joint angles were resolved using the X-Y-Z Cardan angle sequence referencing the distal with respect to the proximal. Spectral analysis of gait kinematics Ten gait strides were analyzed for each limb, in each plane for the FFT analysis. The FFT converted the sagittal, frontal and transverse plane knee kinematics from the time to the frequency domain. The FFT converted the sagittal, frontal and transverse waveforms from the time to frequency domain by representing the waveforms as a summation of sinusoids [16]. Sinusoidal functions have amplitude, frequency and phase components. The waveforms being represented by multiple sinusoids will have multiple amplitude, frequency and phase components for each sinusoid representing the gait waveform. We adopted a well-established method for FFT and power spectrum generation [30,31]. Prior to the FFT analysis the data was padded from 2000 points which is approximately 10 seconds of data to 2048 points. Then the FFT was performed and power and phase spectra were produced from the resulting data. Due to padding the data, we matched the power of the power spectrum to that of the power of the original gait waveform for each frequency [32]. For the spectra the frequencies ranged from 0 to half of the sampling rate and they were normalized to 1Hz bins. The three largest amplitudes obtained from the power spectrum represented the three dominant sinusoidal signals in the gait waveform. The three frequencies at which these dominant amplitudes occurred were stored as they represent the frequency component of the respective sinusoidal functions. The phases for the sinusoids were identified from the phase spectrum. Like the power spectrum, the phase spectrum represents the phase as a function of frequency. To determine the phases, we identified the phases at the same frequencies as the dominant amplitudes. Thus if the power spectrum had a dominant peak of 31 (Deg 2 /Hz) at frequency 1.5Hz, the accompanying phase would be the phase at 1.5Hz. Phase values range from 0 to 360˚; however, here phases were normalized to range from 0˚to 180˚by reducing the values in half. Since this was applied to all phase values it did not affect the results. The FFT analysis was performed using a custom MATLAB program (MATLAB 8.0, The MathWorks, Inc., Natick, Massachusetts, USA). Similar to Lamoth et al. (2002) [33], we used the power spectrum results to determine the cutoff for the number of waveforms analyzed. The first three waveforms were analyzed because they contained 90% of the signal energy (Fig 1). Due to the reduced magnitude, any possible differences observed in that waveform would have had minimal effect on the waveform compared to those observed in the first three waveforms. Gait waveform variability evaluation The first step in the signal-to-noise ratio (SNR) was to subtract the dominant sinusoid (sinusoid 1) from the knee kinematic waveform and becoming the noise signal. The unmodified knee kinematic waveform is the original signal. The SNR computes the ratio of the powers (P) for the original and noise signals, where x is the signal and t is time (Eqs 1-2) [34]. The result of this ratio is provided in decibels (dB). Lower SNRs indicate greater variability in the signal. Patellofemoral pain and control waveform component analysis comparison Ensemble curves of the sagittal, frontal and transverse plane knee kinematic waveforms were generated for the control and PFP groups (Fig 2). The Anderson-Darling Test and Bartlett's and Levene's tests were performed to test the data normality and the equality of the variances, respectively. A two-way ANOVA with a Bonferroni post hoc analysis was performed to compare the main effects of group and limb and the interaction of group and limb on the sagittal, frontal and transverse plane amplitude, frequency and phase measures (α = 0.05). A one-way ANOVA and Bonferroni post hoc analysis was conducted on the SNR data to evaluate between limbs differences in SNR values in the sagittal, frontal and transverse planes (α = 0.05). All statistical analyses were conducted in SPSS (Version 23, IBM, Amonk, NY, USA.). Results The comparison of the control and PFP group's age, mass, height and walking speed reported no between group differences for any of these metrics (Table 1). In the sagittal plane, between group differences were found in the frequency for the second sinusoid signals (p = 0.02) and between limb differences in the PFP group were detected in the phase for the first sinusoidal signal (p = 0.01) ( Table 2). The frequency values were significantly smaller in the PFP group compared to the control group. The phase values for the PFP injured limb were significantly larger than the non-injured limb. In the frontal plane, a group and limb interaction effect was detected in the frequency value for the third sinusoidal signal (p = 0.03) where the PFP group reported lower frequency values than the control group and greater between limb asymmetry ( Table 2). Between group differences were detected in the second sinusoidal signal phase value (p<0.01) as both limbs reported significantly larger phase values. The transverse plane did not report any significant differences in amplitude, frequency and phase between the control and PFP groups for any sinusoid. The SNR analysis found that the PFP injured limb reported significantly lower SNR values than the control right and left limbs in the sagittal plane where the control left limb reported SNR values nearly three times as large as the PFP injured limb (Table 3). In the frontal plane, the PFP non-injured limb reported significantly lower SNR values than the control limbs (p = 0.04). No significant differences were present in the transverse plane. Discussion The objective of this study was to use FFT to assess differences in knee kinematic waveform patterns between the control and PFP limbs based on their amplitude, frequency and phase components during fast walking. Significant differences in frequency and phase values between the control and PFP groups were found in the sagittal and frontal planes only. Since no differences in gait speed were measured between the groups, the reduced sagittal and frontal plane knee frequency in the PFP limbs suggest an altered, slower movement strategy was adopted by the PFP individuals. Furthermore, the significantly larger phase values suggest that the peak knee flexion-extension and abduction-adduction kinematic events were delayed during the gait cycle. These changes could alter when the knee is loaded or unloaded during the gait cycle and lead to injury [35]. The SNR results reported significantly larger sagittal and frontal variability in the PFP individuals which was consistent with previous work [21]. Thus, the SNR was able to detect the knee kinematic variability contributed by the frequency and phase metrics. Overall the FFT and SNR methods were able to delineate differences in knee kinematics between the control and PFP groups and showed that temporal changes in knee kinematics can contribute to the altered joint loading. During walking, delayed vastus medialis activation has been repeatedly observed in individuals with PFP [9][10][11][12]. This delayed vastus medialis activation contributes to poor tracking of the patella and abnormal joint loading [9,10,36]. The delayed sagittal and frontal plane peak knee kinematics measured from FFT reflects the kinematic changes associated with delayed vastus medialis activation. Sasaki and Neptune (2010) [37] found that even small changes in joint kinematics; largely as a result of quadriceps muscle activation, translated to significant increases in joint loading. Therefore, the subtle changes in joint motion detected by FFT may represent the elevated forces experienced on the knee that are causing pain. Since studies often do not investigate temporal changes in gait kinematics, they would not detect these changes that can have a considerable effect on joint loading [37][38][39]. The results support the application of FFT and other techniques that investigate both magnitude and temporal differences in joint kinematics in individuals with PFP. These methods can serve as diagnostic tools to help clinicians implement therapeutic techniques that target altering joint kinematic and muscle activation timing for PFP rehabilitation. Traditionally, SNR is used to detect the presence of noise in a signal to eliminate extraneous information. Here, we found that that extraneous noise was representative of knee kinematic variability. The elevated knee kinematic variability measured using SNR in the PFP group was consistent with work that investigated the temporal component of gait variability in individuals with PFP [21]. This SNR identified that the variability was driven by changes in frequency and phase and showed these changes to be associated with altered joint loading in PFP individuals. While the application of SNR in evaluating movement variability is still new, it appears to be promising and future work will help clarify its clinical significance. A limitation of this study is the assumption that walking biomechanics are sinusoidal. Walking biomechanics are cyclic and repetitive like sine waves which is why we used FFT to model the knee gait waveforms [32]. Additionally, the FFT amplitude, frequency and phase components are easy to show how these parameters alter waveform pattern. This assumption is found in decades of walking and running biomechanics studies, the results of which have provided important findings about differences in joint and muscle characteristics in different populations and we do not believe this assumption affected the results [30,31,40]. FFT was successfully employed to detect differences in sagittal and frontal plane knee kinematic frequency and phase values in individuals with PFP. These changes revealed that individuals with PFP slowed their knee motion and delayed when the knee was loaded during the gait cycle which resulted in elevated forces experienced at the knee. Previous research identified the significance of timing events in individuals with PFP and this work supports this finding. FFT is a quick and quantifiable method for the detection of individuals with PFP. It provides clinicians with a non-invasive assessment tool to enhance the identification of PFP in individuals. In addition, it can be used to monitor and define rehabilitation protocol(s) needed to speed the individual's return to previous activity levels. Supporting information S1 File. This file contains the data used for the study analysis. (XLSX)	v3-fos-license
2022-05-23T15:12:38.595Z	2022-05-01T00:00:00.000	248973232	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1660-4601/19/10/6266/pdf?version=1653120277", "pdf_hash": "3a4f485c7dc04b3f1f31161ab6df49fc4031ca4e", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45401", "s2fieldsofstudy": [ "Environmental Science" ], "sha1": "a4428c12ac0d459e7e9f81827be4bc5ff8676b1f", "year": 2022 }	pes2o/s2orc	Characterizing Time-Series Roving Artisanal and Small-Scale Gold Mining Activities in Indonesia Using Sentinel-1 Data The rapid growth of roving mining camps has negatively influenced their surrounding environment. Although artisanal and small-scale gold mining (ASGM) is a major source of gold production, the mining activities and their activeness are not well revealed owing to their informal, illegal, and unregulated characteristics. This study characterizes the transformations of roving camp-type ASGM (R-C-ASGM) activities in Central of Katingan Regency, Central Kalimantan Province, Indonesia, from 2015 to 2021 using remotely sensed data, such as the time-series Sentinel-1 dataset. The results show that the growth of active R-C-ASGM sites was identified at the center of the Galangan mining region with expansions to the northwest part along the Kalanaman River, especially in 2021. Hence, these approaches identify the transformations of roving mining activities and their active or nonactive status even in tropical regions experiencing frequent heavy traffic rainstorms. They provide significant information on the socioenvironmental risks possibly caused at local and regional levels. Our results also inform the design of timely interventions suited to local conditions for strengthening environmental governance. Introduction The rapid growth of the rove-type mining sector has negatively influenced their surrounding environments. Therefore, detecting such occurrences, determining their development rate, and identifying their active or nonactive status should provide significant insights into identifying possible socioenvironmental problems caused at local and regional levels. This may also allow environmental governance to be promoted at various levels. Artisanal and small-scale gold mining (ASGM) is a major source of gold production using rudimentary technology at individual or community levels despite being informal, illegal, and unregulated [1]. This sector has the largest employer in gold mining at the global level comprising 70% to 80% of informal small-scale workers [2]. Mercury is commonly used to increase the gold extraction process, resulting in highly toxic environmental and health risks due to mercury pollution throughout its emissions and release into water and the atmosphere, respectively [3][4][5]. Such mercury pollution has largely been observed in South America, Africa, and Asian regions. Indeed, environmental impacts, such as deforestation, geomorphic and hydrological changes [6][7][8][9][10][11], and health problems, such as mercury intoxication-oriented movement disorders and various injuries associated with the ASGM activities have been reported [12,13]. Despite its significant socioenvironmental impacts, more than 80 countries have continuously employed ASGM to alleviate poverty for their socioeconomic development [14,15]. Continuous growth has been observed in Indonesia. Active and nonactive ASGM practices have been placed in 93 regencies of 30 of the 34 provinces, estimating 250,000-300,000 miners [16] in more than 1200 hotspots in 2017 [17]. Furthermore, the country has been the fastest increase in polluted sites in the last 20 years on a global scale [2]. In Kalimantan island, one of the ASGM hotspots with alluvial operations, many illegal mining activities have been widespread even in conservation areas, impacting biodiversity and human health [17]. The ASGM sector can be classified into the following two types: "travel-type," in which the miners commute from their local residences to the mining sites, and "camp-type," in which the miners live and conduct mining activities on informal worksites [18] (hereafter referred to as C-ASGM). In the C-ASGM sector, both roving and non-roving practices are observed. The scale of the workforce in the ASGM sector has expanded with the increasing gold prices since 2000 [19]. The strong relationship between ASGM increases and the high price of gold has been confirmed in the literature [7,18,20]. Remote-sensing technologies have been widely used to characterize natural features and physical objects and monitor their spatial changes over time. Additionally, this technology provides a wide variety of continuous data with temporal, spatial, and spectral resolutions. Freely available satellite remote-sensing data, such as the Landsat series, have provided long-term Earth observation data since the 1970s and have been widely used for land cover detection and monitoring [21][22][23][24]. Despite the development in geoinformation technology, few studies have focused on the ASGM sector for quantitative assessments experiencing the harmful environmental and health risks caused by mercury pollution. Even [6][7][8][9][10][11][25][26][27] demonstrated time-series assessments in deforestation, mining area detection, and geomorphic and hydrological changes; however, they mainly examined the travel-type mining sites. To investigate the closed C-ASGM sites, Ref. [18] recently conducted a quantitative time-series analysis of the growth in C-ASGM sites using satellite remote-sensing imagery. Furthermore, Ref. [28] analyzed the transformation of C-ASGM activities by integrating nighttime light (NTL) intensities as a magnitude of mining activities. Although a time-series assessment of the closed C-ASGM sector with non-roving practices has been conducted by [18,28], a roving C-ASGM sector (hereafter R-C-ASGM) has not yet been discovered. The major challenges, such as acquiring an optical cloud-free timeseries dataset [18,28,29], lead to further difficulty in understanding the R-C-ASGM sector, operated at a larger scale in tropical regions experiencing frequent heavy traffic rainstorms. The use of the synthetic aperture radar (SAR), an active independent Earth observation system from solar illumination or day-night cycles [30], is an alternative suited tool for optical data [31]. Further, Ref. [32] reviewed the optical and SAR data for monitoring ASGM sites and ensured results between the datasets. Previous studies have revealed the potential of SAR data usage in mining-induced area detection using SAR sensitivities of radar systems to surface roughness and dielectric properties of materials [27,31,32]. Therefore, SAR data are a powerful tool to overcome weather-related limitations mainly found with optical sensors. This helps detect and monitor closed R-C-ASGM sectors to obtain a qualitative and comprehensive understanding. This study primarily assesses the transformation of the R-C-ASGM activities from 2015 to 2021 in Katingan Regency, Central Kalimantan Province, Indonesia, where active alluvial-based R-C-ASGM activities have been conducted. This study's results are expected to contribute to the understanding of R-C-ASGM development spread in remote rural areas, the prediction of the level of socioenvironmental pollution, and strengthening environmental governance at the regional level. Overall Methodological Workflow The methodological workflow used in this study is demonstrated in Figure 1. This workflow employed three main steps to achieve its primary objective of assessing the transformation of the R-C-ASGM activities. First, the S-1 backscattering coefficients (σ 0 ) were calculated with vertical-vertical (VV) and vertical-horizontal (VH) polarizations. Second, selections of algorithm/polarization were performed to detect the most locally sensitive values. Third, the changes in the R-C-ASGM occurrences during 2015-2021 were calculated based on the S-1 temporal series. This evidence allowed us to understand the historical transformation of the R-C-ASGM activities at the study site. This study presents a discussion based on all the findings described above. The methods used in each step are explained in the following sections. Study Area Indonesia is a well-mineralized metallogenic region with significant gold mineralization, associated with quartz veins in andesite-hosted epithermal settings. One of the major ASGM hotspots in Central Kalimantan with gold-bearing alluvial soils has attracted large ASGM-targeted migrants from Java and South Kalimantan [33]. The Galangan mining region in Central Kalimantan is the geographical and historical center of the land-based mining area, which developed rapidly in the early 1990s [32]. The Hampalit town, especially, was a base for active mining activities for both indigenous miners and a gold company, namely PT Hampalit Mas Perdahana, which closed during the financial crash of 1997. Company-initiated mining activities extracted heavy minerals through an openpit method, digging deep excavation pits. Thus, removing all the soil and vegetation landscape on the surface creates a barren wasteland [13,33]. However, indigenous ASGM communities have extracted gold along with river systems by floating pumps, resulting in disturbances of riverbanks and an increase in sediment volumes [33]. After the company's closure, the lands were taken by migrated miners. They have continuously traveled to the greater areas, from Kalanaman, Pundu to Galangan, to explore newer locations with greater gold production by seasons [33]. This study targets Galangan mining (Central of Katingan Regency, the Central Kalimantan Province, Indonesia), utilizing the alluvial-based mining method ( Figure 2). In this mining region, the Katingan River, one of the major river basins in Southeast Asia, flows north to south. S-1 Imagery A total of seven level-1 grand range detected (GRD) Sentinel-1 datasets, covering 2015-2021, downloaded from the European Space Agency (ESA), were utilized to extract and calculate time-series changes of the ASGM occurrences. Through the EU/ESA Copernicus program, the S-1 mission (S-1A and S-1B) provides an exceptional combination of high spatial (10 m) and temporal (6 days) resolution data by operating two polar-orbiting radar imaging systems working with the C band (~5.7 cm wavelength). The main operational mode is interferometric wide swath mode (IW) with VV and VH polarizations, and images are freely and routinely available [34]. To reduce atmospheric effects, which reduced the quality of images, Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) data was referred to using Google Earth Engine to target months experiencing less rain with local weather station data. Thus, this study focused on July to August from 2015 to 2021. All datasets were acquired from the descending track with relative orbit number 3 of each image's backscatter intensity to better the image. The available S-1 dense time-series offers a unique opportunity to monitor ASGM activities, especially in tropical regions experiencing the magnitude of frequent rainstorms. Image Preprocessing The preprocessing workflow is based on ESA's open-source software, ESA named sentinels application platform (version 8.0.0), and its functionalities. The following steps were implemented in the S-1 Toolbox: orbit correction, thermal noise removal, radiometric calibration, speckle filtering with 5 × 5 windows, and terrain correction using the 3-arcsec digital elevation model (DEM) from the shuttle radar topography mission (SRTM) [35]. Here, the radiometric calibration aims to convert the digital pixel value of the S-1 images into an image intensity value of σ 0 . The data were projected to the World Geodetic System 1984, Universal Transverse Mercator Zone 49 South. Terrain-corrected σ 0 intensities of the VV and VH were used for further analysis. Owing to heavy cloud coverage in the study area, the acquisition of the scenes was extremely limited only to the abovementioned data. However, these images identified mining activities along the Katingan River. According to human visual image interpretation, areas affected by mining activities were separately digitized, and the changed areas were identified by overlaying. Third, 100 points were randomly selected from the datasets to determine the best suitability by polarizations. Fourth, the determined best combination of algorithm and polarization was applied to all datasets post-classification of a majority filter with a moving window size of 5 × 5 pixels to remove isolated pixels. Furthermore, the detected areas observed in the river buffers were eliminated to remove the mudflats in the rivers, possibly caused by changes in the magnitude of precipitation between the acquired years. Consequently, the annual changes in the extent of illegal mining were calculated for the following six temporal series: In previous studies, mining areas in the Central of Katingan Regency were estimated to cover~400 km 2 in 2007 [32]. Hence, the long-term trends in R-C-ASGM sites could be observed from satellite imagery even with a 10-m ground resolution. We summarized the main specifications of the databases used in Table 1. Visualization of Time-Series Color Composites of VV and VH Polarizations The processed VV and VH polarizations were displayed in RGB color composites in six temporal series: 2015/2016, 2016/2017, 2017/2018, 2018/2019, 2019/2020, and 2020/2021, as shown in Figure 3. In this visualization process, the older years were assigned red, and the newer years were assigned green and blue, which detects changes in land covers between two different periods. VV polarizations show slightly brighter intensities compared to that of VH's. VH polarization can detect significant landcover changes along the Katingan and Kalanaman Rivers in 2016/2017 and 2020/2021. Determination of Threshold Both VV and VH polarizations acquired in 2017 and 2018 were primarily used to derive the best combination of algorithm and polarization. The changed areas identified from each result were validated using features extracted from high-resolution Google earth images (GEI), as mentioned in Section 2.5. The most sensitive algorithm and polarization channel indicate changes in illegal mining extents. After processing the optimized thresholding, the locally sensitive methods were found only in the IJ_Isodata and Yen algorithms. Therefore, those were applied both to VV & VH polarization channels. Table 2 The results show no significant value differences in both VV and VH polarizations (IJ_Isodata algorithm). The same values were generated in 2017_VV and 2018_VH from the Yen algorithm; however, 2018_VV showed a larger difference between the two periods. As new mining areas are usually associated with land cover changes from vegetation to bare areas or water, the magnitude of intensity in such areas is expected to be lower intensities in VV and VH polarizations. Thus, the IJ_Isodata algorithm was more sensitive to finding mining activity-induced land landcover changes than the Yen algorithm in this study. Detection of Newly Expanded R-C-ASGM Areas Using the results in Section 3.1, the accuracy assessment was performed to judge their sensitivity. The results show 73.3% and 76.0% for the VV and VH polarizations, respectively. We recalculated the accuracy by omitting the points found at boundaries due to high spectral resolution sensitivity resulting from mixed pixels. As a result, we found 76.7% and 82.1% accuracies for the VV and VH polarizations, respectively. The best combination was found with the IJ_Isodata algorithm with VH polarization. The particular threshold values for each VH polarization were generated for the final classification, possibly leading to better detection of active R-C-ASGM activities (Table 3). Contributions We studied the transformations of the R-C-ASGM activities from 2015 to 2021 using the S-1 time series. A quantitative time-series analysis of the R-C-ASGM sectors can help better understand the rate and pattern of development of such mining activities over time. Detecting such occurrences and their patterns in tropical regions experiencing the magnitude of frequent rainstorms can provide significant information or estimation on the potential rates and levels of socioenvironmental pollution and its human risk resulting from mercury use at R-C-ASGM sites. Understanding the characteristics of R-C-ASGM practices helps strengthen environmental governance at various levels. As described, the establishment of new mining areas is usually associated with changes in land cover from vegetation to bare/water areas. We employed a change detection method based on generating the binary masks using a threshold defined by the image, optimized by the IJ_Isodata algorithm ( Table 2). The analysis reveals that VH polarization was more sensitive than VV polarization, resulting in better separation of areas induced by mining activity (Table 3). While the classification accuracy was 76.0%, a higher accuracy (82.1%) was found with omission of random points at boundary. For 24.0% of errors, influence factors would not be caused by algorithm matter. Instead, the following factors can be considered for this misclassification; SAR specific errors such as foreshortening and layover in mountainous areas owing to the side looking of SAR; differences in data acquired time; weather conditions before the data acquired time; and spatial resolution of data. Previous studies using SAR datasets in the mining sectors only achieved 52.0% [51], 84.9% (producer accuracy), and 72.4% (user accuracy) [31]. Our study does not focus on generating a highaccuracy map of active mining, which can be a replacement for a field survey. Instead, we aim to provide information that leads to and supports the initial survey and social implementation at a local level. Without any field data, we cannot target any destinations for surveying, resulting in huge loss of time and cost. Thus, the generated possible active map helps plan field survey. Our study is not an alternative tool to a field survey; therefore, 76.0% (82.1% highest) accuracy is sufficient for this study. This study demonstrated the transformation of active sites in the R-C-ASGM sector from 2015 to 2021 in the Galangan region, Central Kalimantan, Indonesia, where active alluvial-based R-C-ASGM activities have been historically conducted. This study detected the active mines and their various transformation forms using a quantitative analysis over time ( Figure 5), as described in Section 3.2. Few studies have quantified R-C-ASGM practices with satellite imagery data. A recent study by [18,28] conducted a quantitative time-series analysis of the closed Non-R-C-ASGM sites, employing the vertical tunnel method (shaft) of mining, in Golontato, Indonesia, using optical satellite remote-sensing imagery. However, this work further quantified R-C-ASGM sites where activities are operated at a large scale in tropical regions experiencing frequent heavy rainstorms. Few studies have utilized satellite data to reveal the volume of illegal mining activities. Especially, a recent study by [28] quantified that the extent of illegal mining sites and the magnitude of mining activities in the camps experienced 4.8-and 3.8-fold increases, respectively, from 2014 to 2020. Although the study areas, mining type, and indicator of transformations differ from this study, a similar trend in the occurrence of active illegal mining activities was found in their results. Further, Ref. [32] focused on a study area that is comparable to ours. Their study found an annual expansion of 8 km 2 through the Landsat series in the Galangan region from 1999 to 2002 [32]. This study found a higher magnitude of the occurrence rate during 2015/2021. The possible reason for this may be associated with an increase in the global gold price since 2006. Similarly, the gold price in Indonesia has increased since 2007, with an especially steady increase since 2017, which approximately doubled at the end of 2020 [18]. The magnitude of occurrences was found especially in 2020/2021 when a rapid increase in the gold price was observed globally and nationally, while the lower rate of occurrence observed in 2019/2020 could be due to the globally spread influence of the coronavirus pandemic, which affects the workforce, mining activities, markets' supply chain, and cash flow [52,53]. For the shifts in the occurrence patterns, the result showed that the magnitudes of active areas were found in the western Galangan region and along with the river networks in 2020/2021. Even migrated miners have continuously roved the greater areas by season [33]. Their main mining target sites can be shifted along the Kalanaman Rivers to explore greater gold production. The possible reason for this shift can be associated with the expansions of river extents. From the European Commission Joint Research Centre Yearly Water Classification History dataset [54], which contains yearly water classifications from 1988 to 2020, water extent along the Kalanaman River areas exhibited 1.83 (2015), 2.90 (2016), 2.91 (2017), 2.76 (2018), 2.59 (2019), and 3.40 km 2 (2020), respectively. Thus, alternative R-C-ASGM sites were expected to be further developed that were associated with the expansion of water extents along the Kalanaman River after 2020. Moreover, Ref. [32] previously revealed shifts in mining direction with PALSAR (June-September 2006). The authors found the various shifts in the active area of the western Galangan region. However, a time-series analysis of the study further identifies the detailed characteristics such as volumes, mine status, and trends of active mins' shifting directions in the hidden R-C-ASGM sectors, representing a more comprehensive understanding of R-C-ASGM sectors across the region. The R-C-ASGM sector can operate successfully due to its high productivity of gold. However, it is estimated that approximately 270 tons of mercury are annually released only from Central Kalimantan to the Sea of Java as of 2007 [32]. Furthermore, severe mercury contamination (sediment, local fish, and hair samples) and typical symptoms resulting from mercury intoxications (ataxia, tremor, and dysdiadochokinesia) among workers with high ASGM activities have been observed in the Galangan region [13]. Despite its status as an informal sector, the increase in global gold prices accelerated the massive entry of immigrants into the mining sector, resulting in its massive growth. The growth in those sectors further accelerate to cause mercury-related environmental pollution and health problems at the stages of mining and amalgamation. Therefore, detecting such rapidly developing hidden R-C-ASGM sectors can provide significant insights into the potential rates and the levels of socioenvironmental pollution. This would also strengthen the environmental governance with the participation of different stakeholders at various levels. Limitations This study has certain limitations associated with the characteristics of SAR data. Although SAR data helps in the active independent observation of weather, it causes foreshortening and layover in mountainous areas owing to the side looking of SAR, leading to misclassification. Further, precipitation before the acquisition can decrease the backscatter intensity in polarizations, overestimating illegal mining extents. Moreover, some smaller and complex areas are undetectable due to the used datasets (10 × 10 km grid cell). Finally, because of the S-1 series' operation period, the methodologies applied in this study are limited only to the period after 2014. Although the proposed method cannot detect the existing mining areas before 2015, it identifies the occurrences of R-C-ASGM-related active sites and their changing patterns. Conclusions This study assessed the transformation of the R-C-ASGM sector in Katingan Regency, Central Kalimantan Province, Indonesia, using S-1 time-series data. The results presented herein show the massive occurrence of the active R-C-ASGM sites. In particular, a magnitude of occurrence was found in the center of the Galangan region and along the Kalanaman River in 2021. Therefore, it can be concluded that the active mining sector undertaken by the R-C-ASGM method can be detected from a set of time-series datasets. These results extend our understanding of the transformations of the mining site and the status of their activeness in the hidden R-C-ASGM sectors. Subsequently, it also provides significant insight into the potential for further socioenvironmental problems at the regional level. These findings are expected to assist in developing rapid and appropriate interventions for strengthening environmental governance by involving various stakeholders. Author Contributions: S.K. contributed to designing the research, data analysis, and data visualization. M.S. provided comments. M.N. provided technical advice and critical comments. All authors have read and agreed to the published version of the manuscript.	v3-fos-license
2018-12-05T16:31:27.817Z	2014-12-30T00:00:00.000	55455777	{ "extfieldsofstudy": [ "Environmental Science" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "http://www.tjsr.org/journal/index.php/tjsr/article/download/5/4", "pdf_hash": "7abf7e38b10285edac3468434b55c35a2d8f94d1", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45405", "s2fieldsofstudy": [ "Environmental Science" ], "sha1": "7abf7e38b10285edac3468434b55c35a2d8f94d1", "year": 2014 }	pes2o/s2orc	The dynamics of plant productivity under controlled conditions . Diversity or information exchange ? We result statistical analysis of experimental data on physical modeling of primary soil formation under long and continuous cultivation of plants on initially abiogenous mineral substrates (granite crushed stone, zeolite). The purpose of the experiment was to follow the dynamics of the evolutionary changes in the mineral substrate under condition long-term operation. We used the information approach to quantitative analyze of the relationship of primary soil formation process with the vital activity of plants (tomato, spring wheat) under controlled conditions. We analyzed the dynamics of the diversity of emerging organic matter in the mineral substrate and the biotic community. To quantify the diversity of multicomponent systems, we used information function. We have shown that the dynamics of plant productivity was statistically significant related to the parameter of information exchange between emerging organic matter and biotic community. It has been established that the increase in the total content of organic matter in the mineral substrate does not have a statistically significant correlation with the productivity of Introduction According to the data of our experiments [1] for longterm, uninterrupted and intensive use of mineral substrate for cultivating plants under conditions of a controlled agroecosystem, mineral substrate (granite crushed stone, zeolite) undergo significant changes due to the development of the initial stages of primary pedogenesis.An important phase in this micro-evolutionary process is the formation of a group of organic components of a different nature, and biotic community which in many respects determines the main chemical, physico-chemical, and biological properties of mineral substrate.Processes of accumulation and transformation of organic matter, formation of the biotic community in mineral substrate is similar to the processes that occur in natural conditions.Assing I.A. [2] point out the important role of organic matter in weathering of minerals, elutriation of ash minerals, and their uptake by plants.Transformation of the mineral component of rocks, in Polynov's opinion [3] , is the factor which introduces the progressive element into pedogenesis.It was revealed in many investigations (see, for example, review [4] ) that humic substances include cations of various chemical elements as well as phosphorus, silicon, sulfur, and chlorine, supplying plants with nutrient elements. During lengthy, intensive, and continuous exploitation of mineral substrate under controlled agroecosystem conditions, in it develop processes analogous to natural pedogenic ones related to the transformation of abiogenic rocks into biogenic soil-like bodies, formation of biochemically, multicomponent organic matter in mineral substrate, and development of a multispecies microbiotic community.However, under our experimental conditions the processes of forming of soil-like bodies occur headily.Already to the 15th vegetation (growth period), the enrichment by carbon of the root medium increased compared to the first vegetation about threefold.The biochemical composition of organic matter changed in during operation of mineral substrate.For example, the proportion of nonhydrolyzable residue and alkali-soluble fractions increased and the relative content of cellulose and hemicelluloses in mineral substrate decreased.Subsequent transformation of organic matter led to the formation of humic acids how for new soils. According to the data of our investigations, biological communities forming in mineral substrate under conditions of long exploitation are characterized by a considerably more quantity of living matter per unit of specific surface compared with natural soils.On density of living matter in mineral substrate can be judged on the basis that the numbers of bacteria and fungi in soils are magnitudes of the same order, whereas the specific surface of granite crushed stone is one or two orders lower than in soil (chernozem).As it is known [5] , organic matter, interacting with minerals, participates in destruction of rock, promoting extraction of chemical elements from them, and organic matter also forms labile compounds with them. Processes of mechanical transformation of the mineral substrate are accompanied by a considerable increase in the proportion of fine particles.Destruction of mineral substrate leads to a substantial change in the trophic environment and noticeable variation of the chemical composition of plant and also them productivity. The purpose of the present work was to analyze the dynamics of the plants productivity and to reveal the cause-and-effect interrelations between the dynamics of productivity of plants and dynamics multicomponent organic matter in mineral substrate and also dynamics of multicomponent biotic community.We shall use an information approach, the methods of which allow quantitatively determine of cause-and-effect interrelations between events as well as to indicate the direction of information flows between multicomponent systems.Use of the information approach allows to fulfil quantitative analysis of the interrelationship dynamics of diversity systems of organic matter and biotic community with the dynamics of plant productivity under condition of primary soil formation. Experiment Plants of tomato (variety Ottawa-60) and spring wheat (variety Siete Cerros) were cultivated in the germinator by the method of a small-volume aggregate hydroponic under two-way adjustment of the water regime of the roots. We used also Knop's nutrient solution.The nutrient solution was not changed during entire growth of the plants.We corrected the solution periodically.We used luminaires based on DNaT-400 sodium lamps with a solid-state heat-absorbing filter.The intensity of the radiant flux corresponded to 2 W/m 10 100 ± in the photosynthetically active radiation region.The photoperiod has been 16 h/days with a growth length of 75 days.The plants were grown on an originally abiogenic mineral substrate (granite crushed stone) during twentythree uninterrupted vegetation cycles and on zeolite crushed stone during the twelfth continuous vegetations.The following operations were carried out while growing of the plants.After each growth period of the plants, firstand second-order roots were removed, the mineral substrate was composted for 20-30 days, and then we were carried out by the method of regeneration (control factor) that developed in [6] .After every vegetation cycle the plant samples were taken for study of the chemical elemental composition of the different plant organs. After each growth period, we studied the chemical composition of organic matter, which included cellulose, hemicelluloses, water-and alkali-soluble and alcoholbenzene fractions, and nonhydrolyzable residues according to the experimental scheme given in [5] .To manage plant productivity, the experimental design included a change of crops: after 12th growth period on the mineral substrate on which wheat had been grown, we began were grown tomato, and, conversely.Furthermore, we grown green manure crops instead of tomato and wheat in the 20th and 22nd growth periods. Number of variants 1, 2, 3 and 4, 5, 6 we used for tomato and spring wheat, respectively.Check experiments are the variants 3 and 6.To reduce the total organic matter in the mineral substrate (granite crushed stone), we implemented additionally acid-base regeneration in the experiment (variants 1 and 4).Processing is realized in the following sequence: performs composting, then treated with 0.005-0.01nH 2 SO 4 acid solution, then washed with water, then treated with 0.05n alkali solution, then washed with water again.Such complex effects can reduce the aging process of the mineral substrate.Additionally for variants 2 and 5 we carried out three times the regeneration of the roots of growing plants.Growing plants on the zeolite crushed stone was carried out only for the check experiments. We monitored the dynamics of the changes in the chemical composition (percentage composition of matter) of organic matter quantity in mineral substrate, including cellulose, hemicelluloses, alkaline-soluble and alcoholbenzene fractions and nonhydrolyzed residue, as well of its water-soluble part in the nutrient solution.Simultaneously is analyzed the dynamics of the species composition and the amount of microorganisms (bacteria consuming mineral nitrogen, bacteria consuming organic nitrogen, spore-forming bacteria, cellulose-fermenting bacteria, fungi and actinomycetes) in mineral substrate. We investigated dynamics of content of chemical elements in the composition of the ash of plant roots, reproductive organs, stems and leaves: Ca, K, P, S, Na, Si, Al, Fe, Zn, Mn, Mg, and Cl.Analysis of the elemental composition of plant ash after each growth period we made by using X-ray fluorescent analyzer A-30.Preparation of the plant samples and determination of dry matter and also percentage content of ash in plants were done according to method [7] . Information approach It is known that the traditional methods of processing experimental data -regression, correlation, variance, etc. enable revealing only the correspondence between the phenomena or objects being studied but not the cause-andeffect relation between them.Processes can be correlated but at the same time not cause-related.The information theory method using conditional informational function and conditional probabilities (i.e., the probability of occurrence of event X depends on the occurrence of event Y) when dynamic indices make it possible to unambiguously indicate the direction of information flows into intercoupling of the multicomponent subsystems.In the literature it is noted that processes of transfer of both energy and matter and information are characteristic for soil systems.However, the quantitative measure of information transfer there is not in the literature.Calculation of unconditional and conditional information function permits quantitatively assessment of the measure and direction of information transmission between coupled subsystems.Under the dependence of two events we imply the possibility of occurrence of one of the events from appearing of the other.Here we try to establish a linkage between the dynamics of plant productivity with information exchange between dynamic subsystemsorganic matter and biotic community of the mineral substrate. For a quantitative description of dynamics of multicomponent system (biochemical composition of organic matter and numerical or species compositions of the biotic community) it is convenient to use an integral index -the Shannon information function [8] .Information theory methods allow approach to study of the temporal organization between coupled subsystems from qualitatively new positions.This approach to an analysis of experimental results belongs to the class of statistical methods of reducing the sample size and singling out the most common informative factors.Information function of the finite ensemble of objects can be written in nat units in the following way: under additional conditions where n is a discrete number of object-features of the set, which determine the space of possible values of i P .Information function ( 1) is an integral index of the state of the multicomponent system.The values of i P determine the share of the ith element in the entire collection of the set of elements (e.g., its percentage content or relative content of organic matter fractions of the mineral substrate), i.e., i P assigns the number of realizations or possible ways out.Actually, to calculate the specific number i P , we use A.N.Kolmogorov's combinatorial approach [9] for a collection of n elements entering into this set with normalized mass i P .Function ( 1) is used for a quantitative determination of the measure of organization or diversity of multicomponent subsystems.For example, the values of i P are calculated from the data on the content (percentage) of organic matter components in mineral substrate and biotic community (percentage) for each growth period (vegetation). We analyzed the temporal variations of function (1) for multicomponent systems at discrete moment of times t = T, 2T,…,23T, where T is the duration of one growth period.The number 23 indicates the total number of growth periods.It was established earlier [1,10] that during intensive and long-term exploitation of mineral substrate the trophic environment changes such that the diversity (with respect to percentage content of components) of organic matter and microbiotic community forming in mineral substrate decreases This process accompanied by an increase in the information function.(Figures.1 and 2). However, after crop rotation (12th growth period) the structuredness or diversity of these multicomponent systems begins to increase.This is characterized by a decrease of information function with a simultaneous increase in total organic matter content in mineral substrate. Change of plants productivity (see below) does not match to quantitative change in diversity (information function H) in the subsystems of the organic matter and biotic community (Figures 1 and 2).Productivity of the plants is highest possible for first growth periods for both wheat and tomato.The same vegetations are characterized by a high diversity of subsystems of organic matter and biotic complex.In the area of maximum of the information function elements of multicomponent subsystem are most uniformly distributed that corresponds to the minimum diversity.With increasing duration of exploitation of mineral substrate the diversity of subsystems decreases.At the same time reduced the productivity of plants. However, after the crop rotation (12th vegetation) diversity of the subsystems begins to noticeably is increased.This increase of the diversity was accompanied by a slight increase of plant productivity.Although a diversity of subsystems becomes even higher than for the initial vegetations the productivity of plants remains low.As it is shown by the statistical analysis, this interrelationship of the diversity and productivity has only qualitative character.This result does not allow us unambiguously to relate the dynamics of productivity of the plants with the dynamics of the diversity of the chemical composition of organic matter and the diversity of the biotic community.At the same time, a monotonic increase in the content of organic matter in the rooting medium also does not explain the trend in increase of the plants productivity ranging from 19 to 23th vegetations (see below). In this regard, we analyze quantitatively the dynamics of the exchange of information between multicomponent subsystems: organic matter and biotic community.Methods of information theory allows us to do this by using conditional information functions.Information approach allows us to link quantify the dynamics of variability of multicomponent subsystems with the dynamics of plant productivity. Results and Discussion As is known, information function of a complex experiment (simultaneous realization of the events X and Y) can be determined by the equation [11] : is the conditional probability (in terms of Kolmogorov A. N. [9] ) of realizing state X i of ensemble means that the information about the event X is partially contained in the event Y. Using dynamic conditional information function and conditional proportions uniquely opens the possibility to indicate a preferred direction of information flow in the interconnected subsystems.In accordance with [10] then, we can speak about information flows from source Y to addressee to X .If events X and Y are independent, then obviously quantity information flow equal to zero.The smaller this quantity (positive), the greater the information flow from subsystem Y to subsystem X and the higher the cause-andeffect relation between the interrelated processes of a change in the chemical composition of organic matter and biotic community which develop in time.Information flows determine the objective content of the relation between interacting material objects manifesting itself in a change in the state of these objects.In order to be able to compare the dynamics of productivity of tomato and wheat we will use the relative productivity.Relative productivity, we have determined by dividing the actual productivity at its average value for the entire period of observation for each variant of the experiment.3 shows the dynamics of relative plant productivity with respect to average productivity.For the first vegetation cycle the productivity of plants is maximum and is close to the productivity of plants grown in the black soil.However, the complex changes in the mineral substrate, which occur during long-term operation of the substrate leads to a deterioration in the quality of agro-substrate.This, in turns, is accompanied by a decrease in plant productivity.In continuous operation mineral substrate productivity both tomato and wheat decreased in all variants of the experiment.As follows from our experiments, the intensity of the decrease in plant productivity above for granite crushed stone, compared with the zeolite. The greatest decrease in plant productivity observed after the eleventh growth period.However, after the crop rotation (the twelfth growth period), and the cultivation of green manure (eighteenth, twentieth and twenty-second growth periods) downward trend of productivity is changed to its increase.It is well known that crop rotation improves the quality of organic matter in the roots medium.Instead of tomato we were grown wheat, and instead of wheat was grown tomato. In agricultural practice, it is assumed that the productivity of plants is related to the content of organic matter in the roots medium.Our experiment shows that the progressive accumulation of organic matter in the mineral substrate does not correlate with the productivity of plants (Figures 4A and 4B).That is, the content of total organic matter in the mineral substrate is not the factor determining the productivity of plants. Using both the conditional information function and conditional proportions, opens the possibility of quantitatively point the predominant direction of information flow (causal relationship) in the interconnected subsystems.A comparison of the dynamics of information functions (Figures 1 and 2) for organic matter and biotic community demonstrates that for last vegetation the diversity of these coupled systems is increasing, but it is not correlated with the productivity of plants.This result does not allow us to associate the dynamics of plant productivity with the dynamics of the diversity of chemical composition organic matter and biotic community in mineral substrate. We will carry out statistical quantitative analysis of the relationship between the dynamics of plant productivity and the exchange of information between the multicomponent systems of organic matter in mineral substrate and biotic community.Figures 5A and 5B shows the dynamics demonstrating the nonlinear temporal dependence of the information flow from organic matter to the biotic community. With increase in the duration of use mineral substrate, the cause-and-effect relation between these systems weakens.However, it begins to increase already by the 19th growth period.This temporal dependence of the cause-and-effect relations correlates with nonlinear dynamics of inverse productivity of plants, showing an increase of plant productivity by the 23rd growth period (Figures 5A and 5B). The results don't depend on the plant species being grown or the variant of the experiment.At the same time, the experimental data on the dynamics of accumulation of organic matter indicates its statistically significant increase during the entire observation period.Table 1 summarizes the statistical properties of deterministic mutual relations of plant productivity with the magnitude of information flows between organic matter in the mineral substrate and biotic community.Large t-values point to a close statistical connection between these factors.Table 1 also presents the correlation coefficients, which are significantly greater than statistical tests. Over the entire period of experimental observation the dimensionless quantity parameter S(t) > 0 that is, the diffusion of information flow increasingly from a multicomponent organic matter to biotic community.The weakening of the causal linkage between these systems leads to a statistically significant reduction of plants productivity.Rotation of the crops leads to amplification of the cause-and-effect relations.Thus the exchange of information becomes a quantitative reality as well as the exchange of energy and matter.Thus, the causal-andeffect relation between the dynamics state of organic matter forming in a mineral substrate and dynamics of the biotic community has not only scientific interest but also of practical significance and is an important information factor of the dynamics of system mineral substrateplants related to the production potential of plants. Crop rotation also affected the content of chemical elements in plants [12] .Evidently, intensive saturation of plant tissues with chemical elements is related to progressive weathering of the mineral substrate by organic matter, biotic community and vital activity of root system of plants.Evolutionary processes of primary pedogenesis are accompanied by the formation in the mineral substrate of organic matter which accelerating the rate of weathering of chemical elements from the mineral substrate. .We used the statistical method for rates and proportions [13] for determination of the mutual accordance in the positions of the maxima and minima of the information function dH(t) = H exp (t) -H trend (t) (Figure 6B).We calculated the information function H exp (t) using the experimental data.The methods of information theory allow us quantitative determine the harmonicity of a multicomponent system at a certain time t.If the value of the relative information function , which is equal to 0.618 (this number is related to the sequence of values of the Fibonacci series), then such system is close to the state of structural optimum or structural consistency; max H corresponds to the state of a multicomponent system with equally distributed components.The harmonic states of organic matter are located in the time region after crop rotation (Figure 6A) and cultivation of green manure crops. Discussing the problem of plant productivity, we analyzed the dynamics of nitrate content in plants.Experimentally is observed rising nitrate content in tomato fruits and grains of wheat for the first vegetations.However, with increasing duration of exploitation of mineral substrate nitrate content decreased (Figure 7A). It is known that accumulation of nitrate in plants is closely associated with the complex of organic compounds in the root medium, which affect the vital activity of the roots and metabolism of nitrogen compounds in plants (Figure 7).We can assume that the complex organic compounds are connected with nitric nitrogen in the plants [14] . For ease of comparison of the dynamics of the total carbon content in the root medium and dynamics of nitrate nitrogen content in the reproductive organs (Figure 7A), we use the inverse magnitude of total carbon content (Figure 7B).Figures 7A and 7B also shows the results of an additional experiment.In this experiment, we grow tomatoes on the zeolite during the twelfth uninterrupted vegetation cycles.We found a statistically significant correlation between the inverse magnitude of the total carbon content in the mineral substrate and the content of nitrate nitrogen in plant products (Table 2). Table 2 gives statistical criteria for interrelations variants 2 and 3, as well as variants 5 and 6.An increase in the total carbon in mineral substrate correlates with the reduction of nitrate nitrogen in plants.We found that by the twenty-third of growth period the content of total carbon in the mineral substrate is increased by more than ten times.This increase is accompanied by a decrease of nitrates in tomato fruit by more than two times, and in wheat more than six times.As it is shown by our experiment the dynamical interrelationships between the content of the total carbon in the mineral substrate and content of nitrates in the reproductive organs does not depend on the botanical type of plant and used mineral substrate.It should be noted that in our experiment the nitrate content in the crop significantly below the maximum allowable concentration for plants.In our experiment is not confirmed interrelationship of plant productivity and organic matter content in the rooting medium.The experiment showed that the accumulation of total carbon in the mineral substrate does not increase the yield of plants. Conclusion Our experimental data demonstrate that under condition of primary soil-formation in the mineral substrate arise interrelated processes of formation of multicomponent organic matter and multispecies biotic community.Under the conditions of intensive exploitation of mineral substrates occurs biogenic and physico-chemical weathering of mineral rocks, minerals destruction, formation of fine fractions.These processes have a significant effect on the vital functions of plants and on content of chemical elements in plant tissues [12] .With increasing duration of operation of initially abiotic mineral substrate we observed phenomena that resemble soil fatigue.Plant productivity is reduced.However, the use of crop rotation stops this negative process.Crop rotation returns the system of plants -soil to a more stable state. The experiment showed that, although the productivity of wheat and tomato plants to 23rd vegetation increases, however, the productivity is notably lower than in the early growing season.This relationship is remained for all variants of the experiment and does not depend on the type of mineral substrate.This result does not allow us unambiguously to relate the dynamics of plants productivity with dynamics of the diversity of the biochemical composition of organic matter and biological diversity of the biotic community.At the same time, the collective state of the chemical elements in the different plant organs closely related both for starting vegetation cycle and for final growing cycle and does not correlate with plant productivity [12]. Application of the methods of information theory have allowed us to introduce a quantitative measure that determines the flow of information between linked multicomponent systems.As shown our study the weakening of the cause-and-effect linkage between the subsystems of organic matter and biotic community, leads to a statistically significant reduction in plant productivity.At the same time, a monotonic increase of the organic matter content in the mineral substrate also can not explain the increase in productivity in the range of 19 to 23th vegetation cycles.The use of two methods of acidalkaline regeneration of a mineral substrate does not change the direction of the evolutionary processes of soil formation. Figure Figure3shows the dynamics of relative plant productivity with respect to average productivity.For the first vegetation cycle the productivity of plants is maximum and is close to the productivity of plants grown in the black soil.However, the complex changes in the mineral substrate, which occur during long-term operation of the substrate leads to a deterioration in the quality of agro-substrate.This, in turns, is accompanied by a decrease in plant productivity.In continuous operation mineral substrate productivity both tomato and wheat decreased in all variants of the experiment.As follows Figure 6 Figure 6 shows the dynamics of information function H(t) for total organic matter in mineral substrate and dynamics of multicomponent elemental chemical composition of tomato roots.R 2 = 0.94 is the coefficient of determination; Fisher's test equal to >> = 8 144.F 0 5 05 0 0 1 .F ) cr ( .; .= ; Student's test equal to t = 12.03 > Figure 6 . Figure 6.Comparison of information functions for trends (A) and oscillations (B) about trend of the elemental chemical composition (--) of tomato roots and organic matter (---) in the mineral substrate (granite crushed stone) under tomato.The harmonic state of organic matter is marked by circles	v3-fos-license
2020-12-10T09:03:46.339Z	2020-12-05T00:00:00.000	230626433	{ "extfieldsofstudy": [ "Chemistry" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.26656/fr.2017.5(1).458", "pdf_hash": "f2242421032543eecee0710e3aa415268475952b", "pdf_src": "MergedPDFExtraction", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45408", "s2fieldsofstudy": [ "Agricultural and Food Sciences" ], "sha1": "6bcdfa4b78d4e2918e81dd217a0ff0f394463dad", "year": 2020 }	pes2o/s2orc	Anadara granosa substitution in feed to improve the zinc, protein of the feed, serum albumin, and body weight of malnourished rats The objective of this paper was to evaluate zinc and protein feed levels of Anadara granosa flour substitution and its effect of feed on serum albumin and body weight of a malnourished rat (Rattus norvegicus). This study comprised two stages: 1) analysis of protein and zinc level of the feed, and 2) feed test on rat using separate sample pre–test and post-test control group design. The malnourished rat was treated with dried rice; the dried rice was substituted with 12.5%, 25%, and 50% of Anadara granosa flour. Further, at room temperature, the flour can be kept for six months. This study revealed that the Anadara granosa-substituted feed had zinc levels between 0.999 ppm 2.296 ppm and protein levels of 14.81% 26.39%. On the other hand, the non-substituted feed had 0.791 ppm of zinc, and the protein level was 8.46%. Provision of the feed substituted with Anadara granosa flour increased the albumin serum level (p = 0.000) and the bodyweight of malnourished rats significantly (p = 0.002). This study revealed that substitution of Anadara granosa flour in feed could improve the zinc and protein level of the feed, which in turn improved the growth of malnourished rats (as the albumin level and the bodyweight also improved). Introduction Malnutrition, especially zinc deficit, is largely found on children in developing countries; in childhood stage, children need a higher zinc intake (Yanagisawa, 2004;Parveen and Dipti, 2016). Zinc deficit would influence homeostasis within the biological system (Kaur et al., 2016). A nutrition deficit condition that occurs for a long time might lead to stunting (Chirande et al., 2015;Mardewi et al., 2016). Stunting is one of the prominent health problems in Indonesia as its prevalence is still currently above 20% (Kemenkes, 2018). Provision of zinc supplements is proven to be able to treat stunting (Kusudaryati et al., 2017) since children with stunting have a low level of zinc and albumin serum levels. Thus, the administration of Vitamin A with zinc supplements could reduce infection risk and improve the linear growth of children with stunting (Adriani and Wirjatmadi, 2014). Gibson and Ferguson (1994) reported that zinc deficiency could be solved by increasing zinc intake. As what has been done in Africa, increasing zinc intake through zinc-rich meats and meals that are economically and culturally acceptable in that area, the deficiency problem could be addressed. Zinc is commonly available in high protein food. Animal-based meals are the main source of zinc in the human diet (Agustian et al., 2009;Kaur et al., 2016). Zinc is one of the most essential elements in nutrition for humans and animals. In the growth process, zinc helps in protein synthesis to form new cells, growth, and bone development (Agustian et al., 2009). Zinc also has several physiological characteristics and activates enzymes within the body (Kaur et al., 2016). Anadara granosa is one of the seafood that is rich in zinc contents. Fresh Anadara granosa contain 19.48% of protein and 13.91 ppm of zinc (Nurjanah et al., 2005). It also contains complex amino acids. The protein contents in the Anadara granosa would help absorb zinc and increase protein intake to the body. Therefore, zinc and protein in the Anadara granosa will work in synergy to improve zinc levels of malnutrition rats (Solang et al., 2013). On top of that, zinc also plays a role in inducing metallothionein (zinc-binding protein); thus, it regulates eISSN: 2550-2166 © 2020 The Authors. Published by Rynnye Lyan Resources FULL PAPER the amino acid as the precursor for the synthesis of albumin (Tekeli, 2002). The level of albumin plasma could be used as a sensitive indicator of nutrition status that reflects a diet pattern (Kuwahata et al., 2017). The previous studies above serves as the rationale, as the present study aims to evaluate zinc and protein in Anadara granosa flour-substituted feed and the feed's effect on serum albumin levels and body weight of a malnourished rat (Rattus norvegicus). This study hypothesized that Anadara granosa-substituted feeds can improve zinc and protein feed levels, and the consumption of the feeds can improve albumin levels and body weight of malnourished rat. Research design and sample The Anadara granosa used as substitute feeds in this study were taken from Gorontalo province, Indonesia. Rats were fed in the form of pellets consisting of standard feed, dried rice, and Anadara granosa flour substitution. The standard pellet feed produced by PT Charoen Pokpan contain 13% water, 13-15% protein, 3% fat, 6% ash, and 0.8% calcium. The dried rice feed was used to create malnutrition conditions. The dried rice was made of leftover rice. The rice that will no longer be consumed was set out to dry under the sun. After adding flour, the rice is cooked and stirred until mixed well. The dough is poured into a grinder and was shaped into a pellet. Further, the pellet was set to dry under the sun for the second time. Further, the processes of making the flour are as follows. First, the whole cockles were boiled until they open. Then, the soft tissues of the cockles were removed. After that, the tissues were cut into small pieces. All the cut tissues were dried in the sun to get a constant dry weight. The tissues, after the sun-drying process, were finely milled and sifted to get the cockle flour. Anadara granosa flour contained 27.26% of protein and 0.7913 ppm of zinc and the flour can be kept at room temperature for six months. This study was designed on testing rats using the Separate Sample Pre-Post Test Control Group Design (Campbell and Stanley, 1963;Handley et al., 2018). On the first test, 12 male Wistar rats aged 6 weeks with 115-120 g bodyweight were given standard feed (normal group). Meanwhile, 36 male Wistar rats aged 6 weeks with a bodyweight of 110-120 g were treated using dried rice pellet feed with 8% of protein (nutrition deficit group). In the 8 th weeks, 4 rats from the normal control group and 4 rats from the nutrition deficit group were slaughtered to test their albumin level. The rat with albumin serum less than 3.3 mg/dL was regarded to have nutrition deficit (Giknis and Clifford, 2008;Susanto et al., 2010). In the second phase, normal control rats (8 rats) were treated with standard pellet feed. Thirty-two malnutrition rat were randomly distributed into 4 groups; 8 rats were fed with dried rice, 8 rats were fed with 12.5% Anadara granosa -substituted dried rice (Anadara granosa flour 12.5% + 87.5% dried rice), 8 rats were fed with 25% Anadara granosa -substituted dried rice (Anadara granosa flour 25% + 75% dried rice), and 8 rats were fed with 50% Anadara granosa -substituted dried rice (Anadara granosa flour 50% + 50% dried rice) for 8 weeks. The percentage of Anadara granosa substitution in dried rice is based on the zinc needs of children per day, which is 10 g (Kartono et al., 2012), then converted to rat weighing 200 g. Food and drink for rats were given through ad libitum. Ethics test This study has obtained the ethical certificate No: 11 -KEPK from the Faculty of Public Health, Universitas Airlangga. The implementation methods were presented in front of the ethical committee and have obtained its approval. Procedure Proximate analysis of dried rice feed and feed substituted with Anadara granosa flour were carried out at the Laboratory Unit of Veterinary Faculty of Universitas Airlangga. Acclimatization, cultivation, and surgery of the test animals were carried out at the Veterinary Laboratory of Biochemical Department of Medical Faculty of Universitas Airlangga. Protein content test was carried out with the micro Kjeldahl method (AOAC, 1995). The measurement of protein content in the blood cockle flour was conducted at the Animal Feed Laboratory, Faculty of Veterinary Medicine, Universitas Airlangga. Whereas the Soxhlet method was employed to test the fat level (AOAC, 1995). In addition, the zinc level of the feed was measured using Atomic Absorbant Spectrophotometer (AAS) with the Zenit 700 tool. Further, the analysis of the zinc content of the flour was carried out at the Center for Health Laboratory, Surabaya. As much as 3 mL of a blood sample from the tested rats were taken from the heart and stored within the Blood collection tube EDTA. To obtain the blood serum, the speed was set into 3000 rpm for 15 minutes centrifuged the blood sample. The serum was then separated into Eppendorf tubes. Further, the Albumin level was tested by automatic chemical analysis Prestige 24i. Cat. No. 4-238 with Bromcresol Green (BCG) method (Doumas and Peters, 2009). The eISSN: 2550-2166 © 2020 The Authors. Published by Rynnye Lyan Resources FULL PAPER albumin level was stated in g/dL. The bodyweight of the tested rats was measured using a digital scale of Camry brand with the maximum capacity of 500 g, division 0.1 g with the accuracy level of two numbers behind the coma. Statistical analysis The zinc and protein levels were analyzed in descriptive manner. Meanwhile, the serum albumin level of the tested rat was analyzed using One way ANOVA parametric test in the significance level of 95% and the least significant difference (LSD). Further, the body weight was analyzed using the Kruskal -Wallis test on the significance level of 95% and followed by an advance test of Mann Whitney (Steel and Torrie, 1980;McDonald, 2014). Zinc and Protein level of Anadara granosa floursubstituted feed The analysis of the dried rice feed used to create malnutrition conditions and Anadara granosa substituted feed showed a different protein and zinc levels. The dried feed rice had a protein level of 8.462% and zinc 0.7913 ppm. The Anadara granosa flour had a protein level of 14.81% and zinc 0.9995 ppm. Meanwhile, dried rice feed substituted with 25% of Anadara granosa flour had a protein level of 18.74% and zinc 1.151 ppm. In addition, the 50% substitution of Anadara granosa flour had a protein level of 26.394% and zinc 2.296 ppm ( Table 1). The level of zinc in Anadara granosa's floursubstituted feed increased along with the increase of Anadara granosa composition within the feed ( Table 2). The increase of zinc level percentage on feed substituted with 12.5%, 25%, and 50 Anadara granosa's feed-in the sequence were 26.31%, 45.46%, and 58.35% respectively. Meanwhile, the increase of protein level of the feed substituted with 12.5%, 25%, and 50% of Anadara granosa's flour were 75%, 121%, and 211% in consecutive order. This showed that the substitution of Anadara granosa's flour could increase the zinc and protein level of the feed. Albumin level The results revealed that the Anadara granosa flour substitution feed significantly increased the albumin level of the malnutrition rats (p = 0.000) ( Table 3). The average albumin level of malnutrition rats was 16.16% lower than a normal rat (rat fed with standard feed). In the present work, the percentage was employed to describe the decline in the albumin content due to the differences in the smaller, last number (the level of lownutrient albumin) and the initial, greater number (the normal group of albumin level). On the other hand, the use of the g/dL unit in measuring the decline and increase in the albumin level will only hamper the process of predicting the increase or decline of the albumin content. Moreover, this percentage also applies to the increase in the albumin level; it takes into account the differences in the recent level of the albumin content of the poor-nourished mice (that had been provided with blood cockle flour) and the level of albumin in the normal group mice). The poor-nourished mice have a higher albumin level than the normal group. Another point worth considering is that the percentage provides more significant data (representing the level of albumin content). The average albumin level of nutrition deficit rat who were treated with Anadara granosa flour substituted feed increased along with the increase of Anadara granosa's composition within the feed. The albumin level of the malnutrition rat treated with 25% and 50% of Anadara granosa's substituted feed similarly experienced an increase of 22.59%. The substitution of Bodyweight The result showed that the malnutrition rats fed with Anadara granosa flour substitution experienced a significant increase in body weight (p = 0.002) ( Table 4). The average bodyweight of the malnutrition rats was 56.27% lower than normal rat treated with standard feed. Further, the average body weight of malnutrition rat treated with 25% and 50% of Anadara granosa flour substitution increased by 50.16%. Moreover, the bodyweight of malnutrition rat treated with feed substituted with 50% Anadara granosa's flour was yet to achieve the maximum result, as indicated by the rats' below normal bodyweight. The level of zinc and protein in the Anadara granosa flour-substituted feed The feed substituted with Anadara granosa flour had a higher zinc content due to the natural zinc. Similar studies by Solang et al. (2017) showed that the zinc level in cireng snack made with flour substituted with Anadara granosa flour increased. The increase of zinc in food made with Anadara granosa substitution flour indicates the Anadara granosa flour's potential as an alternative source of zinc. This is shown in several studies, where Anadara granosa are found to contain zinc level below the maximum allowed level, which is under 100 ppm as stipulated in Malaysian Food Regulation 1985(Ministry of Health Malaysia, 1985. In the meantime, the fresh Anadara granosa from Boalemo, Gorontalo Province, Indonesia is found to contain 13.91 ppm zinc, while the dried Anadara granosa contain 54.27 ppm zinc the boiled Anadara granosa contain 12.99 ppm zinc, and the dried, boiled Anadara granosa contain 37.86 ppm of zinc (Nurjanah et al., 2005). On top of that, the Anadara granosa from Pohuwato regency of Gorontalo Province Indonesia was found to contain 2.70 -2.82 ppm (Solang et al., 2013). In addition, the Anadara granosa taken from Tanjung Mas and Wedung water of Semarang, Indonesia contains zinc 68.13 -94.22 mg/kg of wet weight (Taurusiana et al., 2014). This present study also revealed that substitution of Anadara granosa's flour could increase the protein level. This signifies that the substitution of Anadara granosa's flour is one of the alternatives to improve the level of protein within the meal. This present finding supports Subaryono et al. (2003) who found that chips added with Anadara granosa had the protein level of 16.51%. Solang et al. (2017) also described that supplementation of Anadara granosa increased the level of protein in Cireng snack with a percentage of 5.05% -54.49%. Several other studies showed that Anadara granosa were considered as important source of protein in tropical, subtropical, and warm climate regions (Broom, 1985;Ibrahim, 1995;Nurnadia et al., 2011). The protein level of Anadara granosa was 19.8 % (Nurjanah et al., 2005). No. Treatment Average albumin level This present study has pointed out that the higher the composition of Anadara granosa flour into a meal, the meal's zinc and protein level will also increase. Kaji and Nishi (2006) also found that main meals composed of eggs, milk, poultry, and fish have lower zinc: protein ratio compared to meals made of cockles, beef, and other red meats. Albumin serum level This present study shows that the malnutritioned rat given the dried rice were low in zinc level by 0.791 ppm, in protein level by 8.46%, as well as low in albumin level by 46%. It is also shown that the substitution of Anadara granosa flour increased the albumin level on the tested rat. The malnutritioned rats had an albumin level of 3.01 g/dL. The level is considered below the normal level of albumin for a rat, which was 3.3 g/dL (Giknis and Clifford, 2008;Susanto et al., 2010). Meanwhile, the provision of feed substituted with Anadara granosa's flour could increase the albumin level of rat into the normal range, about 3.64 -3.69 g /dL (Table 3). This finding is similar to Giknis et al. (2008) who found that normal male rat had the range of albumin level between 3.4 -4.8 g/dL. This present study is also similar to Kuwahata et al., (2017) who discovered that albumin concentration on rats decreased due to the decrease of protein intake and it returned to normal when the malnutrition rats were given protein intake by 20% of casein through ad libitum method. Khasanah et al. (2015), Abdullahi et al. (2018), and Gounden et al. (2018) stated that protein deficiency in food intake could lead to lack of various essential amino acid in blood serum, which needed to develop cells (synthesis) and for metabolism process as an amino acid is the precursor for albumin synthesis. The lack of amino acid in this serum would lead to a lack of albumin liver production (protein). Provision of feed substituted with Anadara granosa flour could improve the level of albumin in malnutrition rat. This is assumed to correlate with the increase of zinc and protein level on the substituted feed (Tables 1 and 2). Choundhary (2013), described that the provision of zinc could increase protein serum. In turn, the protein contents could increase the absorption and transportation of zinc. Moreover, the presence of protein could increase the availability of amino acid as the precursor for albumin synthesis. Synthesis of albumin depends on the adequate amino acid supply (Marshall, 2012). Meanwhile, Shidhu et al. (2004) showed that the provision of zinc on protein-deficient rats could help increase the level of hepatic protein. This zinc ability is linked to its role to induce metallothionein (zinc-binding protein). Thus, it regulates the amino acid as the precursor for albumin synthesis (Tekeli, 2002). According to the results, the Anadara granosa flour contains the amino acid, glutamate, aspartate, serine, histidine, glycine, arginine, alanine, tyrosine, methionine, valine, phenylalanine, isoleucine, leucine, and lysine. In the present study, the analysis of cysteine and proline was not performed since there was no standard for examining these amino acids. The common amino acid found in the study is glutamate. Azis (2007) reports that Anadara granosa contain cysteine. The amino acids that contribute to the metabolism of zinc are cysteine and histidine (Snedeker and Greger, 1983;Pace and Weerapana, 2014). Both of these amino acids function to transport zinc by facilitating the process of the formation of zinc-histidine and zinc-cysteine complex. Types of amino acids that serve as a precursor of albumin synthesis are lysine, tryptophan, and isoleucine. Isoleucine and tryptophan also play a role in increasing the value of albumin synthesis (Kelman et al., 1972). Hutson et al. (1987, as cited in Harp et al., 1991 argue that the deficit of essential amino acid content, such as lysine, tryptophan, and isoleucine, can decrease the albumin release. Therefore, zinc and protein in Anadara granosa were assumed to work simultaneously to increase the availability of amino acid as a precursor for albumin synthesis; hence, increase the albumin level of malnutrition rat. Bodyweight This present study showed that the provision of dried rice feed could decrease the bodyweight of the test rat, while the Anadara granosa flour substitution feed was proven to increase the bodyweight of the tested rat. A decrease in the bodyweight of the tested rats was assumed to be caused by the low zinc and protein contents in the dried rice feed. The zinc level of dried rice was below 1 ppm (Choundhary, 2013). Meanwhile, the protein level of dried rice was 8%; thus, it was considered unsuitable for the rats' protein needs at 12% (complete protein with 20 amino acids) (Smith and Mangkoewidjojo, 1987). Moreover, Shidhu et al. (2004) reported that protein deficiency decreases body weight. Meanwhile, a decrease in body weight could also blame zinc deficiency (Rossi et al., 2001;Ishikawa et al., 2008;Parveen and Dipti, 2016). The provision of Anadara granosa substituted feed could significantly improve bodyweight. This was suspected due to the increased zinc and protein level in the feed (Tables 1 and 2). Parveen and Dipti (2016) explained that zinc supplementation causes a quick bodyweight increase in the malnutrition rehabilitation phase. Budiastutik et al. (2011) also described that supplementation of zinc eISSN: 2550-2166 © 2020 The Authors. Published by Rynnye Lyan Resources FULL PAPER phosphate and biscuit increases body weight and height. Meanwhile, an increase of protein on the feed also regulates the bodyweight through regulating the mechanism of thermogenesis and body composition, food intake, as well as protein synthesis (Westerterp-Plantenga, 2003;Greco et al., 2017). Further, Westerterp -Plantenga (2003) also explained that animal-based protein-induced higher thermogenesis than plant-based protein. The increase of body weight in malnutrition rat fed with Anadara granosa substituted feed was assumed to correlate with the increase of albumin serum as observed in this study. The increase of albumin levels would optimize the zinc absorption of the feed. Marshall (2012) explained that albumin functions to transport zinc. As the zinc has been appropriately absorbed, the zinc content within the body was assumed to increase. This availability of zinc within the body would accelerate growth and cell differentiation (Parveen and Dipti, 2016). Conclusion This study has shown that the provision of feed substituted with Anadara granosa flour could improve the zinc level and protein of the feed. The increase of zinc and protein could improve the availability of zinc and protein within the body; thus, it will, in turn, improve the growth of malnutrition rats through the mechanism of albumin level improvement, which later increases growth, particularly through the increase of body weight. Conflict of interest The authors do not have any conflicts of interest regarding the content of the present work.	v3-fos-license
2019-12-19T09:17:33.241Z	2019-12-16T00:00:00.000	210989065	{ "extfieldsofstudy": [ "Medicine", "Psychology" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1016/j.ssmph.2019.100531", "pdf_hash": "14e55519a77a61a4dcff3b2a79060048c7461ee5", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45410", "s2fieldsofstudy": [ "Political Science", "Sociology" ], "sha1": "d37b36c59f2a48bc143414f0a30749f67b03cbd5", "year": 2019 }	pes2o/s2orc	Health and voting over the course of adulthood: Evidence from two British birth cohorts Systematic differences in voter turnout limit the capacity of public institutions to address the needs of under-represented groups. One critical question relates to the role of health as a mechanism driving these inequalities. This study explores the associations of self-rated health (SRH) and limitations in everyday activities with voting over the course of adulthood in the 1958 National Child Development Study and the 1970 British Cohort Study. We used data from participants who reported voting in the last general election at least once between the ages of 23 and 55 in the 1958 cohort and between the ages of 30 and 42 in the 1970 cohort. We examined associations controlling for a range of early-life and adult circumstances using random-effects models. Compared with those in good or better health: those in fair health had 15% and 18% lower odds of voting in the 1958 and 1970 cohorts; those in poor or worse health had 17% and 32% lower odds of voting in the 1958 and 1970 cohorts. These effects varied with age and were most marked among those in poor health at the ages of 23/30 in the 1958 and 1970 cohorts. Controlling for SRH, having limitations in everyday activities was not associated with voting in main models. Examining age-based differences, however, we found that reporting limitations was associated with a higher probability of voting at the age of 55 in the 1958 cohort and at the age of 30 in the 1970 cohort. Building on the qualities of the British birth cohorts, we offer nuanced evidence about the role of health on voting, which involves considerable life-course processes. Future studies need to examine how these findings progress after the age of 55, extend to mental wellbeing and health practices, and contribute to explain social inequalities in voter turnout. Introduction One key characteristic of modern democratic societies lies in the capacity of its citizens to influence politics through voting in events such as general elections. The legitimacy of this process depends on the equal opportunity to vote across all groups, independent of age, gender, race/ ethnicity, family background, and other social characteristics. As a dimension of social capital, the high prevalence and fair distribution of voting is also considered as a determinant of population health (Lantz & Pritchard, 2010). Systematic differences in voter turnout, however, remain relatively common (Smets & van Ham, 2013). Health-related outcomes such as physical disability and mental illness have been consistently found to predict lower voter turnout, with voting rights in these groups already championed by advocacy groups (Kamens, Blum, & Styron, 2019;Matsubayashi & Ueda, 2014;Schur, Adya & Kruse, 2013;Schur, Shields, Kruse, & Schriner, 2002, Schur, Shields, & Schriner, 2005. The broader role of health in voting has not received the same level of interest yet has been gaining traction over the past five years (Gollust & Rahn, 2015;Mattila, S€ oderlund, Wass, & Rapeli, 2013). A critical dimension of this debate concerns whether health represents a potential mechanism reinforcing social inequalities in voting over the life-course (Pacheco & Fletcher, 2014;Rodriguez, Geronimus, Bound, & Dorling, 2015). Gollust and Rahn (2015) argued that differences in voting attributable to health may translate into a loss of political power among the socially disadvantaged groups who are less capable to promote their health, thereby representing a new fundamental cause of health inequalities (Phelan, Link, & Tehranifar, 2010). In this context, this paper seeks to challenge two issues cross-cutting the "health-voting" literature, regarding: 1) the methodological approaches used to assess the robustness of its relationship; 2) the life-course moderators, especially age effects, likely to nuance this relationship over the course of adulthood. Health and voting: towards a causal association? Despite newfound interest in the health-voting relationship, public health research on the association between health and voting may be traced back to over twenty years ago (Blakely, Kennedy, & Kawachi, 2001;Smith, 1997;Smith & Dorling, 1996). Studies in the late nineties explored the associations between area-level voter turnout, party affiliation, and mortality rates in the United Kingdom (UK) and found that regions with higher mortality rates were more likely to abstain from voting and vote for left-wing parties (Smith, 1997;Smith & Dorling, 1996). A few years later, Blakely et al. (2001) found that regions in the United States (US) with higher levels of inequalities in voting were more likely to report a high prevalence of poor self-rated health. This work was inspired in part by the rise in popularity of the application of Robert Putnam's social capital theory in public health, which convened that disinvestments in social arrangements worked alongside income inequality to shape population health (Kawachi, Kennedy, Lochner, & Prothrow-Stith, 1997). Because of the relatively small number of individual-level datasets with data on health and voting, a significant portion of studies in this field is still using aggregate (ecological) designs (Kelleher, Timoney, Friel, & McKeown, 2002, Reitan, 2003Kim & Kawachi, 2006). For instance, recent studies reported that areas in the US where individuals voted for the Republican Party were more likely to present higher mortality rates and poorer health outcomes (Bilal, Knapp, & Cooper, 2018;Bor, 2017;Wasfy, Stewart, & Bhambhani, 2017). To infer causal relationships from these associations, however, ecological designs build on the critical assumption that associations at the aggregate level are equivalent to associations at the individual level, an argument that has misled experts on the prediction of voter turnout in the past (Gelman, Shor, Bafumi, & Park, 2007;Gnaldi, Tomaselli, & Forcina, 2018). This first group of studies has been increasingly complemented by individual-level studies. Since the early 2000s, studies on disability and voting have been using large-scale social surveys to examine this relationship (Matsubayashi & Ueda, 2014;Miller & Powell, 2016;Powell & Johnson, 2019;Schur et al., 2002;Schur & Kruse, 2000). Similarly, over the past decade, studies across Canada, the US, and a majority of European countries have found that poor self-rated health is consistently linked to a lower voter turnout (Goerres, 2007;Mattila et al., 2013;Pacheco & Fletcher, 2014;Couture & Breux, 2017;, Rodriguez, 2018. Some of these studies have also started unpacking the specific conditions (alcoholism, cardiovascular diseases, neurological disorders, co-morbidities) that may inform this association (Gollust & Rahn, 2015;Sund, Lahtinen, Wass, Mattila, & Martikainen, 2017). The vast majority of studies on disability and health, however, have used cross-sectional datasets to assess these associations, precluding us from ruling out issues of temporal ordering and unobserved heterogeneity in most cases. The most recent wave of studies is tackling these issues. Burden, Fletcher, Herd, Jones, and Moynihan (2016) used a sibling fixed-effects design with American older adults around the age of 70 in the US Wisconsin Longitudinal Survey and found that self-rated health had an equivalent effect on voter turnout to educational attainment at this age. Ojeda and Pacheco (2019) followed the voting behaviour of young adult participants across US national elections up to the age of 29 in the 1997 National Longitudinal Survey of Youth. They found that poor self-rated health at the end of adolescence was associated with a lower probability of voting across elections, but that subsequent changes in self-rated health did not further influence vote behaviour during young adulthood. Similarly, Rapeli, Mattila, &Papageorgiou (2018) examined the repeated association between self-rated health and voting across the 1992, 1997, 2001, and 2005 UK general elections in the British Panel Household Survey (BPHS). They found that the probability of voting among those who rated their health very poorly had been on average six percentage points (p.p.) lower compared to those who rated their health as excellent. Health and voting: changes over the life-course? This body of work is consolidating a varied evidence base supporting the robustness of the relationship between health and voting. A second issue, however, concerns the disentanglement of the magnitude of this relationship and its underlying mechanisms at different stages of the lifecourse. A large literature has established the importance of age as one of the most important predictors of voting (Goerres, 2007;Plutzer, 2002;Smets & van Ham, 2013;Strate, Parrish, Elder, & Ford, 1989). In the United Kingdom, those who are aged 65 þ have been on average 32% more likely to vote in a general election compared with those who are aged 18-24 (Dempsey & Loft, 2017). Strate et al. (1989) argued that approximately 50% of this association was explained by increases in knowledge and interest, party affiliation, family income, and social networks over the life-course. Goerres (2007) argued that non-political factors such as stability in residence and marital status also contributed to explain this association. Few studies, however, have distinguished age-based differences in regard to health and voting. Mattila et al. (2013) examined the age-graded association between self-rated health and voter turnout across European countries and found a strong gradient, with the association only becoming marked among participants over the age of 50. Corroborating this finding, studies focused on older age groups tend to find higher magnitudes of association for disability and self-rated health measures (Bazargan, Kang, & Bazargan, 1991;Schur et al., 2002, Burden et al. 2016. There is strong reason to believe that changes in health influence voting through different mechanisms at different ages. Examining the voting practices of young adults across US elections, Plutzer (2002) demonstrated that the determinants of initial voting experiences were fundamentally distinct from those of subsequent voting experiences over the course of adulthood. Building on this theoretical argument, Ojeda (2015, Ojeda & Pacheco, 2019 proposed two sets of mechanisms to understand the role of health on voting over the life-course. Before the entry into adulthood, health is hypothesized to influence the capacity to acquire in one's family and school the resources enabling political participation. These include general cognitive skills, political knowledge and interest, feelings of affiliation to a political party, and the establishment of social networks that may reinforce these resources over time. Once resources and initial voting experiences are acquired, health is then hypothesized to influence voting by disrupting a new set of resources driving the capacity to vote, such as knowledge of one's environment, social networks, and physical functioning. Other life-course processes may shape the relationship between health and voting over time. Period effects, such as the importance or closeness of one election, strongly drive voter turnout (Cancela & Geys, 2016;Frenk, Yang, & Land, 2013). Countries, including Canada, Norway, Denmark, the US, and the UK, have also been facing widening inequalities in voter turnout across generations (Kitanova, 2019;Smets, 2012). Smets and Neundorf (2014) argued that these period and cohort effects could also interact, finding in the US that cohorts were more likely to vote in adulthood if they experienced their initial voting experiences in elections with high turnout rates. It is therefore possible that differences in voting attributable to health at different ages are exacerbated in electoral contexts where the resources enabling the capacity to vote are more likely to matter, that is, when elections are less important and among younger generations who are less likely to vote. Objectives Except for the work among US young adults by Ojeda and Pacheco (2019), no study that we know of has disentangled the relationship between health and voting over the course of adulthood in a longitudinal approach. Building on the developmental approach to lifelong voting, we seek to test the hypotheses that: 1) health is associated with initial voting experiences, 2) beyond initial voting experiences, health continues to be associated with voting in adulthood; 3) health has an increased association with voting as individuals become older. In keeping with a life-course approach, we hypothesize that declines in health have an increasing impact on voter turnout as they cumulatively impact on the resources driving the capacity to vote over the course of adulthood. To examine this, the current study examines the progression of the health-voting association using two British birth cohorts, the 1958 National Child Development Study (NCDS) and the 1970 British Cohort study (BCS), which have already been used to study the determinants of voter turnout (Denny & Doyle, 2005, 2007aDeary, Batty, & Gale, 2008;Finlay & Flanagan, 2013;Persson, 2014). Data We used data from the 1958 NCDS, which recruited 17,415 individuals from birth and followed them up to the ages of 55 in 2013, and the 1970 BCS, which recruited 17,196 individuals from birth and followed them up to the age of 42 in 2012 (Power & Elliott, 2006;Elliott & Shepherd, 2006;Chamberlain et al. 2013, University of London, 2008. The 1958 and 1970 cohorts were initially designed to study perinatal mortality and then progressed to become multidisciplinary, collecting information on health, economic, and social circumstances over time. We used the data on participants who reported on voting in the last general election at least once in all valid waves: in the 1958 cohort, six times at the ages of 23, 32, 42, 46, 50, and 55 (14,031 participants); in the 1970 cohort, four times at the ages of 30, 34, 38, and 42 (12,973 participants). While the 1970 cohort was followed for a relatively shorter period of time, its participants voted in the same elections from 1997 to 2010 as the 1958 cohort, enabling us to consider here potential period and cohort effects. Measures For the dependent variable, participants were asked whether they voted in the last general election (Yes/No), referring: 1) in the 1958 NCDS, to the 1979 (age 23), 1987 (age 32), 1997 (age 42), 2001 (age 46), 2005 (age 50), and 2010 (age 55) general elections; 2) in the 1970 BCS, to the 1997 (age 30), 2001 (age 34), 2005 (age 38), and 2010 (age 42) general elections. The 1979 election represented the first election in which the 1958 cohort could vote while the 1997 election represented the second election in which the 1970 cohort could vote. Missingness on voter turnout was relatively marked in the 1970 cohort at the ages of 38 (21.5%) and 42 (13.3%) because these measures were taken retrospectively at the age of 42 from a questionnaire completed by 89% of participants alongside their main interview. To operationalize health, we chose two indicators consistently measured across survey waves in the 1958 and 1970 cohorts: self-rated health and longstanding limitations. For self-rated health, participants were asked to rate their health using different four-or five-point Likerttype scales across waves, with options ranging from "very poor" to "excellent". We recoded responses into three consistent categories: 1) Good to excellent, 2) Fair, and 3) Poor to very poor. For longstanding limitations, participants were asked through various questions whether they felt limited in their daily activities when they reported having a longstanding illness and/or disability (Yes/No). We chose these indicators to represent different facets of health: in particular, self-rated health may indicate shorter-term problems and limitations may indicate longer-term conditions that require social and material adaptations (Stockemer & Rapp, 2019). Question labels and responses on voting and health measures are presented in Supplementary Table 1. To address confounding, we selected a group of variables that were available in each cohort, consistently measured across waves, and likely to be associated with health and voting (Smets & van Ham, 2013). Characteristics at birth included: 1) gender (using sex as a proxy) (Man/Woman); 2) region; 3) mother's age (continuous); 4) mother's smoking during the pregnancy (Yes/No); 5) mother's weight (continuous); 6) father's Registrar General's social class (I Professional/II Managerial and technical/III Skilled/IV or V Partly-skilled or unskilled/Not applicable); and 7) birth weight (continuous). Characteristics at the ages of 23/30 included: 8) intention to vote in the next election (Yes/No/Do not know); 9) educational attainment (No qualifications/NVQ 1: CSE level/NVQ 2: O level/NVQ 3: A level/NVQ 4: Higher qualification/NVQ 5: Degree). Time-varying characteristics in adulthood included: 10) Registrar General's social class (I/II/III/IV or V/Not applicable), 11) employment status (Employed/Unemployed/Homemaker/Other); 12) parenthood (No children/One child/Two or more children); 13) marital status (Single/Married or partnered/Separated, divorced, or widowed); 14) housing tenure (Main owner/Renter/Other). The Registrar General's scheme allocated people to social classes based on employment and occupation, was found to be generally strongly correlated with income, and has been the most common measure of social class in the UK until the early 2000s (Goldthorpe, 2010, Office for National Statistics, 2019). Statistical analyses Before our main analyses, we modelled inverse-probability nonresponse weights for each wave using participants' circumstances at birth (Hawkes & Plewis, 2006;Mostafa & Wiggins, 2015). British birth cohort members are not rejected when they do not answer in one wave. They are re-contacted at following waves unless they refuse to participate. This means that missingness patterns are non-monotonous. Predictors of non-response included: being a man, having no record of the father's social class, having a lower birth weight, and having a mother who was younger, with a lower weight, and who smoked during the pregnancy. For our main analyses, we examined the adjusted associations between the two health indicators and voting using random-effects (RE) logistic models. One of the strengths of RE models is their ability to derive unbiased estimates of person-specific effects in the presence of non-response during the follow-up period, under the condition that it is missing-completely-at-random (MCAR) (Gibbons, Hedeker, & DuToit, 2010). Since only 40% of the 1958 cohort (n ¼ 5578 out of 14,301) and 52% of the 1970 cohort (n ¼ 6700 out of 12,973) participants answered to each wave during the follow-up period, this approach maximizes the data available for analysis. We did not use fixed-effects (FE) modelling despite its capacity to account for time-invariant confounding because of the high variability required in the outcome. Since only 43% of the 1958 cohort and 30% of the 1970 cohort participants with valid data on voting reported both outcomes (i.e., voting and not voting) over the period of follow-up, we argue that this approach would limit the capacity to detect significant differences, especially in the relatively smaller group of participants who reported poor health. We also note that we were unable to implement a multiple imputation (MI) approach to reduce the impact of datamissing-at-random (MAR) because, given the longitudinal nature of our data and the large number and nominal scale of our variables, we faced well-known irreconcilable convergence issues when testing models (De Silva, Moreno-Betancur, De Livera, Lee, & Simpson, 2017). Our analysis therefore builds on the assumption that there is no systematic difference between observations with complete and incomplete data within waves (De Silva et al., 2017). We first estimated the average association of the health indicators with voting across waves in each cohort. We modelled health indicators and covariates in sequential blocks to assess their contribution: Model 1 -Bivariate, Model 2 -Health variables together, Model 3 -Adding characteristics at birth, Model 4 -Adding characteristics at ages 23/30, Model 5 (Full) -Adding time-varying characteristics. Detailed results from the full models are presented in Supplementary Table 2. We then estimated the potential age-based differences in these associations by entering additional age-based interaction terms with the two health indicators after the full model. During the second step, we also produced marginal probabilities to better interpret differences in voting across waves (Muller & MacLehose, 2014). We tested interactions between the two health indicators and between each indicator and sex, and found no additional statistically significant interactions. We also tested if results varied when considering: 1) the number of waves in which cohort members participated (reproducing analyses in sub-samples of cohort members who participated in at least one, two, three, or more waves in each cohort), 2) non-response weights, and 3) intention to vote at ages 23/30, and obtained consistent results (see Supplementary Table 3). Analyses were produced in each dataset separately using Stata 14 (StataCorp 2015). (Dempsey & Loft, 2017). The estimated prevalence of fair or poor self-rated health varied: 1) in the 1958 cohort, between 10% and 24% between the ages of 23 and 55; 2) in the 1970 cohort, from 15% to 21% between the ages of 30 and 42. The estimated prevalence of limitations in everyday activities varied: 1) in the 1958 cohort, from 2% to 20% between the ages of 23 and 55; 2) in the 1970 cohort, between 9% and 17% between the ages of 30 and 42. Comparing these indicators across cohorts at the age of 42, the 1970 cohort was slightly more likely to be in good or better health (85% versus 82%) yet slightly more likely to report limitations in everyday activities (17% versus 13%). Health and voting over the course of adulthood: main associations The first step was to examine the average associations between health and voting across waves in the 1958 and 1970 cohorts. Supporting our first hypothesis, we found that rating one's health as fair or poor was associated with a lower probability of voting in adulthood. Having limitations in everyday activities, however, was not associated with voting after taking into account self-rated health. The magnitude of associations was overall slightly stronger in the 1970 cohort. Table 2 presents the results of the random-effects models estimating these associations in the two cohorts. In the 1958 cohort, participants who rated their health as fair had 15% lower odds of voting (95%CI 0.77-0.94) and participants who rated their health as poor had 17% lower odds of voting (95%CI 0.69-1.00) compared to those who rated their health as good to excellent. Comparing estimates from Models 1 and 5 on the log odds scale, the magnitude of the estimates related to self-rated health decreased by 48% (fair) and 62% (poor). Reporting limitations in everyday activities was Other covariates associated with voting in the 1958 cohort included: being a woman, being older, having a father in a higher social class, having an older mother, having a higher level of education, intending to vote in the next election, being employed, being married, having no children, and being a home owner (see online supplementary material). In the 1970 cohort, participants who rated their health as fair had 18% lower odds of voting (95%CI 0.72-0.95) and participants who rated their health as poor had 32% lower odds of voting (95%CI 0.52-0.90) compared to those who rated their health as good to excellent. Comparing estimates from Models 1 and 5 on the log odds scale, the magnitude of the estimates related to self-rated health decreased by 51% (fair) and 42% (poor). Reporting limitations in everyday activities was not significantly associated with voting in the final model (OR ¼ 1.07, 95%CI 0.87-1.31). Other covariates associated with voting in the 1970 cohort included: being older, having an older mother, having a higher birth weight, having a higher level of education, intending to vote in the next election, having a higher social class, being married, and being a home owner (see online supplementary material). Health and voting over the course of adulthood: age-graded associations The next step was to investigate potential differences in the associations between the two health indicators and voting at different ages over the course of adulthood. Testing age-based interactions terms with the full models (Model 5) in the 1958 and 1970 cohorts, we found evidence of change in the two cohorts. In the 1958 cohort, using the latest time point (age 55) as the reference category: 1) the interaction term for "poor or worse" self-rated health was significant at the age of 23 (p ¼ .025); 2) the interaction terms for limitations in everyday activities were significant at the ages of 23 (p ¼ .005) and 32 (p ¼ .015). In the 1970 cohort, using the latest time point (age 42) as the reference category: 1) the interaction terms for "fair" (p ¼ .014) and "poor or worse" (p ¼ .027) self-rated health were significant at the age of 34; 2) the interaction term for limitations in everyday activities (p ¼ .013) was significant at the age of 30. Contrasting with our first and second hypotheses, we found that: 1) there was no clear age-graded increase in the relationship between self-rated health and voting in the life period covered in the 1958 and 1970 cohorts; 2) there was a positive association between limitations in everyday activities and voting at the age of 55 in the 1958 cohort and at the age of 30 in the 1970 cohort. Table 3 presents the age-graded marginal probabilities of voting according to health indicator categories across waves in each cohort. In the 1958 cohort, the difference in the probability of voting between those who reported "fair" and "good or better" health varied from þ0.1 p.p. to -3.5 p.p. between the ages of 23-55. The difference in the probability of voting for those who reported "poor or worse" health was most marked at the age of 23 (13.4 p.p.), and returned afterwards to similar levels of differences as found in those in fair health, varying from -0.1 p.p. to -3.0 p.p., between the ages of 32 and 55. Differences in voting according to limitations in everyday activities changed direction with age, starting from a disadvantage of -5.7 p.p. at the age of 23 to an advantage of þ3.3 p.p. at the age of 55. In the 1970 cohort, the difference in the probability of voting Estimates represent odds ratios (OR) from weighted random-effects logistic models. Bolded estimates are statistically significant at the p < .05 level. Models 1-5 controlled for age. Model 3 included: gender, region, mother's age, mother's smoking, mother's weight, father's social class, birth weight. Model 4 included: intention to vote in the next election, educational attainment. Model 5 included: social class, employment status, parenthood, marital status, housing tenure. Estimates are marginal probabilities from the full models (Model 5) presented in Table 2, with age-based interactions. Interaction terms for self-rated health and limitations in everyday activities were entered separately. Differences between bolded estimates are statistically significant at the p < .05 level. between those who reported "fair" and "good or better" health varied from 0.0 p.p. to -5.1 p.p. between the ages of 30-42. The differences in the probability of voting for those who reported "poor or worse" health were marked both at the ages of 30 (-8.5 p.p.) and 42 (-8.7 p.p.). Differences in voting according to limitations in everyday activities also followed an age-based pattern, starting from a benefit of þ4.5 p.p. at the age of 30 to a disadvantage of -1.2 p.p. at the age of 42. Discussion Our study sought to corroborate the role of health on voting over the course of adulthood, making use of two British birth cohort studies spanning three decades of life. Supporting our first hypothesis and the bulk of the studies that have examined this question, above the influence of family background and adult achievements, we found a significant negative association with self-rated health, but not with health limitations in everyday activities, and voting in the 1958 and 1970 cohorts. Comparing the magnitude of its effect with other covariates, we find that the importance of self-rated health was comparable to the predictive power of social class, parenthood, marriage, and employment, supporting the argument that health may be more important for understanding political behaviour than previously believed (Mattila et al., 2013, Gollust & Rahn, 2015. Returning to our first objective, we found that the health-voting relationship was robust to a number of circumstances at birth, young adulthood, and midlife. While some early-life circumstances (father's social class, mother's age, birth weight) were significantly associated with voting, they did not appear to influence the magnitude of the health-voting relationship. Between 42% and 62% of this association, however, was attenuated when considering intention to vote, educational attainment, employment, marital status, parenthood, and home ownership over the course of adulthood. Additionally, our analytic approach did not explicitly disentangle the nature of the relationships between self-rated health, other time-varying characteristics, and voting. While poor health is considered to be the result of unfavourable circumstances accumulating over the life-course, it also reinforces disadvantage over the life-course. Poor self-rated health may therefore influence voting by limiting the opportunities to secure relationships and home ownership, which are positively linked with voter turnout (Franke & Kulu, 2017;Smith, 2012). This selection effect would lead us to under-estimate the full size of the health-voting association. Regarding our second objective, improving on other studies with the use of an age-graded analysis in two cohorts that experienced the same elections, we were able to highlight a considerable variation in the magnitude of the association between self-rated health and voting across waves and cohorts. We found that this association was strongest in the first election captured at the age 23 wave in the 1958 cohort and overall slightly stronger in the 1970 cohort. This supports the argument that there are likely to be multiple life-course moderators nuancing this relationship. Following the work of Plutzer (2002) and Ojeda and Pacheco (2019), we found in the 1958 cohort that poor health was most likely to influence voting among participants during their first experience voting in a general election in young adulthood. Similar results found in the second experience voting in a general election (captured at the age 30 wave) in the 1970 cohort support the important role that poor health may play during this life period. In comparison, the association between health and voting was absent during the third opportunity to vote in a general election in both cohorts (captured at age 32 in the 1958 cohort and at age 34 in the 1970 cohort). This suggests that health may be important in shaping voting habits during young adulthood yet decreasingly so in subsequent elections, supporting the findings by Ojeda and Pacheco (2019) in the US. Supporting this in the context of general elections, Smets and Neundorf (2014) found that the first two elections young people could participate in were the most critical in shaping their long-term voting trajectories. Since the voting habit is one of the strongest predictors of voting in future elections, a lower propensity to vote attributable to health may disproportionally lead unhealthy young adults into long-term trajectories of disengagement (Denny & Doyle, 2009). New studies will need to confirm whether these findings are consistent across other instances of voting (local, regional, European) that young adults encounter alongside national elections. Beyond age, we found little evidence supporting substantial period effects when comparing results across elections. For instance, while health was a relatively distant second priority in the 2010 UK election in keeping with the high level of satisfaction with the UK health care system, our estimates suggest that those in poor health were only 2% less likely to vote if they were in the 1958 cohort yet 12% less likely to vote if they were in the 1970 cohort (King's Fund 2010). We found, however, more evidence leading to the consideration of a cohort effect as participants reported a relatively slightly lower propensity to vote if they were in poor health in the 1970 cohort. This finding contributes to the argument that voter turnout in the younger generations that demonstrate political disengagement and dissatisfaction in politics (as was the 1970 cohort around the 2001 UK election) may be shaped to a higher degree by the individual resources driving the capacity to vote, including health (Dempsey & Loft, 2017). Finally, we return to the relatively surprising finding that, controlling for self-rated health and other covariates, reporting limitations in everyday activities was positively associated with voting in the third (1970 cohort) and fifth decades (1958 cohort) of life. It is possible that, compared to self-rated health, the measure of health limitations taps to a lesser extent into the dimensions of health that has a strong influence on the capacity to vote such as physical vitality (Cohen, Forbes, & Garraway, 1995;Mavaddat et al., 2011). Stockemer and Rapp (2019) used a similar measure of health limitations in the European Social Survey and found that it was associated with a lower probability of voting but a higher probability of contacting a politician, signing a petition, wearing a campaign badge, and boycotting in the past year. Two studies in the US and Finland also found that, while many chronic conditions were associated with a decreased voter turnout, individuals with cancer had a higher probability of voting (Gollust & Rahn, 2015;Sund et al., 2017). Finding that finances, physical vitality, and social networks were unlikely to mediate this association, Gollust and Rahn (2015) hypothesized that higher voter turnout could be related to the development of an activist identity related to health conditions, leading those who strongly associate with it to further engage in political processes. One key issue in interpreting this finding resides in our limited ability to unpack the nature of health limitations. In particular, this measure may disproportionally include individuals with milder forms of disability. Studies on disability have consistently found that: 1) associations between disability and voting remain negative over time, 2) they are robust to the nature of the limitation measured (e.g., hearing, visual, cognitive, mobility), and 3) they are age-graded and most marked among older adults (Schur et al. 2002, Matsubayashi & Ueda, 2014. Future studies will need to unpack which limiting conditions are likely to impact voting and whether these fit with an identity hypothesis (Gollust & Rahn, 2015). Strengths and limitations This study builds on the qualities of the British birth cohorts to produce representative estimates of the role of health on voting in adulthood across two generations in the United Kingdom. This includes the capacity to use a large set of variables at different life-course stages to account for participants' family background and changing social circumstances. We highlight a few limitations. First, concerns about unobserved heterogeneity remain plausible, precluding us from interpreting causal relations. We tested modelling additional measures of limitations at the age of 16 and political interest in adult waves, and found that they did not influence our findings. Second, we did not use an analytic strategy to adjust for item-level missingness due to issues of non-convergence (De Silva et al., 2017). Non-random missingness may therefore bias our findings. Third, differences in the measurement of health indicators over the course of the cohorts led us to use fewer categories, which have underestimated differences between extreme categories. Finally, the 1958 and 1970 cohorts were not designed to study voter turnout and waves have been administered years after the last general election. Our findings build on the assumption that variables were valid proxies of the circumstances experienced when the elections occurred. Conclusion Voting is a central element of social cohesion and democracy that is shaped by health. This relationship is not straightforward and may include both the lower propensity to vote with poorer health and the higher propensity to vote when health limits everyday activities. These relationships emerge at different ages, with some of the strongest associations likely to be found during habit formation in young adulthood. Contrasting with the current generation, young adults in poor health today may be even less likely to vote compared to the 1958 and 1970 cohorts as they further delay their transition to adulthood and their participation in politics (Smets, 2016). The findings support the argument that health impacts voting behaviour, and thus elections, long before it may lead to death (Rodriguez, 2018;Rodriguez et al., 2015). The findings also support the argument that those in poor health are likely to suffer a lack of political power over their life-course, which in turn contribute to the reproduction of social inequalities in voting and health (Gollust & Rahn, 2015;Pacheco & Fletcher, 2014). Next research steps include disentangling this association beyond the age of 55 in a longitudinal setting, extending its study to other dimensions of health such as mental wellbeing, and formally testing its contribution in socioeconomic inequalities in voting. Ethics statement There was no need to obtain ethics approval for the secondary data analysis of publicly available data in this study. For the 1958 cohort, ethical approval for 1958-1991 and 2004	v3-fos-license
2021-02-19T14:59:47.394Z	2021-02-18T00:00:00.000	231958106	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://advancesinrheumatology.biomedcentral.com/track/pdf/10.1186/s42358-021-00163-x", "pdf_hash": "c9e42a67fd9cc921de4031471bff172066d4ae27", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45411", "s2fieldsofstudy": [ "Medicine" ], "sha1": "c9e42a67fd9cc921de4031471bff172066d4ae27", "year": 2021 }	pes2o/s2orc	Correlating cognition and cortical excitability with pain in fibromyalgia: a case control study Fibromyalgia is a chronic pain disorder characterized by widespread musculoskeletal symptoms, primarily attributed to sensitization of somatosensory system carrying pain. Few reports have investigated the impact of fibromyalgia symptoms on cognition, corticomotor excitability, sleepiness, and the sleep quality — all of which can deteriorate the quality of life in fibromyalgia. However, the existing reports are underpowered and have conflicting directions of findings, limiting their generalizability. Therefore, the present study was designed to compare measures of cognition, corticomotor excitability, sleepiness, and sleep quality using standardized instruments in the recruited patients of fibromyalgia with pain-free controls. Diagnosed cases of fibromyalgia were recruited from the Rheumatology department for the cross-sectional, case-control study. Cognition (Mini-Mental State Examination, Stroop color-word task), corticomotor excitability (Resting motor threshold, Motor evoked potential amplitude), daytime sleepiness (Epworth sleepiness scale), and sleep quality (Pittsburgh sleep quality index) were studied according to the standard procedure. Thirty-four patients of fibromyalgia and 30 pain-free controls were recruited for the study. Patients of fibromyalgia showed decreased cognitive scores (p = 0.05), lowered accuracy in Stroop color-word task (for color: 0.02, for word: 0.01), and prolonged reaction time (< 0.01, < 0.01). Excessive daytime sleepiness in patients were found (< 0.01) and worsened sleep quality (< 0.01) were found. Parameters of corticomotor excitability were comparable between patients of fibromyalgia and pain-free controls. Patients of fibromyalgia made more errors, had significantly increased reaction time for cognitive tasks, marked daytime sleepiness, and impaired quality of sleep. Future treatment strategies may include cognitive deficits and sleep disturbances as an integral part of fibromyalgia management. Introduction Fibromyalgia syndrome is a common neurological condition that is characterized by widespread musculoskeletal pain and the presence of typical tender points throughout the body. The prevalence has been estimated to be up to 1.75% in general population [1] and remains one of the common reasons for rheumatological referrals [2]. Despite many decades of research, the diagnosis and treatment of fibromyalgia remains a contentious topic. At its core, fibromyalgia is characterized by a heightened pain response to non-nociceptive (allodynia) or nociceptive stimuli (hyperalgesia) [3]. In addition to increased pain, more than 70% of patients of fibromyalgia present a complex symptomatology comprising of fatigue [4], cognitive problems [5,6], and sleep disturbances [7]. In recognition of their importance, the latest definition of fibromyalgia has undergone major revisions [8,9]. The driving force of the revision of definition criteria was to broaden the symptom range of patients of fibromyalgia and ease the diagnostic process at the clinical level [10]. From a research point of view, however, the complex symptomatology brings in a few complications: firstly, the symptom profile has become too varied, especially in terms of the nature and extent of comorbidities; secondly, it can make the management strategy very complicated. An ongoing experience of pain symptoms is traditionally believed to interfere in the cognitive abilities by limiting the neuronal circuitries available to carry out cognitive functions, especially for attention, psychomotor speed, inhibition, and interference. Thus, a progressive decline in cognitive functions in patients of chronic pain is a logical assumption [11]. Mini-Mental State Examination is a common instrument for assessment of mild cognitive impairments [12]. Stroop test evaluates various components of cognition such as processing speed, selective/task attention, automaticity, and parallel processing [13]. Closely related to musculoskeletal pain is the poor quality of sleep. In fact, selfreported unrefreshed sleep is one of the oldest reported symptoms of fibromyalgia. It is believed that the inability to sleep would lead to daytime excessive sleepiness and easy fatiguability. Several attempts have been made to document cognitive decline [5,6,[14][15][16][17][18][19] and disrupted sleep [20][21][22][23][24][25][26]; however, the cognitive decline in relation to sleep has not been extensively studied yet. People with chronic pain often adopt a sedentary lifestyle resulting in a decrease in muscular strength and dysfunction due to the restriction of activities [27,28]. In addition to myographic changes [29], recent evidence suggests altered structure [30][31][32][33][34] and function [35][36][37] of the motor system in patients of fibromyalgia. Under laboratory conditions, parameters of corticomotor excitability using transcranial magnetic stimulation can be used to quantify changes in the excitability of corticomotor and corticospinal tracts [38]. Few reports have assessed corticomotor excitability in patients of fibromyalgia [39][40][41][42] with conflicting results [43]. The present report is a part of a larger project to investigate the role of putative treatment options for the managmenet of fibromyalgia. However, given the current problem of varied symptomatology, it is imperative to conduct an investigation of the co-morbidities and the relation they share with pain symptoms prior to a clinical trial. Study design It was a cross-sectional case-control study conducted in a tertiary health care centre. Diagnosed patients of fibromyalgia were recruited according to the latest definition, based on 2010/2011 criteria with modifications suggested in 2016, given by the American college of rheumatology [8,9]. Age-and sex-matched pain-free controls were recruited from following sources: attendees & care-givers of the patients (genetically unrelated), and other hospital staff (unrelated to the study). Participant recruitment Right-handed female participants between 18 to 65 years were recruited after taking their medical history. Patients of fibromyalgia and pain-free controls were excluded if they had other chronic diseases such as major psychiatric disorders, autoimmune diseases, rheumatic disease; or were contraindicated to transcranial magnetic stimulation (metallic implants, pacemaker, history of substance abuse, epileptic seizure, head injury or brain surgery, pregnancy, use of antidepressants or anticonvulsants) [44]. All tests were conducted between 09:30 am to 11:30 am in a dedicated assessment laboratory at the same health care centre. The participants were asked to refrain from taking analgesics and other neuro-active substances (caffeine, nicotine, or alcohol) 24 h prior to the investigations. Since hormonal fluctuations could affect these outcomes, the tests were scheduled between the 18 th to 23 rd days of the menstrual cycle, keeping start of the last menstrual period as the 1 st day. Pain characteristics were recorded as intensity of pain (using 11-pt Numerical pain rating scale), number of tender points (using manual palpation), and detailed evaluation of pain perception (using McGill pain questionnaire). Methodological details for pain assessments have been detailed in a recent paper from our group [45]. Cognitive assessment i) Mini-Mental State Examination; was used to screen mild to moderate impairments in cognitive functions [12]. The vernacular version allows cognitive screening instrument of people of wide literacy and age range [46]. It has 19-items that have a binary response -'Yes' or 'No'. One mark was given for every correct and zero for an incorrect response resulting in a maximum score of 30. It is a simple bedside test with capacity to examine cognitive functions including attention, registration, orientation, recall, and calculation. ii) Stroop color-word test; was a cognitive parameter that tests for attention span and task execution [13]. The task was designed and adminstered using SuperLab software (version 5.1; Cedrus Corporation, Arizona, USA) on the computer screen [47]. The initial phase of the test acclimatized the subjects to the monitor, response keys, pattern of the task. It also helped us evaluate adequate color perception, and comprehension of the word. Congruent stimuli were words that meant the same as the color of the font, for example, RED written in red color. While in the incongruent stimuli, the two aspects were mismatched. The principal of the task is that incongruent stimuli generates conflict that is then modulated and resolved by top-down cognitive control. During the test phase, the congruent stimuli were presented interleaved with incongruent stimuli. Participants were asked to correspond to font color (in one block), or word comprehension (in another block) with instructions displayed before each block. The following parameters were recorded for color-word blocks: (a) Accuracy; percentage of correct responses was calculated as the number of correct responses averaged across three trials; (b) Reaction time; average time taken to respond correctly. Corticomotor excitability Participants were made to sit on a chair and asked to relax as much as possible with the arm muscles adequately supported with cushions. Scalp landmarks were used to tentatively determine the most excitable region on the scalp overlying the representative area on the motor cortex for the thumb muscles, also known as the 'motor hotspot'. A figure-of-8 coil (70 mm) transcranial magnetic stimulation (NeuroSoft Neuro-MS/D magnetic stimulator, Ivanovo, Russia; max 4 T) coil was placed such that antero-posteriorly and navigated in small steps to locate the actual 'hotspot' of the motor cortex. The point was marked with a pen to ensure the coil position and orientation throughout the experiment. Biphasic pulses were delivered at an interval of 5-7 s [48][49][50]. (i) Resting motor threshold; was defined as the minimum stimulus intensity that elicited an evoked potential with a peak-to-peak amplitude of at least 50 μV in at least five out of ten trials, while maintaining quiescence on the electromyography. Values have been expressed as a percentage of maximum stimulator output. (ii) Motor evoked potential; was identified as the peakto-peak amplitude elicited by averaging ten consecutive stimuli [51] acquired at 100% resting motor threshold. Acquisition window for each trial had a width of 100 msec pre-stimulus and 200 msec poststimulus. Sleepiness Sleepiness was measured using Epworth sleepiness scale in subjects. The questionnaire comprised of situations commonly encountered in daily life situations and had four responses to each item. The investigator gave a score to the option wherein, '0' implies 'would never doze off', and '3' implies 'high chances of dozing', resulting in scoring ranging from 0 to 24. Total scores of Epworth sleepiness scale were obtained by summing the all the responses. It is well established and validated score in Indian population [52]. Sleep quality Pittsburgh sleep quality index was used to assess sleep quality and disturbances during the month preceding the evaluation. There were 19 individual items, each with four options, in increasing order of severity. Additionally, various aspects of sleep quality like subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleeping medications, and daytime dysfunction have also been reported. Each item is scored on a 0-3 interval scale. The global Pittsburgh sleep quality index score is then calculated by summating the seven component scores, providing an overall score ranging from 0 to 21, where lower scores denote a healthier sleep quality [53]. It also validated in the Indian population with an excellent internal consistency and test-retest reliability [54]. Statistical analysis Data summary has been presented as the proportions, mean ± standard deviation, or median (q1 -q3). Patients of fibromyalgia and pain-free control groups were compared using Chi-squared goodness-of-fit test, unpaired Student's t-tests, or Mann-Whitney U-rank tests, as appropriate. The effect size has been presented as partial eta squared (η2) and interpretated as: less than or equal to 0.003 as no effect, 0.10 to 0.039 as small effect, 0.060 to 0.110 as intermediate effect, and 0.140 to 0.200 as large effect [56] Spearman's rank test was used for correlation of the independent variables. Strength of association was interpreted as: less than 0.1 as a small association, less or equal to 0.3 as a medium association, less than or equal to 0.5 as a large association [55]. P-value less than 0.05 was considered statistically significant, and all tests were performed using GraphPad Prism version 8.00 for Mac OS (GraphPad Software, California US). Results Thirty-four patients of fibromyalgia and 30 pain-free controls were recruited for the study, Fig. 1 shows the details of the recruitment. The mean age was 38.6 ± 9.0 with 80% population being urban (all of controls and 21 of patients). Three-fourths of the population belonged to a middle or higher-middle socioeconomic stratum. All participants were literate, with 26% of the subjects being graduates (9 controls and 8 patients) and 29% being post-graduates (11 controls and 8 patients). No differences were found between patients of fibromyalgia and controls. Table 1 shows the sociodemographic features of the participants. On average, patients of fibromyalgia reported moderate to severe pain 7.4 ± 1.3 (range 5 to 9) on the 11point Numerical pain rating scale. The time since the onset of symptoms was 6.8 ± 2.9 years. The total score of McGill pain questionnaire was 32.6 ± 5.3 (range 25 to 48). Clinical examination showed that they had an average of 10.5 ± 2.0 tender points (range 8 to 14). The most common sites of tender points were lower back and calf, forearm and elbow regions. Few patients also reported mild headaches intermittently throughout the day; each bout of pain was felt as dull, boring and aching that often caused the patient to feel extremely tired. The symptoms were aggravated by prolonged sitting, emotional distress, and cold weather. Some degree of relief was felt after using analgesic drugs, local pressure or bandage, and sprays were used to reduce the pain. Corticomotor excitability Most participants were able to tolerate the transcranial magnetic stimulation pulses comfortably. Resting motor threshold and motor evoked potential amplitudes p value = 0.111, U-statistic = 395, z-score = 4.069, η 2 = 0.037, small effect) were found to be statistically comparable between patients of fibromyalgia and pain-free controls. Few participants reported mild headache lasting up to 24 h after the test. No transcranial magnetic stimulation-induced adverse effects were noted. Table 2 shows a summary of Table 2 Parameters of cognition, corticomotor excitability, and sleep parameters compared between participants with fibromyalgia and controls the comparisons between the pain-free controls and patients of fibromyalgia. Correlation No significant correlations were found between pain symptoms (intensity, duration, tender points) and other outcome variables (cognition, corticomotor excitability, or sleep). However, total scores of Mini-Mental State Examination showed a negative correlation with the total score of Pittsburgh sleep quality index (r = − 0.37, p = 0.029). Also, reaction time for word task in Stroop color-word task was negatively correlated with sleep duration (r = − 0.36, p = 0.038) and sleep disturbance (r = − 0.35, p = 0.045) components of Pittsburgh sleep quality index. All three associations were medium in strength. Discussion The present study was aimed at investigating the paininduced symptoms of cognition, corticomotor excitability, sleepiness, and sleep quality in the fibromyalgia patients compared to pain-free controls. Cognition and sleep quality were found to be negatively correlated with each other. No differences were found for the parameters of corticomotor excitability. The Mini-Mental State Examination indicated that patients of fibromyalgia have mild cognitive impairments; also, the participants performed poorly in all parameters of the Stroop Test. The results show that self-reported cognitive decline (the Mini-Mental State Examination scores) are in concordance with the objective test, the Stroop Test. The study is in line with case-control studies [14] and previous meta-analyses of cognitive deficits in fibromyalgia patients [5,6]. Imaging studies reported that anterior cingulate cortex and dorsolateral prefrontal cortex are among the main areas involved. Decision making and response evaluation are specific domains controlled by dorsal anterior cingulate cortex. These areas are also reported for pain perception and modulation. Anterior cingulate cortex has vital role in various behavioral responses such as processing of attention and anticipation of pain [56]. This activity may correspond to shared neural pain representations that allow deduction of the affective state of the pain and facilitating empathy [57]. Dorsolateral prefrontal cortex has shown activation, when resolving conflicts and catching errors, also it assists in memory and other executive functions, so ultimately affects cognition. Dorsolateral prefrontal cortex is involved in encoding and modulating acute pain perception. Elevated high-frequency oscillatory activities in the dorsolateral prefrontal cortex and orbitofrontal cortex correspond with higher affective pain scores, cognitive, and emotional modulation of pain in fibromyalgia patients. Structural brain changes in patients of fibromyalgia, indexed using voxel-based morphometry, found that central sensitization is correlated with reduced gray matter volume in specific brain regions such as anterior cingulate cortex and prefrontal cortex [32,33]. Decreased scores in Stroop test indicates that fibromyalgia symptoms could affected the functioning of anterior cingulate cortex and dorsolateral prefrontal cortex. No significant changes in corticomotor excitability parameters like resting motor threshold and motor evoked potential amplitude were found in patients of fibromyalgia. This is in contradiction to existing literature [39][40][41][42]. The discrepancies in the findings may be attributed to pain characteristics such as chronicity of pain. Latest studies have shown that the time since onset of painful disease could influence the changes in the corticomotor excitability [58]. Another source of dissimilarity could be the methodological differences. Salerno et al., [39] defined resting motor threshold as at least 100 μV amplitude in at least 50% of trials, whereas we used 50 μV cut-off as per the latest guidelines for recording corticomotor excitability by International Federation of Clinical Neurophysiology [49]. Another difference was the site of recording taken for our study is the abductor policies brevis, while the earlier studies used first dorsal interossei [39][40][41] or flexor muscle of forearm [42]. Although it is still unclear if using a different muscle group of the hand recording can contribute to the differences in findings [43] or the differences in the patient cohort being tested is the main cause of discrepancy. Currently available pharmacotherapy for fibromyalgia includes pregabalin, duloxetine, milnacipran, amitriptyline, cyclobenzaprine and tramadol but remain far from satisfactory [59] In the same way, the present cohort of the patients was also allowed to continue the recommended drug as a part of the standard of care [45] (see supplementary file 1). The rationale is based on the fact that these drugs have demonstrated a moderate-tolarge effect size for pain symptoms and somewhat on sleep quality, at least in the short term. Evidence suggests that long-term use of neuro-active drugs may induce cognitive deficits. A decline in cognitive ability has also been found to result in interference in daily activities, low coping, and poorer treatment outcomes. If the evidence can be extended to patients of fibromyalgia , then the current management may compound the issue at hand [59][60][61]. But cognitive abilities as outcome measures of the randomized controlled trials are currently missing from the literature. Going forward, an ideal treatment strategy must address pain symptoms and, at least to some degree, the associated co-morbidities. Limitations The findings of the study only pertain to middle-aged, regularly menstruating females with fibromyalgia who were refractory to standard of care regime. Factors like age and past treatments can affect pain perception, cognition, and sleep quality. Thus, more studies in patients of fibromyalgia are required to see the generalizability of our findings. Conclusions The study confirms a decline in cognitive function and sleep quality in patients of fibromyalgia. Most of the current treatments for fibromyalgia may have some effect on pain perception and to some extent, sleep quality. Introducing adjunct therapies that can help overcome issues of refractoriness and cognitive decline may have better outcomes. More research into treatments that address cognitive deficits vis-à-vis pain and other symptoms need to be conducted.	v3-fos-license
2017-03-31T08:35:36.427Z	2017-03-10T00:00:00.000	14079529	{ "extfieldsofstudy": [ "Biology" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1999-4907/8/3/75/pdf?version=1489369175", "pdf_hash": "1d2ea744535981f03b0ef0c5ed1780f3f5ce3d35", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45412", "s2fieldsofstudy": [ "Environmental Science" ], "sha1": "1d2ea744535981f03b0ef0c5ed1780f3f5ce3d35", "year": 2017 }	pes2o/s2orc	Partitioning Forest-Floor Respiration into Source Based Emissions in a Boreal Forested Bog : Responses to Experimental Drought Northern peatlands store globally significant amounts of soil carbon that could be released to the atmosphere under drier conditions induced by climate change. We measured forest floor respiration (RFF) at hummocks and hollows in a treed boreal bog in Alberta, Canada and partitioned the flux into aboveground forest floor autotrophic, belowground forest floor autotrophic, belowground tree respiration, and heterotrophic respiration using a series of clipping and trenching experiments. These fluxes were compared to those measured at sites within the same bog where water-table (WT) was drawn down for 2 and 12 years. Experimental WT drawdown significantly increased RFF with greater increases at hummocks than hollows. Greater RFF was largely driven by increased autotrophic respiration driven by increased growth of trees and shrubs in response to drier conditions; heterotrophic respiration accounted for a declining proportion of RFF with time since drainage. Heterotrophic respiration was increased at hollows, suggesting that soil carbon may be lost from these sites in response to climate change induced drying. Overall, although WT drawdown increased RFF, the substantial contribution of autotrophic respiration to RFF suggests that peat carbon stocks are unlikely to be rapidly destabilized by drying conditions. Introduction Peatlands contain one of the largest terrestrial carbon (C) stocks, estimated at ~600 Gt C [1], with northern peatland C storage accounting for ~390-440 Gt [1,2].The large C stock has been accumulated as a result of only a marginal difference, over millennia, between photosynthetic C uptake and loss of C as ecosystem respiration, methane (CH 4 ) emissions, and water-borne outflows [3].The stored C is present in the form of highly mineralizable organic C [4,5] protected in water-saturated, anoxic conditions and is highly sensitive to warmer and drier climate [5][6][7].Therefore, any increase in carbon dioxide (CO 2 ) emissions in response to the expected changes in climate has the potential to provide a positive feedback to global warming [4,[7][8][9].In general, many northern peatlands are expected to be drier under future climates [10,11], and while the response of peatland ecosystem respiration to water-table drawdown has been extensively studied [12][13][14][15][16][17], controlled field experimentation for partitioning ecosystem respiration into its source-based major components remains largely unexplored [18].Research is needed to investigate source-based respiration fluxes in relation to potential changes in environmental conditions to improve our understanding of changes in ecosystem C storage or emissions to the atmosphere under climate change scenarios [19,20]. Ecosystem respiration includes the emission of CO 2 to the atmosphere from above and belowground parts of vegetation (autotrophic respiration; R A ), and from microbial decomposition of the soil organic matter including litter (heterotrophic respiration; R H ). The aboveground parts mainly include plant leaves and stems while the belowground parts comprise living roots with their associated mycorrhizal fungi and microbial populations [21,22].Many northern peatlands are treed [23,24].When respiration is measured at the ground layer of a treed peatland, this forest floor respiration (R FF ) includes respiration associated with tree roots, while respiration of the aboveground tree biomass is excluded.This research partitioned the bulk R FF into source-based above and belowground respiration flux components as: (1) forest floor aboveground autotrophic respiration (R FF_A_ag ) and, (2) belowground shrub + herb (roots) autotrophic respiration (R A_SH_bg ) and belowground tree (roots) autotrophic respiration (R A_T_bg ).Separating the R A_SH_bg from R A_T_bg is important as specific vascular plant functional types may respire at different rates [25,26] due to the difference in their respective above and belowground productivities [27], and therefore modify the response of R FF to changes in water-table (WT) level and soil temperature at 5 cm depth (T 5 ) [4,28]. Peatland respiration flux components are highly responsive to environmental changes such as WT level [26,[29][30][31][32], T 5 [15,33,34], and plant functional (shrubs + herbs, trees) type [35][36][37][38].Therefore, increased atmospheric or soil temperatures and subsequent WT lowering [39,40] may increase soil respiration rates [22] and ultimately alter the peatland C sink or source strength ( [37] resulting from increasing atmospheric CO 2 [41].However, increases in R FF alone do not necessarily indicate a loss of soil C if only autotrophic respiration increases, illustrating the importance of determining the source of respiration and the relative response of each component to changing environmental conditions [42]. Ericaceous shrubs (e.g., Rhododendron groenlandicum) are important contributors to ecosystem productivity in many northern bogs [43,44], also making important contributions to ecosystem respiration.The contribution of shrub autotrophic respiration to R FF in a shrub-dominated bog in Patuanak, Saskatchewan, Canada was estimated to be ~75% [45].A median value of root: shoot ratios estimated from 14 sites in boreal forest was found to be 0.39 [46], suggesting that root respiration is likely to make a significant contribution to measured autotrophic respiration.Overall, root/rhizospheric respiration has been found to account for between 10% and 90% of R FF depending upon vegetation type and season of year [28,47].As similar controls apply to the above and belowground productivities of shrubs [48,49], therefore, the response of above and belowground respiration to environmental change should be similar [46]. Experimental partitioning of soil autotrophic and heterotrophic respiration components has been attempted using different methods with varying results.For example, methods include application of stable isotopes [50,51], root biomass regression with R FF to determine belowground root respiration [52], and comparison of soils with and without root exclusion to determine tree root respiration [17,28]; however, evaluating the responses of various respiration sources (R FF_A_ag , R A_SH_bg , R A_T_bg , and R H ) to environmental change has not been completed.Moreover, responses may vary between microforms (hummocks and hollows) due to initial differences in WT level, soil properties, and vegetation community [16,17,32].Therefore, this study focused on partitioning R FF emissions along a microtopographic gradients in order to evaluate responses of each respiration component to short (2 years) and longer term (12 years) water-table drawdown. Evaluate differences in source contributions to R FF along a microtopographic (hummock, hollow) gradient in a boreal forested bog and how these contributions changed in response to WT treatments, and 3. Assess the respiration components' responses to the WT and soil temperature at 5 cm depth (T 5 ) over one growing season. We hypothesized that experimental WT lowering would lead to increases in all respiration components, with greatest increases at hollows.We also hypothesized that increases in R H would be greatest at hollows, while hummocks would have greater increases in autotrophic respiration (both above and belowground) and that all components of R FF would have significant positive correlations with depth to WT and T 5 . Sites Description and Experimental Design During the growing season (1 May to 31 October) of 2012, this research was conducted in a forested bog within the southern boreal forest and near the town of Wandering River, Alberta, Canada.Based on 30-year (1981-2010) averages, the mean growing season (May to October) temperature and precipitation for this region are 11.7 • C and 382 mm, respectively [53].The mean growing season air temperature and precipitation measured during the 2012 study using a meteorological station installed at the study sites were 13.2 • C and 282 mm, respectively. Within the dry ombrotrophic forested bog, three research sites, CONTROL (55 28 W), were chosen or created (Figure 1).The control was an undisturbed site with a mean WT level of −38 cm whereas the experimental site was created adjacent to the control by ditching around, lowering the mean WT level to 78 cm below surface.One year prior to this study, WT level (± Standard Deviation (SD)) at the control (−56 ± 22 cm) and experimental (−57 ± 20 cm) sites were not different (negative values denote belowground WT; ANOVA, F 1, 5 = 0.55, p = 0.492).The drained site (part of the same bog and located 9 km to southwest) was drained inadvertently 12 years prior to the study as a result of a peat harvesting preparation on an adjacent section, and had a mean WT level of ~−120 cm in 2012 (Figure 2).We hypothesized that experimental WT lowering would lead to increases in all respiration components, with greatest increases at hollows.We also hypothesized that increases in RH would be greatest at hollows, while hummocks would have greater increases in autotrophic respiration (both above and belowground) and that all components of RFF would have significant positive correlations with depth to WT and T5. Sites Description and Experimental Design During the growing season (1 May to 31 October) of 2012, this research was conducted in a forested bog within the southern boreal forest and near the town of Wandering River, Alberta, Canada.Based on 30-year (1981-2010) averages, the mean growing season (May to October) temperature and precipitation for this region are 11.7 °C and 382 mm, respectively [53].The mean growing season air temperature and precipitation measured during the 2012 study using a meteorological station installed at the study sites were 13.2 °C and 282 mm, respectively. Within the dry ombrotrophic forested bog, three research sites, CONTROL (55°21′ N, 112°31′ W), EXPERIMENTAL (55°21′ N, 112°31′ W), and DRAINED (55°16′ N, 112°28′ W), were chosen or created (Figure 1).The control was an undisturbed site with a mean WT level of −38 cm whereas the experimental site was created adjacent to the control by ditching around, lowering the mean WT level to 78 cm below surface.One year prior to this study, WT level (± Standard Deviation (SD)) at the control (−56 ± 22 cm) and experimental (−57 ± 20 cm) sites were not different (negative values denote belowground WT; ANOVA, F1, 5 = 0.55, p = 0.492).The drained site (part of the same bog and located 9 km to southwest) was drained inadvertently 12 years prior to the study as a result of a peat harvesting preparation on an adjacent section, and had a mean WT level of ~−120 cm in 2012 (Figure 2).Based on plant species indicators, the studied bog was classified as a forested low shrub bog [55] with two distinct microtopographic features: hummock and hollow.One year prior to this study, the control and experimental site microforms had equal coverage of mosses with sparse shrubs, whereas the drained hummocks had the highest coverage of shrubs and the drained hollows had the greatest coverage of lichens [44].At all sites, black spruce (Picea mariana (Mill.)B.S.P.) was the most abundant type of tree constituting >99% of the tree stand, with 25,766 stems•ha −1 consisting of 37% taller trees (>137 cm height) up to 769 cm high [44].The black spruce stand had an average canopy height of 168 cm, projection coverage of 42%, and basal area of 73.5 m 2 •ha −1 [17].Trees were generally evenly distributed across the study plots (i.e., not clustered).Based on plant species indicators, the studied bog was classified as a forested low shrub bog [55] with two distinct microtopographic features: hummock and hollow.One year prior to this study, the control and experimental site microforms had equal coverage of mosses with sparse shrubs, whereas the drained hummocks had the highest coverage of shrubs and the drained hollows had the greatest coverage of lichens [44].At all sites, black spruce (Picea mariana (Mill.)B.S.P.) was the most abundant type of tree constituting >99% of the tree stand, with 25,766 stems•ha −1 consisting of 37% taller trees (>137 cm height) up to 769 cm high [44].The black spruce stand had an average canopy height of 168 cm, projection coverage of 42%, and basal area of 73.5 m 2 •ha −1 [17].Trees were generally evenly distributed across the study plots (i.e., not clustered).Mean (±SD) pH and electrical conductivity (μS•cm −1 ) of pore water in the control (4.1 ± 0.1 and 16.6 ± 0.7, respectively) and experimental (4.4 ± 0.3 and 15.2 ± 2.5, respectively) sites were similar (ANOVA, pH: F1, 5 = 2.6, p = 0.166; EC: F1, 5 = 0.84, p = 0.401) prior to any manipulation.All the study sites had an average peat depth exceeding 4 m, and were underlain by a sandy clay substrate.Because all the study sites were part of the same bog and their initial primary characteristics of vegetation composition, WT levels, and chemistry were similar, their secondary attributes (e.g., respiration rates) were also assumed to be similar prior to WT manipulation. From the available microtopography, we chose eight hummock and eight hollow microforms at each of the control and drained sites, and four of each microform type at the experimental site (due to its smaller area).In early May 2012, each of the chosen microform plots was fitted with a 60 cm × 60 cm collar having a groove at the top for placing the CO2 flux chamber.The collar was carefully inserted into the peat surface to a depth of ~5 cm to keep the disturbance minimal [56].Outside each collar, a Polyvinyl Chloride (PVC) well (length = 200 cm, diameter = 3.5 cm) with perforation and a nylon cloth covering on the lower 150 cm was inserted into a bore-hole drilled to a depth of 150 cm.WT level was manually measured at all the water wells every time CO2 flux was measured during the growing season of 2012.The WT levels were also monitored during the study period at 20-min intervals using automatic, temperature compensating pressure transducers (Levelogger Junior 3001, Solinst, Georgetown, Ontario, Canada) installed in two randomly selected water wells at each site: one at a hummock and the other at a hollow plot.Mean (±SD) pH and electrical conductivity (µS•cm −1 ) of pore water in the control (4.1 ± 0.1 and 16.6 ± 0.7, respectively) and experimental (4.4 ± 0.3 and 15.2 ± 2.5, respectively) sites were similar (ANOVA, pH: F 1, 5 = 2.6, p = 0.166; EC: F 1, 5 = 0.84, p = 0.401) prior to any manipulation.All the study sites had an average peat depth exceeding 4 m, and were underlain by a sandy clay substrate.Because all the study sites were part of the same bog and their initial primary characteristics of vegetation composition, WT levels, and chemistry were similar, their secondary attributes (e.g., respiration rates) were also assumed to be similar prior to WT manipulation. From the available microtopography, we chose eight hummock and eight hollow microforms at each of the control and drained sites, and four of each microform type at the experimental site (due to its smaller area).In early May 2012, each of the chosen microform plots was fitted with a 60 cm × 60 cm collar having a groove at the top for placing the CO 2 flux chamber.The collar was carefully inserted into the peat surface to a depth of ~5 cm to keep the disturbance minimal [56].Outside each collar, a Polyvinyl Chloride (PVC) well (length = 200 cm, diameter = 3.5 cm) with perforation and a nylon cloth covering on the lower 150 cm was inserted into a bore-hole drilled to a depth of 150 cm.WT level was manually measured at all the water wells every time CO 2 flux was measured during the growing season of 2012.The WT levels were also monitored during the study period at 20-min intervals using automatic, temperature compensating pressure transducers (Levelogger Junior 3001, Solinst, Georgetown, Ontario, Canada) installed in two randomly selected water wells at each site: one at a hummock and the other at a hollow plot. CO 2 Flux Measurements All CO 2 flux measurements were made during the day time of the growing season (May to October, 2012) using the same equipment.We used a closed chamber with dimensions 60 cm × 60 cm × 30 cm (width × length × height), made of opaque acrylic and fitted with a low-speed battery-operated fan to circulate air within the chamber headspace during and between CO 2 concentration measurements.The chamber had no pressure equilibrium port installed.A portable infrared gas analyzer (EGM-4, PP Systems, Amesbury, MA, USA) with a built-in CO 2 sampling pump operating at a flow rate of 350 mL•min −1 and compensating for temperature fluctuations within the chamber headspace was used to measure the instantaneous CO 2 concentration inside the chamber headspace.The chamber headspace temperature was measured using a thermocouple thermometer (VWR International, Edmonton, AB, Canada).The CO 2 concentration and temperature measurements were made every 15 s during a short chamber deployment period [57,58] of 1.75 min.Immediately after the CO 2 concentration measurements at a plot, soil temperature at a depth of 5 cm (T 5 ) was measured using a thermocouple thermometer, and the WT level relative to moss surface was manually measured from a permanently installed water well adjacent to the plot.The CO 2 flux was calculated from the linear change in CO 2 concentration in chamber headspace over time [17], as a function of air temperature, pressure, and volume within the chamber headspace, following the ideal gas law. Forest Floor Respiration (R FF ) Prior to any manipulation at a plot, a CO 2 efflux measurement represented forest floor respiration (R FF ).During a 5-day long R FF measurement campaign in May 2012, we measured the fluxes on four to five occasions in each plot.The measured R FF is divided into major source-based respiration flux components as: where R FF accounts for forest floor aboveground autotrophic respiration (R FF_A_ag ), belowground shrub + herb and tree autotrophic (rhizospheric) respiration (R A_SH_bg + R A_T_bg , respectively), and soil heterotrophic respiration (R H ). At the end of the R FF measurement campaign, we clipped all plots using sharp scissors at the base of capitulum at 1 cm below the moss surface [27,59].All the plots had their surface carefully cleared of any plant litter.The clipped shrubs + herbs were placed in labelled paper bags, taken to the Ecohydrology Lab, University of Calgary, AB, oven dried at 60 • C for 48 hours and weighed to calculate mean biomass (g•m −2 ) at each plot at each site. Partitioning Forest Floor Respiration Following the R FF measurement campaign, a 5-day long campaign for measuring the post-clipping CO 2 emissions from every plot was performed.During the campaign, we measured the emissions on four to five occasions in each plot.At plot level, the CO 2 emissions (g To estimate R A_T_bg at hummock or hollow microform at each site, we used a trenching method [17,52].One half of the instrumented hummock or hollow plots (60 cm × 60 cm) were chosen at each site for the trenching procedure, whereas the other half of the plots were left untrenched.The trenched hummock or hollow plot R FF was not significantly different from those of the untrenched plots at control (ANOVA, Hummock: F 1, 25 = 0.667, p = 0.422; Hollow: F 1, 23 = 0.316, p = 0.580), experimental (ANOVA, Hummock: F 1, 7 = 0.000, p = 0.990; Hollow: F 1, 7 = 0.605, p = 0.466), or drained (ANOVA, Hummock: F 1, 23 = 0.041, p = 0.841; Hollow: F 1, 23 = 0.070, p = 0.790) site.Therefore, in June 2012, we incised around the trenched plots to a depth of 30 cm and installed a thick polythene sheet to prevent root ingrowth while keeping the disturbance minimal.All the trenched and intact, untrenched plots were measured for CO 2 emission between July and September, 2012 to quantify the difference in the respiration rate for estimation of R A_T_bg .The trenching method at this site had already been used at this bog [17]. We did not measure R A_SH_bg and calculated it by using regression equations (y = a + bx) generated by regressing the aboveground biomass of shrubs + herbs (x) with R FF_A_ag (y) following a previous study [52].We did not sample/measure shrub + herb root biomass (B SH_bg ) to avoid disturbance to plots used for ongoing research and instead calculated B SH_bg as: B SH_ag × 0.39 [46], as previous research found the median root:shoot ratio of shrubs + herbs for 14 data points in boreal forest to be 0.39.Similar factors control net production both above and belowground, and the two rates were found to be related with each other [47].The generated regression equations were used to determine the (R A_SH_bg ) by substituting B SH_ag with B SH_bg in the equation.The R H was calculated from Equation (1) once all other components were estimated. Seasonal Modeling Only the measured respiration components (R FF , R FF_A_ag , R A_T_bg ) during the growing season (May to October) of 2012 were modeled using a multiple linear regression model with T 5 and WT level as: where a, b, and c are regression coefficients (Table 1).Seasonal R FF , R FF_A_ag , and R A_T_bg were estimated for each 20-min period between 1 May to 31 October 2012, averaged daily, and summed separately for the growing season using T 5 (Onset®, HOBO®, Bourne, MA, USA) and WT level (Levelogger Junior, Solinst Canada Ltd., Georgetown, ON, Canada) measurements made on site. As the environmental variable logs were missing for the first 21 days of May 2012, they were filled by assuming that the first measured value was representative of the whole missing period.The field measured values of R FF , R FF_A_ag , and R A_T_bg were plotted against the model predicted values obtained using SPSS 24.0.0.1.Validation of the models showed excellent agreement between the measured and the modeled values (Appendix A: Figure A1).The seasonal respiration rates (R FF , R FF_A_ag , R A_T_bg ) at hummock and hollow microforms were up-scaled by multiplying mean estimated growing season respiration by their corresponding coverage of 56% and 44% at the control, 55% and 45% at the experimental, and 52% and 48% at the drained site, respectively [60].The seasonal value of R A_SH_bg was calculated separately for each microform/site combination by determining it as a proportion of corresponding instantaneous R FF value and then estimating it as this proportion of the modeled seasonal R FF .Seasonal R H was determined by difference according to Equation (1). Statistical Analyses To test the effects of WT level and microform on R FF , R FF_A_ag , R A_SH_bg , and R A_T_bg , we conducted a repeated, linear mixed-effects model analysis (LMEM; IBM SPSS Statistics 24.0.0.1,IBM corporation, Armonk, New York, USA) with WT level (control, experimental, drained) and microform (hummock, hollow) as predictor variables and R FF , R FF_A_ag , R A_SH_bg , or R A_T_bg as response variables, including the random effect of plot and repeated effect of time (as the same plots were used sequentially for all the measurements).We also tested the effect of B SH_ag in combination with site and microform on R FF_A_ag with random effect of plot using a repeated, LMEM.In this case, non-significant terms were removed from the model one at a time, starting with the highest p-value and the model was re-run until only significant terms remained.In all the LMEMs used in this study, a combined symmetry covariance structure was used.The relationships of WT level and T 5 with R FF , R FF_A_ag , R A_SH_bg , and R A_T_bg were also tested for their significance using linear regression model fitting where applicable.Differences between regression slopes were tested [61] where applicable. Results The microclimate of the study sites was monitored over the growing season (May to October) of 2012 and was warmer by 1.4 • C and drier by 79 mm than the 30-year (1981-2010) regional averages measured at Athabasca, Alberta, Canada.The WT levels at the experimental and drained sites were as much as 36 cm and 82 cm lower than at the control site (Figure 2).In general, as a result of 12 years of drainage, the mosses at the drained hummocks were replaced by shrubs and mosses at the drained hollows were replaced by lichens.Detailed site hydrological responses to the warmer and drier climate at these sites have been reported [17]. Controls on Respiration Flux Components All the measured component fluxes (R FF , R FF_A_ag , R A_SH_bg , R A_T_bg ) were well correlated to T 5 and WT (Figure 3) and therefore the modeled values matched the measured values well (Appendix A: Figure A1).Generally, the highest respiration occurred with warm temperatures and deep WT position. Mean Forest Floor Respiration Rate (R FF ) There were significant effects of each of the WT (control, experimental, drained) and microform (hummock, hollow) types on R FF measured in May, 2012; however, their interaction term did not significantly affect R FF (Table 2).The drained site had significantly higher R FF value (±SE) of 21.0 ± 2.1 g•CO 2 •m −2 •day −1 compared with those measured at the experimental (13.2 ± 2.5 g•CO 2 •m −2 •day −1 ; p = 0.042) and control (9.3 ± 2.1 g•CO 2 •m −2 •day −1 ; p = 0.006) sites that were not different (p = 0.455) from each other.Bonferroni pairwise comparisons revealed that the hummocks had overall significantly higher R FF (17.9 ± 1.8 g•CO 2 •m −2 •day −1 ) than hollows (11.2 ± 1.8 g•CO 2 •m −2 •day −1 ).Comparing microforms across sites (Figure 4) revealed that R FF at hummocks was significantly different among all sites with highest rates at drained, followed by experimental, and then control, while hollows were not different (p = 0.526) across sites.12 = 0.9 0.437 * All models included a random effect of plot at sites to account for repeated measurements made at each site. Comparing microforms across sites, drained hummocks had significantly higher RFF_A_ag emissions than all other plots (p = 0.005; Figure 4) while RA_SH_bg at hummocks was significantly different There were significant effects of WT and microform treatments individually and interactively on both the R FF_A_ag and R A_SH_bg (Table 2).Similar to R FF , the R FF_A_ag and R A_SH_bg values were highest at the drained site (5.2 ± 0.6 g•CO 2 •m −2 •day −1 ; 2.8 ± 0.6 g•CO 2 •m −2 •day −1 , respectively), followed by values at the experimental (3.−1 , respectively) sites.The R FF_A_ag and R A_SH_bg fluxes were also overall higher at the hummocks (4.9 ± 0. day −1 , respectively).Comparing microforms across sites, drained hummocks had significantly higher R FF_A_ag emissions than all other plots (p = 0.005; Figure 4) while R A_SH_bg at hummocks was significantly different between all three sites.There were no significant differences at hollows across the WT treatment sites for either R FF_A_ag or R A_SH_bg. Forests 2017, 8, 75 9 of 17 between all three sites.There were no significant differences at hollows across the WT treatment sites for either RFF_A_ag or RA_SH_bg.We used BSH_ag to calculate BSH_bg (BSH_bg = BSH_ag × 0.39; Table 3) and determine RA_SH_bg (Figure 4, 5) using the regression equations we generated by regressing BSH_ag with RFF_A_ag (explained in detail in Methods section).The BSH_ag was not different among sites or microforms due to large variation between plots; however, the drained site had the highest, while experimental had the lowest BSH_ag of all sites, and drained hummocks had higher BSH_ag than those of control and experimental hummocks in that order (Table 4, Figure 4).The BSH_ag and microform type significantly explained RFF_A_ag emissions individually and interactively (Table 4).Also, the RFF_A_ag was significantly related to an interaction between BSH_ag, site and microform, where BSH_ag was significantly related to the RFF_A_ag at all sites and microforms except at the experimental hollows which had the lowest or inconsistent BSH_ag values (Table 4; Figure 5).The overall regression lines' slopes differed significantly at the hummocks and hollows (z = 4.43; 3.12, respectively).Regarding hollows, the slope was steeper at the drained site compared to control.* all values are mean ± SD (n = 6 for each of the BSH_ag and BSH_bg, and n = 3 (except experimental site with n = 1) for BT_ag) [17].Site biomass was determined by weighting the forest floor by the proportion of hummock and hollow microforms at each site (hummocks: control = 56%, experimental = 55%, drained = 52%), where applicable.We used B SH_ag to calculate B SH_bg (B SH_bg = B SH_ag × 0.39; Table 3) and determine R A_SH_bg (Figures 4 and 5) using the regression equations we generated by regressing B SH_ag with R FF_A_ag (explained in detail in Methods section).The B SH_ag was not different among sites or microforms due to large variation between plots; however, the drained site had the highest, while experimental had the lowest B SH_ag of all sites, and drained hummocks had higher B SH_ag than those of control and experimental hummocks in that order (Table 4, Figure 4).The B SH_ag and microform type significantly explained R FF_A_ag emissions individually and interactively (Table 4).Also, the R FF_A_ag was significantly related to an interaction between B SH_ag , site and microform, where B SH_ag was significantly related to the R FF_A_ag at all sites and microforms except at the experimental hollows which had the lowest or inconsistent B SH_ag values (Table 4; Figure 5).The overall regression lines' slopes differed significantly at the hummocks and hollows (z = 4.43; 3.12, respectively).Regarding hollows, the slope was steeper at the drained site compared to control. Modeled Respiration Components Overall, all modeled seasonal (May to October) respiration components (g•CO 2 •m −2 .growingseason −1 ) were in the order of drained > experimental > control site, with greater overall increases at hollows than at hummocks at all the sites (Table 5).R H accounted for approximately 48%, 43%, and 37% of R FF at control, experimental, and drained sites, respectively.The R FF and R A_T_bg fluxes at the drained site were significantly higher than those at the control site, while R FF_A_ag and R A_SH_bg were not different among sites, but were different between drained hummocks and drained hollows.The seasonal R A_T_bg and R H values were highest at the drained hollows. Modeled Respiration Components Overall, all modeled seasonal (May to October) respiration components (g•CO2•m −2 .growingseason −1 ) were in the order of drained > experimental > control site, with greater overall increases at hollows than at hummocks at all the sites (Table 5).RH accounted for approximately 48%, 43%, and 37% of RFF at control, experimental, and drained sites, respectively.The RFF and RA_T_bg fluxes at the drained site were significantly higher than those at the control site, while RFF_A_ag and RA_SH_bg were not different among sites, but were different between drained hummocks and drained hollows.The seasonal RA_T_bg and RH values were highest at the drained hollows. Discussion In agreement with the previous seasonal (2012) RFF estimates made at these sites [17], this research found the greatest growing season RFF values at the drained site (422 ± 22 g•CO2•m −2 ), smaller values at the experimental site (354 ± 16 g•CO2•m −2 ), and the smallest values at the control site (255 ± 10 g•CO2•m −2 ; Table 2); however, in general, the values at hollows across sites were slightly lower in Discussion In agreement with the previous seasonal (2012) R FF estimates made at these sites [17], this research found the greatest growing season R FF values at the drained site (422 ± 22 g•CO 2 •m −2 ), smaller values at the experimental site (354 ± 16 g•CO 2 •m −2 ), and the smallest values at the control site (255 ± 10 g•CO 2 •m −2 ; Table 2); however, in general, the values at hollows across sites were slightly lower in the present study, as these R FF values were modeled using measurements made over a shorter duration and from different plots (located in an area adjacent to the previous study plots [17]).The increased losses of CO 2 at the short-and long-term drained sites that we observed compare well with those reported by others from experimentally drained boreal peatlands [9,32,44,62,63].Declining WT level promotes desiccation of aquatic vegetation and soil that progresses over time.The R FF was partitioned into major respiration components (R FF_A_ag , R A_SH_bg , R A_T_bg ) which were then modeled for seasonal estimates with WT level and T 5 as covariates.Model validations across sites/microforms showed excellent agreements within RSE of 0.24-2.41g•CO 2 •m −2 growing season −1 (Table 1, Figure A1).Many investigations on peatland or forest respiration components have shown that warm and dry conditions enhance autotrophic [9,30,64,65] and heterotrophic [25,26,39] respiration emissions with greater impact over longer time scales [17,32,66].Warmer air and soil temperatures stimulate microbial activity, resulting in increased respiration fluxes; however, T 5 response of R H depends on substrate type and availability of nutrients and moisture [67,68].Water-table lowering in peatlands prompts increased respiration emissions that are enhanced with increase in peat surface temperature, as we noticed that R FF values at our sites were well correlated to both WT level and T 5 (Tables 2 and 3; Figure 3).In contrast, a few studies suggest that the vascular vegetation (shrubs + herbs and trees) are less sensitive to WT lowering, as they can increase their rooting depth with deeper WT [15,30].Peatland microforms have been observed to respond to changes in WT level and T 5 with different magnitudes and in different directions [16,60,69] mainly due to differences in vegetation coverage and composition.Counter to our hypothesis, we observed a significant increase in R FF in response to WT lowering at hummocks, while R FF at hollows was not significantly affected.Hummocks were drier and had significantly higher coverage of shrubs + herbs compared to hollows that were dominated by Sphagnum mosses at the research sites prior to WT manipulation.Therefore, differences in R FF response to WT drawdown may be driven by changes in either autotrophic or heterotrophic respiration, or both. At the control site, autotrophic respiration components (R FF_A_ag , R A_SH_bg , R A_T_bg ) were similar at hummocks and hollows.As for R FF , drainage increased autotrophic respiration at hummocks, but only resulted in increased R A_T_bg at hollows.The greater R FF_A_ag and R A_SH_bg at hummocks (dominated by shrubs + herbs) was found to be related to B SH_ag and B SH_bg (Tables 4 and 5), respectively, indicating that drainage-induced increase in shrubs + herbs biomass had a strong control on above and belowground autotrophic respiration components at hummock microforms, which is similar to the findings of several studies [44,70,71].Minimal change in B SH_ag and B SH_bg at hollows (Table 3) resulted in the non-significant change in R FF_A_ag and R A_SH_bg .As tree roots likely extend across all microforms, particularly as the depth of aerated peat is thickened by WT drawdown, the overall increase in tree productivity in response to drainage [17] resulted in increased R A_T_bg across both hummocks and hollows.Although R FF_A_ag did not increase significantly at hollows following drainage, the slope of the R FF_A_ag versus B SH_ag became steeper, indicating increasing autotrophic respiration rates per unit increase in biomass.This may reflect the shift from moss dominated vegetation at the control site to more herbs and shrubs at the drained site, as vascular plants tend to have higher respiration rates than bryophytes [72,73]. Although R H increased at some microforms (i.e.drained hollows) in response to WT drawdown, there was no significant difference across the WT treatments sites (Table 5).We hypothesized that R FF increase would be greatest at hollows, as the WT drawdown would shift this microform position from largely anoxic to oxic conditions, resulting in large increases in R H that would drive shifts in R FF .We did observe substantial increases in R H at hollows following WT lowering; however, these were masked by autotrophic respiration.Previous studies have also reported that R FF_A_ag can account for the majority of peatland respiration [45].As WT drawdown enhanced plant productivity at the study site [17], R H accounted for a declining proportion of growing season R FF from 48% at control to 36% at the drained site.Hummock R H did not change substantially in response to WT drawdown.As WT was initially deep below hummocks at the control site, due to the continental climate at the study site, the surface peat was well-aerated initially and R H was likely rarely limited by saturated conditions.Drainage may actually result in desiccation of the surface peat that could result in conditions too dry for optimal rates of R H [74].The limited change in R H between the WT treatments is therefore partially driven by the differential microform response where R H is enhanced at hollows, with little change, or slight reduction at hummocks. Overall, the substantial contribution of autotrophic respiration to R FF suggests that the large increase in R FF observed in response to WT drawdown will result in only slow loss of the C accumulated in peat, with these losses likely greatest at hollows [17].However, comparison of R H is partially complicated by the fact that it was calculated by difference once all the other partitioned flux components were estimated, and thus also contains error associated with the estimation of individual components.Isolated partitioning of the source-based respiration components remains to be developed, although the few manipulative field experiments that have investigated how climate change factors interact with one another to alter soil respiration [41,75] were not able to separate soil respiration into its components without significantly disrupting the soil [28].Separating peat soil C into various major components is an important challenge for improving our understanding of peatland C cycling response to climate change [76]. Conclusions Experimental water-table (WT) drawdown in a treed boreal bog increased forest floor respiration (R FF ).While all measured and estimated respiration components also increased following WT lowering, these were generally only significantly increased at hummock microforms.Increases in R FF at hummocks were largely driven by increases in autotrophic respiration as shrub biomass increased.Drainage increased heterotrophic respiration at hollows, with less response at hummocks, suggesting that carbon is more likely to be released from stored peat at hollows.Overall, shifts in R FF were largely driven by autotrophic respiration, indicating that rapid destabilization of peat carbon stocks under drying conditions are unlikely.Partitioning R FF into its subcomponents accurately and without substantial disturbance to the soil is difficult and the development of partitioning methods are needed to better understand the fate of peat carbon stocks under various disturbances. Figure 1 . Figure 1.Geographical map of the Wandering River study sites located in a forested peatland complex within boreal forest in Alberta, Canada[54]. Figure 1 . Figure 1.Geographical map of the Wandering River study sites located in a forested peatland complex within boreal forest in Alberta, Canada[54]. Figure 2 . Figure 2. Daily mean site air temperature (°C), soil temperature (°C) at 5 cm depth, total precipitation (mm), representative hollow and hummock water-table (WT) level (cm) over the 2012 growing season (May to October).Note the two y-axes on the right side: daily mean WT level, and air and soil temperatures. Figure 2 . Figure 2. Daily mean site air temperature ( • C), soil temperature ( • C) at 5 cm depth, total precipitation (mm), representative hollow and hummock water-table (WT) level (cm) over the 2012 growing season (May to October).Note the two y-axes on the right side: daily mean WT level, and air and soil temperatures. Figure 4 . Figure 4. Mean (±SD) RFF, RFF_A_ag, RA_SH_bg, and RA_T_bg measured during the growing season (May to October) of 2012 at all sites/microforms.Mean RA_SH_bg was determined by using a biomass regression method (Tables3 and 4; Figure5)[46,52].Bars having no letters in common are significantly different (p < 0.05) while bars with same letters indicate no significant difference (p > 0.05); letters should be compared only within one flux component across all microforms. Figure 4 . Figure 4. Mean (±SD) R FF , R FF_A_ag , R A_SH_bg , and R A_T_bg measured during the growing season (May to October) of 2012 at all sites/microforms.Mean R A_SH_bg was determined by using a biomass regression method (Tables 3 and 4; Figure 5) [46,52].Bars having no letters in common are significantly different (p < 0.05) while bars with same letters indicate no significant difference (p > 0.05); letters should be compared only within one flux component across all microforms. Figure 5 . Figure 5. Aboveground autotrophic respiration of forest floor (RFF_A_ag; g•CO2•m −2 •day −1 ) versus aboveground biomass of shrubs + herbs (BSH_ag; g•m −2 ) at (a) hummock and at (b) hollow microforms at each site.Regression lines were plotted for each microform type at each site when statistically significant at p < 0.05. Figure 5 . Figure 5. Aboveground autotrophic respiration of forest floor (R FF_A_ag ; g•CO 2 •m −2 •day −1 ) versus aboveground biomass of shrubs + herbs (B SH_ag ; g•m −2 ) at (a) hummock and at (b) hollow microforms at each site.Regression lines were plotted for each microform type at each site when statistically significant at p < 0.05. Table 1 . Fitted model parameters, their values (±SE), residual standard error (RSE), p values, adjusted r 2 , and the number of values (n) included in the regression analyses for the forest floor respiration (R FF ), forest floor aboveground autotrophic respiration (R FF_A_ag ), and belowground autotrophic respiration of tree roots (R A_T_bg ) models (Equation (2)) . R FF , R FF_A_ag , and R A_T_bg models were developed for each microform type (n = 3) at the control, experimental, and drained sites for the growing season of 2012.a and b are soil temperatures at 5 cm depth and WT level (below-ground) coefficients, respectively, and c is a regression constant.All modeled parameters are significant at α = 0.05. Table 2 . Statistical results of repeated, linear mixed effects models.The models tested the fixed effects of site and microform on respiration of forest floor (RFF), aboveground autotrophic respiration of forest floor (RFF_A_ag), and belowground autotrophic respiration of shrubs + herbs and trees (RA_SH_bg and RA_T_bg, respectively), separately . Table 2 . Statistical results of repeated, linear mixed effects models.The models tested the fixed effects of site and microform on respiration of forest floor (R FF ), aboveground autotrophic respiration of forest floor (R FF_A_ag ), and belowground autotrophic respiration of shrubs + herbs and trees (R A_SH_bg and R A_T_bg , respectively), separately .All models included a random effect of plot at sites to account for repeated measurements made at each site. * Table 4 . Statistical results of a repeated, linear mixed effects model with fixed effects of site (control, experimental, drained), microform (hummock, hollow), and aboveground biomass of shrub + herb (B SH_ag ; covariate), random effect of plot, and an outcome variable of aboveground autotrophic respiration of shrubs + herbs at the forest floor (R FF_A_ag ) . Random effect of plot was included in the model to account for the repeated measurements made at each site. Table 4 . Statistical results of a repeated, linear mixed effects model with fixed effects of site (control, experimental, drained), microform (hummock, hollow), and aboveground biomass of shrub + herb (BSH_ag; covariate), random effect of plot, and an outcome variable of aboveground autotrophic respiration of shrubs + herbs at the forest floor (RFF_A_ag) .Random effect of plot was included in the model to account for the repeated measurements made at each site.	v3-fos-license
2021-05-07T00:03:55.132Z	2021-03-01T00:00:00.000	233799385	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GREEN", "oa_url": "https://www.researchsquare.com/article/rs-220148/v1.pdf?c=1631891791000", "pdf_hash": "b2dde5d8b46ab8df62e3756361f5f93e1ddfce60", "pdf_src": "Adhoc", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45413", "s2fieldsofstudy": [ "Medicine" ], "sha1": "3c8e8ced539139ed3bd2a75c4f7f0f960c5fb77a", "year": 2021 }	pes2o/s2orc	The Circadian Temperature Trajectory Predicts the Severity and Prognosis in Critically Ill Patients Background: Disrupted circadian temperature is commonly observed in patients in the intensive care unit (ICU). The aim of this study is to examine the association between body temperature (BT) circadian rhythm and mortality critically in patients receiving ICU admission for at least 24h.Method: Adult patients with a complete record of temperature during the first 24 hours of ICU stay in the Multi-parameter Intelligent Monitoring in Intensive Care III (MIMIC-III) database were included in this retrospective cohort study. Body temperature circadian rhythm ratio (BTCRR) was calculated according to the value of mean nighttime divided by daytime mean temperature. All patients were divided into the nocturnal BT rising (NBTR) group (BTCRR >1) and the non-NBTR group (BTCRR≤1). Five subgroups were also built according to different quantile of BTCRR (5%, 10%, 30%, 50%). The associations of NBTR, subgroup, and BTCRR with 28-day mortality were assessed separately using Cox proportional hazards model.Findings: The overall cohort comprised 32419 patients. The non-NBTR group (n=20148) had higher 28-day mortality than the NBTR group (n=12271). After adjusting for covariates, the analysis showed that NBTR was significantly associated with mortality at 28 days (hazard ratio: 0.923; 95% CI, 0.888–0.960, P<0.05). All results of subgroup analysis showed obvious statistical significance, and similar results persisted in the patients with different groups. The % BTCRR had a significant non-linear (p < 0.05) association with 28d-mortality after adjusting for other variables (p < 0.05). Increasing the percentage up to 101% resulted in a hazard ratio (HR) to reduced mortality (HR: 0.96; 95%, 0.941–0.972, P<0.001), while increases above 101 % didn’t make a significant suggestion in mortality.Conclusions: The findings of this study suggest that both low and high BTCRR indicates a poor outcome, such that having a BTCRR of 101% had a survival advantage. BTCRR may aid in the early identification of critically ill patients at high risk of 28-day mortality. These findings may provide a basis for future randomized controlled trials comparing temperature control of ICU patients. Introduction The condition of intensive care unit (ICU) patients is exposed to their own pathophysiological state as well as psychological, environmental, iatrogenic and other factors [1,2]. Early physiological monitoring and systematic assessment can aid clinicians in making effective interventions to improve patient outcome [3]. However, existing severity scoring systems and prediction tools give rise to challenges in integrating a comprehensive panel of physiologic variables and presenting to clinicians interpretable models early in a hospital admission. In particular, most approaches to early prediction of the severity of patients in ICU rely on the rst 24h, 48h or 72 h of a patient's ICU stay [4][5][6]. It is clear that the exposure factors of ICU patients induce disorders by affecting physiological homeostasis, including body temperature (BT), circadian rhythm, blood pressure, etc. [7]. To the best of our knowledge, this series of homeostasis imbalances is a major contributor to poor outcomes in ICU patients, leading to emotional depression, delirium, immune disorders, and cognitive impairment [8]. It is generally recognized that homeostasis balances of BT typically re ected as approximately 0.5°C around a mean of 37.0°C in healthy individuals [9]. On the contrary, it's worth noting that abnormal circadian body temperature (CBT) is associated with worse Acute Physiology and Chronic Health Evaluation III scores (APCHE III), suggesting that CBT monitoring can be a good predictor of potential ICU patients or that adjusting their CBT may improve outcomes [7]. From this perspective, CBT in the intensive care unit may be a useful predictor of illness. However, there is still a lack of convincing evidence and forecasting strategies. In this study, we extracted the BT records of the rst 24 hours of ICU admission from the Multi-parameter Intelligent Monitoring in Intensive Care III (MIMIC-III) database [10]. We aimed to explore the associations of CBT characteristics with mortality to facilitate early risk strati cation in the ICU. Study Design We conducted a retrospective cohort study of adult ICU admissions from the MIMIC-III database maintained by Beth Israel Deaconess Medical Center (BIDMC, Boston, MA) and the Massachusetts Institute of Technology (Cambridge, MA). MIMIC-III contains data from 61532 ICU admissions and includes demographic characteristics, vital signs, laboratory, physiologic information, medications, comorbidities, nursing notes, and survival outcomes in the adult ICUs of BIDMC from 2002 to 2011. Any researcher who agrees to the terms of the database must complete "Protection of Human Subjects" training before they can access the data. Patient Population The inclusion criteria were (1) age ≥ 16years at ICU admission and (2) day and night body temperature (BT) records during the rst 24 hours in the ICU. The exclusion criteria were (1) multiple ICU admissions, (2) length of ICU stay < 1 day, (3) BT measured < 6 times on the day 1 of admission. All diseases in our study were classi ed using the International Classi cation of Diseases, Ninth Revision (ICD-9). Given that all ICU admission records in the MIMIC-III database were anonymized, the requirement for individual ICU admission consent was waived by the institutional review board of Beth Israel Deaconess Medical Center. Circadian Rhythm of BT In this study, we focused on differences between average nighttime and daytime BT (℃). Consequently, the circadian rhythm of BT was described by body temperature circadian rhythm ratio (BTCRR) calculated as mean nighttime BT divided by mean daytime BT. Mean nighttime BT was calculated as all BT values during the night divided by the number of BT examinations, and the mean daytime BT was calculated as all BT values during the day divided by the number of BT examinations. Patients were divided into 2 groups according to the circadian rhythm of BT: the nocturnal BT rising (NBTR) group (> 1) and the non-NBTR group (≤ 1). 5 subgroups were also bult according to different quantile of BTCRR (5%, 10%, 30%, 50%): group A (≥ 50th percentile), group B (< 50th and ≥ 30th percentile), group C (< 30th and ≥ 10th percentile), group D (< 10th and ≥ 5th percentile), group E (< 5th percentile). The mean daytime BT value was computed as the average of all BT values from 6 AM to 10 PM, and mean nighttime MAP was from 10 PM to 6 AM the next day. Outcomes We reviewed the MIMIC-III database for information on therapy durations (admission time, discharge time, and death time). We obtained data on ICU stay, hospital stay and 28-d mortality by calculating relevant time points. The primary outcome measure in our study was 28-d mortality. Statistical Analysis Statistical analyses were performed using R software, version 3.6.2 (R Foundation for Statistical Computing). Continuous variables in the data were presented as mean value ± SD or median with interquartile range and categorical variables were presented as absolute numbers and percentage. Kolmogorov-Smirnov test was used to test the normality. The student t test or Mann-Whitney U test and Chi square or Fisher's exact tests were used to investigate the differences in quantitative and categorical variables, respectively between these groups. To assess the association of NBTR with 28-day mortality, we used univariate and multivariate Cox proportional hazards models for 28-day mortality and calculated Kaplan-Meier curves. In multivariate regression analyses, 5 models were built to show the modeling process, which could prove the stability of the association of NBTR with mortality. Model 1 was adjusted for age, gender, ethnicity. Model 2 was adjusted for model 1 plus an average of body temperature. Model 3 was adjusted for model 2 plus SOFA and APSIII. Model 4 was adjusted for model 3 plus primary diagnoses such as sepsis, cardiovascular disease, ARDS, renal disease, and type II diabetes. Model 5 was adjusted for model 4 plus laboratory analytical values such as WBC, glucose, hematocrit, and platelet. To assess the association of BTCRR with 28-day mortality, we divided the population into ve groups (A, B, C, D and E), and used univariate and multivariate Cox proportional hazards models for 28-day mortality. In multivariate regression analyses, 3 models were built. Model 1 was adjusted for age, gender, ethnicity, an average of body temperature. Model 2 was adjusted for model 1 plus WBC, glucose, hematocrit, and platelet. Model 3 was adjusted for model 2 plus SOFA, APSIII, sepsis, cardiovascular disease, ARDS, renal disease and type II diabetes. To show the relationships between BTCRR and 28-d mortality, Kaplan-Meier curves were calculated, and cause-speci c Cox proportional hazard models were implemented using restricted cubic splines, to predict the relative hazard of death in ICU with 95% con dence intervals. Results Data from a total of 61532 ICU admissions were accessed from the MIMIC-III database. We excluded 12821 admissions with ICU length of stay less than 24 hours, 3887 admissions with age less than 16 years and 3119 patients with multiple ICU admissions. Among the remaining patients, there were 39530 admissions with both day and night body temperature records during the rst 24 hours in ICU. After 7111 patients with < 6 times BT measured on the day 1 of admission were excluded, 32419 patients were included in the analysis (Fig. 1). Characteristics and Outcomes Patients in the non-NBTR group had lower age (P < 0.001), higher SOFA (P < 0.001), and higher APSIII (P < 0.001). The proportion of patients with sepsis, cardiovascular disease, or renal disease in the non-NBTR group was higher (P < 0.001). On the rst day of admission, there were more patients receiving ventilation therapy in the NBTR group (P < 0.001). Patients with ARDS or type II diabetes in the NBTR was more than the others (P < 0.001). Table 1 presents the characteristics across the classi cation of the circadian rhythm of BT. The patients in the non-NBTR group had higher ICU, hospital, and 28-day mortality than those in the NBTR group (8.0% versus 5.7%, 12.2% versus 8.9%, 11.6% versus 8.8%, respectively). Association of NBTR with the 28-d mortality The associations between NBTR and the 28-d mortality were further con rmed by COX regression (Table 2). After adjusted for age, gender, ethnicity, average of body temperature, SOFA, APSIII, sepsis, cardiovascular disease, ARDS, renal disease, type II diabetes, WBC, glucose, hematocrit, and platelet, the NBTR was independently associated with the 28-dmortality with a hazard ratio (HR) of 0.923 [95% con dence interval (CI) 0.888-0.960]. Kaplan-Meier survival curves revealed that the 28-d probability of survival was higher in the NBTR group than that in the NBTR group (Fig. 2). As observed in the Kaplan-Meier curves, Group E had the worst survival, whereas Group A survived the longest (log-rank P < 0.05) (Fig. 3). Figure 4 shows the Cox regression hazard ratios for BTCRR (HR value at 50% population with BTCRR as reference). The relationship between HR and BTCRR was U-shaped. It can be seen from the U-shaped gure that the population with a ratio of 1.01 has the highest survival advantage. Increasing the percentage up to 101% resulted in a hazard ratio (HR) to reduced mortality (HR: 0.96; 95%, 0.941-0.972, P < 0.001), while increases above 101 % didn't make a signi cant suggestion in mortality (P = 0.44). Discussion Circadian rhythm is a universal, built-in timing system that lasts nearly 24 hours and can be assessed through chronobiologic analysis of the time series of melatonin, cortisol and temperature [11]. This system was developed by the changes the human body respond to the external environment, speci cally, the periodic changes in light and darkness caused by the Earth's revolution around the sun. Prior research generally con rms that circadian rhythms are controlled by a master clock in the suprachiasmatic nucleus (SCN) of the hypothalamus and regulated by the nerve hormone including the hypothalamopituitary-adrenal (HPA) axis and melatonin in the pineal gland [1,12]. Patients hospitalized in the ICU are subject to changes in various zeitgebers due to light/dark cycles, social interactions, dietary patterns, medication, etc., which may cause changes in circadian rhythms such as core body temperature, leading to further adverse consequences [13,14]. Our study explores the role of diurnal temperature trend in ICU patients in predicting illness and prognosis, and for the rst time uses temperature models to distinguish the clinical trends of different groups. Traditionally, temperature has been considered a dichotomous variable and patients are classi ed as febrile or nonfebrile based on absolute value. However, evidence suggests that temperature pattern analysis can convey meaningful clinical information whether or not patients meet the criteria for fever [15][16][17]. Varela et al. investigated the role of temperature change analysis in predicting survival in critically ill patients and found that temperature analysis was similar in its ability to predict mortality as the sequential organ failure assessment score [18]. Some scholars have reported that typical changes in temperature patterns include changes in amplitude, increases in frequency, or increases or decreases in baseline variability [19]. Therefore, it provides support for patients in intensive care units to explore effective diurnal temperature trend prediction models. Published studies have reported the relationship between BT and biological rhythm or progression of disease or all-cause mortality in hospitalized, and ICU patients. Benjamin et al performed a cohort of severe trauma patients in France and reported that early exacerbation of the temperature rhythmicity is associated with the development of sepsis [20]. In another cohort of 6759 neurologic ICU patients from a 20-bed neurology ICU in the USA, elevated body temperature was found to predict higher mortality rate and worse outcome [21]. BT is one of the vital signs that can be used to evaluate APCHE III and mortality in critically ill patients [18]. However, no studies have reported circadian BT variation and its prognostic value in the ICU. In our study, we extracted complete BT records of patients' rst 24 hours in the ICU from a public database and investigated the circadian characteristics of BT to determine how to identify patients with a higher risk of mortality early in the hospital. We found that NBTR served as an important protective factor for higher survival in the 28-d mortality (HR: 0.923; 95% CI, 0.888-0.960) after adjusting for a series of covariates. To the best of our knowledge, this study is the rst in which the relationship between the circadian rhythm of BT and ICU mortality has been evaluated. Wu's study showed that the prognosis for sepsis patients in ICU became worse with decreased temperature minimum (T min), as well as increased T max and T max-min [22]. Further analysis indicated that A36.5-37°C (A: the area under the temperature curve) was associated with a positive prognosis. Meanwhile, A38-38.5°C, A38.5-39°C, and A39-39.5°C could result in a poor prognosis. From that perspective, strati ed comparison can better distinguish the survival conditions of different risk ratios. Therefore, it is necessary to scienti cally distinguish the risk levels of different groups for accurate prediction of clinical outcomes of diseases. To exclude the effect of clinical indicators included in the analysis on the relationship between BTCRR and survival outcomes, we divided the participants into 5 groups according to various BTCRR proportion and then conducted interaction and subgroup analysis. Findings suggested that the higher the proportion, the higher the survival. The association between BTCRR and mortality remained signi cant in various subgroups. We have demonstrated a non-linear, signi cant association between the percent BTCRR and mortality by 28-d. The results suggest that increasing the BTCRR to approximately 100 % was associated with decreased mortality, while increases above that point were associated with increasing mortality. From the perspective of biorhythm, moderate elevation of body temperature at night in ICU patients may be a positive embodiment of immune protection. This is consistent with the normal body temperature regulation, indicating that the patient has better immune regulation function [23][24][25][26][27]. Therefore, BTCRR could be used as a reliable risk factor for ICU mortality. An abnormal circadian status of BT is mainly caused by dysfunction of the thermoregulation center and day/night differences in physical activities [28]. Patients in the ICU usually experience tremendous acute stressors like infections, trauma, multiple organ dysfunctions, arti cial light, noise, mechanical ventilation, enteral nutrition, and medications. These factors further lead to abnormal thermoregulation [29][30][31]. Previous studies showed that abnormal body temperature is determined by the outcome of energy metabolism [32]. In addition, abnormal BT variation observed in patients is associated with autonomic nervous system dysfunction and poor sleep quality, which is also common in ICU patients, unless they are sedated or unconscious [33,34]. Nevertheless, there is still no de nitive study of the mechanism behind the circadian changes in body temperature, which provides a train of thought for further exploration. However, there are some limitations to our analysis. Firstly, MIMIC-III is a single-center database, and thus obvious selection bias cannot be ignored. On the positive side, the recruited patients were enrolled from various ICUs, in other words, their data may re ect real-world situations encountered by clinicians. Secondly, given the retrospective design, the data were previously collected. Therefore, some of the information is incomplete such as the frequency of BT monitoring, noise level, patient/nurse ratio. Although we have adjusted for as many covariates as possible and conducted a series of sensitivity analysis, a multicenter prospective study with adequate covariates is needed to further con rm the association between BTCRR and prognostic outcomes in critically ill patients. Thirdly, our study is only an association between BTCRR and mortality, not a cause-and-effect relationship. Subsequently, a highquality prospective research is urgently needed to evaluate causality between BTCRR and mortality. Conclusions The ndings of this study suggest that both low and high BTCRR indicates a poor outcome, such that having a BTCRR of 101% had a survival advantage. BTCRR may aid in the early identi cation of critically ill patients at high risk of 28-day mortality. These ndings may provide a basis for future randomized controlled trials comparing temperature control of ICU patients. Declarations Ethics approval and consent to participate Not applicable Consent for publication All authors have agreed to publish this article Availability of data and material The datasets used or analysed during the current study are available from the corresponding author on reasonable request. Competing interests The authors declare that they have no competing interests	v3-fos-license
2014-10-01T00:00:00.000Z	2010-10-22T00:00:00.000	7470475	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://alzres.biomedcentral.com/track/pdf/10.1186/alzrt54", "pdf_hash": "eef7da5d12adfdda3379e12597c3fdfe93b06f29", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45417", "s2fieldsofstudy": [ "Medicine" ], "sha1": "eef7da5d12adfdda3379e12597c3fdfe93b06f29", "year": 2010 }	pes2o/s2orc	Herbal therapy: a new pathway for the treatment of Alzheimer's disease It has been a clinical challenge to treat Alzheimer's disease (AD). In the present commentary we discuss whether herbal therapy could be a novel treatment method for AD on the basis of results from clinical trials, and discuss the implications for potential therapy for AD pathophysiology. There is evidence to suggest that single herbs or herbal formulations may offer certain complementary cognitive benefits to the approved drugs. The current evidence supporting their use alone, however, is inconclusive or inadequate owing to many methodological limitations. Herbal mixtures may have advantages with multiple target regulation compared with the single-target antagonist in the view of traditional Chinese medicine. Several clinical trials using herbal mixtures are being conducted in China and will hopefully show promising results for treating AD in the near future. Introduction Th e ultimate aim of Alzheimer's disease (AD) therapy is to stop or slow down the disease progression. Cholinesterase inhibitors have a modest clinical eff ect on the symptoms, however, and memantine -the currently available N-methyl-d-aspartate receptor antagonistdoes not prevent the deterioration of dementia [1,2]. Finding an eff ective method to treat AD still poses a signifi cant clinical challenge. Herbal medicine has long been used in China as therapy for dementia. Th e Complete Work of Jingyue published in 1624 contains the earliest known description in the world of a herbal therapeutic strategy for dementia. In the past 10 years, however, herbal drugs have seldom been approved for use alone in treating dementia. Overall, systematic review has identifi ed a few single herbs and herbal formulations as possible eff ective medicine for AD (Table 1). According to the current evidence, some of these therapies show promising results in terms of their cognitive benefi ts. In the present commentary we discuss whether herbal therapy could be a novel pathway to treat AD, on the basis of the results from clinical trials, and the implications for potential therapy of AD pathophysiology. Ginkgo biloba Ginkgo biloba extract is among the most widely used complementary therapies. A Cochrane review included 36 trials of gingko biloba, but most trials were small and of duration <3 months [3]. Nine trials were of 6 months duration and of adequate size, and were conducted to a reasonable standard. Of the four most recent trials to report results, three studies found no diff erence between Ginkgo biloba, at diff erent doses, and placebo [3], and one study found very large treatment eff ects in favor of Ginkgo biloba, but the trial sample size was very small [4]. Another recent trial reported negative results in reducing cognitive decline in older adults with normal cognition or with mild cognitive impairment [5]. Th e current overall evidence that Ginkgo has a predictable and clinically signifi cant benefi t for people with dementia or cognitive impairment therefore seems inconsistent and unreliable. Serrate clubmoss Huperzine A extracted from the serrate clubmoss herb is a potent, reversible and selective inhibitor of acetylcholinesterase. Considering the available evidence from six trials, Huperzine A seems to have some benefi cial eff ects on improvement of general cognitive function, global clinical status, behavioral disturbance and functional performance, with no obvious serious adverse events for patients with AD [6]. Only one study was of adequate quality and size, but the period during this study that found very large treatment eff ects was only 12 weeks [7]. Overall the current evidence supporting clinical use of Huperzine A is presently inconclusive or inadequate. Abstract It has been a clinical challenge to treat Alzheimer's disease (AD). In the present commentary we discuss whether herbal therapy could be a novel treatment method for AD on the basis of results from clinical trials, and discuss the implications for potential therapy for AD pathophysiology. There is evidence to suggest that single herbs or herbal formulations may off er certain complementary cognitive benefi ts to the approved drugs. The current evidence supporting their use alone, however, is inconclusive or inadequate owing to many methodological limitations. Herbal mixtures may have advantages with multiple target regulation compared with the single-target antagonist in the view of traditional Chinese medicine. Several clinical trials using herbal mixtures are being conducted in China and will hopefully show promising results for treating AD in the near future. Ginseng Panaxi ginseng's main active ingredient is panaxsaponin, which can enhance psychomotor and cognitive performance, and can benefi t AD by improving brain cholinergic function, reducing the level of Aβ and repairing damaged neuronal networks [8]. Th e high-dose ginseng group showed statistically signifi cant improvement on the Alzheimer Disease Assessment Scale (ADAS) and Clinical Dementia Rating (but not on the Mini-Mental State Examination) at the end of the study, when compared with the control group. Th is study was poorly designed, with an insuffi cient description of randomization and without blinding. Furthermore, the sample size was small (n = 15 for each group), and there was also a confounding eff ect due to concurrently administered western medications [9]. Th e evidence for ginseng as a treatment of AD is thus scarce and inconclusive. Further rigorous trials seem warranted [10]. Salvia offi cinalis Salvia offi cinalis has been used in herbal medicine for many centuries. After 4 months of treatment, salvia offi cinalis extract produced a signifi cantly better outcome on cognitive functions than placebo -as seen on the ADAS cognitive subscale and the Clinical Dementia Rating Sum of Boxes scale in patients with mild to moderate AD aged between 65 and 80 years [11]. Th ere were no signifi cant diff erences between salvia offi cinalis and placebo in terms of the observed side eff ects. In addition, salvia offi cinalis may reduce agitation in patients. More high-quality large-scale randomized controlled trials are needed, however, for further determination of the herb's effi cacy [11]. Herbal formulations or mixtures of herbal ingredients Herbal formulations or mixtures of herbal ingredients may have advantages with multiple target regulation compared with the single target antagonist in the view of traditional Chinese medicine, although there have been few clinical trials examining the effi cacy and safety of herbal formulations in AD patients. Shenwu capsule, a mixture of six herbs that is thought to reduce amyloid cytotoxicity, increased the memory score from baseline (n = 83) -but without signifi cant diff erence from aniracetam (n = 83) -in a 12-week phase II trial for patients with mild cognitive impairment [12]. A phase III trial is now underway. Stilbene glycoside, an extract of Shenwu capsule, has been evaluated in a phase I trial for AD. Further results for both of these formulations will be available in the next few years. GEPT, a combination of fi ve active components extracted from Chinese herbs, may be valuable for the treatment of AD -reducing the level of Aβ via the inhibition of γsecretase (presenilin-1) and the promotion of insulindegrading enzyme and neprilysin, which has been reported in the brain of APPV717I transgenic mice [13]. A 24-week preliminary study of GEPT showed a signifi cant improvement on cognitive function in patients with amnestic mild cognitive impairment, an early stage of AD (n = 101), consistently across diff erent cognitive scales; for example, an improvement in the ADAS cognitive subscale from baseline of -4.19 points (95% confi dence interval = -5.74 to -2.63), which declined at 24 weeks of follow-up after the GEPT withdrawal. Th is level of effi cacy was comparable with that of -4.23 points found in the subjects taking Donepezil (n = 100) ( Figure 1) [14]. GEPT is planned to apply for a shape II trial. Furthermore, the herbal preparations Ba Wei Di Huang Wan and Yi-Gan San are individually reported to significantly improve cognition or behavior and function on the Mini-Mental State Examination, the Neuro psychi atric Inventory and the Barthel Index in the patients with AD [15,16]. Conclusion Single herbs or formulations may be able to complement approved drugs for AD. No serious adverse events have been reported. Th e current evidence to support their use alone, however, is inconclusive or inadequate. Th is uncertainty is mainly caused by methodological limitations such as poor study design, relatively small sample sizes without a power calculation, inappropriate outcome measures and primary and secondary end-point selection, and invalid statistical analysis. In addition, the herbs' potential value for prevention and treatment of AD only results from symptomatic changes and short treatment periods (<6 months). Several studies currently underway or in early-stage development in China to evaluate herb mixtures will hopefully show promising results in the near future. Abbreviations AD, Alzheimer disease; ADAS, Alzheimer Disease Assessment Scale. Competing interests The authors declare that they have no competing interests.	v3-fos-license
2019-04-03T13:10:15.823Z	2019-01-01T00:00:00.000	91743922	{ "extfieldsofstudy": [ "Biology" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.1016/j.gecco.2019.e00544", "pdf_hash": "34e895348b6f0ca57acc787fdf69c2e4f5467c4e", "pdf_src": "Adhoc", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45418", "s2fieldsofstudy": [ "Environmental Science", "Biology" ], "sha1": "406a1312a292c79f20f1f91aed94f8a7f81f4e0a", "year": 2019 }	pes2o/s2orc	Genetic differentiation in leaf phenology among natural populations of Adansonia digitata L. follows climatic clines Lea ﬁ ng phenology is an important component of climatic adaptation in semi-arid regions. The questions are to what extent phenology is under genetic control and represent adaptation to local climates? In the present study, we compare leaf phenology among Adansonia digitata L. trees of 27 different origins from West and East Africa and test if the differences follow climatic clines. Timing of bud burst was largely synchronized with the start of rainy season, but some few individual trees showed bud burst before the ﬁ rst rain. Timing of leaf senescence was under genetic control with substantial differences among origins. The timing of senescence was for some origins at the end of rainy season and for some in the beginning of the dry season. Differences among origins in timing of leaf senescence were related to the variation in drought just before- and in the ﬁ rst months of the rainy season at the sites of origin. Populations from drier sites had the earliest leaf shed at the common test site indicating that trees have been adapted to the prevailing climatic conditions at the sites of origin. We discuss the results in the light of possible triggering factors. Introduction Tropical trees in many arid and semi-arid environments with a marked seasonality in rainfall have phenology synchronized to the periods of humidity and aridity. It most likely reflects that timing of phenological events is crucial for survival and in arid environments, and that selection therefore works against genotypes that are out-of-tune with the season. For example, setting or maintaining leaves in the dry season can lead to excessive loss of water, leading to leaf dieback and loss, depletion of nutrients and carbohydrates due to the loss of leaves and other drought associated damages such as xylem cavitation. For tropical deciduous tree species, investigations in situ have suggested that temperature, photoperiod, soil water availability, tree water status, air humidity or vapor pressure deficit influence the initiation of the leaf phenology (e.g. Njoku, 1964;Chidumayo, 2001;Borchert et al., 2005;Do et al., 2005;Seghieri et al., 2012). In tropical Africa, satellite imagery demonstrates a pre-rain greening of trees, which suggests that phenology may be triggered by changes in photoperiod or an increase in solar irradiation (Ryan et al., 2017). A greenhouse experiment, where Adansonia digitata L. (African baobab) populations were subjected to combinations of water stress-and day-length treatments, showed that leaf flushing was depending on both day-length and water regime and that population responded differently on the water regime (Di Lucchio et al., 2018). Studies of genetic variation in phenology of tropical trees of temporal dry zones are however rare (Raebild et al., 2011). Therefore, it is not clear whether reported differences in phenology are purely phenotypic reactions to varying environmental conditions, or whether there are genetic differences between origins, reflecting adaptation to the prevailing climate. This is a serious lack of knowledge, as climate change may challenge a current adaptation of trees to their local growth conditions (Ouedrago, 2014). Sub-Saharan Africa will likely face increased annual temperatures between 2 and 6 C during the next century. The development in annual precipitation is more uncertain (Niang et al., 2014), but there is a risk for shorter rainy seasons with extreme precipitation in West Africa (Sylla et al., 2016). If there are large genetic differences in phenology reflecting adaptation to prevailing climates, it will lead to the question whether species will be able to cope with rapid changed in climate by changing phenology through phenotypic plasticity or natural selection. It will also stress the need for gene conservation to retain genetic diversity. If the phenotypic plasticity and adaptation potential of the species is considered poor, an alternative will be to actively increase genetic diversity through assisted migration (e.g. Aitken and Bemmels, 2016;Lobo et al., 2018). Studies of genetic variation in phenology are therefore highly needed to guide management of plant genetic resources. The aim of this study is contribution to a better understanding of tropical tree phenology, using the A. digitata as a study species. We examine the variability in leaf phenology of A. digitata originating from different geographic and climatic zones, and test if the differences among origins follow geographical or climatic clines. The hypothesis is that differences between sites of origin are reflected in differences in leaf phenology among the origins as result of local adaptation. To study this, we take advantage of a common garden trial with a unique pool of genotypes representing a large part of the distribution area. A. digitata (African baobab) belongs to the family of Malvaceae (Baum, 1995) and is widely distributed in the driest parts of the Savannas of West Africa, East Africa and South East Africa. Present populations of West Africa, East and South Africa are isolated from each other due to a gap in Central Africa (Wickens, 1982;Sidib e and Williams, 2002). The distribution of A. digitata is mainly associated with precipitation and temperature rather than soil (Sanchez et al., 2010). The species is often found in smaller groups or as single trees mainly due to human influence in West Africa or wildlife in East and South East Africa (Dhillion and Gustad, 2004;Assogbadjo et al., 2005;Duvall, 2007;Venter and Witkowski, 2010;Larsen, 2010). A. digitata is highly adapted to semi-arid areas with frequent wild fires with its thick bark and thick fruit shells (Kempe et al., 2018). In natural populations, A. digitata sheds its leaves during the start of the dry season and flushes new leaves towards the end of the dry season (Sidib e and Williams, 2002;Assogbadjo et al., 2005), but some genotypes has been reported to retain leaves during the dry season (Gebauer and Luedeling, 2013). A. digitata is tetraploid. Seed and pollen of the species is mainly dispersed by fruit bats (Baum, 1995). Genetic marker studies indicate regional population structures and a modest gene flow (Kyndt et al., 2009;Larsen, 2010). A. digitata is a priority species for domestication in several African countries (Gebauer et al., 2016) and is of high importance for the food security in Africa providing minerals, vitamins and proteins also in years with famine (Yazzie et al., 1994;Nordeide et al., 1996;Sidib e et al., 1998;Glew et al., 1997;Diop et al., 2005;Parkouda et al., 2009;Gebauer et al., 2016) as well as medicine and fibers for ropes (Owen, 1970;Diop et al., 2005). Plant material and experiment Leaf phenology was studied among 27 origins in a field trial at Samanko in Mali (12.53 N,8.07 W). The origins were from 11 East and West African countries representing widely different annual rainfalls ranging from 260 to 1200 mm ( Table 1). The origins represented baobab populations from both North and South of the Equator (Fig. 1), and each population was represented by plants from seeds collected from approximately 20 open pollinated mother trees and mixed in equal proportions (10 seeds per sampled mother tree). Seedlings were germinated and raised in the nursery at the Regional Centre of Agronomic Research of Sotuba (Institute of Rural Economy, Bamako, Mali) from July 2007. The 1-year-old seedlings were planted in the field in August 2008 in a randomized block design with 4 replications and where each population was represented by one to five non-contiguous randomized 4-tree-plots in each replication. The trees of each population were chosen randomly among the seedlings raised from the seed mix of the 20 open pollinated mother trees. The spacing between trees was 2 Â 2 m. The long-term aim of the trial was to study the variation among African Baobab populations in their growth and development. Annual precipitation estimates for the field trial site was 781 mm and 630 mm in 2012 and 2013, respectively, with a long dry season extending from November to April (Fig. 2). Average monthly temperatures in 2012 and 2013 ranged between 21.4 and 32.8 C based on gridded climate estimates from the University of East Anglia Climatic Research Unit Harris et al., 2014). The growth of the plants from May 2009 to May 2011, and survival in the nursery are reported in Korbo et al. (2012), while the present study reports the variation in phenology in 2012 and 2013 among the different origins. Phenology monitoring Leaf phenology was assessed 2012e2013, approximately every two weeks, from 29th May 2012 to 25th June 2013 (day 150, 164, 178, 206, 220, 234, 248, 276, 290, 304, 318, 332, 346 and 360 in 2012 and day 9, 23, 37, 51, 65, 79, 93, 107, 121, 135, 149 and 163 in 2013). At each assessment, the phenological stage of all trees was recorded using 4 phases: phase 0 without leaves, and no signs of bud swelling; phase 1 with some leaf buds swelling, and bud breaking; phase 2 when young leaves are opened, but not totally expanded; phase 3 when leaves are fully opened, and adult; phase 4 when the leaves start to show signs of senescence in terms of a change in colors from normal green to yellow brown. Different parts of the trees could be in different stages of phenology, but only the dominant phase of the tree was used for further analysis. The change between two subsequent dominant phases was estimated as the median day between the two assessments where the change was recorded. Two measures of the timing of senescence were estimated: (i) change to phase 4 and (ii) change to phase 0. Time of bud burst was estimated as change from phase 0 or phase 1 to a more progressed phase. The mean number of living trees per provenance used for senescence assessment in autumn 2012 was 26 and the minimum number per provenance was four trees. The mean number of living trees per provenance used for the assessment of bud burst in spring 2013 was 23 and again the minimum number of trees was four trees. Not all populations were represented in all replications at the time of the phenology assessments (Table 1). Climate estimates For each of the sites of origin, estimates of monthly precipitation, evapotranspiration and mean temperature were obtained from gridded data available from the University of East Anglia Climatic Research Unit (CRU), Harris et al., 2014). A monthly drought index (DI) was estimated as the difference between monthly precipitation and potential evapotranspiration. Cumulated DI -and precipitation were calculated for all possible combinations of consecutive months from January to December and for all possible combinations of consecutive months from June to May. Temperature means were calculated for the same periods. Statistical analysis The following linear model was applied to test for differences among origins concerning the estimated days for changes in dominant phenology phases: Where Y ijk is the estimated day of the individual tree where it is changing from one phenology phase to another, m is the general mean, b i is the random effect of replication i in the trial P j is the fixed effect of origin number j, r ij is the random effect of plot expressing the origin by block interaction and e ijk is the residual, which is assumed to follow the normal distribution Nð0; s 2 e Þ. The significance of origins was tested using the Satterthwaite, (1946) option in the procedure GLM (general linear models) in SAS (SAS Inc. 2015), i.e. the origin effects were mainly tested by the origin by block interaction. Least square means (LSM) for origins were obtained from the procedure MIXED (SAS Inc. 2015) using model [1] above. Regressions of phenology on monthly precipitation, DI, and cumulated precipitation and DI over different periods at the sites of origins were tested. Similarly, mean monthly temperatures and mean temperatures for different periods at the sites of origins were tested. The reference period was 1961e2006. Origins from north and south of Equator were analyzed separately, because the annual patterns in rainfall are different for the two groups with a rainy season starting JanuaryeJune north of Equator and SeptembereNovember south of Equator (including tendencies to bimodal rainy seasons) (Fig. 3). To account for different precision of origin least square means, the least square means (LSM) were weighted by the reciprocal of their variance in the procedure REG in SAS (SAS Inc. 2015). The regressions showing lowest Criterion of Akaike (AIC) were chosen (Burnham and Anderson, 2002). Variation between origins All origins were mostly leafless during the dry season and with foliage during the rainy season (Fig. 4). Analyses of variance showed significant (P < 0.001) differences among origins in the timing of senescence in 2012 (Table 2) with mean date ranging among the origins from 25th of august 2012 (day 238) to 20th of November 2012 (day 325), (Table 1). When assessed as change to phase 0, the origins ranged from 9th of September 2012 (day 253) to 6th of December 2012 (day 341), i.e. a difference of 88 days (Table 1). Origins with early senescence were shedding the leaves at a time when there was still considerable rainfall, while origins with late senescence shed leaves long time after rainfall had stopped at the trial site (for rainfall at trial site see Fig. 2). The majority of the origins initiated bud burst around the time where the first small rain showers were recorded, but some individual trees set new leaves by the end of March before the first rain occurred in April (Fig. 4). The mean date of bud burst (estimated date for change from either phase 0 or 1 to a higher phase) did not vary significantly among origins at a 5% level (P < 0.090 and P < 0.065, respectively; Table 2). Climatic clines in timing of senescence Among origins from the northern hemisphere, the day of senescence was significantly earlier among populations from drier climates, i.e. with highly negative cumulated DI from AprileJuly/MayeJuly (Table 3, Fig. 5a). DI estimates for September, October and November at the start of the dry season at the sites of origins were not, or only vaguely significant as explanatory variables with P < 0.06, P < 0.03 and P < 0.07, respectively. Latitude proved significant as explanatory variable (Table 3). Latitude and DI at the sites of origins north of Equator are highly confounded with a correlation between latitude and DI AprileJuly of 0.87 and between latitude and DI MayeJuly of 0.84. For origins south of the Equator, DI in October (start of the rainy season at the sites of origins), showed the highest significant association with the timing of senescence (Table 3, Fig. 5b). DI estimates for April, May and June at the start of the dry season at the sites of origins were not significant as explanatory variables with P < 0.41, P < 0.17 and P < 0.18, respectively. Neither latitude, nor longitude was significant as explanatory variable, but longitude was at the borderline (Table 3). The correlation between longitude and DI in October was 0.68 so the DI in October is to some extent confounded with longitude. Overall, the main result is that origins from drier sites, in this case expressed in terms of DI in spring or autumn, show earlier senescence. Timing of senescence This study is to our knowledge the first to show that the timing of leaf senescence of A. digitata is under genetic control. The results are based on a single year of observations at one site, and more studies are required to test for genotype by year Fig. 3. Monthly precipitation estimates for the origins from the northern (a) and southern (b) hemispheres, monthly DI estimates (precipitation minus potential evapotranspiration) for the origins from the northern (c) and southern (d) hemispheres and monthly mean temperatures for the origins from the northern (e) and southern (f) hemispheres. Climate estimates from University of East Anglia Climatic Research Unit, Climate Research Unit Jones and Harris, 2014). Vertical lines equals the equinox in spring and autumn, respectively. Legend numbers refer to the origin numbers in Table 1. and genotype by environment interactions and the observed phenology-climate clines. Nevertheless, the clear relationship between timing of senescence and climatic parameters at the site of origin supports that the patterns are generated by local adaptation. More studies are also needed to clarify to what degree the adaptive patterns are due to epigenetics as observed recently for several tree species (see Sow et al., 2018 for a recent review). Genetic differences in timing of bud burst and senescence was reported in Parkia biglobosa among West African origins tested in Burkina Faso, showing a north-south cline in senescence (Ouedrago, 2014). Although studies on more species are needed, the two studies collectively indicate substantial variation in the timing of leafing senescence among origins of semiarid species, and that such variation is linked to adaptation to the prevailing climates at the origins. Table 2 Results from the test of differences between origins concerning the day with change in phenology phase. It remains uncertain if the timing of leaf senescence is induced by day-length, temperature, reduced soil water availability, or other factors. However, reduced water availability can possibly be ruled out for the origins starting to show leaf senescence already at the end of August (day 238 in 2012, Table 2), where the rainfall was still reasonably high at the test site (Fig. 2). Interestingly, Korbo et al. (2013) found variation in leaf production among A. digitata origins along the course of the year in a trial where plants were continuously irrigated. This indicates that timing of leaf production is not simply controlled by soil water availability in A. digitata. A decrease in leaf production despite irrigation was also observed among seedlings of species from drier climates in Nigeria (Njoku, 1964). Populations from drier sites of the Central American tropical dry forest species Quercus oleoides, also shed their leaves earlier during a drought treatment (Ramírez-Valiente and Cavender-Bares, 2018). Conversely, among irrigated trees of the arid savanna tree species Colophospermum mopane, there was a weak seasonality in leaf area index compared with water stressed trees (Stevens et al., 2016). Studies on Madagascan Adansonia species indicate that the trees use water stored in the parenchyma rich wood to support leaf flushing (Chapotin et al., 2006). The early development requires that the trees are not losing excessive amounts of water from the stems after the end of the rainy season due to the transpiration. The large genetic differences in timing of senescence may therefore reflect that trees from drier regions retrieve nutrients and shed their leaves in time to secure that the water Table 3 Results from the regression analyses of relations between changes in phenology phases and climate of origins. Origins Year reserves in the stems are sufficiently large to support leaf flushing the next season. This is a hypothesis that deserves to be tested in further studies. Bud burst No firm conclusion can be made concerning the genetic variation in bud burst from this study. It might be that the occurrence of first rain is an important trigger of budburst in A. digitata and that early rain at the test site synchronized the timing of the bud burst among the different origins. Yet, in our study, a few trees set leaves in March before the first occurrence of rain, and the coincidence between the spring equinox and start of leaf flushing before the start of the rain makes it tempting to suggest a role for daylength as a determining factor for the start of bud burst. In the study of Di Lucchio et al. (2018), longer photoperiods resulted in higher levels of meristematic activity, also suggesting a role for daylength in relation to bud break. Even at small distances from the Equator, there are differences in daylength along the course of the year and the difference between the shortest and the longest day exceed 1 h at many of the sites of the origins (Table 1). Early leaf flushing is an important strategy to benefit from early rainfall and to make leaves ready to start photosynthesizing when the rainy season starts (Chapotin et al., 2006). Bud burst starts earlier than the rainy seasons over several vegetation types and years in the tropics, suggesting that bud burst is not controlled by rainfall (Swanepoel, 1993;Ryan et al., 2017). The role of daylength for tropical tree phenology constitutes an interesting topic for follow up research (Borchert et al., 2005) and additional studies with control of day length will allow a specific test of the hypothesis. Adaptation in semi-arid areas and adaptation potential Tree species in dry seasonal climates can be divided into groups according to their phenology: deciduous species with short or long leafing periods, semi-evergreen species which shed their leaves over a short period, and evergreen species in which old leaves are still on the tree when the new leaf flush occurs (Seghieri et al., 2012). While A. digitate represents a deciduous species, there may be reason to extend the study to other species representing other areas of distribution and representing other phenology groups. To our knowledge, few studies on genetic variation in phenology from arid and semiarid areas are based on experiments where population and environmental effects are not confounded, emphasizing that we have only started to understand how species may have adapted their phenology to the prevailing climates. An important question is also to what extent observed variations in phenology among populations are due to epigenetics as observed in temperate tree species (Br€ autigam et al., 2013;Sow et al., 2018). In times with climatic changes the degree of phenotypic plasticity including epigenetics and speed of adaptation in phenology is crucial. We do not yet know to what degree the observed differences in phenology of A. digitate influence fitness, but the trial will be followed in the future to help in determining whether e in response to climatic changes -it will be advisable to resort to assisted migration to increase the genetic variation in the species, as suggested for temperate trees (Aitken and Bemmels, 2016). In this context, the quite large genetic variation in leaf senescence shown in this study combined with an indicated limited gene flow could make A. digitata populations vulnerable to climate change.	v3-fos-license
2023-04-09T15:08:57.561Z	2023-04-07T00:00:00.000	258036428	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://onlinelibrary.wiley.com/doi/pdfdirect/10.1002/jha2.687", "pdf_hash": "d69598f50ef8927494cd2e2fe356c1edcec9d2bd", "pdf_src": "Wiley", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45419", "s2fieldsofstudy": [ "Biology" ], "sha1": "64aeb027dfad3d72cbbf29dbdc5812b77b9c3d70", "year": 2023 }	pes2o/s2orc	A phase 1 first‐in‐human study of GS‐0189, an anti‐signal regulatory protein alpha (SIRPα) monoclonal antibody, in patients with relapsed/refractory (R/R) non‐Hodgkin lymphoma (NHL) Abstract Signal regulatory protein alpha (SIRPα) is the receptor for cluster of differentiation (CD)47, a potent “don't eat me” signal for macrophages. Disruption of CD47‐SIRPα signaling in the presence of prophagocytic signals can lead to enhanced phagocytosis of tumor cells, resulting in a direct antitumor effect; agents targeting this pathway have shown efficacy in non‐Hodgkin lymphoma (NHL) and other tumor types. GS‐0189 is a novel anti‐SIRPα humanized monoclonal antibody. Here we report: (1) clinical safety, preliminary activity, and pharmacokinetics of GS‐0189 as monotherapy and in combination with rituximab from a phase 1 clinical trial in patients with relapsed/refractory NHL (NCT04502706, SRP001); (2) in vitro characterization of GS‐0189 binding to SIRPα; and (3) in vitro phagocytic activity. Clinically, GS‐0189 was well tolerated in patients with relapsed/refractory NHL with evidence of clinical activity in combination with rituximab. Receptor occupancy (RO) of GS‐0189 was highly variable in NHL patients; binding affinity studies showed significantly higher affinity for SIRPα variant 1 than variant 2, consistent with RO in patient and healthy donor samples. In vitro phagocytosis induced by GS‐0189 was also SIRPα variant–dependent. Although clinical development of GS‐0189 was discontinued, the CD47‐SIRPα signaling pathway remains a promising therapeutic target and should continue to be explored. phagocytosis induced by GS-0189 was also SIRPα variant-dependent. Although clinical development of GS-0189 was discontinued, the CD47-SIRPα signaling pathway remains a promising therapeutic target and should continue to be explored. K E Y W O R D S CD47, GS-0189, SIRPα, non-Hodgkin lymphoma, monoclonal antibodies INTRODUCTION Non-Hodgkin lymphoma (NHL) is one of the most common cancers in the United States, accounting for about 4% of all cancers [1]. Systemic treatment options include chemotherapy, immuno-and targeted therapy, chimeric antigen receptor (CAR) T-cell therapy, and stem cell transplant [2]. Many patients have relapsed/refractory (R/R) disease following frontline treatment for NHL [3][4][5][6]. The overall prognosis and long-term survival for R/R patients after multiple lines of therapy are often poor [7][8][9]. CAR T-cell therapy provides impressive response rates in heavily pretreated R/R follicular and other B-cell NHLs, but toxicities, complicated logistics, limited access, and delays in delivering treatment may limit use of this therapy in many patients [10][11][12]. The development of more effective and tolerable therapies for R/R NHL represents a high unmet medical need. GS-0189 was designed as an alternative to magrolimab, an anti-CD47 monoclonal antibody (mAb) in clinical development in hematologic malignancies and solid tumors, as it had been found not to impact red blood cells in preclinical experiments [23]. Here we report: (1) clinical safety, preliminary activity, and pharmacokinetics (PK) of GS-0189 as monotherapy and combination therapy with rituximab, from a phase 1 first-in-human (FIH) clinical trial in patients with R/R NHL (NCT04502706); (2) in vitro characterization of GS-0189 binding potency to SIRPα variants; (3) and in vitro phagocytic activity of macrophages derived from donors possessing different SIRPα variants relative to the activity of a control pan-SIRP blocking antibody with an inert Fc, KWAR23. Design and participants This was an open-label FIH trial to evaluate GS-0189 safety, PK, and preliminary efficacy of monotherapy and in combination with rituximab in patients with select R/R NHL histologies (Supplemental Methods; Figure S1): here, we report monotherapy dose escalation Endpoints The primary endpoint was incidence of adverse events defined by National Cancer Institute Common Terminology Criteria for Adverse Events, Version 5.0. Secondary endpoints included GS-0189 PK, objective response rate per Lugano response criteria [24], duration of response (DOR), and progression-free survival (PFS). Statistical analysis Statistical analyses were performed on all patients who received ≥1 dose of GS-0189. For categorical variables, frequencies and percentages were calculated. Nonclinical and ex vivo assays Antibodies and reagents used in these studies are listed in Table S1 GS-0189 and KWAR23 binding to SIRPα variants Binding experiments quantifying the affinity of GS-0189 or KWAR23 for recombinant SIRPα variants were performed on either a Biacore T100 or T200 instrument using a CM5 sensor chip. GS-0189 and KWAR23 as active control [25,26] were captured and regenerated according to manufacturer instructions. SIRPα V1 and SIRPα V2 were injected using maximum concentrations of either 0.3 μM and three-fold serial dilutions for the higher affinity interactions, or 3 μM and 4-fold serial dilutions for the lower affinity interactions. Data were fitted to a simple kinetic model to derive k on , k off , and K D , using the relationship K D = k off / k on . 2.3.3 In vitro phagocytosis assay Clinical study results Nine patients were enrolled and treated between December 2, 2020 and Table 1. Reasons for GS-0189 discontinuation were patient's decision (n = 2) and progressive disease (n = 7). Reasons for study discontinuation were death (n = 1), withdrawn consent (n = 2), and study terminated by sponsor (n = 6). at the time of last adequate tumor assessment. Study treatment was discontinued due to patient decision, which coincided with the time of sponsor decision to discontinue the study. Two patients had stable disease (1 each, from MDE2 and CDE1) with estimated PFS of 5.55 and 3.71 months, respectively; 5 had progressive disease (Table 4); and 1 withdrew consent prior to the first response assessment. Pharmacokinetics Concentration-time profiles for patients in MDE1, 2, and 3 and CDE1 are shown in Figure 1. Relevant PK parameters for MDE and CDE (Cohort 1) are listed in Table 5. The area under the curves for GS-0189 serum concentration-time profiles in the given dose range of 10 to 100 mg were lower than projected based on cyno PK-PD studies. This could potentially be due to target-mediated drug disposition for the antibody in humans. GS-0189 RO over time from three patients in CDE1 is shown in Figure S2 for comparison with concentration-time profile. The antidrug antibody incidence rate for GS-0189 cannot be interpreted due to a small sample size. TA B L E 5 Median pharmacokinetic parameters for GS-0189 post first dose (cycle 1, day 1). GS-0189 and KWAR23 binding to SIRPα variants Binding affinities of GS-0189 to recombinant SIRPα V1 and SIRPα V2 compared with those of KWAR23 are shown by kinetic and equilibrium constants (Table 6) Antibody Antigen values for k on ,k off , and K D varied less than 2-fold between experiments. 2 The equilibrium dissociation constant K D = k off /k on . F I G U R E 3 GS-0189 binding to SIRP isoforms and variants shown by sensogram for GS-0189 (A and B) or KWAR23 (C and D) binding to SIRPα V1 (A and C), and SIRPα V2 (B and D). The highest concentration injected was 3 μM for lower affinity interactions, and 0.3 μM for higher affinity interactions for SIRPα V1 and SIRPα V2 for GS-0189. The highest concentration injected was 0.3 μM for all SIRP variants for KWAR23. Black lines denote binding data; orange lines represent the kinetic fit. Abbreviations: SIRPα, signal regulatory protein alpha. Potency of GS-0189 in phagocytosis assays is dependent on SIRPα polymorphism Maximal phagocytosis of Raji cells by macrophages from PBMC donors with SIRPα V1/V1 and SIRPα V2/V2 variants was induced by GS-0189 combined with rituximab ( Figure S3), with only a small increase in maximal phagocytosis induced by the combination versus rituximab alone. Therefore, we identified a human colorectal cancer cell line DLD-1 for which phagocytosis was more dependent on inhibition of the CD47-SIRPα axis and for which phagocytosis was significantly induced in response to CD47-SIRPα blockade alone [25]. The ability of GS-0189 to potentiate phagocytosis of DLD-1 cells by human PBMCderived macrophages from donors with different SIRPα variants was dose dependent (Figure 4). The average half-maximal effective concentration (EC 50 ) of GS-0189 with macrophages from donors expressing SIRPα V1/V1 was 0.02 μg/mL, expressing SIRPα V1/V2 was 13.8 μg/mL, and expressing SIRPα V2/V2 was 9.5 μg/mL. KWAR23 induced phagocytosis across SIRPα variants with EC 50 ranging from 0.04 to 0.10 μg/mL ( Figure S4). Regardless of SIRPα genotype, maximal phagocytosis induced by GS-0189 was equivalent to that induced by KWAR23 at concentrations above 10 μg/mL (Figure 4). DISCUSSION GS-0189 up to 100 mg as monotherapy and in combination with rituximab was well tolerated by patients with R/R NHL in this phase 1 study. There were no DLTs, no grade 4 TEAEs, and no deaths related to TEAEs. Since SIRPα expression is mainly on macrophages and GS-0189 does not have a functional Fc, it was theorized that GS-0189 would result in less anemia compared to most CD47-targeting agents [23,27]. In SRP001, anemia occurred in 2 patients, 1 of whom had anemia at baseline. Lymphocyte count decreases observed with GS-0189 in combination with rituximab were transient and resolved quickly. Although lymphopenia has been reported with rituximab monotherapy [28], the causality of this phenomenon will remain unclear unless a randomized clinical study is conducted to evaluate each drug's contribution. GS-0189 is an anti-SIRPα antibody with an aglycosylated (inert) Fc region that was theorized to benefit from the presence of another drug that provides an "eat me" signal, such as rituximab, to induce phago- GS-0189 was intended as a low anemia-risk alternative to magrolimab. With priming-dose regimen and clinical management, acute anemia observed with magrolimab is no longer expected to limit magrolimab clinical development [29,30]. Therefore, the decision was made to terminate the SRP001 study and discontinue clinical development of GS-0189. Nevertheless, the CD47-SIRPα interaction/phagocytic mechanism remains a promising target for treatment of patients with solid tumors and hematologic malignancies and should continue to be explored. AUTHOR CONTRIBUTIONS MN, NLB, SI, LP, AS, JB, VG, TC, and MP conceived of or designed the study. MN, LP, JB, YL, and MP acquired and provided data. All authors analyzed or interpreted the data, drafted, or critically reviewed the manuscript, approved the final version, and agreed to be accountable for all aspects of the work. ACKNOWLEDGMENTS We thank the patients who participated in the clinical study and their families, the study coordinators, and the support staff at the clinical sites. We would also like to acknowledge the contributions of the GS-	v3-fos-license
2019-04-17T15:38:39.454Z	1977-11-25T00:00:00.000	122439470	{ "extfieldsofstudy": [ "Mathematics" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.annalsofgeophysics.eu/index.php/annals/article/view/4832/4879", "pdf_hash": "79caadf98826847d27d07a37d878de23b34c12cd", "pdf_src": "Adhoc", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45420", "s2fieldsofstudy": [ "Geology", "Physics" ], "sha1": "96434f78ebdeee7da840581c565079ea9414f9e6", "year": 1977 }	pes2o/s2orc	The pattern of eigenfrequencies of radial overtones which is predicted for a specified Earth-model In 1974 Anderssen and Cleary examined the distribution of eigenfrequencies of radial overtones in torsional oscillations of Earth-models. They pointed out that according to Sturm-Liouville theory this distribution should approach asymptotically, for large overtone number m, the value nnz/y, where y is the time taken by a shear-wave to travel along a radius from the core-mantle interface to the surface, provided elastic parameters vary continuously along the radius. They found that, for all the models which they considered, the distributions of eigenfrequencies deviated from the asymptote by amounts which depended on the existence and size of internal discontinuities. Lapwood (1975) showed that such deviations were to be expected from Sturm-Liouville theory, and McNabb, Anderssen and Lapwood (1976) extended Sturm-Liouville theory to apply to differential equations with discontinuous coefficients. Anderssen (1977) used their results to show how to predict the pattern of deviations —called by McNabb et al. the solotone effect— for a given discontinuity in an Earth-model. Recently Sato and Lapwood (1977), in a series of papers which will be referred to here simply as I, II, III, have explored the solotone effect for layered spherical shells, using approximations derived from an exacttheory which holds for uniform layering. They have shown how the form of the pattern of eigenfrequencies, which is the graph of S — YMUJI/N — m against m, where ,„CJI is the frequency of the m"' overtone in the I"' (Legendre) mode of torsional oscillation, is determined as to periodicity (or quasi-periodicity) by the thicknesses and velocities of the layers, and as to amplitude by the amounts of the discontinuities (III). The pattern of eigenfrequencies proves to be extremely sensitive to small changes in layer-thicknesses in a model. In this paper we examine a proposed Earth-model with six surfaces of discontinuity between core boundary and surface, and predict its pattern of eigenfrequencies. When seismological observations become precise enough, and can be subjected to numerical analysis refined enough, to identify the radial overtones for large m, this pattern of eigenfrequencies will prove to be a severe test for any proposed model, including he one which we discuss below. theory which holds for uniform layering. They have shown how the form of the pattern of eigenfrequencies, which is the graph of S -YMUJI/N -m against m, where ,"CJI is the frequency of the m"' overtone in the I"' (Legendre) mode of torsional oscillation, is determined as to periodicity (or quasi-periodicity) by the thicknesses and velocities of the layers, and as to amplitude by the amounts of the discontinuities (III). The pattern of eigenfrequencies proves to be extremely sensitive to small changes in layer-thicknesses in a model. In this paper we examine a proposed Earth-model with six surfaces of discontinuity between core boundary and surface, and predict its pattern of eigenfrequencies. When seismological observations become precise enough, and can be subjected to numerical analysis refined enough, to identify the radial overtones for large m, this pattern of eigenfrequencies will prove to be a severe test for any proposed model, including he one which we discuss below. PEM-A. We select for our purpose the model PEM-A (parametrically simple Earth-model A), presented by Dziewonski, Hales and Lapwood (1975), which has the following values at and between discontinuities: [l] where pis density and /?,• is shear velocity adjacent to the (/, /+1) interface in the j' h layer and p' ;+ 1,/?',•+1 are adjacent to the same interface in the (/'+1)"' layer. Since density and shear velocity increase downwards in PEM-A, except at the (4, 5) boundary, all reflection coefficients but one are negative. THE PATTERN OF EIGENFREQUENCIES. Lapwood and Sato proved (III, § 2) that for a shell of three uniform layers, the j"' being bounded by spherical surfaces r=r,-i and r = r/, a measure of the solotone effect is given for any co by where R, is the reflection coefficient at normal incidence of a ray from region (/') on the (/, y'+l) interface, and Xi is the change of phase as a shear-wave passes through the j"' layer. Xi = CJ Xi'< where A and T for any arc being the angle subtended at the centre of the sphere and the time taken on the arc respectively. As shown in the Appendix, the extension of the formula [3] to a shell of n layers is Ri . 0 Ri . 0 -sin 2 xi sin 2 xs, where r= ^ xi- We have computed S from [5] for 1 = 2 and m= 1,2,... , 150. The values of S are graphed against m in Fig. 1. Our earlier work has shown that S is rather insensitive to the value of / (III, § § 1,2), so that the pattern of eigenfrequencies which emerges from Fig. 1 may be said to characterise the radial overtones of all modes of torsional oscillations of the model. We observe immediately that in Fig. 1 there is a pattern, of period 10 in m, which changes slowly as m proceeds from 1 to 150. We have taken this large range of m to exhibit another period of S, which is shown in the long sinusoidal arc of period about 110. We now look for the explanation of these two recurrence periods. If into the general formula [5] we insert the values defined for the Earth-model PEM- The coefficients of the sines are R\, Ri,... , Rt respectively. The largest amplitudes are those of the fifth and first terms, while the contribution of the third term is very small. If we now ask what lowest values of m will produce in the successive arguments of the sines almost exact multiples of 2TT, we find respectively 10, 10, 10, 80, 105, 150. [8] Thus the first three terms of [7] have the same sign and almost the same period in m, and together account for the dominant period of 10 in m in the pattern of eigenfrequencies shown in Fig. 1. The first term contains R\, which is the coefficient of reflection at the 667 km discontinuity; the second contains R 2 , which is the coefficient of reflection at the 417 km discontinuity, while the third contains R3, corresponding to a small discontinuity at 217 km. The fifth term has Rs as factor, corresponding to reflection at the base of the crust. Thus the shape of the pattern of eigenfrequencies seems to be determined mainly -in fact almost completely -by resonances among reflections at the 667 km and 417 km discontinuities and the Moho. 2 shows the graph of S* against m. We see that the essential shape of the small pattern of period 10 in Fig. 1 is reproduced in Fig. 2, though Fig. 2 inevitably shows no evidence of the longer arc on which, the small patterns lie in Fig. 1. Fig. 3 shows the values of S' lying on a sine curve of period 105 in m. In the same figure are plotted averages of eigenvalues in successive sets of 10. The locus of averages eliminates the small patterns. There is very good agreement between the two graphs. One would not expect perfect agreement, since the period 10 is not exact for any of the first three terms of [7]. The lengthening of the recurrence period in the graph of averages to about 112 is probably mainly due to the last term of [7\|, which arises from reflection within the crust. In the formulae [5] and [7] multiple internal reflections have been neglected. In [9] and [10] radial travel times have been used instead of Xp of [5]. For these reasons our exposition must be seen as only a first order approximation. Nevertheless, since the second order effects are all small we may be confident that even when superposed they will not affect our main conclusions: in our previous papers we compared miz/y + S with m wi computed with great precision from exact frequency equations, and found excellent agreement. CONCLUSIONS. We have found the pattern of eigenfrequencies for the Earth-model PEM-A, and have shown that it is determined mainly by the depth of the Moho and the depths of discontinuities in the neighbourhood of 400 km and 650 km. When it becomes possible, with increased accuracy of observation and increased delicacy of computational analysis, to identify and measure the higher overtones of torsional oscillations of the Earth itself, the predicted pattern of eigenfrequency can be put to the test. Then the model will receive additional strong support, or else will be shown to need modification. ACKNOWLEDGEMENT. This paper is dedicated with our sincere congratulations to Professor Caloi. It was written when E. R. L. was a guest at the Earthquake Research Institute, University of Tokyo, being supported by a grant from the Royal Society of London. APPENDIX. McNabb, Anderssen and Lapwood have shown (1976) that the asymptotic form of the frequency equation for a Sturm-Liouville system with discontinuous coefficients is obtained as the condition that the following set of homogeneous linear equations is consistent: n j=i r p-j+i	v3-fos-license
2021-04-04T06:16:23.030Z	2021-03-28T00:00:00.000	232772684	{ "extfieldsofstudy": [ "Chemistry", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1999-4923/13/4/459/pdf", "pdf_hash": "13f3e44d134540b61c180b91fb977405a979cff5", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45421", "s2fieldsofstudy": [ "Medicine", "Chemistry" ], "sha1": "819c0d7f8d9cb97c31f13f89713ad8681b7744ce", "year": 2021 }	pes2o/s2orc	Oligonucleotide Delivery across the Caco-2 Monolayer: The Design and Evaluation of Self-Emulsifying Drug Delivery Systems (SEDDS) Oligonucleotides (OND) represent a promising therapeutic approach. However, their instability and low intestinal permeability hamper oral bioavailability. Well-established for oral delivery, self-emulsifying drug delivery systems (SEDDS) can overcome the weakness of other delivery systems such as long-term instability of nanoparticles or complicated formulation processes. Therefore, the present study aims to prepare SEDDS for delivery of a nonspecific fluorescently labeled OND across the intestinal Caco-2 monolayer. The hydrophobic ion pairing of an OND and a cationic lipid served as an effective hydrophobization method using either dimethyldioctadecylammonium bromide (DDAB) or 1,2-dioleoyl-3-trimethylammonium propane (DOTAP). This strategy allowed a successful loading of OND-cationic lipid complexes into both negatively charged and neutral SEDDS. Subjecting both complex-loaded SEDDS to a nuclease, the negatively charged SEDDS protected about 16% of the complexed OND in contrast to 58% protected by its neutral counterpart. Furthermore, both SEDDS containing permeation-enhancing excipients facilitated delivery of OND across the intestinal Caco-2 cell monolayer. The negatively charged SEDDS showed a more stable permeability profile over 120 min, with a permeability of about 2 × 10−7 cm/s, unlike neutral SEDDS, which displayed an increasing permeability reaching up to 7 × 10−7 cm/s. In conclusion, these novel SEDDS-based formulations provide a promising tool for OND protection and delivery across the Caco-2 cell monolayer. Introduction Gene therapy offers highly specific therapeutics without significant side effects. It covers a wide variety of approaches, ranging from the replacement of malfunctioning genetic information, often in the form of a large nonviral plasmid DNA (pDNA), to the currently more investigated oligonucleotide (OND)-based therapy acting on the level of mRNA destabilization or translation [1,2]. As for OND-based therapy, synthetic 20-30-mer nucleotides connected by phosphodiester bonds form a polyanionic backbone that determines the physicochemical properties of this hydrophilic macromolecule independently of the nucleotide sequence [3]. Therefore, there should be versatility in formulating OND for various applications. Nevertheless, the nucleotide structure is unstable in the presence of nucleases. Rapid enzymatic degradation combined with insufficient cellular uptake, often followed by entrapment in endosomes, makes the delivery of OND challenging [4]. Nonviral drug delivery systems, mainly lipid-based nanoformulations, have emerged as a promising option to overcome not only these limitations but, also, to enable oral delivery [5]. Oral delivery is the preferred route of administration-in particular, in the case of local gastrointestinal diseases. Many of these pathologies are defined or accompanied by intestinal inflammation, such as inflammatory bowel disease [6,7]. Novel colloidal drug delivery systems benefit from the enhanced permeation and retention effect of the inflamed tissue, which leads to increased accumulation of nanocarriers (~100-500 nm) within the leaky inflammatory site, which is also accompanied with an altered mucus layer [8,9]. Moreover, the increased accumulation of nanocarriers enables their uptake by phagocytic immune cells invading the area affected by inflammation; their rapid elimination due to diarrhea commonly present is avoided. In addition to the size of carriers, negative surface charge and hydrophilic surface decoration are also strategies utilized for passive targeting to the inflamed intestinal tissue [9][10][11]. The local delivery of OND has been investigated using many kinds of polymeric nanoparticles [12], lipid nanoparticles [13], microencapsulated nanogel [14], thioketal nanoparticles [15] and nanoparticles-in-microspheres multicompartment systems [16]. Self-emulsifying drug delivery systems (SEDDS) are isotropic mixtures of oils, surfactants, cosurfactants and cosolvents forming oil-in-water nanoemulsions upon gentle dispersion in an aqueous environment, e.g., gastrointestinal fluids [17]. These oral lipidbased drug delivery systems were originally established to improve the bioavailability of small poorly water-soluble molecules [18]. Recently, the delivery of hydrophilic macromolecules in SEDDS has gained increasing attention in order to take advantage of its beneficial properties, such as easy upscaling, nanosize of formed droplets and protection of the loaded substance from chemical and enzymatic degradation [19]. Nevertheless, the hydrophilicity of a potential hydrophilic substance needs to be reduced, usually by hydrophobic ion pairing [19]. The hydrophobic ion pairing of nucleic acids is a method based on replacing the counterions associated with negatively charged phosphate groups with surfactants carrying the positive charge. This method leads to a lipophilicity increase of the resulting ion-paired complexes [20]. For this purpose, positively charged surfactants, such as dimethyldioctadecylammonium bromide (DDAB) and 1,2-dioleoyl-3trimethylammonium propane (DOTAP), are frequently utilized [21][22][23][24]. In addition, these quaternary ammonium salts enhance endosomal escape so the nucleic acid can reach the cytosol [25,26]. ONDs as hydrophilic macromolecules use predominantly paracellular transport through the intestinal monolayer; therefore, their permeability under physiological conditions is very limited [27]. Oral formulations containing intestinal permeation enhancers (PEs) have been widely investigated to enhance oral delivery of hydrophilic macromolecules [28]. PEs based on medium-chain fatty acids (MCFA) have been found to enhance paracellular transport via the opening of tight junctions (TJs) in a reversible manner by interaction with the cytoskeleton [29,30]. In addition, at higher concentrations the action of MCFA might be supported by transcellular permeation as a result of cell membrane alterations [28]. Hauptstein et al. pioneered the delivery of nucleic acids in SEDDS. In their research, several hydrophobic pDNA-cationic lipid complexes were tested. Among the tested cationic lipids, the complexes with cetrimide delivered in SEDDS showed a good transfection efficiency of HEK-293 cells comparable to Lipofectin ® , a gold standard transfection reagent [31]. Mahmood analogical SEDDS by the incorporation of a cell-penetrating peptide and confirmed an internalization of 50-nm nanoemulsions in Caco-2 cells [32]. Both studies focused on the delivery of large pDNA into cells. In contrast to considerably larger pDNA molecules, this study focuses on the delivery of short OND sequences as emerging, highly potent therapeutics. The aim of this study was to prepare and characterize MCFA-based SEDDS loaded with hydrophobized OND and to investigate the ability of this system to deliver OND across the intestinal Caco-2 monolayer. This formulation approach has a considerable potential to overcome OND low stability and poor permeability. In this study, firstly, hydrophobized complexes of cationic lipids (DDAB or DOTAP) and a model fluorescently labeled 20-mer OND were thoroughly described and subsequently loaded into SEDDS (Figure 1). Dispersed SEDDS (negatively charged or neutral SEDDS) were characterized in terms of size, zeta potential, lipolysis, protective effect against nucleases and OND permeability. Pharmaceutics 2021, 13, x 3 of 27 efficiency of HEK-293 cells comparable to Lipofectin ® , a gold standard transfection reagent [31]. Mahmood et al. improved the transfection efficiency of pDNA-cetrimide in analogical SEDDS by the incorporation of a cell-penetrating peptide and confirmed an internalization of 50-nm nanoemulsions in Caco-2 cells [32]. Both studies focused on the delivery of large pDNA into cells. In contrast to considerably larger pDNA molecules, this study focuses on the delivery of short OND sequences as emerging, highly potent therapeutics. The aim of this study was to prepare and characterize MCFA-based SEDDS loaded with hydrophobized OND and to investigate the ability of this system to deliver OND across the intestinal Caco-2 monolayer. This formulation approach has a considerable potential to overcome OND low stability and poor permeability. In this study, firstly, hydrophobized complexes of cationic lipids (DDAB or DOTAP) and a model fluorescently labeled 20-mer OND were thoroughly described and subsequently loaded into SEDDS (Figure 1). Dispersed SEDDS (negatively charged or neutral SEDDS) were characterized in terms of size, zeta potential, lipolysis, protective effect against nucleases and OND permeability. Complex Preparation The OND and a cationic lipid (DDAB or DOTAP) were dissolved in 1 mL of Bligh-Dyer monophase consisting of chloroform:methanol:water 1:2.2:1, as previously described [33]. The monophase was separated into a two-phase system by the addition of an aliquot of 250 µL of water and chloroform. Subsequently, the mixture was vigorously vortexed for 1 min and centrifuged at 600× g for 5 min to achieve complete phase separation. The chloroform phase containing OND complexed with DDAB and DOTAP was collected, and the solvent was evaporated under nitrogen flow. The amount of complexed OND was quantified indirectly from the amount of noncomplexed OND in the aqueous phase. The fluorescence of FAM-labeled OND was measured in the aqueous phase at the excitation and emission wavelengths of 495 nm and 520 nm, respectively, in a microplate reader (FLUOstar Omega, BMG Labtech, Ortenberg, Germany). Complexation efficiency (CE) was calculated as follows: Various charge ratios (+/−) of the cationic ammonium head of a cationic lipid and the anionic phosphate group of the OND backbone were tested in order to achieve efficient CE (>95%), starting at the ratio 1:1 and increasing the amount of cationic lipid. Prepared complexes were stored at −20 • C until further use [22,23]. To evaluate the changes introduced by complexation of a cationic lipid and OND, a physical mixture was prepared at the charge ratio of cationic lipid:OND 3:1. The required amount of a cationic lipid was weighed in a 4 mL screw thread glass vial (Thermo Fisher Scientific (Waltham, MA, USA). It was dissolved in a volume of chloroform sufficient to completely dissolve the lipid. Chloroform was subsequently evaporated under a nitrogen stream. The corresponding amount of OND solution was added into the glass vial, and the solvent was evaporated under the nitrogen stream. Solid lipid and OND were mixed into a physical mixture. Effect of SEDDS Lipid Excipents on Complex Stability Cationic lipid-OND complexes were prepared in the Bligh-Dyer monophase (chloroform:methanol:water 1:2.2:1) and two phases separated with an aliquot of water and chloroform, as described in Section 2.2 [23]. The effect of the lipid excipients (Captex 300, Labrasol, LPC, Citrem, Maisine CC and Peceol) in the SEDDS on the complex stability was studied. A tested SEDDS lipid excipient in the amount relevant to SEDDS composition was dissolved in chloroform and added directly to the chloroform phase containing preformed complexes. As a control, pure chloroform was added. Water:methanol solution (ratio 1:1) was added to the aqueous layer in order to keep the ratio of the top and bottom layer constant. The system was vigorously vortexed for 1 min and subsequently centrifuged at 600× g for 5 min. The amount of OND in the aqueous phase was determined by fluorescence, as described in Section 2.2. The CE of OND was calculated by Equation (1) and is indicated relative to the control (no added lipid). Atomic Force Microscopy (AFM) The morphology of OND and complexes was investigated by AFM imaging. AFM was performed on the Park NX20 (Park Systems, Suwon, South Korea) with tapping mode in the air, using PointProbe®Plus Non-Contact/Tapping Mode -High Resonance Frequency -Reflex Coating (known as PPP-NCHR) probes with a nominal tip radius of 7 nm. Samples deposited on freshly cleaved mica were used. For proper fixation of the nucleic acid, a divalent cation Mg 2+ (10 mM) to electrostatically bridge the negatively charged surface of mica substrate to the anionic OND backbone was used [34,35]. Therefore, the mica was treated with 10-mM MgSO 4 for 5 min before the OND solution was applied. Ten microliters of 1-nM OND solution was allowed to attach for 5 min, rinsed with DI water and dried under nitrogen flow. Mica surface treatment was not necessary to absorb lipophilic complexes. The samples of the complexes were initially dissolved in chloroform and subsequently diluted in methanol. Ten microliters of 0.01-nM complex solution in methanol were spin-coated on fresh mica at 1000 rpm. Results were analyzed by SPIP (Image Metrology A/S, Hørsholm, Denmark) using the Particle and Pore analysis module, which enables height and volume analyses of the particles. Differential Scanning Calorimetry (DSC) DSC was performed with the DSC 200 F3 Maia instrument (NETZSCH-Gerätebau GmbH, Selb, Germany). The samples, bulk lipids and their respective complexes were prepared as described in Section 2.2, and the physical mixtures of the same ratio between a cationic lipid:OND 3:1, prepared by mixing, were sealed in standard aluminum crucibles. The samples were cooled down to −50 • C and kept at this temperature under isothermal conditions for 10 min. The analysis was performed at a heating rate 10 K/min in the range from −50 to 150 • C. An empty sealed crucible was used as a reference. The obtained thermograms were evaluated by Proteus Software (NETZSCH-Gerätebau GmbH, Selb, Germany). Attenuated Total Reflectance-Fourier-Transform Infrared (ATR-FTIR) Spectroscopy FTIR spectra were recorded with the MB 3000-PH apparatus equipped with the MIRacle ATR sampling kit and the software Horizon MB (ABB MEASUREMENT & ANALYTICS, Zurich, Switzerland). ATR-FTIR measurement was performed on samples in a solid state placed over ZnSe ATR crystal. Typical spectra were accumulated over the spectral range 4000-400 cm −1 with the coaddition of 64 scans at a resolution of 16 cm −1 in triplicate. Reference spectra were also obtained and subtracted from the raw spectra. Spectra of blank lipids, complexes at different charge ratios, physical mixtures of cationic lipid:OND 3:1 and OND were obtained. In order to observe complex-specific bands, spectra of the respective cationic lipids and OND were subtracted from the spectra of a complex. Preparation of SEDDS Formulations The composition of the utilized SEDDS is depicted in Table 1, as described by Ramakrishnan Venkatasubramanian et al. (data not published, manuscript in preparation). The structure of utilized SEDDS excipients is shown in Figure 2. All excipients were weighed in a glass vial and homogenized by stirring with a magnetic bar at 37 • C overnight. Two SEDDS were used, each differing in the surface charge: negatively charged Citrem SEDDS and neutral Standard SEDDS. The chosen SEDDS were reported to differ only in the surface charge while maintaining a comparable size suitable for targeting in leaky inflammatory lesions. achieved by dissolution of the loaded substance in the SEDDS under stirring with a magnetic bar at 37 °C overnight. The loaded substances and their respective concentrations utilized throughout the study are summarized in Table 2. The amount of loaded cationic lipids was equivalent to the amount present in the ion-paired complexes. Figure 2. Chemical structures of SEDDS excipients (the structure of Labrasol was adapted from [36], and the structure of Citrem was adapted from [37]). Name of Formulation Loaded Substance Concentration of Loaded Substance in SEDDS DAB-OND in SEDDS Complex of DDAB and OND at the charge ratio 3:1 100 nmol of OND/g SEDDS OTAP-OND in SEDDS Complex of DOTAP and OND at the charge ratio 3:1 100 nmol of OND/g SEDDS Figure 2. Chemical structures of SEDDS excipients (the structure of Labrasol was adapted from [36], and the structure of Citrem was adapted from [37]). The SEDDS were loaded with the OND complexed with a cationic lipid, the cationic lipid itself or orlistat. The loaded substance was weighted into a vial. Subsequently, the amount of SEDDS was added to reach the required concentration of 100 nmol of a complex per 1 g of SEDDS, 3.8 mg of DDAB or 4.2 mg of DOTAP per 1 g of SEDDS. Orlistat inhibits the activity of gastrointestinal lipases, and in this study, it was used at a concentration of 0.25% (w/w), prepared as 2.5 mg of orlistat dissolved in 1 g of SEDDS. Loading was achieved by dissolution of the loaded substance in the SEDDS under stirring with a magnetic bar at 37 • C overnight. The loaded substances and their respective concentrations utilized throughout the study are summarized in Table 2. The amount of loaded cationic lipids was equivalent to the amount present in the ion-paired complexes. Characterization of Dispersed SEDDS The dispersions of both Citrem and Standard SEDDS were characterized in terms of size and polydispersity index (PdI) by dynamic light scattering; zeta potential was measured by laser Doppler electrophoresis using Zetasizer Nano ZS (Malvern Instruments, Worcestershire, UK). The dispersions of blank and complex-loaded SEDDS were dispersed 1:100 (w/w) in DI water and 10-mM MES-HBSS buffer (pH 6.5) immediately before the experiment and evaluated at 37 • C. Data were collected in triplicate. Dynamic In Vitro Lipolysis of SEDDS Dynamic one-compartment in vitro lipolysis was carried out under human fasted-state conditions, as previously described [38][39][40]. Nonloaded Citrem and Standard SEDDS both without and with the addition of orlistat 0.25% (w/w) were tested. The addition of orlistat, a lipase inhibitor, into the SEDDS composition has previously been described to inhibit the digestion of SEDDS [41]. SEDDS (0.5 g) was added into a temperature-controlled vessel containing fasted-state simulated intestinal fluid (FaSSIF) used as the lipolysis medium (2.95-mM bovine bile, 0.26-mM soy phospholipids, 2.0-mM maleic acid, 2.0-mM Tris, 50.0-mM NaCl and 1.4-mM CaCl 2 , adjusted to pH 6.5). The digestion of the lipids was initiated by the addition of 5 mL of pancreatin solution with a final activity of 500 USP units/mL. The pancreatin solution was prepared by combining the appropriate amount of porcine pancreatin and adding intestinal lipolysis medium. The resulting suspension was vortexed and centrifuged (7 min, 6500× g). pH 6.5 was maintained by an automated pH stat (Metrohm Titrino 744, Tiamo Version 1.3, Herisau, Switzerland), as the released free fatty acids were continuously titrated by 0.4-M NaOH. The lipolysis was performed for 60 min in three independent experiments (n = 3). Cryogenic Transmission Electron Microscopy The structures formed upon the dispersion of the SEDDS 1:100 (w/w) in FaSSIF (Biorelevant, London, UK) at 37 • C were observed by cryogenic transmission electron microscopy (cryo-TEM). Individual samples (3.5 µL) were loaded on freshly glow-discharged TEM grids (Quantifoil, Cu, 200 mesh, R2/1) inside the Thermo Scientific Vitrobot Mark IV (Thermo Fisher Scientific, Waltham, MA, USA) and plunged frozen into liquid ethane. The Vitrobot chamber was kept at 4 • C and 100% relative humidity during the entire process. Each sample was incubated for 30 s on the grid and blotted for 5 s prior to vitrification. The grids were subsequently mounted into the Autogrid cartridge and loaded into the Thermo Scientific Talos Arctica transmission electron microscope (Thermo Fisher Scientific, Waltham, MA, USA). The microscope was operated at 200 kV, and cryo-TEM data was collected on the Thermo Scientific Falcon 3EC direct electron detector operating in charge integration mode. The micrographs were acquired using SerialEM software at magnifications corresponding to the calibrated pixel sizes of 22.16 Å/px and 3.15 Å/px, respectively. Each micrograph collected at 3.15 Å/px comprised an overall dose of 10 e/Å 2 . Degradation by Nucleases In order to assess the ability of the SEDDS to protect the encapsulated OND complexes against degradation by nucleases, formulations were incubated with S1 nuclease (100 U/µL), an enzyme that specifically degrades single-stranded nucleic acids. Stability studies were carried out in the following formulations: aqueous solution of naked OND, aqueous dispersion of complexed DDAB-OND in the SEDDS and DOTAP-OND in the SEDDS and dispersion of the nonloaded SEDDS in an aqueous solution of naked OND. The SEDDS were dispersed immediately before the experiment. The S1 nuclease assay (Thermo Fisher Scientific, Waltham, MA, USA) was performed according to the manufacturer's instructions. Briefly, equivalents containing 1.2 µg of OND were incubated with 12 U of S1 nuclease in the presence of the reaction buffer pH 4.5 at 25 • C. As controls, the same formulations were incubated in an acetate buffer of pH 4.5. The reaction was terminated after 30 min by the addition of 0.5-M EDTA and incubation for 10 min at 70 • C. Nondegraded intact OND was precipitated via ethanol precipitation. From each formulation, 60 µL was taken, and 1.3 µL of 5%NH 4 OH, 30 µL of 7.5-M ammonium acetate and 400 µL of 96% ethanol were added into the samples, which were kept at −80 • C for at least one hour. Following this, the samples were centrifuged at 15,000× g for 30 min at 4 • C. Intact OND was pelleted and dissolved in 500 µL of 100-mM triethylammonium acetate (TEAA). The samples were subjected to HPLC analysis. Additionally, 100 µL of supernatant was diluted with 500 µL of 100-mM TEAA. Thirty microliters of the samples were analyzed using the Waters Symmetry C18 (3.5 µm; 4.6 × 75 mm) HPLC column (Waters Corporation, Milford, MA, USA). The mobile phase consisted of 2 eluents, eluent A of 100-mM TEAA (pH 7.0) and eluent B of acetonitrile, with a linear gradient starting from 1 min at 10% to 30% B in 4 min. Thirty percent B eluent was maintained for 1 min, then decreased to 10% B in 1 min and maintained at this level for 4 min before analysis of the following sample. The flow rate was 1 mL/min. FAM-labeled OND was detected by fluorescent detection excitation/emission 495 nm/520 nm according to the procedure described previously [42]. Each formulation was tested in triplicate. The percentage of intact OND was calculated as a ratio of area under the curve (AUC) of protected OND found in the pellets after incubation with S1 nuclease and pelleted OND recovered from the respective formulations, as described by Equation (2). Caco-2 Cell Monolayer Transport Study Caco-2 cells, a model of human enterocyte epithelial cells, were obtained from the American Type Culture Collection (ATCC) (Manassas, VA, USA). The cells were seeded on T12 filter inserts (pore size 0.4 mm, 1.13 cm 2 growth area, Corning, Sigma-Aldrich, Copenhagen, Denmark) at a final density of 1 × 10 5 cells/cm 2 . The cells were cultured for 20-23 days before the transport study was conducted. Transport experiments were conducted on a horizontal shaker at 37 • C using a pH gradient mimicking physiological conditions. A volume of 500 µL of 10-mM MES-HBSS buffer (pH 6.5, 0.05% w/v BSA) was used on the apical side and 1000 µL of 10-mM HEPES-HBSS buffer (pH 7.4, 0.05% w/v BSA) on the basolateral one for 120 min. The experiment was initiated by applying 500 µL of a formulation on the apical side. The following formulations were tested: OND solution, DDAB-OND in the SEDDS, DOTAP-OND in the SEDDS, DDAB in the SEDDS dispersed in OND solution, DOTAP in the SEDDS in dispersed OND solution and nonloaded blank SEDDS dispersed in OND solution for both the Citrem and Standard SEDDS. The SEDDS were dispersed 1:100 (w/w) shortly before the experiment in a MES-HBSS-based buffer preheated to 37 • C. All formulations finally contained 1 nmol/mL of OND. Transepithelial electrical resistance (TEER) was monitored with the EVOM Epithelial Voltohmmeter (World Precision Instruments, East Lyme, CT, USA). Before and after the experiment, the cell culture cup Endohm chamber was used at 25 • C in MES-HBSS and HEPES-HBSS applied apically and basolaterally, respectively. In addition, TEER values were also monitored using chopstick electrodes throughout the experiment at 30, 60, 90 and 120 min at 37 • C. Samples of 100 µL were taken from the basolateral side at predetermined time points (15,30,45,60, 90 and 120 min) and replaced with the same volume of fresh prewarmed HEPES-HBSS buffer. The withdrawn samples were analyzed for fluorescently labeled OND by measuring fluorescence excitation/emission wavelengths at 495/520 nm in a microplate reader (FLUOstar Omega, BMG Labtech, Ortenberg, Germany). The apparent permeability coefficient (P app ), and transported OND accumulated basolaterally across the Caco-2 monolayer were calculated using the following Equations (3) and (4). where J indicates the flux over a steady-state time period (pmol/s/cm), and C 0 stands for the initial concentration of OND in the apical compartment (pmol/mL). Q is the accumulated mass of OND transported across the monolayer on the basolateral side (pmol) of the area A (cm 2 ) of membrane filters. In Vitro Cytotoxicity Study Cytotoxicity of the formulations was investigated as the loss of cell membrane integrity accompanied by release of a cytoplasm enzyme lactate dehydrogenase (LDH). The cytotoxicity study was conducted by the Cytotoxicity Detection Kit (LDH) (Roche, France) according to manufacturer's instructions after 120 min of the transport experiment in the Caco-2 cells. Briefly, a 40-µL sample of apical medium was diluted 5-fold by DI water. Fifty microliters of the diluted sample were pipetted into a clear 96-well multiwell plate in duplicates, and 50 µL of LDH reaction mixture was added. The 96-well plate was incubated for 30 min at 37 • C in a horizontal shaker in the dark. Subsequently, the absorbance was measured at 492 nm. Results are depicted as cell viability (%). Relative cytotoxicity (%) was calculated according to the manufacturer's instructions, and the resulting values were deducted from 100% to obtain cell viability (%). Uptake Study Caco-2 cells were seeded onto confocal dishes (Nunc TM 4-well Rectangular Dishes, Thermo Fisher Scientific, Waltham, MA, USA) at a density of 10 5 cells/cm 2 and were cultured for 7 days until 100% confluence was reached. The experiment was initiated by the addition of 0.5 mL of a test sample. In order to mimic the intestinal environment, the test samples were prepared in 10-mM MES-HBSS buffer (pH 6.5, 0.05% w/v BSA). The uptake of OND was evaluated by testing the following samples: OND solution, nonloaded SEDDS dispersed in OND solution, DDAB-OND in SEDDS and DOTAP-OND in SEDDS dispersed in MES-HBSS buffer. The experiment was carried out for both Citrem and Standard SEDDS, and SEDDS were dispersed at 1:1000 (w/w). The cells were incubated with the samples for 2 h at 37 • C and 5% CO 2 without shaking. Fifteen minutes before the termination of the incubation, cell nuclei were stained by the addition of 10 µL of Hoechst 33342 (1 mg/mL). At the same timepoint, 10 µL of propidium iodide (PI) (0.1 mg/mL) was added to detect dead cells. Immediately after incubation, cells were washed twice with prewarmed Opti-MEM® medium (Thermo Fisher Scientific, Waltham, MA, USA). Images were acquired with the Nikon Ti-E (Nikon, Minato, Japan) epifluorescence microscope equipped with a cooled scientific CMOS camera (AndorZyla 5.5; Andor Technology, Belfast, UK) and LED fluorescence source (CoolLED pE-300; CoolLED, Andover, UK). DAPI, FITC and Cy3 filter sets were used for image Hoechst, FAM-labeled OND and PI, respectively. Three representa-tive images of each sample were taken using NIS Elements AR 4.20 software (Laboratory Imaging, Prague, Czech Republic). Statistical Analysis Statistical analysis was performed using Graph Prism 8 (GraphPad Software, San Diego, CA, USA). One-way ANOVA followed by Tukey's post-hoc test was carried out for the evaluation of size and zeta potential of the dispersed SEDDS. Two-way ANOVA followed by Tukey's test was performed to evaluate lipolysis data and in vitro data. Oneway ANOVA followed by Dunnett's post-hoc test was used, as the impact of the other lipids on the complex stability was evaluated as well, as in the case of the in vitro data. Data are presented as mean ± standard deviation (SD). Significant difference was considered at p < 0.05. Complex Preparation First, complexes of a cationic lipid and 20-mer fluorescently labeled OND were prepared using the Bligh-Dyer method [23,33]. Various charge ratios between a quaternary ammonium head group of a cationic lipid (N + ) and a nucleotide phosphate group in the backbone of 20-mer OND (PO 2 − ) were investigated. Upon complexation, no visible precipitated material in the aqueous phase nor at the interface was observed. The CE of DDAB-OND and DOTAP-OND are depicted in Table 3. Efficient CE (>95%) was achieved for the charge ratio of cationic lipid:OND 3:1, which equals the molar ratio 60:1. Moreover, at this charge ratio, there was no significant difference in CE between DDAB and DOTAP. It is noteworthy that DOTAP also enables a more efficient hydrophobization of OND at a lower charge ratio than DDAB. Results are presented as mean ± SD (n = 3). Effect of SEDDS Lipid Excipents on Complex Stability Next, the impact of the SEDDS excipients on the destabilization of the complexes was tested. The tested excipients included anionic Citrem, zwitterionic LPC and neutral Captex 300, Labrasol, Maisine CC and Peceol. The results from the combination of LPC and DOTAP-OND could not be determined as emulsification occurred, disabling phase separation under the method's conditions. Only anionic Citrem had a negative impact on the stability of both DDAB-OND and DOTAP-OND (p < 0.001) ( Table 4). Zwitterionic LPC and neutral excipients had no impact on the complex stability. These results are in accordance with the observations by Wong et al. [23]. Atomic Force Microscopy (AFM) OND and the cationic lipid-OND complexes were imaged by AFM in order to compare their size and morphology. It is important to note that AFM imaging only provides information on surface immobilized structures, and the absolute accuracy is limited by the physical convolution between the tip (radius around 7 nm) and the sample. Nevertheless, AFM topography images are useful for comparing the sizes and for providing a direct observation of the morphology of the OND and complexes on the surfaces. Figure 3A shows a topographic image of OND molecules dispersed on mica. The molecules were compacted instead of stretched out over the mica substrate, suggesting a potential effect of Mg 2+ (a bridging agent used for deposition on the mica substrate) on the molecular conformation [34]. An analysis of height cross-sections indicates a step height of around 1 nm, which is in agreement with the size of a single strand DNA OND, as a double helix of DNA has a diameter of 20 Å (=2 nm) [43,44]. The AFM topographic image of DDAB-OND complexes ( Figure 3B) reveals a mixture of nanoparticulate complexes and membrane patches which are presumably the result of collapsed lipid capsules during sample drying. The height of the membrane patches is 4 to 5 nm. In contrast, the topographic image of the DOTAP-OND complexes ( Figure 3C) shows mostly homogenous and intact particles. Average height of the particles was 15.1 ± 6.3 nm (n = 167) and 20.9 ± 3.5 nm (n = 70) for DDAB-OND and DOTAP-OND, respectively. The actual height of both complexes in solution should be higher, as significant deformation is expected during drying and immobilization on the substrate during sample preparation. The spherical lipid capsule collapsed into a shape of a flattened disc, having maintained its volume. Using the reported particle volume from SPIP software, the diameter of the particles was calculated for the original thermodynamically favoured spherical shape. The ATR-FTIR Spectroscopy In the region of the FTIR spectra depicting changes in base pairing (1800-1500 cm −1 ), OND bands of 1643 cm −1 and 1605 cm −1 were observed ( Figure 5A,B). Upon complex formation, these bands moved to the higher wavenumbers of 1659 cm −1 and 1697 cm −1 . The positive shift in this FTIR region suggests the disintegration of the hydrogen bonding network involved in base pairing [24,46]. Although the model OND is single-stranded, partial base pairing could arise. The list of complex-specific bands is summarized in Table 5. The OND peak at 1211 cm −1 shifted positively to 1257 cm −1 in both complexes but remained at the same position (1211 cm −1 ) in the mixture. A new band appeared at 1095 cm −1 and 1088-1080 cm −1 in DDAB-OND and DOTAP-OND, respectively. These bands characterize P=O vibrations. The peak at 1018 cm −1 , typical for the P-O-C bond, gained more intensity upon complexation. This suggests an increase in the vibrational frequencies of anionic PO2 in the OND backbone as a result of an interaction with the cationic head of a lipid. An intense peak at 802 cm −1 was observed in all complexes, suggesting a change in the conformation of deoxyribose ( Figure 5A,B). Together with this intense peak, a broader peak at 864 cm −1 in the case of DOTAP-OND suggests an N-type (C3′-endo) sugar puck- Upon mixing and complexing DOTAP with OND, the melting temperature decreased to 1.3 and 3.5 • C, respectively ( Figure 4B). ATR-FTIR Spectroscopy In the region of the FTIR spectra depicting changes in base pairing (1800-1500 cm −1 ), OND bands of 1643 cm −1 and 1605 cm −1 were observed ( Figure 5A,B). Upon complex formation, these bands moved to the higher wavenumbers of 1659 cm −1 and 1697 cm −1 . The positive shift in this FTIR region suggests the disintegration of the hydrogen bonding network involved in base pairing [24,46]. Although the model OND is single-stranded, partial base pairing could arise. The spectra of the cationic lipids carried several specific bands ( Figure 5C,D). The ester group of DOTAP gave rise to a peak at 1744 cm −1 [47,48] that also remains in the same position after complexation with OND. A weak band of around 1489 cm −1 characteristic of the trimethylammonium headgroups [47] was not observed in any of the cationic lipids. This weak band is likely overlapped by the CH2 scissoring deformation at 1466 cm −1 The list of complex-specific bands is summarized in Table 5. The OND peak at 1211 cm −1 shifted positively to 1257 cm −1 in both complexes but remained at the same position (1211 cm −1 ) in the mixture. A new band appeared at 1095 cm −1 and 1088-1080 cm −1 in DDAB-OND and DOTAP-OND, respectively. These bands characterize P=O vibrations. The peak at 1018 cm −1 , typical for the P-O-C bond, gained more intensity upon complexation. This suggests an increase in the vibrational frequencies of anionic PO 2 in the OND backbone as a result of an interaction with the cationic head of a lipid. An intense peak at 802 cm −1 was observed in all complexes, suggesting a change in the conformation of deoxyribose ( Figure 5A,B). Together with this intense peak, a broader peak at 864 cm −1 in the case of DOTAP-OND suggests an N-type (C3 -endo) sugar puckering mode [46]. For DDAB-OND, no other markers of the N-type puckering mode existed. This could be due to the DDAB peak occurring at a similar wavelength (887 cm −1 ), resulting in an interference with other N-type marker bands. In the physical mixtures, there were less intensive and broader peaks at 879-887 cm −1 and 840 cm −1 in DDAB-OND and DOTAP-OND, respectively. These bands are characteristic of S-type sugars [46]. In contrast, the bands found in OND showed a rather broad shoulder peak at 818 cm −1 . The position of this broad shoulder peak is unspecific and might suggest the presence of both sugar puckering modes. The spectra of the cationic lipids carried several specific bands ( Figure 5C,D). The ester group of DOTAP gave rise to a peak at 1744 cm −1 [47,48] that also remains in the same position after complexation with OND. A weak band of around 1489 cm −1 characteristic of the trimethylammonium headgroups [47] was not observed in any of the cationic lipids. This weak band is likely overlapped by the CH 2 scissoring deformation at 1466 cm −1 seen in both lipids. In the formed complexes, this peak did not change its position, which correlates with the behavior of lipid peaks at higher wavelengths in the C-H stretching region. In DDAB, the symmetric (2849 cm −1 ) and asymmetric (2916 cm −1 ) CH 2 vibrations and, similarly, the symmetric (2854 cm −1 ) and asymmetric (2924 cm −1 ) CH 2 vibrations of DOTAP remained at the same vibrational frequency upon complexation at all tested ratios [47], which indicated no changes in the lipid chain arrangement. Table 5. Bands specific to the complexes of both cationic lipids and OND. Fourier-transform infrared (FTIR) spectra of the respective cationic lipid and OND were subtracted from the FTIR spectra of the complexes to obtain complex-specific bands. Bands were observed in the complexes at all ratios if not otherwise specified. Characterization of Dispersed SEDDS Both Citrem and Standard SEDDS preconcentrates were clear oily formulations that immediately dispersed into milky emulsions both in DI water and MES-HBSS buffer (pH 6.5). The dispersion in DI water was evaluated in order to assess properties of the formulations and in MES-HBSS to mimic the pH, as well as a more complex environment of in vitro testing media, as shown in Table 6. The size (z-average) was found between 180 and 270 nm for both SEDDS in both investigated media, with the PdI ranging from 0.125 to 0.360. In the case of the Citrem SEDDS, there was no significant difference in size across the formulations and media. The Standard SEDDS loaded with DOTAP-OND formed significantly larger nanostructures in the DI water than nonloaded Standard SEDDS (* p < 0.05). In MES-HBSS buffer, a similar phenomenon was observed; in this case, the DDAB-OND-loaded Standard SEDDS created larger structures than the nonloaded formulation (* p < 0.001). The zeta potential was significantly influenced by the composition of SEDDS and the dispersing media. In DI water, both nonloaded SEDDS showed a negative surface charge: −35.5 ± 0.6 and −5.2 ± 2.1 mV for Citrem and Standard SEDDS, respectively. After loading, the surface charge of the Citrem SEDDS increased to about −25 mV ( ### p < 0.001) and turned positive in the loaded Standard SEDDS to about 13 mV ( ### p < 0.001). Both complexes increased the zeta potential of the nonloaded Citrem and Standard SEDDS to the same extent. The Citrem and Standard SEDDS in MES-HBSS buffer (pH 6.5) showed no significant differences within the same kind of SEDDS among the nonloaded and loaded SEDDS; about −9.5 and 0 mV for Citrem and Standard SEDDS, respectively. Dynamic In Vitro Lipolysis of SEDDS Human intestinal digestion of SEDDS under fasted-state conditions was simulated by dynamic in vitro lipolysis. Lipolysis of the SEDDS was performed with the nonloaded Citrem and Standard SEDDS, as well as both SEDDS with orlistat, in order to evaluate the degree of digestion ( Figure 6). The amount of released free fatty acids was the same for both SEDDS. The addition of orlistat significantly decreased lipolysis to~10% of the formulations with no orlistat (* p < 0.001). No significant difference was observed between the Citrem and Standard SEDDS upon the addition of orlistat. Moreover, lipolysis of DDAB in SEDDS and DOTAP in SEDDS was performed. None of cationic lipids had any impact on the digestion process (data not shown). degree of digestion ( Figure 6). The amount of released free fatty acids was the same for both SEDDS. The addition of orlistat significantly decreased lipolysis to ~10% of the formulations with no orlistat (*** p < 0.001). No significant difference was observed between the Citrem and Standard SEDDS upon the addition of orlistat. Moreover, lipolysis of DDAB in SEDDS and DOTAP in SEDDS was performed. None of cationic lipids had any impact on the digestion process (data not shown). All formulations dispersed into nanosized structures. Predominantly, spherical uniand multilamellar vesicles (darker and more pronounced borders) were observed, while oil droplets occurred less frequently (darker spheres with brighter borders). Wu et al. described similar differences between vesicles and oil droplets in a soybean oil-in-water emulsion stabilized by egg yolk lecithin [50]. The nanostructures varied in size from 100 to 300 nm. In addition, cryo-TEM detected more complex spherical and elongated vesicular structures (such as in Figure 5A,B,D), usually of larger sizes (>300 nm). No apparent differences between the six different SEDDS could be observed. Protection Against Nuclease Degradation The S1 nuclease completely degraded OND in an aqueous solution of naked OND (data not shown). Dispersions of both nonloaded SEDDS in the aqueous solution of naked OND showed the same pattern. This confirmed that neither the composition of Citrem SEDDS nor Standard SEDDS interfered with the activity of the enzyme. If OND becomes cleaved by a nuclease, short OND fragments are detected in the supernatant that do not precipitate into pellets. The amount of OND that remain intact upon contact with the S1 nuclease was detected in the pellets and is shown in Table 7. All formulations dispersed into nanosized structures. Predominantly, spherical uniand multilamellar vesicles (darker and more pronounced borders) were observed, while oil droplets occurred less frequently (darker spheres with brighter borders). Wu et al. described similar differences between vesicles and oil droplets in a soybean oil-in-water emulsion stabilized by egg yolk lecithin [50]. The nanostructures varied in size from 100 to 300 nm. In addition, cryo-TEM detected more complex spherical and elongated vesicular structures (such as in Figure 5A,B,D), usually of larger sizes (>300 nm). No apparent differences between the six different SEDDS could be observed. Protection Against Nuclease Degradation The S1 nuclease completely degraded OND in an aqueous solution of naked OND (data not shown). Dispersions of both nonloaded SEDDS in the aqueous solution of naked OND showed the same pattern. This confirmed that neither the composition of Citrem SEDDS nor Standard SEDDS interfered with the activity of the enzyme. If OND becomes cleaved by a nuclease, short OND fragments are detected in the supernatant that do not precipitate into pellets. The amount of OND that remain intact upon contact with the S1 nuclease was detected in the pellets and is shown in Table 7. There was a significant difference in the amount of OND protected by the Citrem and Standard SEDDS, as the Standard SEDDS showed about 3.5-fold higher protection (p < 0.001). On the other hand, no difference was observed between DDAB-OND and DOTAP-OND in any of the SEDDS. Table 7. Percentage of OND protected from degradation after 30 min of incubation in the presence of the S1 nuclease. Caco-2 Cell Monolayer Transport Study Several parameters of the prepared formulations were investigated in vitro in the Caco-2 monolayer. As an indicator of permeation enhancement, the TEER values were determined by an Endohm chamber at 25 • C before and after the experiment. After 120 min, the TEER decreased significantly if incubated with the Citrem or Standard SEDDS. No changes occurred after incubation with OND solution or the MES-HBSS buffer serving as the control. There were no significant changes among the formulations based on the Citrem SEDDS nor Standard SEDDS, indicating that the TEER changes can be attributed to the two SEDDS. Nevertheless, the extent to which the two SEDDS decreased the TEER differed significantly. The final TEER at 120 min of the Citrem SEDDS and Standard SEDDStreated wells represented 66% ± 10% and 28% ± 4% of the initial TEER values, respectively (p < 0.01) ( Figure 8A). In addition, the TEER values seem to be influenced differently by the Citrem and the Standard SEDDS. Results are presented as mean ± SD (n = 5 to 6). Table 8. Transported OND accumulated in the basolateral compartment after 120 min presented as the total mass (pmol) and % of the initial apical mass. Citrem SEDDS Standard SEDDS Transported OND Accumulated Basolaterally at 120 min Results are presented as mean ± SD (n = 5 to 6). * p < 0.05, p < 0.01 and * p < 0.001 = statistically significant differences in transported OND formulated into SEDDS in comparison to the OND solution. Figure 8C,D). Figure 9A,B shows the cumulative OND transport into the basolateral compartment. At 120 min, both SEDDS displayed a significant increase of transported OND compared to the OND solution (Citrem SEDDS # p < 0.05 and Standard SEDDS ## p < 0.01 and ### p < 0.001). In the case of Citrem SEDDS ( Figure 9A), no difference was observed between the two complexes. For the Standard SEDDS ( Figure 9B), DDAB and DOTAP in the SEDDS and, also, the nonloaded SEDDS enabled the transport of higher amounts of OND than DDAB-OND and DOTAP-OND in SEDDS over 120 min (* p < 0.05). The total transported mass of OND after 120 min is summarized in Table 8. In Vitro Toxicity Study The viability of the Caco-2 cells was evaluated by an LDH assay after 120 min of treatment with all SEDDS ( Figure 8B). The OND solution was confirmed to be nontoxic. For both the Citrem and Standard SEDDS, there were no significant differences between the applied SEDDS formulations, which suggests that the main contribution to the cytotoxicity is from the surfactants in SEDDS and not the cationic lipids or the complexes. All Citrem SEDDS had viability values close to 100%, while the Standard SEDDS had a decreased viability to about 70% (* p < 0.05 and ** p < 0.01). These data might indicate increased cell stress due to compromised integrity of the cell membrane. Figure 9. The transported OND accumulated in the basolateral compartment for Citrem SEDDS (A) and Standard SEDDS (B). The apparent permeability coefficient (P app ) values for specified time intervals in the Citrem SEDDS (C) and Standard SEDDS (D). (A,B) # , ## and ### represent significant differences of the flux curves of SEDDS-treated cells relative to the OND solution ( # p <0.05, ## p < 0.01 and ### p < 0.001). (B) * represents a significant difference between complex-loaded Standard SEDDS and a group of other Standard SEDDS formulations (* p < 0.05). (C,D) Bars marked by # showed a significant difference in the P app relative to the respective bars of the OND solution (p < 0.05). (D) The significant differences between the time intervals in the Standard SEDDS is denoted as * p < 0.05. The results are presented as mean ± SD (n = 5 to 6). Table 8. Transported OND accumulated in the basolateral compartment after 120 min presented as the total mass (pmol) and % of the initial apical mass. Results are presented as mean ± SD (n = 5 to 6). * p < 0.05, p < 0.01 and * p < 0.001 = statistically significant differences in transported OND formulated into SEDDS in comparison to the OND solution. Formulation A linear steady state of the flux curves was taken as the basis for the calculation of P app . In order to investigate in detail changes in permeability across the Caco-2 monolayer during 120 min, the time frame of the experiment was divided into three intervals, i.e., ∆15-45 min, ∆30-60 min and ∆60-120 min, with the P app values calculated for each interval. As can be seen in Figure 9C, the P app was stable for all Citrem SEDDS formulations at all time intervals. OND solutions containing dispersions of the Citrem SEDDS with or without DDAB or DOTAP already showed a significant increase at the interval of ∆30-60 min relative to the OND solution. At the final interval of ∆60-120 min, all Citrem SEDDS significantly improved the permeability of OND relative to the respective time intervals of the OND solution (p < 0.05). In contrast, for the Standard SEDDS, the P app increased over time for all formulations (p < 0.05) ( Figure 9D). Only the OND solution with dispersions of DDAB and DOTAP in the Standard SEDDS already increased the P app at the initial time interval ∆15-45 min. At the following interval of ∆30-60 min, the P app of DOTAP-OND in the Standard SEDDS and the nonloaded Standard SEDDS was also enhanced significantly. Finally, at the interval of ∆60-120 min, all formulations of the Standard SEDDS showed an increase in P app relative to the OND solution. In Vitro Toxicity Study The viability of the Caco-2 cells was evaluated by an LDH assay after 120 min of treatment with all SEDDS ( Figure 8B). The OND solution was confirmed to be nontoxic. For both the Citrem and Standard SEDDS, there were no significant differences between the applied SEDDS formulations, which suggests that the main contribution to the cytotoxicity is from the surfactants in SEDDS and not the cationic lipids or the complexes. All Citrem SEDDS had viability values close to 100%, while the Standard SEDDS had a decreased viability to about 70% (* p < 0.05 and ** p < 0.01). These data might indicate increased cell stress due to compromised integrity of the cell membrane. Uptake Study The uptake of the fluorescently labeled OND formulated into SEDDS was examined in confluent Caco-2 cells using fluorescent microscopy. Figure 10 shows merged images: blue Hoechst staining cell nuclei, red PI staining dead cells and green FAM-labeled OND. Images depicting staining with a single dye can be found in the Supplementary Materials ( Figure S1). Similar to the OND solution ( Figure 10B), no OND was observed in living Caco-2 cells upon incubation with the Citrem or Standard SEDDS or when OND was administrated in the aqueous phase (nonloaded SEDDS; Figure 10C,D) or complexed inside the SEDDS (DDAB-OND and DOTAP-OND; Figure 10E,F and Figure 10G,H, respectively). The green signal of FAM-labeled OND was observed only in combination with the red signal indicating that OND was localized only inside dead Caco-2 cells. Preparation of Cationic Lipid-OND Complexes Cationic lipid-DNA complexes have been shown to be a less immunogenic but, also, a less efficient alternative to viral vectors for the delivery of nucleic acids [4]. These macromolecules are often formulated as lipoplexes with preformed liposomes to enhance the cell transfection of large DNA molecules (several kb) [51], as well as shorter siRNA (tenths of bp) [52], for parenteral delivery to the systemic circulation. In this study, we prepared hydrophobic complexes consisting of cationic lipids and OND in order to load them into SEDDS, a well-established lipid-based oral delivery system. The presented experiments were conducted with a model nonspecific fluorescently labeled OND. Nevertheless, the structural features involved in the key procedures, such as the preparation of ion pairs with cationic lipids and, thus, hydrophilicity reduction, are applicable for potential therapeutic OND sequences. Therefore, the easily detectable fluorescently labeled model OND was used to investigate this innovative formulation process. Monovalent cationic lipids with two lipophilic chains, DDAB and DOTAP, are known to be less toxic than their single-tailed counterparts [53]. Both aliphatic chains of DDAB are saturated, unlike the chains of DOTAP, which have a double bond in the C9 Similar to the OND solution ( Figure 10B), no OND was observed in living Caco-2 cells upon incubation with the Citrem or Standard SEDDS or when OND was administrated in the aqueous phase (nonloaded SEDDS; Figure 10C,D) or complexed inside the SEDDS (DDAB-OND and DOTAP-OND; Figure 10E,F and Figure 10G,H, respectively). The green signal of FAM-labeled OND was observed only in combination with the red signal indicating that OND was localized only inside dead Caco-2 cells. Preparation of Cationic Lipid-OND Complexes Cationic lipid-DNA complexes have been shown to be a less immunogenic but, also, a less efficient alternative to viral vectors for the delivery of nucleic acids [4]. These macromolecules are often formulated as lipoplexes with preformed liposomes to enhance the cell transfection of large DNA molecules (several kb) [51], as well as shorter siRNA (tenths of bp) [52], for parenteral delivery to the systemic circulation. In this study, we prepared hydrophobic complexes consisting of cationic lipids and OND in order to load them into SEDDS, a well-established lipid-based oral delivery system. The presented experiments were conducted with a model nonspecific fluorescently labeled OND. Nevertheless, the structural features involved in the key procedures, such as the preparation of ion pairs with cationic lipids and, thus, hydrophilicity reduction, are applicable for potential therapeutic OND sequences. Therefore, the easily detectable fluorescently labeled model OND was used to investigate this innovative formulation process. Monovalent cationic lipids with two lipophilic chains, DDAB and DOTAP, are known to be less toxic than their single-tailed counterparts [53]. Both aliphatic chains of DDAB are saturated, unlike the chains of DOTAP, which have a double bond in the C9 position. This influences their physicochemical properties; DDAB exists in the gel state and DOTAP in the liquid crystalline state at room temperature [54]. In the present study, it was found that the more flexible DOTAP chains seem to be more easily assembled around the OND core, leading to CE > 95% already at the charge ratio 2:1 (N + :PO 2 − ). At the ratio 3:1 (N + :PO 2 − ), there was no significant difference between the CE of the cationic lipids (Table 3). At all employed ratios, no lipid was observed at the interface during the Bligh-Dyer extraction, suggesting that most of the lipid contributed to the complexation. The ability of DDAB to create a more ordered structure was shown also by DSC, as several melting peaks of crystalline assembles were observed (Figure 4). The more ordered structure and, consequently, tighter molecular packing could explain the smaller size of the DDAB-OND complexes, as detected by AFM (Figure 3). The formation of complexes between a cationic lipid added as a monomer or in micellar form and plasmid DNA (pDNA) has already been described and well-characterized [23,55]. In both published preparation methods, a sufficient recovery of pDNA was reported at the charge ratio 1:1. The recovery process was more efficient (>90%) if a lipid was added in excess of 1.5:1 (N + :PO 2 − ) [55]. For shorter siRNA, the need of higher lipid/OND ratio was reported [52,56]. It is widely accepted that the mechanism of complex formation is common for any type of nucleic acid and cationic lipid [57]. In other words, both components of the complex self-assemble into a core-shell structure based on the ion pairing-more specifically, primarily on electrostatic interactions between the cationic headgroup of a lipid and anionic phosphate of the nucleic acid backbone. The ATR-FTIR analysis ( Figure 5 and Table 5) showed complex-specific bands that occurred mostly in the region characteristic of changes in the phosphate-deoxyribose backbone and confirmed the interactions with the OND backbone in the complex formation process. These interactions could not be observed in physical mixtures prepared with the same cationic lipid:OND ratio. Previous studies showed that electrostatic interactions help to form inverted micelles, encapsulating the nucleic acid in the core surrounded by the shell of the cationic lipid [23]. This is in agreement with the AFM topography images showing collapsed lipid shells around the cores, as observed in both the DDAB and DOTAP complexes (Figure 3). Loading the Complexes into Citrem and Standard SEDDS and Evaluation of the Formulations Both nonloaded and complex-loaded SEDDS were characterized in terms of size and zeta potential in DI water and MES-HBSS buffer (pH 6.5). The Citrem SEDDS created more uniform nanostructures in terms of size, irrespective of the investigated loading and media. Dispersion of the Standard SEDDS seemed to be more influenced by a loaded substance, as a significant difference in size among the Standard SEDDS formulations was observed (Table 6). However, all SEDDS remained in the range suitable to trigger phagocytosis by macrophages, which are highly active in inflamed intestinal tissue [11,58]. Moreover, in the affected area, positively charged proteins are overexpressed [9]. Thus, negatively charged formulations such as Citrem SEDDS present an attractive approach for local targeting to the diseased tissue. In DI water, a significant difference in zeta potential between nonloaded and complexloaded SEDDS was observed for both the Citrem and Standard SEDDS. After complex loading, the zeta potential becomes more positive, suggesting the interference of both cationic lipids from the complexes with the surface charge. Given that OND was complexed with cationic lipids in excess, the shift of zeta potential towards more positive values can be expected. Nevertheless, these differences were not observed in the MES-HBSS buffer (pH 6.5). An interplay between ions present in the buffer and decreased pH seemed to mitigate the differences in zeta potential for individual SEDDS. The Citrem and Standard SEDDS dispersed in FaSSIF were imaged by cryo-TEM (Figure 7). Previous studies imaging dispersed SEDDS by Tran et al. [59] and Fatouros et al. [60] showed oil droplets after SEDDS dispersion, unlike the Citrem and Standard SEDDS, for which vesicular nanostructures prevailed over oil droplets. In comparison to SEDDS characterized in the previous cryo-TEM studies, Citrem and Standard SEDDS contain a higher percentage of lipophilic surfactants. High concentration of lipophilic surface-active molecules can lead to the formation of vesicles, as well as higher complexity and diversity of the vesicular structures. The majority of the observed nanostructures were between 100 and 300 nm. This range correlates with the size measured by dynamic light scattering ( Table 6). As the impact of the SEDDS excipients on the complex stability was evaluated, a slight but significant decrease in complex stability induced by anionic Citrem was observed (Table 4). However, this slight decrease seems to have translated into a low ability of the Citrem SEDDS to protect OND. In the presence of the S1 nuclease, only 16% of the loaded OND amount was protected against hydrolysis ( Table 7). The Standard SEDDS excipients showed no interference with the complex stability, which improved the protection of the loaded OND 3.5-fold (up to 58% of the loaded OND amount). This suggests that there are also other factors in addition to the negatively charged Citrem that lead to complex destabilization. Both the Citrem and Standard SEDDS disperse into nanodroplets and vesicular structures (Figure 7). Complexed OND is likely to remain inside the nanodroplets; the vesicle formation, however, could impair the complex stability. It would require further examination to track the complex inside the vesicles and to evaluate the impact of vesicular structures on the stability of the complexes. Thus, OND protection could be further improved by optimization of the SEDDS composition. Upon oral administration, SEDDS excipients undergo digestion by lipases present in the gastrointestinal tract. The impact of gastric lipases, together with low gastric pH, can be circumvented by administering SEDDS preconcentrates in enteric-coated capsules. In the small intestine, the majority of triglycerides are digested by lipases [61]. In order to maintain the dispersed colloidal structures intact and, thus, to prolong the protection of OND, a lipase inhibitor, orlistat, was added to the SEDDS. Only 0.25% (w/w) of orlistat sufficed to decrease the level of lipolysis in the tested SEDDS to about 10% ( Figure 6). This indicates that orlistat incorporation is a potentially useful approach to inhibit SEDDS digestion and to enable the delivery of well-protected OND. Previous studies have shown that the inhibition of lipolysis of a MCFA-based delivery system did not interfere with its permeation-enhancing effect [62,63]. In Vitro Performance of the Formulations The ability of the Citrem and Standard SEDDS to enhance the permeation of OND was investigated in a transport study across the Caco-2 cell monolayer. A significant increase in OND in the basolateral compartment of the SEDDS-treated cells correlated with a decrease in TEER. No changes in TEER after 30 min in the Citrem SEDDS treated samples led to a stable P app . The TJs opening by the Standard SEDDS plateaued after 60 min, and the final TEER was reached ( Figure 8C,D). This resulted in a significantly enhanced permeation for the Standard SEDDS formulations compared to the respective Citrem SEDDS formulations. Finally, all SEDDS formulations enabled a significantly higher amount of transported OND to the basolateral compartment relative to the OND solution ( Figure 9). For the Citrem SEDDS, there was no significant difference among the formulations, while the DDAB, DOTAP-loaded and nonloaded Standard SEDDS dispersed in OND solution showed a superior permeation of OND compared to the complex-loaded Standard SEDDS. Due to the complexation, OND is associated within the dispersed SEDDS and, thus, is not readily available for permeation, unlike the case of the noncomplexed OND dissolved in the dispersion medium. Nevertheless, loading OND complexed with a cationic lipid into SEDDS enables the crucial codelivery with PEs into the site of the action [64]. In addition, specific delivery through the intestinal monolayer provided by the Citrem and Standard SEDDS was confirmed, as no fluorescently labeled OND was localized inside the Caco-2 cells after the two-hour-long uptake study ( Figure 10). This observation is in accordance with Li et al., who reported only intercellular localization of a macromolecular loading when it was delivered in MCFA-based formulations. In contrast, the loading was observed perinuclearly when employing a formulation based on long-chain fatty acids [65]. The performed experiments did not offer a conclusive answer regarding the choice of a cationic lipid for OND complexation. Although differing in the minimal amount of lipid needed for the complexation (Table 3), as well as the size of the complexes (Figure 3), no differences in the ability to protect OND (Table 7) and in the in vitro performance were observed (Figure 9). Utilized SEDDS excipients are generally recognized as safe (GRAS) and/or approved by the European and US Pharmacopeia [37,66,67]. All SEDDS disperse into nanoparticles of a comparable size of about 200 nm, but they differ in zeta potential. The addition of Citrem into SEDDS decreases the surface charge of the formed colloidal nanostructures. The negative surface charge diminishes the interactions between SEDDS droplets and the negatively charged cell membranes of the Caco-2 cell monolayer. Due to the lack of this kind of repulsion between the cell surface and Standard SEDDS, there will be more frequent interactions between the Standard SEDDS and the cell membrane, resulting in membrane disturbance and LDH release ( Figure 8B). Simultaneously, the opening of TJs by Citrem SEDDS occurs in a less intense manner. Nevertheless, it is noteworthy that McCartney et al. showed the reversible nature of Labrasol-induced reduction in TEER. The same study also reported that a static system such as the Caco-2 monolayer has a limited repair capacity compared to rat colonic mucosae [62]. Medium-chain triglyceride-based SEDDS confirmed their potential in vivo as they outperformed SEDDS based on long-chain triglycerides for the delivery of insulin [63]. However, generally, the SEDDS in vitro-in vivo correlation is known to not always be satisfactory [19], mainly due to the dynamic in vivo environment of complex intestinal fluids of both endogenous and dietary origin and the challenging crossing of the mucus layer [28]. Pathophysiological inflammatory changes, such as the overexpression of positively charged proteins and highly active phagocytes, create conditions for passive targeting. These changes specific to the inflammatory site might favor the negatively charged SEDDS in terms of enhanced effectivity at the target tissue. On the other hand, the permeation enhancement effect of the neutral Standard SEDDS was shown to be more pronounced in the Caco-2 monolayer. The preferential uptake of negatively charged particles by phagocytic cells [68] might limit the ability of the neutral SEDDS to deliver OND into targeted phagocytes. Conclusions The SEDDS based on MCFA was proven to be a viable delivery system across the Caco-2 monolayer for OND drugs. Hydrophobic ion pairing of OND with a cationic lipid, either DDAB or DOTAP, led to the formation of specific complexes, as confirmed by AFM and FTIR. Complexation reduced the hydrophilicity of OND and increased its solubility in lipids. The resulting ion-paired complexes subsequently allowed for their formulation into SEDDS: negatively charged or neutral SEDDS. A morphological evaluation of these SEDDS using cryo-TEM showed the presence of both nanosized oil droplets and vesicular structures. Negatively charged SEDDS did not significantly affect the Caco-2 viability. Additionally, neutral SEDDS enabled a higher OND permeability across the Caco-2 monolayer into the lamina propria, as well as better OND protection against hydrolytic decomposition in comparison to its charged counterpart. The lipolysis of both SEDDS was inhibited with the addition of orlistat, enabling the prolonged protection of OND in the gastrointestinal tract. The evaluation of further properties of the tested formulations, e.g., passive targeting and their effect on intestinal histology, in an appropriate in vivo model could reveal more details of this potent delivery system for local OND delivery. Data Availability Statement: The data is available on reasonable request from the corresponding author.	v3-fos-license
2020-07-19T13:05:36.207Z	2020-07-01T00:00:00.000	220629520	{ "extfieldsofstudy": [ "Chemistry", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://res.mdpi.com/d_attachment/microorganisms/microorganisms-08-01054/article_deploy/microorganisms-08-01054-v2.pdf", "pdf_hash": "c08b90ba9f6752b23c1b34f1b615056280937f99", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45422", "s2fieldsofstudy": [ "Biology", "Chemistry", "Environmental Science" ], "sha1": "c36fd46ea52c88a409335fd7b6922caead4bdf36", "year": 2020 }	pes2o/s2orc	Lipids by Yarrowia lipolytica Strains Cultivated on Glucose in Batch Cultures Oleaginous microorganisms, such as Yarrowia lipolytica, accumulate lipids that can have interesting applications in food biotechnology or the synthesis of biodiesel. Y. lipolytica yeast can have many advantages such as wide substrate range usage and robustness to extreme conditions, while under several culture conditions it can produce high lipid productivity. Based on this assumption, in this study, 12 different Yarrowia lipolytica strains were used to investigate microbial lipid production using a glucose-based medium under nitrogen-limited conditions in shake-flask cultivations. Twelve wild-type or mutant strains of Yarrowia lipolytica which were newly isolated or belonged to official culture collections were tested, and moderate lipid quantities (up to 1.30 g/L) were produced; in many instances, nitrogen limitation led to citric acid production in the medium. Lipids were mainly composed of C16 and C18 fatty acids. Most of the fatty acids of the microbial lipid were unsaturated and corresponded mainly to oleic, palmitic and linoleic acids. Linolenic acid (C18:3) was produced in significant quantities (between 10% and 20%, wt/wt of dry cell weight (DCW)) by strains H917 and Po1dL. Introduction Microbial lipids or single-cell oils (SCO) contain triacylglycerols (TAGs) (lipids of energy reserve), glycolipids (lipids of membrane structure), phospholipids and sterylesters. Among these lipids, TAGs are the main accumulated component in microbial cells. The synthesis of TAGs by microorganisms begins during the stationary phase of growth with the formation of oil droplets in the cytoplasm, when glucose, other hexoses or polysaccharides or similarly metabolized compounds (i.e., glycerol) are employed as the sole carbon source. The quantity and composition of TAGs accumulated in the cytoplasm of cells depend on the physiology of the microorganism and the culture composition applied, such as the type of carbon and nitrogen sources, as well as physicochemical conditions. In most cases, single-cell oils have a fatty acid composition similar to common plant oils [1][2][3][4][5][6][7]. As a potent oleaginous yeast, Yarrowia lipolytica can accumulate lipids to more than 20% of its dry biomass, consisting mostly of unsaturated fatty acids, as in plant oils, and this is of industrial importance in food biotechnology. Beside this, it has been widely used in the production of lipids and lipid-derived compounds such as biodiesel, edible oils or dicarboxylic acids, which are used as building blocks for polymers synthesis [8]. Y. lipolytica has many advantages over other oleaginous microorganisms: The initial cell concentration and pH of the medium were 10 7 cells/mL and 6.00, respectively. During cultivation, pH was manually regulated between 5.00 and 6.00 by the aseptic addition of a 500-600 µL volume of 5 M KOH within a period of 12 h. Culture samples were collected daily for the determinations of biomass (OD 600 and dry cell weight), pH, lipids and sugar concentrations. Methods of Analysis Dry cell weight (DCW), optical density (OD600) and cell density were used for the determination of biomass. Samples collected from culture medium were centrifuged at 5000 rpm for 15 min and washed twice with distilled water. They were dried at 105 • C until a constant weight for dry cell weight determination. OD 600 was measured at 600 nm in a spectrometer. Cell density was determined by counting live cells (by staining with methylene blue) in a microscopic chamber (Thoma Lam chamber). The total lipids of the collected biomass ere determined according to the method of Folch, et al. [28]. Cellular lipids were extracted from biomass by using a 2/1 concentration of chloroform/methanol solution. Extracted cellular lipids were esterified for fatty acid profile analysis. n-Heptane and 2 N KOH in methanol were used for the esterification process. Fatty acid methyl esters (FAMEs) were analyzed in a gas chromatograph-mass spectrophotometer (GC-MS) (Hewlett-Packard, USA), equipped with mass spectrometer (MS) detector and an HP-NOWAX (Code: 1909IN-136) column (60 m × 0.25 mm × 0.25 µm). The injection volume was 1 µL (splitless). Helium was used as a carrier gas with a flow rate of 1 mL/min. The column temperature was started at 120 • C (held for 3 min), then increased at 10 • C/min to 180 • C (held for 19 min) and then at 10 • C/min to 250 • C (held for 19 min). The sugar contents in the culture supernatant were determined by a high-performance liquid chromatograph (HPLC) (Shimadzu brand LC-20AD model), equipped with a Bio-Rad HPX-87H (300 × 7.8 mm) column and an RID detector (RID-10A model). Then, 20 µL of supernatant was injected into the HPLC column and eluted at 50 • C at a flow rate of 0.5 mL/ min using a 5 mM H 2 SO 4 solution as a mobile phase. The prior analysis of protein content in supernatants was precipitated by the addition of 50 µL of perchloric acid (70%) into 1 ml of sample, which was then centrifuged and filtered through 0.45 µm syringe filters. In total, 20 to 35 dilutions were applied to samples. Each analysis was performed in duplicate. Statistical Analysis The data of obtained results were analyzed for analysis of variance (ANOVA) and multiple range tests by using the statistical program SPSS (v.20) for Windows 8 (IBM, Armonk, New York, NY, USA). Principle component analysis (PCA) was performed to simplify the interpretation of the results by using the statistical program PAST (v.3.21) (University of Oslo, Oslo, Norway). Microbial Lipid Production by Y. lipolytica Strain in Shake-Flask Cultivation An example of the lipid content variation for Y. lipolytica strain W29, monitored over eight days of cultivation in shake-flasks, is illustrated in Figure 1. As shown in Figure 2, there is a wide diversity between strains regarding lipid accumulation. The lipid accumulated between 24 to 48 h; at the later stage, lipid accumulation remained constant or even decreased in agreement with previous results of Sabra, et al. [29] (Figure 1). In general, the higher lipid content varied between 0.4 to 1.30 g/L on the second or third day of cultivation. In some cases, lipid accumulation was very low, especially for When glucose, glycerol or similarly catabolized compounds are employed as carbon sources of Y. lipolytica in batch nitrogen-limited cultures, there are three different types of lipid accumulation metabolism that are observed. In the first type-called typical oleaginous metabolism-high quantities of lipid are accumulated after nitrogen exhaustion in the medium with a simultaneously lower amount of extracellular metabolite production, such as citric acid (CA) and polyols [30,31]. In the second type-referred to as atypical oleaginous metabolism-lipids are stored at the beginning of cultivation after nitrogen depletion in the medium, and at a later stage, citric acid production occurs and continues uninterrupted, accompanied by decreasing lipid concentration [25,[32][33][34]. The third type of metabolism is an atypical metabolism, in which lipids in yeast cells accumulate at slow rates without degradation and with the simultaneous secretion of citric acid [29]. According to this assumption, as can be seen in Figure 2, while strains Peggy and DVPG 5858 showed a typical oleaginous metabolism, strains CBS 6303 and K57 displayed an atypical metabolism. Moreover, the third type of atypical oleaginous metabolism was observed for strains Zu110 and W29, in which lipid accumulation occurred with continuous citric acid production (Tables 1 and 2). When glucose, glycerol or similarly catabolized compounds are employed as carbon sources of Y. lipolytica in batch nitrogen-limited cultures, there are three different types of lipid accumulation metabolism that are observed. In the first type-called typical oleaginous metabolism-high quantities of lipid are accumulated after nitrogen exhaustion in the medium with a simultaneously lower amount of extracellular metabolite production, such as citric acid (CA) and polyols [30,31]. In the second type-referred to as atypical oleaginous metabolism-lipids are stored at the beginning of cultivation after nitrogen depletion in the medium, and at a later stage, citric acid production occurs and continues uninterrupted, accompanied by decreasing lipid concentration [25,[32][33][34]. The third type of metabolism is an atypical metabolism, in which lipids in yeast cells accumulate at slow rates without degradation and with the simultaneous secretion of citric acid [29]. According to this assumption, as can be seen in Figure 2, while strains Peggy and DVPG 5858 showed a typical oleaginous metabolism, strains CBS 6303 and K57 displayed an atypical metabolism. Moreover, the third type of atypical oleaginous metabolism was observed for strains Zu110 and W29, in which lipid accumulation occurred with continuous citric acid production (Tables 1 and 2). When glucose, glycerol or similarly catabolized compounds are employed as carbon sources of Y. lipolytica in batch nitrogen-limited cultures, there are three different types of lipid accumulation metabolism that are observed. In the first type-called typical oleaginous metabolism-high quantities of lipid are accumulated after nitrogen exhaustion in the medium with a simultaneously lower amount of extracellular metabolite production, such as citric acid (CA) and polyols [30,31]. In the second type-referred to as atypical oleaginous metabolism-lipids are stored at the beginning of cultivation after nitrogen depletion in the medium, and at a later stage, citric acid production occurs and continues uninterrupted, accompanied by decreasing lipid concentration [25,[32][33][34]. The third type of metabolism is an atypical metabolism, in which lipids in yeast cells accumulate at slow rates without degradation and with the simultaneous secretion of citric acid [29]. According to this assumption, as can be seen in Figure 2, while strains Peggy and DVPG 5858 showed a typical oleaginous metabolism, strains CBS 6303 and K57 displayed an atypical metabolism. Moreover, the third type of atypical oleaginous metabolism was observed for strains Zu110 and W29, in which lipid accumulation occurred with continuous citric acid production (Tables 1 and 2). Table 1. Values of DCW (g/L), consumed glucose (g/L), maximum lipid content (g/L) and lipid in DCW (g/g) for Y. lipolytica strains in shake-flask cultivations (N = 2). Generally, the maximum lipid in DCW (g/g) values-rather than lipid content values-are preferred to evaluate the lipid-producing capacity of oleaginous microorganisms. The maximum lipid in DCW (g/g) values and lipid contents of 12 Y. lipolytica strains were determined. Strains K57 and CBS 6303 obviously did not display lipid-producing characteristics, with lipid in DCW (g/g) values of 0.09 and 0.03, which were very low quantities when compared with the other strains (Table 1). This indicated that the cultivation of these two Y. lipolytica strains in a glucose-based nitrogen-limited environment did not exhibit de novo TAG synthesis and thus lipid accumulation, as the percentage of intracellular lipid content for de novo accumulation should be more than 20%. This behavior can be explained by a switch of metabolic pathways toward the synthesis of citric acid instead of lipid de novo synthesis and accumulation [32]. However, strains Po1dL, DBVPG 4558, Zu110, Ain 19, Ain 16 and H917 had higher lipid in DCW (g/g) values, ranging between 0.39 and 0.61 g/g, respectively. In addition, the variation of lipid in DCW (g/g) of the 12 tested strains is illustrated in Figure 3, in which higher values are generally observed on the second day of cultivation and later remained constant or even decreased. This characteristic of strains was also observed in other studies reported in the literature [25,29,32,35]. Dourou, et al. [5] demonstrated that the lipid metabolism of Y. lipolytica during glucose consumption in shake-flask cultivation occurs in three different phases; i.e., balanced growth, oleaginous and lipid turnover phases. The balanced growth phase concluded with the exhaustion of nitrogen in the growth medium. The oleaginous phase began after the depletion of nitrogen and finished with glucose in the medium. The final phase, the "lipid turnover phase", occurred after glucose depletion in the cultivation medium. Relatively high amounts of lipid were stored from glucose during the transition from the early oleaginous to late oleaginous phase. As can be seen in Figure 3, the time interval from 0 to 24 h of cultivation shows a balanced growth phase, while the interval of 24 to 48 h (some cases until 72 h) indicates the oleaginous phase, and the sixth day of cultivation can be inferred as the lipid turnover phase of most strains (except strain H917). Incubation Zu110, Ain 19, Ain 16 and H917 had higher lipid in DCW (g/g) values, ranging between 0.39 and 0.61 g/g, respectively. In addition, the variation of lipid in DCW (g/g) of the 12 tested strains is illustrated in Figure 3, in which higher values are generally observed on the second day of cultivation and later remained constant or even decreased. This characteristic of strains was also observed in other studies reported in the literature [25,29,32,35]. Dourou, et al. [5] demonstrated that the lipid metabolism of Y. lipolytica during glucose consumption in shake-flask cultivation occurs in three different phases; i.e., balanced growth, oleaginous and lipid turnover phases. The balanced growth phase concluded with the exhaustion of nitrogen in the growth medium. The oleaginous phase began after the depletion of nitrogen and finished with glucose in the medium. The final phase, the "lipid turnover phase", occurred after glucose depletion in the cultivation medium. Relatively high amounts of lipid were stored from glucose during the transition from the early oleaginous to late oleaginous phase. As can be seen in Figure 3, the time interval from 0 to 24 h of cultivation shows a balanced growth phase, while the interval of 24 to 48 h (some cases until 72 h) indicates the oleaginous phase, and the sixth day of cultivation can be inferred as the lipid turnover phase of most strains (except strain H917). The maximum lipid in DCW (g/g) values and lipid contents varied from 0.03 to 0.61 g/g and 0.15 to 1.28 g/L, respectively, and were remarkably high values when compared with the results obtained by Y. lipolytica in shake-flask cultivations using glucose-based media reported in other studies. For instance, maximum lipid in DCW (g/g) values lower than 0.04 g/g were obtained for strains ACA-YC 5033, ACA-YC 5029 and W29 [32]. In the present study, only strains K57 and CBS 6303 exhibited lipid in DCW (g/g) values lower than 0.09 g/g. Moreover, strain W29 generated a higher lipid in DCW (g/g) value, at 0.35 g/g, than reported by Papanikolaou, et al. [32]. Similar results were also reported by Papanikolaou, et al. [35] with regards to the lipid in DCW (g/g) of Y. lipoytica ACA-DC 50109, which were found to range between 0.05 and 0.12 g/g for different glucose concentrations (34 to 150 g/L glucose concentrations) during shake-flask cultivations. Moreover, in another study with the same strain, Dourou, et al. [5] reported a maximum lipid in DCW (g/g) value of 0.33 g/g from a glucose concentration of 40 g/L in the late oleaginous phase of a shake-flask cultivation. Recently, Sabra, et al. [29] also reported a lipid in DCW (g/g) value of 0.12 g/g from a starting glucose concentration of 50 g/L using the same strain in shake-flask cultivation. Despite these low lipid in DCW (g/g) values produced by wild-type Y. lipolytica strains, a metabolically engineered strain of Y. lipolytica, YL-ad9, was capable of accumulating a lipid content of 67% (g/g) from glucose in a fed-batch cultivation [36]. Although this value was found to be close to the 0.61 g/g obtained for strain Po1dL in the current study, the final biomass and lipid content obtained for this strain were very low (1.29 and 0.42 g/L, respectively). 0.002 ± 0.00 0.003 ± 0.00 0.06 ± 0.00 0.03 ± 0.00 0.04 ± 0.00 0.02 ± 0.00 0.03 ± 0.00 0.05 ± 0.00 0.03 ± 0.00 0.04 ± 0.00 0.04 ± 0.00 0.14 ± 0.05 Q C 0.11 ± 0.00 0.13 ± 0.00 0.03 ± 0.00 0.11 ± 0.00 0.002 ± 0.00 0.004 ± 0.00 0.05 ± 0.00 0.01 ± 0.00 0.11 ± 0.00 0.03 ± 0.00 0.08 ± 0.00 0.003 ± 0.00 (Xmax-maximum biomass (g/L), Cmax-maximum citric acid (g/L), Lmax-maximum lipid (g/L), Y X/S -biomass yield (g/g), Y C/S -citric acid yield (g/g), Y L/S -lipid in DCW (g/g) (g/g), Qc-volumetric productivity of citric acid (g/L/h)). On the basis of the lipid in DCW (g/g) graph (Figure 3), the FA composition of each strain was examined at two different stages of the early stationary phase (48 h after inoculation) and late stationary phases (144 h after inoculation). The early stationary phase is a time period in which lipid accumulation increases; in contrast, during the late stationary phase-referred to as the lipid turnover period-accumulated lipids and stored lipids inside cells decrease [32]. The results of FA composition are presented in Table 3, in which oleic acid (C18:1), palmitic acid (C16:0) and linoleic acid (C18:2) are the most abundant FAs produced inside the cell of all examined Y. lipolytica strains. Moreover, linolenic acid (C18:3) was produced in significant quantities (between 10 and 20%, wt/wt), especially by strains H917 and Po1dL. Other FAs, such as C15:0, C17:0 and C17:1, were also found in small amounts for all the strains except strain H917 (approximately 17% of C17:0 and 6% of C17:1 were found). Another point of view regarding FA composition involves the FA profile between the early and late stationary phases, as a slight change in FA profile was obtained for strains Ain 16, Ain 19, DBVPG 4558, K57 and N155, whereas drastic differences were found for strains CBS 6303, DBVPG 5858, H917, Po1dL, Peggy, W29 and Zu110. For instance, the proportions of C16:0, C18:2 and C18:3 in the early stationary phase, increased whereas they decreased for C16:1, C18:0 and C18:1 in the late stationary phase. Generally, a conversion into the group of C18 chains (C18:0, C18:1, C18:2 and C18:3) and C16 chains (C16:0 and C16:1) was observed at different stationary phases with varying concentrations. In addition, the ratio of unsaturated to saturated fatty acids (UFAs/SFAs) indicated that lipids produced in the early stationary phase are, in general, characterized as more unsaturated than those in the late stationary phase, as the ratio of UFAs/SFAs decreased; i.e., the unsaturated profile decreased through the late stationary phase. Among the FA profiles, only strains Peggy and Po1dL in the stationary phase, strain H917 in the early phase and strain Zu110 in the late stationary phase showed a saturated profile. When the results of the FA composition from this study are compared to those from the literature [29,32,35], it can be observed that similar FA compositions were obtained. The most dominant FAs in cellular lipid produced by all strains were oleic, palmitic, linoleic and stearic acids. The only difference was the linolenic acid content, which in some cases was higher than previous results reported in the literature. Moreover, the cellular FA composition produced by strain W29 during cultivation on 30 g/L glucose-based media investigated by Papanikolaou, et al. [32] was different to this study in that the concentration of C16:0 (around 19% in the work performed by Papanikolaou, et al. [32] and 25% in this study) was lower but C18:1 (around 55% in the work performed by Papanikolaou, et al. [32] and 39% in this study) was higher. Table 3. Fatty acid composition of total lipids produced by Y. lipolytica strains during early (second day of cultivation) and late stationary (sixth day of cultivation) growth phases on glucose-based media. UFAs/SFAs: ratio of unsaturated to saturated fatty acids. PCA analysis was performed to characterize the fatty acid composition generated by 12 Y. lipolytica strains in shake-flask cultures. For each strain, two samples, collected on the second and sixth day of cultivation, were considered. In general, on the second day of cultivation, the lipid in DCW (g/g) value was higher; on the sixth day of cultivation, lipid turnover was observed. Figure 5 shows a bi-plot of PCA which explains overall 63.48% of the total variance, composed of 41.37% of F1 and 22.10% of F2. Three different distinct groups of strains on the bi-plot are notable due to their fatty acid composition. In general, fatty acids of palmitic (C16:0), stearic (C18:0), linoleic (C18:2), linolenic (C18:3), palmitoleic (C16:1), oleic (C18:1), pentadecanoic (C15:0), margaric (C17:0) and heptadecenoic acids (C17:1) were observed. The first group was generated by strains Ain19, CBS 6303, W29, K57, DBVPG 4558, N155, Zu110-2 and DBVPG 5858-2, characterized by palmitoleic, oleic and linoleic acids and UFAs/SFAs; the second group consisted of strains H917 and Ain16, identified by heptadecenoic, margaric, pentadecanoic and linolenic acids. Finally, the third group was formed by strains Po1dL, Peggy, Zu110-6 and DBVPG 5858-6, differentiated by palmitic and stearic acids. Considering the PCA results, it can be said that most of the strains were characterized by the property of unsaturation; i.e., oleic, palmitoleic and linoleic acids. However, strains Peggy, Po1dL and sixth-day samples of Zu110 and DBVPG 5858 showed more saturated characteristics consisting of a high level of palmitic and stearic acids. In addition, linolenic acid was generated to a greater degree by strains H917 and Ain16. Conclusions Among the examined strains, of the 12 strains, ten Y. lipolytica strains exhibited an oleaginous property, producing more than 20% (g lipid/g dry cell weight) of the lipid in dry cell weight value (in some cases, 45% of lipid in DCW)-two strains did not exhibit this property (K57 and CBS 6303). According to results, lipids were produced more in the early stationary phase than late stationary phase of cultivation. PCA analysis allowed us to distinguish three different groups of strains according to their FA profile. Most of the tested strains generated in the first group were characterized by an unsaturated profile, while a saturated profile was observed for the second group (strains Peggy and Po1dL). Moreover, the third group (consisting of strains H917 and Ain16) consisted of unusual fatty acids such as linolenic, pentadecanoic, heptadecenoic and margaric acids. This is the first study that has shown that Y. lipolytica strains can be differentiated according to their FA profile. The lipid content and fatty acid profile synthesized by Y. lipolytica are also dependent on the feedstock used. Different types of cheap carbon sources were employed in lipid production via Y. lipolytica cultivation. For instance, 15-25% (wt/wt) of lipids, consisting of major fatty acids such as oleic acid and palmitic acid, were produced from olive mill wastewater-based media [37]. Moreover, in a study of microbial lipid production from industrial derivative of tallow, 52.0% of lipid was accumulated in Y. lipolytica strain ACA-DC 50109 [38]. In the same study, the fatty acid composition was found to be rich in saturated fatty acids, mainly stearic acid (78% wt/wt) and palmitic acid (17% wt/wt). In addition, glycerol-based media (crude glycerol) have been considered as a cheap feedstock, and in nitrogen-limited conditions, lipids accumulated in Y. lipolytica cultıvated on industrial glycerol contained higher amounts of oleic, linoleic, palmitic and palmitoleic acid [39]. In another study, a similar fatty acid profile to Y. lipolytica strain S6 cultivated on 25 g/L pure glycerol or glycerol fraction was obtained [12]. Leiva-Candia, et al. [40] and Qin, et al. [3] also reviewed potential agro-industrial waste utilization using oleaginous yeast for lipid production and explained that Y. lipolytica can produce different amounts of lipid with varying fatty acid compositions depending on the carbon source chosen or culture conditions employed. The fatty acid profiles obtained by Y. lipolytica cultivation on different agro-industrial feedstocks reported in these studıes were similar to vegetable oils; e.g., palm oil. Results from this study also showed that, in general, the major fatty acids of oleic, palmitic, linoleic, stearic and palmitoleic acid, corresponding to the fatty acids profile of vegetable oils, were produced by different Y. lipolytica strains during cultivation on glucose-based media. This lipid, which is similar to the fatty acid profile of vegetable oils, produced by Y.lipolytica grown in glucose-based media has the potential to be used in the sustainable and renewable biodiesel industry. Conclusions Among the examined strains, of the 12 strains, ten Y. lipolytica strains exhibited an oleaginous property, producing more than 20% (g lipid/g dry cell weight) of the lipid in dry cell weight value (in some cases, 45% of lipid in DCW)-two strains did not exhibit this property (K57 and CBS 6303). According to results, lipids were produced more in the early stationary phase than late stationary phase of cultivation. PCA analysis allowed us to distinguish three different groups of strains according to their FA profile. Most of the tested strains generated in the first group were characterized by an unsaturated profile, while a saturated profile was observed for the second group (strains Peggy and Po1dL). Moreover, the third group (consisting of strains H917 and Ain16) consisted of unusual fatty acids such as linolenic, pentadecanoic, heptadecenoic and margaric acids. This is the first study that has shown that Y. lipolytica strains can be differentiated according to their FA profile.	v3-fos-license
2017-08-27T06:50:55.179Z	2009-05-02T00:00:00.000	33526564	{ "extfieldsofstudy": [ "Political Science" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://www.ccsenet.org/journal/index.php/jpl/article/download/2306/2165", "pdf_hash": "18085eb3dd5ff49f8cd5c8b0d8245f9303991102", "pdf_src": "ScienceParseMerged", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45423", "s2fieldsofstudy": [ "Political Science" ], "sha1": "9b09b089cf008d927284a22e6d6702fc77c1493c", "year": 2009 }	pes2o/s2orc	The Impact of Singapore ’ s Military Development on Malaysia ’ s Security In this intense era of military and defense development in South East Asia, Singapore has emergence as the fastest country in the development of military capabilities. The rapid military development that started in 1965 has made Singapore become the strongest and finest in military and defense compared to other Southeast Asia nations. Singapore’s decision to be independent from Malaysia has forced it to be self-reliant, especially in terms of security and defense. Singapore adopted the approach to develop and strengthen its defense and military system after achieving independence in 1965. Its increasing economic development in1990 has influenced the military development process and defense system. This rapid expansion has made Singapore emergence as the strongest and most advanced in military capabilities country in the Southeast Asian region. The offensive defense doctrine practiced such as forward defense, poison shrimp, pre-emptive strike and strategic weaponry ownership had raised concerns among leaders in the Southeast Asian countries. At the same time, Malaysia has also taken action to speed up its military development, diversifying the defense doctrine including total defense, complete military with modern and sophisticated defense equipment. It is speculated as a result of the security impact that Malaysia face from Singapore’s military development. Hence, this study tries to elaborate the impact or security implications on Malaysia resulting from Singapore’s military development from the Malaysian military perspective. Introduction The history of the Singapore's military development has essentially been conducted before the separation of Singapore from Malaysia in 1965.However many researchers agree that the military development in Singapore has been conducted during the British rule.According to Huxley (2000:1-4) the initial Singapore's military development was aimed to protect the British's autonomy, in which the latter controlled the island of Singapore as the administration centre particularly in Southeast Asia.The importance of Singapore as an island port had influenced British to build a defense system as a move to protect the island.Since 1927, Singapore has owned system of defense which includes the army, navy and air force.Within 1948-1960, Singapore's military was controlled by two limited battalions, Singapore Infantry Regiment (SIR) and the navy force which is under the Malayan Naval Forces (MNF) base in Woodland, Singapore and the air force which is known as Malayan Auxiliary Air Forces (MAAF).Yet at that period, these defense forces known as SIR, MNF and MAAF are defense forces that were under the authority of the Federation of Malaya. Singapore has rapidly developed its military forces since 1965 upon its separation from Malaysia, as an independent nation under the leadership of Prime Minister Lee Kuan Yew (Nasibah Harun,2005:17).Realizing that in terms of geography condition the nation is small and its defense force is limited, Singapore was forced to rely on British to ensure its security through the establishment of Anglo Malayan Defense Agreement (AMDA) in 1957 and Five Power Defense Arrangement (FPDA) in 1971(Chamil Wariya,1989:49;Chin,1983:chapter3&4).During that period Singapore has taken the initiative to draft and build its military and defense system.SIR was changed to Ministry of Interior and Defense at the end of 1960 and the administration of all three arm forces became Singapore Army (SA), Singapore Navy (SN) and Republic Singapore Air Force (RSAF) and placed under one authority called Singapore Armed Forces (SAF).While since 1970 the Ministry of Defense Singapore (MINDEF Singapore) was established and Ministry of Home Affairs was founded to manage the internal affairs (Huxley, 2000:37-40). Singapore's Military Development and Defense Process The right after independence and due to the human resources scarcity in military forces, Singapore has decided to set up a volunteer organization known as People's Defense Forces in 1966 to strengthen the force of 6 battalion army at that time.Singapore also introduced the National Service program base on the national service model applied by Israel in 1967 that obligated 18-year-old citizen or permanent resident to join the National Service.As a result, the total force of Singapore's military doubled to 12 battalions with the increase of 6 more army battalion from the National Service at the beginning of 1972 (Yong,2001:286-288).It is a process to make sure Singapore's objective to form an army of citizen is achieved through the national service program.To ensure that the planning and the defense system are well-built, Singapore has brought in military experts from Israel, Britain and Sweden to train and help develop its military capabilities since 1965 (Hussin Mutalib,2001:41).These experts are responsible in providing the training and planning to Singapore's military officers at the military training institute known as Singapore Armed Forces Training Institute .During 1965During -1975, Singapore implemented the defense doctrine which is defensive through the approach of protecting the country from threats using the deterrence system also knows as the doctrine of poison shrimp.This doctrine warns enemies not to take any action that can affect Singapore's security and sovereignty.It depicts the readiness of Singapore to act upon enemies who threaten its security and also popular through the phrase: "eat it and you may die" (Mauzy and Milne,2002:170).After the United State's defeat in Vietnam in 1975, Singapore started to practice a defense doctrine similar to the defense doctrine of Israel which is more offensive in nature and known as preemptive strike (attack before the enemy strikes base on accurate intelligence information) using the air force, land force (amour), landing and mobility.Furthermore, Singapore Armed Force (SAF) received help from Israel Defense Force (IDF) who has introduced the defense doctrine named forward defense that stresses on the importance of air defense development, total military and sustainable defense.At the same time, Singapore has also reinforced its security system by practicing the dependence on superpowers policy to make sure Singapore receives support (Tan,1998:458). Since 1980, Singapore has made changes in policy and ownership of sophisticated and strategic armaments parallel with the doctrine of defense that it has applied.It too is an approach to ensure the safety and considered as the process to guarantee Singapore's survival.During the 1980s, Singapore has possessed modern weaponry such as 270 light tanks and Main Battle Tanks (MBT), 720 carrier vehicles and artillery cannons 155mm (land), 26 F-5 battle aircrafts and Skyhawk aircrafts, F-16, Bloodhound missile, RBS-70, Rapier, Surface-to-Air Missile (SAM) and Airbone Early Warning system type E-2C Hawkeye.Singapore even has successfully produced its own aircraft called Super Skyhawk at the end 1980s.Huxley (2000:459) explained, around 1991 Singapore military power is far more establish compare to Malaysia and Indonesia.Military development in Singapore during 1990s involved the purchase and ownership of weaponry such as light tanks and Main Battle Tanks (MBT), F-16 aircrafts, helicopters, missiles, modern artillery equipment and submarines.Singapore's progress in the field of defense has proven its ability to produce Infantry Fighting Vehicle (IFV) in 1998, making it the first Southeast Asian country to successfully manufacture an IFV.In 2000, Singapore has already diversified its modern defense equipment that was imported from various countries like the United States, Russia, Sweden, Israel and France.Among the weaponry that Singapore possess is 12 AH-64 D Apache Longbow helicopters, 20 F-16 aircrafts and aircrafts with Dassault Aviation Rafale technology. Singapore has also ordered as many as 6 Frigate La Fayette ships from France and by year 2004 Singapore is scheduled to receive 4 SSK submarine (please refer to the schedule below).Directly this will make Singapore become the strongest and the best navy force in Southeast Asia.According to Tan (1998:459), Singapore's strength at this moment is the best compared to other countries in Southeast Asia.Mazy and Milne (2002:169) said the rapid development and increase on allocation of expenditure has placed Singapore as a Southeast Asian country which owns the best defense and security system in the region of Southeast Asia.Dibb (1997) stated that a countries RMA process in Southeast Asia is still vague except Singapore's.This is because since 1992 Singapore have started envisioning and directing its military to confront the challenges of the 21st century, parallel to the development of current technology.Realizing the current development of technology and world threats, Singapore have started to take measures in ensuring that the military moves together with technological development and current threat especially when encountering electronic warfare (EW).For example, the widespread usage of electronic combat radio in its military operation has directly shown Singapore's seriousness in applying EW during military operation activities such as survey, disruption and deception of the enemy (Tan,1998:467). Singapore's seriousness has brought changes in its military RMA and it is visible through the efforts by trying to apply technological advancement that completes high technology military.For example the application of C3I's that is to Command, Control, Communication and Intelligence in its military with the use of electronics.It is based on the military's need of adaptation and technological advancement that could strengthen national defense and security system.The C3I application is especially noticeable through the use of electronic equipment in performing military operations including spying or investigating by utilizing the airborne early warning (AEW), unmanned aerial vehicle (UAV), with high tech aircraft surveyor, satellite, ground base radar, use to decide target through computer usage, purchase of aircrafts and other high technology defense equipment (Brooke,2004:4-7). Malaysia's Concern on Singapore's Military Development This military development has indirectly raised the concern of Singapore's neighboring countries, Malaysia in particular.Singapore military development is seen as a security threat to Malaysia.Rustam A. Sani (1998:23) stressed bilateral issues between the two countries such as the issue of water, newspaper, border invasion, territorial claims, racial problem and other lingering issues has influenced how Malaysia views the security threats from Singapore.According to Sarimah Othman (1998:15), there were reports in the Malaysian press regarding Singapore actions on the bilateral relation with Malaysia and has the purpose to create a strategic enemy (Malaysia) to achieve a higher purpose.Mohd Zuki Pileh (2003:36) quoted the Malaysia Foreign Minister statement, Datuk Seri Syed Hamid Albar concerning Singapore's actions regarding the bilateral ties:-"They (Singapore) want to show when they separated from Malaysia, they were a small and weak country but now they have the ability to defeat Malaysia.Therefore, Singapore thinks they are more superior". In the 1990s and 2000, Malaysia has started to modernize its armed force to be ready to face any threats from aggressors.Malaysia's Defense Minister, Datuk Sri Najib Tun Razak has stressed that the modernization efforts of Malaysian military will focus on mobility, fire power, increase the amount of battleships and possessing the Airbone Warning and Control System (AWACS) (Asian Defence Journal, Oktober 2003:16-23).According to Jayasankaran (2002:20) the Malaysian military development is Malaysia's reaction on Singapore rapid military progress.Malaysia has also taken measures in acquiring weaponry which are multi-function or defensive and offensive in nature.The purchase of FA-18 Hornet aircraft, MIG-29N Fulcrum, Hawk MK108, SU-30MKM, PT-91M tank, Scorpene's submarine, close range missile launcher ASTROS II, G5 MK III , Styer portable rifle among others are Malaysia's military development process to face any security threat (Nasibah Harun,2006:3-6;Tempur, July 2003:39-40).Badrul Azhar Rahman (1998) stated that:-"Statement of Singapore's Minister of Trade and Industry, Brig Jen. Lee Hsien Loong, that Malay are not allowed to become pilot and hold high rank position in Singapore's military forces because concerned over their loyalty… indicates that Singapore is getting ready for war". Hamdan Hj Abu (2003:10) quoted former Malaysian Prime Minister's statement Tun Dr Mahathir Mohamad relating to Singapore's threat on Malaysia's security:-"…there is a country (Singapore) who has declared Malaysia as its battlefield…there is a proud country (Singapore) who claimed they have the right to take preemptive strike and forward defense towards our country.We promise if anyone tries to invade our country's independence with action such as preemptive strike or forward defense, they will get what westerners call a bloody nose". Security threat Singapore's military development and defense system in its early stage was initially defensive in nature, since 1971 Singapore has practiced the poison shrimp doctrine.This doctrine perceived as defense doctrine extracted from the Israel's doctrine of defense which affirms that Singapore warns any aggressor not to attack them.This doctrine takes into account the regional geo-political condition similar to Israel's position which is surrounded by Arab countries.This doctrine is only a warning towards aggressors, however if attacks or threats are thrown at Singapore then the aggressor are forced to face Singapore reaction.The emergence of offensive doctrines known as preemptive strike doctrine is Singapore's preparation to attack the enemy if the enemy is believed (base on accurate intelligence information) to try and threaten its security.Singapore will not attack any country Malaysia in particular, as long as Malaysia does not threaten the security of Singapore.This doctrine is categorized as a need to warn the enemy not to invade or attack Singapore.Hence, to complement the doctrine of preemptive strike, Singapore has implemented another defense doctrine called forward defense whereby the military development and defense must always be advance.This doctrine affects planning and war strategy, hence, Singapore would always need to stay ahead in the development of military in terms of physical and non physical features. According to Arrifin Omar (2007) Singapore will not attack and threaten Malaysia's national security.It is due to Singapore will take into consideration various aspects, not only from Malaysia's military aspect but the economical aspect (especially Singapore's investment in Malaysia and Singapore's own economy) and the geography of the region in case it opts for war.Dent (2001:1-23) stated the survival of Singapore does not solely depend on the power of its military capability but also on its economy.As a nation with limited resources, Singapore emphasizes more on its economic security that relies heavily on foreign country.Singapore would be forced a pay high price if it opt for war because its economic prosperity depend largely from foreign countries, Malaysia in particular.If Singapore attacks or strikes Malaysia, then Singapore's economy and its economy and investment dependence on Malaysia will surely be affected.It directly will jeopardize Singapore's security and survival.In evaluating whether Singapore can cause a threat to Malaysia, Ahmad Ghazali Abu Hassan (2007) stated that:-"Singapore is not the main threat of Malaysia's security and sovereignty.Basically, Singapore is the second-largest investor in Malaysia after United States.The total export to Malaysia is believed to be 15.8% from the whole of its total export worldwide.The total imports from Malaysia are also significant with a total of 16.8% from all its import.In other words, both countries are interdependent of each other.Hence a military conflict on Malaysia will directly affect Singapore's own survival." The concept of total defense that is implemented by Singapore is still insubstantial and will risk Singapore's ambition to react upon Malaysia.The concept of total defense is a doctrine of defense that stresses the use of all assets and resources of a nation to increase its ability to face any form of threats, be it domestic or international.Among the evident characteristics of total defense implementation is its activation of volunteer defense and security force in any related organization.The concept of total defense practiced by Singapore is a concept that is deemed unsuccessful because loyalty and nationalism of Singaporeans is believed to be fragile.Singaporeans that are one of the elements of total defense prioritize more on economic stability, their own possessions and life.It is different from the total defense concept of Malaysia whereby Malaysian are perceived to have high level of loyalty and nationalism (Mohd Zackry Mokhtar, 2006:38-43).Arrifin Omar (2007) stated that:-"As a nation controlled by Chinese, Singapore must take into account the regional countries total population surrounded by the Malay Archipelago, if Singapore plans to attack Malaysia.In this context, it is positive that Singapore should not take the risked and bear the consequences in case it attacks Malaysia which is majorly populated by Malays." Hence, in the context of survival of both countries during war is different.Here, Malaysia is deemed to survive when a war breaks compare to Singapore for the spirits of Singapore citizen patriotism is weak.Therefore, Singapore would not adopt a military force approach against Malaysia due to the fact that there is still weakness in its doctrine of defense that it practices.Assessing from the aspect of its military capabilities, Singapore has assets that are offensive, yet it is characterized as merely a need for a country that has insecurity issues.Although Singapore has carried out many operation, training and war strategy in the jungle (jungle warfare), yet Singapore emphasize more on its military training performed in the city (urban warfare).This means Singapore's preparation is to defend the nation and not an attack on Malaysia which undeniably needs the jungle warfare strategy.Ariffin Omar (2007) said, if a country plans to go for war, the country must take into account the geographical aspect and its military ability and compare it to the environment of the battle field.This is because the decision to carry out war with no knowledge of the geographical condition would risk the country to face immense destruction and defeat.Although there exist several bilateral issues which rises tension between Malaysia and Singapore, the issues can be solved at a diplomatic level and not through the use of military force.The issues of water, border invasion (air and sea), islands dispute and other issues are issues that can be negotiated through diplomacy.According to Ahmad Ghazali Abu Hassan (2007):-"What is apparent is the existence of action reaction approach which often fluctuate the relationship.Historically it is evident that both countries have shown that they are prioritizing the method of diplomacy in settling bilateral issues" The rapidness of Singapore's military development is not at threat to Malaysia's security.This is because the relationship of both countries leaders is close and amicable.It must be understood that a decision to go to war or a conflict are in the hands of the leader (head of government) of a country.With the good ties among the leaders, it is the pioneer to increase confidence and trust in strengthening the relationship between Malaysia and Singapore.Both countries are manifestly serious in conducting and implementing measures in building confidence (Confident Building Measures or CBM).CBM is viewed from different perspective.Holst and Melander (1977:147) explained the concept of CBM as such, building confidence (CBM) involves the notification of credible prove that there exist no perturbing threats.Alford (1981:134) also explained CBM as a measures that can explain military actions or objectives, while Borawski (1986:3) describes CBM as a management tool to seek a way to control and notify how, when, where and why a military activity will be executed.According to Chalmers (1996:161), in Southeast Asia, CBM has started to be recognized since December 1993 when governments in this region started to implement CBM.He explained that:-"It has become imperative that confidence building measures (CBM) be introduced into the region with greater vigor.CBM possess a genuine promise for reducing the chances of unintended conflict and for improving the basic quality of a region political environment.They basically aim at enhancing transparency between states….CBM also seeks to make explicit military intentions in order to promote confidence by increasing the flow of information to make relations more predictable, thereby reducing the chances of conflicts and surprise attacks" . The move to build confidence pioneered by the top level has also been conducted at ministerial level and government official, either officially or informally In fact Singapore's readiness in sharing intelligence information since 2001 is an act of willingness to foster good relationship both military powers.Basically Singapore's military development does not give any threat to Malaysia's safety.This is because, since Singapore separation from Malaysia in 1965, there has been no security threat on Malaysia.From Malaysia's military perspective, Singapore is a country that is not classified as a major threat.On the other hand, Indonesia and Thailand are believed to be major threats on Malaysia's security compared to Singapore.This is because according to Ahmad Ghazali Abu Hassan (2007:-"If we assess which country has the ability to threaten Malaysia, it is not Singapore but Indonesia.This is because historically Malaysia has faced armed confrontations with Indonesia during the era of Sukarno.We should be reminded about the vision of a Greater Indonesia that was introduced by Sukarno, symbolizing that Indonesia has had the objective and agenda to conquer Malaysia" . The perception of this country regarding a threat is based on the history of Malaysia's confrontations with Indonesia that took place in 1963 (Patmanathan, 1980:23).The military was sent to confront Indonesian military attack that landed in Johor and was facilitated by Singapore to stop intelligence information to Indonesia in Malaysia (Aelina Surya, 1992:18).In fact, according to Ahmad Ghazali Abu Hassan (2007) this confrontation between Malaysia-Indonesia claimed a number of Malaysian troops in Borneo during the effort to protect national security.It is believed to be the sign and measurement of Malaysian military of Indonesia's ability to use its military force upon Malaysia (Tempur, April 2003:24).Ariffin Omar (2007) explained that:-"Although Singapore is strong in term of economy, political and military power, it is not a country that can easily set out a war because Singapore realizes that it is still lacking in terms of nationalism or patriotic spirit.The countries that can afford to threaten the security of Malaysia are Indonesia and Thailand." Ahmad Ghazali Abu Hassan (2007) perceives Indonesia and Thailand as nations that are able to threaten Malaysia's security.This is because Indonesia and Thailand are regarded as unstable states base on the instability of internal politics.Internal problems such as poverty, internal rebellion, ethnic conflict, weak government and terrorist issues make Malaysia prone to security threat through the spread of these internal problems to Malaysia (Jasbir Singh, 2003:66-68).Indonesia's and Thailand's weakness and failure to prevent internal problems would provide a direct impact on Malaysia such as the excessive immigration into Malaysia, making Malaysia a hide-out and the spread of terrorist activity are all other factors that formulate the threats from Indonesia and Thailand (Allan Gyngell,1983:116).The close ties between Malaysia and Singapore either from bilateral aspect or through international organizations, has been the pioneering of confidence and belief between both countries.The basic principle of ASEAN countries, that is not to intervene, will affect both countries to prevent from threatening each other sovereignty and security.Cooperation and agreement spirit emphasize stability of the region has become the essence of policy implementation of each country.Hence, any implementation of policy from any ASEAN member will take into account the region's interest.Five Nations Protection Law (FPDA) consisting of Malaysia, Singapore, Britain, New Zealand and Australia in 1971 has influenced understanding between countries to mutually help each another and can prevent military violence among members (Wariya,1989:79).In conclusion Singapore's military development does not cause security threat to Malaysia.However the issue is why have several statements by leaders and scholars questioned that Singapore's military development can cause security threat to Malaysia? Security dilemma In explaining the issue on the action of leaders and scholar that perceive Singapore's military developmental as threatening, we need to understand the concept of security.According to Snow (1998:23) the concept of security involves the freedom of mind from fear and danger pertaining two aspects: physical and psychological.In other words, security does not only exist in a physical form but also non-physical form (security dilemma).According to Nye (2005:38) and Mingst (2001:153&288), security dilemma exists when a country adopts to enhance military capability, consequently it would affect another country which will perceive the enhancement of military as a form of threat.Directly it would encourage the country to adopt the same approach because by increasing the military ability of one country it will raise insecurities to other country.If one country develops a military force it would effects directly to another country and makes it feel weak.Collin (2000:32-34) said the dilemma phenomenon of security in Southeast Asia is represented in two aspects: in the country (inter-state) and between countries (intra-state).One of the aspects that can raise security dilemma are conflict border, rapid military development in Southeast Asia such as in Singapore, Indonesia, Thailand and Malaysia that had a positive effect on stimulating and created the phenomenon of dilemma security among Southeast Asian countries.To ensure Malaysia's security and sovereignty, Malaysia's military power has used strategy which involves four stages namely detection, survival, strike and control.Each stage has approaches and specific measures.In explaining the impact of dilemma security on Malaysia from Singapore's military development, Ahmad Ghazali Abu Hassan (2007) explained that:-"At this time, we (the military) have detected several signs in Singapore like its rapid military development and problems involving Malaysia-Singapore's bilateral ties.Hence, we (the military) as much as possible would strongly inclined and encourage solution through diplomacy.In efforts to face this scenario our strategy (the military) is to strengthen the relationship with Singapore so that security threat risk can be eliminated.At the same period we (the military) must continue military development planning that was planned by the government to ensure our safety.Malaysian cannot avoid security dilemma when our neighboring country Singapore is developing its military power aggressively.Yet the existing security dilemma is under control.This is because we ourselves (Malaysia's military) have our own approach in handling this phenomenon". Although the above statements depicts that Malaysia is facing psychological impact (security dilemma), yet it is handled through strategy and approach whereby Malaysia's military especially has enhance confidence building measures (CBM) between Malaysia and Singapore.The military's strategy uses CBM as an approach to eliminate security dilemma and threat, through cooperation between both countries and defense force in ensuring confidence and belief between both countries that can be strengthened.It is consistent with Malaysia stance as a country that practice and emphasizes on peace and constructed relationship policy with neighboring countries, particularly Singapore.In conclusion, military development Singapore has a psychological effect (security dilemma) on Malaysia.Yet this effect was handled by Malaysia and its military through strategies and approaches that prevented this effect from affecting the relationship between both countries.(1993:136-152) believes that there are scholars trying to prove that the arms race phenomenon in Southeast Asia has existed since 1990s and described that after the Cold War (1991), NAT countries competed to obtain modern weapons to strengthen their defense system especially in the purchase of defense and weaponry equipment.In fact he observed that this phenomenon had made Asia the region that recorded a very high arms trade and if one country's military development process is uncontrollable (abnormal), then it would spur a phenomena of arms race between countries.Buzan (2000: 88-108) believes that one country's rapid military development will invite other countries reactions to develop its military to establish the balance of power.But does this arms race phenomenon also happen in Malaysia?The development of Malaysia's military force should not be looked as a reaction (arms race) against rapid development of Singapore's military force.Malaysia's military modernization planning was drafted earlier and implemented stage by stage according to the national economic position. Klare In fact the rapid military development in Malaysia has been conducted since the early 1990s and it is a normal development.The purchase of several strategic equipment like battleship KD Jebat and Kasturi in 1992 which at the time was the most modern and up-to-date warship in Southeast Asia, is seen as need for Malaysia in ensuring maritime sovereignty and security.It is corresponding to the fact that Malaysia is a country whose behave maritime state.Increase in assets and the purchase of 4 more Frigat's battle ships ( 2006) and 2 Scorpene submarines in 2002 conducted by Malaysia is not a reaction to counter Singapore's military capability, whereby at that time Singapore owns 4 submarine and warship furnished with missiles (1 Destroyer battle ship called Formidable, 6 corvette warships Victory, 6 boat with missiles Sea Wolf, 6 Fearless warship). Instead, the purchase of strategic defense equipment in Malaysia is seen as need of one country to make sure the level of power and capability of sea defense in Malaysia is ready to face any security threats forms outside.The purchase of two submarines which is a strategic equipment of sea defense and a process to make Malaysia have the ownership of underwater strategic weapons is to strengthen the maritime defense.According to the Director of Centepis Center UTM, Azmi Hasan (2007), the purchase of submarines is not characterized as a weaponry competition against Singapore; instead it is process in strengthening the national under water marine defense weaponry.The Air force development in Malaysia, through the purchase of 26 MIG 29 aircraft from Russia, 8 FA / 18D Hornet aircraft from US and latest 18 Sukhoi SU-30MKM aircraft said to be the most sophisticated from Russia, cannot be seen as a phenomenon of arms race with Singapore.Malaysian military development process is a step to ascertain that it is able to function as a credible army in handling any forms of threat against national sovereignty and interest.Malaysia does not wish to engage in any arm race with Singapore and at the same time also don't want to be left behind in the field of national defense.Arrifin Omar (2007) also said :-"In terms of sophistication Malaysia has made weaponry purchases and owns up-to-date aircraft enabling to increase Malaysia's strength.Yet in terms of quantity Singapore's has far greater ownership of aircrafts.Hence, arms race between Malaysia and Singapore is non-existent". According to Gray (1983) among the features in arms race is the involvement of two or more parties behaving aggressively hostile.Parties involved will compete to match one another in term of quantity (army personnel, weapons) or quality (military, weapons, organization, doctrine, location).Arms race phenomenon must possess a continuous increase in quantity and quality.The arms race assumption between Singapore Malaysia from the result of Singapore rapid development military is not conclusive.This is because military development process in Malaysia is based on plan and the planning is perform according to a specific time frame and also base on the countries financial capacity.What is certain is that the allocation of Malaysia's defense expenditure, since the country achieved independence in 1957, has never exceeded 5 %.This means there is no prove of Malaysia trying to match Singapore expenditure and military power.According to Acharya (2001), although there is an increase of defense expenditure allocation among Southeast Asian countries after the Cold War, it is only a record of figure increment.One country's military development is often misinterpreted by other country as a potential threat.This is influenced by psychological effect from the military development impact of the other country (Acharya,2001:136-141).Ahmad Ghazali Abu Hassan (2007) stated that:-"The development and purchase of our (Malaysian military) strategic and conventional equipment should not be considered as arms race.In fact, if we carefully analyze from our purchase of 8 FA-18 aircrafts, this lot is not qualified to be a squadron which must consist of 16 aircrafts.So where is the validity of the arms race assumption between Malaysia and Singapore?Moreover in terms of submarine purchase, it cannot be concluded as arms race because the purchase of 2 Scorpene submarines cannot match Singapore who owns 4 so far". Additionally, in his opinion the purchase and development should be viewed as prevention because it is to make Singapore aware that Malaysia also has strategic equipment and Malaysia would be able to fight if war would ever erupt between both countries.Thus the military development and modernization process in Malaysia currently are process to strengthen the existing deterrence system. Deterrence system There are raising questions concerning Malaysia's military development goals and objective that been increasingly active since the early 1990s.What are Malaysian military goals and objectives in its development and modernization process?Ahmad Ghazali Abu Hassan (2007) explained that:-"The military will continuously be modernized through phases according to its concepts which are deterrence and forward defense.At the same time, the development towards the direction of approaches will take into account the need of all three forces army, navy and air, to ensure their abilities are more effective." What is deterrence?Deterrence can be defined as a social and political contact especially to enable one party to influence the other party.It aims to ensure the enemy or opponent abides to the party who implements deterrence.Kegley and Wittkopf (1989:377) said that deterrence is a strategic capability to avoid from being attack by the enemy and it also is a move to convince the enemy not to take any action so that war can be prevented.For further explanation, please refer to the diagram 1. The application of deterrence concept in international relations is to make sure B does not take action or implement a policy that could threaten A's position.A will threaten a severe act if B continues its plan of attack and will have to pay the consequences.Therefore A's threats are aimed to warn and prevent B's harmful purpose.This can be concluded that the deterrence concept can be a tool of diplomacy, also known as diplomatic bargaining in the international relation arena.Threat which is used by a country to its enemy is a psychological tactic without involving the use of physical force that can produce.The implementation deterrence system has directly given the enemy an opportunity to weigh and reconsider the impact of their policy or strike before taking action. Usually a country would not execute an action that is unfavorable to itself.According to Buzan (1987:69), deterrence is a strategic capability to avoid any attack from the enemy.In other words, deterrence is a tentative to convince an enemy not to initiate a war.In explaining the goal of military development in Malaysia, it should not be considered as Malaysia's a preparation to act aggressive upon Singapore.Instead Malaysia is trying to create peace and stability with Singapore through the deterrence system.With this system it should influence both countries to weigh on decision of adopting military force against each other.Here the statements which were issued by Malaysia's leader announcing that Malaysia will respond to the attack if it was threatened are forms of deterrence approach.Both countries have already built their own deterrence system either through development of military physical approach or non physical approach.This system has successfully influenced both countries to weigh upon the impact if one launches an attack.The peace and stability between Singapore since 1965 until today are influenced by the success of the deterrence system applied by both countries.According to Ahmad Ghazali Abu Hassan (2007):-"If we compare military and defense capabilities of Malaysia with Singapore, Malaysia is lagging far behind.But with the strategic strength, even lacking in number, Malaysia still has succeeded in creating prevention of attack." Additionally he thinks that the deterrence system is not only limited to development of a defense force branch.For example Malaysia's purchase of 2 Scorpene submarine in 2002, although it cannot match the strength of Singapore's armada equipped with 4 submarines, it does not mean that Malaysia's deterrence failed.This is because Malaysia's defense concept is not only subjected to the navy but is also a Comprehensive Defense Concept or Total Defense (HANRUH).The concept of total defense involves all forces of security and defense, official public (volunteers force), which has given a great impact on Singapore in realize that Malaysia is strategically stronger.With the introduction and practice of joint warfare in Malaysia, this can send a message to Singapore not to belittle Malaysia's military powers.Indirectly this is another form of prevention to avoid any threats from Singapore.The military development in Malaysia has the purpose to establish a deterrence system for the enemy.It also is the aim of the government in ensuring the system and Malaysia's defense equipment is reliable and able to face enemy any threats.It does not aim to present a threat to another country, on the contrary it is the country's obligation in modernizing and upgrading the strength and military capabilities of Malaysia (Tempur, Julai 2003:2-5). Conclusion Singapore's rapid military development has been conducted since 1965 and has raise national security concern among Malaysian leaders.The emergence of bilateral issues such as water problem, border conflict and others have been the signs in assessing the possibility of threats that Singapore could inflict upon Malaysia.Actually these assumptions are irrelevant according to the Malaysian military's perspective, because it is just viewed as a mere development to fulfill this small country's need.The military development in Singapore doesn't intend to prepare a war on any country, but it is solely a deterrence system.The implementation deterrence system is the best asset for a country in guarantying its sovereignty and security without involving in war.Hence, offensive physical and non physical development is Singapore's approach to strengthen its deterrence system.To make sure the deterrence is strong, it has to be made aware among regional countries thus vast publicity is needed to convey the message of the deterrence system that is being implemented in Singapore. Singapore is a small country that is very depended on service oriented economy, apart from trade and port.Singapore's shortages of natural and labor resources are elements that influence Singapore's policy to not be an initiator of triggering war.Its economic dependence with Malaysia is also another factor why Singapore would avoid not adopting military force and not affect their Malaysia-Singapore relationship.A sour relationship as a result of inflicting military force would directly destroy Singapore's economic strength.It is unarguable that the military development in Singapore will give psychology implications to Malaysia especially security dilemma.This is because according to the theory security, any increase in military development will give implications to other countries and make them feel insecure.This security dilemma does not affect Malaysia's security and sovereignty, Malaysian military has plan a strategy that is to implement the confidence building measures (CBM) concept with Singapore.This strategy not only creates good ties among head of states but also involves relationship of both countries' ministries of defense.Sharing intelligence information and cooperation among both countries relating to security aspects will eventually affect both countries confidence and belief to increase and become stronger.This progress is sure to erase the security dilemma and security threat assumptions. The military development in Singapore has actually triggered Malaysia's awareness to address in modernizing its possession of military and defense.The purchase of military and defense equipment either strategic or conventional is Malaysia's action to ensure the system of defense and military are able to face any security and sovereignty threats.Malaysia's military development and modernization is to meet the planning and strategy which are fixed in accordance with a particular time span.In fact the purchase of defense equipment lies on the limited financial allocation.Hence, the assumption speculating an arms race by Malaysia's military from the development of Singapore's is not concluded, instead it is merely a development and modernization of both countries defense force.Basically Malaysia's military development and modernization is a process to strengthen the existing deterrence system.This is because, to ensure security and sovereignty are secured without launching a war, the implementation and enhancement of this system is the finest move.The purchase of modern equipment either conventional or strategic, is an approach that sends a warning to the enemy to consider the impact if they launch an attack on Malaysia. Table 1 . Singapore's military Budget 1996-2006 Source : Altered from The Military Balance 2006, International Institute for Strategic	v3-fos-license
2019-04-06T00:41:28.849Z	2019-04-01T00:00:00.000	96435925	{ "extfieldsofstudy": [ "Medicine", "Computer Science" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://jamanetwork.com/journals/jamanetworkopen/articlepdf/2729808/yu_2019_oi_190088.pdf", "pdf_hash": "3807b1b8783d1e767193f2d01dbbafe492071c48", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45424", "s2fieldsofstudy": [ "Computer Science" ], "sha1": "3807b1b8783d1e767193f2d01dbbafe492071c48", "year": 2019 }	pes2o/s2orc	Assessment of Automated Identification of Phases in Videos of Cataract Surgery Using Machine Learning and Deep Learning Techniques Key Points Question Are deep learning techniques sufficiently accurate to classify presegmented phases in videos of cataract surgery for subsequent automated skill assessment and feedback? Findings In this cross-sectional study including videos from a convenience sample of 100 cataract procedures, modeling time series of labels of instruments in use appeared to yield greater accuracy in classifying phases of cataract operations than modeling cross-sectional data on instrument labels, spatial video image features, spatiotemporal video image features, or spatiotemporal video image features with appended instrument labels. Meaning Time series models of instruments in use may serve to automate the identification of phases in cataract surgery, helping to develop efficient and effective surgical skill training tools in ophthalmology. eAppendix. Supplemental methods eTable 1. Metrics to evaluate performance of algorithms to classify phases in cataract surgery eTable 2. Differences in area under the receiver operating characteristic curve between pairs of algorithms for phase classification This supplementary material has been provided by the authors to give readers additional information about their work. eAppendix. Supplemental methods This section supplements the Methods described in the manuscript to provide sufficient detail to replicate our experiments. We used Python 3.5.2 with OpenCV and skvideo.io packages for video processing, scikit-learn for nearest neighbors, and PyTorch 0.4.1 as our deep learning Python library. Support Vector Machine (SVM) We used a SVM in Algorithm #1. The input to the SVM is labels for instruments that are in use at a given image or frame in a video, (i.e., cross-sectional data). We adopted a one-vs-the-rest strategy to implement the SVM, i.e., we fitted one SVM per phase to classify whether a given vector of instrument labels (in use at the time) belongs to the particular phase or any of the other phases. We fitted 10 such classifiers, one for each phase. We represented labels of instruments in use in a given video frame as a dimensional indicator vector where the -th dimension is 1 if the corresponding tool is in use. Let … be a set of labels of instruments in use within phase of video clip . We select a subset * … * from such that * is a unique set of features. For example, if a particular phase contained only two unique combinations of instrument label, we included the two as part of the training data. The collection of * for all s and s constituted the dataset to fit our multi-class SVMs. We used a linear SVM implementation provided by sklearn and performed a grid search over the penalty parameters ∈ 0.1, 0.2, 0.5, 1, 2, 5 of the error terms. Temporal Recurrent Neural Network (RNN) We used the RNN in Algorithms #2, #4, and #5. For the RNN, we implemented a standard long short-term memory (LSTM) model. We implemented three separate RNN models with different inputs: 1. Spatial features obtained from the spatial CNN 2. Spatial features obtained from the spatial CNN appending instrument labels 3. Instrument labels alone Given a single frame in the video, the output dimension of SqueezeNet (spatial CNN model) is a vector of size 512 and the dimension of the instrument labels (annotations) is 14. Thus, the inputs for the three separate RNN models itemized above are 512 by temporal length , 526 by temporal length , and 14 by temporal length , respectively. We structure the RNN as follows:  An LSTM layer that outputs a 512 dimensional vector. The input at each time-step is fed into this layer and saved.  A global average pooling across time-steps.  A fully connected layer that outputs a feature vector of length 128.  The resulting feature is trained to classify phase of the clip using softmax activation and cross-entropy loss. The following are the optimization parameters to replicate our study. We optimize the temporal RNN using the ADAM optimizer with the following parameters:  Spatial Convolutional neural network (CNN) We used the CNN in Algorithms #3, #4, and #5. To implement the spatial CNN, we utilized the SqueezeNet architecture. SqueezeNet consists of multiple fire modules. Each module contains a series of 1 × 1 convolutional filters to "squeeze" the information followed by a set of 1 × 1 and 3 × 3 filters to expand. In the standard architecture, there are 8 fire modules. The number of filters in each of the modules is given in Figure 1. SqueezeNet is implemented as follows: Convolution with n filters, where n is the number of classes. In our case, n is 10. Softmax. Figure 1: Information on the number of filters in each fire module. Note that in the SqueezeNet paper, these modules are labeled "fire2" through "fire9". We take PyTorch's implementation of SqueezeNet from their pretrainedmodels package, already pre-trained on ImageNet. We then replaced the last convolutional layer with one that has 10 classes rather than 1000 in order to allow for classification of the 10 phases. The model is then finetuned to classify phase on cataract surgery images. To do this, we first extract training examples from our videos through the following steps.  One frame per second is extracted from all surgical videos. This is done while still maintaining the split between training and evaluation data.  This image training data is then balanced so there is an even distribution of images for each phase. We then finetuned the model on the above data set using the ADAM optimizer with the following parameters:  Initial Learning Rate: 0.001	v3-fos-license
2022-10-01T15:02:42.886Z	2022-09-28T00:00:00.000	252630554	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2227-9032/10/10/1898/pdf?version=1664360012", "pdf_hash": "ccec6624d7d26d0a245bf6b9aae66e12a33a6f3f", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45425", "s2fieldsofstudy": [ "Medicine" ], "sha1": "c203c292b5bb6306a51dc7eb58257838ba57b86e", "year": 2022 }	pes2o/s2orc	Disappearing Colorectal Liver Metastases: Do We Really Need a Ghostbuster? The development of new systemic treatment strategies has resulted in a significant increase in the response rates of colorectal liver metastases (CRLM) in the last few years. Although the radiological response is a favorable prognostic factor, complete shrinkage of CRLM, known as disappearing liver metastases (DLM), presents a therapeutic dilemma, and proper management is still debated in the literature. In fact, DLM is not necessarily equal to cure, and when resected, pathological examination reveals in more than 80% of patients a variable percentage of the tumor as residual disease or early recurrence in situ. Moreover, while a higher incidence of intrahepatic recurrence is documented in small series when surgery is avoided, its clinical significance for long-term OS is still under investigation. In light of this, a multidisciplinary approach and, in particular, radiologists’ role is needed to assist the surgeon in the management of DLM, thanks to emerging technology and strategy. Therefore, the aim of this review is to provide an overview of the DLM phenomenon and current management. Introduction Colorectal cancer (CRC) is the third leading cause of cancer-related death worldwide in both males and females [1]. Over 15% of patients present liver metastases at the time of diagnosis, and up to 50% develop them within 3 years [2]. Among patients with metastatic colorectal cancer (mCRC), approximately 20% of them survive beyond 5 years from diagnosis [3]. Although systemic chemotherapy is broadening its indications in patients with mCRC, surgical treatment is the only definitive treatment, and thus, multidisciplinary decisionmaking is mandatory in the different subsets of patients (e.g., multiple sites of liver-limited disease that is not resectable, or resectable disease with a high risk of recurrence). In this subset of patients, multimodal ablative therapy may result in improved outcomes compared with continued systemic therapy alone [4][5][6]. Currently, the number of patients with colorectal liver metastases (CRLM) who are a candidate for hepatic resection is significantly increased due to new surgical strategies and oncological therapies, which increase the rate of resectable disease [7][8][9]. In fact, while upfront surgery is the treatment of choice in patients with resectable disease, over 60% of patients with CRLM present an objective response to chemotherapy, according to RECIST 1.1. criteria, and this is usually an indicator of favorable prognosis and can be resected after chemotherapy [10,11]. Moreover, a complete radiological response with the disappearance of CRLM can occur in a minority of patients. This phenomenon, named disappearing liver metastases (DLM), also known as vanishing, missing or ghost CRLM, is based on the complete shrinkage of liver metastases after chemotherapy. DLM incidence is highly variable, depending on the accuracy and quality of restaging imaging with up to 37% of patients undergoing systemic treatments [12]. Among different strategies, magnetic resonance imaging (MRI) and contrast-enhanced intraoperative ultrasound (CEIOUS) are the most accurate imaging modalities and should be adopted routinely for detecting DLM [13]. However, the optimal management of DLM is still controversial due to the uncertainty of residual microscopic disease and effective long-term outcomes in resected versus un-resected patients [14,15]. This review aims to provide an overview of the state of the art in the management of DLM, which is still a critical challenge in clinical practice. Systemic Treatment of CRLM The liver is the most common metastatic site for CRC patients, and 20 to 34% present with synchronous hepatic involvement at diagnosis [16][17][18][19]. Chemotherapy represents the best treatment for most of these patients. However, especially in the case of metastatic liverlimited disease, a conversion approach could be considered, and surgical re-evaluation could be planned every 2 months from the beginning of therapy. To date, there is no standard for this attempt, and chemotherapy backbone choice, represented by 5-fluorauracil/leucovorin (5FU/LV), irinotecan (IRI) and oxaliplatin (OXA), depends on clinical preference and patient clinical conditions. Previous retrospective studies on initially unresectable mCRC treated with 5FU/LV combined with irinotecan (FOLFIRI) or oxaliplatin (FOLFOX) showed positive conversion rates for both the combined schemes (32.5 and 40%, respectively) [20,21]. However, secondary hepatotoxicity due to both irinotecan and oxaliplatin, represented by steatohepatitis and sinusoidal liver injury, respectively, does not support the use of one drug or the other [22,23]. Besides FOLFOXIRI, prior therapy international guidelines suggest that all patients with mCRC should be tested for exons 2, 3 and 4 of rat sarcoma virus (RAS) (K-and N-) and BRAF genes mutations as well as for microsatellite instability (MSI) proteins or mismatch repair (MMR) gene deficiencies. These genetic tests are needed to determine a tumor's biological features and support clinicians in selecting the best monoclonal antibody to add to chemotherapy [25]. RAS/RAF wild-type patients are eligible to be treated with anti-epidermal growth factor (EGFR) monoclonal antibodies (e.g., cetuximab and panitumumab). Previous posthoc analysis of patients' liver conversion rate treated with FOLFOX or FOLFIRI alone or plus cetuximab showed a positive trend from the addition of anti-EGFR in KRAS wild-type selected cohort (60 vs. 32%) as for the CELIM trial [7]. Similar results were obtained in a single-centre pan-Asian experience, in which KRAS wild-type liver-only mCRC were randomized to receive chemotherapy alone or with cetuximab, confirming the superiority of the combined treatment on FOLFOX or FOLFIRI alone in terms of R0 and objective response rate (25.7% vs. 7.4% and 57.1% vs. 29.4%, respectively) [26]. Considering the triple drug combination efficacy, the phase II trial VOLFI tested the efficacy of panitumumab when combined with FOLFOXIRI vs. chemotherapy alone. Overcoming the predefined objective response rate cut-off of 75% (87.3 vs. 60.6%), data support FOLOFOXIRI plus panitumumab [27]. Finally, further evidence will be available from the Italian phase III TRIPLETE trial. The TRIPLETE trial, which aimed to compare FOLFOXIRI plus panitumumab vs. standard of care (FOLFOX Panitumumab), just finished enrolment, and their results will soon be published [28]. The presence of RAS/RAF mutations contraindicates anti-EGFR agent use due to intrinsic drug resistance [29], while promoting the use of the anti-vascular endothelial growth factor (VEGF) monoclonal antibody bevacizumab in the first line setting. However, data about the additive effect of this agent on two-agent chemotherapy in terms of resection rate are controversial. While negative results were obtained in a big cohort of 1400 mCRC patients, on the other hand, the BECOME trial results sustained the addition of monoclonal antibodies to the oxaliplatin-based scheme [30]. In the BECOME trial, among 241 initially un-resectable RAS mutant patients, randomized to receive FOLFOX with bevacizumab or alone, 27 of 121 in the experimental arm (vs. 7 of 120 patients of the chemo-only arm) underwent R0 surgery and reached a statistically significant objective response rate (p < 0.01). Stronger-performing results on overall response rate, as above, were obtained in several phase II and III trials by the triplet scheme FOLFOXIRI with bevacizumab [31][32][33]. In this regard, in a pooled analysis of 3 prospective Italian trials on more than 200 patients with liver-limited disease receiving the triple combination plus anti-VEGF, 69% of patients reached a radical surgery attempt, resulting in R0 for 30.7% of them. Moreover, R0/1 resected patients had longer survival compared to other patients, both in terms of PFS and OS, independently from mutational status, including BRAF mutant tumours, which still represent an oncological challenge due to their aggressive biology [34]. The advent of immunotherapy changed the treatment paradigm in MSI-high or MMRdeficient (dMMR) mCRC patients. Anti-programmed death (anti-PD1) antibody, alone or in combination with anti-CTLA4, showed significantly longer survival periods, reaching higher rates of complete and sustained complete response in this cohort compared to previously obtained chemotherapy results. The overall response rate of pembrolizumab alone, nivolumab alone or in combination with ipilimumab was 43.8, 31.1 and 55%, respectively [35][36][37]. In light of this, future studies will need to investigate the potential role of surgery or local therapy in this subset of patients after a first immune checkpoint inhibitor (ICI) approach. Predictors of Complete Radiological Response after Chemotherapy In the last decades, several factors have been identified as possible predictors of developing DLM. Among them, low carcinoembryonic antigen (CEA) levels at diagnosis and its normalization after chemotherapy, patients younger than 60 years, re-staging with magnetic resonance image (MRI), synchronous disease, size and number of LM at diagnosis, number of chemotherapy cycles, agents included in regimens and the use of hepatic arterial infusion (HAI) chemotherapy have been associated with the confirm of DLM with a complete pathologic response [38][39][40][41][42]. Currently, the focus is shifting to tumor burden. Xu et al. showed that the early primary T stage (T1-2 vs. T3-4, OR [95% CI]: 3.131 [1.213-8.082], p = 0.018) was an independent predictor of pathologic complete response after preoperative chemotherapy [43]. Factors predisposing to the development of DLM are summarized in Table 1. However, there is no absolute consensus on the criteria for complete pathologic response, and in fact, several authors analyzing the correlation between the size of CRLM prior to chemotherapy and the onset of DLM revealed wide variability in the mean or median size of CRLM. Moreover, most of these studies were performed before the advent of target therapy. The likelihood of developing a "true" complete pathological response after chemotherapy is higher in patients with a combination of these factors, and it is associated with both prolonged survival and decreased risk of recurrence [42]. The "Histological Truth" behind Complete Radiological Response According to the new guidelines established through RECIST 1.1 criteria [10], the goal of the treatment of solid tumors is defined as the disappearance of the target lesions. However, a trend toward an increased rate of DLM leads the surgeon to a decision-making dilemma: to resect or not the sites of the now-missing targets. Data collected through an international survey [40] revealed that hepatobiliary surgeons do not have a unified attitude in the management of this challenging setting. In several series, the pathologic examination of the DLM on imaging revealed a variable percentage of macroscopic or microscopic residual disease or early recurrence in situ up to more than 80% [12,44]. The histological results obtained from the analysis of the resected DLM are shown in Table 2. The Role of the Radiologist It is sometimes difficult in routine radiological practice to detect DLM. Most of the time, when patients have MRIs to monitor their disease, they have already received neoadjuvant therapy. The clinician then asks about the status of the liver parenchyma and the evolution of liver metastases, which are difficult to detect on ultrasound and/or MRI due to the effect of primary medical treatment [45]. In order to detect residual tumors, a classical protocol (T1 without and with multiphasic gadolinium injection, a diffusion sequence and with ADC mapping, a T2 Fat-SAT sequence) is often used. In most cases, these sequences are sufficient to assess CRLM but are not the only ones used. This is when the role of the radiologist is pivotal to detect any sign of remnants, which are not always visible on other sequences [46]. For instance, some metastatic lesions are no longer visible on T1 sequences without and with injection and are, therefore, extremely difficult to visualize. Sometimes the only sequence that allows the radiologist to detect residual tumor lesions is the diffusionweighted sequence. Indeed, at high "B" values, these "vanishing lesions", which are not visible on the other sequences, may still show a diffusion restriction. We, therefore, know that they are the sites of residual tumor cells [47]. However, when the classical protocol is not sufficient to assess CRLM, other strategies are needed. In the past two decades, MRI has demonstrated enhanced accuracy thanks to novel hepatocyte-specific contrast agents such as gadolinium ethoxybenzyl dimeglumine (Gd-EOB-DTPA) [55]. Gd-EOB-DTPA, which behaves similarly to traditional contrast agents in the dynamic phases, adds substantial information in the hepatobiliary phase, detecting and characterizing focal liver lesions or diffuse liver disease [56]. The importance of Gd-EOB-DTPA has been described in a retrospective analysis by Morin et al., where among 110 patients with liver metastasis, in 43% of patients, Gd-EOB-DTPA revealed a different number of liver lesions, and it potentially modified surgical planning in more than 17% of patients [55]. Furthermore, as the future of radiology and hepatobiliary surgery will tell us, the role of the radiologist is changing, and a radiologist must be able to use artificial intelligence tools, including 3D reconstruction and radiomics, that surely will be involved in solving the DLM dilemma [57][58][59]. In fact, radiologists' role is also to help the surgeon plan the surgical procedure by positioning the residual lesion (sometimes visible only on the diffusion-weighted sequence) in relation to the other vessels using 3D modelling. In this way, the surgeon in the operating room will be able to locate the lesion more easily during the operation, because, very often, even with intra-operative ultrasound, the surgeon cannot find the residual tumor site [60]. Moreover, 3D models can be integrated into navigated image guidance systems (IGS) to guide surgery and provide additional information to the surgeon, merging all the information provided from the CEIOUS and preoperative imaging [61]. The surgeon will therefore have to trust the 3D model previously designated by the radiologist on the basis of the MRI and the diffusion sequence. Furthermore, 3D printed models generated from medical images that are graspable and patient-anatomy-specific will improve the understanding of preoperative surgical liver anatomy and, eventually, surgical resection accuracy [62]. However, radiologists' role is not focused on guiding surgical plans and positioning the residual lesions, but their role is critical in multidisciplinary teams. For instance, in the initial setting, during a multidisciplinary assessment, radiologists should detect the DLM high-risk lesions, for which a fiducial marker could reduce the risk of DLM, representing the real ghostbuster [63,64]. Image interpretation by radiologists could guide initial therapy discussions as well as interpret post-treatment imaging following liver-directed therapy, providing additional information to all the members of multidisciplinary team. Moreover, radiologists in the perioperative settings could help the surgeon beyond CEIOUS, with local percutaneous treatment or in case of complications (e.g., stenting or abdominal drainage) [65][66][67]. Novel Strategies in the Management of Patients with DLM Although radiological morphology could be considered a surrogate of complete pathologic response (CPR), the finding of DLM in cross-sectional imaging does not mean that those lesions have been cured. Furthermore, a recent consensus on hepatic resection for CRLM established that surgery should include all sites of liver metastases described before chemotherapy [68]. However, no recommendation based on hard evidence supports whether DLM should be resected or left in situ [13]. In fact, surgical treatment is the only curative treatment for CRLM, and the development of novel chemotherapy, plus the advance in radiology, as mentioned before, and surgery has ameliorated the clinical outcome of CRLM [69]. Advocates of DLM resection mention the low incidence of CPR (20%) and higher rates of CRLM recurrence in situ (70%) due to the microscopic residual disease or the presence of a supportive microenvironment for tumor relapse [19,[70][71][72]. In surgical planning, even with the 3D reconstruction, CEIOUS represents a valuable instrument in the hand of the surgeon that can decipher the DLM dilemma, increasing the detection rate of DLM when compared with Gd-EOB-DTPA alone. In fact, Gd-EOB-DTPA + CEIOUS, with a sensitivity of 93% and specificity of 73%, could detect the clinically relevant DLM with viable tumors, thus at risk of local recurrence [49]. Regarding the surgical procedure, one-stage parenchymal-sparing hepatectomy (PSH), when feasible, had comparable safety and efficacy when compared with anatomical resection (AR) and did not compromise oncological outcome, reducing the hepatic parenchyma resected [4]. Furthermore, whereas R1 resections increase the likelihood of locoregional recurrence and redoing liver surgery, R1 resection with detachment from major intra-hepatic vessels (R1Vasc) was demonstrated to achieve an outcome equivalent to R0 resection [4,73]. However, in some cases, complete eradication of the original site of the disease could be very complex to perform due to the localization deep in the parenchyma; moreover, an extensive resection could involve the risk of an insufficient liver remnant. For these reasons, traditional surgery has moved towards new strategies such as two-stage hepatectomy, portal vein embolization (PVE) and associating liver partition and portal vein ligation for staged hepatectomy (ALPPS) [74]. Finally, alternatives to surgical strategy that should always be evaluated are cryotherapy and radiofrequency ablation (RFA). Microwave ablation (MWA) can also be evaluated in the case of extrahepatic disease, the deepest position where the surgical procedure could compromise too much parenchyma, patients' comorbidities [75]. For instance, a recent metanalysis indicates how MWA can represent a valid tool to treat CRLM, especially in those smaller than 3 cm, alone or in combination with hepatectomy to expand the pool of resectable patients [65,75]. Besides, a surgical strategy should be driven not only by the anatomical and technical variables but also considering molecular disease characteristics, such as RAS mutational status, to decide margin width and, in the case of DLM, the indication to proceed with resection [76][77][78]. Some studies have already started to compare different treatment approaches (watchful waiting vs. resection). Goere et al. reported a 5-year OS and recurrence-free survival (RFS) of 80% and 23%, respectively and a median recurrence time between 13.8 and 21 months. In their study cohort, adjuvant chemotherapy with oxaliplatin-based hepatic arterial infusion resulted in a lower rate of intrahepatic relapse [50]. Similar results were confirmed by Tanaka et al., who reported, after a median follow-up of 44 months, a lower recurrence rate in DLM resected patients when compared with watchful waiting (24.4% vs. 40.7%, respectively) [41]. Another study by Owen et al. demonstrated no difference in terms of RFS between DLM left in situ (360 days) and resected (483 days) [46]. However, the clinical significance of intrahepatic recurrence may not automatically determine a detrimental effect of OS. In fact, although Van Vledder et al. reported a higher rate of 3 years intrahepatic recurrence in the surveillance group when compared to the surgical group, in the same series, no survivorship advantage was reported at 1, 3, and 5 years [42]. In light of this, current evidence demonstrates that DLM resection could determine a benefit in terms of RFS, and DLM resection seems not to affect survivorship despite the high risk of residual disease in the previous tumor bed [41,42]. Moreover, the COVID-19 emergency determined a change in colorectal patients, leading to a higher rate of delayed presentation, delayed treatment, and oncological emergency with novel real-life evidence that could determine a change in clinical practice [79][80][81][82][83]. On the basis that maximizing resection of CRLM still remains the main objective, some authors have experimented with marking lesions at high risk of disappearing with a fiducial prior to initiation of chemotherapy or immediately before surgery. Vujic et al. demonstrated the importance of positioning a CT-guided marker in DLM after neoadjuvant treatments and observed a complete histological response in only 18% of resected sites [63,64]. However, the fiducial marker placement is not without possible complications; therefore, it should be considered, especially in CRLM at high risk of being missed. Kepenekian et al. described the fiducial placement in lesions smaller than 25 mm in diameter, deeper than 10 mm in the hepatic parenchyma and sited outside the field of a planned resection [63]. Fiducial placement could change the paradigm in the high-risk DLM treatment by allowing the surgeon to move towards tailored and more radical resections. Conclusions The complete curing of CRLM with systemic therapy is a rare phenomenon that now occurs in less than 5% of cases. However, thanks to the development of innovative oncological strategies, it is safe to state that in the future, a higher rate of patients will develop DLM [84]. Still, most complete radiological responses actually harbor macro or microscopic foci of residual disease. Yet, resection of DLM can be technically troublesome. For this reason, it is necessary to perform detailed restaging after and during chemotherapy with accurate localization of all sites of CRLM previously described as the key point for the correct treatment [85]. An algorithm for preoperative DLM management has been proposed ( Figure 1). Still, most complete radiological responses actually harbor macro or microscopic foci of residual disease. Yet, resection of DLM can be technically troublesome. For this reason, it is necessary to perform detailed restaging after and during chemotherapy with accurate localization of all sites of CRLM previously described as the key point for the correct treatment [85]. An algorithm for preoperative DLM management has been proposed ( Figure 1). Despite preliminary experiences demonstrating how watchful waiting could represent a safe alternative to DLM, future studies are needed to determine the risk of relapse in DLM. While waiting for artificial intelligence, radiomics or other innovative diagnostic techniques, fiducial marker placement in CRLM at high risk to become DLM will represent, in the next years, the practice-changing procedure that will help as a ghostbuster, reducing the rate of undetected DLM and providing a guide to solve the DLM dilemma with resection of the fiducial tissue with adequate margins.	v3-fos-license
2023-02-14T14:07:02.842Z	2022-07-03T00:00:00.000	256830046	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://doi.org/10.1080/13632752.2022.2092274", "pdf_hash": "1f9e39a5e705eb1c3740852a3f62c1fbe3bf5cbd", "pdf_src": "TaylorAndFrancis", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45428", "s2fieldsofstudy": [ "Law", "Education", "Sociology" ], "sha1": "f2d06b22270784acf85f8eb63334868d76f026da", "year": 2022 }	pes2o/s2orc	Illegal school exclusion in English education policy ABSTRACT This article draws upon legislation, policy, guidance and recently published research in order to explore why illegal exclusion, sometimes known as off rolling, is an under-researched area of education, and what the most significant barriers are to gaining a clearer understanding of the scale of, and reasons behind, the practice. In doing so, the article draws attention to policy obfuscation, inconsistent definitions and a desire to ‘name and shame’ offending schools. Furthermore, the contention is that school exclusion policy and safeguarding policy are not aligned, which means that many children, particularly those who are statistically more vulnerable to exclusion are denied their right to both education and to be kept safe from harm. I conclude by arguing that a shift in priorities, away from punishment and towards upholding safeguarding legislation, is required. Introduction Illegal exclusion from mainstream schools in England is a chronically under-researched area of education. This article draws upon key legislation, policy and guidance concerning both school exclusion and safeguarding, as well as recently published research on this topic to explore why this may be the case, and what the most significant barriers are to gaining a clearer understanding of the scale of, and reasons behind, illegal exclusion. In doing so, tensions are highlighted between the pressure for mainstream schools to ensure the inclusion and safeguarding of all pupils while navigating the quasi-market, illustrating how the illegal exclusion of pupils is embroiled in a perfect storm of cuts to key support services (Davies, Diamond, and Perry 2021), dominance of a punitive discourse in behaviour policy (Thompson, Tawell, and Daniels 2021;Ferguson 2021), and variations in the definition of what means to illegally exclude a child from school, combined with a lack of awareness of the law (CCO (Children's Commissioner's Office) 2013; Timpson 2019; Ofsted 2021a). This article argues that school exclusion policy and safeguarding policy are not aligned, which means that many children, particularly those who are statistically more vulnerable to exclusion are denied their right to both education and to be kept safe from harm (Porter 2016; Davis and Marsh 2020). The failure to align safeguarding policies with exclusion policies in a system riddled with perverse incentives to exclude (Thompson, Tawell, and Daniels 2021), increases the risk of vulnerable children and young people becoming 'collateral casualties of policy' (Thompson, Tawell, and Daniels 2021, 42). Struthers (2021) claims that this situation is not helped by the fact that the current safeguarding training given to teachers both pre-service and as continual professional development fails to incorporate human rights education, therefore preventing teachers from being empowered to act by a legal knowledge of human rights. Such an argument serves to illustrate a potential solution to the policy disconnect that is referred to throughout this article. In addition, the article explores the presence of policy obfuscation, where the very guidelines that are supposed to ensure the legal exclusion of pupils from school fail to provide an agreed, consistent definition on what an illegal exclusion is and the circumstances in which it may occur (DfCSF (Department for Children, Schools and Families) 2008; Timpson 2019; Ofsted 2021a) -leaving the children and parents who are vulnerable to this process in, what Robert Halfon MP (Con) described as, a 'Wild West of exclusions' (TES 2018, para.13). While large scale quantitative data is limited on the topic of illegal exclusion/offrolling we do know that it is children and young people who belong to most vulnerable groups who are most likely to be impacted by this practice (Hutchinson and Crenna-Jennings 2019) and swept up in this seemingly simple, but in reality very complex, legal tangle surrounding the use of school exclusion. Illegal school exclusion/off-rolling in England The true nature and extent of illegal exclusion from school in England is currently unknown however, articles published in the media do allow for an insight into the reasons why illegal exclusion occurs. In 2017, the Guardian published a story which detailed how a top performing grammar school located in Orpington, London had illegally excluded sixteen pupils who did not perform as well as they were expected to in their AS levels. By issuing illegal exclusions, the school prevented these pupils from obtaining their full A level at the school while ensuring the best possible league table result for the school itself. The article goes on to reveal that issuing illegal exclusions for the purpose of maintaining/obtaining excellent league table results is not uncommon amongst high achieving grammar schools, naming at least one other institution where this practice was rumoured to have occurred (Weale 2017). More recently, in 2021, the Telegraph and Argus reported that schools in West Yorkshire were attempting to obtain Education, Health and Care (EHC) plans for low achieving students in order to ensure their grades did not contribute to the schools league table results -some schools were also 'trying to get them into other settings' (Beecham 2021b, para. 8). Reports of these illegal exclusions for the purposes of manipulating league table positions prompted Leeds City Council to launch an investigation into off-rolling in their jurisdiction (Beecham 2021a). In addition, media articles also provide insight into how schools engaging in the process of off-rolling are reprimanded. In 2019, the Times Educational Supplement challenged Ofsted's consistency in reporting off-rolling when it was revealed that of two schools found to be illegally removing pupils from the school roll, only one was downgraded to Requires Improvement (Roberts 2019). When questioned about this, Ofsted stated 'The report speaks for itself and we have no further comment to make' (Roberts 2019, para.12). Furthermore, in 2021, Schools Week reported that three academy leaders had been found guilty of serious misconduct relating to historic off-rolling occurring between 2013 and 2016 which, again, involved the removal of pupils from the school roll for the purposes of improving/maintaining league table results (Belger 2021). In an attempt to address this issue, Ofsted commissioned YouGov to conduct a survey of 1,018 teaching professionals in an attempt to establish what people considered off-rolling to be and how widespread the practice was in English schools (Ofsted 2019a). Ofsted followed up this survey with a blog post which detailed how they identify and report off-rolling when inspecting schools (Owen 2019). Despite this, Ofsted have since admitted that, due to linguistic variations in Ofsted reports, some instances of off-rolling were not flagged and therefore, were not investigated (Roberts 2021). Timpson review of school exclusion The Timpson Review arose from the Race Disparity Audit (Cabinet Office, 2017) in which exclusions were highlighted as a specific site of racial inequalities within education. The audit was commissioned by then Prime Minister, Teresa May, who in 2016 spoke of the 'burning injustices' present in the UK; specifically referencing the disproportionate exclusion of Black Caribbean boys (BBC News, 2016). Timpson's review comprised evidence gathered from nine local authorities in England and, in each location, discussions with local authority representatives, parent groups, school leaders and charities. In addition, roundtable meetings with practitioners, pupils, leaders and academics were held (Timpson 2019). Timpson also commissioned the re-analysis of Department of Education exclusion data, a literature review exploring research on commonly excluded groups, and research into the experiences of children, parents and carers who have experienced exclusion; the latter was undertaken by the children's charity, Coram. The purpose of such an extensive review was to gain a better understanding of how exclusion policy is implemented in the school setting and why certain groups of pupils, namely, pupils with special educational needs (SEN), pupils eligible for free school meals (FSM), pupils who have experienced the social care system and pupils from certain ethnic groups, are more likely to be excluded from school. Timpson concluded by setting out 30 recommendations which were intended, when implemented, to ensure a fair, reasonable and lawful school exclusion process (Timpson 2019). Timpson's focus was on what might be categorised as lawful permanent exclusions and suspensions from school. However, it is significant that illegal exclusion is referred to several times throughout the document when it is primarily described as off-rolling. Timpson defines off-rolling as a practice where children: [Children] are made to leave their school and are removed from the school roll without a formal permanent exclusion or by the school encouraging the parents to remove their child from the school, which is done in the school's interests, and at the school's request (Timpson 2019, 10) Although Timpson was not looking for evidence of illegal exclusions and/or off-rolling, eight of the nine local authorities visited admitted to a knowledge of off-rolling taking place in local schools; such admissions were corroborated by parents, carers and even schools. Some parents and carers involved in the review recalled instances of being encouraged to educate their children at home or move them to another school in order to avoid a permanent exclusion appearing on the child's record, and such advice was being given even where appropriate alternative educational provision was not available; one headteacher admitted to a knowledge of off-rolling taking place in their own school (Timpson 2019). Timpson is very clear about his views on off-rolling, describing it as 'quite simply wrong' and a 'rare but unacceptable practice' that is both a safeguarding issue in itself and has the potential to expose children to further safeguarding risks (Timpson 2019, 11). However, research on this issue has shown off-rolling to be anything but rare. Prior to the Timpson review being published, the CCO (Children's Commissioner's Office) (2013) in a report on illegal exclusion claimed, 'even at the most conservative estimates . . . an unacceptably large proportion of schools are acting illegally' (CCO (Children's Commissioner's Office) 2013, 10). Further, research from the Education Policy Institute found that 1 in 10 of the pupils in the cohort they studied 'experienced an unexplained exit at some point during their time at secondary school' (Hutchinson and Crenna-Jennings 2019, 9). In addition, at no point does Timpson (2019) shed any light on which pupils are more likely to be off-rolled and why that is the case. Fortunately, the Education Policy Institute research does dedicate attention to this, finding that many of the children who experienced unexplained exits in their sample were from vulnerable groups including; 'one in six children ever identified with SEND and children ever eligible for free school meals', 'one in seven of those with low prior attainment and of those from Black ethnic backgrounds', and 'close to a third of current or formally looked after children' (Hutchinson and Crenna-Jennings 2019, 10). As such, Timpson's instance that off-rolling/illegal exclusion are rare practices serves to further obscure and minimise concerns about the practice (Bei, Knowler, and Butt 2021). Timpson's (2019) only recommendation that specifically addresses off-rolling goes no further than to encourage Ofsted to routinely consider patterns to off-rolling and, where evidence is found of this practice, this 'should always be reflected in inspectors reports and, in all but exceptional cases, should result in a judgment that the school's leadership and management is inadequate' (Timpson 2019, 101). Yet, as Ofsted themselves have admitted, some instances of off-rolling have not been reported or investigated due to linguistic variations in how this practice is described in Ofsted reports (Ofsted 2021b). The fact that the Timpson Review (Timpson 2019) never intended to go beyond an investigation of official and lawful school exclusions reveals a significant oversight that suggests a degree of naiveté, perhaps even wilful ignorance, on the part of Timpson and his team of researchers around the reality of exclusions in English schools. As such, the experiences of tens of thousands of children (Hutchinson and Crenna-Jennings 2019) were discounted because the practice of off-rolling was, 'not originally in the scope' (Timpson 2019, 27) of the Timpson review. Moreover, in neglecting to investigate the conditions and cultures which allow pupils from Black ethnic backgrounds to be disproportionately off-rolled (Hutchinson and Crenna-Jennings 2019), the Timpson Review fails in its fundamental aim to address the 'burning injustices' present in education (BBC News 2016). Furthermore, it demonstrates 'tacit intentionality' (Gillborn 2007) in that while the review did not explicitly intend to further racial inequity, the failure to meaningfully address racial disproportionality in off-rolling practices means that this inequity will continue to be experienced. For this reason, the Timpson Review should not be hailed as a landmark report as it has been by so many (DfE (Department for Education) 2019; George 2019b; Booth 2021). Hence, it is important to consider its findings and recommendations regarding illegal exclusion in the context of other policy documents in order to establish how the issue is being approached at governmental and policy level. Legislation School exclusion and its use as a disciplinary measure is outlined in legislation and accompanying guidance. The current law regarding school exclusion is set out in four key policy documents: the Education Act (2002); the Education Act (2011), the School Discipline (Pupil Exclusions and Reviews) (England) Regulations (2012) and the Exclusion from Maintained Schools, Academies and Pupil Referral Units in England (DfE 2017) statutory guidance. Under the current law, only suspensions and permanent exclusions are legal and the exclusions process is reserved for use on disciplinary grounds only, i.e. where a pupil either persistently breaches or commits a serious breach of the school behaviour policy, or where a pupil's behaviour would be detrimental to other learners should they remain in the school setting. Headteachers are the only members of school staff with the authority to issue an exclusion and when doing so they must have due regard for the 2017 statutory guidance mentioned above. Illegal exclusions only came to light in 2008 with the publication of the Department for Children, Schools and Families, 'Improving behaviour and attendance: guidance on exclusion from schools and Pupil Referral Units' which stated: If a head teacher/teacher in charge is satisfied that, on the balance of probabilities, a pupil has committed a disciplinary offence and needs to be removed from the school site for that reason, formal exclusion is the only legal method of removal. Informal or unofficial exclusions are illegal regardless of whether they are done with the agreement of parents or carers (DfCSF (Department for Children, Schools and Families) 2008,15) The document makes it clear that sending a pupil home to 'cool off' is also unlawful if not formally recorded with the duration of the exclusion also specified. When a formal school exclusion occurs at a maintained school or academy, whether it be permanent or for a fixed period, the Education Act (2002, s.51A[3b]) ensures that the governing body of the school considers whether a pupil should be reinstated. Where the governing body make the decision to not reinstate a pupil, the Education Act (2002, s.51A[3c]) gives the local authority power to arrange for an independent review panel to review the decision. Furthermore, where a pupil is lawfully excluded, the school is responsible for setting and marking work during the first five days of the exclusion. After this period, if a fixed-term suspension has taken place, then the governing board must ensure suitable educational provision for the remaining duration of the suspension; where a permanent exclusion has taken place, the local authority are responsible for arranging suitable full-time alternative provision no later than the sixth day of the exclusion (DfE (Department for Education) 2017). This right to challenge an exclusion and to continue to access education illustrates why it is so important that headteachers follow correct and lawful procedures when deciding to exclude a child from their school. However, it is important to note here that since 2010, when independent appeal panels (IAP) were replaced by independent review panels (IRP) (Dyke 2011), even in circumstances where an independent review panel recommend that an exclusion should be reconsidered, they no longer have the power to rescind the exclusion, therefore meaning that the governing board of the school have the final say in this process. Thus, while the visibility of illegal exclusion is increasing, the avenues for overturning even lawful school exclusion are becoming increasingly restricted. Definitions Unfortunately, regardless of what the law on school exclusion dictates, illegal exclusions do occur and they are not a new phenomenon. The CCO (Children's Commissioner's Office) 2013) published a report entitled 'Always Someone Else's Problem' that focused attention on the use of illegal exclusions in schools, claiming that such practices may take the form of sending a child home for a cooling off period, or encouraging a parent to home-school their child to pre-empt the school formally excluding them. The Children's Commissioner found that a small but significant number of schools were illegally excluding pupils, and that the pupils most likely to be illegally excluded were the same pupils facing higher rates of formal exclusion; pupils with SEN were especially overrepresented (CCO (Children's Commissioner's Office) 2013). The report also claimed that some headteachers felt that they were acting in the best interests of the child by illegally excluding them since having an exclusion on their school record would render it difficult for a child to be accepted into another school. The Commissioner concluded that parents, pupils and teachers were not sufficiently aware of the law surrounding exclusion, thus allowing for the unquestioned use of illegal exclusions (CCO (Children's Commissioner's Office) 2013). Consequently, it was recommended that the DfE, produce more statutory guidance on exclusions in collaboration with the Equality and Human Rights Commission and the Government Equalities Office. It was also recommended that Ofsted review their approach to identifying the existence of illegal exclusion and give parents an opportunity to share their experiences of this practice (CCO (Children's Commissioner's Office) 2013). The DfE responded to this recommendation by claiming that additional guidance was not needed as only a minority of schools act outside of legal boundaries (DfE (Department for Education) 2013a). However, Ofsted did appear to have taken the issue of illegal exclusion more seriously and in 2019 the Education Inspection Framework (Ofsted 2021a) and the accompanying handbook (Ofsted 2021c) were updated to include sections specifically dedicated to the off-rolling of pupils. Ofsted (2021a) defined this practice as: The practice of removing a learner from the provider's roll without a formal, permanent exclusion or by encouraging a parent to remove their child from the school roll, when the removal is primarily in the interests of the school rather than in the best interests of the learner -off-rolling in these circumstances is a form of "gaming". (Ofsted 2021a, "Leadership and Management" para. 3) The guidance also states that, where evidence of off-rolling according to the above definition is found, the school leadership and management should be judged as 'Inadequate'. However, when considering the evidence uncovered by the Children's Commissioners Office (CCO (Children's Commissioner's Office) 2013), the definition of off-rolling put forward by Ofsted does not appear to be fit for purpose, primarily since some headteachers surveyed believed that they were acting in the best interests of the child. Furthermore, this definition fails to fully align with the definition of offrolling provided by Department for Children, Schools and Families (DfCSF (Department for Children, Schools and Families) 2008) guidance; hence, the decision of Ofsted to specifically refer to permanent exclusion in their definition creates ambiguity around the legality or illegality of schools sending pupils home to cool off for a short period without registering this as a suspension. It is perhaps unsurprising that in a recent update to inspectors, Ofsted (2021c) admitted that, even where practices considered to be off-rolling were found at inspections, the inspection report did not explicitly adopt the term 'off-rolling'. However, the update failed to provide clear guidance on offrolling, instead issuing a somewhat contradictory statement regarding the role of inspectors in deciding what is in the best interests of a pupil and, therefore, whether an unlawful exclusion has occurred: It is not for inspectors to decide what is and is not in pupils' best interests, so they will ask leaders to explain why a move off the roll was in a pupil's best interests. When leaders' explanations are not convincing, and the evidence supports it, inspectors are likely to come to the conclusion that they have evidence of off-rolling (Ofsted 2021b). Ofsted (2021a) definition of off-rolling specifically refers to the act of gaming, where gaming describes the deliberate manipulation of academic performance monitoring systems in order to gain competitive advantage over other schools Knowler 2021, 1039). Knowler (2021, 1039) argue that, by defining off-rolling with reference to gaming, Ofsted discourage 'recognition of exclusionary practices that are not irrefutably related to academic performance as such'. It appears that, by Ofsted's definition, off-rolling is only a concern when it contradicts the principle of free and fair competition that sustains the quasi-market through which education is organised in England and which Ofsted supports. This definition fails to recognise that off-rolling may be driven by other factors such as discrimination and/or the inability to accommodate difference, especially as pupils with special educational needs and pupils experiencing disadvantage are most commonly offrolled (CCO (Children's Commissioner's Office) 2013). Safeguarding Whilst policy and guidance relating to school exclusion is extensive, specific policy and guidance on the topic of illegal school exclusion is insufficient and sometimes contradictory, permitting inappropriate interpretations of the law and under-reporting of a significant safeguarding issue. Illegally excluding pupils from school is, undoubtedly, a significant safeguarding concern, especially where the exclusion occurs through parents being encouraged to educate their child at home. Where a pupil is removed from the school roll in order to be home educated, it is the duty of the school to inform the local authority of this decision under regulation 12(3) of the Education (Pupil Registration) (England) Regulations 2006. Once removed from the school roll, it is the responsibility of the parent to provide a suitable education for their child and at this point the local authority is no longer involved. This process of 'elective' home education also applies to children and young people with SEN as the Special Educational Needs and Disability (SEND) Code of Practice (2015) states that where a parent decides to home educate and has notified the school of this decision, their child's name must be removed from the school roll. The Children and Families Act (2014) (s.42[2])) states that the local authority must only arrange for SEN provision for a home educated child if they have an Education, Health and Care (EHC) plan. Furthermore, section 10.34 of the SEND Code of Practice states: Local authorities do not have the right of entry to the family home to check that the provision being made by the parents is appropriate and may only enter the home at the invitation of the parents (DfE 2015, 215) While such legislation is appropriate for parents and carers making fully informed decisions to home educate their child, there may be parents who have been pressurised or coerced into home education (Parliament. House of Commons 2021) who do not fully understand what it entails and/ or lack the time and resources to home educate therefore meaning that their children may not in receipt of a 'suitable education' as defined by the Children Act 1989 (s.36). Concerningly, a survey conducted by the Association of Directors of Children's Services (2017) revealed that 65% of the local authorities surveyed had only one member of full-time staff, or less, to monitor home schooling within their locality, making it impossible to offer parents the support needed to fulfil a home educator role; according to s.31 of the Children Act 1989, failing to provide a child with a suitable education satisfies the threshold for 'significant harm' to the child. Parents in such situations may resort to provision provided by illegal schools. An illegal school is one that is unregistered and operating independently (Ofsted 2019b). Ofsted (2021d) claims that children being educated in such settings are exposed to safeguarding and health and safety risks, and in the period 1 January 2016 to 31 March 2021 Ofsted have received 857 referrals concerning unregistered and illegal schools. Of this number, 386 were inspected and 105 were found to be attended by home educated pupils (Ofsted 2021d). The Education Act (2002) (s.175) stipulates that local authorities have a duty to ensure and maintain the welfare and safeguarding of children in their jurisdiction and, where a child finds themselves in unsuitable education following off-rolling or exclusion, this is also a failure of the school. The statutory guidance on keeping children safe in education (DfE 2021) makes it clear that all individuals working in a school or college have a duty to safeguard and ensure the welfare of children and young people in their setting. The practice of illegal exclusion demonstrates how punitive approaches towards behaviour management, in an environment fuelled by high-stakes accountability and a lack of financial resources to accommodate difference, are prioritised over statutory duties to ensure children's safety and wellbeing, and fulfilment of potential. Implications for research Research on the topic of illegal exclusion is minimal in comparison to research regarding lawful permanent exclusions and suspensions. Research by Gill (2017) suggests that official exclusion statistics, collected and distributed by the DfE, significantly underestimate the scale of school exclusion. The basis of this claim was data gathered in the period spanning 2013/14-2016/17 indicating that 'the number of pupils educated in schools for excluded pupils is five times higher than the number of officially permanently excluded pupils' (Gill 2017, 13). It was noted that these figures may not accurately represent the scale of the problem as illegal exclusion can take various forms. Research by Nye and Thomson (2019) for the Education Data Lab arguably presents more accurate figures on the scale of illegal exclusion; the latest publication in a 'Who's Left' series claims that between 6,700 and 9,200 pupils from the 2018 cohort went missing from the education system, i.e. these children and young people 'remained in England and yet did not take any qualifications or, if they did, did not count in results anywhere'. This figure is higher than the 6,200-7,700 pupils that were estimated to be missing from the 2017 cohort (Nye and Thomson 2018). However, again, these findings should be approached with a degree of caution since they are estimates and 'without knowing individual children's circumstances, leaving state education can't be treated as synonymous with off-rolling' (Nye and Thomson 2018). What these estimates do imply is that the practice of illegally excluding children and young people from the state school system in England is far more prevalent than the Timpson review of school exclusion (DfE (Department for Education) 2019) would suggest. A survey conducted by YouGov on behalf of Ofsted (2019b) did provide valuable insights on the nature and scale of off-rolling; 1,018 teaching professionals were surveyed and 14 educators interviewed across England, including one headteacher. However, due to the ordering of the questions, the results of this survey may be biased in favour of Ofsted's perceptions of the nature and causes of off-rolling. For example, at the beginning of the survey participants are asked to select, from a multiple choice list, what they perceived to be as the correct definition of off-rolling -this list included Ofsted's definition: 'A pupil being taken off the school roll in order to try and manipulate reported exam results/league tables' (Ofsted 2019a, 7), 68% of those surveyed identified this as the correct definition. Therefore, it can be argued that it is unsurprising to see that when participants were later asked to identify key drivers of off-rolling, 51% of the sample identified 'achieve/maintain a high position on a league table' (Ofsted 2019a, 12). Furthermore, of the 288 teachers who reported that they had witnessed off-rolling, 57% reported persistent behavioural issues as the main reason for this practice (Ofsted 2019a, 13). Whilst this research does provide some insight into which pupils are more vulnerable to off-rolling and the extent of teachers' awareness of the practice, the small sample size and Ofsted's apparent commitment to a narrative of gaming as the primary reason for off-rolling do raise questions around the validity of these findings. Knowler (2021, 1051) argue that the 'hegemonic status' of Ofsted's definition of off-rolling is likely to militate against other types of off-rolling and reasons for off-rolling being recognised. Moreover, when researching illegal exclusion, sampling and recruitment of participants often appears to be an issue. McShane (2020) describes difficulty in recruiting participants and details the experiences of only three participants. McShane identifies that concerns around anonymity were marked amongst participants, with one stating 'that she would love to have the courage to "whistle blow", but that such an action would be a "career-wrecker"' (McShane 2020, 67). Similar issues were found by Done and Knowler (2021) when attempting to recruit senior leaders to a survey on offrolling; despite repeated efforts to reassure potential participants of their ability to contribute anonymously, only 21 senior leaders took part in the survey. Nevertheless, such research provides insightful contributions to discussions surrounding illegal exclusion; for example, by illustrating the sense of professional risk evoked in acknowledging illegal exclusionary practices and, possibly, the 'shame at having to work in this way' (Done and Knowler 2019, para.6). The inclination of both the Children's Commissioner's Office and Ofsted to name and shame offending schools (Allen-Kinross 2019; George 2019a) may explain the reluctance of education professionals to engage in dialogue around off-rolling. While it is increasingly agreed that illegal exclusion and, indeed, legal exclusion is an archaic practice that should have no place in a modern English school system (CCO (Children's Commissioner's Office) 2013; No More Exclusions 2021; Psychologists for Social Change n.d.) simply punishing schools that engage in such practices by publicly shaming them and issuing a judgement of 'Requires Improvement' in their Ofsted report only serves to marginalise school leaders and encourage new ways to remove 'difficult' pupils from the school roll. A shift in priorities, away from punishment and towards upholding safeguarding legislation, is required. Conclusion This article has explored the issue of illegal exclusion from the perspective of policy and legislation, addressing the contributions of the Timpson review of school exclusion (Timpson 2019) before outlining how various policy documents have contributed to the widely accepted definition of offrolling provided by Ofsted and now accepted by many educational professionals (Ofsted 2019a;Done and Knowler 2021). Discussion of safeguarding legislation and the complexities involved in researching illegal exclusion was intended to provide context to the analysis of relevant policy. Two main conclusions can be drawn from this analysis. Firstly, the narrow definition of illegal exclusion or off-rolling that is dominant in legal guidance means that illegal exclusion will continue to be obscured; Ofsted's insistence on gaming as a key element of its definition raises concerns around the power of the principles that govern the quasi-market in education and consequent inability of mainstream English schools to provide a truly inclusive learning environment for all of their pupils. In terms of research and, in particular, the recruiting of participants, the dominance of Ofsted's definition is not conducive to either understanding the scale of the problem or to encouraging open dialogue about the reasons why headteachers and senior leaders decide to remove a pupil from the school roll illegally. Secondly, a culture of sanctioning and punishing schools rather than one of understanding, investment and reform, further discourages educators from sharing their lived experiences of illegal exclusion and does nothing to ensure that safeguarding legislation and exclusion guidance are upheld in the future. The two main conclusions outlined here serve to further reinforce the need for a shift in priorities within education policy. More needs to be done to meaningfully recognise, address, and understand the impact of the disproportionality in the use of off-rolling for pupils with SEN, pupils eligible for FSM and pupils from Black ethnic backgrounds. In order to address this problem, more needs to be known about the true extent of the issue however, in a hostile policy environment that seeks to 'name and shame' (Allen-Kinross 2019; George 2019a), or even wilfully ignore the problem of off-rolling, the potential for research is significantly curtailed. As such, it is recommended that policymakers should concurrently engage with research that illustrates the perils of running an education system on principles of a quasi-market (Davies, Diamond, and Perry 2021;Done and Knowler 2019;Ferguson 2021;Thompson, Tawell, and Daniels 2021), while also making meaningful attempts to actively pursue the safeguarding and equity of education for all children and young people in England. It is also recommended that policymakers develop an agreed, consistent definition on what an illegal exclusion is and the circumstances in which it may occur, doing so may serve to both raise awareness and reduce the use of illegal exclusion practices.	v3-fos-license
2020-12-11T05:04:09.834Z	2020-12-01T00:00:00.000	228084151	{ "extfieldsofstudy": [ "Medicine", "Biology" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://virologyj.biomedcentral.com/track/pdf/10.1186/s12985-020-01461-4", "pdf_hash": "7a7cd57b297ea57ca0d296178ab92427a988772a", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45429", "s2fieldsofstudy": [ "Medicine" ], "sha1": "0cda0d6fa07e1abb58b35677627cde502708c4c1", "year": 2020 }	pes2o/s2orc	Comparative analysis of clinical features of SARS-CoV-2 and adenovirus infection among children Background The new emerging coronavirus disease 2019 (COVID-19) overall shares similar symptoms with other common respiratory viral infections. We aimed in this study to compare COVID-19 and human adenovirus (HAdV) infections in pediatric patients regarding the frequencies of major clinical symptoms and the potential disparities in laboratory and imaging parameters. Methods Following a case–control-like design, we built 72 age-matched pediatric COVID-19 and HAdV patient pairs. Their early symptoms and laboratory and imaging characteristics were then retrieved and compared. Results Fever and cough were the most common symptoms for both infections but were seen more often in HAdV than in COVID-19 patients (92% vs. 66% and 60% vs. 18%, respectively). Compared with COVID-19 patients, children with HAdV infection had statistically significantly higher values of neutrophil count, neutrophil percentage, activated partial thromboplastin time, prothrombin time, lactate dehydrogenase, C-reactive protein, procalcitonin but lower values of lymphocyte percentage, total bilirubin, potassium and sodium. Thoracic computed tomography also revealed more anomalies in HAdV patients than in COVID-19 patients (95% vs. 67%). Conclusions COVID-19 is an overall less symptomatic and less severe infection at admission compared to HAdV respiratory infection in pediatric population. Background In late December 2019, a novel respiratory infectious disease caused by a new strain of coronavirus, namely severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), made its first appearance in Wuhan, China and spread rapidly around the world [1]. The disease was later on named coronavirus disease 2019 (COVID- 19). As of September 27, 2020, a total of 32,730,945 confirmed COVID-19 cases, including 991,224 deaths, have been reported globally [2]. Children appear to be less affected by COVID-19 and only account for 1-5% of diagnosed COVID-19 cases [3]. Fever and cough are the most commonly seen symptoms in children [4,5]. Compared with adult COVID-19 patients, pediatric patients manifest relatively mild symptoms and rarely develop severe pneumonia [4,5]. Human adenoviruses (HAdV) are a family of viruses causing 4-10% of respiratory illnesses in children and infants worldwide [6,7]. Although COVID-19 and HAdV respiratory infection show similar symptoms such as fever and cough [8,9], detailed comparisons of the two infections with respect to their early clinical characteristics among pediatric patients have not yet been reported. The aim of this study was to compare the two infections regarding their early clinical, laboratory, and radiological characteristics among pediatric patients. Study design and participants From January 23 to February 23, 2020, a total of 106 children (age range: 0-11 years) were diagnosed with COVID-19 and hospitalized in Guangzhou Women and Children's Medical Center, Third People's Hospital of Shenzhen, or Wuhan Children's Hospital. The diagnosis of COVID-19 followed the Protocol for Novel Coronavirus Pneumonia Diagnosis and Treatment issued by the National Health Commission of the People's Republic of China [10]. All the patients were laboratory confirmed using real-time reverse transcriptase polymerase chain reaction (RT-PCR) assays on nasopharyngeal swab specimens. Laboratory confirmation was also performed on other common respiratory pathogens including influenza-A virus (H1N1, H3N2, H7N9), influenza B virus, respiratory syncytial virus (RSV), parainfluenza virus, adenovirus, SARS coronavirus (SARS-CoV), and MERS coronavirus (MERS-CoV), and any co-infections were precluded from this study. In the electronic medical database of Guangzhou Women and Children's Medical Centre, we identified 570 pediatric patients who were hospitalized with laboratoryconfirmed adenovirus respiratory infections between November 1, 2016 and March 31, 2020. Please note that patients with laboratory-confirmed co-infections of other respiratory pathogens were also excluded. Following a case-control-like design, we selected an age-matched patient with respiratory HAdV infection for each of the COVID-19 cases, where an eligible match was defined as the absolute difference in age at admission for a patient pair was no greater than 30 days. If a COVID-19 patient had multiple age-matched HAdV patients, priority would be given to the one of the same sex. Data collection procedures Data on demographics and clinical features were collected from electronic medical records. Chest computed tomography (CT) and laboratory findings were collected from the Picture Archiving and Communication Systems and the Laboratory Information System, respectively. Radiologic abnormalities were determined according to descriptions in the clinical charts. Statistical analysis Quantitative measurements were presented as medians and interquartile ranges (IQRs); qualitative measurements were presented as counts and percentages. Abnormally high or low levels of laboratory findings were defined using age-specific or otherwise universal reference ranges (Additional file 1). Because of the pairing design, differences between groups were analyzed by Wilcoxon signed-ranks test (for continuous data) or McNemar's chi-square test (for binary data), with p value less than 0.05 considered statistically significant. All analyses were conducted with R (version 3.4.6; The R Foundation for Statistical Computing, Vienna, Austria). This study was approved by the ethics committees of participating hospitals. Written informed consent was obtained from parent or guardians of the pediatric COVID-19 patients. Results Among all the pairs, 72 met the matching criterion and therefore 72 age-matched patient pairs of COVID-19 patients and patients with HAdV respiratory infection were formed for analysis. The median age was 2 years (range: 0-11 years) for both groups, and boys accounted for 63% (45) for HAdV and 61% (44) for COVID-19. Twenty-two COVID-19 patients were laboratory confirmed after epidemiological exposure but were clinically asymptomatic at admission. These patients and their HAdV pairs were excluded when we compared the frequencies of clinical symptoms between the two infections. In the remaining 50 symptomatic COVID-19 and HAdV patient pairs, the average time between symptom onset and admission was 3 days for COVID-19 and 7 days for HAdV. The frequencies of the signs and symptoms at admission were reported in Table 1 and Fig. 1. Symptoms at illness onset were similar among the two groups, with fever and cough being the most common symptoms, but the percentages of cases with fever [46 (92%) vs. 33 (66%), p = 0.006] and wet cough [30 (60%) vs. 9 (18%), p < 0.001] in HAdV were higher than those in COVID-19. More headaches, nasal congestion, shortness of breath, diarrhea, nausea or vomiting, fatigue, and chills were reported in the HAdV group than in the COVID-19 group, but none of these symptoms showed statistically significant difference between the two groups. Laboratory tests were performed on the day of admission for the HAdV patients and within the first 2 days after admission for the COVID-19 patients. Table 2 In addition, patients with HAdV infection had averagely lower levels of lymphocyte percentage, total bilirubin, potassium, and sodium than those with SARS-CoV-2 infection. Other laboratory findings did not significantly differ between the two groups (Additional file 2). As shown in Table 3, the odds of having abnormally low neutrophil count were lower in the HAdV patients than in the COVID-19 patients, but the odds of having higherthan-normal levels of hemoglobin, albumin, APTT, CRP, and procalcitonin were higher in HAdV patients. Data for other laboratory variables showing no statistically significant difference are provided in Additional file 3. Discussion In this study, we presented the early clinical symptoms and signs, radiologic and laboratory findings of of pediatric patients who had HAdV respiratory infection and Table 2 Haematological and blood biochemical measurements of the 72 age-matched pairs of pediatric COVID-19 patients and patients with adenovirus respiratory infection For each measurement, the exact number of patient pairs included in the analysis varied due to missing values. Distribution of the measurements is denoted by median and interquartile range (in parentheses). The p values were calculated using Wilcoxon signed-rank test COVID-19, respectively. Fever, cough, nasal congestion, shortness of breath, diarrhea, nausea or vomiting, and chills were the most often reported early symptoms in both infections but occurred more often in patients with HAdV infection. HAdV patients had higher levels of neutrophil count, neutrophil percentage, LDH, CRP, procalcitonin, and APTT, but lower levels of lymphocyte percentage and total bilirubin. CT abnormality, in particular bilateral patchy shadowing, occurred more often in HAdV patients than in COVID-19 patients. Children with COVID-19 often appear to have a mild clinical course [11][12][13]. In the present study, only 1 case was treated in ICU, and no one died during hospitalization. In a previous report, asymptomatic patients were common at admission as patients with a history of household exposure or travel to epidemic areas were required to undergo free testing, isolation, or hospitalization [14], while children with respiratory HAdV infection were often admitted because of certain symptoms. Thus, we only retained symptomatic COVID-19 and HAdV patient pairs for the symptomological comparison. Our findings were consistent with prior research [11,13,15,16] where fever and cough were found to be the most common clinical manifestations. We found that children with HAdV infection were more likely to exhibit clinical symptoms than COVID-19 cases. Previous research assumed that low frequencies of clinical symptoms in pediatric COVID-19 patients might relate to children's less mature and functional binding receptors of the target cells [11]. It may also be because the COVID-19 cases in the present study had an averagely shorter time between symptom onset and hospital admission than patients with HAdV infection. Previous studies showed that elevated neutrophils and decreased lymphocytes signify adverse clinical progress and an increased risk of poor prognosis in both SARS-CoV-2 and HAdV infections [17][18][19]. In this study, children with HAdV infection showed a higher neutrophil count and a lower level of lymphocyte percentage than COVID-19 patients, indicating that two infections might be different in their severity. Previous literature showed that longer APTT and PT were related to poor outcome in COVID-19 [20]. In this study, markedly longer APTT and PT were seen in children with HAdV infection, implying that pediatric HAdV patients might have more severe coagulation disorder than COVID-19 patients. With respect to immunologic biomarkers, significantly higher levels of C-reactive protein and procalcitonin were observed for children with HAdV infection. The elevation of these two markers points to the development of a systemic inflammatory response syndrome (SIRS) in patients with a severe form of respiratory disease [17,21,22]. In the present study, other hematological tests than what we reported above showed statistically non-significant differences in their levels between the COVID-19 patients and patients with HAdV infection. However, some of these tests might also be indicators of COVID-19 severity. A recent meta-analysis reports severe COVID-19 associated with lower lymphocyte and higher leukocyte counts, although only adult patients were considered in the analysis [10]. Increased serum creatinine was also reported in pediatric patients as a marker of acute kidney injury [23]. However, none of these hematological tests in the present study showed statistically significant differences in their levels between the COVID-19 patients and patients with HAdV infection. CT investigations in pediatric COVID-19 patients are limited but consistently show that bilateral involvement occur more often than unilateral involvement and ground-glass opacity is the most common CT abnormalities [24]. In this study, the rate of bilateral patchy shadowing was also higher than the rate of unilateral patchy shadowing but only 19% of our COVID-19 cases showed ground-glass opacity, possibly due to the fact that CT investigation was performed at the very early stage of the disease. This study, to the best of our knowledge, is the first study that compares the SARS-CoV-2 and HAdV infection in the pediatric population. Considering its small sample size and inclusion mainly of patients with mild clinical manifestations, future studies should focus on more representative pediatric COVID-19 patients and identification among them of clinical features that can help distinguish this infection from other common respiratory viral infections in their early stage. Conclusions COVID-19 is an overall less symptomatic and less severe infection at admission compared to HAdV respiratory infection in pediatric population.	v3-fos-license
2019-12-12T10:49:25.665Z	2018-10-05T00:00:00.000	211548704	{ "extfieldsofstudy": [ "Political Science" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://sd-magazine.eu/index.php/sd/article/download/9/9", "pdf_hash": "951fbf4f015418a423b6ee1478132a213f5932d9", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45430", "s2fieldsofstudy": [ "Political Science" ], "sha1": "265d52d2367e5a101bb25a4690b1e4cc0507dd6c", "year": 2018 }	pes2o/s2orc	Armed Forces Operation in the Scope of the Civilian Health Protection during Peacekeeping and Stabilization Missions: A Short Review Nowadays, the diversity of armed conflicts determines the participants of international relations to undertake various actions in the scope of civilian health protection. It should be noted that tasks resulting from civilian protection are fulfilled in numerous manners, depending on the situation of the armed conflict. The article presents actions undertaken by the armed forces in the scope of the civilian health protection during peacekeeping and stabilization missions. There are also presented engagement of Polish armed forces in Afghanistan and their actions to improve the civilian population. Introduction Nowadays, the diversity of armed conflicts determines the participants of international relations to undertake various actions in the scope of civilian health protection. Regardless of the circumstances, the respect for human rights is the duty of all parties involved. The right to respect the dignity of each human being is inalienable, and the compliance with this law is the basis of all military actions. The Latin maxim Inter arma silent leges should not be applied in XXI century. Providing safety for civilians is not only a duty but also is in the best interest of all involved. Under the term protection of civilians during armed conflicts, one should understand protecting human life, but also the protection of their property and the environment, so it does not suffer due to the outside interference. It should be noted that tasks resulting from civilian protection are fulfilled in numerous manners, depending on the situation of the armed conflict. On the one hand, civilian protection is the element of country's defense system and constitutes a set of planning, organizational, training, investment, technical, and supplying undertakings. Those undertakings are realized through central and local government entities, and through organizational units. Civilian protection aims at protecting civilians, workplaces, public utility facilities, cultural goods, saving, and providing aid to those affected by the war, as well as collaboration in combating natural disasters and environmental hazards and removing their effects. [1] On the other hand, in the case of participation in an armed conflict which does not result from the protection of own territory, civilian protection is the part of international human rights norms, which also limit using the methods and the means to conduct hostilities. Poland's Participation in Peacekeeping and Stabilisation Missions The history of peacekeeping missions began in 29 May 1948, when the United Nations Security Council (UN) decided (UN Security Council Resolution 50) to station military observers in order to supervise compliance with conditions of the first truce (11 June -8 July 1948) and not to allow the strengthening of the military positions in the region of the fights of the I Israeli-Arab War in 1948-1949. [2] Two weeks after the decision was made by the UN Security Council, group of 36 unarmed Belgian, French, Swedish and American officers came to the Middle East. That UN Security Council decision laid grounds for peacekeeping missions and contributed to the creation of United Nations Truce Supervision Organization (UNTSO) whose primary role is to observe and supervise the ceasefire. To commemorate that first mission, the UN Secretary-General, Kofi Annan announced the 29 of May to be the Day of International Peacemaking Missions' Participants. [3][4][5] Poland has continuously taken part in peacemaking and stabilization missions since 1953. To this day, over 90 thousand soldiers and civil-military workers have taken part in more than 70 operations outside of Poland. At first, they were traditional peacekeeping operations, mainly consisting of monitoring the separation of conflicted parties after the armed operations were ceased. [4] At that time, Polish engagement did not constitute a permanent feature of the national security policy, while the principles of the Polish Armed Forces engagement in missions were developed on ad hoc basis, adapting them to the needs of a given operation. Between 1953-1973, Polish military and civil personnel took part in the works of Neutral Nations Supervisory Commission (NNSC) in the Korean Conflict (Poland still refers their two representatives to NNSC), International Control and Supervisory Agency in Indo-China, and International Observers Group in Nigeria. Between 1953Between -1988 Poland took part in seven military initiatives, including four Disputes Committees and three peacekeeping operations. There were over 17 thousand people involved. [4] The change of the political system in 1989 had a great impact on the Polish engagement in missions outside its territory. International missions were started to be deemed as an essential element of state's defense system. Between 1989-2009, the number of the personnel involved in the missions quadrupled. Over 67 thousand soldiers and civil workers served in 64 operations, including 30 UN peacekeeping operations, 13 Allied Force Operations, nine Monitoring Missions of the Organisation for Security and Cooperation in Europe (OSCE), six European Union missions, and six operations of international communities. New political and military conditions, especially joining NATO and the European Union, increased terrorism risk, as well as military benefits, resulted in the gradual change of Polish attitude towards the missions. Our involvement in UN operations decreased, while the involvement in NATO, and international coalitions' missions aimed at enforcing peace in regions where it was most at danger, increased. [4] The sign of undertaking the new challenges in the sphere of security was assigning a contingent to an anti-terrorist operation "Enduring Freedom" conducted by an international coalition in Afghanistan in 2002. In 2004, the scale of actions carried out in this country grew rapidly. Today, the engagement in the said operation is the priority of the Polish Army. One of the biggest challenges of Polish Armed Forces was 2003 operation "Iraqi Freedom," during which allied states carried out military operations in Iraq. In September 2003, Poland decided to join multi-task stabilization mission in this country. In 2008, Poland was assigned the command of Central-South Zone within International Stabilization Force in Iraq, i.e., to withdraw the Polish contingent from the region. Until this day, it remains one of the most demanding tasks in the whole history of Polish participation in international missions. Currently, the involvement of Polish Armed Forces in international military operations is, next to national defense, the key element of the national security strategy. Modern operations are complex activities, which take into consideration the meaning and the impact of various political, military, economic and social factors. Current international operations are often the reaction to inner-state conflicts and cases of mass violations of human rights. The amount of interventions against repressive countries, to protect democracy and human rights, is also on the rise. Today's operations are the part of the broader spectrum of "pro-development" activities, within which military and civil instruments are parallelly used on each stage of restoration and maintaining of peace and safety, as well as during the preventive and post-conflict activities. [4] Currently, Poland is taking part in numerous international operations (including those of Polish UN observers) carried out along with UN, NATO, and EU. Polish operation in Afghanistan with NATO is one of the most significant and biggest missions of Polish Armed Forces. Poland desired to give its engagement in Afghanistan a more complex character. The essential aim of the Polish activities is to supervise the safety and support the stabilization and reconstruction of Afghanistan. Therefore, along with strictly military tasks, Poland seeks to assist the development. [6,7] Activities in the Frame of Protection of Civilian Health Undertaken by Task Force White Eagle in Afghanistan It needs to be stated, that according to the Concept of Medical Support of the Task Force White Eagle Afghanistan, health service of the Polish Task Force does not provide regular humanitarian aid, rather it offers assistance and support in such actions to governmental and non-governmental organizations. Humanitarian aid is limited to saving lives, limbs, and sight (only for a short period of time). Providing medical support for the local civilians is limited only to the situations, in which it is possible to provide it, and the provision thereof does not endanger the mission of Polish Task Force. [4] Furthermore, the conflict in Afghanistan is not only a war against the Taliban but a combination of several minor conflicts that involve -in addition to an international coalition gathered under the banner of the ISAF -all sorts of entities (both state and private), international terrorist organizations, and criminals. This would also include different tribes, mercenaries, religious and ideological leaders, and intelligence services that have broken away from the control of the state. Therefore, ending the conflict in Afghanistan is an extremely difficult task, and probably one that will prove impossible to achieve for a long time. Additionally, in the case of a "new war," it is very important that international humanitarian law is not toothless and is fully applied. [8] During each shift of Polish Military Contingent in Afghanistan, Polish medical personnel not only provided aid in the events of imminent threat to human lives but also assisted those in need, who reported their problems directly to one of the Polish Military Contingent bases. Most frequently those cases related to trauma consequences, e.g., burns, sprains, broken bones, which demanded immediate treating. However, the work of the paramedics, nurses, and doctors in the Polish bases did not only include the help in case of the injuries. [9] Many people who reached Polish bases were poor, exhausted and dehydrated. "Sometimes ambulatory care or consultation is sufficient after an agreement by phone with a doctor of Bravo combat training camp located in the Warrior base." In the cases of imminent threat to human life, MEDEVAC (Medical Evacuation) helicopters are summoned, and the injured are evacuated to the field hospital in Ghazni. The MEDEVAC helicopter environment is one of the most difficult, if not the most demanding, critical care environments. Medical evacuation is the movement and en route care of injured and medically compromised patients by medical care providers via helicopter. Military MEDEVAC platforms provide lifesaving interventions that improve survival in combat. [10,11] The most common diseases among the local society are lower and upper respiratory tract infections, anemia, malnutrition, bacterial infections, virus and parasite digestive tracts infections, dermatophytosis and scabies. Low awareness of maintaining appropriate levels of hygiene is also an enormous problem. Most diseases require radiological diagnosis and laboratory tests, which are virtually unattainable for the poorest. There are entities within the Polish Military Contingent which systematically work, adequately to the needs and possibilities, to monitor and help the poorest of Afghans. These include the Civil-Military Cooperation (CIMIC), and the Provincial Reconstruction Team (PRT). The help offered by the Polish reaches schools, universities and medical centers. They also assist those under the care of the Commission for Refugees and the orphanage in Ghazni. The support is also given to the poorest residents, including people of the Kuchi tribe living in the province. [12] Polish soldiers from the CIMIC group, along with the medical personnel and the leaders of the combat training camp, regularly provide medical help to the local community. Most frequently, it takes place during so-called "white Sundays" (Medical Civil Action Program -MEDCAP), during which they try to reach the poorest residents of Afghanistan, e.g., those living in Gelan and Moqur district. During every "white Sunday" Polish medical personnel grants aid and distributes necessary medications, as well as providing extra medical consultations and training local medical personnel. Each adult patient receives a first aid kit, while the younger patients are given clothes, shoes, toys, and sweets. From the reports of the medical personnel, it appears that a major issue they encounter during "white Sundays" is convincing women to undergo the examination. "If they are finally convinced, the examination has to be done by the female personnel, and the local women who are not used to medical examination, do not take off their burkas, under which there are numerous layers of shawls, therefore auscultation of the heart borders on miracle". [13] The aid granted by the specialists of Provincial Reconstruction Team focuses mainly on reconstructing damaged infrastructure and on conducting training. Primarily, it consisted of investments in infrastructure development. The main focus was the development of communal services in the city. Mechanical sewage treatment plant was created because the lack of the sewer economy in Ghazni has a direct impact on the health of the residents. [14] Polish team of PRT Ghazni specialists from Polish Military Contingent in Afghanistan realized over 100 projects in less than three years. All of the actions of PRT and CIMIC aim at improving the living standards of Afghanistan's residents, one of the poorest regions in Asia. The actions for granting medical aid are the immediate reaction to numerous requests of the local community and district's authorities (very common recently) where the Polish soldiers are stationed. [11] Conclusion Modern international humanitarian law provides protection for the civilians. It needs to be remembered that even if the engagement of the armed force is morally and legally justified, there are specific means which cannot be applied. The undeniable need for the fight against the plague of the 21st century, the terrorism, does not justify the usage of certain forms of violence, especially against the civilians. [15] Attacking civilians constitutes a violation of the Geneva Convention and is deemed as a war crime and a crime against humanity. The parties taking part in the military conflict are obliged to grant aid to the civilians. All countries which ratified the Geneva Convention and the additional protocols have the duty to respect their rules in the time of war and to spread them during the time of peace. Nowadays, the Geneva Conventions apply in all cases where hostilities are ongoing, regardless of whether the war is declared or not. Moreover, the classification of the armed conflict by its participants does not matter. In addition to war, the Geneva Conventions are concerned with the occupation, even if there is no armed resistance. Furthermore, the Conventions apply to all states, including situations where one of the countries involved in the conflict may not be a party to the Convention. [16] In accordance with Articles 47, 48, 127, and 144 of Geneva Conventions I, II, III, and IV, respectively, it is a legal obligation of countries to spread knowledge of these Conventions and Protocols. [17] Active participation of Polish Armed Forces in Afghanistan has a greater meaning, outside of the military one. Polish program of developmental collaboration for Afghanistan, realized through the Foreign Affairs Ministry, aims not only at the support for the better government but also to expand the public infrastructure, including schools, water, and electricity supply. New workplaces are created, and the support is given to the refugees. There are also projects conducted in the health service sector. International humanitarian law provides special protection devices and medical personnel during armed conflicts. In today's wars, the lack of respect for the protective emblems of the red cross and red crescent and the lack of respect for medical activities become more frequent. [18] Military personnel, as well as the police and civil Polish Military Contingent in Afghanistan, can take pride in excellent results when it comes to improving the respect for human rights, reacting to the vulnerable situation of the civilians, and triggering proper support for them.	v3-fos-license
2019-03-09T18:35:05.377Z	1981-01-01T00:00:00.000	110452149	{ "extfieldsofstudy": [ "Engineering" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://downloads.hindawi.com/journals/apec/1981/140578.pdf", "pdf_hash": "eba75e16ec4bfecae63954cf36a21a2f3e24391a", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45431", "s2fieldsofstudy": [ "Materials Science" ], "sha1": "eba75e16ec4bfecae63954cf36a21a2f3e24391a", "year": 1981 }	pes2o/s2orc	COPPER MATERIALS SYSTEM FOR MICROCIRCUITRY : A STATUS REPORT The electronics industry has become interested in a thick film materials system based on copper for both technical and economic reasons. Copper conductors offer excellent conductivity, solderability, and solder leach resistance as well as lower intrinsic metal cost and price stability. INTRODUCTION Industry interest in a thick film materials system based on copper has increased in the latter part of the 1970's for both technical and economic reasons.As a thick .filmconductor, copper offers excellent conductivity, solderability and solder leach resistance relative to silver-bearing conductors.In addition a copper conduc- tor offers price stability and, with adoption in higher volumes, will offer lower material costs to microcircuit manufacturers than silver-bearing conductors. Significant adoption of copper in thick film applica- tions requires a complete system of materials including resistors and higher K dielectrics for printed capacitors.Since compositions of the copper materials system are fired in nitrogen, sufficient processing information to help the industry get started is also needed.Industry has begun the adoption of copper in selected applica- tions.Design of new applications and experience in processing will be required to move this technology forward in the industry. Du Pont has a commercial copper conductor and compatible low K dielectric compositions and an initial resistor system available.In addition, significant pro- cessing studies in the nitrogen firing area have been completed.This paper is a status report with particular emphasis on potential applications for copper, the resis- tor properties and recently developed processing information. PRODUCTS AVAILABLE Four different copper compositions have been developed for fired thick film applications on alumina and beryllia substrates. 9922is a high adhesion, low resistance conductor used for resistor terminations and microwave applications.9923 is compatible with nitrogen-fireable dielectric and was developed for multilayer circuit applications. 9924is similar to 9923 but produces thinner fired films which facilitates print- ing high resolution dielectric patterns over the copper.4234 was designed for printing through holes to con- nect one side of a substrate with the other.Dielectric composition 4175 was developed for use with copper to make multilayer structures.It is fireable in nitrogen and has a dielectric constant of about eight. A nitrogen-fireable resistor series was designed to be compatible with 9922 copper and is processable under the same conditions.The series is based on new non- noble metal technology and presently consists of com- positions with resistivities of 10, 100, 1000, and 10,000 ohms per square.Corresponding product numbers are 4239, 4238, 4237 and 4299.The nitrogen-fireable resistor compositions are similar in properties to the 1600 series BIROX(R) compositions. Products under development are a lower-loss micro- wave copper composition for alumina and a 100K ohm per square resistor composition to extend the range of the nitrogen-fireable series.Additional pro- ducts being developed are copper, dielectric and resis- tor compositions suitable for firing at 625C on porcelain-enamelled steel. PROCESSING COPPER COMPOSITIONS The copper compositions are printed with 325-mesh stainless steel screen to produce fired films which are about 15 micrometers thick.The prints are dried about 10 minutes at 120C in an air atmosphere using either convection or infrared driers.To avoid oxidizing the copper the drying temperature should not exceed 150C.Infra-red ovens must be adjusted carefully; when subjected to direct radiation, the surface of the copper may reach higher temperatures than indicated by the oven controller, resulting in oxidation of the cop- per.Dried copper films which are oxidized will produce grey or purple fired films which are difficult to solder. Copper is fired in a nitrogen atmosphere in a thick film conveyor furnace similar to those shown in Figure 1.Commercially available furnaces may be used.The typical temperature profile shown in Figure 2 55 minutes long and holds at a peak temperature of 900C for 6-10 minutes.Time and temperature may be varied.Copper has been successfully fired in profiles as short as 20 minutes and at peak temperatures as high as 950C and as low as 850C.During firing of dried copper films, the oxygen concentration in the hot sec- tions of the furnace should be maintained below 15 to 20 ppm to avoid oxidizing the copper.Oxygen concen- tration of 6-8 ppm in the burnout section has been found satisfactory for firing copper.When properly processed, fired thick film copper is moderately reflec- tive and copper-pink in color. PROCESSING DIELECTRIC COMPOSITIONS The dielectric is printed with 200-mesh stainless steel screen to produce fired films which are 25 micrometers thick.To minimize pinholes in the dielectric the squeegee speed should not exceed about 7 centimeters per second (2.8 inches per second).In addition, the squeegee stroke should be adjusted so that the screen has lifted from the pattern area before the squeegee lifts for the return stroke.The dielectric film is dried about 10 minutes at 120C in an air atmosphere.Dielectric films are fired in the same type furnace as copper using the same 900C firing profile shown in Figure 2. Although addition of oxygen to the burnout section is not necessary, it is helpful under some conditions in removing organic vehicle components from the dielec- tric film to prevent blister formation in multilayer struc- tures.Depending on the temperature in the burnout section, up to 600 ppm oxygen may be added without significant oxidation of fired copper accompanying the dielectric. Two layers of dielectric, each individually fired, are used between copper layers in the multilayer structures. PROCESSING RESISTOR COMPOSITIONS 6. NITROGEN FURNACE DESIGN Resistor compositions are printed with 325-mesh stain- less steel screen over pre-fired 9922 copper termina- tions and are dried 10 minutes at 120C in an air at- mosphere.The dried films are about 25 micrometers thick.The resistor films are fired in the same type furnace (Figure 1) with the same firing profile as used for .copper (Figure 2).The firing sensitivity over the range of peak firing temperatures from 875C to 930C is shown in Table I.Resistance change for the 10K composition is 2.5% per C. The change for other members of the series is between 1 and 2% per C. The variation of resistivity and hot TCR with oxygen concentration in the burnout section is shown in Table II.Both resistivity and hot TCR are relatively insensi- tive to oxygen concentration in the range from 1 to 25 ppm which is within the practical limits of furnace con- trol.Resistors may be fired with oxygen concentration in the burnout section of 6 to 8 ppm which is the same as used for copper.aSheet resistivity measured on 2.5 mm x 5 mm resistors normalized to 25/m dried print thickness.b25C to 125C. Commercial nitrogen furnaces suitable for firing copper, dielectric and resistor compositions are available from several vendors.Two typical designs are shown in Figure 1.The thick film nitrogen furnace must perform the same functions as an air furnace: remove organic vehicle components, sinter the functional phases and effect bonding to the substrate.The oxygen concentra- tion must be maintained at a relatively low level in the hot sections of the furnace to avoid oxidation of copper.To prevent oxygen from entering the hot sections, purge chambers are provided at each end of the fur- nace.In addition, tight seals are necessary at flange joints and piping connections to prevent air from leak- ing into the furnace.Inadequate seals have been a source of serious processing problems. The muffle construction may be stainless steel or quartz.Quartz provides a cleaner atmosphere, but it can be more difficult to obtain tight seals between a quartz muffle and a metal purge chamber.Furnaces have been provided with external nitrogen purges around the seals to solve that problem. Furnaces may have natural convection vents or forced vents.Both have provided satisfactory service. Nitrogen should be introduced to both the top and the bottom of the entrance purge chamber.Without th6 flow of nitrogen from beneath the belt, air can be trapped between the substrate and the floor of the purge chamber.The air can oxidize prefired copper on the underside of the substrate when it reaches the hot zones of the furnace. NITROGEN FURNACE AUXILIARY EQUIPMENT An oxygen analyzer, sample pump and sample lines are useful auxiliary equipment for monitoring the furnace atmosphere.The analyzer should have a range from 1 ppm to at least 1000 ppm.The sample pump should deliver about 1 liter per minute (2 ft per hour).Sample lines near the hot sections of the furnace should be 1/4" stainless steel.A filter should be installed in the sample line between the furnace and the pump to prevent foul- ing the pump and analyzer with contaminants from the furnace atmosphere.A permanent sample line should be installed to the burnout section.In addition flexible 1/8" stainless steel sample lines are useful for inserting into the furnace at the ends to monitor the atmosphere anywhere along the belt. NITROGEN FURNACE OPERATION It is important to balance the nitrogen flows and vent flows so that the pressure inside the furnace is above atmospheric.That will prevent air from being drawn into the furnace from the ends.Figure 3 shows practical nitrogen flow rates versus muffle width.Copper has been successfully fired with nitrogen flow rates shown at the bottom of the range.A range is shown because some furnaces may have higher muffles and some furnaces may be operated with larger vent flows which will require greater nitrogen flow to the furnace.Ten to twenty percent of the nitrogen should be introduced to each purge chamber and sixty to eighty percent should be added to the muffle. The ends of the furnace should be protected from room air currents and should be enclosed as much as possible to prevent air from entering. Typical commercial nitrogen has an oxygen specifica- tion of less than 10 ppm and is usually supplied with 4 to 8 ppm oxygen.It is not difficult to maintain the oxygen concentration below 15 ppm in the hot sections of a furnace which does not have air leaks. COPPER COMPOSITION PROPERTIES AND APPLICATIONS Copper compositions provide high adhesion conductors with resistivity values in the range of one to two mil- liohms per square per mil.Resistivity of copper is com- pared with precious metal thick film conductors in Copper is somewhat lower in resistivity than platinum-silver and gold conductors and is signific- anti3/lower than palladium-silver and platinum-gold.Thick film copper has been shown to produce low loss microwave circuits with a quality factor exceeding that obtained with thick film gold. Thick film copper forms excellent bonds to alumina and beryllia substrates.Wire peel adhesion 2 values with 2 mm by 2 mm conductor pads range as high as 31 new- tons (7 pounds) initially and 29 newtons (6.5 pounds) after aging 48 hours at 150C.Because of its high adhe- sion copper may be used in place of Mo-Mn metalliz- ing. Thick film copper conductors have outstanding solder leach resistance, a property which will permit extensive reworking of soldered components.Solder leach resistance is illustrated in Figure 5 160 Sn/40 Pb heated at 225C.2Each cycle consists of dip in mildly activated flux followed by 10-second dip in solder followed by solvent cleaning. solder erosion of the edges of conductor pads was minimal. When used with nitrogen-fireable dielectric composi- tion 4175, copper may be used to make multilayer interconnect structures with up to four conductor levels. DIELECTRIC COMPOSITION PROPERTIES AND APPLICATIONS Dielectric 4175 forms a relatively dense film with the electrical properties shown in Table III.At one kHz the dielectric constant is in the range of 7 to 9 and the dissipation factor is less than 0.005.Breakdown voltage is greater than 500 V/mil and insulation resistance is greater than 1 1012 ohms.When used with thick film copper, 4175 dielectric is useful for making multilayer interconnect circuits.Using a screen with 0.30 mm (12 mil) vias, 4175 dielectric will produce via prints which are 0.15 to 0.20 mm (6 to 8 mils) in size. In addition, the dielectric may be used to protect cop- per from the environment and to define conductor areas for subsequent soldering.IV.Coefficient of variationS" of resistiv- ity (CV) ranges from 2.3 to 5.7%.Temperature coeffi- cient of resistance (TCR) ranges from 150 ppm at 10 ohms per square to near zero at 1 OK ohms per square.Both hott and colder TCR have approximately the same value for a given resistivity.In addition the TCR values are not affected by the geometry of the resistors so different size resistors have good tracking characteris- tics. Resistivity change with retiring ranges from + 10% for the 10 ohm per square resistors to 11% for 1 OK ohm per square resistors. fStandard deviation divided by the mean. tHot TCR temperature range is 25 to 125C; cold TCR temperature range is 55 to 25C.Laser trim stability results are shown in Table V.The resistors were trimmed with a YAG laser at a rate of ten mm per second.Trim speeds as high as 75 mm per second have been used.The resistors were one milli- meter in size and were trimmed with a plunge cut to about 1.5 times the original value.They were not en- capsulated during subsequent tests. Resistors were exposed to the following no-load test conditions for up to 800 hours: 25C ambient, 150C storage, 90% relative humidity at 40C and 2 ten- second immersions into 62 Sn/36 Pb/2 Ag solder at 220C.In addition resistors were run with 31 watts per Type laser-YAG Trim speed-10 mm per sec. b31 watts per square centimeter.Drift levels off after 100 hours. cm 2 (200 watts/in2) of power loading in a time cycle of 1.5 hours with power on and 0.5 hour with power off.All compositions changed 0.1% or less at 25C dur- ing the first 24 hours after trimming. After 800 hours the average change in resistivity values for the 10, 100, and 1 K resistors was 0.3% or less during each of the tests.The change during most tests was less than 0.1%.After 300 hours the 10K resistors changed less than 0.3% during all tests except for load conditions.For load conditions of 15.5 watts per cm 2 (100 watts/in2) the resistance change was- 1.24%; for 31 watts per cm 2 (200 watts/in2) loading the change was-2.65%.The change occurred during the first 100 hours under load with little change after 100 hours. Resistivity is relatively independant of the length of the resistor as shown in Figure 6.These resistor compositions have characteristics which make them particularly useful for resistor net- works and hybrid microcircuits.In addition they are compatible with nitrogen-fireable dielectric 4175 and may be applied on the top dielectric layer of multilayer structures. 12. CONCLUSIONS A copper materials system has been developed with high adhesion low resistivity conductors, a dense low K dielectric and resistor compositions from 10 to 10K ohms per square with characteristics similar to the BIROX(R) 1600 series. The compositions are printed and dried using the same equipment and procedures as used for precious metal thick film compositions. The copper materials system compositions are fired in nitrogen furnaces which are commercially available using commercial nitrogen.Care must be taken to prevent air from leaking into the hot sections of the fur- nace to avoid oxidizing the copper. Figure 4 . Figure 4. Copper is somewhat lower in resistivity than FIGURE 4 FIGURE4 Resistivity of thick film copper vs. typical preci- ous metal thick film conductors. FIGURE 5 FIGURE 5 Solder leach resistance: copper conductor pad width vs. number of solder dips. TABLE II Effect of oxygen in burnout section of R and HTCRb. TABLE IV Typical nitrogen-fireable resistor properties. TABLE V Laser trim stability typical drift values.	v3-fos-license
2018-12-22T14:26:41.862Z	2011-08-10T00:00:00.000	62898535	{ "extfieldsofstudy": [ "Biology" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://academicjournals.org/journal/AJB/article-full-text-pdf/F2C46FB30274.pdf", "pdf_hash": "f4bda2e292bab21cf1f627cde09f179305386200", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45432", "s2fieldsofstudy": [ "Environmental Science" ], "sha1": "f4bda2e292bab21cf1f627cde09f179305386200", "year": 2011 }	pes2o/s2orc	Environmentally safe in vitro regeneration protocol for Curcuma, Kaempferia and Zingiber 1 Gene Research Center, Graduate School of Life and Environmental Sciences, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8572, Japan. 2 Institute of Biotechnology and Genetic Engineering (IBGE), NWFP Agricultural University Peshawar-Pakistan. 3 Plant Breeding Laboratory, Faculty of Agriculture, Padjadjaran University, Indonesia. 4 Plant Genetic Resources Program, National Agricultural Research Center, Islamabad45500, Pakistan. INTRODUCTION The sustainable utilization and conservation of plant genetic resources is assisted and accelerated if collections are correctly evaluated, characterized and properly maintained (Hussain et al., 2008).Due to the fast occurring genetic erosion phenomenon, it is very important to conserve indigenous plant genetic resources representative of the prevailing genetic diversity especially from under-utilized agro-ecological regions.Field gene banks and in vitro gene banks are playing an important role in this perspective (Ravindran et al., 2005). Zingiberaceae, known as ginger family, is one of the largest flowering families comprising 53 genera and more than 1200 species (Ahmad, 2010).Members of ginger family notably Curcuma longa and Zingiber officinale have been used for centuries in Chinese, Korean, Japanese and Indian cuisines and traditional Southeast Asian medicines.These species are important natural sources of spices, herbal medicines, dyes, perfumes and multipurpose aesthetic compounds (Chaveerach et al., 2008). Myanmar, situated in Southeast Asia, is considered as one of the center of origin of Zingiberaceae plants (Ahmad et al., 2009).Myanmar is the habitat of 161 Zingiberaceae species belonging to 21 genera where Curcuma and Zingiber are the main species (Jatoi, 2008).The favorable climatic conditions and habitat diversity are the main reasons of plants genetic diversity in Myanmar (Ahmad et al., 2009).The Zingiberaceae genetic resources of Myanmar have not been well characterized and due to excessive uses and urbanization these valuable native plants need to be collected, characterized and properly conserved.In Myanmar, most of the Zingiberaceae species grow wild or as backyard plantation that demands conservation efforts (Jatoi et al., 2006(Jatoi et al., , 2007;;Ahmad et al., 2009). Tissue culture is a common tool for conservation and rapid multiplication of asexually propagated species (Anish et al., 2008).Direct in vitro regeneration of explants without passing through callus phase, in order to keep integrity of the genome, offers a mean for conservation of vegetatively propagated plant species through tissue culture.Different in vitro culture protocols have been used for: Z. officinale (Bhagyalakshmi and Singh, 1988;Hosoki and Sagawa, 1977;Rout et al., 2001); Kaempferia galanga (Shirin et al., 2000); C. longa (Salvi et al., 2001(Salvi et al., , 2002)); Curcuma amada (Prakash, et al., 2004;Barthakur and Bordoloi, 1992) and Curcuma xanthorhiza (Mukhri and Yamaguchi, 1986).Tissue culture for in vitro regeneration of the rhizome bud explants have not been previously reported for Zingiber barbatum species growing widely in Myanmar. Most previously used protocols involved the use of HgCl 2 for disinfection of the explants.HgCl 2 is a serious environmental pollutant and potentially toxic to human and plants (Brandenberger and Maes, 1997;Ma and David, 2006).Ma and David, (2006) reported a disinfection protocol where HgCl 2 was not used for explants disinfection, but this protocol was used for a single species (Z.officinale only).This study reported a disinfection protocol exclu-ding HgCl 2 for explants pretreatment effective for Curcuma, Kaempferia and Zingiber the three very important genera of Zingiberaceae. In Zingiberaceae, different plant organs have been used as explants; the most successful for direct regeneration are the rhizome buds (Ma and David, 2006;Prakash et al., 2004), but culturing efficiency of buds from "mature rhizomes" (rhizomes obtained after plants are harvested at the end of the season) and "immature rhizomes" (rhizomes obtained while plants are still growing in the field) have not been studied previously. This study is a continuation of the previous efforts regarding characterization and conservation of Myanmar Zingiberaceae genetic resources (Jatoi et al., 2007;Ahmad et al., 2009).The specific objectives of this research work were; to extend the environmentally safe pretreatment protocol to other species of Zingiberaceae, to establish a single in vitro direct regeneration protocol effective for different species belonging to Zingiberaceae and to report on the culturing efficiency of buds from mature verses immature rhizomes.In addition, this study presented for the first time the in vitro plantlet production system for Z. barbatum. Plant material Rhizomes of different Zingiberaceae species were collected from different sources in Myanmar.Detail descriptions of the germplasm are given in Table 1.These rhizomes were sown in pots in Tsukuba Japan; in a non replicated mode the growth period extended to 8 months (from end of April to end of December).The mother rhizomes produced new rhizomes referred here as "immature rhizomes" with new buds on them which were used as explants.At the end of the season plants were harvested in December and the rhizomes obtained referred here as "mature rhizomes", were stored in the green house and were carefully checked for aeration and fungal contamination from time to time.In the month of March when the dormancy period came to an end, the mature rhizomes developed new buds which were also used as explants (Figure 1A). Rhizomes pre-treatment The buds from immature and mature rhizomes were used as explants depending upon their availability.The immature rhizomes buds were available in large number but only in the growing season when the plants were young.The mature rhizomes buds were available in small number but for longer time even when the rhizomes were not growing.The rhizomes were soaked in water, scarified and washed carefully.Each rhizome was sliced into small pieces (1 to 3 cm) with buds on them (Figure 1B).The rhizomes pieces were briefly treated with Tween 20 and rinsed with deionized distilled water for 5 min.Later on, the rhizomes were immersed in hot water (50°C) for 10 min.After the hot water treatment, rhizomes were shifted to clean bench and rinsed with autoclaved distilled water 4 to 5 times.Then, the rhizomes pieces were completely submerged in 10% common bleach (pH 12.6) for 30 min and rinsed once with autoclaved distilled water.The buds (5 to 10 mm) were cut apart from the rhizomes pieces.The buds were treated with 0.5% plant preservative mixture (PPM) for 10 min, rinsed in autoclaved distilled water, submerged in 70% ethanol and again rinsed with sterile distilled water. Culturing conditions Murashige and Skoog (MS) medium (Murashige and Skoog, 1962), containing 3% (w/v) sucrose and 0.8% (w/v) agar, was prepared with pH 5.6 ± 0.2.The media was microwaved till the agar was dissolved.When the temperature dropped down to 50°C the media was supplemented with a growth regulator, 6-benzyladenine and a commercial fungicide, Benlate (50% of benomyl, methyl 1-(butylcarbamoyl) benzimidazol-2-ylcarbamate) purchased from Sumitomo Chemical Garden Products Inc. Japan.Five ml media was poured into each test tube.The test tubes containing MS medium were autoclaved and later on stored at 4°C for subsequent use.The sterilized buds were cultured in the test tubes containing sterilized MS medium supplemented with benzyl adenine and benlate (Figure 1c).The test tubes were shifted to the growth room with room temperature maintained at 25 ± 1°C with 16 h light and 8 h dark period under 60 to 70% relative humidity in the culture room. Environmentally safe protocol The use of metals free pre-treatment protocol is very important to avoid environmental contamination and insure health safety, especially for places where the dumping facilities of metals are not according to the internationally recommended standard.Heavy metals like mercury are known for immunotoxic and neurotoxin properties (Silva et al., 2005).Although, HgCl 2 is a very effective disinfectant in plant tissue culture explant pretreatment protocols, but it was shown in this study that its use is not imminent and can be avoided. One of the novelties of this work was to establish a single in vitro direct regeneration protocol efficient for different species of Zingiberaceae simultaneously.Figure 2 represents the work flow of the protocol, from rhizomes harvesting to explant culturing, effective for Curcuma, Kaempferia and Zingiber species.Following this protocol bud explants from Curcuma, Kaempferia and Zingiber after sterilization, were successfully established on the MS media and were contamination free even after successive sub-culturing. Minimizing explants contamination The immature rhizomes were directly taken from the soil during the growing period of the plants while mature rhizomes were taken from the green house and therefore, needed to be surface sterilized to eliminate the microorganisms present on the surface of the explants.The use of Tween 20 (20%) and sodium hypochlorite solution (5%) effectively controlled the exogenous contamination.Exogenous contamination can effectively be controlled by using surface sterilants and disinfectants (Bajaj, 1989). Plants harbor endophytic fungi and bacteria which cause serious contamination of in vitro cultures (Sarasan et al., 2006).The explants were passed through a series of pre-treatment steps including hot water and plant preservative mixture (PPM) treatment which helped to minimize the endogenous contamination.Endogenous Harvest the rhizomes from soil at the end of growing period ↓ Remove all the mud and wash with tap water ↓ Treat with sodium hypochlorite solution (5%) ↓ Sun dry and Store in a well aerated place with no direct sunlight ↓ At the end of dormancy period the new buds start emergence ↓ Scarify the rhizomes carefully and wash with tap water ↓ Slice the rhizomes into small pieces (1 to 3 cm) with buds intact ↓ Treat the rhizomes pieces with Tween 20 for 1 min ↓ Rinse with tap water for 5 min ↓ Immerse in hot water (50°C) for 10 min ↓ Shift to clean bench ↓ Rinse with autoclaved distilled water 4 to 5 times ↓ Submerge completely in common bleach (10%) for 30 min ↓ Rinse once with autoclaved distilled water ↓ Cut the buds (5 to 10 mm) apart from the rhizome pieces ↓ Treat the dissected buds with 0.5% PPM for 10 min ↓ Rinse with autoclaved distilled water ↓ Submerge in 70% ethanol ↓ Final rinse with sterile distilled water ↓ Culture the sterilized buds on MS medium in a test tube ↓ Shift the test tubes to growth room contaminants can make their appearance in the culture even after a long period of time (Bajaj, 1989).To address this problem, Benlate (fungicide) and PPM was added to MS media. The contamination was recorded for all accessions used in this study (Table 2).Bacterial contamination was a lesser problem compared to fungal contamination.The contamination percentage varied with a range of 0 to 39% for different accessions; while, the cultures of some accessions were totally free of contamination (Figure 3).Loc et al. (2005) showed 40% of zedoary explants cultures which were contaminated.In the case of Zingiber, only 10% cultures were contaminated (Ma and David, 2006).In this work, the overall explants contamination less than 12% for Curcuma, less than 34% for Zingiber and less then 7% for Kaempferia (Table 2). Bud-explants growth efficiency The culturing efficiency of buds from mature rhizomes, which were stored for three months in the green house to pass through the dormancy period, versus immature rhizomes, which were obtained directly from the soil during the growth period of the plants, have not previously been studied in Zingiberaceae.When buds from the immature rhizomes were passed through explants pretreatment steps, these buds were unable to response to the MS media and most of them died (data not shown); the regeneration ranged from 5 to 20%.On Accessions No of the germplams used for explant percentge of survival of bud explant Curcuma Zingiber Kaempferia Percentage of survival of bud explant the other hand, when buds from the mature rhizomes were cultured in MS media, after passing through the same pretreatment protocol, they responded amazingly to the media and more than 40 to 95% buds started induction and produced contamination free plantlets (Figure 4).The data in this paper was based on the bud explants from the mature rhizomes.Figure 1 shows the type of buds selected from a mature Curcuma rhizome, the buds induction after culturing on MS media and the development of contamination free plantlet. A variable number of bud explants both from Curcuma and Zingiber were cultured depending on the availability of the material (Table 3).The Zingiber buds explant started induction after two days of culturing, while it took 4 days for Curcuma explants to start induction on MS media.A total of 152 buds were planted for Curcuma where 115 grew contaminated free and produced plantlets, while the remaining 37 bud explant either did not grow or got contaminated (Table 3).In the case of Zingiber, 132 buds were planted where 76 buds grew contaminated free and made plantlets, while 56 buds did not grow (Table 3).In the case of Kaempferia, only 15 buds were planted where 8 buds grew contaminated free and made plantlets, while 7 buds did not grow (Table 3).In Zingiberaceae, different plant parts including buds, leaf sheets and inflorescence have been used as explants, where some studies reported rhizomes buds producing better results (Ma and David 2006;Prakash et al., 2004).The buds growth percentage varied from 40 to 95% in Curcuma, while this percentage decreased in Zingiber (53 to 75%).The C. amada accession ZO18-1 showed the highest survival percentage (Figure 4).Different stages of in vitro plants production including bud explants induction, shooting, rooting and growth at different days are shown in Figure 5. Direct in vitro regeneration system for Z. barbatum It is the first report on the in vitro plantlet production using Z. barbatum buds explants.The same protocol, used for other species in this study, was followed for Z. barbatum in vitro plantlet production (Figure 2).Unlike other species studied, Z. barbatum produced more plantlets (1 to 6) from a single bud where other species produced lesser plantlets (1 to 3) from a single bud in the test tube on MS media.Similarly, Z. barbatum explants survival rate was the highest (75%) among the other accessions of Zingiber (Table 3).The type of bud explants used, stages of shooting and rooting and the fully grown plantlet in the test tube on MS media are depicted for Z. barbatum in Figure 6. Conservation through direct regeneration As pointed out by Hussain et al. (2008), it is very important to select the representative diversity germplasm from fields and properly conserve them in vitro.Only a few Zingiberaceae species like turmeric and ginger are cultivated in some areas of Myanmar, while others like Z. barbatum and C. amada grow wild or are kept as backyard plantation.These plants are very essential not only to keep the integrity of the agro-ecosystem, but they are principal sources of traditional medicines.Due to the human activities including both urbanization and excessive use of the native plants without re-plantation, it is not only needed to collect these valuable genetic resources and characterize them but also their proper conservation is essential for their sustainable use.Being vegetatively propagated plants, plant tissue culture is a desirable method for their conservation.Direct regeneration is preferred for conservation through tissue culture in order to minimize soma clonal variation and genotype dependency which are likely to arise in the case of regeneration through callus, Tyagi et al., 2004).Using a single in vitro direct regeneration protocol for different species belonging to different genera of a family, can certainly reduce the cost and time needed for in vitro conservation.This protocol needs a further investigation to check the effect of different concentrations of different hormones in various combinations on the multiplication of the plantlets. Figure 1 . Figure 1.Different stages of buds explant culturing, (A) Mature Curcuma rhizome with newly developed buds at the end of the dormancy period; (B), Rhizomes piece possessing the bud explant to pass through disinfection before culturing; (C), a bud explant cultured on MS media; (D), a regenerated plantlet on MS media.The arrows indicate the buds explant on the rhizome. Figure 2 . Figure 2. The work chart of the protocol effective for Curcuma, Kaempferia and Zingiber for the in vitro plantlet regeneration from rhizome harvest to in vitro culture. Figure 3 . Figure 3.The comparison of the contamination percentages that occurred in the different accessions of Curcuma, Zingiber and Kaempferia after the explants were cultured on the MS media. Figure 4 . Figure 4.The comparison of explants survival percentages of different accessions of Curcuma, Zingiber and Kaempferia after the culturing on the MS media. Figure 5 . Figure5.The regenerated Curcuma (ZO130) plantlets after buds were cultured on MS media in test tubes; the shooting started in the first week after culturing followed by rooting of the buds (the photographs were taken before changing the media each time). Figure 6 . Figure 6.Different stages of Z. barbatum buds explant culturing.(A), a mature rhizome with the arrows pointed to the buds emergence on the rhizome at the end of the dormancy period, These buds were passed through sterilization and then cultured on MS media; (B), emergence, shooting and rooting of the explants; (C), a 50 days old plantlet. Table 2 . The contamination details of bud explants of the different accessions infected with bacteria and fungus. Table 3 . The number of bud explants cultured from different accessions and their percentage of growth.	v3-fos-license
2023-05-21T15:05:39.534Z	2023-05-01T00:00:00.000	258822740	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.3390/ejihpe13050066", "pdf_hash": "6dee4c1940f3e3ef140a669b43c792525021a383", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45435", "s2fieldsofstudy": [ "Sociology" ], "sha1": "686b2200156ce397cc3aad67772301c721257beb", "year": 2023 }	pes2o/s2orc	A Causal Analysis of Young Adults’ Binge Drinking Reduction and Cessation Background: This study, using the multiple disadvantage model (MDM), sought to identify factors (disadvantaging social disorganization, social structural, social integration, health/mental health, co-occurring substance use, and substance treatment access factors) in young adults’ binge drinking reduction and cessation in the United States. Methods: We extracted data on 942 young adult binge drinkers (25–34 years, 47.8% female) from the National Longitudinal Study of Adolescent to Adult Health (Add Health), carrying out a temporal-ordered causal analysis, meaning the evaluation of select variables’ impacts on an outcome at a subsequent time. Results: MDM found a relatively high reduction likelihood for non-Hispanic African Americans and respondents with relatively more education. MDM found a relatively low reduction likelihood accompanying an alcohol-related arrest, higher income, and greater number of close friends. Change to nondrinking was found more likely for non-Hispanic African Americans, other non-Hispanic participants having minority ethnicity, older respondents, those with more occupational skills, and healthier respondents. Such change became less likely with an alcohol-related arrest, higher income, relatively more education, greater number of close friends, close friends’ disapproval of drinking, and co-occurring drug use. Conclusions: Interventions incorporating a motivational-interviewing style can effectively promote health awareness, assessment of co-occurring disorders, friendships with nondrinkers, and attainment of occupational skills. Introduction The present study was intended to identify risk and protective factors in the reduction and cessation of binge drinking by young adults in the U.S. In 2019, more than 34% of American adults aged 18-25 and nearly 25% aged 26 or more were binge drinkers. Binge drinking is defined as consuming four or more alcoholic drinks, for females, and five or more alcoholic drinks, for males, on the same occasion, on at least 1 day out of the past 30 days [1]. Binge drinking has been linked to the mental health disorders depression and anxiety [2] and to the physical disorders liver disease, cancer, high blood pressure, and heart disease [3]. It has also been associated with injury-causing vehicular accidents as well as falls and burns. Research also links binge drinking to three forms of violence-homicide, suicide, and intimate partner violence. Furthermore, 30% of annual alcohol poisoning deaths involve binge drinking [4]; indeed, binge drinking is involved in 18,000 alcoholrelated deaths per year [2]. Binge drinking is obviously an important public health issue. Literature Review The multiple disadvantage model has been successfully applied to understand the onset of adult drinking [5], victimization in intimate partner violence [6], and homicide [7,8], as well as access to substance treatment [9]. In the present study, the model tested whether drinking behavior-reduction and cessation, specifically-was linked to six types of disadvantaging factors (see Figure 1). [7,8], as well as access to substance treatment [9]. In the present study, the model tested whether drinking behavior-reduction and cessation, specifically-was linked to six types of disadvantaging factors (see Figure 1). First were social-disorganization factors including residence in an unsafe or rundown neighborhood. Next were the three social structural factors: race/ethnicity, income, and education. Next were social integration factors, comprising measures of social relationships and social support. Additionally, we examined health and mental health factors, measures of the respondent's use of substances beyond alcohol, and factors describing access to substance treatment. The literature links the social disorganization factor disadvantaged neighborhood, reflecting economic deprivation and multiple social problems, to drinking among adults [5,10,11]. It seems plausible that the challenges of life in disadvantaged neighborhoods could hinder any effort or even desire to give up alcohol. Still, one prior study has reported that living in a disadvantaged neighborhood showed no effect on binge drinking reduction [12]. On the other hand, alcohol-related, possibly binge drinking-related, problems including violence, injury, drunk driving, and property damage clearly contribute to such neighborhoods' social problems [10]. These risks should logically foster the reduction and cessation of binge drinking [13]. However, the literature notes that binge drinking reduction has been found to diminish where alcohol-related problems are observed [14]. The social structural factors race/ethnicity, income, and education have relationships with binge drinking reduction and/or cessation. A prior study conducted within one state of the United States showed that African Americans and Asian Americans were less likely than White Americans to engage in binge drinking [15]. Compared to White Americans, a different study of binge drinkers found that Americans of a minority ethnicity were less likely to stop binge drinking but more likely to reduce binge drinking [13]. An additional study on pregnant women, showed that respondents of a minority ethnicity were less First were social-disorganization factors including residence in an unsafe or rundown neighborhood. Next were the three social structural factors: race/ethnicity, income, and education. Next were social integration factors, comprising measures of social relationships and social support. Additionally, we examined health and mental health factors, measures of the respondent's use of substances beyond alcohol, and factors describing access to substance treatment. The literature links the social disorganization factor disadvantaged neighborhood, reflecting economic deprivation and multiple social problems, to drinking among adults [5,10,11]. It seems plausible that the challenges of life in disadvantaged neighborhoods could hinder any effort or even desire to give up alcohol. Still, one prior study has reported that living in a disadvantaged neighborhood showed no effect on binge drinking reduction [12]. On the other hand, alcohol-related, possibly binge drinking-related, problems including violence, injury, drunk driving, and property damage clearly contribute to such neighborhoods' social problems [10]. These risks should logically foster the reduction and cessation of binge drinking [13]. However, the literature notes that binge drinking reduction has been found to diminish where alcohol-related problems are observed [14]. The social structural factors race/ethnicity, income, and education have relationships with binge drinking reduction and/or cessation. A prior study conducted within one state of the United States showed that African Americans and Asian Americans were less likely than White Americans to engage in binge drinking [15]. Compared to White Americans, a different study of binge drinkers found that Americans of a minority ethnicity were less likely to stop binge drinking but more likely to reduce binge drinking [13]. An additional study on pregnant women, showed that respondents of a minority ethnicity were less likely than White respondents to cease or reduce binge drinking [16]. Yet, another study reported no such relationship [17]. Furthermore, one study in the literature reports that binge drinking reduction diminished with higher family income [18], and another that binge drinking reduction and/or cessation increased with additional education [16,17]. We could not locate any prior research about possible relationships between binge drinking reduction and cessation and level of occupational skill. According to the literature, binge drinking reduction is promoted by social support from family and friends and by the presence of networks of nondrinking relatives and friends [19]. In addition, binge drinking reduction appears to be promoted by being unmarried and also by being a parent [18]. It is reasonable to think, then, that social support, social networks, and marital status might also have impacts on binge drinking cessation. Since participation in religious activities has been observed to cultivate cessation of alcohol use [20][21][22][23], we speculated in the present study that participation in religious or other community-based activities would facilitate cessation and reduction of binge drinking. Generally, among adult drinkers, being ill or in poor physical health is linked to cessation of alcohol use [24,25]. Often, binge drinkers with depression cease consuming alcohol only to continue battling depressive symptoms [26]. Some studies in the literature report that depressive clients are relatively less likely to stop drinking [21,27]; other studies found no such relationship [28]. In addition, the literature reports that access to Medicaid facilitates cessation of binge drinking, and yet the lack of such access demonstrates no impact on reduction in binge drinking [16]. Nevertheless, receiving treatment for alcohol use enabled participants in several studies to reduce their binge drinking [29][30][31][32][33][34]. Co-occurring use of other substances, however, hinders the cessation of alcohol use in general [24,27,28]. These many mixed results need to be addressed as research continues. Therefore, we hypothesized, in our study, that among young adults, the reduction and cessation of binge drinking would be associated negatively with social disorganization (i.e., unsafe neighborhood, drinking-related arrest), with physical health, and with drug use. We also hypothesized that these outcomes would be associated positively with social structural factors (i.e., gender, age, race/ethnicity, income, education, occupation) and social integration (i.e., marital status, spousal relationship, close friends, spousal disapproval of binge drinking, close friends' disapproval of binge drinking, religiosity, volunteering). Finally, we hypothesized that the outcomes would have a positive association with mental health (i.e., depression), access to health insurance (i.e., health insurance coverage), and access to substance treatment (i.e., receipt of counseling). Sample The present sample was drawn from the National Longitudinal Study of Adolescent to Adult Health, or Add Health, data set [35]. Between 1994 and 2008, Add Health collected data from over 15,000 nationally representative respondents orchestrated through four waves of interviews. The original core sample's response rate was 77% at Wave 3 and 78% at Wave 4 [36]. Through in-home interviews in Wave 3 and Wave 4, Add Health researchers collected information on respondents' substance use, social and economic factors, physical and mental health, family, peers, school, and community [37]. Most respondents were aged 18-26 at Wave 3 (in 2001-2002). Since Add Health collected longitudinal data from a nationally representative sample, such data allows the examination of changes in young adults' binge drinking behaviors over time. Furthermore, the data set collected information on neighborhoods and on individuals' alcohol-related arrests. Add Health's public-use data included only 6504 adolescents in its Wave 1 core sample. To examine possible changes in binge drinking mode over time, we isolated the Wave 3 data of 942 young adult binge drinkers and sought predictors of their drinking reduction or cessation at Wave 4. We defined binge drinker as a respondent who self-reported consuming at least 4 alcoholic drinks, for a female, or at least 5 alcoholic drinks, for a male, on the same occasion, on at least 1 day out of the past 30 days [1]. Measures For our temporal-ordered causal analysis of the extracted data, the outcome variable binge drinking type had 3 categories. Binge drinking maintained indicated that, from Wave 3 to Wave 4, the respondent did not change their binge drinking behavior. Change to non-binge drinking denoted that a respondent had a binge drinker status at Wave 3 but had a non-binge drinker status at Wave 4. Change to nondrinking indicated that a respondent had a binge drinker status at Wave 3 but had a nondrinker status at Wave 4. Binge drinkers, again, were respondents who self-reported consuming at least 4 drinks, if female, or at least 5 drinks, if male, on the same occasion, on at least 1 day out of the past 30 days. Non-binge drinkers were those respondents who reportedly consumed fewer than 4 drinks, if female, or fewer than 5 drinks, if male, in that same time frame. Nondrinkers were those respondents who self-reported no alcohol consumption in the 12 months preceding the interview. Temporal-ordered causal analysis is, basically, an examination of the impact of an explanatory variable at one time (here, at Wave 3) upon the outcome variable at a subsequent time (here, at Wave 4). In conducting this temporal-ordered causal analysis, we measured most of our explanatory variables at Wave 3, and measured each variable's Wave 3-to-Wave 4 difference. A variable's Wave 3 value constituted its baseline. If an explanatory variable's wave-to-wave difference had a positive value, that meant its value at Wave 4 had risen above its baseline. If a variable's wave-to-wave difference had a negative value, that meant its value at Wave 4 had fallen below its baseline. For the Add Health research, there were few explanatory variables that were measured at Wave 3 but not Wave 4. Several explanatory variables, however, were measured solely at Wave 4, because they affected the outcome fairly instantaneously [38]. Our temporal-ordered causal analysis involved measures of ethnicity and gender as time-invariant variables. Our study involved 4 sets of explanatory variables. The first comprised social disorganization factors. Unsafe neighborhood at Wave 3, a dichotomous (yes/no) variable, indicated whether a Wave 3 Add Health interviewer had perceived a respondent's neighborhood or surroundings to be unsafe, implying a disadvantaged neighborhood. Unsafe neighborhood difference indicated any Wave 3-to-Wave 4 change in the characterization of respondents' neighborhoods. Involved in fights at Wave 3 was the number of times in the 12 months preceding the Wave 3 interview that a respondent had engaged in physical fighting because of drinking. Being arrested at Wave 4 was the number of times a respondent had been arrested, per Wave 4 self-report, on a charge of disturbing the peace or driving while impaired or any drinking-related infraction (0 indicating never, 1 indicating once, 2 indicating more than once). Our second set of explanatory variables included social structural and demographic factors. A respondent's socio-economic status was represented by 3 variables: personal income, in USD; education level (1 indicating no high school graduation, 2 indicating high school graduation, 3 indicating some college, 4 indicating undergraduate degree, and 5 indicating graduate school or higher); and occupational skill (0 indicating not in labor force, 1 indicating service worker, 2 indicating operative/farmer, 3 indicating clerical/sales worker, 4 indicating craftsman, 5 indicating professional/manager). Our classification of occupational skill employed a modified version of Rexroat and Shehan's well-known scale [39]. Personal income difference, education level difference, and occupational skill difference measured Wave 3-to-Wave 4 changes in personal income, education level, and occupational skill, respectively. The demographic variables we considered in our study were male (versus female), age (in years), and ethnicity (non-Hispanic White [the reference group], Hispanic, non-Hispanic African American, other non-Hispanic ethnic minority). Our third set of explanatory variables included social integration factors. Being married at Wave 3, a dichotomous (yes/no) measure, denoted if the respondent was married, cohabiting, or romantically/sexually involved with a partner at Wave 3. Being married difference described a change in marriage, cohabitation, or involvement as time elapsed from Wave 3 to Wave 4. The variable satisfactory spousal relationship at Wave 3 rated respondents' satisfaction with any spouse/partner reported at the Wave 3 interview. Satisfaction was rated on a 5-point Likert scale with responses ranging from 1 (very unsatisfied) to 5 (very satisfied); respondents who were single were assigned 0. Satisfactory spousal relationship at Wave 4 similarly rated respondents' satisfaction at the Wave 4 interview. It was assessed as a global construct using the sum of seven 5-point Likert scale items (with a Cronbach's alpha of 0.87). The 7 items were "enjoyed things together", "handled problems", "listened to partner", "expressed affection", "enjoyed sex life", "trusted faithful partner", and "handled finances" [40,41]. Higher scores indicated more satisfactory relationships. Additionally, the third set of explanatory variables included partner disapproved binge drinking at Wave 3. This variable indicated if, according to the respondent at the Wave 3 interview, the respondent's spouse or other partner disapproved of binge drinking. The measures were obtained using a 5-point Likert scale ranging from 1 (strongly approved) to 5 (strongly disapproved). Additionally in our third explanatory variables set, number of close friends measured how many persons a respondent reported being at ease with, comfortable talking to about private matters, or confident calling on for help. Add Health researchers specified five responses for the measure: 1 (no close friend), 2 (1-2 close friends), 3 (3-5 close friends), 4 (6-9 close friends), and 5 (10 or more close friends). Number of close friends difference measured any Wave 3-to-Wave 4 change in number of close friends. Close friends disapproved binge drinking at Wave 3 rated close friends' disapproval of binge drinking, reported at Wave 3. A 5-point Likert scale ranging from 1 (strongly approved) to 5 (strongly disapproved) was used to measure this variable. Four further explanatory variables made up the third set. Religiosity at Wave 3 represented frequency of engagement in religious activities of any type in the 12 months preceding the Wave 3 interview. Offered responses ranged from 0 (never) to 5 (more than once per week). Religiosity difference denoted Wave 3-to-Wave 4 changes in frequency of religious engagement. The dichotomous (yes/no) variable volunteering noted if a respondent had performed volunteer work or served a community in some way in the 12 months preceding the Wave 3 interview. Volunteering difference tracked any Wave 3-to-Wave 4 change in such work or service of the respondents. The fourth set of explanatory variables employed in this study presented health, mental health, and healthcare access factors. The respondents self-reported their general physical health at Wave 3 as either 5 (excellent), 4 (very good), 3 (good), 2 (fair), or 1 (poor). General physical health difference measured any change in general health from Wave 3 to Wave 4. Depressive feelings at Wave 3 was measured by the 9-item version of the Center for Epidemiologic Studies Depression Scale (CES-D). The scale yields respondents' total score on self-reported "feeling depressed", "losing appetite", "experiencing hopelessness", and other symptoms of depression. The same CES-D scale was used to rate each of the 9 symptoms or feelings. The responses for each item ranged from 0 (never/rarely) to 3 (most/all of the time); higher scores indicated more frequent depressive feelings. The scale generated a Cronbach's alpha of 0.82 at Wave 3 and of 0.79 at Wave 4. Depressive feelings difference denoted any change in the frequency of depressive feelings from Wave 3 to Wave 4. Two dichotomous (yes/no) measures represented healthcare access. Private health insurance at Wave 4 indicated any possession of health coverage through employment, a spouse or parent, union, school, or private purchase. Public health insurance at Wave 4 indicated coverage by Medicaid or the federal Indian Health Service. For this measure, the reference group comprised respondents with no health insurance. Additionally, a dichotomous (yes/no) answer for attended drug treatment at Wave 3 measured respondent self-reports of alcohol or drug treatment received in the 12 months preceding the Wave 3 interview. Attended counseling services at Wave 4 (yes/no) indicated if a respondent had been counseled for a psychological or emotional problem in the 12 months preceding the Wave 4 interview. The final variable in the set, drug use at Wave 4 (yes/no), measured whether a respondent had used marijuana, cocaine, inhalants, or another illicit drug in the 30 days preceding the Wave 4 interview. Data Analysis Because our outcome variable had 3 categories and Add Health employed a clustering sample design, our data analysis involved applying Stata software's version 15.1 survey procedures for multinomial logistic regression (featuring linearized variance estimations of robust standard errors). The analysis used statistically significant relative risk ratios (RRRs) to identify significant predictors of our outcome variable. With the clusters and Wave 4 sampling weights provided by Add Health, we attained a final sample of 129 clusters. Preliminary analysis of tolerance statistics (≥0.49) and correlations (−0.60 ≤ r ≤ 0.27) suggested no multicollinearity problems among the explanatory variables. Results Of the 942 binge drinkers in our sample, 44.7% remained binge drinkers at Wave 4, while 46.6% became non-binge drinkers and 8.7% became nondrinkers (see Table 1). The average age at Wave 3 was 28.4 years and 52.2% of the sample was male. Non-Hispanic Whites constituted 77.0% of the sample, with non-Hispanic African Americans constituting 9.8%, Hispanics constituting 10.2%, and other non-Hispanic ethnic minorities constituting 3.0%. At Wave 3, 21.2% of the sample were married, 3.0% resided in unsafe neighborhoods, 28.6% volunteered, and 4.4% attended drug treatment. Of the sample at Wave 4, 70.3% had private health insurance, while 5.5% had public health insurance. At Wave 4 as well, 10.3% of the sample attended counseling, and 28.2% self-reported drug use. The For seven variables, moving from Wave 3 to Wave 4, we observed increases in average scores. Personal income had an increase of USD 23,344.20. For the variable education difference, the average score rose by 0.6. For the variable occupational skill difference, the average score rose by 1.1. For the variable being married difference, the average score rose by 0.4. In addition, for the variable number of close friends difference, the average score rose by 1.1. Finally, for volunteering difference, the average score rose by 1.1, while for depressive feelings difference, the average score rose by 0.3. In contrast, for two variables, moving from Wave 3 to Wave 4, the average scores showed decreases. The average score for religiosity difference showed a decrease of −0.1, while that for general physical health difference showed a decrease of −0.3. Multivariate Analysis Results The results of multinomial logistic regression confirmed that the hypothesized model differed significantly from the null model (F = 5.26, p < 0.01; see Table 2). The likelihood of change to non-binge drinker at Wave 4 was negatively associated with reported arrest at Wave 4 (RRR = 0.62, p < 0.01); with personal income difference (RRR = 0.99, p < 0.05); with number of close friends at Wave 3 (RRR = 0.18, p < 0.01); and with number of close friends difference (RRR = 0.75, p < 0.01). Such a likelihood was associated positively with non-Hispanic African American ethnicity (RRR = 2.28, p < 0.01); with education level at Wave 3 (RRR = 1.45, p < 0.01); and with education level difference (RRR = 1.60, p < 0.01). Other variables' associations with the likelihood of change to non-binge drinker were insignificant. Eight of the tested factors proved to reduce the likelihood of change to nondrinker at Wave 4. They included self-reported arrest at Wave 4 (RRR = 0.61, p < 0.01); relatively high personal income at Wave 3 (RRR = 0.99, p < 0.05); relatively more education at Wave 3 (RRR = 0.60 p < 0.05); and relatively larger wave-to-wave increase in education (RRR = 0.51, p < 0.01). The other four factors were relatively high number of close friends at Wave 3 (RRR = 0.03, p < 0.01); relatively large wave-to-wave increase in number of close friends (RRR = 0.56, p < 0.01); close friends' relatively greater disapproval of binge drinking at Wave 3 (RRR = 0.63, p < 0.01); and drug use at Wave 4 (RRR = 0.38, p < 0.01). In turn, six of the tested factors increased likelihood of change to nondrinking at Wave 4. They included non-Hispanic African American ethnicity (RRR = 5.16, p < 0.01); other non-Hispanic ethnic minority (RRR = 3.22, p < 0.05); and relatively older respondents (RRR = 1.45, p < 0.01). These six factors also included relatively greater occupational skill (RRR = 1.41, p < 0.05); relatively greater wave-to-wave increase in occupational skill (RRR = 1.22, p < 0.05); and relatively greater wave-to-wave increase in physical health (RRR = 1.47, p < 0.01). The remaining variables showed no significant associations with the likelihood of change to nondrinker at Wave 4. Discussion The 942 binge drinkers constituting our sample made up 24.5% of the 3844 young adults interviewed in Wave 3, which took place during 2001-2002. This percentage of binge drinkers was less than the proportion reported in a prior study [1]. The discrepancy may suggest growth, over the years, in the number of binge drinkers in the young adult population of the U.S. The present study also found a relatively large proportion of binge drinkers, 46.7%, had become non-binge drinkers eventually, although less than 9% were observed to become nondrinkers. In other words, among young adults, becoming a nondrinker is uncommon. Furthermore, almost 45% of binge drinkers in our study maintained their binge drinking behavior. The present study's multivariate results partially supported the hypotheses that (1) binge drinking reduction and cessation is negatively associated with social disorganization (i.e., unsafe neighborhood, drinking-related arrest), with physical health, and with drug use; (2) these outcomes are positively associated with social structural factors (i.e., gender, age, race/ethnicity, income, education, occupation) and social integration (i.e., marital status, spousal relationship, close friends, spousal disapproval of binge drinking, close friends' disapproval of binge drinking, religiosity, volunteering); and (3) the outcomes have a positive association with mental health (i.e., depression), access to health insurance (i.e., health insurance coverage), and access to substance treatment (i.e., receipt of counseling). In observing no link between binge drinking and unsafe neighborhoods, our study confirms some prior results [12]. Our multivariate results also support other published research that found, as our study did, a negative association between arrest for drinking-related offenses and likelihood of change to non-binge drinker, as well as to nondrinker [14]. Such results imply that, in a vicious circle, binge drinkers may use binge drinking to manage the challenges of residing in disadvantaged neighborhoods whose disadvantages are fueled by the disruptions caused by their binge drinking. It is implied that an effective intervention would be public health education specifically for young adults who live in disadvantaged neighborhoods and have arrest records involving or suggesting typical drinking-related problems. The present study showed non-Hispanic African American respondents to be 1.2 times more likely to achieve binge drinking reduction and 4.2 times more likely to achieve drinking cessation than non-Hispanic White respondents; similarly, other non-Hispanic ethnic minority respondents were 2.2 times more likely than non-Hispanic White respondents to achieve drinking cessation. Such results support certain published results [13]. Probable explanations for African Americans' likelihood of binge drinking cessation included their strong ethnic identity [42] and neighborhood norms of rejection of African American binge drinkers [43]. In contrast, cessation by Asian Americans may be explained by the rejection of binge drinking by some Asian subgroups. It could also be explained by some of these subgroups' adjustment to their host nation's cultural discouragement of binge drinking (however, in tandem with acceptance of drinking) [44,45]. Consistent with prior results [16,17], our present findings confirmed that the likelihood of binge drinking reduction was positively associated with education level, as well as with wave-to-wave differences in education level. At the same time, our findings indicated that having more education, as well as increasing attainment of education, were linked to a lower likelihood of drinking cessation. This contradicted the results of a prior study involving solely pregnant women [16]. The present study observed that a relatively high income, as well as an increasing income, diminished the likelihood of drinking cessation and likelihood of binge drinking reduction, respectively. This observation supported results of a prior study [18]. Our study also showed that having more occupational skills as well as improving occupational skills increased the likelihood of drinking cessation. Such results can be plausibly explained. The attainment of relatively more education probably motivates individuals to cut back on binge drinking but not actually cease alcohol consumption. In addition, a growing income increases spending power, including spending on alcoholic beverages. In contrast, growing occupational skill may well be attended by growing responsibility at work, with the employer expectations perhaps facilitating drinking cessation. One earlier study with a small sample of Native American women [18] led to results unlike those of our present study, in which marital status, satisfaction with relationship, and partner disapproval of binge drinking showed no links to binge drinking reduction or cessation. In addition, our findings show that religious or volunteer community activities were not associated with binge drinking reduction or cessation. This contradicts prior findings concerning alcohol use in general [20][21][22][23]. In addition, our study found that respondents with more close friends or a growing number of close friends were less likely to reduce binge drinking or experience binge drinking cessation, compared to respondents with few close friends. Moreover, the likelihood of binge drinking cessation was observed to dwindle with relatively strong Wave 3 disapproval of binge drinking by close friends. An implication of such results is that young adults whose binge drinking continues unabated have, and make new, "drinking friends" instead of friends who abstain and also likely disapprove of binge drinking. Our findings in this area underscore the importance of maintaining a network of non-drinking friends [19], regardless of the existence of any spousal partner or religious commitment. Our study observed that depressive feelings were not associated with binge drinking reduction or cessation. Furthermore, the absence of any apparent link, in our study, between substance treatment and binge drinking reduction or cessation puts the study at odds with prior work [29][30][31][32][33][34]. We also found that self-reported counseling at Wave 4 was not associated with the outcome, nor was the insurance type. Unsurprisingly, this study showed that drug use at Wave 4 reduced the likelihood of binge drinking cessation. Additionally, the present study found no association between binge drinkers' health or improving health and their likelihood of binge drinking reduction; nevertheless, those whose health improved from Wave 3 to Wave 4 were the most likely to give up binge drinking. It seems plausible that many binge drinkers have emotional/behavioral problems that they continue palliating, temporarily, with alcohol or drugs. Only those who come to fully acknowledge alcohol's role in their health are likely to pursue cessation in order to regain health. These findings imply a need for interventions helping binge drinkers who also use illicit drugs to gain insight on how drug use negatively affects their health and mental health. Three notable limitations affected our present study. First, the data were based on self-reports and may be susceptible to social desirability bias. Second, the study's exclusive focus on young adults means its results may not be generalizable to those outside that age group. A third limitation is that Add Health interviewers, not respondents themselves, described the level of safety of the respondents' neighborhoods. Conclusions The most important findings of the present study were that alcohol-related arrests, drug use, a relatively higher income, and relatively large network of friends were risk factors that curtailed binge drinking reduction and cessation within this study's sample of young adults. In contrast, a non-Hispanic minority ethnicity fostered the group's reduction and cessation of binge drinking. Additionally, the factors education, occupational skill, and friends' disapproval of drinking affected the reduction and/or cessation of binge drinking. The present study apparently observed an intertwined group of risk and protective factors characterizing young adults' binge drinking reduction and drinking cessation that could provide broader implications for interventions and practice. To be efficient and successful, our study's results suggest that any campaign fostering reduction/cessation of binge drinking would need to focus on young adult binge drinkers living in disadvantaged neighborhoods who have been arrested for alcohol-related offenses. The campaign would need to inform the individuals about behaviors that can improve health. It would be essential to reach young adult binge drinkers who use another substance as well as alcohol or who have a dual diagnosis. To curtail binge drinking, interventions should stress expanding clients' awareness of how one's alcohol-using friends can influence one's choices in unhealthy ways. Interventions should foster the development of networks of nondrinking friends. Completing this kind of intervention among college students who are young adults-especially while using a motivational-interviewing style-should prove effective [46]. Interventions provided to young adults of a minority ethnicity should, moreover, promote ethnic identity and explore various cultural beliefs about binge drinking. One further profitable aim of intervention would be to equip young adult binge drinkers with better occupational skills. Future research similar to ours is needed. The role played by cultural factors like ethnic identity and acculturation in binge drinking reduction/cessation by young non-Hispanic minority adults should be a topic of future studies; so should the role of mental disorders other than depression. Furthermore, research remains to be performed exploring bingedrinking behavior's relationship to social disorganization factors such as neighborhood norms or racial discrimination. Future research might also ask why education and income demonstrate such opposite effects on the reduction in binge drinking. Future research could, moreover, explore factors in binge drinking reduction and cessation by individuals in later adulthood. Author Contributions: T.C.C. and C.C.L. made equal contribution to the preparation of the manuscript as well as to the performance of the data analysis. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: Not applicable. The institutional review board of the author's university of affiliation exempted the present secondary research of public-use data from review. Informed Consent Statement: Not applicable. This research employed a public-use data set that does not contain identifiable information of participants. The original researchers that produced the data set obtained informed consent from the participants. Data Availability Statement: The Add Health was obtained through Inter-university Consortium for Political and Social Research; www.icpsr.umich.edu (accessed on 22 March 2023).	v3-fos-license
2022-08-18T15:12:16.452Z	2022-08-01T00:00:00.000	251631149	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/1422-0067/23/16/9158/pdf?version=1660566276", "pdf_hash": "ed56c55f3ae402537238e7e48e2f6a67c5d7a16a", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45436", "s2fieldsofstudy": [ "Biology", "Medicine" ], "sha1": "5dab762fae6be770388ce26f11554d23f176de4f", "year": 2022 }	pes2o/s2orc	Looking into the Eyes—In Vitro Models for Ocular Research Animal research undoubtedly provides scientists with virtually unlimited data but inflicts pain and suffering on animals. Currently, legislators and scientists alike are promoting alternative in vitro approaches allowing for an accurate evaluation of processes occurring in the body without animal sacrifice. Historically, one of the most infamous animal tests is the Draize test, mainly performed on rabbits. Even though this test was considered the gold standard for around 50 years, the Draize test fails to mimic human response mainly due to human and rabbit eye physiological differences. Therefore, many alternative assays were developed to evaluate ocular toxicity and drug effectiveness accurately. Here we review recent achievements in tissue engineering of in vitro 2D, 2.5D, 3D, organoid and organ-on-chip ocular models, as well as in vivo and ex vivo models in terms of their advantages and limitations. Introduction The number of factors that can damage human tissues increases every year. For example, smog, substances contained in cosmetics, unnatural food additives, and UV radiation have a harmful effect on our skin and eyes. Moreover, every year industry delivers thousands of new chemical substances which are necessary for new medicines, chemicals, or food additives. Therefore, biocompatibility assessments of each new compound, especially one that involves animal testing, is impossible. Moreover, the number of studies performed on animals has to be limited in Europe by the law (i.e., Directive 2010/63/EU). Optic neuropathies, such as glaucoma, anterior ischaemic optic neuropathy (AION), traumatic optic neuropathies, optic neuritis, etc., need new treatment options, which in turn require the development of disease models [1]. However, the pathophysiological mechanisms of these diseases are not fully understood; therefore, developing an animal model is a tough challenge. Moreover, due to the physiological differences, animal models differ significantly from human diseases [2]. For example, rodents' eyes do not have maculae or foveae, and 85-90% of their optic nerve axons decussate to the other side of the brain [1]. On the other hand, monkeys' anatomy of the retina and optic nerve is almost identical to that of human eyes. Still, monkey breeding is complicated, very expensive, and time-consuming; therefore, the number of tests performed on individual animals is limited. As a result, monkeys are often used in the stage just before clinical trials on humans [1]. Historically, one of the most popular experiments performed on animal eyes is the Draize test. Developed in 1944 by American toxicologists John H. Draize and Jacob M. Spines, it was widely used to study cosmetics and other chemicals. However, the test arouses many controversies due to the lack of reliable and objective results. In fact, the test was never correctly validated. Briefly, the test is based on applying the test substance directly to the eye, but the exposition time is not well defined. After observing the eye reaction for some time, the substance is washed from the eye, and the animal is observed for another two weeks. The result is subjectively assessed by the operator [3]. Moreover, the test is considered incorrect mainly due to anatomical and biochemical differences between the human and the animal (mostly the rabbit) eye. Therefore, currently, the Draize test is not performed. Instead, chemicals are usually tested using the EpiOcular eye irritation test, in vitro cytotoxicity assay, and irritation tests on the rabbit's skin. Each of these tests has advantages and disadvantages, but none of them allows for multifactorial compound-eye interaction evaluation. The EpiOcular eye irritation test is an in vitro alternative which allows for the assessment of acute eye irritation in response to the topical administration of chemicals onto the EpiOcular cornea epithelial model. The test makes cytotoxic effect measurement possible and provides a tool for eye-hazardous chemical identification. For these reasons, it is crucial to develop new in vitro tissue models to study all substances for ocular treatment and to understand the development and molecular causes of eye diseases [4]. Here, we summarized all cellular and tissue-specific animal models used in in vitro eye studies. Eye Structure The eye is a highly complex biological machine ( Table 1). The human eye has the shape of a sphere about 24 mm in diameter. It is filled with a vitreous body that allows the shape of the ball to be maintained. It is located in the eye socket, which reduces the risk of mechanical damage. The eye is divided into two parts: external and internal. The eye's outside layers are tough, elastic structures, with a white sclera and transparent cornea which provide eye shape [5]. The second inner layer is the vascular membrane, including the iris, ciliary body, and choroid. The third layer is the retina. It consists of light receptors, cells, nerve fibers, and blood vessels originating from the central retinal artery [6]. The primary function of the eye is to convert light pulses into electrical signals, which are transmitted to the brain and converted into images. Light is refracted by the cornea and the lens, which results in a sharp, inverted, reduced image formed on the retina. The amount of light reaching the receptors is regulated by the iris, which changes pupil diameter [6]. The retina in vertebrates is characterized by light-sensitive structures and covers 60% of the back of the eyeball. It is located above the choroid. It consists of 10 cell layers and contains photoreceptors called rod cells and cone cells [7]. These cells contain a visual pigment which is located in the cell membrane. Rod cells allow the recognition of shapes and motion at low light intensity [8]. Cone cells are responsible for seeing color and detail in a more intense light than rod cells. Individual cone cells differ in sensitivity at wavelength, which allows us to distinguish colors. In humans, cone cells are concentrated in the macula-a small pocket of the retina center. The optic nerve's axons that leave the retina at the ONH form a blind spot. There are no photoreceptors in that spot [9]. The eyeball is surrounded by the connective tissue-the sclera. It is a hard layer protecting the inner structures of the eye. In addition, it stiffens the eyeball like a bag of collagen and elastic fibers [10]. The sclera is thinner, more permeable to light, and creates a transparent cornea on the front. It refracts the light rays so that they fall on the lens. It is susceptible to pain and able to partially regenerate. It is avascular and consists of six layers (epithelial, Bowmann, stroma, Dua's, Descemet, and endothelial). The conjunctiva lines the inner surface of the eyelid. It contains a lot of mucous cells, which ensures the constant humidity of the eyeball-it produces mucus and tears. It covers the eyeball up to the edge of the cornea. It has a very high regeneration ability. The conjunctiva is sensitive to any irritation, such as smoke, dust, or chemical substances. These factors can lead to conjunctivitis. During this type of inflammation, blood vessels are firmly filled with blood, causing redness and swelling of the eye [11]. Glands are dispersed in the conjunctiva, and the lids secrete mucus, water, and lipids, forming a tear film whose primary function is to moisturize and cleanse the eye from undesired foreign bodies if needed. There are many glands dispersed in the conjunctiva and the lids that secrete mucus, water, and lipids, thus forming the tear film. However, the main lacrimal gland (responsible for emotional tears) is outside the eye structure. Even though the eye is a very specialized organ, there is significant progress in the development of tissue engineering, and newer and more suitable in vitro models are emerging. The search for such models is caused, among others, by increasing awareness of the welfare of animals used in experiments, including toxicological effects [12]. In Vitro Ocular Models Over the years, many alternative assays were developed to accurately evaluate ocular toxicity and drug effectiveness. Here we present recent achievements in tissue engi-neering of various ocular models in terms of their advantages and limitations (see also Figures 1 and 2). 2D Eye Models Currently, 2D models are the most popular ones in ocular research. Two-dimensional cell line culture is an inexpensive, well-established model providing results that are easy to compare with the vast literature. However, the unquestionable drawback of these culture systems is the lack of predictivity in research connected with the fact that cells growing on a flat surface are not an equal representation of the cell environment in the organism. Pigment Epithelium Cell Lines One of the most common 2D models is immortalized retinal pigment epithelium (RPE) cell lines. Primary cultures of retinal cells are challenging to handle. Obtaining a homogenous cell line that is not contaminated with other eye cells is challenging. Furthermore, isolated cells often quickly change their properties. For example, cells can lose keratin-containing intermediate filaments [13]. Cell transformation using the SV40 virus managed to obtain a line that retains the characteristics of retinal cells [14]. These cells are characterized by appropriate polarization and monocellular epithelial cell formation. In 1995, RPE cells were first isolated by Davis et al. from a patient. However, the RPE cell line was only used in toxicity tests because the cells lost the characteristics of normal metabolism, adequate cytoskeleton polarization, and enzyme activity [14]. In the literature, primary models of RPE cell culture obtained from mice (i.e., Mouse Retinal Pigment Epithelial Cells-Hpv16 E6/E7, Immortalized) [15], rats (RPE primary cells isolated from PVG rats susceptible to experimental uveitis development; RPE isolated from Long Evans rats) [16,17], chickens (primary RPE cells isolated from domestic chickens embryos at stages 29-31 of development) [18], bovines (primary RPE cells) [19], and frogs (Xenopus laevis isolated primary RPE cells) [20] have also been described. The human cell line ARPE-19 has structural and functional properties characteristic of RPE cells in vivo (in rats, RPE-J) [11]. This line is essential because the number of tissue donors is limited [21]. Studies on ARPE19 showed several features confirming the usefulness of this line for retinal pigment epithelial examination, such as expression of characteristic RPE cell markers, CRALBP, and RPE65, secretion of IL-6 and IL-8, as well as morphological polarization in monolayers, and ability form tight-junctions [22,23]. RPE-340 are primary cells isolated from humans which have epithelial morphology, but after several passages, their ability to replicate is limited [24]. Human RPE cells are good models for pharmacodynamic and physiological evaluation of a drug's effect on the choroid-RPE-photoreceptor, but after 40-60 population doublings, they go into a senescence state [24]. To develop a cell line with an extended lifespan, RPE-340 was transfected with a plasmid expressing the human telomerase reverse transcriptase subunit (hTERT), creating a new cell line-hTERT-RPE-1 (human retinal pigment epithelial RPE-1) [25]. This way, the lifespan of hTERT-RPE-1 is extended without any alterations in the population, doubling time and RPE-340 characteristic features [26]. hTERT-RPE-1 is reported to be an excellent model for epigenetic regulation studies [27][28][29]. Unfortunately, this line still has its limitations. The handling lasts 20 passages longer than RPE-340, but after this period, the cells change their morphology and function under the phenomenon called deadaptation [30,31]. Due to the low availability of primary human RPE cultures, validating and comparing this cell line with immortalized cell lines is challenging. The perfect line of human RPEs has yet to be developed. 2D Eye Models Currently, 2D models are the most popular ones in ocular research. Two-dimensional cell line culture is an inexpensive, well-established model providing results that are easy to compare with the vast literature. However, the unquestionable drawback of these culture systems is the lack of predictivity in research connected with the fact that cells growing on a flat surface are not an equal representation of the cell environment in the organism. Pigment Epithelium Cell Lines One of the most common 2D models is immortalized retinal pigment epithelium (RPE) cell lines. Primary cultures of retinal cells are challenging to handle. Obtaining a homogenous cell line that is not contaminated with other eye cells is challenging. Furthermore, isolated cells often quickly change their properties. For example, cells can lose keratin-containing intermediate filaments [13]. Cell transformation using the SV40 virus managed to obtain a line that retains the characteristics of retinal cells [14]. These cells are characterized by appropriate polarization and monocellular epithelial cell formation. In 1995, RPE cells were first isolated by Davis et al. from a patient. However, the RPE cell line was only used in toxicity tests because the cells lost the characteristics of normal metabolism, adequate cytoskeleton polarization, and enzyme activity [14]. In the literature, primary models of RPE cell culture obtained from mice (i.e., Mouse Retinal Pigment Epithelial Cells-Hpv16 E6/E7, Immortalized) [15], rats (RPE primary cells isolated from PVG rats susceptible to experimental uveitis development; RPE isolated from Long Evans rats) [16,17], chickens (primary RPE cells isolated from domestic chickens embryos at stages 29-31 of development) [18], bovines (primary RPE cells) [19], and frogs (Xenopus laevis isolated primary RPE cells) [20] have also been described. The R28 immortalized retinal precursor cell line originating from postnatal day 6 rat retinal culture has been frequently used in in vitro and in vivo studies [32]. R28 provides an important system for understanding retinal cell behavior aspects such as differentiation, cytotoxicity, light stimulation, and neuroprotection. Although R28 originated from single clones, they remained highly heterogenous, suggesting the precursor character of these cells [32]. This cell model has been used in various toxicity experiments in vitro [33][34][35][36][37]. In addition, R28 exerts a high potential for studying the neuroprotective properties of chemical compounds [38][39][40]. Latanoprost was one of the drugs validated on R28 under the angle of cytoprotective properties [33]. RGC-5 was previously described as a rat-derived, transformed retinal ganglion cell line and is widely used in glaucoma research [41]. After more than 220 published papers worldwide involving the use of the RGC-5 cell line, it was reported that these cells are in fact 661W, a mouse SV-40 T antigen transformed photoreceptor cell [41,42]. The 661W cell line was present in the laboratory of origin of RGC-5; therefore, the most probable scenario was the cross-contamination of the newly developed cell line with 661W. This incident has shown how crucial the proper culture protocols and DNA profiling of newly-developed cell lines are. 661W is a model of cone photoreceptor cells. This cell line was widely used as a model for research on macular degeneration, but studying retinal ciliopathies such as retinitis pigmentosa is believed to be possible [43]. 661W shows properties of both retinal ganglion and photoreceptor cells, providing a functional photoreceptor model [43,44]. Moreover, 661W are believed to be an alternative model to the hTERT-RPE-1 cell line previously used for small molecule screening to identify new treatments for retinal ciliopathies [45]. 661W shows potential in studying ciliopathy disease genes not expressed or expressed at a low level in hTERT-RPE-1 cells [43]. Cornea Cells The primary cornea cultures on which the individual in vitro models were developed come mainly from rabbits. Rabbit corneal epithelial cells (RbCEpC) help assess drug safety, pharmaceutical effects, corneal development, pathology, glaucoma, viral infections, keratitis, ocular hypertension, and even special contact lenses that provide sustained, extended-release of ophthalmic drugs [31]. Human corneal epithelial cells (HCEpC) have been used as models for studying corneal damage and reconstruction, re-epithelialization of the eye following surgery, and the effects of degradative enzymes. Corneal research mainly focuses on developing a model of drug permeation through this structure [46]. The models of corneal culture used in cellular research concern simple monolayers, the multilamellar epithelium, and very complex three-dimensional (3D) tissues resembling the functional cornea. HCEpC was used to create a single cellular layer later used for transplantation [47]. Several commercially available in vitro cornea models are destined to be cultured in 2D models (monolayers). One such model is HCE-T, in which cells are grown on the collagen membrane and are located at the air-liquid interface with the serum-free medium. The cells have the features of the primary cell line and form a stratified epithelium whose morphology can be modulated with calcium. Moreover, the cells expressed specific corneal epithelial cell markers such as epidermal growth factor (EGF), EGF receptor, basic fibroblast growth factor (basic FGF), transforming growth factor-beta 1 (TGF-beta 1), and interleukin-1 alpha (IL-1 alpha) [48]. Corneal Endothelial Cells The role of corneal endothelial cells (CECs) is to control corneal transparency. Unfortunately, the cells exhibit limited proliferative capability; therefore, their dysfunction may be one of the causes of blindness. One of the gold standards in treatment of corneal endothelial dysfunction is the donor isolated corneal transplant [49]. The first culture of human corneal endothelial cells from donors was established by Pistov et al. in 1988 [50], and from that time, plenty of protocols and newly designed biomaterials for the propagation of these cells were developed [51]. However, due to the low proliferation rate of primary CECs culture, protocols for immortalized cells were established [52,53]. Currently, several immortalized human CECs are available in the market and are used mainly to understand corneal endothelial cell dysfunctions. Two clonal cell lines derived from the immortalization of human corneal endothelial cells (obtained from the donor) were described by Valtink et al. [54]: B4G12 and H9C1 cells. B4G12 cells are polygonal, strongly adherent cells, which form a strict monolayer and H9C1 cells are less adherent and formed floating spheres. Both cell lines exhibited the characteristic expression of corneal endothelial cell markers; however, on different levels. Therefore, the authors concluded that the B4G12 cell line is a good model of differentiated CECs, and H9C1 is a good model for developing or transitional CECs. Alternatives for donor corneal endothelial cells or cornea endothelial cell lines may be pluripotent stem cell-derived corneal endothelial cells. The cells generated from a cryopreserved human embryonic stem cell (hESC) are stable, express corneal endothelial cell markers, and have an improved proliferation rate compared to primary CECs [55,56]. Conjunctival Cells The most popular eye conjunctival test model is the rabbit conjunctiva. For the first time in 1996, Saha et al. isolated the primary culture of the rabbit's conjunctiva. The model created by him represents a tight epithelial barrier [57]. This model is constantly being improved. The most significant difficulty was adapting the cells of this model to contact with air, just like in the natural eye. The use of additional filters (for example, the Transwell filter) allowed cell growth at the air-liquid interface. The layers of the conjunctival epithelial cells showed transepithelial resistance and a difference in potential [58]. The conjunctival epithelial cells are polygonal with many microvilli [59]. The primary culture of conjunctival cells was also obtained from bovines [59] and rats. The immortalized rat conjunctival (CJ4.1A) cell line was created by transfection of SV 40 [60]. CJ4.1A expresses the SV40 T antigen, conjugal cytokeratin 4, and cytokeratin specific for goblet cells 7, but not the cytokeratin 12. The line's lifespan is very long-line cells can be cultured for over 60 passages, and the population doubling times were 22 ± 7 h [60]. The development of methods for obtaining the primary culture of conjunctival cells has contributed to the development of transplantation techniques for heterotopic or allogeneic grafts. After severe damage to the conjunctiva, it is possible to restore its function by taking a piece of epithelium from a healthy eye, multiplying it in a cell culture, and implanting it in the affected conjunctiva [61]. Besides the primary cell culture, several established conjunctiva cell culture lines also exist. An example would be two human immortalized conjunctival cell lines: HCjE [62] and IOBA-NHC [63]. These cells have a typical epithelial morphology of the human epithelium, and after exposure of the cells to inflammatory mediators (IFNγ and/or TNFα), they increase the expression of the intercellular adhesion molecule (ICAM)-1 and MHC class II cell surface receptor (HLA-DR) [63]. The above-mentioned 2D models have their advantages but also their limitations. Firstly, these models are exceptionally delicate, and their manipulation must be meticulous. For example, the layer is easily damaged and dried. In addition, the models do not take into account cell to cell communication and the influence of immunological factors, which probably have a tremendous impact on the regeneration of this structure. Finally, based on these models' results, it is impossible to recapitulate all the processes occurring in the cornea in the human eye [64]. Therefore, the researchers decided to develop more complex, multicellular eye models. 3D Models Three-dimensional models better replicate the organism-environment compared to two-dimensional cultures. Cells grow in every dimension and closely replicate tissue in vitro, which complements 2D cell culture [65]. Although 3D models provide us with more information than 2D models, they are more challenging to handle. Multilayer models respond to more and more questions about corneal damage and disfigurement, but they are still far from the complex equipment that the eye is. For example, they lack the lacrimal apparatus responsible for cleansing and supporting regeneration [66]. Few 3D cornea models have been developed to this point. EpiOcular™, developed in 2010, was obtained from cultured human epithelial cells. The cells showed a morphology and expression of biomarkers similar to the intact human cornea and maintained its thickness and permeability [67]. The EpiOcular model was used to assess the eye irritation potential of surfactant and surfactant-based formulations. Based on the protocol, the compound is considered to be an irritant when more than 50% of the cell die as compared to the negative control [68,69]. This test is validated and under review by the European Center for the Validation of Alternative Methods (ECVAM). Clonetics (cHCEC) was developed in 2011 and was obtained from human corneal epithelial cells. Research using the model provides information on the assessment of corneal penetration by various chemical compounds (e.g., ophthalmic drugs) [68]. cHCEC was examined by RT-PCR for the expression profile of drug-metabolizing enzymes (e.g., CYP P450s and UGT1A1) and transporters in cHCE in comparison to the human cornea [70]. The SkinEthic (HCE) 3D cornea model comprises immortalized human mucosa cells; cells are grown at the air-liquid interface using a polycarbonate membrane. Under appropriate conditions, the cells differentiate and form the three-dimensional (3D) stratified epithelium and have non-keratin structures [69]. The advantage of the model is the possibility of administering dissolved substances in organic and inorganic solvents at any concentration (a very concentrated solution or minimal drug application concentrations can be applied) and the preservation of conditions similar to the eye mucosa of the human eye. Immortalized cells are grown in a dedicated medium and form a histologically multilayered construct with a thickness of 60 µm. The HCE secretes the same mucins found in the human cornea in vivo and expresses CD44 and keratin. This model is used to study phototoxicity, irritation, corrosivity, and the transport of substances [71,72]. The LabCyte CORNEA-MODEL is produced from normal human cornea epithelial cells [73]. It was developed by differentiating and stratifying cornea epithelial cells and is meant to be used to identify irritant chemicals in eye irritation tests. The corneal epithelial cells are cultivated on an inert filter substrate for 13 days with a medium containing 5% FBS. Proliferating cells build up in a multilayer structure consisting of a fully differentiated epithelium with features of the average human corneal epithelial tissue [74]. The limited source of corneal tissue to form a 3D-model, the short-lived life cycle of the corneal cells themselves, and the time-consuming culture contribute to problems with the industrialization of culture. The use of many commercially available 3D models was limited by the rapid differentiation of cells leading to problems with maintaining cell culture [75]. Many attempts have been made to increase the in vitro culture cycle of corneal epithelial cells concerning telomerase reverse transcription gene transfection, viral transfection, and the induction of spontaneous mutations. Nevertheless, the abnormal phenotype of these cells, which can lead to the potential risk of tumorigenesis, is not desired in the construction of new cornea models. Therefore, Li et al. enriched cornea cells with limbal stem cells providing additional expansion and development stimulation [75]. The addition of limbal stem cells promoted development and cell expansion. Moreover, it enabled the large-scale production of a new 3D model. Use of the corneal stromal layer of the animal to stimulate a specific microenvironment for limbal stem cells resulted in their differentiation into cornea epithelial cells. Zuguo et al. proposed a new in vitro xeropthalmia model by dissecting the conjunctival epithelium and subconjunctival matrix, culturing it on a collagen I coated dish submerged in a culture medium. The in vitro dry eye model is obtained after 4-20 days. The invention can be used to research dry eye squamous metaplasia, ocular surface epithelial barrier damage, epithelial mucin change, to test new drugs, or to find new methods for dry eye treatment [76]. The model created by Minami et al. consists of bovine epithelial, stromal, and endothelial cells in a collagen gel matrix. The epithelium consists of five to six layers, and the epithelial cells produce keratin, which is a fundamental multilayer model for the cornea [77]. In addition, some corneal models use cell lines from different animals. In these models, individual layers come from mice, rabbits, bovines, and pigs [65,78,79]. 2.5D Models 2.5D models seem to be an alternative approach compared to 2D and 3D cultures. In 2D, cells are grown on a flat surface, while 3D models are based on cells embedded in an extracellular matrix (ECM) and/or scaffolds that provide a proper three-dimensional environment. In 2.5D cultures, cells are grown in an extracellular matrix (ECM) layer which often is not flat but unregular with projections and grooves, thus, providing an intermediate between 2D and 3D conditions [80]. Ex Vivo Models One alternative to the Draize test is harvesting organs for examination from animals used for meat (ex vivo model). Eyeballs are isolated from bovines (BCOP), rabbits (IRE), pigs (PCOP), and chickens (ICE). This test was accepted internationally in 2009 and is used to research if significant tissue damage can occur [81]. Tests on the models mentioned above are based mainly on histological and light transmittance through cornea analysis. The pigs' cornea provides the highest degree of similarity to the human cornea, especially in tests involving substances dissolved in water [82]. Unfortunately, all the models mentioned above have serious drawbacks, primarily resulting from anatomical differences. In addition, these models can only be used to study individual eye structures. Therefore, they do not allow for general-purpose research. It is vital to create cell microenvironments that support tissue differentiation and changes, tissue-tissue communication, and spatiotemporal chemical and mechanical gradients of the microenvironment of living organs [83]. Yu F. proposed an ex vivo mammalian cornea culture system used for chemical tests of consumer products [84]. This system closely resembles in vivo testing by maintaining the corneal structure, architecture, and epithelial cell interaction. The cornea or the whole eye is excised and placed on an agar or collagen scaffold. It is then submerged in a culture medium until the medium covers the limbus. The upper part of the cornea is not submerged in medium. The tested reagent is administered directly to the cornea. The inventor claims that the system may be used to replace the use of Draize's test in many situations. This system allows drug testing without using live animals. The corneas or eyeballs may be, for example, easily acquired when dissecting rabbits for meat or fur industry purposes. Spheroids, Organoids, and Organ-on-Chips New techniques and technologies in cell culture allow the development of more proper and scientific-useful models for ocular research. Here we described three types of it: spheroids, organoids, and organ-on-chips, a summary of which is presented in Table 2. Spheroids Spheroids are self-assembly aggregated cells that spontaneously organize themselves into spherical-shaped structures. This phenomenon occurs naturally during embryogenesis, morphogenesis, or organogenesis. In in vitro culture, single cells may constitute multicellular spheroids after applying appropriate cell culture techniques (i.e., pellet culture, the hanging drop method, culture in the extracellular matrix, or others) [85]. Spheroids may have a different biological response to various factors due to the presence of a concentration gradient of nutrients, oxygen, or metabolites between cells from the outside and the inside part of the spheroid. Spheroids are mainly used in cancer research [86,87]. However, the technique of 3D multicellular culture with spheroids is also used in cellular research. Lu et al., using air-lifting 3D spheroid formation techniques, developed an in vitro model for research on the ocular surface and tear film systems. The model was composed of rabbit conjunctival epithelium and lacrimal gland cell spheroids [88]. The model allowed for the creation of the aqueous and mucin layers of the tear film, which may facilitate research on dry eye. A Japanese-German research group generated multicellular spheroids from human-donor RPE cells cultured in a methylcellulose matrix [89,90]. The model mimics the in vitro drusen model, which might help understand the pathogenesis of drusen-related diseases such as AMD. Sherwin's group from New Zeeland developed methods for isolation and propagation of spheroid human peripheral cornea using a clear cornea component of the rim isolated from a donor [91,92]. They found that generated spheroids implanted into frozen-stored corneoscleral tissue worked as limbal stem cell centers and proliferated to reproduce limbal cells. Spheroids are also used in ocular cancer research. There are several spheroid models of retinoblastoma (cells isolated from human intraocular tumors) which are used to develop new cancer treatments [93] or to understand retinoblastoma pathophysiology [94,95]. Organoids Organoids are stem cell derived 3D structures with organ-level functions. They are composed of self-organizing organ-specific cells derived from embryonic stem cells, induced pluripotent stem cells, or organ-restricted adult stem cells [96,97]. One of the most well-known ocular organoid models is a model described by Eiraku et al. [98]. The authors used mouse embryonic stem cells and show that ESCs in differentiation medium are self-organizing into optic-cups in 3D culture. Susaimanickam et al. developed an organoid model based on human embryonic stem cells (ESCs) or human induced pluripotent stem cells (iPSCs) cultured in a retinal differentiation medium supplemented with noggin [99]. The addition of noggin is crucial because of the protein's (a BMP inhibitor) involvement in the retinal differentiation of pluripotent stem cells during embryonic and organoid development. After two weeks, the culture gave rise to retinal and corneal primordia, and after six to eight weeks, primordia developed into minicorneas with specific morphological and marker similarities to the human cornea. This model may be used in basic research and regenerative applications. In addition, the use of organoid models with different ranges of time culture could provide us with data regarding drug toxicity in different stages of eye development. A congruous model or cornea organoids was developed by Foster et al. [100]. In this model of the cornea, three distinct cell types with the expression of key epithelial, stromal and endothelial cell markers were obtained. Mellough et al. in 2012 showed that ESC and iPSC cultured in ventral neural induction media (VNIM) supplemented with noggin, Dickkopf-1, Insulin-like growth factor 1, Lefty A, Human Sonic Hedgehog, and 3, 30, 5-triiodo-L-thyronine may develop retinal photoreceptor cells [101]. Later, they showed that VNIM can differentiate both EPS and iPSC cells, but the presence of IGF-1 is essential for the development of 3D ocular-like structures containing retinal pigmented epithelium, neural retina, primitive lens, and corneal-like structures [102]. In the latest work, Mellough et al. found that different embryoid bodies' (EBs) generation protocols affect the method and maintenance conditions that determine the later differentiation and maturation of retinal organoids [103]. The generation of more advanced in vitro multiocular organoids from human iPSCs cells was proposed by Isla-Magrané et al. [104]. In this protocol, organoids are differentiated in three different media, which leads to obtaining multicellular organoids after 150 days. Firstly, 75% confluent hiPSCs were cultured on Matrigel in an induction medium (DMEM/F12, 5% fetal bovine serum, nonessential amino acids, GlutaMax, N2, B27, β-glycerol phosphate, nicotinamide, Noggin, DKK1, bFGF) for 30 days. Next, all-trans retinoic acid (ATRA) was added for the next 60 days. Finally, cells maintained in a medium with ATRA for the next 60 days develop multiocular and corneal organoids, and cells cultured without ATRA and with triiodothyronine develop retinal organoids, RPE organoids, and multiocular organoids. Recently, the National Centre for the Replacement Refinement & Reduction of Animals in Research (NC3Rs) and The National Eye Institute established a relationship that will result in the construction of organoids for drug screening, disease modeling, and regenerative medicine [105]. Therefore, a retinal 3D model is constantly being developed to fulfill those criteria [105][106][107]. The retinal 3D model contains bioprinted Müller cells, microglia, neurons, and RPE cells [108]. Organ-on-Chips Organ-on-chips (OoC) are structures created by combining microfluidic technology, biomaterials, and cell culture methods [97]. Many organ-on-chips were used to research the permeability of the epithelium. Puleo et al. created a microfluidic device consisting of a bilayer structure of a corneal epithelial layer, a layer of stromal cells, and collagen vitrigel substrate [109]. Bennet et al. invented a cornea organ chip including epithelial layers, Bowman's membrane, basement membrane, and a device simulating tear flow dynamics. The measurement of epithelium permeability underflow showed results similar to in vivo measurements [110]. Cornea and retina chips are powerful and promising in vitro tools to study drug effects and therapeutic approaches, yet the chips are still minimal and straightforward 60 [97]. Recently, Seo and Huh proposed a cornea-on-chip "human blinking eye model" [111]. The system mimicked spontaneous eye blinking in humans with keratinocytes cultured to mimic the epithelial cells and form a corneal structure. Blinking imitation was performed by integrating a tear chamber in a 3D-printed eyelid [112]. DynaMiTES' Dynamic Micro Tissue Engineering System was developed from cornea immortalized cells. The system allowed for the measuring of transepithelial electrical resistance in real-time by implementing two electrodes into the system, providing a noninvasive way to monitor cell conditions [113]. Although organoids and organ-on-chips carry indisputable benefits, their potential in drug testing has yet to be closely examined. The main issues concerning drug assays relate to permeation and accessibility of the ocular surface of the tested models [97]. In Silico Analysis In silico analysis is often used to meet the 3Rs regulations (replacement, reduction, and refinement) [114]. Many in silico models have been proposed up to this point in time. One of them is a quantitative structure-property relationship (QSPR) model proposed by Vincze et al. to study corneal permeability. The model is based on corneal-PAMPA (Parallel artificial membrane permeability assay) experimental data and different in silico drug transport parameters (Caco-2 and jejunal permeability) [115]. The test provided good predictions and is suitable for efficiently shortening the examined drugs list, provided we have comparable experimental data at our disposal. However, although promising, in silico studies currently do not provide us with enough data to regard drugs as safe. Therefore, experimental testing should be carried out to confirm the result of the studies. Conclusions Tissue engineering is one of the most rapidly developing scientific disciplines. It allows an easy and more reliable study of the effects of various factors and substances (including drugs). The development of this field will contribute to the invention of more advanced methods of combating diseases, repairing damaged tissues as a result of trauma, and to the ability to change and improve the function of given structures. At the same time, it will limit the number of animals used for experiments, which are now often indispensable research models. Currently, scientists are trying to fine-tune in vitro models and combine as many elements as possible to create a fully functional organ. One of the paths leading to this goal is the development of bioreactors. Bioreactors extend the time of in vitro culturing through specific, periodic exchanges of the culture medium. Physical factors are strictly controlled, e.g., temperature, pH, oxygen, and carbon dioxide. Additionally, they enable the precise delivery of nutrients and the removal of unnecessary metabolites from the nutrient solution [116,117]. All of the research models mentioned above have their limitations and advantages. Different Draize test alternative models provide more extensive flexibility in our research. Currently, 2D cultures are the most common research models. The reason is that 2D cultures are relatively inexpensive, more modulable, and easy to maintain [118]. Because of reproducible results obtained in controlled conditions [119], big-scale screening assays should be performed on these models. The main weakness of 2D models is their low ability to recreate the complexity of different cell classes and matrices interaction [118]. On the other hand, 3D multilayer models seem to be sufficient for small-scale drug toxicity and irritation assays. These models more closely resemble the eye microenvironment and consider cell-to-cell interactions, providing more relevant results. Moreover, 2D and 3D models seem to be limited in immunological disorders, such as allergy or sensitivity, because of their low of complexity. This problem could be addressed with organoids, which generate remarkable research outcomes, but only after long and arduous steps of standardization and testing [118]. Both in vitro and ex vivo models share one major limitation: the lack of vascularization [120]. The immune cells and vascularization should be introduced to these models to address this problem more appropriately. Organ-on-chip technology may be applied to facilitate the manipulation of more complex research models [120,121]. For example, including blood vessels in the model is possible by applying a forced flow supplied by on-chip technology. All that remains is to hope that the current development of in vitro models in ocular research allows for the complete elimination of the need to conduct tests on living organisms in the near future. Data Availability Statement: The data reviewed in this study are available on request from the corresponding author. The data are not publicly available due to founding agreement limitations. Conflicts of Interest: The authors declare that they have no conflict of interest.	v3-fos-license
2017-08-15T11:38:04.059Z	2013-02-20T00:00:00.000	9928861	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "https://doi.org/10.5772/55375", "pdf_hash": "bc3cfa3c457ec89a84e63519d2ec12045a9cff6e", "pdf_src": "ScienceParseMerged", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45437", "s2fieldsofstudy": [ "Computer Science" ], "sha1": "942655baf1d6caa40dbe534674104b7e4f1e3c36", "year": 2013 }	pes2o/s2orc	Ocular Movement and Cardiac Rhythm Control using EEG Techniques There exist different methods to analyze and study biomedical signals. Some of these methods are based on medical imaging, involving Magnetic Resonance Imaging (MRI), Computed Tomography (CT), Nuclear Scintigraphy, etc. This techniques show the image of a specific part of the human body. But there exits other methods which study the different signals based in 1-dimensional analysis, such as Electroencephalography (EEG), Electrocardiography (ECG), etc. These do not show an image, but they also show relevant information. Introduction There exist different methods to analyze and study biomedical signals.Some of these methods are based on medical imaging, involving Magnetic Resonance Imaging (MRI), Computed Tomography (CT), Nuclear Scintigraphy, etc.This techniques show the image of a specific part of the human body.But there exits other methods which study the different signals based in 1-dimensional analysis, such as Electroencephalography (EEG), Electrocardiography (ECG), etc.These do not show an image, but they also show relevant information.These techniques are not only used to detect any anomaly, but in the case of the EEG, it is also possible to develop communication systems by Brain Computer Interface (BCI) [1].The BCI technology has numerous applications which can improve the quality of life of those people who need external help at the time of communication or controlling their movements. EEG records the electrical activity along the scalp produced by the neurons.This activity happens since the brain cells communicate each other giving place to tiny electrical impulses. The impulse begins with a chemical discharge which origins a current in the membrane of the emitting cell.Once the impulse gets the extreme of the connection between cells, the neuron secretes a protein that inhibits or excites another neuron. After receiving the signal, the neuron releases ions to the outside of the cell.When lots of ions are expelled at the same time they can stimulate other neurons.At the time this wave of ions gets to the electrodes, the ions can attract or push the metal of the electrodes.This difference of pressure can be measured by a voltmeter and the record of this activity along time is the EEG signal.Different studies show that, with the correct training, it is possible to control different external devices with the BCI technology.In [2], Galán et al. present a study where two users have to control a wheelchair using a system based in BCI.They have concluded that the success has been higher after learning the interaction with the system. With respect to the communication, the most used technique is the detection of the P300 wave as result of a photostimulus [3].In these cases, the user has to watch one keyboard where the columns and rows are illuminated randomly.If the letter the person is watching is illuminated, the brain sends a signal as a response to that photostimulus.This answer is called P300.In [4] the authors use a system based on evoked potential to control one robotic arm.The system illuminates different actions which can be made by the arm.To know which action the user is focused, they detect the P300 and the N2pc. There are also different applications of BCI technology in different games, such as Mindball [5], which consist on moving a ball depending on the brain activity: the more relaxed and focused the user is, the more the ball will move. But electrodes are not only used to measure brain signals, but they also register the electrical activity from other parts, such as the heart (ECG) or the skeletal muscles (EMG).It is also possible to pick up the muscular activity, as it could be the measurement of the eye movements (EOG). We have decided to use an EEG as electrocardiograph (ECG) and as electrooculography (EOG).The used device is the same for both applications: a portable EEG with 4 different channels. This device was initially acquired to work in one application in which it was necessary to carry a portable EEG.Due to its portability, it can collect data during long periods of time and outdoors. The electrooculography (EOG) is based on the electrical activity of the eyes movement and it has been used for the user's communication.There are other applications based on the eye movements to control devices.They detect the user's iris, so it can know the position the user is looking through the coordinates of the iris.In [6] the study presents a method to control a wheelchair with one camera establishing three points as reference: the eyes and the nose.This three points form a triangle. The ECG is a register of the electrical activity of the heart in order to know if there is any anomaly depending on the heart behavior. We have used two different placements of the electrodes: • Bipolar placement: each channel represents the difference between two adjacent electrodes. This montage has been used for the EOG. • Referential or monopolar placement: each channel represents the difference between one electrode and another one established as reference.The ECG has been acquired with this montage. The chosen electrodes are adhesive because of their facility to be placed both in the face as well as in the breast.The diameter of each electrode is 24 millimeters.The electrodes are responsible for registering the voltage inside a single cell.An electric impulse, known as the action potential, is an electric discharge which travels along the cellular membrane.The action potential is used to transport the information between the tissues.Although they can be generated for different kind of corporal cells, the most active are the cells of the nervous system [7].For the monitoring and the posterior analysis we have developed two user's interface in Matlab: one for the ECG and another one for the EOG. Relation between EEG and EOG During the register of an EEG, there can appear different signals called artifacts.These signals can be originated by multiple causes: muscular movements, skin impedance, technical problems, etc.One of the causes of these artifacts can be the ocular movement, such as eye blinking.This electrical activity is registered in the frontal region of the brain, being FP1, FP2 the channels where can best be appreciable. Figure 6. EEG with two eye blinking But to register the eye activity it is not necessary to place the electrodes on the head, it can be registered placing the electrodes in the face, which is called EOG. An EOG is a method to register the eyes movement by placing little electrodes near the eyes muscles.Under normal conditions there exists a potential difference (10µV to 5 mV approximately) between the cornea and the Bruch´s membrane, located on the back of the eye.This is known as cornea-retinal potential.The cornea corresponds to the positive extreme and the retina corresponds to the negative of the dipole. Figure 7 represents the differential of potential between the cornea and the retina: In [8] it is well explained the different kinds of eye tracking systems and the evolution of them. For studying the eyes movement the criteria for divide the different techniques is different.There are three different kinds of eye tracking: • Invasive technique: this technique uses a special contact lens with an incorporated mirror or a magnetic field senor.The data acquired are more detailed because it is in contact with the eyeball, but is more uncomfortable because the user has to wear the contact lens. • Electrical potential: it uses electrodes placed around the eyes to detect the movement [9].It is a robust method to measure the ocular movements when there are eyes blinking or change of the gaze.The main disadvantage is that it may become uncomfortable and the register can be affected by the electrodes movement.One of the main advantages is that it registers the eye movement with closed eyes so they can be used in the analysis of sleeping problems [10] and [11].The EOG is within this kind of eye tracking. • Non -invasive technique: there is no contact with the eyes.The eyes movement is located through a camera or an optical sensor.The infrared light generates different reflections of the cornea of the user's eyes known as Purkinje images [12] and [13].The difference between channels 2 and 3 shows whether the user is looking up (positive) or down (negative).The difference between channels 4 and 5 determines if the user has looked to the left (positive) or to the right (negative). The final application will have one user interface divided into two tasks: • Communicate between the device and the main application via Bluetooth. • Interpret what the user wants to express.For the first task, we have used the API provided by the fabricant.To allow the user's communication, the application will show the same keyboard as Figure 12: The user moves one circle through the keyboard by moving his eyes to the left, to the right, up or down.Once the user selects "OK", the application will use the T9 dictionary to interpret the correct word.The dictionary has been programmed taking in count the words which meaning is related with expressing feelings and emotional states. Figure 13 show the difference between looking up and looking down.The subtraction of CH2 and CH3 accentuates the difference between the eye movements.In Figure 14, which shows the difference between looking to the left or to the right, it can be seen that the situation is similar than looking up and down: looking to the left produces an increment of the signal value, while looking to the right decrements it. Relation between EEG and ECG An ECG is the graphical representation of the electrical activity precedent from the heart.It is a non -invasive method to study and to analyse the condition of the heart with the aim of detecting possible anomalies or diseases. In order to synchronize the cyclical contraction of the heart, the fibers of the cardiac muscle transmit electrical impulses. The electrical activity is the potential difference generated by the cardiac cells (each cell acts as a voltage generator) which are appreciable on the skin surface, where the electrodes pick up this activity in order to get a graphical representation. This representation shows 4 significant parts: "P Wave": the cardiac cycle begins with the depolarization, which leads on the contraction of the atriums in order to deposit the blood in the ventricles.The sinus node (SA node) indicates the atrium muscles that they have to contract to begin the sequence.The potential action is propagated through the specialized cells of the cardiac muscle to the AV nodule.The electrodes are places along the trunk of the users, taking the shoulders as reference.Figure 21 shows the position of the electrodes for the correct register of the ECG.In this section, it will be described how the EOG has been used to move a mark along a keyboard on the screen in order to help the people who are not able to communicate with the rest due a disease. First of all, the application calibrates the position of the electrodes and establishes the correspondent thresholds to know the position of the eyes.After that, it will show the keyboard the user will use for the communication.Once the application shows the keyboard, it will do the operations described in Figure 24 and Figure 25 to know if the user has made any action with the eyes. Figure 24. Up and down criterion The signal is obtained during 200ms and then the application calculates the average during this time interval.After that, it compares if there has been an enough increment or decrement to detect if the user has moved the eyes intentionally.The EOG has been tested with 5 different users (3 female, 2 male) with ages between 23 and 30.They have been asked to spell 5 different words: WATER, FISH, SUN, DOG and MIND. ECG This section describes the ECG used to monitor the heart activity and to detect possible anomalies.In this case, the montage is monopolar, so we can obtain information from 4 different points (4 channels), establishing as a reference the user's shoulders.The signals the ECG records are smoothed before being monitored.After acquiring the signal (Figure 29), it is filtered with a Butterworth bandpass filter between 15Hz and 50Hz to detect the QRS complex (Figure 30). EOG We have asked the users to move a circle along a keyboard to spell five different words.We have divided the results into areas: the spelling of the whole word and the spelling of the single letters. It has been established that if the user expends more than three attempts to move the mark, the objective letter is considered as an error. Table 1 shows the obtained results for the different users: After establishing different thresholds, it is possible to select the wished letter controlling the movement with the eyes with an average of the 90.53%.The following two graphs are presented where the word FISH is correctly spelled (Figure 34) and wrongly spelled (Figure 35).It can be seen that an exception has happened with the EOG at the time the user was spelling fish. ECG The ECG allows the detection of the principal waves for the posterior analysis of anomalies.The ECG shown in Figure 37 presents a premature ventricular contraction after the T wave.The three graphs explained above correspond to three different users, showing the channels acquired for the 4 electrodes.The first one presents a normal ECG while the other two present two kinds of anomalies: premature ventricular contraction and arrhythmia. Comparing the normal ECG with the rest, it can be seen that there are differences between the form of the wave and the RR intervals.In the second graph, there is a beat which is advanced in respect of a normal frequency.Because of that, the second graph presents few numbers of RR intervals (6 beats in 10 seconds). Regarding to the third graph, the form of the waves is similar than the normal ECG, but the heart frequency is not constant: it can be appreciated that the heart frequency is not constant due to a little pause between the seconds 83 and 86. Conclusions As it has been demonstrated, the sensors used in BCI have an extra potential for applications which need high sensitivity. The signals recorded by an EEG, an EOG and an ECG are bioelectrical signals; that is why the techniques to obtain the electrical activity of the brain, the eye movements and the heart activity are very similar.In the three cases, it is necessary to place electrodes in order to measure the biologic potentials. The electrical activity of the eyes and the heart movement can be better appreciated than the activity from the brain.Because of that, it has been possible to use one EEG device as ECG and EOG as the resolution to distinguish the different signals is less than in an EEG.On the contrary, it could be not possible to use as EEG a device which has not been designed for that. The election of the type of electrodes that are needed to be used is very important depending on the signal to measure and the needed resolution.The chosen electrodes are good for the recording of an EOG and an ECG, but their resolution may not be sufficient in some EEG applications. Regarding to the EOG, the users have communicated the different letters with an average of 90.53%.With the appropriate training, the times of spelling could be reduced and it would be possible to increase the hit rate. As future work there could be three different lines: • Use the EEG for analyzing more bioelectrical signals, for example EMG. • Analyze the signals in real time in order to detect any possible anomaly. • Improve the results of the EOG adding new functions to the interface and check the most adequate trainings to increase the percentage of success. Figure 7 . Figure 7. Cornea -retinal potentialFigure 8 show the behavior of the eye as a dipole: Figure 9 . Figure 9. Purkinge imagesThe montage of the electrodes is bipolar: we register the difference between two electrodes.As the device is a 4 channel EEG, we have used them to perform 2 different positions: one for register if the user is looking on the left or on the right; and a second one to check if the user is looking up or down.The reference electrodes are placed in the ears. Figure 11 . Figure 11.User's interface of EOG Figure 13 . Figure 13.Difference between looking up and down Figure 14 . Figure 14.Difference between looking to the left and to the right Figure 15 . Figure 15.Atrial systole "QRS complex": the wave of depolarization arrives to the ventricles through the Bundle of His and the Purkinje fibers, producing the contraction of the ventricles. Figure 16 . Figure 16.QRS complex "ST segment": indicates the time between the end of the contraction of the ventricles and the beginning of the resting period. Figure 17 . Figure 17.ST segment "T wave": indicates the repolarization of the ventricles. Figure 18 . Figure 18.T wave Figure 19 is the representation of a normal ECG.There can be appreciated the different waves explained before. Figure 21 . Figure 21.Electrodes placement The user's interface has the aspect of Figure 22.The user can select the channels in both graphs. Figure 22 . Figure 22.User's interface for ECG Figure 25 . Figure 25.Left and right criterion Figure 26 and Figure 27 show the calibration state of the EOG before showing the keyboard.The first two images correspond to looing up (yellow rectangle) and down (green rectangle). Figure 26 . Figure 26.Looking up and down Below, the pictures show the user looking to the left (dark blue) or to the right (cyan). Figure 27 . Figure 27.Looking to the left and to the right Figure 28 . Figure 28.General diagram of ECG monitoring Figure 32 . Figure 32.Eye movements for the word DOG Figure 32 and Figure 33 present the different movements for the spelling of the words DOG and WATER respectively. Figure 33 . Figure 33.Eye movements for the word WATER Figure 37 . Figure 37. ECG for different channels with anomaly Figure 38 contains an ECG with arrhythmias: the cardiac frequency presents an alteration in the second 83 and the second 86. Figure 38 . Figure 38.ECG for different channels with arrhythmias Table 1 . Spelled words by users Medical Imaging in Clinical Practice Table 2 contains the average of each word by letter and by users.The number inside the parenthesis indicates the total number of letters of each word involving the five sessions (five different users). Table 2 . Average of each word	v3-fos-license
2021-05-21T16:56:37.960Z	2021-04-19T00:00:00.000	234816546	{ "extfieldsofstudy": [ "Business" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2071-1050/13/8/4517/pdf", "pdf_hash": "24f78c39248e3599c1b14e2499c1776e29df3eb3", "pdf_src": "ScienceParsePlus", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45441", "s2fieldsofstudy": [ "Medicine", "Political Science" ], "sha1": "7f0feb5252ada293c40823b45f5c73cef26a7ba7", "year": 2021 }	pes2o/s2orc	Integrated Healthcare and the Dilemma of Public Health Emergencies : Traditional healthcare services have demonstrated structural shortcomings in the delivery of patient care and enforced numerous elements of integration in the delivery of healthcare services. Integrated healthcare aims at providing all healthcare that makes humans healthy. However, with mainly chronically ill people and seniors, typically suffering from numerous comorbidities and diseases, being recruited for care, there is a need for a change in the healthcare service structure beyond direct-patient care to be compatible in peacetime and during public health emergencies. This article’s objective is to discuss the opportunities and obstacles for increasing the effectiveness of healthcare through improved integration. A rapid evidence review approach was used by performing a systematic followed by a non-systematic literature review and content analysis. The results conﬁrmed that integrated healthcare systems play an increasingly important role in healthcare system reforms undertaken in European Union countries. The essence of these changes is the transition from the episodic treatment of acute diseases to the provision of coordinated medical services, focused on chronic cases, prevention, and ensuring patient continuity. However, integrated healthcare, at a level not yet fully deﬁned, will be necessary if we are to both deﬁne and attain the integrated practice of both global health and global public health emergencies. This paper attains the necessary global challenges to integrate healthcare effectively at every level of society. There is a need for more knowledge to effectively develop, support, and disseminate initiatives related to coordinated healthcare in the individual healthcare systems. Introduction Although there might be national differences in defining Integrated Healthcare (IHC), one can simply describe it as efforts needed to provide all healthcare services that make humans healthy [1]. Consequently, such provision necessitates valid and functional delivery systems at all times. The World Health Organization (WHO) defines an integrated delivery system as, "the organization and management of health services so that people get the care they need, when they need it, in ways that are user-friendly, achieve the desired results, and provide value for money" [2]. These definitions work well in peacetime and with intact resources and capabilities. However, the last decades' economic strain on healthcare systems and an increasing number of public health emergencies highlight the insufficiency of delivering integrated healthcare, particularly to vulnerable groups of a society [3]. IHC is one of the important concepts related to the management and organization of healthcare systems. At present, efforts to integrate the process of providing healthcare, in addition to the integration of other areas of socio-economic life with the healthcare system, are a major concern for many international entities like WHO [2]. Attention to this concern is seen through a variety of projects that either directly or indirectly orient towards coordinating the provision of healthcare. By implementing these projects, healthcare systems aim to obtain the interrelated goals of improving the effectiveness of treatment, attribution of costs, and leveling inequalities in the access to and utilization of medical care [2]. In consideration of the concept of efficiency with these goals, there are three dimensions: market efficiency, cost-effectiveness, and social efficiency [4]. Market efficiency describes the maximization of health benefits to a population through the allocation of healthcare services that consider the cost incurred per activity. Cost-effectiveness describes the construction of a particular number of services at a minimum of cost. As such, costeffectiveness allows a more accurate assessment of the costs of a particular service [5] through an outcome analysis of a particular healthcare entity's service activities concerning the amount of cost incurred. In practice, efforts to improve efficiency aim to maximize the outcomes of service activities at specific input levels or by obtaining desired outcomes with minimal expenditure [6]. Social efficiency is associated with progress towards the implementation of social tasks, which benefit the healthcare system as seen in improving the combined health of societal groups or improving the quality and availability of healthcare at differing levels for a certain population. Market and social efficiency are macro-level constructs, which consider the ratio of costs to benefits in the healthcare sector with the broader national economy and population equity [7]. Efficiencies derived at the micro-level do not equate with efficiency at the macrolevel, particularly where a fragmented and non-equal healthcare industry is present. As such, healthcare expenditure is viewed in light of, and competing with, other expenditures at a national scale. This raises legitimate economic and social concerns on healthcare influences upon human health. This results in a problem of weighing the opportunity cost of various wants and needs, which can contribute to a fragmented healthcare system leading to the abandonment of frail patients. These are the same vulnerable victims of public health emergencies. Stakeholders and politicians [8][9][10] have long praised integration. It is supported by mutual respect and prospects and is facilitated by endurance [10,11]. However, any integration, especially among independent parties, may create conflicts [8][9][10][11][12][13][14], resulting in failing to deliver or delivering insufficiently, if pushed too far [10,12]. Within healthcare, integration is "an approach to strengthening patient-centered health systems through the promotion of the comprehensive delivery of quality services across the life-course designed according to the multidimensional needs of the population and the individual and delivered by a coordination of health-care providers working across settings and levels of care". Although this model ensures optimal outcomes and the appropriate use of resources based on the best available evidence, utilizes feedback loops to continually improve performance, tackles upstream causes of ill health, and promotes well-being through inter-sectoral and multi-sectoral actions, it may also create some difficulties, particularly times of crisis and public health emergencies and among underserved communities and vulnerable groups of a society [3,15]. The purpose of this study is to review the current concerns and shortcomings of IHC systems and to analyze the available opportunities and existing obstacles to improve their efficiency in delivery. Materials and Methods A Rapid Evidence Review approach was used by performing a systematic literature search, followed by a non-systematic literature review during January-March 2021. This approach summarizes the available state-of-the-art even if quantitative outcome data is unavailable in the literature [14]. The following search engines were used: Google Scholar, PubMed, Scopus, Science Direct, and University Library search engines in Poland and Sweden. Using the following keywords: "Integrated Healthcare" AND "Delivery" AND "Utilization" AND "Pros" AND "Improvement", a high number of hits was obtained. The search was then limited to the years 2010-2021 and for the publications in English. Two researchers (KG, and A K-M) conducted the search independently, and the outcomes were then matched. The included articles were chosen based on consensus between KG and A K-M, and if needed a third author was contacted. Included studies: Original publications and reviews from January 2010 to March 2021. Excluded studies: Proceedings, editorials, meeting notes, news, abstracts, and nonrelevant papers. Finally, a qualitative thematic analysis of the included literature based on an inductive approach was applied. This content analysis aimed to study all included articles, focusing on similarities and differences in the findings to present the tentative results [16]. Results The search started by using each keyword alone or in combination. The initial keyword, such as Integrated Healthcare or Integrated Health Systems, returned too many hits, which gradually were filtered by adding new keywords and limiting the search period. Table 1 shows the reduction of hits after adding new keywords ( Table 1). The obtained hits from all search engines were selected according to the study's inclusion and exclusion criteria to achieve a feasible number of articles for this review ( Figure 1). The thematic results were then presented in sections and topics (Section 3.1, Section 3.2). Table 1. The results of initial filtering of the hits in each search engines. Healthcare Integration between Different Levels According to the World Health Organization, IHC is a combination of inputs, delivery, management, and the organization of services associated with diagnosis, treatment, care, rehabilitation, and health promotion. Integrated care aims to improve the access, quality, user satisfaction, and efficiency of healthcare services [17]. Others define integrated care as a worldwide trend in healthcare reforms and new organizational arrangements focusing on more coordinated and integrated forms of care provision [18]. The system of integrated provision of health services requires integration at three levels: organizational, clinical, and individual [19]. However, these levels, divided into subgroups, are interrelated and have mutual factors, which may affect the outcomes of IHC. Healthcare Integration between Different Levels According to the World Health Organization, IHC is a combination of inputs, delivery, management, and the organization of services associated with diagnosis, treatment, care, rehabilitation, and health promotion. Integrated care aims to improve the access, quality, user satisfaction, and efficiency of healthcare services [17]. Others define integrated care as a worldwide trend in healthcare reforms and new organizational arrangements focusing on more coordinated and integrated forms of care provision [18]. The system of integrated provision of health services requires integration at three levels: organizational, clinical, and individual [19]. However, these levels, divided into subgroups, are interrelated and have mutual factors, which may affect the outcomes of IHC. Organizational Integration Integration can occur vertically or horizontally. By using both vertical and horizontal directions of integration, an integrated service delivery network can create a system of Organizational Integration Integration can occur vertically or horizontally. By using both vertical and horizontal directions of integration, an integrated service delivery network can create a system of cooperating programs, entities, or health plans at the equivalent level of care and between its various levels [20]. While the influence of tasks enacted by this type of system mainly affects the economic and clinical levels, integration might also benefit patient satisfaction by improving relationships with patients through better continuity of treatment and safeguarding a sense of health security [21]. This multi-level system offers opportunities to deliver high-quality care to patients if working appropriately at all levels, otherwise, may generate more suffering. While the approach to healthcare integration and coordination strategies might vary by country, the process of transformation is similar. Key factors on the supply side include potential economic pressure on the cost management of healthcare systems and the development of medical technologies and IT systems [22]. Considerations on the demand side include demographic and epidemiological changes in society, patient empowerment in the form of stronger voices for their needs and preferences, and organized consumer movements. The Aging Patient The growing life expectancy rate and the growth of the segment of people in the 65+ group undoubtedly affect the volume and structure of the demand for healthcare. Over the past 30 years alone, the percentage of the population aged 65+, calculated as the European Union (EU) average, increased from 13% in 1980 to around 20% in 2019 [23,24]. The trend toward higher life expectancy alone does not allow for accurately estimating the demand for health services. Elements, such as individual lifestyle, environment, and genetic risk factors must also be considered in calculations and weighed alongside the emergence of health problems that are known to arise with old age and new diseases [25][26][27]. This explains why the demand for health services in an aging society, especially in Europe, is described by functional, psychological, and social consequences of chronic and multi-organ diseases. The most appropriate answer to dealing with these problems synchronously is IHC, which primarily refers to the holistic approach to human health [28]. Disease Panorama Treatment of increasingly common ascending chronic diseases will require moving away from the fragmentation of healthcare and focusing on so-called acute diseases, and short-term hospitals adapted to combat the management of the patient's disease unit together with a social and environmental support system. A transition to a patient-centered healthcare model from existing disease-, technology-, physician-, and hospital-centric models is essential [29,30]. A consequence of patient-centered healthcare is that changes occur to both the supply and demand side of the health system. On the supply side, the distribution of financing health services is based on need and solidarity and on the demand side, patients take increased responsibility for their health, in addition to the financial implications. A fundamental issue that will need to be resolved is the sharing of risk between all parties involved in the provider-payer-patient relationship [31]. Another opportunity results from patients having greater access to medical information and globalization, which results in higher expectations from patients than what the current healthcare delivery system in most countries can support [32]. The increasing demand from patients for the availability of healthcare, continuity of care, and comprehensiveness translates to the increased responsibility of the healthcare institutions to provide greater management, health promotion, and extensive rehabilitation. Meanwhile, the rise of patient empowerment is linked to improved disease self-management [32][33][34][35] and a desire to take control of one's health through lifestyle change and establishing a healthy environment [36]. The Advancing Technology Ongoing advances in information and communication technologies (ICT) are a factor on the supply side of healthcare that is thought to have the most significant impact on progress towards integration [37]. ICT, by facilitating information flows between treatment participants at differing levels, improves the quality of care. These technologies also enable the interested system stakeholders to monitor the effectiveness of the care process through feeding back relevant metrics and data. This feedback enables vertical integration of organization and information to support the processes involved in providing medical services [38]. At the meso level, the integration of professionals and institutions into interdisciplinary teams enables sharing of experience and knowledge and is facilitated by ICT [39]. This represents a form of horizontal integration in terms of sharing know-how. For the patient, the advantages of ICT are an increase in the speed of diagnosis and that treatment can be started remotely without in-person attendance [40]. Telemedicine, a popular form of ICT, increases the availability of services for patients living outside large specialist centers, enhancing support for social efficiency. When considering the macro level, the influx of epidemiological data and results provided by ICT enables more effective planning for the allocation of healthcare services. Furthermore, the costs of obtaining medical research data and evidence are reduced, hence increasing the evidence base for the provision of care based upon scientific research [41,42]. Over time, the adoption of new medical technology will force an adjustment in the existing structures and forms of healthcare service delivery. Additional opportunities for integration are supported where ICT can, for example, cut down queues in busy walk-in centers in populated areas or prehospital triage, and provide decision support tools for advisory call centers. Community Commitments: Supply and Services Increasing outpatient and community level care service provision, including home care, can reduce patients' length of stay (LOS) and lower admissions in hospitals [43]. More innovative approaches such as gene therapy will necessitate completely new forms of healthcare delivery and considerations for novel organizational and logistical solutions. However, new technologies like gene therapy may prove more cost-effective than current ones as coordination between existing outpatient and specialized healthcare levels is a challenge [44]. On the supply side, one must consider the phenomenon of demand induced by supply, a well-known macroeconomic problem [45]. This leads to an increase in expenditure. For example, if we were able to shorten the LOS for a hospital in-patient through new technology, we may indirectly induce an increase in the demand for beds. This would in turn lead to new hospital admissions and potential misuse of resources where patient admissions are shifted to other parts of the hospital leading to worsening patient safety and increased risk for spreading infectious disease. However, social efficiency is increasing due to the increased service capacity at various levels of healthcare, improving the access to services [46]. Improving efficiency at the macro scale is possible through system harmonization and consolidation of links between individual levels of care. In light of the progressing specialization and fragmentation of medicine, this goal may be difficult to achieve [47]. IHC Model Across International Healthcare Systems The scope of competence and tasks of a general physician (GP) and the care carried out is characterized by certain features, which are different from the specifics of the work of specialists in narrow fields of medicine [48,49]. Specialists focus more on a targeted organ such as the heart or kidney. The family doctor provides comprehensive care for the patient, family, and the local community. Characteristic features of GP activity include, among others: comprehensiveness, continuity, patient orientation, and coordination [50]. GPs may play a significant role in the execution of IHC, however, overcrowded Emergency Departments in most western countries indicate that the GP's position as an integrating actor is still undeveloped. In Europe, Scandinavian countries have been successful in advancing IHC delivery throughout the national health system [51,52]. In other countries, such as Poland, healthcare providers have also succeeded in introducing integrated care solutions [53][54][55]. However, the introduction of integrated care differs between countries in terms of the scope and type of solutions used, which may be systemic, aimed at reducing barriers between sectors of the healthcare system or improving cooperation between medical professionals. This can be seen in Scandinavian countries, with almost similar cultural and behavioral backgrounds. While Norway approaches IHC by individual plans and legislation, with a strong local character, the Danish approach is conducted in a top-down manner and a traditional bureaucratic culture. Sweden, on the other hand, practices IHC as a voluntary collaborative task in a locally initiated and conducted development work to deliver an improved health service to the patients [56]. In other EU countries, initiatives integrating health and social care have been implemented by creating housing estates adapted to the lives of the chronically ill [57][58][59][60][61][62][63][64]. A review by Nolte and Knai indicated the existence of numerous chronic disease management initiatives across 12 European countries [65]. Although the majority of approaches focused on populations with defined conditions (e.g., diabetes, asthma, cardiovascular disease, cancer), they often involve elements of integration between providers and sectors. It could involve the implementation of multidisciplinary teams and care coordination among different professions as well as building providers' networks. In a recent study, integrated care models were systematically reviewed and compared between the UK and several other countries. The result indicated that higher patient satisfaction, increased perceived quality of care, and increased patient access to services are the three most frequent indicators of care integration [66]. The same results were also found in another review, comparing six Asian countries, including Singapore, China, India, Vietnam, etc. [67]. In developed countries, coordination of healthcare reflects the cooperation between primary healthcare and other healthcare systems [65,68]. It assures the optimization of medical decisions from a financial perspective. From the patient's point of view, a physician's action consists of medical counselling and acting as a guide to a complex healthcare system [69]. From the system administrator's point of view, it is perceived as an element of access control to cost-intensive specialists and hospital services [70]. The concept of coordination is also extended to activities that go beyond healthcare, perceived as cross-sectoral cooperation, which includes the system of social care, education, environmental protection, and the workplace [71]. Nearness to Death (NTD) Studies show that some 55% of a person's entire lifetime hospital bed use occurs in the last year of life, and thus disproves the claim that population growth and ageing are the causes of steady healthcare demands [72]. Considering that this period of life is associated with higher costs, in terms of investigations and bed occupancy, it is logical to believe that major changes need to occur in terms of end-of-life costs. NTD should thus be considered in capacity planning models to provide realistic estimates of future bed demands [73]. New studies in the UK show that the demand for medical beds is likely to increase by 39% until 2060 [74]. Planning is needed for both adults and pediatric patients [75]. Supporting legislation is an absolute necessity for successful integrated healthcare to not only guide the care of patients in peacetime and emergencies but also for the implication of nearness to death as the patients must at some point be diverted from further aggressive care into more palliative alternatives [76][77][78]. Discussion, Challenges, and Recommendations Although IHC provides high quality of care from national and global perspectives and is both praised and supported by stakeholders and politicians, it must function in harmony and mutual respect and prospects for all involved parties. The imbalance between needs and demand in underserved communities and during national and international emergencies may influence its efficiency and create some difficulties, resulting in insufficient or failed delivery [8][9][10][11][12][13][14][15]. Within healthcare, a chief concern is the fragmentation between public and private healthcare models within a common context. In some countries, a mix of Beveridge, Bismarck, and out-of-pocket healthcare systems is practiced side by side. Such a mix may provide freedom of choice but may also hinder equity because of difficulties to transfer patients between systems [54,[58][59][60][61][62][63][64]. Such an imbalance and health inequity would be more during crises. Digitalization may be a tool to improve transferability, but an optimized flexible system of resource allocation is a necessary remedy to limit fragmentation. To prevent frag-mentation, all steps in the digitalization process of health services have to place the patients first and the process itself has to help the patients through the procedures and services available at their disposal. However, a major challenge for modern medicine is an ageing society, which generates increasing medical costs. According to a United Nations (UN) report, the ageing of society is an unprecedented demographic trend determining significant changes in all areas of life by at least 2050 [79], emphasizing why the social support system for the elderly as part of policymaking for seniors is becoming an enormous challenge. There is a need to develop a long-term healthcare strategy. This requires systematic measurement, and comparison of health outcomes, nationally and globally, to improve the outcomes. Maps of health needs are undoubtedly a tool enabling such measurements and analysis of health data, whereas, according to many authors, the essential tool supporting the planning and coordination of healthcare, used for communication with the patient, is electronic medical documentation [42][43][44][45][46]. This documentation is also necessary for the effective use of telemedicine and e-health mobile solutions in health service delivery [80][81][82][83]. However, digitalization may create some difficulties in extreme situations and during disasters in obtaining the necessary authorization to access patients' data and should be considered in the early planning phases. Applying the Pareto principle, modern healthcare systems typically see 80% of their expenditures consumed by 20% of patients [83,84]. Driving this trend is the recruitment of mainly seniors and chronically ill people who suffer from numerous co-morbidities and diseases. As such, a need for change in the overlying structure of health services may emerge. A further challenge, particularly for GPs is to consider patient priorities and values including the need to involve patients in joint decision-making so that the best solutions can be planned regarding prevention, diagnosis, and therapy. Quality improvement to care for chronically ill people is also called for by numerous reports which highlight pressures to keep expenses down and to manage these patients more efficiently [85,86]. Several organizational and individual factors affect the outcomes of IHC. These factors, not only play a significant role in daily healthcare delivery to underserved communities but also enhance the negative outcomes of public health emergencies. The world has been successful in globalizing communications and travel through integrated policies and practices not seen before. Whereas traditional healthcare practices can manage infectious disease outbreaks and epidemics, it has failed in the current Coronavirus Disease 2019 pandemic, which requires several elements of integration beyond current direct-patient care [87,88]. Worse, in many countries, especially the United States, the pandemic destroyed the one highly IHC system, that of public health, forcing it to acquiesce to economic and political demands. Integrated healthcare, at a level not yet fully defined, will be necessary if we are to attain the goal of an integrated practice of both global health and global public health [89][90][91][92]. To overcome these global challenges, we must first effectively integrate healthcare at every level of society. Limitations This article does not discuss all initiatives and solutions to improve the effectiveness of IHC, rather we have focused on identifying the solutions that seem to be the most beneficial due to a large number of policy proposals, especially in domestic markets. Another limitation to this study is its dependency on English literature. Other definitions or discussions about IHC in other languages were not searched for and thus some important data might be missing. Finally, although the literature selection was conducted independently by two authors, according to a defined topic and consensus, there might still be some bias and some of the excluded studies might have had an impact on the final discussion and conclusion. Conclusions Healthcare systems in many European countries have been experiencing severe difficulties for many years, mainly due to inadequate funding of the healthcare system [93]. These problems manifest themselves by demanding access to specialist care and long queues awaiting benefits, the unsatisfactory quality of health services, low wages in the healthcare sector, mass emigration of specialist medical staff to wealthier EU countries, and the indebtedness of public healthcare facilities [94]. Coordinated global healthcare is an essential and effective strategy to achieve the needed reform in health systems across Europe. It can take many different forms and can be implemented using a variety of organizational and financial models [95]. However, its implementation in individual systems requires the integration of health and social policy and cooperation between the market regulator, the payer, and medical service providers, as well as the patient's involvement [6][7][8][9][10][11][12][13]. Coordinated healthcare is necessary for the patient, the service provider, the payer himself, and the governmental institutions. These policies might also need to be compatible in both peacetimes and during public health emergencies. Such compatibility needs educational initiatives [96] in all levels of disaster management to be implemented during a period and resource scarcity. More knowledge is needed to effectively develop, support, and disseminate initiatives related to coordinated healthcare in individual healthcare systems of different countries and during different scenarios [97][98][99][100][101][102]. It is equally important to pay attention to initiatives, which will allow the creation of better and enhanced technical and organizational solutions, as well as change the behavior of patients and healthcare providers. 1. The most crucial element to achieving integrated healthcare for use in both peacetimes and public health emergencies is interagency or interdisciplinary collaboration. 2. Community support and resources are significant factors to unburdening emergency healthcare. Mutual planning and considerations between various healthcare branches for the care of chronically ill patients and those in terminal disease are mandatory. 3. Nearness to death should be included as one important factor in future healthcare planning and integrated healthcare. 4. Technological advances should be used based on evidence and according to the prognostic outcome of patients.	v3-fos-license
2023-10-27T15:33:21.273Z	2023-10-25T00:00:00.000	264498765	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2072-6651/15/11/625/pdf?version=1698195413", "pdf_hash": "517496080266051f23f2ff260ede0f8ce485a54a", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45443", "s2fieldsofstudy": [ "Medicine", "Biology" ], "sha1": "9fa93ca32153b0d3ffc750dfddb46cbf61107072", "year": 2023 }	pes2o/s2orc	Exosome Liberation by Human Neutrophils under L-Amino Acid Oxidase of Calloselasma rhodostoma Venom Action L-Amino acid oxidase (LAAO) is an enzyme found in snake venom that has multifaceted effects, including the generation of hydrogen peroxide (H2O2) during oxidative reactions, leading to various biological and pharmacological outcomes such as apoptosis, cytotoxicity, modulation of platelet aggregation, hemorrhage, and neutrophil activation. Human neutrophils respond to LAAO by enhancing chemotaxis, and phagocytosis, and releasing reactive oxygen species (ROS) and pro-inflammatory mediators. Exosomes cellular nanovesicles play vital roles in intercellular communication, including immune responses. This study investigates the impact of Calloselasma rhodostoma snake venom-derived LAAO (Cr-LAAO) on human neutrophil exosome release, including activation patterns, exosome formation, and content. Neutrophils isolated from healthy donors were stimulated with Cr-LAAO (100 μg/mL) for 3 h, followed by exosome isolation and analysis. Results show that Cr-LAAO induces the release of exosomes with distinct protein content compared to the negative control. Proteomic analysis reveals proteins related to the regulation of immune responses and blood coagulation. This study uncovers Cr-LAAO’s ability to activate human neutrophils, leading to exosome release and facilitating intercellular communication, offering insights into potential therapeutic approaches for inflammatory and immunological disorders. Introduction L-Amino acid oxidase (LAAO) is an enzyme widely distributed in several species, such as insects, fungi, plants, and snake venoms [1][2][3][4][5].The mechanism of action of LAAO is not yet fully elucidated but involves oxidative effects on various pathogens, such as fungi and viruses, as well as on various cell lines.Some of the observed effects are attributed to the production of hydrogen peroxide (H 2 O 2 ) during oxidative reactions [6,7]. While leukocytes are implicated in several inflammatory pathologies, little is known about the state of activation of these cells during Cr-LAAO action.Recent research indicates that extracellular vesicles may be used by cells to transfer inflammatory mediators and receptors [25].Majumdar et al. [26] demonstrated that these vesicles carry mediators such as LTB 4 , acting in an autocrine manner, sensitizing neutrophils, and amplifying the inflammatory response.Additionally, it was shown that these vesicles acted in a paracrine manner, promoting cell communication [27]. Exosomes, nanovesicles secreted by cells into the extracellular space, have an endocytic origin, and their size ranges from 40 to 100 nm in diameter [28].Their formation occurs through the production of endocytic vesicles from the plasma membrane, called primitive endosomes.Subsequently, they become late endosomes, or multivesicular bodies (MVB), containing numerous intraluminal vesicles, which can be degraded by lysosomes or fuse with the cell plasma membrane, releasing their contents into the extracellular medium [29]. Through the interaction of exosomal membrane proteins with the target cell via juxtacrine communication, vesicles can initiate cell signaling events, activating the target cell [30].Exosomes also hold potential as a novel therapeutic technique since they offer a method for delivering materials such as medicines, proteins, and siRNAs to specific cells or tissues through laboratory manipulation [31,32]. The ability to use exosomes as biomarkers, released under both normal and pathological conditions, has been demonstrated by the literature in various bodily fluids, including blood, tears, saliva, urine, and cerebrospinal fluid [33,34].Recently, the importance of studying the intravesicular components of exosomes, such as proteins, mRNA, and miRNA, has been highlighted, as they are implicated in the progression of viral pathologies and their prognosis [35][36][37][38].Moreover, its components can influence immuno-regulatory processes, such as tumor escape mechanisms, and regenerative and degenerative processes [39]. These vesicles are composed mainly of lipids and membrane proteins, similar to those present in the cell from which they originate.Furthermore, they may contain proteins and mRNA transcripts derived from the intracytoplasmic environment [40].Their ability to Toxins 2023, 15, 625 3 of 20 interact with different cell types makes them an important mechanism for intercellular communication in the body.Furthermore, their role in the vascular environment depends on the cellular origin and the type of stimulus used to promote vesiculation, giving them specialized functions in their cellular targets [41,42]. The literature shows that exosomes are involved in cell-cell communication and can carry molecules such as MHC-II (linked to antigenic peptides), amplifying and modulating the immune response [43][44][45][46][47]. Additionally, exosomes are found in biological fluids such as plasma, bronchoalveolar lavage, breast milk, and saliva [48].However, due to their small size and heterogeneity, detecting and classifying them remains a challenge [25].Based on these considerations, studying the activation, formation, and release pattern of human neutrophil exosomes induced by Cr-LAAO may provide insights into their role in the inflammatory response and better characterize them for future clinical applications. Results The electrophoretic profile of exosome samples from RPMI (negative group), PMA (positive control group), and Cr-LAAO (experimental group) has revealed the presence of proteins at approximately 60 kDa under all conditions (Figure 1A).The size of the exosome vesicles present in the extracellular medium particles ranges from 8 to 70 nm for the negative control group, 6 to 50 nm for the positive control group, and 8 to 70 nm for the Cr-LAAO group (Figure 1B-D).Immunoreactivity of antibodies to CD63 and CD81 was detected in the samples of exosomes from neutrophils stimulated with Cr-LAAO, as well as in the negative control.However, cells stimulated with PMA did not exhibit immunoreactivity (Figure 1E).CD63 is a marker for extracellular vesicles in general, while CD81 is a specific marker for exosomes [49,50]. The proteomic analysis identified a total of 722 protein hits in all experimental groups.According to the maximum parsimony analysis, this study identified 330 proteins among the RPMI (negative control group), PMA (positive control group), and Cr-LAAO (experimental group).Detailed parameters for the identification of these proteins are provided in Supplementary Material S1. Protein expression analyses revealed 206 common proteins among the three experimental groups.Specifically, 41 proteins were common between the control and Cr-LAAO groups, 81 proteins were exclusive to the control group compared to the Cr-LAAO group, and 28 proteins were exclusive to the Cr-LAAO group compared to the control group.Additionally, 9 proteins were common between the Cr-LAAO and PMA groups, 38 proteins were unique to the Cr-LAAO group compared to the PMA group, and 18 proteins were unique to the PMA group compared to the LAAO group.Finally, 30 proteins were common between the control and PMA groups, 89 proteins were unique to the control group compared to the PMA group, and 9 proteins were exclusive to the PMA group compared to the control group (Supplementary Materials S1 and S2). The Volcano Plot indicated proteins that were differentially expressed in Cr-LAAOstimulated neutrophils compared to the control group.These proteins exhibited a fold change of ≥2 and a T-test analysis p-value < 0.05 (Figure 2A).Spectral count data for proteins from both groups were subjected to PLS-DA analysis, and the plot scores demonstrated a clear separation of the two groups and their respective biological replicates.This separation indicated high reproducibility among the replicates and a significant differential effect in the presence of Cr-LAAO (Figure 2B).In Figure 2C, the results of the VIP score analysis are displayed, highlighting the most important proteins associated with Cr-LAAO's impact on neutrophils.Proteins with a VIP score ≥ 1.7 were considered statistically significant.Table 1 presents the up-regulated and down-regulated proteins in Cr-LAAO-stimulated neutrophils obtained through Volcano Plot Analysis.and Cr-LAAO (D) groups after ultracentrifugation using dynamic light scattering (DLS) technique employed to determine the nuclear dimensions of the nanoparticles using the Zetasizer Nano ZS system (Mallvern ® Instrument, Malvern-United Kingdom).(E) Western blot for CD81 and CD63 using the supernatant of neutrophils (1 × 10 7 ) stimulated with PMA (500 ng/mL) for the positive control (C+), RPMI for the negative control (C−), and Cr-LAAO (100 µg/mL) after 3 h of incubation in a humidified atmosphere of CO2 (5%) at 37 °C. The proteomic analysis identified a total of 722 protein hits in all experimental groups.According to the maximum parsimony analysis, this study identified 330 proteins among the RPMI (negative control group), PMA (positive control group), and Cr-LAAO (experimental group).Detailed parameters for the identification of these proteins are provided in Supplementary Material S1. Protein expression analyses revealed 206 common proteins among the three experimental groups.Specifically, 41 proteins were common between the control and Cr-LAAO groups, 81 proteins were exclusive to the control group compared to the Cr-LAAO group, and 28 proteins were exclusive to the Cr-LAAO group compared to the control group.Additionally, 9 proteins were common between the Cr-LAAO and PMA groups, 38 proteins were unique to the Cr-LAAO group compared to the PMA group, and 18 To gain a better understanding of the biological processes, molecular functions, and cellular components associated with the differentially expressed proteins in exosomes from Cr-LAAO-stimulated neutrophils, gene ontology analysis was conducted.Among the proteins corresponding to biological processes, 16 proteins (27% of them) were involved in cellular processes; 9 (15%) in metabolic processes; 7 (12%) in localization activities; 6 (10%) in response to stimuli; 5 (8%) in biological regulation; 4 (7%) in the regulation of biological processes; 4 (7%) in immune system processes; 3 (5%) in the organization of cellular components or biogenesis; 2 (3%) in the upregulation of biological processes; 2 (3%) in developmental processes; and 2 (3%) in multicellular organizational processes (Figure 3A).Regarding the proteins involved in molecular functions (Figure 3B), it was found that 33 proteins (48%) were related to binding functions, 24 (35%) to catalytic activity, 7 (10%) to structural functions, and 5 (7%) to transport activation functions.In the cellular components category (Figure 3C), it was observed that 24 proteins (44%) were associated with cellular anatomical entities, 20 (36%) with intracellular regions, and 11 (20%) with components of protein complexes.Figure 4 is a graphical representation of information related to proteinprotein interactions.These interactions involve proteins that were expressed differently in neutrophils after being treated with Cr-LAAO. Discussion In the present study, to identify the presence of exosomes in the extracellular medium, isolation was standardized by filtration through 0.45 and 0.22 µM nylon membranes, followed by ultracentrifugation.The precipitate was collected, and subsequently, a Western blot was performed using antibodies against surface protein markers (CD63 and CD81) present in these vesicles.Exosome isolation derived from stimulated neutrophils under different experimental conditions (control, PMA, and Cr-LAAO) was successfully obtained, consistent with the results published by Lasser et al. [48] and Majumdar et al. [26].By stimulating neutrophils isolated from peripheral blood with fMLP, Majumdar et al. [26] observed bands corresponding to the intravesicular components of myeloperoxidase, metallopeptidase 9, and 5-lipoxygenase, as well as the surface markers found on exosomes. Simultaneously, the size of the exosomes present in the extracellular medium was determined using dynamic light scattering (D.L.S) with a ZetaSizer, which measures particle size.In this study, particles ranging from 10 to 100 nm were obtained, demonstrating exosome heterogeneity.Yamashita et al. [51] identified a similar size and Discussion In the present study, to identify the presence of exosomes in the extracellular medium, isolation was standardized by filtration through 0.45 and 0.22 µM nylon membranes, followed by ultracentrifugation.The precipitate was collected, and subsequently, a Western blot was performed using antibodies against surface protein markers (CD63 and CD81) present in these vesicles.Exosome isolation derived from stimulated neutrophils under different experimental conditions (control, PMA, and Cr-LAAO) was successfully obtained, consistent with the results published by Lasser et al. [48] and Majumdar et al. [26].By stimulating neutrophils isolated from peripheral blood with fMLP, Majumdar et al. [26] observed bands corresponding to the intravesicular components of myeloperoxidase, metallopeptidase 9, and 5-lipoxygenase, as well as the surface markers found on exosomes. Simultaneously, the size of the exosomes present in the extracellular medium was determined using dynamic light scattering (D.L.S) with a ZetaSizer, which measures particle size.In this study, particles ranging from 10 to 100 nm were obtained, demonstrating exosome heterogeneity.Yamashita et al. [51] identified a similar size and distribution pattern of exosomes in cell supernatant samples from murine melanoma after ultracentrifugation. The Western blot assay confirmed that the vesicles were indeed exosomes, as they exhibited the characteristic tetraspanins, including CD63, on their cell surface.Tetraspanins contain four transmembrane domains and are responsible for forming networks with a wide variety of proteins, creating tetraspanin-enriched microdomains (TEMs) [52,53].Tetraspanin molecules can associate with various adhesion molecules such as β1 integrin and receptors, performing functions like cell proliferation, adhesion, activation, and migration.CD63, the first tetraspanin characterized, was found to be present in all samples in this study. Furthermore, CD61, known as β3 integrin, is also a cell surface protein belonging to the integrin family.Its relationship with exosomes is linked to the role played by β3 integrin in exosome biogenesis and release.Studies have shown that β3 integrin (CD61) is involved in forming integrin complexes on exosome membranes.These complexes may play a role in selecting and incorporating specific proteins into exosome vesicles during their formation.Additionally, the presence of CD61 on exosome membranes may influence their interaction with target cells through the recognition of specific ligands [54,55].Data from the present study indicate that CD61 was expressed in exosomes from Cr-LAAOstimulated neutrophils and the control groups, but there was no expression in the PMAstimulated neutrophils group.The differences in cell activation induced by PMA and Cr-LAAO may occur through different mechanisms that trigger distinct cellular responses.These differences in activation mechanisms can be attributed to several factors, including variations in intracellular signaling pathways.These pathways have different targets and effects within the cell, resulting in diverse responses [56,57]. Regarding the protein-protein interactions among proteins differentially expressed in Cr-LAAO-stimulated neutrophils, several interesting associations were identified.Among these, proteins from the α and β tubulin family were observed.Deficiency in these proteins or dysfunction of microtubules can lead to various human conditions and disorders, including neurodegenerative diseases such as Alzheimer's, Parkinson's, and Huntington's disease [58], as well as breast cancer [59] and pancreatic cancer [60]. Two immune system proteins, myeloperoxidase (MPO) and complement 3 (C3), were also detected.These proteins play vital roles in the inflammatory response and the body's defense against infections.Studies have suggested that MPO may influence complement activity [61,62].MPO can modify C3, leading to its activation and the release of C3 fragments, which in turn can trigger inflammatory responses and attract immune cells to the site of infection [63,64]. Myosin and actin are the primary contractile proteins involved in muscle contraction.Myosin also plays a crucial structural role in maintaining the organization of these proteins in the sarcomere.Disruption in the function of these proteins can result in diseases such as cardiomyopathy and muscle paralysis [65][66][67].It is important to note that these proteins have specific functions and interactions in various cellular contexts and physiological processes.There is no direct interaction between all of them in a single system or cellular event.Each of these proteins has its specific functions and roles in different biological systems and processes. Additionally, the proteomic data from exosomes derived from Cr-LAAO-stimulated neutrophils, in comparison to the control, revealed the presence of several other proteins, including: [86,87]. Among the data, proteins related to exosome composition and involved in cell signaling were also detected, including the 14-3-3 proteins (zeta, epsilon, beta, eta, gamma, and theta), which are encoded by the genes YWHAZ, YWHAE, YWHAB, YWHAH, YWHAG, and YWHAQ, respectively.These regulatory proteins are primarily found in the cytosol and the extracellular environment.They have the capability to interact with over 200 signaling proteins, including kinases, phosphatases, transmembrane receptors, enzymes, structural and cytoskeletal proteins, proteins involved in transcriptional control, and proteins associated with apoptosis.Moreover, they have demonstrated effects on the inflammatory process and are involved in the regulation of cell proliferation, differentiation, and metabolism [91][92][93][94][95]. Peroxiredoxin-2, encoded by the PRDX2 gene, plays a role in cellular protection against oxidative stress by detoxifying peroxides.It also functions as a sensor of hydrogen peroxidemediated signaling events and may participate in signaling cascades involving growth factors and tumor necrosis factor-alpha (TNF-α), regulating intracellular concentrations of H 2 O 2 [96,97] which have a role in neutrophil activation induced by Cr-LAAO [17][18][19][20][21]. Histone H2B and H3.3 histone B, encoded by the genes H2BC15 and H3-3B in humans, are involved in the structure of eukaryotic cell chromatin.They play a role in the process of compacting and decompressing DNA, which is important for the formation of NETs [98,99] as observed in neutrophils stimulated by Cr-LAAO [17]. L-lactate dehydrogenase A chain and L-lactate dehydrogenase B chain, encoded by the LDHA, LDHB, and LDH genes, are primarily responsible for converting pyruvate into lactate and transforming NADH into NAD+.These proteins are related to the pyroptosis process [100][101][102][103][104]. Arginase 1, encoded by the ARG1 gene, is a key component of the urea cycle and induces a profound suppression of T cell proliferation and cytokine synthesis by neutrophils [105][106][107]. Table 2 displays the proteins identified in the proteomics data when comparing neutrophils stimulated by Cr-LAAO with those stimulated by PMA. Involved in the pentose phosphate pathway and plays an important role in the generation of NADPH. Nebulin Acts mainly in the regulation of muscle structure and function, it is crucial for the proper functioning of skeletal muscles.[123,124] Leukotriene A 4 hydrolase Plays a central role in the synthesis of leukotrienes, which are an important inflammatory mediator for immune response.[125,126] Vitamin D binding protein Plays a crucial role in the transport and regulation of vitamin D in the body.[127] Cathelicidin antimicrobial peptide Plays a key role in host defense against microbial infections, including bacteria, viruses, fungi, and even parasites.[128,129] Fibronectin Plays multiple roles in cell adhesion, migration, wound healing, embryonic development, and immune response, produced by various cells, including fibroblasts, hepatocytes, endothelial cells, and muscle cells.[128,130] Proteins related to the composition of exosomes and involved in cell signaling were also detected, such as annexin A2, encoded by the ANXA2 gene in humans.This protein can bind to cell membranes in a calcium-dependent manner and is involved in a variety of cellular functions, including signaling processes, cytoskeletal regulation, and interactions with lipids [131,132]. Rab GDP dissociation beta inhibitor protein, encoded by the GDI2 gene in humans, plays a key role in the GTPase cycle of Rab proteins.It helps maintain Rab in its inactive GDP-bound state, contributing to the precise regulation of intracellular vesicular trafficking and membrane transport.This regulation is essential for normal cell function and for maintaining the structure and function of cell compartments [133,134]. Coronin-1A, encoded by the CORO1A gene in humans, plays a key role in regulating the actin cytoskeleton.It influences cell migration, phagocytosis, cell shape, and signaling pathways in different cell types, especially in cells of the immune system [135,136] which were observed in neutrophils stimulated by Cr-LAAO [17][18][19][20][21]. Talin-1, encoded by the TLN1 gene in humans, is a protein involved in the formation of focal adhesions, which are points of contact between the cell and the extracellular matrix.These adhesions allow the transmission of mechanical signals and regulate cell adhesion and migration [137][138][139]. In conclusion, the data from the present study revealed that Cr-LAAO activates human neutrophils in vitro, stimulating the production of extracellular vesicles called exosomes, which may play an important role in cell-to-cell communication and the regulation of inflammatory processes.Proteomic analysis revealed the presence of several proteins involved in fundamental cellular processes, such as the regulation of the immune response, muscle contraction, and blood clotting.These findings highlight the importance of investigating the role of exosomes and differentially expressed proteins in Cr-LAAO-activated neutrophils, both in the physiological and pathological context.Understanding these mechanisms can provide valuable insights for the development of new therapeutic strategies and biomarkers in inflammatory and immunological diseases. Cr-LAAO Isolation The Cr-LAAO isolation was performed following the protocol published by Pontes et al. [17] and Paloschi et al. [20].The presence of endotoxin in Cr-LAAO samples was determined using the Quant kit derived from Pierce™ Chromogenic Endotoxin.The Cr-LAAO preparation exhibited the presence of 0.1 EU/mL of endotoxin, which was within the acceptable threshold of 1 EU/mL [21]. Isolation and Activation of Neutrophils Neutrophils were isolated from peripheral blood obtained from healthy self-donors aged 18 to 40 years.Donors provided informed consent before blood collection, and all procedures were conducted per applicable regulations.The study was approved by the Ethical Committee of the Center of Tropical Medicine Research (CEPEM, Rondônia, Brazilapproval number CAAE 77529817.8.0000.0011),and participants provided informed consent before participating in the study according to the method described by Pontes et al. [17] and Paloschi et al. [21]. Exosomes Isolation For exosome isolation, 1 × 10 7 human neutrophils were obtained from three donors (3 Ns) and were incubated for 3 h at 37 • C, in a humidified atmosphere with 5% CO 2 with 100 µg/mL of Cr-LAAO (experimental group), 500 ng/mL of PMA (positive control) and RPMI without supplementation (negative control) in a final volume of 400 µL.After the end of the incubation, the samples were centrifuged at 400× g for 10 min at 4 • C to remove the cells.Protease and phosphatase inhibitors (1 µg/mL) were added to the supernatant and subjected to sequential filtration through 0.45 and 0.22 µm filters to retain cell debris and particles larger than 200 nm.Then, the samples were subjected to ultracentrifugation at 100,000× g for 3 h at 4 • C. The precipitate containing the exosomes was resuspended in 200 µL of PBS containing protease and phosphatase inhibitors (1 µg/mL).The homogeneity of exosome size was determined using Zetasizer Nano ZS90 (Malvern Instruments, Orsay, France).A total of 1 µL of each preparation (experimental group, negative and positive control groups) was diluted in 999 µL of PBS, and the parameters of size distribution and zeta potential (electronegativity) of the exosomes were analyzed at 37 • C according to Khashayar et al. [140]. Protein Profile of Neutrophil Supernatants An aliquot of each homogeneity of the exosomes was taken to determine the protein concentration by the BCA colorimetric method (Bicinchoninic Acid Protein Assay Kit, Sigma-Aldrich, St. Louis, MO, USA).To evaluate the protein profile after ultracentrifugation, the samples (20 µg) were incubated with 0.02% bromophenol blue, 10 mM mercaptoethanol, and 10% sodium dodecyl sulfate (SDS) for 5 min at 95 • C for protein denaturation.Then, they were separated on a 12% (w/v) polyacrylamide gel (SDS-PAGE).In the electrophoretic run, a constant current of 90 volts was fixed and, between 10-15 • C, to avoid band distortions and, consequently, resolution problems.For comparison purposes, 5 µL of Molecular Weight Standard (10-230 kDa Biolabs) was added to each electrophoretic run.The gel was stained with Coomassie Blue, and the bands were visualized directly. Western Blot Protein Expression For protein expression analysis, the proteins were electrotransferred to the adsorbent 0.45 µm nitrocellulose membrane, previously hydrated in water for 2-3 min in a transfer buffer for 5 min.The membrane was blocked with 3% (v/v) bovine serum albumin (BSA) for 1 h.After three 5 min washing steps with TPBS (PBS and 0.1% Tween), the primary antibodies to CD63 and CD81 were diluted in the titration established by the manufacturer and dispensed on the membrane, followed by incubation for 18 h at 20 • C. The washing steps were repeated, and then a secondary antibody conjugated with peroxidase (Sigma-Aldrich) was added and incubated for 2 h.After the washing steps, the membrane was developed using diaminobenzidine (DAB), 1 M Tris HCl pH 6.8, and 30% H 2 O 2 [19]. Proteomic Analysis Protein bands of each sample from RPMI (negative group), PMA (positive control), and Cr-LAAO (experimental group) were excised and submitted to in-gel digestion protein according to Shevchenko et al. [141] using trypsin as a proteolytic enzyme.Then, peptides were extracted from the gel matrix, followed by desalting in homemade C18 stage tips.The peptides were analyzed by LC-MS/MS in triplicate in an EASY nLC 1000 coupled to an LTQ Orbitrap XL ETD mass spectrometer (Thermo Scientific) equipped with a nanoelectrospray ion source (Phoenix S&T) at the Mass Spectrometry Facility RPT02H at Instituto Carlos Chagas, FIOCRUZ (Curitiba, Parana, Brazil) according to Forrest et al. [142]. Mass Spectrometry Data Analysis Raw mass spectrometry data in RAW format was submitted to PatternLab software [version 4.0.0.84] [143] to obtain protein identification.The main parameters used in this tool were the UNIPROT database (Taxonomy Homo sapiens; 74,788 proteins, 5 November 2019); trypsin enzyme; allowance for 2 stray cleavages; post-translational modification carbamidomethylation of cysteine residues; variable post-translational modification oxidation of methionine residues; and MS 40 ppm and MS/MS tolerance errors of 0.0200 ppm.The maximum FDR (False Discovery) rate was considered to be ≤1%.A matrix compatible with the MetaboAnalyst 4.0 ® program (https://www.metaboanalyst.ca/accessed on 4 February 2022.) was constructed using the proteomic data.The spectral counts were normalized for each protein identified by the weighted average of the duplicate of each sample [144].Proteins that have been identified in less than 70% of the samples were excluded from the analysis.Partial Least Squares (PLS) was used as the main method of multivariate analysis.Only signals present in 80% of the samples were considered for the generation of statistical models.Proteins were subjected to enrichment analysis for Gene Ontology (GO) terms "molecular function", "biological process", and "cellular component" using Panther software (www.pantherdb.org(accessed on 4 March 2023)).Protein interactions were also investigated regarding their biological processes using the STRING software, version 12.0 (http://string-db.org(accessed on 15 March 2023)), using the basic parameters: cut-off score of 0.90, confidence as network edges, and PPI enrichment p-value of <1.0 × 10 −16 . Figure 1 . Figure 1.Exosome characterization.(A) Protein profile of the negative control (RPMI), positive control (PMA), and experimental group (Cr-LAAO) samples by SDS-PAGE 12% (w/v).MW: molecular weight.Histogram of the particle profile in the supernatant of the RPMI (B), PMA (C), and Cr-LAAO (D) groups after ultracentrifugation using dynamic light scattering (DLS) technique employed to determine the nuclear dimensions of the nanoparticles using the Zetasizer Nano ZS system (Mallvern ® Instrument, Malvern-United Kingdom).(E) Western blot for CD81 and CD63 using the supernatant of neutrophils (1 × 10 7 ) stimulated with PMA (500 ng/mL) for the positive control (C+), RPMI for the negative control (C−), and Cr-LAAO (100 µg/mL) after 3 h of incubation in a humidified atmosphere of CO2 (5%) at 37 °C. Figure 1 . Figure 1.Exosome characterization.(A) Protein profile of the negative control (RPMI), positive control (PMA), and experimental group (Cr-LAAO) samples by SDS-PAGE 12% (w/v).MW: molecular weight.Histogram of the particle profile in the supernatant of the RPMI (B), PMA (C), and Cr-LAAO (D) groups after ultracentrifugation using dynamic light scattering (DLS) technique employed to determine the nuclear dimensions of the nanoparticles using the Zetasizer Nano ZS system (Mallvern ® Instrument, Malvern, UK).(E) Western blot for CD81 and CD63 using the supernatant of neutrophils (1 × 10 7 ) stimulated with PMA (500 ng/mL) for the positive control (C+), RPMI for the negative control (C−), and Cr-LAAO (100 µg/mL) after 3 h of incubation in a humidified atmosphere of CO 2 (5%) at 37 • C. Figure 2 . Figure 2. Volcano Plot analysis of Cr-LAAO ̶ stimulated neutrophils.Important features selected by volcano plot with fold change threshold (x) and t-tests threshold (y).The pink circles represent features above the threshold.Note both fold changes and p values are log-transformed.The further its position away from the (0, 0), the more significant the feature is (A).PLS-DA analysis of differentially expressed proteins in response to Cr-LAAO treatment in cells (control group vs. LAAO group) (B).VIP score (Variable Importance in Projection score) represents the importance of each variable in the protein projection.The important factors identified by the PLS-DA analysis are listed in decreasing order of importance.VIP scores ≥ 1.7 were considered statistically significant.High VIP scores indicate a significant contribution of proteins to the separation between the control and Cr-LAAO groups.The boxes on the right indicate the relative concentration of the corresponding protein in each study group.Red boxes indicate higher protein abundance, while green boxes indicate lower abundance (C). Figure 2 . Figure 2. Volcano Plot analysis of Cr-LAAO-stimulated neutrophils.Important features selected by volcano plot with fold change threshold (x) and t-tests threshold (y).The pink circles represent features above the threshold.Note both fold changes and p values are log-transformed.The further its position away from the (0, 0), the more significant the feature is (A).PLS-DA analysis of differentially expressed proteins in response to Cr-LAAO treatment in cells (control group vs. LAAO group) (B).VIP score (Variable Importance in Projection score) represents the importance of each variable in the protein projection.The important factors identified by the PLS-DA analysis are listed in decreasing order of importance.VIP scores ≥ 1.7 were considered statistically significant.High VIP scores indicate a significant contribution of proteins to the separation between the control and Cr-LAAO groups.The boxes on the right indicate the relative concentration of the corresponding protein in each study group.Red boxes indicate higher protein abundance, while green boxes indicate lower abundance (C). Figure 3 . Figure 3. Ontology analysis of proteins differentially expressed in exosomes from Cr-LAAOstimulated neutrophils.Human neutrophils were stimulated with 100 µg/mL Cr-LAAO for 3 h.After incubation, the supernatant containing the exosomes was collected and subjected to ultracentrifugation (100,000× g) for 3 h.The exosomes were subjected to polyacrylamide gel electrophoresis for protein separation and subsequent proteomics analysis.Data referring to the biological process (A), molecular function (B), and cellular component (C) were analyzed using Panther software.Unique proteins are present in the Cr-LAAO experimental group when compared to the control group. Figure 3 . Figure 3. Ontology analysis of proteins differentially expressed in exosomes from Cr-LAAOstimulated neutrophils.Human neutrophils were stimulated with 100 µg/mL Cr-LAAO for 3 h.After incubation, the supernatant containing the exosomes was collected and subjected to ultracentrifugation (100,000× g) for 3 h.The exosomes were subjected to polyacrylamide gel electrophoresis for protein separation and subsequent proteomics analysis.Data referring to the biological process (A), molecular function (B), and cellular component (C) were analyzed using Panther software.Unique proteins are present in the Cr-LAAO experimental group when compared to the control group. Figure 4 . Figure 4. Protein-protein interactions of differentially expressed proteins in neutrophils upon Cr-LAAO treatment.Data were generated using String software (https://string-db.org,accessed on 4 February 2022).Circles denote proteins, and straight lines illustrate interactions between these proteins.The thickness of the lines corresponds to the strength of associations. Figure 4 . Figure 4. Protein-protein interactions of differentially expressed proteins in neutrophils upon Cr-LAAO treatment.Data were generated using String software (https://string-db.org,accessed on 4 February 2022).Circles denote proteins, and straight lines illustrate interactions between these proteins.The thickness of the lines corresponds to the strength of associations. [1,69, B4, encoded by the SERPINB4 gene, can act as an inhibitor of parasitic and bacterial proteases, influence cell apoptosis, stimulate cell proliferation, and suppress immune defense against tumors[1,69,70]; [68]min, encoded by the DES gene, represents a muscle-specific III-type intermediate filament crucial for proper muscle structure and function[68];• Table 2 . Proteins differentially expressed in Cr-LAAO compared to PMA-stimulated in exosomes neutrophils.	v3-fos-license
2017-09-15T11:58:32.111Z	2014-11-14T00:00:00.000	6922557	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0112668&type=printable", "pdf_hash": "62a02f2c9d475acfa8618e7b15d779b3208e2e7c", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45444", "s2fieldsofstudy": [ "Medicine" ], "sha1": "f9c8af6c45d7b9d920ccc78634855dce01d17eb5", "year": 2014 }	pes2o/s2orc	Combination HIV Prevention among MSM in South Africa: Results from Agent-based Modeling HIV prevention trials have demonstrated the effectiveness of a number of behavioral and biomedical interventions. HIV prevention packages are combinations of interventions and offer potential to significantly increase the effectiveness of any single intervention. Estimates of the effectiveness of prevention packages are important for guiding the development of prevention strategies and for characterizing effect sizes before embarking on large scale trials. Unfortunately, most research to date has focused on testing single interventions rather than HIV prevention packages. Here we report the results from agent-based modeling of the effectiveness of HIV prevention packages for men who have sex with men (MSM) in South Africa. We consider packages consisting of four components: antiretroviral therapy for HIV infected persons with CD4 count <350; PrEP for high risk uninfected persons; behavioral interventions to reduce rates of unprotected anal intercourse (UAI); and campaigns to increase HIV testing. We considered 163 HIV prevention packages corresponding to different intensity levels of the four components. We performed 2252 simulation runs of our agent-based model to evaluate those packages. We found that a four component package consisting of a 15% reduction in the rate of UAI, 50% PrEP coverage of high risk uninfected persons, 50% reduction in persons who never test for HIV, and 50% ART coverage over and above persons already receiving ART at baseline, could prevent 33.9% of infections over 5 years (95% confidence interval, 31.5, 36.3). The package components with the largest incremental prevention effects were UAI reduction and PrEP coverage. The impact of increased HIV testing was magnified in the presence of PrEP. We find that HIV prevention packages that include both behavioral and biomedical components can in combination prevent significant numbers of infections with levels of coverage, acceptance and adherence that are potentially achievable among MSM in South Africa. Introduction The identification of a single HIV intervention that is capable of preventing large numbers of infections, such as a highly effective vaccine, remains elusive. Nevertheless, in recent years there have been enormous successes in identifying moderately effective HIV prevention interventions. These interventions include both behavioral and biomedical strategies. The question is how to combine these moderately effective interventions into highly effective prevention packages [1][2][3]. The idea is that multiple interventions when used in combination could prevent more infections than any single intervention used in isolation. Furthermore, the effectiveness of interventions when used in combination may be synergistic. Although the efficacies of various interventions applied in isolation have been evaluated in a number of rigorous randomized controlled trials, there is little direct evidence about the efficacy or effectiveness of combinations of these interventions [4][5][6][7]. Estimates of the effectiveness of combinations of interventions are important for guiding the development of prevention strategies. Characterizing the effect sizes of combination prevention interventions are critically important before embarking on large scale randomized controlled trial to rigorously evaluate such combination interventions to insure that the trials are adequately powered [8][9][10]. The drivers of the global HIV MSM epidemics have been previously reviewed highlighting the sustained HIV prevalence and often increasing HIV incidence among these men in several epidemic contexts [11,12]. The majority of randomized trials in the context of generalized HIV epidemics have focused on heterosexual or vertical transmission of HIV with limited data evaluating efficacy of interventions focused specifically on men who have sex with men (MSM) [13]. Here we report the results from agent-based modeling of the effectiveness of combination prevention interventions for MSM in South Africa. The work is part of the Sibanye Health Project to develop and test HIV prevention interventions among MSM in South Africa. Agent-based modeling has been used to evaluate the drivers of the HIV epidemic in several MSM populations [14]. Here our focus is the use of agent-based models to estimate the overall effectiveness of combination prevention in terms of percentages of infections prevented among MSM in South Africa. Recently, there has been important modeling work of HIV prevention strategies in MSM populations focusing on specific interventions in various regions of the world such as HIV testing in New South Wales [15], circumcision in Peru [16], and antiretroviral treatments including testing and linkage to care (but not preexposure prophylaxis) in China [17]. While previous modeling work on combination HIV prevention has been performed in South Africa [18], those models have not focused on the MSM population. This paper is focused on the evaluation of combination HIV prevention in the MSM population in South Africa. We examine four components of combination prevention: treatment of HIV infected persons with ART; prophylactic treatment of high risk HIV uninfected persons to reduce risk of acquisition of HIV infection (PrEP); counseling and condom promotion to reduce the frequency of unprotected anal intercourse; and HIV antibody testing. We perform a detailed statistical analysis of the simulation results of the agent-based model to assess the stochastic variability in the results and to borrow strength across all the simulations to improve estimates of the effects of combination prevention. Overview of Agent-based Model Agent-based models are stochastic simulations of interacting agents (e.g. individuals) who may alter behaviors in response to other agents or changes in the environment [19][20]. We developed an agent-based model to evaluate combination HIV prevention interventions among MSM in South Africa. Here, we describe the main features of the model which are also summarized in Table 1. Further details and specific model parameter values are given in the Supporting Information S1. Each person (agent) is assigned values for variables associated with risks for transmission and acquisition of HIV. The distributions of these variables were chosen to match available data from South Africa [21]. To simulate the heterogeneities in risk across MSM populations, the values of the variables (e.g., level of sexual activity, numbers of partners) were drawn from probability distributions. The variables include level of and predominant type of sexual activity (e.g., receptive or insertive unprotected anal intercourse [22], numbers of regular partners, whether or not the person is in a main partnership, frequency of HIV antibody test screening, HIV infection status at baseline, and for HIV infected persons, their CD4 levels and whether or not receiving they were receiving ART at baseline. Persons were assigned into variable sized networks of regular sexual partners. One of those regular partners could also be assigned to be the person's main sexual partner. Persons were also allowed to have sexual contact with persons outside their network of regular partners (i.e., casual partners). The probability of sexual contact on any day between two persons depended on whether the partnership was between main partners (most likely), regular partners (somewhat less likely) or casual partners (least likely). The agent-based simulation proceeded day by day, starting at baseline (which is defined as calendar time t = 0). On each day, we simulated whether an HIV uninfected person had sexual contact with an infected person for every possible pair of persons. If an HIV uninfected person had contact with an infected person, we simulated transmission occurrence. The probability of transmission was determined by factors that included the type of sexual contact (e.g., insertive or receptive role in unprotected anal intercourse (UAI)), antiretroviral treatment for the infected partner, and oral Truvada-based pre-exposure prophylaxis (PrEP) for the uninfected partner [5,[23][24][25]. Persons also had an opportunity to receive an HIV test. At the end of each day, the infection status and CD4 cell count were updated [26]. Persons were removed from the simulation when death occurred. We considered four possible components of combination prevention interventions and different intensity levels of these components. One component was treatment of HIV infected persons with ART [4]. Consistent with current South African national standards, HIV infected persons with a CD4 ,350 who had an HIV test within the preceding 6 months were eligible to receive ART. We considered a range of values for the percentage (X 1 ) of persons eligible for ART and not already in treatment at baseline who receive ART. Here X 1 is measuring additional ART coverage among eligible persons over and above those already receiving ART at baseline. Data from South Africa indicated approximately 50% of eligible persons are on ART and that value was taken to be the baseline level of ART coverage [21]. A second component of combination prevention was prophylactic treatment of high risk HIV uninfected persons to reduce risk of acquisition of HIV infection with tenofovir/emtricitabine (Truvada) (PrEP) [5]. HIV uninfected persons who had an HIV test within the preceding 6 months and were at high risk (defined as either .12 UAI acts in the preceding 6 months or having a main partner who is HIV infected) were eligible to receive PrEP. We considered various values for the percentage (X 2 ) of eligible persons who were offered and accepted PREP. Persons who received PrEP were classified as either low or high adherers (see online supplement for details). The model allowed adherence level to modify the effectiveness of PREP in reducing risk of HIV acquisition [28]. The third component was counseling and condom promotion to reduce the frequency of UAI [29]. We considered a range of values for the proportionate reduction in UAI contacts (X 3 ). For example, X 3 = 15% refers to an intervention that successfully reduces the rate of UAI by 15%. The fourth component was a program to increase HIV antibody testing. We considered an intervention component that decreased by one half the proportion of persons who never receive an HIV antibody test, from 1/3 to 1/ 6. The presence of this component in combination HIV prevention intervention is indicated by X 4 = 1 (otherwise X 4 is set to 0). We considered combination prevention interventions consisting of one or more of these four components: ART treatment coverage; PrEP coverage, UAI reduction, and HIV testing increase. We considered a range of for the levels of each component (values for X 1 , X 2 , and X 3 , ranged between 0.0% and 95%, and X 4 took values 0 or 1. For example, a combination HIV prevention intervention with X 1 = 75%, X 2 = 25%, X 3 = 5%, and X 4 = 1 corresponds to a combination prevention intervention with four components: ART given to 75% of all eligible persons who were not already receiving ART at baseline; PrEP given to 25% of eligible persons; a 5% reduction in the rate of occurrence of UAI; and a halving of the proportion of persons who never received an HIV test. We considered various combination prevention interventions by varying the values of X 1 , X 2 , X 3 and X 4 , and performed multiple replications for each of those combination prevention interventions. We performed replications of the simulations to assess stochastic variation [27]. The number of replications was chosen to control the standard error. Specifically, we calculated the standard error of the mean proportion infected over 5 years after each replication using all replications performed up to that point. If the standard error was above 0.01 we proceeded and performed an additional replication. We stopped replications when the standard error fell below.01. The mean number of replications performed for a combination intervention was 13 with a minimum of at least 5 replications performed for each combination intervention. In addition, we performed 60 replications for the control setting of no intervention (i.e., each X i = 0). We considered all packages corresponding to 4 levels each of ART coverage, 4 levels of PREP coverage, 4 levels of UAI reduction and 2 HIV testing levels. In addition we considered a number of additional packages of interest including when one or more of the levels were 0 as well as some additional packages when the UAI reduction was fixed at X 3 = 15% which was a value thought to be potentially achievable. In total we studied 163 HIV prevention packages and a total of 2252 simulations run of the agent-based model across all packages. Each simulation run was carried out for a five year period. The agent-based models were implemented in the statistical programming language R with the multithreading package 'snowfall' to address the highly intensive computational demands [30][31]. Statistical Analysis We performed statistical analyses of the dataset of results from the 2252 simulation runs of the agent-based model. The dependent variable (y) was the cumulative proportion of MSM that became HIV infected over 5 years from each run of the simulation. We developed a statistical model to relate y (the dependent variable) to the levels of the components (X 1 , X 2 , X 3 and X 4 ) in the combination prevention intervention (the explanatory variables). We used a generalized linear model with a logistic link of the form, log(y/(1-y)) which we arrived at after model fitting and regression diagnostics [32]. We considered both linear and higher order polynomial terms (e.g. quadratic and cubic terms) for the levels of the components (the X's). We also considered interaction terms between the components. Because we performed replications of each combination interventions we were able to evaluate the variance of y and found that the variance of y was not constant across interventions but varied with the magnitude of y. We found that a cubic polynomial adequately described the relationship between the variance of y and the expected value of y. Accordingly, we used iteratively reweighted least squares to account for the non-constant variance and to estimate the regression coefficients in the model for y [33]. We used the resulting model for y to calculate the percent of HIV infections prevented for a combination intervention with component levels X 1 , X 2 , X 3 , and X 4 compared to no intervention (i.e., when all the X's are equal to zero). We calculated confidence intervals for the percent of HIV infections prevented that accounted for the covariance between regression coefficients (see Supporting Information S1 for detail). The model allows us to predict the percent of HIV infections that could be prevented among MSM for any levels of the components of combination prevention intervention. The statistical analyses were performed with the R programming language [30]. Figure 1 is a graphical display of the 2,252 simulation runs of the agent-based model prior to any statistical modeling of the results. Each data point corresponds to one of the 163 combination prevention interventions including the control setting of no intervention. Each data point plots the average of the cumulative percent of MSM who become HIV infected over five years (y) versus the standard deviation of y based on all replications performed for that intervention. The mean cumulative percent infected over 5 years ranged as high as 26.4% when there was no intervention (X 1 = X 2 = X 3 = X 4 = 0). As shown in the figure, interventions that reduced the rate of UAI by at least 25% succeeded in reducing the cumulative incidence of infection to less than 15%, and thereby preventing at least 100 x (26.4-15)/ 26.4 = 43.2% of HIV infections. The figure also shows that the standard deviation of y increased with y; there was a decreasing trend in the coefficient of variation of y (i.e., standard deviation/y) from approximately 0.20 to 0.16. Results The regression model equation for y is given in the online supplement. Figure 2 is based on that equation and shows the percentage of HIV infections prevented for a range of combination interventions that include: UAI reduction of 0 or 15%; PrEP coverage of 0 or 50%; HIV testing increase that reduced the proportions of persons who never test by 50%; and a continuous range for incremental ART coverage (X 1 ) over and above persons already receiving ART at baseline. The figure shows that an intervention with only a 15% UAI reduction and no other component was superior to all other interventions that did not include a UAI reduction component. The figure illustrates a positive association between the percentage of infections prevented and increasing ART coverage (X 1 ), but the positive slope is small and that finding is explained because X 1 , as defined here, refers only to the additional ART coverage of eligible persons (,350 CD4 and an HIV test within the previous 6 months) who were not already receiving ART at baseline. We consider this point further in the Discussion Section. Figure 3 is similar to Figure 2 except it includes interventions with UAI reduction of 25% and PrEP coverage of 25%. The figure shows that a combination prevention intervention with a 25% UAI reduction, 25% PrEP coverage, and HIV testing increase (X 2 = 25%, X 3 = 25%, X 4 = 1) can prevent more than 35% of infections over five years. To understand how high-impact prevention packages could be constructed, we assumed that a basic prevention package would include ART coverage of 50% of eligible persons. Table 2 shows the effects of sequentially adding components to ART coverage to create HIV prevention packages. We find that the addition of a component that reduces UAI by 15% (X 3 = 15%) would prevent an additional 20.3% of infections over and above the base package (Package 1 in Table 2). That reduction in HIV infections from the addition of a condom promotion/UAI reduction component to the base package is considerably greater than that achieved from the addition of a PrEP component with 50% coverage (which yields an additional 9.5% of infections prevented) or an increase in Figure 2. HIV infections prevented over 5 years from combination prevention interventions with four components. ART coverage of eligible persons who were not already receiving ART at baseline, PREP with 50% acceptance (dotted lines), 15% UAI reduction (blue lines; no UAI change are in red) and increase in HIV testing (black triangles). See Table 1 for further details about the components of the prevention interventions. doi:10.1371/journal.pone.0112668.g002 Figure 3. HIV infections prevented over 5 years from combination prevention interventions with four components. ART coverage of eligible persons who were not already receiving ART at baseline, PREP with 25% acceptance (dotted lines), 25% UAI reduction (blue lines; no UAI change are in red) and increase in HIV testing (black triangles). See Table 1 for further details about the components of the prevention interventions. doi:10.1371/journal.pone.0112668.g003 HIV testing of previously untested men (which yields an additional 2.9% of HIV infection prevented). If we start with a package that includes both 50% ART coverage and 15% UAI reduction components (Package 2 in Table 2), we find that HIV infections could be reduced by 10.1% with the addition of a PrEP component (with 50% coverage) or, alternatively, reduced by 3.1% with the addition of an HIV testing component to reach men never tested for HIV. If we start with a package that includes three components, ART coverage, UAI reduction and PrEP components (Package 3 in Table 2), we find that the addition of the HIV testing component would further reduce infections by 4.9%. We find that the impact of an HIV testing component is greater in the presence of a PrEP component than without a PrEP component (compare 4.9% to 2.9% and 3.1%). Discussion In recent years significant progress has been made in HIV prevention science. Findings from HIV prevention trials have demonstrated the effectiveness of behavioral and biomedical interventions such as earlier initiation of ART, PrEP, condoms and behavioral change. The effectiveness of these interventions will depend on their availability in communities as well as levels of uptake and adherence by persons at risk. Each of these interventions is only partially effective in preventing HIV infections, and as such, no single intervention is expected to be sufficient to eliminate the spread of HIV. HIV prevention packages offer the potential to significantly increase the effectiveness of any single intervention. HIV prevention packages offer multiple approaches for reducing risks and possibilities of synergies between the interventions. Quantification of the effectiveness of HIV prevention packages is important for developing combinations of interventions and for designing prevention trials of combination prevention. Unfortunately most research to date has focused on testing only a single intervention rather than HIV prevention packages. In this report we used agent-based models to evaluate the effectiveness of HIV prevention packages among MSM in South Africa. We identified a four component HIV prevention package for MSM in South Africa which could prevent approximately 34% of infections over five years. We single out this intervention for discussion because it is a potentially achievable combination package that we found to be particularly effective. That four component package consists of 50% ART coverage for eligible persons who were not already receiving ART at baseline, 50% PrEP coverage for high risk eligible persons, 15% UAI reduction and a 50% reduction in those who never test for HIV. The component with the largest incremental impact on infections was the 15% UAI reduction which prevented an additional 21% of infections when added to a package of the other three components. PrEP coverage had the second largest incremental impact, followed by HIV testing and additional ART coverage over baseline levels. We find that even small reductions in UAIs can have huge effects. We believe the target goals for coverage of each intervention component of the four component package outlined above are achievable in the MSM population in South Africa with concerted commitments and prioritization for HIV prevention for MSM. While the target goals for each component of the prevention package are modest, those goals are at present not being met. Currently, there is essentially no uptake of PrEP among MSM in South Africa; most anal intercourse acts are not protected by condoms; and significant numbers of men in South Africa have not been tested for HIV. In order to achieve the scale of interventions required for public health impact, the coordinated efforts of government, clinicians, and community are required. We found that the impact of HIV testing in prevention packages depended on which other components were in the package, and specifically, the impact of the HIV testing was magnified when PrEP was included in the package. Such synergies make sense because HIV testing is a gateway to access to PrEP. In this report, we examined the effect of reducing the numbers of persons who never receive an HIV test by half. The effect of an HIV testing component in a prevention package would be greater if the never testers were reduced by more than half or the HIV testing frequencies were increased among persons that do test. We found a modest effect of ART coverage relative to the other components of the package. This finding was initially surprising because other modeling work has demonstrated that ART can have a significant impact on HIV incidence. For example, Eaton and colleagues [34] performed a systematic review of models to address the question of the impact of ART in a treatment-naïve population. They found that HIV incidence would be considerably lower after 8 years if large numbers remain on ART compared to a counterfactual scenario in which there is no ART. However, Eaton and colleagues emphasize that their result assumed a treatment-naïve population at baseline, that is, the Eaton work assumed ''ART was introduced into the population beginning in 2012 with no treatment provision prior to this which is in contrast to the rapid scale up of treatment that has actually occurred prior to 2012 in South Africa'' [34]. An important difference of the Eaton work from our work is that we are evaluating the impact of extending ART coverage in the context of significant numbers (50% of eligible persons) already receiving ART at baseline rather than in a treatment-naïve population. Because we had significant numbers of persons receiving ART at baseline, the impact of additional coverage of newly eligible persons is smaller than if the population was treatment-naïve at baseline. ART eligibility requires an HIV test in the preceding 6 months and a CD4 count less than 350. Furthermore, the numbers of persons becoming eligible for ART over the 5 year period (who were not already eligible at baseline) were staggered over the 5 years and were not becoming eligible all at once in a bolus. For example, in our simulation of the control (no intervention) scenario, the number of persons receiving ART at baseline (t = 0) was 55 persons out of 255 infected persons at baseline. In a simulation of the four component package (50% ART coverage for eligible persons who were not already receiving ART at baseline, 50% PrEP coverage for high risk eligible persons, 15% UAI reduction and a 50% reduction in those who never test for HIV), the additional numbers of persons who would go on ART at some time post-baseline during the subsequent 5 years is only 71 persons in addition to the 55 persons already receiving ART at baseline. Furthermore, more than half of these additional 71 persons going on ART would in fact not begin ART until after 2.5 years post baseline (t.2.5). These numerical results illustrate that in our simulation work, the incremental ART coverage over baseline is relatively modest. In our modeling setting, the benefit of ART is limited by the number of persons with clinical indication (,350 CD4) who were not receiving ART at baseline and who had an HIV test. As pointed out by Eaton and colleagues ''comparing results and conclusions across models is challenging because models have addressed slightly different questions'' [34]. The impact of ART is driven by the numbers of treatment at baseline, the treatment threshold, the sufficiency of HIV testing to identify those living with HIV, and of course the extent to which treatment is efficacious in reducing infectiousness. If the guideline for treatment were to shift to CD4 count below 500 or an even higher threshold, the impact of ART would be greater. The impact of a PrEP component in a prevention package depends on the eligibility requirement. In our work, the PrEP eligibility requirement was either being in a sero-discordant main partnership or having a very high rate of UAIs (12 per 6 months). If that threshold for eligibility is lowered to expand the numbers who are eligible, then the impact of PrEP would be greater than reported here. We performed considerable numbers of replications of our agent-based modeling to account for stochastic variation. The confidence intervals we report account for the stochastic variation. However, as in all agent-based models, our model is based on numerous assumptions and input parameters. Many aspects of our model such as the networks of sexual partners, distributions of numbers of partners and baseline frequencies of UAIs relied on limited data. Furthermore, the regular and main partners did not change over the 5 years of the simulation. We did however allow persons to have contacts outside their network of regular partners (casual partners). We only forecast five years in an attempt to limit the sensitivity of the results to these model simplifications. More reliable information about partner formation and dissolution among the MSM population on South Africa is important to further inform models of HIV prevention. We focused on the MSM population and did not attempt to model the dynamics of Components include ART (50% ART coverage of eligible persons from among those not already receiving ART at baseline); PREP (50% acceptance of PREP among eligible persons); UAI reduction (15% reduction), and HIV testing increase (50% reduction of persons who never have an HIV test). 1 The percent infections prevented due to component i refers to the percentage decrease in the 5 year cumulative HIV incidence with the HIV package that includes all four components compared to the HIV prevention package that includes three of the four components leaving out component i. transmission within and across other risk groups such as intravenous drug users. We also did not model variable infectiousness over time. We did not account for new incoming MSM to the population although that simplification may have a small effect over the five years the simulations were run. We recognize that caution should be exercised when interpreting the findings from agent-based models that rely on many simplifications and assumptions. As such, we focused on presenting the results in terms of relative effects of a prevention package (e.g., the percent of infections averted with a prevention package compared to no intervention) because relative effects may be less sensitive to model assumptions than the absolute cumulative number of infections. Nevertheless, caution should still be exercised as with all modeling results. In spite of these limitations, we believe agentbased modeling offers a useful tool for approximating the effectiveness of HIV prevention packages when direct empirical data from comparative studies of combination HIV prevention is unavailable. Further research on understanding associations within individuals with regard to uptake and adherence levels across the various components of a prevention package will help to refine our models. For example, identification of subgroups that are resistant to accepting or adhering to any intervention would be important for modeling and also for helping to design packages to overcome barriers to acceptance of HIV prevention. The HIV epidemic among MSM in South Africa continues to grow. Obtaining sufficiently high levels of coverage, acceptance and adherence with any single biomedical or behavioral intervention is a major obstacle to controlling epidemic growth. Combination HIV prevention offers the possibility of preventing significant numbers of infections with sufficient levels of coverage, acceptance and adherence; these levels are achievable with the concerted efforts of multiple stakeholders. In the context of a vigorous debate about the roles of behavioral and biomedical interventions, our results are reconciling in that we demonstrate that traditional HIV prevention activities, such as condom promotion and HIV testing programs, still play vital roles in the context of biomedical prevention. HIV prevention packages that include both behavioral and biomedical components can, in combination, prevent significant numbers of infections among MSM in South Africa.	v3-fos-license
2019-05-22T14:27:05.677Z	2019-05-21T00:00:00.000	160010221	{ "extfieldsofstudy": [ "Chemistry", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.nature.com/articles/s41419-019-1627-9.pdf", "pdf_hash": "0149493a80e1a1d77219311a67cc68c5c506fac3", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45445", "s2fieldsofstudy": [ "Medicine", "Chemistry" ], "sha1": "0149493a80e1a1d77219311a67cc68c5c506fac3", "year": 2019 }	pes2o/s2orc	XPO1 inhibitor KPT-330 synergizes with Bcl-xL inhibitor to induce cancer cell apoptosis by perturbing rRNA processing and Mcl-1 protein synthesis XPO1 (exportin1) mediates nuclear export of proteins and RNAs and is frequently overexpressed in cancers. In this study, we show that the orally bioavailable XPO1 inhibitor KPT-330 reduced Mcl-1 protein level, by which it synergized with Bcl-xL inhibitor A-1331852 to induce apoptosis in cancer cells. KPT-330/A-1331852 combination disrupted bindings of Mcl-1 and Bcl-xL to Bax, Bak, and/or Bim, elicited mitochondrial outer membrane permeabilization, and triggered apoptosis. KPT-330 generally mitigated mRNA expression and protein synthesis rather than mRNA nuclear export or protein stability of Mcl-1. KPT-330 inhibited mTORC1/4E-BP1 and Mnk1/eIF4E axes, which disrupted the eIF4F translation initiation complex but was dispensable for Mcl-1 reduction and KPT-330/A-1331852 combination-induced apoptosis. Mature rRNAs are integral components of the ribosome that determines protein synthesis ability. KPT-330 impeded nucleolar rRNA processing and reduced total levels of multiple mature rRNAs. Reconstitution of XPO1 by expressing degradation-resistant C528S mutant retained rRNA amount, Mcl-1 expression, and Bcl-xL inhibitor resistance upon KPT-330 treatment. KPT-330/A-1331852 combination suppressed growth and enhanced apoptosis of non-small cell lung cancer xenografts. Therefore, we clarify the reason of apoptosis resistance of cancer cells to XPO1 inhibition and develop a potential strategy for treating solid tumors. Introduction Exportin1 (XPO1, also known as chromosomal maintenance region 1, or CRM1) mediates nuclear export of proteins and RNAs, and ribosome biogenesis, which are important for cancer growth and survival 1 . XPO1 is frequently amplified or mutated in several hematological and solid tumors. XPO1 overexpression correlates with poor prognosis in various cancers, whereas either targeting XPO1 alone by the selective inhibitors of nuclear export (SINE) or in combination with other targeted therapies or chemotherapies shows broad anticancer effect and acceptable tolerance [2][3][4] . SINE compounds degrade XPO1 protein by specific binding to its C528 residue in the cargo-binding groove. One of the first-generation orally bioavailable SINEs, KPT-330 (selinexor) is under testing in patients in 64 phase I/II/III trials (ClinicalTrials.gov), whilst the brain-associated adverse effects like anorexia and weight loss, and hematologic adverse effects like thrombocytopenia limit its dose 5 . The second-generation SINE, KPT-8602 has proven its activity against hematological malignancies, with improved tolerability than KPT-330 owing to its lower brain penetration in preclinical animal models 6,7 . The balance between the antiapoptotic (Bcl-2, Bcl-xL, Mcl-1, and less studied Bcl-W and BFL-1) and proapoptotic Bcl-2 family proteins (Bax, Bak, and BH3 domain-only proteins) determines the activity of mitochondrial apoptotic signaling 8 . The functional redundancy of antiapoptotic proteins safeguards cancer cells from apoptotic induction when some of the proteins are compromised. Whereas high Bcl-2 expression dominates the survival of some liquid tumors making targeting Bcl-2 sufficient to kill them 9,10 , Bcl-xL and Mcl-1 often act as double insurance for solid tumor survival increasing the apoptotic threshold and entailing dual targeting for apoptosis induction [10][11][12][13] . The development of the dual Bcl-2/Bcl-xL inhibitor ABT-263 ended up in vain due to thrombopenia resulted from Bcl-xL inhibition. However, the Bcl-xL-selective inhibitors A-1155463 and A-1331862 demonstrated tolerability and efficacy in preclinical solid tumor models 14 . Mcl-1 is a short-lived protein that is vulnerable to suppression of protein expression on the transcriptional, post-transcriptional, translational, or post-translational levels 11,[15][16][17] . Recently, Mcl-1-selective inhibitors evolved and one of them showed exceptional anticancer efficacy 12,18 . Furthermore, it was demonstrated that SINE compounds including KPT-185, KPT-276, and KPT-330 downregulated Mcl-1 protein [19][20][21] , but the underlying mechanism and function of Mcl-1 upon SINE treatment are unclear. It was hypothesized in one prior study that nuclear retention of Mcl-1 mRNA caused Mcl-1 downregulation 20 . In this study, we investigated the effect and regulatory mechanism of KPT-330 on Mcl-1 expression and developed combination therapy to enhance the anticancer activity of KPT-330. We demonstrated that KPT-330 decreased Mcl-1 protein synthesis through mitigating rRNA processing and global protein synthesis, making cancer cells more susceptible to Bcl-xL inhibitors like A-1331852. KPT-330 synergized with A-1331852 to induced apoptosis in a range of cancer cells in vitro and suppressed tumor growth in a non-small cell lung cancer (NSCLC) model. XPO1 and Bcl-xL inhibitors synergistically induce apoptosis in cancer cells We interrogated the effect of XPO1 inhibitors on antiapoptotic Bcl-2 proteins to gain insights on the molecular mechanism conferring their inefficient apoptosis-inducing capacities. The XPO1 inhibitor leptomycin B (LMB) and KPT-330 consistently downregulated Mcl-1 but not Bcl-2 or Bcl-xL in a dose-dependent manner in U87 and U251 glioblastoma cells and H1299 NSCLC cells (Fig. 1a, b). LMB and KPT-330 also consistently downregulated Bim but not other proapoptotic Bcl-2 proteins in H1299 cells ( Fig. 1b). Mcl-1 reduction correlated well with XPO1 reduction upon KPT-330 treatment (Fig. 1a, b). Although Bcl-2, Bcl-xL, and Mcl-1 have different preference in binding antiapoptotic and BH3 domain-only Bcl-2 proteins, they play redundant roles in blocking mitochondrial outer membrane permeabilization (MOMP). Therefore, Mcl-1 downregulation by XPO1 inhibitor was insufficient to induce apoptosis in cancer cells but likely made cancer cells more susceptible to inhibitors targeting of Bcl-2 and/ or Bcl-xL. Indeed, in glioblastoma (A172, U87, U118, and U251), NSCLC (H1299 and A549), and cervical cancer cells (HeLa), inhibitor of Bcl-xL (A-1331852) or Bcl-2/ Bcl-xL (ABT-263) but not Bcl-2 (ABT-199) further reduced the viability of cells treated with KPT-330 at the dose capable of downregulating Mcl-1 (Fig. 1c), indicating that the remaining Bcl-xL rather than Bcl-2 confers to KPT-330 resistance in these cells. Combination of KPT-330 and different Bcl-xL-selective inhibitors triggered intense apoptosis in U87, U251, H1299, and A549 cells (Fig. 1d). In glioblastoma, NSCLC, and cervical cancer cells, KPT-330 plus A-1331852 had a strong synergistic effect on viability inhibition, as evaluated by their combination index (Figs. 1e and S1). JC-1 staining showed that such combination elicited MOMP in U251 cells (Fig. 1f). These results indicate a strong synergism of XPO1 and Bcl-xL inhibitor combination in apoptosis induction in cancer cells. KPT-330-downregulated Mcl-1 facilitates mitochondriamediated apoptosis upon KPT-330 and A-1331852 Although KPT-330 treatment counteracted the effect of Mcl-1 overexpression, a slight increase of Mcl-1 level partially reversed KPT-330/A-1331852-triggered apoptosis in U251 and H1299 cells (Fig. 2a). Overexpression of Mcl-1 also made H1299 cells more resistant to long-term KPT-330 treatment (Fig. 2b). KPT-330 diminished Mcl-1 binding to Bax, Bak, and Bim while enhanced Bcl-xL binding to Bax and Bcl-2 binding to Bim possibly compensating for Mcl-1 loss. A-1331852 prevented Bax and Bim from Bcl-xL binding. Their combination thereby freed and activated both Bax and Bak (Fig. 2c). We failed to detect Bax in Bim immunoprecipitant but showed no alteration of Bak/Bim binding upon drug combination possibly due to enhanced Bim sequestration by Bcl-2 ( Fig. 2c). Simultaneous knockdown of Bax and Bak reversed KPT-330/A-1331852-triggered apoptosis and MOMP in U251 cells (Fig. 2d, e). Noxa knockdown slightly reduced KPT-330/A-1331852-triggered apoptosis while Bim knockdown upregulated Noxa (at least in U251 cells) and increased such apoptosis (Fig. 2f, g degradation by Act D and MG-132 respectively (Fig. 3d, e), reflecting a defect of Mcl-1 translation. XPO1 can export RNA into the cytosol and mRNA nuclear retention impairs its translation. However, KPT-330 did not uniformly decreased cytosolic Mcl-1 mRNA levels in all tested cells (Fig. 3f). Nor did KPT-330 affect the cytosol/ nucleus distribution of LRPPRC, an XPO1 export cargo assisting eIF4E-dependent mRNA export 22 , in U251 and H1299 cells (Fig. 3g). Unexpectedly, the OPP incorporation assay demonstrated a reduced nascent synthesized protein content upon KPT-330 treatment (Fig. 3h). We further analyzed synthesis rates of some other proteins and found that the production of fast degrading proteins KPT-330 inhibits rRNA processing RNA content was reduced in KPT-330-treated cells (Fig. 5a). As rRNA constitutes the majority of total cellular RNA, we evaluated levels of different rRNA in these cells. KPT-330 reduced 5.8S, 18S, and 28S rRNA, processing products of 45S pre-rRNA synthesized in the nucleolus but not 5S rRNA synthesized in the cytosol in U87 and U251 cells (Fig. 5b). In contrast, none of these rRNAs is downregulated in KPT-330-treated H1299 cells (Fig. 5b). To further clarify whether KPT-330 disrupts nucleolar rRNA synthesis or processing, the nucleoli of U87, U251, and H1299 cells treated with KPT-330 or Act D were purified and the nucleolar RNAs were extracted. Fragment analysis showed that Act D blocked the synthesis of 45/47S and 32/34S pre-rRNAs while it increased 28S rRNA. Similarly, KPT-330 treatment resulted in reduced 45/47S and 32/34S rRNAs, while it enhanced 28S in the nucleolus of U87 and H1299 cells. Neither KPT-330 nor Act D increased 28S but tended to reduce 45/47S in U251 cells. Expressing a degradationresistant form of XPO1 (C528S) partially reversed KPT-330-induced defects of rRNA processing ( Fig. 5c-g). Consistently, KPT-330 suppressed nascent RNA synthesis in U87, U251, and H1299 cells (Fig. 5h). Not as reported recently 24 , KPT-330 did not significantly interfered rRNA nuclear export in any of the tested cells h, then subjected to western blot or flow cytometry analysis for the sub-G1 phase (mean ± SD, n = 3). P < 0.05. b H1299 cells expressing Mcl-1 (T92A) were treated KPT-330 (1 μM) for 3 or 4 days and subjected to flow cytometry analysis for the sub-G1 phase (mean ± SD, n = 3). P < 0.05. c U251 and H1299 cells were treated with KPT-330 (1 μM) for 24 h and further with A-1331852 (1 μM) for 6 h. Total and active forms of Bax and Bak in cell lysates were immunoprecipitated followed by western blot. d, e U251 and/or H1299 cells expressing shBax and shBak were treated as in a and subjected to western blot (d), flow cytometry analysis for the sub-G1 phase (d), or mitochondria membrane potential (e) (mean ± SD, n = 3). P < 0.05. f, g U251 and H1299 cells expressing shBim and/or shNoxa were treated with KPT-330 (1 μM) for 24 h and further with A-1331852 (1 μM) for 6 h or 24 h, then subjected to western blot (6 h) or flow cytometry analysis (24 h) for the sub-G1 phase (mean ± SD, n = 3). P < 0.05. GAPDH was used as the loading control (Fig. 5i). These results suggest that KPT-330 inhibits rRNA processing, which will definitely influence ribosome complex. XPO1 reconstitution restores rRNA and Mcl-1 expression and drug resistance upon KPT-330 and A-1331852 To verify that XPO1 degradation upon KPT-330 contributes to Mcl-1 reduction and cancer cell apoptosis following KPT-330 and A-1331852 treatment, we reconstituted XPO1 expression by overexpressing wild-type XPO1, degradation-resistant mutant C528S, or recurrent hotspot mutant E571K or R749Q in U251 and H1299 cells. C528S mutant restored protein levels of XPO1 and Mcl-1 in cells treated with medium (1 μM) or high dose (10 μM) of KPT-330, while wild-type, E571K, and R749Q variants had weak effects (Fig. 6a). C528S mutant also restored rRNA and Mcl-1 mRNA in KPT-330-treated U251 cells (Fig. 6b). Accordingly, U251 and H1299 cells expressing C528S mutant but not cells expressing other XPO1 variants were resistant to KPT-330/A-13331852induced viability reduction and apoptosis (Fig. 6c, d). However, C528S mutant failed to rescue KPT-330/A-13331852-induced apoptosis in Mcl-1-deficient U251 cells (Fig. 6e, f). Unexpectedly, C528S augmented apoptosis in A-1331852-treated Mcl-1-deficient cells (Fig. 6f). Bim was not involved in this effect since C528S did not enhanced Bim expression and Bim knockdown failed to abolish elevated apoptosis (Supplementary Fig. 5). Thus, these results demonstrate that KPT-330 targets XPO1 for degradation to disrupt rRNA and Mcl-1 expression, , and H1299 cells treated as in a (mean ± SEM, n = 3). P < 0.05. g Western blot analysis of indicated proteins in the cytosol and nucleus of U251 and H1299 cells treated as in a. h Nascent protein synthesis in U87, U251, and H1299 cells treated as in a was measured by OPP incorporation (mean ± SEM, n = 3). P < 0.05. GAPDH, α-tubulin, and MeCP2 were used as the loading control of total, cytosolic, and nuclear protein, respectively Combination of KPT-330 and A-1331852 suppresses tumor growth in vivo Finally, to evaluate the anticancer activity of KPT-330/A-1331852 combination in vivo, we inoculated NOD-SCID mice with H1299 cells and treated them with KPT-330 (10 mg/kg, p.o., Monday, Wednesday, and Friday) and/or A-1331852 (25 mg/kg, p.o., every day) or vehicle for 10 days when tumor volume reached~50 mm 3 . Exposure to KPT-330 or A-1331852 alone resulted in inhibition of tumor growth throughout the treatment and lower tumor weight in the end, while cotreatment with two drugs further suppressed tumor growth (Fig. 7a-c). However, the statistical difference of tumor volume or weight between cotreatment group and either monotherapy group was insignificant probably owing to good performance of either drug and relatively low synergistic effect of these drugs in H2199 cells in vitro (Fig. 7c). Mice in cotreatment group lost 15.5% of their body weight post treatment but were all alive (Fig. 7d). Despite not downregulating XPO1 level post treatment, KPT-330 generally maintained lower Mcl-1 level and, when combined with A-1331852, induced apoptosis in terms of caspase-3 cleavage (Fig. 7e, f). These results suggest that KPT-330/A-1331852 exerts anticancer effect in NSCLC xenografts and is tolerant in mice. Discussion The clinical SINE compound KPT-330/selinexor is a specific and reversible XPO1 inhibitor, with oral bioavailability and tolerability. Preclinical studies have demonstrated its apoptotic-inducing effect in various types of cancers and different associated molecular Real-time PCR analysis of indicated rRNAs in U87, U251, and H1299 cells treated as in a (mean ± SEM, n = 6 for U87 and U251 and n = 8 for H1299). P < 0.05. c Nucleolar RNAs from XPO1 (C528S) expressing U87, U251, and H1299 cells treated with KPT-330 (1 μM) for 24 h or Act D (500 ng/μl) for 3 h were subjected to fragment analysis. Quantification of the relative percentage of 18S, 28S, 32/34S, and 45/ 47S was shown in d-g (mean ± SEM, n = 3-6), respectively. P < 0.05. h Nascent RNA synthesis in U87, U251, and H1299 cells treated as in a was measured by EU incorporation (mean ± SEM, n = 5). P < 0.05. i Real-time PCR analysis of cytosolic and nuclear rRNAs indicated in U87, U251, and H1299 cells treated as in a (mean ± SEM, n = 4 for U87 and n = 5 for U251 and H1299). P < 0.05 f Cells treated as in e were further treated with A-1331852 (1 μM) for 24 h and subjected to flow cytometry for the sub-G1 phase (mean ± SD, n = 3). P < 0.05. α-Tubulin was used as the loading control mechanisms, like IκBα nuclear retention, NF-κB signaling inhibition, and survivin transcriptional inhibition 3,25 ; nuclear accumulation of p53 and FOXO3a 26 . However, whether it directly regulates the mitochondrial apoptotic signaling leading to apoptosis remains elusive. SINE was shown to downregulate the antiapoptotic Bcl-2 protein Mcl-1 that counteracts MOMP, but no study scrutinized the underlying mechanism and associated phenotype [19][20][21] . Given that targeting Mcl-1, directly or indirectly, alone or combinatorial proved to be promising apoptosis-based anticancer therapeutic strategies 13,17,18,27,28 , we reckon that delineating the overlooked aspect is worthwhile and combination therapy based on such information may extend the application range and improve the performance of KPT-330 in cancer treatment. In this study, we c Tumor weight of the corresponding H1299 xenografts. d Relative body weight curves of mice (mean ± SEM, n = 5). P < 0.05. e Western blot analysis of xenografts isolated from mice treated as indicated. GAPDH was used as the loading control. f Grayscale ratio quantification of indicated proteins to GAPDH in e by Photoshop software (mean ± SEM, n = 3 for vehicle and n = 4 for other groups). *P < 0.05. g Schematic overview of the purposed pathway for KPT-330/A-1331852 combination therapy to cotreatment of KPT-330 and A-1331852, while cells expressing recurrent hotspot mutant E571K and R749Q are as vulnerable to the cotreatment as those with wildtype XPO1 (Fig. 7g). Life of Mcl-1 protein is short. Strategies like inhibition of mTORC1/4E-BP1 signaling to constrain capdependent global protein translation 13,17 or signaling modifying Mcl-1 and coupling Mcl-1 to the ubiquitinproteasome pathway 27 commonly decrease Mcl-1 protein and suppress tumor growth. Coincide with the previous report that XPO1 inactivation dampened mTOR signaling 29 , we showed that KPT-330 inhibited mTORC1/4E-BP1 axis and Mnk1/eIF4E axis to diminish cap-dependent translation initiation activity, but dephosphorylation of 4E-BP1 and eIF4E lagged behind and was dispensable for Mcl-1 downregulation. Nor did we observe mTOR nuclear retention in our system (Fig. 3g) as before 29 . XPO1 regulates ribosome biogenesis. An iTRAQ analysis showed that SINE compound KPT-185 downregulated a series of ribosome proteins by~10-27% 30 , while a recent study demonstrated that KPT-330 crippled nuclear export of 5S and 18S rRNA, ribosome assembly, and protein synthesis in glioblastoma cells 24 . However, our data challenge such explanation showing that KPT-330 basically did not alter the cytosolic/nuclear distribution of nucleolar processed 5.8S, 18S, and 28S rRNA, but rather reduced their total expression in glioblastoma cells, which were concordant with the decrease of RNA content and nascent RNA synthesis ability in these cells. One probable reason for the rRNA distribution difference is that we used U6 small nuclear RNA as housekeeping gene to normalize the relative nuclear RNA levels in real-time PCR analysis instead of Actin used in the previous study 24 . Consist with a previous study revealing that XPO1 inhibition using LMB disrupts rRNA synthesis and processing 31 , we found KPT-330 caused decreased pre-rRNAs while increased 28S levels in the nucleolus of U87 and H1299 cells, which further reduced total expression of mature rRNAs in glioblastoma cells. Despite enhanced 28S levels in the nucleolar, KPT-330 resulted in reduced nuclear 28S of glioblastoma cells, suggesting that KPT-330 suppressed nucleolar/nuclear export. Furthermore, KPT-330-mitigated RNA synthesis ability may reflect the fact but more likely reflects the difficulty of rRNA processing as 45S pre-rRNA is unstable if not processed and reduction of mature rRNAs impairs nascent RNA accumulation. Coincidence with more severe rRNA processing deficiency, RNA and protein synthesis rates, and RNA content were lower in H1299 cells than in U251 cells. We speculate that the defect of Actin expression may contribute to the failure of detecting rRNA downregulation in H1299 cells. The XPO1 ribosome export adaptor NMD3 facilitates XPO1 nucleolar localization and they cooperatively regulate rRNA synthesis and processing 31 . According to the TCGA database, XPO1 and NMD3 genes are frequently altered in lung squamous cell carcinoma and to a less extent in lung adenocarcinoma ( Supplementary Fig. S3a). Accordingly, mRNAs of XPO1 and NMD3 are high in lung squamous cell carcinoma ( Supplementary Fig. S3b). Lung cancer samples with XPO1 alteration tend to express higher level of NMD3 (Supplementary Fig. S3c). In addition, NMD3 mRNA upregulation tends to accompany XPO1 mRNA upregulation (Supplementary Fig. S3d). Many coaltered genes in samples with XPO1 mRNA alteration function in RNA metabolism (Supplementary Fig. S3e). These bioinformatics information emphasize the key role of XPO1 in RNA metabolism and rRNA processing in cooperation with NMD3. Although mTORC1 controls ribosome biogenesis, including rRNA transcription and processing 32 , it is not the case here given that protein synthesis was attenuated 1 h after KPT-330 treatment 24 and mTORC1 substrate 4E-BP1 was dephosphorylated 24 h after treatment (Fig. 4d). Besides, 45S pre-rRNA expression was hardly changed (Fig. 5b). Interaction of mTORC2 and ribosome improves the activity of mTORC2/Akt signaling and thereby activates mTORC1 33 . However, KPT-330 did not inhibit Akt phosphorylation (Supplementary Fig. S4). In addition, phosphorylation of ERK and p38, upstream regulators of Mnk1, were paradoxically upregulated (Supplementary Fig. S4). Therefore, expression inhibition of components like mTOR and Mnk1 rather than inactivation of upstream regulators possibly resulted in the suppression of mTORC1/4E-BP1 and Mnk1/eIF4E axes. Since these axes are less important in regulating Mcl-1 expression here, we did not explore the associated molecular mechanism. We checked the status of several antiapoptotic, proapoptotic, and BH3 domain-only Bcl-2 proteins in H1299 cells following LMB or KPT-330 treatment. We observed a concurrence of band shift (probable phosphorylation) and downregulation of BimEL and downregulation of Mcl-1. Phosphorylation and degradation of Bim by kinases such as ERK upon anticancer treatment causes drug resistance 34 . We believe that Bim downregulation cooperates with or contributes to Bcl-xL/Bax interaction to make cancer cells adapt to XPO1 inhibition. Thus, fully neutralizing the activity of antiapoptotic Bcl-2 proteins can raise KPT-330 sensitivity. We chose Bcl-xL inhibitor A-1331852 to potentiate XPO1 efficacy considering the intolerance of Bcl-2/Bcl-xL inhibitor ABT-263 in the clinic and the primary role of Bcl-xL in apoptosis resistance in solid tumors after Mcl-1 inhibition 10-13 . Moreover, A-1331852 proves effective against solid tumors alone or in combination with other drugs in preclinical animal models and less toxic than Bcl-2 inhibitor or Bcl-2/Bcl-xL inhibitor to granulocytes like neutrophils 10,14 . Hence, it is reasonable and meaningful to evaluate the clinical application value of A-1331852 in terms of efficacy and safety. In summary, we define the molecular basis of Mcl-1 reduction and apoptosis resistance upon KPT-330 treatment. Based on this mechanism, we develop a potential therapeutic strategy combining KPT-330 and A-1331852 against solid tumors. Such treatment is effective regardless the clinical relevant mutation status (E571K and R749Q) of XPO1. These findings provide a strong rationale for its further investigation in the clinic. Western blot After collection, cells were lysed in RIPA supplemented with PMSF, phosphatase inhibitor, and protease inhibitor cocktail. Western blot was carried out as previously described 35 . Grayscale of protein bands was analyzed by Photoshop CS4 software. MTT assay Cell viability was measured by the MTT assay performed as previously described 35 . Flow cytometry assay To measure apoptosis, fixed cells were stained with propidium iodide as previously described 35 . Mitochondrial membrane potential was determined using the Mitochondrial Membrane Potential Assay Kit with JC-1 according to the manufacturer's protocol. After staining, cells were analyzed using a BD LSR II flow cytometer. Sub-diploid cells were considered apoptotic. The proportion of sub-dipliod cells were analyzed by FlowJo 7.6.1 software. Immunoprecipitation Cells were lysed on ice in cell lysis buffer for western and IP (20 mM Tris (pH 7.5), 150 mM NaCl, 1% Triton X-100). Lysates were adjusted to have equal protein concentrations and incubated with indicated antibodies overnight, and with additional protein A agarose or protein G agarose at 4°C for 3 h. Precipitates were washed three times with lysis buffer before adding SDS-PAGE loading buffer and denaturation. Precipitates and lysates were then subjected to western blot. Nucleoli isolation Nucleoli were prepared as previously described 36 . Briefly, Cells detached with trypsin/EDTA were lysed in hypotonic buffer (10 mM Hepes, PH 7.5, 10 mM KCl, 1.5 mM MgCl 2 , 0.5 mM DTT) with a Dounce tissue homogenizer until at least 90% of the nuclei had been released. The nuclear fractions were separated from cytoplasmic fractions by centrifugation at 300 g for 5 min at 4°C and purified by sucrose cushion (0.35 M sucrose, 0.5 mM MgCl 2 ) centrifugation. The nucleoli were prepared by sonicating the resuspended purified nuclear fractions in 0.35 M sucrose and 0.5 mM MgCl 2 solution and purifying with sucrose cushion (0.88 M sucrose, 0.5 mM MgCl 2 ) centrifugation. RNA extraction, real-time PCR, and RNA fragment analysis RNA was extracted, reverse transcribed, and analyzed by quantitative real-time PCR as previously described 35 . Primer sets for PCR are listed in the supplementary information. RNA fragment analyses were performed with the Fragment Analyzer™ Automated CE System (Agilent formerly Advanced Analytical Technologies, California, USA) following the manufacturer's protocol. Nascent RNA and protein synthesis assay Cells were plated in 96-well culture plates. After drug treatment, cells were incubated with 1 μM 5-ethynyl uridine (EU) for 1 h to probe nascent RNA or with 20 μM Opropargyl-puromycin (OPP) for 0.5 h to probe nascent protein. EU and OPP detection were performed according to the manufacturer's protocols. Fluorescence was measured by multi-mode microplate reader. FITC intensity was normalized against DNA-bound NuclearMask Blue stain intensity as the internal control. Bicistronic luciferase reporter assay The luciferase reporter assay was conducted using a dual-luciferase reporter assay system according to the manufacturer's protocol. Cells were transfected with a bicistronic luciferase reporter plasmid, pcDNA3-rLuc-polioIRES-fLuc, directing cap-dependent translation of Renilla luciferase gene and cap-independent polioIRESdependent translation of the firefly luciferase gene 37 using Lipofectamine 3000. After 24 h transfection, Cells were treated with 1 μM KPT-330 for 24 h. Luminescence was measured by multi-mode microplate reader. The renilla/ firefly luciferase luminescence was calculated for capdependent translational activity. Cap-binding assay Cell lysates were incubated with m 7 GTP agarose at 4°C for 3 h to capture eIF4E and its binding proteins. Following procedures were performed in the same procedure as immunoprecipitation. Isolation of cytosolic and nuclear RNA Cytosolic and nuclear RNAs were separated using the PARIS kit according to the manufacturer's protocol. Human NSCLC xenografts study Male NOD-SCID mice (5 weeks, Shanghai Lingchang Bioscience Company, China) were maintained in the pathogen-free environment. All experimental procedures were approved by the Institutional Animal Care and Use Committee of the Institute of Neuroscience, Chinese Academy of Science. H1299 cells (9 × 10 6 ) in 150 μl serumfree DMEM were injected under the skin of mice. When average tumor volume reached~50 mm 3 , mice were randomized into four groups (n = 5 per arm) and treated with KPT-330 (10 mg/kg, Monday, Wednesday, and Friday of every week) and/or A-1331852 (25 mg/kg, daily) by oral gavage. The vehicle for KPT-330 was 18.5% DMSO, 0.6% Pluronic F-68, and 0.6% PVP in water. The vehicle for A-1331852 was 2.5% DMSO, 10% ethanol, 27.5% PEG 300, and 60% Phosal 50 PG 14 . Tumor volume was calculated using the formula: V = 0.5 × length × width 2 . The treatment lasted for 10 days. One day after the final drug administration, mice were euthanized and tumors were isolated. Statistical analysis OriginPro 8 software (OriginLab Corporation, Northampton, USA) was used for data analysis and graphing. Results are expressed as mean ± SD or mean ± SEM. The one-way ANOVA test was used for nucleolar rRNAs processing tests and the two-tailed unpaired t-test was used for other tests to determine significant differences between the mean values of groups, with statistical significance defined as p < 0.05.	v3-fos-license
2020-06-08T13:05:54.612Z	2020-06-08T00:00:00.000	219528427	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GREEN", "oa_url": "https://www.medrxiv.org/content/medrxiv/early/2020/06/08/2020.06.06.20124123.full.pdf", "pdf_hash": "7f93cbc008cd3ba19038b4216eff810ec489a4d1", "pdf_src": "MedRxiv", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45447", "s2fieldsofstudy": [ "Medicine" ], "sha1": "7f93cbc008cd3ba19038b4216eff810ec489a4d1", "year": 2020 }	pes2o/s2orc	Clinical evaluation of self-collected saliva by RT-qPCR, direct RT-qPCR, RT-LAMP, and a rapid antigen test to diagnose COVID-19 Background The clinical performance of six molecular diagnostic tests and a rapid antigen test for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) were clinically evaluated for the diagnosis of coronavirus disease 2019 (COVID-19) in self-collected saliva. Methods Saliva samples from 103 patients with laboratory-confirmed COVID-19 (15 asymptomatic and 88 symptomatic) were collected on the day of hospital admission. SARS-CoV-2 RNA in saliva was detected using a quantitative reverse-transcription polymerase chain reaction (RT-qPCR) laboratory-developed test (LDT), a cobas SARS-CoV-2 high-throughput system, three direct RT-qPCR kits, and reverse-transcription loop mediated isothermal amplification (RT-LAMP). Viral antigen was detected by a rapid antigen immunochromatographic assay. Results Of the 103 samples, viral RNA was detected in 50.5-81.6% of the specimens by molecular diagnostic tests and an antigen was detected in 11.7% of the specimens by the rapid antigen test. Viral RNA was detected at a significantly higher percentage (65.6-93.4%) in specimens collected within 9 d of symptom onset compared to that of specimens collected after at least 10 d of symptom onset (22.2-66.7%) and that of asymptomatic patients (40.0-66.7%). Viral RNA was more frequently detected in saliva from males than females. Conclusions Self-collected saliva is an alternative specimen diagnosing COVID-19. LDT RT-qPCR, cobas SARS-CoV-2 high-throughput system, direct RT-qPCR except for one commercial kit, and RT-LAMP showed sufficient sensitivity in clinical use to be selectively used according to clinical settings and facilities. The rapid antigen test alone is not recommended for initial COVID-19 diagnosis because of its low sensitivity. in Japan [18]. On the day of admission, saliva specimens (~500 μ L) were self-collected by all Saliva specimens were diluted with phosphate-buffered saline at a volume 1-5 times in 1 1 2 accordance with the consistency and mixed with a vortex mixer. The suspension was 1 1 3 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020 centrifuged at 20,000 × g for 30 min at 4 °C and the supernatant was used in the following 1 1 4 molecular diagnostic and RAT. (NIID) protocol which is nationally recommended for SARS-CoV-2 detection in Japan [18]. primer and probe sets indicated the presence of viral RNA. Direct RT-qPCR methods without RNA extraction were performed using three Detection of viral RNA by automated RT-qPCR device The cobas SARS-CoV-2 test (Roche, Basel, Switzerland) [7,21] was performed on the SARS-CoV-2 RNA was defined as "detected" if targets 1 and 2 were detected or 1 4 5 "presumptive positive" if target 1 was not detected but target 2 was detected. RT-LAMP detection of SARS-CoV-2 was performed using a Loopamp® . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.06.20124123 doi: medRxiv preprint which was provided by a RAT kit into the saliva specimen and then into the sample of the antigen assay. Subsequently, 2 drops of buffer were added and the results were Detection of SARS-CoV-2 viral antigen by rapid antigen test interpreted after a 30 min incubation. The saliva sample collection day was defined as day 1. Symptomatic cases were Written informed consent was obtained from each enrolled patient at the Self-Defense Forces Central Hospital. This study was reviewed and approved by the Self-Defense Forces Central Hospital (approval number 02-024) and International University of Health and Welfare (20-Im-002-2). 6 9 Statistical analysis 1 7 0 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.06.20124123 doi: medRxiv preprint 1 0 Continuous variables with a normal distribution were expressed as mean (± SD) and Kruskal-Wallis test was used for nonparametric analysis with over three independent samples. Linear regression analysis was used to assess the relationship between each molecular sensitivity than either Method A or C. Only 12 patients tested positive using the RAT. The Ct values for the N-1 and N-2 primer sets for the direct RT-qPCR Method C (35.5 1 8 8 ± 2.2 and 34.8 ± 2.4, respectively) were significantly (p < 0.001) greater than those for LDT 1 8 9 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. was significantly lower in saliva samples that tested positive by RAT compared to that of 1 9 5 samples that tested negative (25.4 ± 1.8 vs. 30.8 ± 2.7, respectively; p < 0.001; Figure 2B). On the day of admission, 15 patients (14.6%) who did not display any symptoms were phase, and asymptomatic patients tested positive by molecular diagnostic tests 65.6-93.4%, from asymptomatic patients (p < 0.01). There were no significant differences in prevalence of 2 0 7 positive results by RAT among the three groups. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Effect of clinical background on the prevalence of viral RNA in saliva The baseline clinical characteristics of the 103 patients enrolled in this study are 2 1 0 presented in Table 2 The effect of clinical background against the prevalence of viral RNA in saliva was that tested positive for the virus compared to that of samples which tested negative (69.0% vs. 42.1%, respectively; p = 0.035) ( Table 2). There were no significant differences in 2 2 1 distribution by age or disease activity between patients detected or undetected with viral RNA 2 2 2 (p > 0.05). 3 A summary of clinical symptoms and disease severity is shown for 88 symptomatic 2 2 4 patients in with viral RNA in their saliva compared to 4 of 14 patients (28.6%) who did not test positive 2 2 6 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 8, 2020. Here, we present evidence for the clinical usefulness of saliva specimens in diagnosing collected during the early phase of symptom onset to increase sensitivity. the salivary gland and tongue tissues as well as nasal mucosa, nasopharynx, and lung tissue 2 4 5 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 8, 2020. affect the viral load in saliva and be associated with the difference in diagnostic sensitivity 2 5 0 between males and females. We did not observe significant differences in disease severity or was detected in more than 50% of the asymptomatic patients. These findings support asymptomatic patients [14]. Therefore, our findings revealed that saliva, collected in the early 2 5 7 phase of symptom onset, is a reliable and practical source for the screening and diagnosing of 2 5 8 COVID-19. 5 9 The clinical performance of direct RT-qPCR kits and RT-LAMP, and any correlation for SARS-CoV-2 using upper and lower respiratory tract specimens has been reported as 2 6 3 equivalent to 6,28]. However, our results indicate that the sensitivity of 2 6 4 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020 12. 5 6 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 8, 2020. Fisher's exact test for categorical variables. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 8, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 8, 2020. Fisher's exact test for categorical variables. 4 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.06.20124123 doi: medRxiv preprint	v3-fos-license
2018-07-06T13:13:02.090Z	2018-07-05T00:00:00.000	49573031	{ "extfieldsofstudy": [ "Biology", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://parasitesandvectors.biomedcentral.com/track/pdf/10.1186/s13071-018-2921-6", "pdf_hash": "62c16d1d781ab7b67f8cc874419137f0814e4035", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45451", "s2fieldsofstudy": [ "Biology" ], "sha1": "265254a23c97e5cd33ef324c909ec5a534be45a6", "year": 2018 }	pes2o/s2orc	Laboulbeniales (Fungi: Ascomycota) infection of bat flies (Diptera: Nycteribiidae) from Miniopterus schreibersii across Europe Background Bat flies (Diptera: Nycteribiidae and Streblidae) are obligate, blood-sucking ectoparasites of bats with specialized morphology, life-cycle and ecology. Bat flies are occasionally infected by different species of Laboulbeniales (Fungi: Ascomycota), microscopic fungal ectoparasites belonging to three genera: Arthrorynchus spp. are restricted to the Eastern Hemisphere, while species of Gloeandromyces and Nycteromyces occur on Neotropical bat flies. Little is known about the distribution and host specificity of Arthrorynchus spp. on bat flies. In this study, we focused on sampling bat flies from the cave-dwelling bat species Miniopterus schreibersii. Bat and ectoparasite collection took place in Albania, Croatia, Hungary, Italy, Portugal, Slovakia, Spain and Switzerland. Flies were inspected for Laboulbeniales infections. Results Six hundred sixty seven bat flies of five species were collected: Nycteribia latreillii, N. pedicularia, N. schmidlii, Penicillidia conspicua, and P. dufourii. Laboulbeniales infection was observed on 60 specimens (prevalence = 9%). Two Laboulbeniales species, Arthrorhynchus eucampsipodae and A. nycteribiae, were present on three bat fly species. All observations of A. eucampsipodae were on N. schmidlii, and A. nycteribiae was present on P. conspicua and P dufourii. Arthrorhynchus eucampsipodae is, for the first time, reported from Slovakia and Spain. Arthrorhynchus nycteribiae represents a new country record for Portugal and Slovakia. There were no significant differences among infection rates in different countries. Females of N. schmidlii showed a higher infection rate than males with an observable trend (P = 0.0502). No sex differences in infection rate for P. conspicua and P. dufourii were detected. Finally, thallus density was significantly lower in N. schmidlii compared to P. conspicua and P. dufourii. Conclusions With this study, we contribute to the knowledge of the geographical distribution and host specificity of Laboulbeniales fungi associated with ectoparasitic bat flies within Europe. We discuss parasite prevalence and host specificity in the light of our findings and the available literature. Penicillidia conspicua is unambiguously the main host species for A. nycteribiae based on our and previous findings. Differences in parasite intensity and sex-biased infections of the fungi are possible depending on the species. Electronic supplementary material The online version of this article (10.1186/s13071-018-2921-6) contains supplementary material, which is available to authorized users. Background The distribution of parasites is shaped by the distribution of their hosts, although complete overlaps are infrequent. Hosts can lose their parasites or gain new ones when colonizing new areas [1,2]. Whether parasites are lost or gained is driven by a combination of abiotic and biotic factors. Abiotic factors, such as climate or habitat type can strongly affect parasite occurrence [3]. Biotic factors, for instance host behavior or immune response to parasitism, may be essential in determining factors in parasite distribution [4,5]. Studying geographical differences in parasite distributions is the first step in understanding how parasite loss or gain is shaped. Bats represent the second largest mammal order with a worldwide distribution and have highly specific and diverse micro-and macroparasites [6], which can be subject to parasites of their own [7]. These multilevel trophic systems may be shaped by many parameters, such as the ecology, immunology, behavior and sex of the bat hosts as well as their parasites (e.g. [4,8,9]). This study focuses on Miniopterus schreibersii, a cave-dwelling bat which is widely distributed across southern Europe, Asia Minor and North Africa, and represents the only European member of its genus [10]. Miniopterus schreibersii hosts a myriad of highly specific parasites as for example Spinturnix psi mites (Acari: Mesostigmata: Spinturnicidae), Nycteribia schmidlii and Penicillidia conspicua bat flies (Diptera: Hippoboscoidea: Nycteribiidae), or Polychromophilus melanipherus blood parasites (Alveolata: Apicomplexa: Plasmodiidae) [9,11,12]. Additionally, non-specific ectoparasites, such as the bat fly species Penicillidia dufourii, Nycteribia latreillii and N. pedicularia, can parasitize M. schreibersii. Even though these non-specific associations are considered the result of accidental host choice by the parasites, they cannot be considered rare [9]. Altogether, M. schreibersii represents an outstanding target species in parasitology research. Bat flies are obligate blood-sucking ectoparasites of bats belonging to two families, the Nycteribiidae and the Streblidae. In Europe, 16 nycteribiid and one streblid species have been reported so far [9]. The morphology and life-cycle of bat flies are unique among Diptera. Nycteribiids are wingless and possess reduced ocelli, both being adaptations for living in the fur of their hosts. Bat flies give birth to a single third-instar larva (larviposition) on the roost wall of their hosts. Before larviposition, the larva develops in the uterus-like organ of the female, nourished by milk-glands. The larva immediately pupates after larviposition and the emergence of imagoes from pupae is influenced by the presence of bats; after emergence bat flies actively search for bat hosts [6]. Laboulbeniales are ectoparasitic fungi that associate with representatives of three subphyla of Arthropoda: Chelicerata, Myriapoda and Hexapoda [13]. Of all described Laboulbeniales species, 80% are associated with Coleoptera and 10% with Diptera. The rest occur on many other different taxa, belonging to Arachnida, Diplopoda and Hexapoda [13]. The Laboulbeniales are different from most other fungal groups because they lack hyphae and instead form microscopic fruiting bodies or thalli as a result of determinate growth. Most Laboulbeniales are moderately to extremely host-specific. Many members of the order are associated with a single host species or several species of a same genus. Some species exhibit position specificity and are found on determined parts of their host's integument [14]. What drives host specificity in nature is unknown. Ecological specificity is the last but most interesting level; shifts between phylogenetically unrelated hosts that share the same microhabitat is a significant trigger for speciation [15][16][17]. Three genera and eight species of Laboulbeniales are known from bat flies [7,18], with the potential of many undescribed species, especially in the Neotropics [19]. The genera Gloeandromyces and Nycteromyces are reported on streblid bat flies from Central and South America. Arthrorhynchus is apparently restricted to the Eastern Hemisphere and has only been reported from Nycteribiidae. Four species are known: Arthrorhynchus acrandros, A. cyclopodiae, A. eucampsipodae and A. nycteribiae (although A. acrandros is disputed [7]). Arthrorhynchus nycteribiae has been most widely reported [7,19]. It is known in Europe (Austria, Bulgaria/Slovakia, Croatia, the Czech Republic, "Czecho-slovakia" [sic.], Denmark, France, Hungary, Italy, Poland, Romania, Russia, Serbia, Spain, Sweden, Switzerland and the Netherlands), Africa (Kenya and Zambia), Asia (Sri Lanka) and Oceania (Australia). In a recent study, 1494 bat flies (11 spp.) from 1594 bats (28 spp.) collected in Europe were screened for the presence of Laboulbeniales fungi [7]. Many bat flies parasitized by these fungi have been collected from M. schreibersii. The prevalence of Laboulbeniales on bat flies from this species was highly disparate among bat fly species. Nycteribia schmidlii had a parasite prevalence of 3.1% (n = 147), and was infected by both Arthrorhynchus eucampsipodae and A. nycteribiae. Penicillidia conspicua had a prevalence of 23.1% (n = 142); only A. nycteribiae was found. All P. dufourii from M. schreibersii were found to be uninfected (n = 22). In the present study, we expanded capturing efforts of M. schreibersii to focus on Laboulbeniales infections of specific versus generalist bat flies. Using this tripartite system, we attempted to assess parasite distributions within and between host populations. Collection of bats and bat flies Bats were captured from April through September in 2009-2016, using mist nets and harp-traps, placed in front of caves where M. schreibersii colonies occur. Sampling took place in Albania, Croatia, Hungary, Italy, Portugal, Slovakia, Spain and Switzerland (Fig. 1). Exact localities are given in Additional file 1: Table S1. Age, sex, and morphological characteristics were collected for most individuals. Bat flies were removed with forceps and placed in 70-90% ethanol. Bats were released immediately after processing in the vicinity of the capture site. Bat fly species and sex were determined following Theodor's key (1967) [21]. Voucher bat fly specimens are deposited under the accession number 16CH12-XB07 at the Museum of Zoology, Lausanne, Switzerland. Collection and identification of Laboulbeniales The presence or absence of Laboulbeniales was determined using a stereomicroscope (Leica M205C, Leica Mcrosystems AG, Heerbrugg, Switzerland). For each infected bat fly, the total number of Laboulbeniales thalli was counted. Thalli were removed from the host at the point of attachment with an entomological pin and slide-mounted in Amman's solution for identification [22]. Identification was based on the original descriptions and drawings by Thaxter [23] and recent amendments by Haelewaters et al. [7]. Slides will be deposited at the mycology herbarium of Ghent University, Belgium (Ghent). Statistical analyses Fisher's exact tests were used for prevalence comparison among countries, bat fly species and sexes. In addition, Mood's median tests were performed to compare the median thallus density and bootstrap tests were used to compare mean thallus density of Laboulbeniales among infected bat fly hosts (based on 1000 bootstrap replications), performed in Quantitative Parasitology v.3.0 [24]. Bat flies and Laboulbeniales We collected 667 bat flies from 270 M. schreibersii bats. Five bat fly species were encountered: N. schmidlii (n = 468), P. conspicua (n = 144), P. dufourii (n = 52), N. pedicularia (n = 2) and Nycteribia latreillii (n = 1). Of all bat flies, 60 specimens were infected with Laboulbeniales fungi (prevalence of 9%). Nycteribia latreillii and N. pedicularia were represented in very low sample numbers and were uninfected with Laboulbeniales, therefore we excluded them from further analyses, figures and tables. Arthrorhynchus eucampsipodae was found in Hungary, Slovakia and Spain (Table 1). All observations of A. eucampsipodae were made on a single bat fly host, Nycteribia schmidlii. Additionally, we reported A. nycteribiae from five countries: Croatia, Hungary, Portugal, Slovakia and Spain (Table 1). Hosts for A. nycteribiae were P. conspicua and P. dufourii. The presence of A. eucampsipodae is reported here, for the first time, from Slovakia and Spain. Additionally the presence of A. nycteribiae is also a new record from Portugal and the first undoubtful record from Slovakia. Prevalence rate in different countries In Albania, Italy and Switzerland, Laboulbeniales infection was not detected among the 361 collected specimens of Nycteribia schmidlii, Penicillidia conspicua and P. dufourii. The highest parasite prevalence was observed in Slovakia (25.6%). We found the lowest overall parasite prevalence in Portugal (10%, only P. conspicua sampled). In Croatia, Hungary and Spain, overall fungal prevalence was 11.7%, 15.8% and 11.9%, respectively. There were no significant differences in parasite prevalence between the different countries. Parasite prevalence and host specificity Of the 468 collected N. schmidlii, 23 flies were infected (4.9%) with Laboulbeniales. We sampled 52 specimens of P. dufourii of which 4 individuals carried Laboulbeniales (7.7%). Of 144 P. conspicua specimens, 33 were infected (22.9%). Penicillidia conspicua had a significantly higher parasite prevalence compared to the other two species (Fisher's exact test, P < 0.0001). Infection by A. eucampsipodae was found exclusively on N. schmidlii, while A. nycteribiae infection was detected only on Penicillidia species. Differences in parasite prevalence between female and male bat flies Of 269 females and 199 males of N. schmidlii, we found 18 infected females and 5 infected males, which shows a marginally significant trend in the infection between the sexes (6.7 and 2.5%, respectively; Fisher's exact test, P = 0.0502). Of 81 females and 63 males of P. conspicua, 21 females and 12 males were infected (25.9 and 19%, respectively; Fisher's exact test, P = 0.424) with Laboulbeniales. Of P. dufourii, 25 females and 27 males were collected and only two individuals were infected for each sex (8 and 7.4%, respectively; Fisher's exact test, P = 1.0; Fig. 2), therefore the prevalence is not significantly different between the sexes, neither in P. conspicua, nor in P. dufourii. Geographical distribution, host range, and prevalence of Laboulbeniales The geographical distribution and host range of bat-fly associated Laboulbeniales in Europe has been reported by several studies [7,20,25,26]. Our study presents additional occurrence data focusing on the tripartite system of the M. schreibersii cave-dwelling bat, its ectoparasitic Blackwell [20] reported the prevalence of Arthrorhynchus species to be 2.2% (n = 2517). Regarding the bat fly species discussed in our study, she found parasite prevalences of 2.5% on N. schmidlii (n = 316), 18.6% on P. conspicua (n = 86) and 0.7% on P. dufourii (n = 289). Haelewaters et al. [7] reported a total prevalence of 3% on all screened bat flies (n = 1494) but the parasite prevalence varied depending on the host species. These and our results allow us to suggest that these low (N. schmidlii, P. dufourii) to moderate (P. conspicua) infection rates in these species are not particularly variable. Based on our study and the recent work by Haelewaters et al. [7], A. eucampsipodae seems to be highly specific towards N. schmidlii. However, Blackwell [20] reported other host species for A. eucampsipodae with prevalences ranging between 0.3-8.3%. Although A. eucampsipodae displays preferences for N. schmidlii, strict host specificity does not seem to be the rule for this Laboulbeniales species. The same specificity pattern is observed for A. nycteribiae [7,19]. Based on our results, we can conclude that Penicillidia dufourii is merely a secondary host species for A. nycteribiae compared to P. conspicua. This confirms findings by Haelewaters et al. [7] who reported a prevalence for A. nycteribiae of 25% on P. conspicua (n = 152), whereas on P. dufourii prevalence was much lower (2.0%, n = 102). In addition, Blackwell's [20] data, also show that P. conspicua has the highest prevalence with A. nycteribiae (18.6%, n = 86). Taken together, P. conspicua is unambiguously the main host species for A. nycteribiae, but this fungus probably has the capacity to also grow on many other bat fly hosts. Although Laboulbeniales prevalence on bat flies varied among countries, we did not find significant differences. Since infection rates can be influenced by habitat type on a smaller geographical level, such as differences between wet or dry habitats (see [26] and references therein), future work should focus on identifying factors that shape the distribution of infection. Prevalence of Laboulbeniales between bat fly sexes In parasitological studies, it is widely observed that different sexes often show different infection rates throughout several taxa [8,27,28]. Recently, it was found that among bat flies females are more likely to be infected with Arthrorhynchus spp. compared to males [7]. In our study, we only found a trend supporting this observation in the case of N. schmidlii (P = 0.0502), but no support was found for P. conspicua and P dufourii (Fig. 2). A commonly reported species of Laboulbeniales for which infection patterns can be significantly different between male and female hosts is Hesperomyces virescens, a parasite of ladybirds (Coleoptera: Coccinellidae). Sexual differences (e.g. prevalence and/or position specificity) in H. virescens infection on Harmonia axyridis ladybirds are presumed to be the result of host mating behavior [29][30][31]. Regarding bat flies, females are known to live longer [32]. Pregnant bat flies are significantly larger than male flies, accounting for more integument surface, and have fat reserves organized as lobes in their haemolymph, presenting higher nutritional resources [33]. Different sexes can exhibit different levels of parasite resistance [27,33,34]. These differences are most commonly explained by variances in hormone levels between sexes, for example steroid reproductive hormones [28]. In conclusion, sex bias in parasitism can occur also as a consequence of the different immune status of the hosts. Conclusions During this study, we collected and analyzed the occurrence of Arthrorhynchus spp. in a wide range of geographical distributions within Europe on the cave-dwelling bat species, Miniopterus schreibersii. Five bat fly species were collected in eight countries, on which three species showed fungal infections, each with a different parasite prevalence. Prevalence can differ among host species and between host sexes. Arthrorhynchus eucampsipodae was only reported on N. schmidlii. In addition, A. nycteribiae was observed on two Penicillidia species, of which P. conspicua appeared to be the "main host" for this fungus, while P. dufourii is considered a secondary or accidental host. Our work has also resulted in new country records: A. eucampsipodae is newly reported from Slovakia and Spain, while A. nycteribiae is newly reported from Portugal and represents the first undoubtful record for Slovakia. Additional files Additional file 1: Table S1. Authors' contributions TS initiated the study and identified the bat fly parasite specimens. WPP identified fungal specimens. LC, TS, PC and OG participated in parasite collection. TS, DH, WPP, PC and OG contributed toward writing the first draft of the manuscript. All authors read and approved the final manuscript. Ethics approval Animal capture was conducted according to the Swiss Animal Legislation (legislation number 2964).	v3-fos-license
2022-04-23T15:07:30.901Z	2022-04-20T00:00:00.000	248341627	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2076-3417/12/9/4136/pdf?version=1650446546", "pdf_hash": "4393cc544b133c530bff14da1497b2f61c3ade4f", "pdf_src": "MergedPDFExtraction", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45456", "s2fieldsofstudy": [ "Computer Science" ], "sha1": "ce4fee4ff035b0a7f02df456c3238bf8b94102b6", "year": 2022 }	pes2o/s2orc	Remaining Useful Life Estimation of Rotating Machines through Supervised Learning with Non-Linear Approaches : Bearings are one of the most common causes of failure for rotating electric machines. Intelligent condition-based monitoring (CbM) can be used to predict rolling element bearing fault modes using non-invasive and inexpensive sensing. Strategically placed accelerometers can acquire bearing vibration signals, which contain salient prognostic information regarding the state of health. Machine learning (ML) algorithms are currently being investigated to accurately predict the health of machines and equipment in real time. This is highly advantageous towards reducing unscheduled maintenance, increasing the operational lifetime, as well as mitigation of the associated health risks caused by catastrophic machine failure. Motivated by this, a robust CbM system is presented for rotating machines that is suitable for various industrial applications. Novel non-linear methods for both feature engineering (one-third octave bands) and wear-state modelling (exponential) are investigated. The paper compares two main types of feature extraction, which are derived from Short-Time Fourier Transform (STFT) and Envelope Analysis (EA). In addition, two types of supervised learning, Support Vector Machines (SVM) and k -Nearest Neighbour ( k -NN) are explored. The work is tested and validated on the PRONOSTIA platform dataset, with remaining useful life (RUL) classiﬁcation results of up to 74.3% and a mean absolute error of 0.08 achieved. Introduction Electric machines are a vital component for major industries, such as manufacturing, mining, agriculture, energy and transport sectors. It is fair to say that all of these sectors are currently undergoing major growth and technological innovation. These machines are typically required to operate under harsh environmental conditions and demanding drive cycles, which gives rise to premature degradation and the occurrence of catastrophic failure modes [1,2]. It is imperative for the future viability and sustainability of these sectors that we have efficient, robust and highly reliable electric machines. Sudden catastrophic machine breakdown results in acute manufacturing downtime, dramatic reductions in productivity and health and safety concerns. Moreover, performing critical maintenance is both labour-intensive and costly. Faults can be difficult to diagnose and troubleshoot for maintenance teams [3][4][5]. Broadly, electric machines can be broken down into the bearings, stator, rotor and other elements as shown in Figure 1a. The statistics of failure for three classes, low voltage, medium voltage and high voltage, are presented in Figure 1b. Bearings are the dominant failure mode at low and medium voltages, followed closely in the latter class by stator fault modes. The final class, high voltage, is dominated by stator fault modes, as seen in Figure 1b. Research studies have shown bearings to be responsible for up to 75% of lowvoltage electric machines breakdowns and up to 41% of all rotating machine failures [2,[6][7][8][9]. Condition-based monitoring has received considerable attention over the past years [2,[10][11][12]; hence, a rich literature exists at present. The areas of detection and diagnosis of fault modes has received the most attention with mature industrial technologies available. More recently, advanced methods of prognosis are being investigated, and these focus on the more challenging problem of predicting the Remaining Useful Life (RUL) of the machines or sub-components [13][14][15]. This is illustrated in Figure 1c. Knowing the RUL of a component ensures that the state of health of a machine is known and that suitable maintenance can be performed at the optimal times. The maximum usable life is achieved without the threat of total machine breakdown occurring [16][17][18][19]. The Short-Time Fourier Transform (STFT) feature extraction method has been extensively used to extract useful time-frequency features, which reported to achieve high levels of classification accuracy [32][33][34][35]. Envelope Analysis (EA) has also been used extensively for prognostic and diagnostic purposes as the method is simple yet versatile, making it applicable to many different types of mechanical fault monitoring processes [21,22,[36][37][38]. Other methods of time-frequency feature extraction, which have shown good promise for bearing prognostic use cases are Wavelet Transformation (WT) [39][40][41][42] and Empirical Mode Decomposition (EMD) [43][44][45][46][47][48], both of which were reported to achieve highly accurate performance scores. In [59], an RUL prediction method was proposed based on a long short-term memory (LSTM) neural network framework and deep features, which were learned adaptively from the two health states. Benali [48] proposed a method to characterize and classify seven different bearing classes using statistical features, EMD energy entropy and an artificial neural network (ANN). Li [56] used a combination of two supervised ML techniques; a regression model and multi-layer ANN to predict the RUL of rolling element bearings. Here, in this paper, a novel ML method for RUL is proposed, using non-linear signal processing techniques to perform feature engineering based on STFT and EA with Onethird octave band feature compression. The rationale for using Fourier and EA-based feature extraction was as a result of detailed vibration signal analysis conducted on the bearing signals from the dataset. This motivated the incorporation of non-linear feature compression of the multidimensional feature space using Octave bands as the prognostic information signatures are highly concentrated in the lower portion of the spectra. These features form part of the ML method recipes alongside twelve different supervised learning algorithms based on k-NN and SVM to determine the optimal choice for RUL classification. SVM and k-NN algorithms were chosen because of their robustness for supervised learning problems, in particular problems with datasets of limited size, which greatly limits the suitability of applying other more advanced deep-learning approaches, e.g., ANN and LSTM. The work also highlights the importance of using non-linear wearstate models to track the degradation severity levels; this has been shown to greatly improve the performance of the ML classifiers overall for this RUL task. The time frequency analysis conducted on the vibration signals also motivated the investigation of non-linear wear state models as the bearing degradation typically does not follow linear trends. The remainder of this paper is organised as follows. Section 2 presents a graphical and statistical analysis of the vibration signals. The proposed ML method is detailed in Section 3. The experimental procedure is illustrated in Section 4. In Section 5, the results from the proposed CbM system are presented. Section 6 presents the major trends and findings from the results and statistical analysis, and the limitations of this work are discussed. Finally, Section 7 concludes the work and highlights future research avenues to explore. Vibration Signal Analysis The typical degradation of bearings is a gradual and slowly evolving process. Ordinarily, it would take many years to acquire signals characterising the entire process, starting with a new bearing and progressing to a fully degraded bearing at the end of its life cycle. Typically, applied experiments are performed in the laboratory under controlled conditions to acquire bearing data. These accelerated ageing experiments can involve artificially inducing faults by strategically drilling small holes or etching the surface of the bearings, applying excessive loads and operating speeds, or elevating the temperature and humidity, as reported in [60,61]. Dataset The proposed condition monitoring methods described in this research were tested and validated using the award winning FEMTO-ST Institute 2012 PHM Challenge dataset [62], which has found widespread use in this field [63][64][65]. This complete dataset comprises of vibration signals from the accelerated degradation of 17 bearings, performed at three different operating speed and load conditions: (1800 rpm and 4000 N), (1650 rpm and 4200 N) and (1500 rpm and 5000 N) [62]. This dataset provides realistic fault modes achieved under accelerated ageing conditions as opposed to data generated from artificial fault modes, e.g., drilled holes, machined narrow cuts or indentation lines to emulate the occurrence of hairline cracks. Accordingly, for each test case, the details on the specific type of failure mode or element (e.g., ball, inner race or outer race or cage) is not known or provided. This work is more concerned with monitoring degradation rather than diagnosis of the specific failure mode. The test setup was composed of three main parts: a rotating part, a degradation generation (loading) part and a sensing measurement part. The bearing's vibration amplitudes were recorded by two miniature accelerometer sensors, Dytran, Model 3035B, placed orthogonally to one another on the vertical and horizontal axis of the bearing under test. This sensor pair was placed radially on the external race of the bearing. The acceleration measures were sampled by the analogue to digital converter (ADC) at 25.6 kHz [62]. Each recording consists of bursts of 2560 samples (0.1 s duration), which were obtained every 10 s throughout the test bearings' lifetimes up until the point of failure. Failure for these test bearings was defined to be the point where the amplitude of the vibration signals surpassed the reference acceleration threshold value of 20 g. This threshold was carefully chosen as it also avoided any considerable propagation that could severely damage the test-bed mounts and fixtures [62]. Signal Analysis The testing and validation of the proposed wear state estimation system was focused towards the first operating condition, with speeds of 1800 rpm and an applied force of 4000 N, for seven bearing test cases. The duration of each test case varied in length with the longest test case reaching almost 8 h and the shortest lasting only 2 h and 25 min as illustrated in Table 1 and Figure 2. Since the run-to-failure test was conducted with no artificial mechanical tampering of the constituent bearings, it can be expected that a spread of different fault types would have occurred in the bearings, involving the rolling elements, the cage and the inner-race and outer-race parts. [62]. The MA interval is 512 points and a reference level of 1 µm/s 2 was used, as per [66]. Figure 3 demonstrates examples of typical time domain vibration signals where different fault types occur. The signal depicted in the three top panels show a very gradual increase in the vibration amplitude for bearing S. 01 before the fault occurs. The fault in bearing S. 04, depicted in the three lower panels, manifests itself as a sudden change in the vibration amplitude about three quarters through the lifetime. This indicates the occurrence of a very different, abrupt fault mode, such as the rapid formation of a catastrophic crack, a part snapping or sudden deformation due to heat induced by friction. These two examples, that of the slowly evolved degradation mode and the rapid formation fault mode, go some way to demonstrate the inherent degree of complexity and challenges that exist in developing robust condition-monitoring systems using machinelearning methods. It is difficult to ascertain trends and patterns in the time domain signals for the vibration amplitudes alone. Hence, a conversion to the frequency domain is necessary to observe spectral signatures and trends throughout the duration of the test to failure. Accordingly, a Short-Time Fourier Transform (STFT) process was applied to each bearing test case with the parameters set to produce a multivariate spectral description of the data. Table 2 details the STFT algorithm parameters used in this work. The concept of STFT analysis is fundamental for describing any quasi-stationary (slowly time varying) signal. In general, one can define the STFT in terms of the output of an arbitrary bank of filters. The amplitude spectrum of each frequency component of the signal was converted to a decibel scale. Figure 2 illustrates the variance of mean STFT spectral component energies for each time sample for all 7 of the bearing test cases, using the first 50 time samples of each as their reference baseline value. The natural degradation trend of a bearing does not represent a gradual, linear pattern as represented in Figure 2. All seven bearings test cases vary in the number of time samples as no two experiments were equal in duration, however, the avalanche-like pattern of degradation is apparent in all cases. This sudden effect makes the RUL estimation more difficult as the wear states leading to this stage share a great deal of the same feature values, and some do not vary from their initial baseline values until roughly 75% of their usable lifetime has passed. The shaded error bar illustrates the mean frequency spectral amplitudes ± one standard deviation. Proposed ML Method This study investigated and compared a variety of algorithm options for bearing wear-state classification. The proposed method begins by taking raw accelerometer data and concludes with assigning a predicted wear state class. A number of intermediate steps are involved. This section describes each method stage and describes how each stage can branch into alternative steps as illustrated in Figure 4. Feature Extraction The proposed method begins with a feature extraction step and two feature extraction methods were performed for comparison. The two techniques of extracting classification features from the raw time-series data were: (1) applying a STFT to the discrete-time signals and (2) calculating the signal envelopes of each discrete-time signal. Discrete-Time Signal Analysis The non-stationary time series data recorded from the 2012 PHM Data Challenge bearing dataset are presented as a 2-D vector. The vibration amplitude sampled at 25.6 kHz by the Dytran Model 3035B accelerometer was transferred from the discrete-time domain to the frequency domain using the Short-Time Fourier transform (STFT) parameters detailed in Table 2. The Short-Time Fourier transform (STFT) can be defined as a sequence of Fourier transforms of a windowed input signal. STFT is used to extract time-localized frequency information, for situations in which frequency components of a signal vary drastically over time, such as non-stationary bearing vibration signals [67]. The STFT, shown in Equation (1), involves calculating a windowed Fast Fourier transform (FFT) of the discrete time samples, with each window overlapping the previous by a factor of 75% to obtain the complex feature vector signatures across time. The average value of these absolute complex feature vectors for each sample is calculated to extract a spectral feature vector consisting of 512 spectral points (bins). The spectral points are spaced at 25 Hz intervals and represent the spectral amplitudes content over a range of 0 to 12,800 Hz (Nyquist frequency). This frequency range is determined by specifying the number of Discrete Fourier transform (DFT) points calculated for each window and by the sampling frequency of the DFT. The sampling frequency was matched to the frequency of the accelerometers sampled the bearing signals at, 25.6 kHz, and the number of DFT points calculated for each window was 1024 points (the same value as the length of the sample window g(n)). is the DFT of windowed data centred about time mR, g(n) is a window function, and R is the hop size between successive DFTs. The STFT has been favoured as a feature extraction method to obtain useful features for both RUL estimations and fault classification and achieved extremely high results as seen in [34,35,68,69]. Envelope Analysis An Envelope Analysis (EA) approach, involving the extraction of the signal envelope of the discrete-time vibration signals, was also used as an alternative method to extract timefrequency features. Two different filtering approaches were applied to the non-stationary signals, linear and non-linear. The linear filtering process uses equidistant frequency spacing, whilst the non-linear option involves applying the one-third octave band scale to the vibration signal in the time domain. This produced 25 representations of the signal for both the linear and the non-linear cases for classification performance comparison. The filter employed was a 25 Finite Impulse Response (FIR) sixth order Butterworth filter. The discrete-time domain signal is demodulated by taking the absolute value of the non-stationary signal points, to produce x r [n]. By taking the Hilbert Transform of this rectified signal, we can produce the signal, x i [n], which enables the creation of the complex analytic signal, defined as z(n) and shown in Equation (2). The final step to calculate the envelope signal is to take the magnitude value of the complex analytic signal, as described in Equation (3). where EA[n] represents the envelope signal. Feature Compression Feature compression was applied to both the discrete-time STFT-and EA-generated spectral feature sets referenced in Sections 3.1.1 and 3.1.2. For the STFT features, dimensionality reduction from 512 down to 25 spectral features was applied in order to extract the most useful features to train the learning model algorithm. In addition, dimensionality reduction simplifies the complexity of the computations so that more optimal and accurate estimations could be obtained. This was to avoid the well-known phenomena often defined as the curse of dimensionality [70,71]. This term describes the inherent problem caused by the exponential increase in volume associated with adding extra dimensions to a Euclidean space [71]. Feature reduction was achieved using a filter band approach as described in Equation (4) for both linear and non-linear sized bands. The linear bands, L[m, k], consisted of equidistant bands applied evenly to 512 spectral features, whereas the non-linear approach, O[m, k], comprises a one-third octave band filter being applied. The one-third octave band filter places a higher emphasis on the lower end of the frequency spectrum by having smaller bands that increase non-linearly in size as the frequency increases. where Wear-State Temporal Models When analysing the horizontal vibration signal from the bearing's external race, a degradation trend can be identified in the signal amplitude values in the discrete-time domain. The degradation trend can be identified as having a non-linear increase in amplitude as the bearing failure stage is approached. In this study, five temporal wear-state classes were used to characterise the RUL of the seven bearings under test. Two different wear-state models, one linear and one non-linear, were considered and tested to determine the optimum scale to determine the RUL of the components accurately and robustly. The linear wear-state model divides the data into five equidistant temporal classes, as illustrated by Equation (5), where each temporal class represents a 20% portion of the bearing's overall lifetime. In contrast, the non-linear wear-state approach uses five classes that are strategically spaced to add greater granularity or compression to the class boundaries towards the latter stages of the bearing's life. This is achieved as follows: the first 63% of the bearing's lifetime is allocated to class 1 (healthy), the second class consists of the 86% of the bearing's lifetime and so on as presented in Equation (6). where α i and β i define the linear and non-linear temporal class boundaries, respectively, and the index i = {1, 2, 3, 4} corresponds to those class numbers. Note: α 5 = β 5 = 1 as the boundaries are normalised with respect to time. Figure 5 shows a graphical representation of the linear and non-linear wear-state class boundaries. Classification Supervised ML algorithms were used to detect trends and patterns in the pre-processed data and classify the health wear-state of the bearing test samples. Two widely used classical methods of ML were studied-that of support vector machines (SVM) and k-Nearest Neighbour (k-NN). Support Vector Machines A support vector machine (SVM) is a supervised ML algorithm. The SVM algorithm is used to classify pre-labelled test cases (targets) by analysing the training cases (predictors) and finding a separator between classes. The use of SVM-classification algorithms has been recorded to achieve highly accurate results in high dimensional feature spaces [52,53,55,72,73]. Another key benefit that the SVM algorithm option offers is memory efficiency. Only a small subset of the training features, the support vectors, are required to calculate the location of optimised hyperplanes between wear-state classes. The pre-labelled training data, also referred to as the predictor features, is mapped to a higher-dimensional feature space, so that data points can be categorised as shown in Figure 6. This mapping process is often referred to as kernelling, as the transformation can be achieved through the application of various kernel functions. The predictor features are transformed in such a way that the separator can be formed as a hyperplane, which can be considered as a line function representing the largest separation, or margin, between classes (wear-states). The data points whose positions lie closest to the calculated hyperplane are identified as support vectors. To achieve the most accurate SVM prediction model, the hyperplane should be at the maximal distance possible from the nearest support vectors. This distance from a support vector to the hyperplane is identified as the margin. The classification of target instances is achieved by inputting the unseen, feature data values, without their corresponding wear state label, into the hyperplane function shown in Equation (7). where w represents the weight vector, y is the input vector and b is the bias. The result determines whether the data point is an instance of the class above or below the hyperplane. This means that only a fraction of the overall predictor feature points are processed for calculation unlike other ML methods, such as Decision Trees, Logistic Regression, Naive Bayes and k-NN classification, which require all feature points to be included in each calculation. This significantly reduces the complexity of calculations while increasing the efficiency and speed of producing RUL estimations. The biggest challenge associated with the SVM classification algorithm as a prediction model is its tendency to over-fit data. Over-fitting would be most prevalent when the feature number (dimensionality) is high relative to the number of predictor instances. To counteract this, numerous bearing test-cases with large numbers of time-samples are used to train and test the performance of the SVM classification algorithms. The performance of six different kernel functions were investigated and compared in this study. Kernel options used to map the data into a higher dimensional feature space included Linear, Quadratic, Cubic, Fine Gaussian, Medium Gaussian and Coarse Gaussian functions. k-Nearest Neighbour The k-Nearest Neighbour (k-NN) classification algorithm is one of the most widely used supervised ML methods for categorising unknown signals into a discrete set of classes [49][50][51]. The k-NN method is used to classify target instances based on their similarities to predictor features (training data). The most similar predictor cases are referred to as the "neighbours", hence, the title associated with the classification method. Classification is achieved by first defining a value of k. The optimal k value is dependant on the input data. Choosing a low k value often produces an over-fitted prediction model, which produces inaccurate predictions on out-of-sample target instances. A higher k value makes the prediction model too generalised as the classes with more predictor instances become prioritised as the target instance. The optimal k value may only be determined through trial and error, using multiple values to compare accuracy results. The next step to achieve a RUL prediction using the k-NN framework involves calculating the distance from the features of the target instance from all other predictor instance features. This distance metric can be calculated in a number of ways, including Euclidean, Mahalanobis, City block and Minkowski distances. As we are dealing with distance metrics to determine classification, it is important that each of the features used are standardised before training the prediction model. Min-max normalisation is applied to each feature to put all training data into the 0-1 scale. The target instance is then normalised using the same min-max values as the training data. The same min-max value is used for the normalization process to eliminate the occurrence of data dredging, the statistical inference performed after looking at the complete dataset. The k nearest observations in the training data that are nearest to the unknown target data point are selected as the "neighbour" points. A case is classified by determining the mode class value of its neighbours, with the case being assigned to the class most common amongst its k nearest neighbours measured by a distance function. Six methods of k-NN classification were used to obtain RUL predictions for the bearings, including Fine, Medium, Coarse, Cosine, Cubic and Weighted k-NN. The specified parameters varied in each method are presented in Table 3. Experimental Procedure This section describes the experimental procedure, which can be summarised under three main strands: the ML method recipes, the round robin framework and the performance metrics. ML Method Recipes The experimental procedure for this research involved varying the following parameters as described in the previous section: (a) feature extraction using either STFT of the discrete-time domain signal or the envelope of the vibration time-series data, (b) feature selection using full spectra from 0 to 12,800 Hz X[m, f ], linear bands L[m, k] or one-third octave bands O[m, k] as feature vectors, (c) the wear-state classification model using either linear, α, or non-linear, β, temporal class boundaries, (d) model training and testing using a SVM or k-NN method approach. In the case of SVM kernelling six function options including: linear, quadratic, cubic, fine, medium and coarse Gaussian, were applied to convert the input signals to a higher dimensional feature space. Six classification methods for determining the target class for the k-NN algorithm were investigated including fine, medium, coarse, cosine, cubic and weighted. Round Robin Framework All seven bearing test cases were incorporated into each RUL estimation process in a round robin framework that seeks to maximise the data set as well as mitigate problems relating to over-fitting. The experimentation process involved allocating six bearing signals as training datasets to teach the ML algorithms. The seventh bearing test-case was used for testing purposes. Once RUL estimations were obtained for each of the out-of-sample test signals, the bearing was added back to the in-sample testing pool, and the next sequential bearing was transferred to the testing pool. The ML prediction model was retrained, and this was iterated until RUL estimations had been obtained for all seven bearing test cases. The incorporation of this framework to train and test the performance of each prediction algorithm greatly reduces over-fitting as we are only using out-of-sample test signals. All model training data comprises of signals from a completely different bearing for each test case. This gives an extremely accurate interpretation of how the models would perform in a real-world application using signals from different bearings used to train the models in every case. The classification process involves dealing with a multi-class and multi-label classification model, which comprises five temporal wear-state classes to be estimated for thousands of consecutive time samples. A moving-average (MA) filtering technique is incorporated to smooth out any undesirable whipsawing or erratic transitioning between the five temporal wear-state classes. The MA technique involves taking a window length of nine discrete-time samples, consisting of the current target prediction, the previous four predicted targets and the following four predictions. The mode of the nine predicted values is then assigned to the current target sample. In a real-time application, these nine samples comprise a 40 s temporal time period. This short time period is extremely negligible over a bearing's lifetime, which is typically years for a real system. Performance Metrics The performance of each ML approach investigated was analysed by computing the Jaccard Index [74,75], Equation (8) and multiplying by a factor of 100 to obtain a percentage accuracy value. J(z,ẑ) = \|z ∩ẑ\| \|z ∪ẑ\| (8) where z represents the true class of a time-sample andẑ represents the class prediction from the ML algorithm. The Mean Absolute Error (MAE) was calculated for each of the classification models and feature selection options [76,77]. Using Equation (9), the absolute error between the predicted target instances and the real expected values was calculated for each time-sample. This resulted in a natural number in the range 1 to 4, as we are dealing with five wear-state classes and the maximum error a prediction could possibly be classified as would be four classes away from the expected real class value. The error for each individual timesample was summed and divided by the total number of test-samples to calculate the MAE. These MAE values for each classification model were then normalised by dividing each MAE result by 4 and are compared in the tables below. where M represents the total number of target instances to be classified, z represents the true class of a target instance andẑ represents the class prediction from the ML algorithm. Results This section presents the results obtained from the proposed ML framework for RUL classification. Linear Wear-State Classification Approach The linear wear-state classification accuracy and MAE error results achieved using the STFT features extracted from the discrete-time signals are presented in Table 4. In the case of the SVM classification method, the lowest accuracy results were recorded at 27.5% for the one-third octave band feature set using a fine Gaussian kernel function. The highest classification performance achieved was 59.5% using the one-third octave band feature selection and a coarse Gaussian kernelling method. For the k-NN classification method, the lowest classification performance recorded was 39.9% for the linear band feature set using fine k-NN. The highest classification accuracy achieved was 54.2% by the one-third octave band features with cosine k-NN. The MAE results indicate that the coarse Gaussian kernel using the one-third octave band features was also the best-performing model with the lowest MAE score. The lowest error scores of 0.17 were achieved by both the cubic and cosine k-NN models. The results of the experiment using the signal envelope-derived features and linear wear-state classification are presented in Table 5. In the case of the SVM model, the lowest performance was recorded at 30.4% accuracy for the one-third octave band FFT features using a fine Gaussian SVM classifier. The highest classification performance, on the other hand, was achieved at 62.5%, for the one-third octave band summed envelope features using a Linear SVM kernel function. For the k-NN classification results, the lowest performance was recorded at 44.2% accuracy for the squared one-third octave band features from a fine k-NN classifier. The highest classification accuracy was achieved at 57.9% for the squared linear band features using a coarse k-NN classifier. The MAE results indicate that the linear, coarse and medium Gaussian kernel functions using the summed one-third octave band features were the best-performing model with the lowest MAE score. The coarse k-NN model proved to be the best option from the MAE values also, proving to be the most accurate in both Jaccard Index and MAE aspects. Non-Linear Wear-State Classification Approach The results achieved using non-linear wear-state classes and STFT features are presented in Table 6. For the SVM experiments, the lowest performance recorded was 53.9% for the linear band features using a cubic kernel function SVM model. The highest performance of 73.6% was achieved by the one-third octave band features using a medium Gaussian kernel function. In the case of the k-NN experiments, the lowest performance recorded was 60.1% from the linear frequency band features with the Fine k-NN. The highest performance accuracy of 73.2% was achieved using the one-third octave band features with Coarse k-NN. This was also the highest classification performance achieved overall for the STFT analysis feature study. The normalised MAE results for the STFT spectral features using SVM classification models indicate that the Medium and Coarse Gaussian kernel function using the one-third octave band features was also the best-performing model with the lowest normalised MAE score. Similarly for the k-NN experiments, the one-third octave using Coarse k-NN achieved the lowest error score. The non-linear wear-state results for the Signal-Envelope-derived features using are presented in Table 7. For the SVM models, the lowest performance recorded was 9.5% for the squared one-third octave band features using a cubic kernel function, whereas the highest performance of 73.1% was achieved for the summed one-third octave band envelope features using a linear SVM kernel function. In the case of the k-NN classification results, the lowest performance recorded was 62.0% for the squared one-third octave band features using a Fine k-NN classifier. The highest performance accuracy was 74.3% for the one-third octave band FFT features using a cosine k-NN classifier. This was also the highest classification performance achieved overall for the Envelope Analysis k-NN study and importantly for all of the experimental studies reported in this work. The normalised MAE results using the signal envelope derived spectral features applying SVM classification models indicate that the Linear kernel function using the one-third octave band features were the best-performing model with the lowest MAE score of 0.09. The Cosine k-NN model proved to be the best options for the non-linear temporal classes, with an error score of 0.08, which was also the lowest overall error score across all four experimental studies presented. Accordingly, these best MAE values also correspond with the best classification accuracy achieved, which is not unexpected. Discussion The results presented in Section 5 highlight the best-and worst-performing supervised ML approaches for both the linear and the non-linear wear-state classes that were investigated. The SVM-based algorithm approach that employed one-third octave band-based features was found to yield the best performance for the linear wear-state classes. This was the case for both STFT-and EA-based features, achieving scores of J = 59.5% with MAE = 0.13 for O[m, k] using SVM (Coarse G) and J = 62.5% with MAE = 0.12 for EA O using SVM (Linear), respectively, as shown in Tables 4 and 5, for the linear wear-state classification. Again, for the non-linear wear-state classification the SVM (Medium G) based algorithm performed extremely well using the O[m, k] STFT features by achieving scores of J = 73.6% with MAE = 0.10 as shown in Table 6. Using the same octave band feature set, the k-NN (Cubic) approach had comparable performance coming in at slightly less accuracy at J = 73.2% with the same MAE = 0.10, as shown in Table 6. In the case of the EA features for non-linear wear-state classification, the SVM (Linear) achieved J = 73.1% with MAE = 0.09, see as shown in Table 7. However, the k-NN (Cosine) had superior performance using the spectral-based EA features, F EA O , achieving J = 74.3% with MAE = 0.08. This was the best performance achieved for all the ML approaches over this entire investigation, with the best classification accuracy and the lowest MAE. In order to better analyse and interpret the results more closely, confusion matrices are presented for the best-performing approaches for both the linear and non-linear wear-state classification approaches investigated. Accordingly, these confusion matrices correspond to the approaches that have their values highlighted in bold font in Tables 4-7 as discussed previously. At the class level, these confusion matrices enable the classification results to be examined, and they allow the MAE to be quantified and better appreciated-for instance, regarding how many samples from class 1 were incorrectly classified as class 5. The confusion matrices shown in Figure 7 allow us to see the classification performance for each class by observing the percentage score along the diagonal. It is noted that the vast majority of classification inaccuracy (MAE) tends to be predicting the neighbouring class, which is significant, while this information was captured collectively for the entire class set using the MAE metric for ML approaches; however, these confusion matrices offer the granularity to identify which specific classes were the most challenging to estimate. All of the ML recipes presented performed very well on class 1, the max and min range being 92.6% to 89.0% in comparison to the max and min range for class 5 of 68.1% to 41.6%. The ML recipe with the best performance overall at 62.5% with MAE of 0.12 for the linear wear-state classification was that of the EA O features with a SVM (Linear) algorithm, this is shown in matrix (c) of Figure 7. This particular ML recipe achieved the highest classification for class 4 at 65.1% and was second best in class 1, 3 and 5, which ultimately led to it achieving the best overall score. Similarly, as shown in Figure 8, the classification performance for class 1 for the nonlinear wear-state classes was good; however, the range was wider, with max and min values of 95.5% and 78.9%, respectively. Whereas in class 5, the max and min range was from 63.2% to 39.3%. The best performance overall at 74.3% with MAE of 0.08 for the linear wear-state classification was the ML recipe that comprised of the EA 2 L features with k-NN (Coarse) algorithm, this is shown in matrix (d) of Figure 8. This ML recipe strong performance in class 1 and average performance in class 2 and 3 with poor performance in class 5. However, the performance in class 4 at 25.7% significantly outperformed the other ML recipes shown, and this lead to it scoring the best overall. These trends in individual class performance also be viewed in Figure 9, without the benefit of visualising where the incorrectly classified test cases have been predicted. These points along with a mean value correspond to the diagonal values for the confusion matrices in Figures 7 and 8. The high performance of class 1 for both the linear and non-linear wear-state class options is identifiable as well as the decreasing trend as the classes approach the failure stage of the bearings under test. Prior work by Sutrisno et al. [78], Singleton et al. [79]a nd Lei et al. [80], presented ML methods that achieved percentage accuracy scores of 76.2%, 67.20% and 77.44%, respectively, using the PRONOSTIA bearing dataset. However, these proposed methods utilised a framework where only bearings S. 01 and S. 02 were used for training the algorithm, and the remaining five bearings were used for testing. The round-robin experimental framework presented in this paper presents the mean percentage accuracy of all seven bearing signals whereas the prior work only presented the mean of five signals. Furthermore, the MAE performance metric was used for analysis purposes to ascertain the severity of the misclassifications. Feature subset compression using the non-linear one-third octave-based filtered for both linear and non-linear wear-states performed very well in both cases. This can be attributed to placing a higher emphasis on the lower portion of the spectra for feature extraction. From a feature-engineering perspective, this was shown to offer more valuable diagnostic trend information for characterising the health condition of the bearings under test [11]. Importantly, this reduces the multivariate dimensionality [70,71] of the feature space in a more optimal way compared with linear filtering as the results demonstrated by yielding superior classification performance. Moreover, using a non-linear wear-state model approach to classification is more suitable as the ageing mechanisms typically follow an non-linear exponential trend; hence, we can see significantly higher performances achieved. Clearly, a trade-off exists as if the size of class 1, is too large with respect to the others, then the suitability of the RUL framework for taking timely action, such as equipment maintenance and critical parts replacement diminishes. As the subsequent classes would therefore be too short in time, the severity of degradation between these classes would occur rapidly. Our non-linear exponential model described in Equation (6) strikes a suitable balance and was found to work extremely well in this proposed approach. The highest overall classification scores were achieved using the Cosine k-NN classifier. This was achieved across all seven bearing test-cases using the round robin framework and, hence, demonstrates how the proposed ML methodology performs on unseen raw vibration signals. However, these k-NN and SVM ML algorithms are heavily reliant on the depth of the training data, which is common in the field of supervised learning and hence makes these methods prone to producing over-fitted prediction models. While this paper introduced a valuable and robust approach for RUL estimation, future work might investigate applying this proposed method on larger data-sets. The dataset used here in this research was limited in terms of the total number specimens aged and captured using vibration signals In addition, the level of accelerated ageing is perhaps too rapid, which led to a high proportion of abrupt failure modes occurring approximately 43% of the time. These have a completely different degradation trend to the gradual ageing mechanisms; hence, this places limits on testing the true efficacy of the ML frameworks and recipes due to model over-fitting. If the datasets were significantly improved by increasing the number of specimens and reducing the level of accelerated ageing, this would offer the potential to explore ML approaches that employ advanced deep learning using neural networks. In testing the versatility and robustness of the proposed ML method recipes on different bearing types and sizes under different speed and load conditions, work could explore vibration data gathered from research testbeds where the shaft speed changes. This will require developing extensive experimental campaigns to create more advanced datasets that better reflect typical real-world operating conditions. Conclusions Traditionally, condition-based monitoring (CbM) of electric and rotating machines has focused heavily on two primary areas, the detection and the diagnosis of fault modes. More recently, research efforts have investigated the more challenging area of prognosis to determine the remaining useful life (RUL) of the machine under test. This paper introduced a valuable machine learning (ML) approach to estimate the RUL of rolling element bearings, which are a core component of rotating machines. The proposed ML recipes and approaches comprise of signal processing techniques and ML algorithms applied to real-world vibration signals, which were acquired from the outer-race of bearings degraded over time using an accelerated ageing test-rig. The paper reports the results for linear and non-linear wear-state models using novel feature engineering derived from Short-Time Fourier Transform (STFT) and Envelope Analysis (EA) representations. In addition, two different classification algorithm approaches, k-Nearest Neighbour (k-NN) and Support Vector Machines (SVM), were investigated and compared. This work achieved classification accuracy results of up to 74.3% with a mean absolute error (MAE) of 0.08, which demonstrates the method's efficacy for performing the task of RUL. This ultimately offers a robust and low complexity approach that is highly valuable for advanced predictive maintenance purposes in industry. Data Availability Statement: Publicly available datasets were analysed in this study. This data can be found here: (https://ti.arc.nasa.gov/tech/dash/groups/pcoe/prognostic-data-repository/#bearing, accessed on 11 March 2022), also see [62]. Conflicts of Interest: The authors declare no conflict of interest. The Institute of Technology Carlow have endorsed this manuscript to go forward for peer review and publication. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. Abbreviations The following abbreviations are used in this manuscript:	v3-fos-license
2022-02-06T05:14:56.934Z	2022-02-04T00:00:00.000	246556024	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0263186&type=printable", "pdf_hash": "9f6e75488d48d9360aa15cbf0f65fcd28a201a42", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45458", "s2fieldsofstudy": [ "Medicine" ], "sha1": "9f6e75488d48d9360aa15cbf0f65fcd28a201a42", "year": 2022 }	pes2o/s2orc	Association between Chlamydia trachomatis, Neisseria gonorrhea, Mycoplasma genitalium, and Trichomonas vaginalis and Secondary Infertility in Cameroon: A case-control study Objective Data on the prevalence and etiology of infertility in Africa are limited. Secondary infertility is particularly common, defined as the inability of a woman to conceive for at least one year following a full-term pregnancy. We describe a prospective study conducted in Cameroon designed to test the hypothesis of an association between common treatable sexually transmitted infections (STI): Chlamydia trachomatis (CT), Neisseria gonorrhoeae (NG), Mycoplasma genitalium (MG), and Trichomonas vaginalis (TV) and secondary infertility in women. Methods In this case-control study, we enrolled women in Fako Division, Cameroon between November 2017 and December 2018 with secondary infertility (cases) or current pregnancy (controls). We conducted a baseline survey to collect sociodemographic, and sexual and medical history information. Nucleic acid amplification testing using Aptima (Hologic, San Diego, CA, US) was performed on endocervical swabs for CT, NG, MG, and TV. Multivariable logistic regression was used to assess the relationship between active STI and secondary infertility. Results A total of 416 women were enrolled: 151 cases and 265 controls. Compared to controls, cases were older (median age 32 vs 27 years) and had more lifetime sexual partners (median 4 vs 3) (p<0.001). Cases were more likely to report dyspareunia, abnormal menses, prior miscarriage, and ectopic pregnancy (all p<0.05). STI positivity was not significantly different among cases and controls (2.7% vs 5.4% for CT, 1.3% vs 2.9% for NG, 6.0% vs 7.0% for MG, respectively), with the exception of TV which was more common in pregnant controls (0.7% vs 5%; p = 0.02). Conclusion Study findings did not support an association between active STI and secondary infertility in Cameroon. Given high rates of pre-existing tubal damage, routine STI screening and treatment in younger women may be more impactful than costly STI testing during infertility assessments. Introduction Infertility affects 10-15% of couples, an estimated 80 million people globally [1]. Secondary infertility is defined as the inability to conceive for at least one year by a woman with a history of prior delivery. In Cameroon, a country in central Sub-Saharan Africa (SSA), the prevalence of infertility ranges from 15% to 30% and secondary infertility is twice as common as primary infertility [2,3]. Stigma associated with infertility is common in SSA and women in couples with infertility are more likely to experience domestic violence, divorce, and social marginalization compared to women without infertility [2,4]. Additional studies are needed to understand the etiology of infertility in SSA to inform prevention and treatment options in a region where access to assisted reproductive technology is limited [4,5]. Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) infections are common bacterial sexually transmitted infections (STI) that cause cervicitis in women and ascend to the upper reproductive tract in 10-20% of cases, leading to pelvic inflammatory disease (PID). PID can cause inflammation and damage to fallopian tubes that results in tubal factor infertility (TFI) [6,7]. CT infection is associated with a 3.4-fold increase in TFI [8]. Data on the association between Mycoplasma genitalium (MG) infection and infertility in women are limited and Trichomonas vaginalis (TV) infection is not known to impact fertility. Studies of MG, TV and infertility using highly sensitive PCR molecular diagnostic testing are ongoing but this STI testing is mostly unavailable in SSA [9,10]. Since women with secondary infertility have a prior history of delivery, they may be more likely to have a reversible cause of TFI, such as a recently acquired STI. If active infection is present, it is possible that STI treatment could reduce tubal inflammation and improve fertility. This case-control study was designed to test our hypothesis of an association between active common STIs and secondary infertility among women in Cameroon. Study area and setting The study was conducted in semi-urban areas of Fako Division (Buea, Limbe Tiko, and Mutengene) in the Southwest region of Cameroon in Central Africa. Participants were seen at the main reference hospital in the region. Cases with secondary infertility were enrolled at the gynecological unit and pregnant controls were enrolled at antenatal clinics in the same region. Study design and participants This unmatched case-control study enrolled women in the age range of 15-46 years. Cases were women with secondary infertility according to self-report or as identified by a gynecologist. Secondary infertility was defined as the inability to conceive by a woman despite at least one year of unprotected sexual intercourse following a prior delivery. Controls were randomly selected pregnant women seen in prenatal clinic who reported no history of infertility. Approximately two controls were selected for each case identified. Women were ineligible to participate if they refused to grant consent and if they had taken any antibiotics during the past 7 days. Study procedures Women were enrolled in the study after providing informed consent. Gynecologists, nurses, and study staff surveyed women about socio-demographics, sexual history, obstetrical and gynecological history, including any infertility evaluations. Medical records were reviewed for STI history, pregnancy history, and other medical history, when available. Questionnaire and study procedure was formulated to follow hospital patient management protocols at the various study sites. Biological specimen collection Endocervical swabs were collected by a medical provider using Dacron swabs and placed directly into Hologic 1 Aptima 1 vaginal swab transport media (Hologic Inc San Diego, USA). Specimens were transported to the University of Buea Laboratory and stored at -80˚C until transport to the University of Alabama at Birmingham STI Laboratory (USA) for STI nucleic acid amplification testing (NAAT). Ethical considerations The study was approved by the Faculty of Health Sciences Institutional Review Board in Cameroon (No 2016/459/UB/FHS/IRB). Administrative clearance was also obtained from the Regional Delegation of Public Health for the Southwest Region and all the health facilities where study was conducted. Statistical analysis For a power calculation assuming two controls per case, a sample size of 408 (136 cases and 272 controls) would provide >80% power at an alpha level of 0.05 to detect an association between STI and secondary infertility if STI positivity in the control group was 19%.Chisquare or Fisher's exact testing was used for comparisons of categorical variables where appropriate; and, analysis of variance or the Kruskal-Wallis test was used for continuous variables. Univariate logistic regression was performed to identify predictors of secondary infertility. Variables of interest, specifically socio-demographics, sexual behaviors, and a composite STI measure were included in the full multivariable logistic regression based on published data to explore the association between STI positivity and secondary infertility. Unadjusted and adjusted ORs, 95% CI, and p-values were calculated with significance set at p<0.05. SAS version 9.4 (Cary, NC) was used for analyses. Participant characteristics A total of 416 women were enrolled: 151 cases and 265 controls (Table 1). Cases were older than controls, median (interquartile range) 32 years (27, 36) vs 27 years (24, 32), p<0.001. Sexual behavior differed somewhat between the groups: 26% of cases and 7% of controls had �5 lifetime partners (p<0.001) although the median number of lifetime partners was <5 in both groups. Two-thirds of women in both groups were married and cases were more likely to report multiple current sexual partners (4% cases vs 0.8% controls; p = 0.03). Among cases, 46 had hysterosalpingography (HSG) results in their medical record. Of these 46, 31 women (67%) had documented evidence of tubal obstruction +/-hydrosalpinx (20 with bilateral obstruction, 11 with unilateral obstruction), 6 had uterine adhesions, and 1 had radiographic evidence of cervicitis. The STI positivity (CT/NG/MG/TV) amongst cases with HSG was 3/46 (6.5%). None of the controls had HSG results as a comparison group since they had no prior evidence of infertility according to inclusion criteria. Factors associated with secondary infertility In unadjusted models with secondary infertility as the outcome, older age, white collar occupation, higher number of lifetime sex partners and having more than one current sex partner were associated with infertility (Table 2). When compared to the youngest age category of 15-24 years, the crude odds ratios (OR) and 95% confidence intervals for secondary infertility increased in a stepwise fashion from 2.9 (CI 1.5-5.6) for women aged 25-29 years to 9.6 (95% CI 4.5-20.6) for women aged 40-46 years. Women with more than one current sex partner (OR 5.4; 95% CI 1.1-27.3) and �5 lifetime sex partners (OR 4.7; 95% CI 2.5-8.9) had greater odds of secondary infertility. Women with dyspareunia, abnormal periods, history of miscarriage and prior ectopic pregnancy also had greater odds of secondary infertility ( Table 2). Discussion This is one of the first studies to focus on the association between active STI and secondary infertility in Africa using highly sensitive molecular diagnostic testing to detect CT, NG, MG, and TV in women of reproductive age in Cameroon. Our prospective case-control study did not support our hypothesis of a possible association between active STI and secondary infertility although classic STI risk factors (higher number of lifetime sexual partners, more than one current sexual partner) were more prevalent in cases compared to controls. STI prevalence among women in this study was similar to other studies in Africa [15][16][17][18][19][20]. Among the four STIs detected, only TV showed a significant difference between the two groups, with higher positivity in controls. Additional studies are needed to explore the association noted in unadjusted models between TV and pregnancy. Any assessment of the association between prior STI infection and secondary infertility in Cameroon is limited by the fact that, unlike high-income countries, routine STI testing among women is not recommended and sensitive diagnostic testing is mostly unavailable. The performance of syndromic STI management in SSA as recommended by the World Health Organization is demonstrably poor, particularly in women who normally have asymptomatic bacterial STI [21]. In terms of STI and TFI, tubal blockage has been detected in 50% of women with infertility (mostly secondary infertility) in urban Cameroon [22]. This is consistent but lower than the 67% of women with secondary infertility shown to have tubal blockage in the semi urban setting of this present study. The mechanism of pathogenesis of NG and CT fallopian tube infection and inflammation resulting in tubal infertility has been well-described [23]. In the literature, tubal obstruction is seen in 20-30% of cases of primary and secondary infertility [24]. Egbe et al. found that a history of CT was strongly associated with tubal factor infertility in Cameroon [25]. A meta-analysis demonstrated a higher prevalence of NG in infertile populations compared to the general population [16]. Although data on MG are fewer, a recent US study by Peipert et al found an association between serologic evidence of prior MG infection and infertility [26,27]. Our study did not show an association between active infection with CT, NG, MG, and secondary infertility. These findings do not rule out a possible association between bacterial STI and secondary infertility since HSG test results among participants with secondary infertility suggested that tubal damage had already occurred. One relevant implication of our study is that current infertility management protocols in Cameroon that rely on CT serology testing and antibiotic treatment, may be ineffective and costly for the patient if infection occurred in the past. Therefore, the WHO recommended syndromic STI management and treatment (which fails to detect asymptomatic infections) may not be an effective strategy to prevent tubal damage and adverse pregnancy outcomes. More frequent and earlier STI screening and treatment of women in SSA may be needed to prevent upper genital tract complications [28]. In addition to highly sensitive molecular diagnostic testing, it would be useful to measure STIspecific host responses to detect prior infection and further understand the role of each STI in terms of a causal pathway for TFI. A serologic study to compare immunoglobulin-specific responses to CT between cases and controls in our population will be informative about prior CT exposure and is currently ongoing. We have previously shown that CT antibody testing such as IgG1 to elementary bodies and CT OmcB can help elucidate the timing of prior infection [29]. Some factors associated with secondary infertility in our study align with other studies; age, adverse pregnancy outcomes (miscarriage and ectopic pregnancy) and dyspareunia [30,31]. Modifiable risk factors such as marital status, education, and age of sexual debut had no association with secondary infertility in our study. This differs from findings by Larsen et al. in Cameroon where secondary infertility was associated with early sexual debut (before the age of 15) and single marital status [32]. Our study had limitations. Importantly, cases and controls were not age-matched, duration of secondary infertility varied from 1-18years among cases and there was heterogeneity in lifetime number of sexual partners between groups (although a majority reported 1-5 partners). Since the risk of STI generally decreases with age and cases were older than pregnant controls, this may have biased our sample toward higher STI rates in pregnant controls (not observed for CT/NG/MG). Participant recall error and bias in STI history information are also possible since STI testing can be performed as part of the initial infertility evaluation. If cases received empiric STI treatment during infertility evaluation, they may have had more historical antibiotic exposure compared to controls. Finally, since the study sample was isolated to women in Southwest Cameroon, findings may not generalize to all women with secondary infertility. Future studies should include detailed diagnostic evaluations to better understand the most common etiologies of primary and secondary infertility among women and men in sub-Saharan Africa as well as the frequency and etiology of tubal damage. Once the population attributable fraction due to STI is understood, prospective prevention studies can be designed. Conclusion This prospective case-control study of women with secondary infertility in Cameroon did not show an association with active STI (CT, NG, MG, or TV). Since rates of tubal damage were high, the benefit of routine CT/GC NAAT testing and prompt treatment in young sexually active women for the endpoint of infertility prevention should be investigated further.	v3-fos-license
2020-10-06T13:37:19.377Z	2020-10-01T00:00:00.000	222147096	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://onlinelibrary.wiley.com/doi/pdfdirect/10.1111/os.12788", "pdf_hash": "ff627f5c30deb7c5d35b7128383706f4d241d838", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45460", "s2fieldsofstudy": [ "Medicine" ], "sha1": "d9738544b65cd8e45ed73b5552f6b8917d37dce3", "year": 2020 }	pes2o/s2orc	Shoulder Osteonecrosis: Pathogenesis, Causes, Clinical Evaluation, Imaging, and Classification The humeral head is the second most common site for nontraumatic osteonecrosis after the femoral head, yet it has attracted relatively little attention. Osteonecrosis is associated with many conditions, such as traumatism, corticosteroid use, sickle cell disease, alcoholism, dysbarism (or caisson disease), and Gaucher's disease. The diagnosis is clinical and radiographic with MRI, with radiographs being the basis for staging. Many theories have been proposed to decipher the mechanism behind the development of osteonecrosis, but none have been proven. Because osteonecrosis may affect patients with a variety of risk factors, it is important that caregivers have a heightened index of suspicion. Early detection may affect prognosis because prognosis is dependent on the stage and location of the disease. In particular, the disease should be suspected in patients with a history of fractures, steroid usage, or sickle cell disease, and in divers. This report reviews osteonecrosis of the humeral head, with an emphasis on causes, clinical evaluation, imaging, and classification. Introduction A fter the femoral head, the humeral head is the most frequent location for nontraumatic osteonecrosis 1 . This pathology is underestimated, and a proper clinical evaluation must be presented as knowledge for the orthopaedic surgeon. In addition to post-traumatic avascular osteonecrosis (AVN) 2 , causes of humeral head necrosis ( Fig. 1) are sickle cell disease (SCD) 3 , the first genetic disease in the world, and corticosteroid therapy 4 ; as a result, a large population of patients are at high risk for shoulder osteonecrosis. Isolated osteonecrosis of the humeral head in the absence of involvement of the hip is rare; in fact, it is frequent in patients with multifocal osteonecrosis 5,6 . Shoulder Osteonecrosis: A 250 Million Year History Mesozoic diving marine reptiles have been characterized on the basis of avascular necrosis caused by decompression. The ichthyopterygians reptiles appeared around 250 Ma; they were marine reptiles and became extinct 90 Ma 7 . The humeral head revealed avascular bone necrosis in 16% of these reptiles. Bone death was caused by intravascular bubble formation during decompression. After deep diving and during decompression, nitrogen changes from dissolved phase to free phase and gas bubbles may be formed in many tissues. As nitrogen is more soluble in fat 7 , it will dissolve in the fatty marrow of the long bones. Normally, bubbles are formed during decompression in the venous circuit, and the lung is effective as a filter. This is true when the systemic and the pulmonary circuit are separated, as for mammals and whales. However, an opening in the heart with the possibility of right to left shunting in those reptiles allowed venous nitrogen-oversaturated blood to flow into the systemic circulation, compromising the vascularization of bone. necrosis of the shoulder in compressed air workers. The humeral head has aroused rather limited research interest as compared with hip osteonecrosis. In 1960, Heimann and Freiberger 10 provided initial descriptions of shoulder osteonecrosis; in 1968, Cruess et al. 11 presented a classification of shoulder osteonecrosis. More recent studies have reported on patients with humeral head osteonecrosis. Etiology and Risk Factors T here are many causes of humeral head osteonecrosis ( Table 1) described in the literature. We will focus on the most frequent. Trauma Trauma leads to humeral head osteonecrosis by disrupting the vascular supply. Humeral head osteonecrosis is frequently identified in patients who have undergone internal fixation (Fig. 1). This occurs when the blood supply to the humeral head is disrupted by a fracture. Vascular anatomy allows us to understand the mechanism of necrosis. The ascending branch of the anterior humeral circumflex artery is thought to be predominant for vascularization 12 . This finding was supported by Gerber et al. in 1990 13 . They also showed that, through anastomoses of these two arteries (Fig. 2), the anterior circumflex and posterior circumflex arteries supply the humeral head with other arteries. The anterior circumflex artery reaches the surgical neck of the humerus at the inferior border of the subscapularis after passing beneath the coracobrachialis and the short head of the biceps. Despite the posterior circumflex artery having a diameter three times larger than that of the anterior circumflex artery, it supplies only the greater tuberosity and the posteroinferior part of the humeral head. Although it may be suspected that a history of dislocation could lead to osteonecrosis via capsular injury and vascular shearing, this has not been proven. Solberg et al. 14 find osteonecrosis to be the most common complication of fractures (16% affected). There is a strong correlation between osteonecrosis and the length of the initial metaphyseal contact point or hinge (Fig. 3) attached to the articular fragment, as demonstrated by Hertel et al. 15 . Robinson et al. 16 investigated patients who had sustained proximal humeral fracture dislocations that were treated with osteosynthesis. Robinson proposed as classification for risk of osteonecrosis after fracture the following criteria. Type I injuries: fracture capsular attachment, 2 cm in length, with cancellous bone arterial bleeding. Type II injuries: minimal capsular attachment, <2 cm in length, with no bleeding present. Using this classification, osteonecrosis with type II injuries was more frequent. Robinson et al. 16 also included bone bleeding as a predicting factors; however, some humeral heads with signs of perfusion went on to develop osteonecrosis. Finally, the age at which the injury occurs may also be an important risk factor, but this is not proven clinically. Patients sustaining three-part and four-part fractures have been investigated for humeral head viability through Patel estimated the number of posttraumatic osteonecrosis after fracture 2 . The risk of osteonecrosis ranges from 0% to 25% 18,19 for three-part fractures and 0%-77% for four-part fractures, with an increased rate with longer follow-up 20 . Corticosteroids Corticosteroids are a common cause of atraumatic humeral head necrosis, but the exact mechanism is unknown. Hernigou 4 evaluated the natural history and the rate of disease progression with a long-term follow up. The outcomes of 215 shoulders (125 adult patients) with osteonecrosis related to corticosteroids were evaluated at early stages before collapse with radiographs and MRI. With an average 15 years of follow up (range 10 to 20), there was pain in 65% of asymptomatic shoulders and collapse had occurred in 50% of shoulders. In symptomatic shoulders, collapse has developed in most at the final follow up. The time between the diagnosis and collapse was an average lof 6 years. At the most recent follow up (average 15 years), 50% of shoulders had required surgical treatment. Stage, occurrence of pain, and continuation of peak doses of corticosteroids were risk factors for progression of osteonecrosis. The role of steroid dose remains controversial. Suspected as early as 1957 21 , whether this is a dose response, threshold, peak-dose, or idiosyncratic phenomenon is unclear. The high dose of steroids during the first several weeks seems to be more important than the total cumulative dose. In addition, patients may have a predisposition toward osteonecrosis, with a possible genetic susceptibility for the disease. The most commonly accepted theory surrounding corticosteroids and shoulder osteonecrosis involves fat accumulation in marrow (Fig. 4), leading to increased intraosseous hypertension and decreased blood flow. This concept has been extrapolated 22 into a multiple-hit theory in which corticosteroids alter bone homeostasis, injure bone cells, impair blood flow, and suppress bone cell precursors in susceptible patients. Corticosteroids inhibit angiogenesis and promote a hypercoagulable state, which could contribute to the formation of intravascular thrombosis. Sickle Cell Disease Sickle cell disease, an autosomal recessive disorder, is also called sickle cell anemia (SCA) due to the hemolytic anemia characterized by abnormally shaped (sickled) red blood cells (RBC), which are removed from the circulation and destroyed at increased rates, leading to anemia. Of greater clinical importance, the sickled RBC cause vascular occlusion, which leads to tissue ischemia and infarction. The patients who are homozygous for the sickle cell gene are named "hemoglobin SS" or Hb SS patients. There have been at least four mutational events occurring independently, with three in Africa and a fourth in Saudi Arabia or central India. These events occurred from 3000 to 6000 generations ago, approximately between 70,000 and 150,000 years ago. The distribution of Hb SS in the world (Fig. 5) is indicative of two factors: selection for carriers with survival advantages in malaria-endemic regions and subsequent migrations. The first descriptions of humeral head osteonecrosis were published by Chung 23 and Millner 24 . Patients who are homozygous for the sickle cell gene have a high risk of bone osteonecrosis due to microvascular occlusion (Fig. 6) in relation to the disturbance in the erythrocyte architecture. Although precise data are not available, a recent estimation suggests that SCD affects 0.74% of births in sub-Saharan Africa. By comparison, only 0.15% of the black population of the United States and of Europe is affected by SCD. SCD is also an important etiology of hip osteonecrosis in the Indian subcontinent, in South America, in the Persian Gulf, in the Mediterranean countries and those from the Caribbean, and in Central America. It is estimated that each year over 300,000 babies (every decade 3 million) with severe forms of these diseases are born worldwide, with the majority in low-income and middle-income countries. Approximately 5% of the world's population are healthy carriers of a gene for SCD or thalassaemia. In high-income countries, the survival of people with SCD has increased steadily, often to adulthood. In contrast, infant mortality related to SCD in Africa remains between 50% and 90%, with less than half of affected children reaching their fifth birthday. An index of the high mortality rate throughout childhood is the observation that the prevalence of Hb-SS in adults is 10 times lower than the incidence of births (0.2%-0.3% against 2% to 3%). Nearly 90% of the global population with SCD live in three countries: India, Nigeria, and the Democratic Republic of Congo, where 2% of the population is affected by the disease and the prevalence rate of carriers (trait sickle cell) reaches 10% to 30%. Nigeria alone would have at least 150,000 newborns a year. According to these data and to some extrapolations, SCD affects several million adults worldwide. Specific orthopaedic manifestations of SCD and its sequelae include osteonecrosis, infection, and bone marrow hyperplasia. Osteonecrosis of the humeral head (shoulder) has been reported in up to 50% of patients based on the type of hemoglobinopathy. With this frequency (50%) and the presence of bilateral osteonecrosis most often in this population, the number of shoulder osteonecrosis to SCD is probably several milli on in the world, and it is probably the most frequent cause of shoulder osteonecrosis. Hernigou 25 and Poignard 26 reported the natural evolution of 82 adult patients with SCD and osteonecrosis of the humeral head. A total of 104 cases of shoulder osteonecrosis were identified with MRI. Shoulder osteonecroses were graded with MRI and radiographs. Partial or total repair with a decrease in the size of the osteonecrosis or stage regression was never observed on MRI. At average follow up 20 years (range 15 to 24 years), collapse had occurred in 86% of shoulders. The mean interval between onset of pain and collapse was 6 years. The principal risks for shoulder osteonecrosis in adults with SCD were the presence of hip osteonecrosis, and the genotypes S Beta or SC. Alcohol The connection between alcohol use and shoulder osteonecrosis has been known for almost a century, and, as with corticosteroids, the exact mechanism is unknown. Animal studies of bone marrow treated with alcohol have shown increased pressure and adipogenesis and decreased hematopoiesis, which could lead to osteocyte injury and shoulder osteonecrosis 27 . There seems to be a dose response relationship; however, there is no established threshold beyond which patients are proved to be more at risk 28 . Given the relatively low prevalence of shoulder osteonecrosis among all alcohol users, there is likely a multifactorial relationship between shoulder, genetic susceptibility, environmental factors, and medical comorbidities. Dysbaric Osteonecrosis The risk of decompression illness 29 is directly related to the depth of the dive, the amount of time under pressure, and the rate of ascent. Dive tables, such as the US Navy Dive Tables, provide general guidelines as to what depths and dive times are less risky for the development of decompression sickness. In humans, decompression illness is seen among scuba divers and compressed air workers, where the uptake of nitrogen in the blood and tissues is continuously taking place 30,31 . If appropriate decompression procedures have not been followed, independently of the risk of death, these divers and workers (Fig. 7) will be at risk of avascular bone necrosis 32 . Although dysbaric osteonecrosis is well known by paleontologists 7 , it is now less well known by orthopaedists because divers use dive tables. However, it is important to understand (and usually rheumatologists and orthopaedic surgeons do not know) that even divers who follow decompression schedules and tables may still experience decompression illness in cases of foramen ovale. The human heart has four chambers, and the systemic and pulmonary circuit is normally fully separated. In the fetus, the left and the right atrium communicate with one another through the foramen ovale. This opening closes during the first weeks of life. In the fetus, the lungs are not functioning and the fetal blood is oxygenated in the placenta. In the atrium, 60% of the blood is shunted from the rightd to the left side (R-L shunt). In some individuals 32 , this intracardiac opening is persistent to some degree in adult life as a patent foramen ovale (PFO). Small-and medium-sized openings are normally symptomless and may be present in as many as 30% of normal subjects. In humans, the risk of decompression illness is five times higher in individuals with patent foramen ovale; this condition allows blood shunting from the venous circuit to the systemic circuit. There is convincing evidence of a connection between PFO and decompression illness in divers; Torti et al. 33 found that PFO was correlated with a five times increased risk of major decompression illness events and the risk augments with the PFO size. Situations that elevate right atrial pressure relative to left atrial pressure would increase the tendency for any right to left shunting, such as breath holding, coughing, and the Valsalva maneuver (all common during scuba diving). The Valsalva technique, used to equilibrate the pressure in the middle ear, increases pressure in the right atrium and thereby increases the right-left shunt if PFO is present. Miscellaneous Many other factors have been associated with shoulder necrosis. Gaucher disease is a disorder where an enzyme deficiency causes accumulation of glucocerebroside in macrophages. The result is infiltration of the bone marrow with vascular compromise. A study of patients with Gaucher disease and shoulder necrosisfound that it was more prevalent in patients who had a prior splenectomy 34 . Clinical Evaluation I nsidious pain with movement is frequent in nontraumatic necrosis of the humeral head. Pain at night is not frequent; however, in one study, 70% of patients had difficulty sleeping 35 . Symptoms may prevent physical work in up to 80% of patients. A "click accompanying certain movements" results from joint incongruity, cartilage flap, or a large loose body. Patients with osteonecrosis of the humeral head are usually younger than patients with osteoarthritis. Physical examination may reveal tenderness, but motion is often preserved until the late stages. The discomfort is greater with the arm abducted or elevated 90 , corresponding to maximum glenohumeral loading. Depending on the history, useful tests may include blood count, erythrocyte sedimentation rate, and C-reactive protein to help rule out infection. Specific serology for rheumatoid arthritis is useful to rule out inflammatory conditions. SCD is confirmed by analysis of hemoglobin. Gaucher's disease is characterized by elevated serum acid phosphatase, but the diagnosis should be confirmed by enzymatic and mutational analysis. In many cases, the history, physical examination, and laboratory tests serve only to raise the suspicion of osteonecrosis. Imaging I f history and physical examination findings are suspicious for shoulder osteonecrosis, plain radiography is the next step in diagnosis. Although very early shoulder osteonecrosis may be undetectable on plain radiography, early shoulder osteonecrosis shows cystic and/or sclerotic changes in the humeral head. The term "crescent sign" describes an area of subchondral lucency in the humeral head that indicates subchondral fracture due to bone necrosis and subsequent attempts at repair. shoulder osteonecrosis at a later stage shows humeral head flattening, collapse, and degenerative changes. MRI is the modality of choice for patients with a suspicious history, along with physical examination with examination of normal radiographs. The diagnosis of osteonecrosis on MRI was based on band-like abnormal signals, and bandlike hypo-intense zones on T1-weighted images (Fig. 8). It is 99% sensitive and specific for detecting early shoulder osteonecrosis, which usually presents as an area of low-intensity signal on T1-weighted and high-intensity signal onT2weighted images (Fig. 9). Bone marrow edema and a joint effusion may also be present. The later stages of shoulder osteonecrosis are better assessed using plain radiography and, in some cases, CT scans. CT scans are useful for evaluating patients with a suspected subchondral fracture, which may not be seen on MRI. The percentage of the humeral head affected by the lesion can be calculated as the ratio of the volume of the lesion in relation to the volume of the humeral head, considered as half of a sphere. The extent of involvement may be graded as A, indicating mild (<15%), B, indicating moderate, and C, indicating severe extent (>30%), according to the percentage extent of the lesion in the humeral head. The extent of the lesion in contact with the glenoid can be measured on the transverse image with MRI. The diameter of the glenoid may be divided into two parts: anterior and posterior. The necrotic area is expressed as anterior if in contact with less than half of the glenoid rim, posterior if in contact with less than half of the glenoid rim, and medial if in contact with all the glenoid rim (Fig. 10). Once a lesion is identified, radiographs of the contralateral shoulder should be obtained. If the radiograph is negative, MRI should be considered. Alternatively, radionuclide imaging can be performed, to exclude disease in other joints. Classification O steonecrosis of the humeral head is frequently classified using the description of Cruess (1978) 36 ; it a variation of the system used by Ficat and Arlet 37 to describe hip osteonecrosis. It is split into five stages. Stage 1 is characterized by normal radiographs and abnormal MRI. The hyperintense areas of bone marrow are replaced by hypo-intensity on T1-weighted images, while a hyper-intense focus is seen in T2 weighted images (Fig. 11). In stage 2, a reparative process, including sclerotic or diffusely mottled osteopenia, is seen; sphericity of the humeral head is maintained. Stage 3 is characterized by the crescent sign. It is a subchondral radiolucent line; in this stage, the subchondral collapse may also cause minor depressions in the joint surface. Stage 4 disease has collapse where the articular surface presents destruction of the underlying trabecular pattern. Osteo-cartilagenous flaps of bone may occur and break free to become loose intra-articular bodies. Stage 5 is when the glenoid has developed articular changes due to joint incongruity. Other authors have proposed similar classifications 4, 38 .There is a correlation between different exams and histology (Fig. 12). Treatment C onservative treatment options are proposed and effective in patients with stage I and II osteonecrosis. Two types of core decompression have been described: a percutaneous technique using a classical delto-pectoral approach or an arthroscopic technique. After collapse, the indication is usually an arthroplasty. Osteonecrosis as a surgical indication accounts for approximately 5% of all shoulder arthroplasties performed. However, relatively few studies have evaluated the outcomes following arthroplasty for osteonecrosis of the humeral head and many studies include only a few cases or have a short-term follow-up. The decision to use hemiarthroplasty or a total shoulder arthroplasty is usually based on the glenoid status and the surgeon's opinion. One absolute contraindication for arthroplasty is active infection, and relative contraindications include significant brachial plexus injury and concomitant deltoid and rotator cuff insufficiency, where a reverse shoulder arthroplasty can be proposed. The challenge when it comes to shoulder arthroplasty in patients with atraumatic osteonecrosis remains prosthesis longevity, as these patients are relatively young. Conclusion shoulder osteonecrosis is a complex condition that is not fully understood. It most commonly arises from trauma or corticosteroid and alcohol use, but is also associated with a variety of other risk factors, including blood dyscrasias and metabolic and coagulation disorders. Initial evaluation of the shoulder should include a thorough history and physical examination, as well as assessment of plain radiographs of the hip and pelvis. Early stage shoulder osteonecrosis is best evaluated by MRI, and CT scans can be helpful in identifying subchondral fractures.	v3-fos-license
2018-04-03T01:20:30.520Z	2014-10-01T00:00:00.000	14646571	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "http://www.cardiologyres.org/index.php/Cardiologyres/article/download/331/370", "pdf_hash": "3240313f83a1d1ef6206b7c095f186e933df8445", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45463", "s2fieldsofstudy": [ "Medicine" ], "sha1": "3240313f83a1d1ef6206b7c095f186e933df8445", "year": 2014 }	pes2o/s2orc	A One-Sided Affair: Unoriginal Origin of the Left Coronary Artery, a Case Report Coronary artery anomalies constitute a group of congenital malformations that have a multitude of clinical manifestations and highly variable pathophysiology. We report a 56-year-old male with angina due to an anomalous origin of the left main coronary artery; approach and management. Introduction We report a 56-year-old gentleman found to have an anomalous origin of the left main coronary artery. Case Report A 56-year-old gentleman with a history of coronary artery disease of undefined anatomy or intervention in the past, presented to our cardiology clinic for evaluation of new onset angina. He described a substernal chest pressure occurring occasionally on exertion, intermittent, non-radiating, progressively getting worse, not associated with other symptoms. He endorsed a history of similar presentation and underwent percutaneous coronary intervention with stent placement. Manuscript accepted for publication July 28, 2014 Investigations He was referred for a Cardiolite stress study which showed a fixed inferior defect with ischemia. On coronary angiogram ( Fig. 1), the left main coronary artery was found to have an anomalous origin of the right coronary cusp (from the aorta). It was found to have a 75% distal lesion. A 99% stent restenosis was also revealed. The rest of the angiogram had patent flow and the ventriculogram showed a ventricular function of 75%. A coronary CT angiogram ( Fig. 2-4) also revealed this peculiar origin of the left main artery from the posterior margin of the right coronary cusp, extending posterior to the aortic root, between the aorta and left ventricle. Outcome He was referred for a coronary bypass in which he got a left internal mammary artery (LIMA) to diagonal (artery), saphenous vein graft (SVG) to marginal (obtuse) and right coronary artery. He did well in recovery, meeting his goals and was discharged to follow-up. Discussion Coronary artery anomalies constitute a group of congenital malformations that have a multitude of clinical manifestations and highly variable pathophysiology. The normal anatomy of the coronary arteries is characterized by two ostia centrally placed in the left and right aspects of the sinus of Valsalva just above the left and right coronary cusps respectively. The main left coronary artery (LCA) originates from the left ostium and branches into the left anterior descending artery (LAD) and the circumflex artery (LCX) which travels around the left atrioventricular groove. The right coronary artery (RCA) arises from the right ostium providing an infundibular branch to the anterior side of the heart; then courses back up in the atrioventricular groove. The main coronary arteries branch superiorly to the atria and inferiorly to the ventricles; they end in short fanning branches that penetrate the myocardium [1]. In order to address the normal spectrum of variations in this architecture however, it has been suggested that any form prevalent in > 1% of the general population be considered normal [2,3]. Hemodynamically significant coronary artery anomalies may be classified as primary or secondary. The primary congenital coronary artery anomalies include anomalous origin of the coronary arteries from the aortic sinus, anomalous origin from the pulmonary artery, coronary artery stenosis, absent coronary artery, or coronary artery fistula. The secondary forms occur in conjunction with congenital heart disease [1]. In a necropsy study reported by Alexander and Griffith, the incidence of primary coronary anomalies in a series of 18,950 patients was 0.3% [4]. More recent studies however, suggest a much higher figure of 5.6% in patients studied by coronary angiography [2]. The large difference may be a result of strict diagnostic criteria and warrants further study into the subject. These rare anomalies are usually detected with abnormalities in myocardial perfusion or hemodynamic abnormalities or in cases with progressive atherosclerosis. They can result in mild symptoms such as dyspnea to having severe manifestations such as sudden death. Cases with milder symptoms are usually not detected during life and even on post-mortem examination [1]. Anomalous origin of the LCA is a very rare clinical entity. It may arise from a number of sites, most importantly the pulmonary artery and right sinus of Valsalva (RSOV). Ectopic origin of the coronary artery from the opposite sinus of Valsalva may have important clinical manifestations such as exertional sudden cardiac death especially in younger patients [5]. In a study of 1,950 patients undergoing coronary angiography, only 0.15% were reported to have anomalous origin of the LCA from the RSOV [2]. Sudden cardiac death has particularly been notified among young adults with this anomaly [6]. The clinical syndrome depends on the route of the aberrant left main; it may have an intramural course between the aorta and pulmonary artery, an intraseptal course, a wraparound course in the posteroanterior interventricular groove, or it may run anterior to the pulmonary outflow [7,8]. Compression of the LCA can occur because of the anterior course where it runs between the pulmonary artery and the aorta. This can result in poor perfusion of the myocardium leading to ischemia manifested by chest pain or even sudden death [9]. It has not yet been established whether those patients in whom the left main courses within the septal muscles as opposed to taking an interarterial route, are at a lower risk of sudden cardiac death [6]. Coronary angiography remains the gold standard for the diagnosis of coronary anomalies [1] as opposed to echocardi- ography which may give false negatives [10]. However, echocardiography, computed tomography and magnetic resonance imaging have been recommended as screening entities [11,12]. Intravascular ultrasound may be used to assess the extent of stenosis and the need for interventional treatment [13,14]. There are three options to treat symptomatic patients with such anomalies. Medical treatment (B-blockers) may be as effective as lifestyle modifications (avoidance of strenuous activity) in such patients [15]. Percutaneous coronary intervention with stent placement may be a reasonable option for anomalies with interarterial coursing and risk of systolic compression [13]. Early surgical correction has been the mainstay of treatment for many years and has shown the best outcome [16]. For anomalous origin of the LCA from the right cusp, surgical intervention has been highly recommended and it may be in the form of coronary artery bypass grafting to re-implant the ectopic artery into the left sinus of Valsalva, or by creating a longitudinal opening in the wall of the aorta in the intramural segment of the anomalous artery (osteoplasty) [8,14].	v3-fos-license
2022-09-30T15:02:36.184Z	2022-09-27T00:00:00.000	252617946	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2309-608X/8/10/1015/pdf?version=1664274690", "pdf_hash": "2d66520ccab0033728b0e64b9471a727abd5c566", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45464", "s2fieldsofstudy": [ "Biology" ], "sha1": "7d80db004ebae5464461691f8784a5c58c227261", "year": 2022 }	pes2o/s2orc	Sordaria macrospora Sterile Mutant pro34 Is Impaired in Respiratory Complex I Assembly The formation of fruiting bodies is a highly regulated process that requires the coordinated formation of different cell types. By analyzing developmental mutants, many developmental factors have already been identified. Yet, a complete understanding of fruiting body formation is still lacking. In this study, we analyzed developmental mutant pro34 of the filamentous ascomycete Sordaria macrospora. Genome sequencing revealed a deletion in the pro34 gene encoding a putative mitochondrial complex I assembly factor homologous to Neurospora crassa CIA84. We show that PRO34 is required for fast vegetative growth, fruiting body and ascospore formation. The pro34 transcript undergoes adenosine to inosine editing, a process correlated with sexual development in fruiting body-forming ascomycetes. Fluorescence microscopy and western blot analysis showed that PRO34 is a mitochondrial protein, and blue-native PAGE revealed that the pro34 mutant lacks mitochondrial complex I. Inhibitor experiments revealed that pro34 respires via complexes III and IV, but also shows induction of alternative oxidase, a shunt pathway to bypass complexes III and IV. We discuss the hypothesis that alternative oxidase is induced to prevent retrograde electron transport to complex I intermediates, thereby protecting from oxidative stress. Introduction Fruiting bodies are complex three-dimensional structures that are formed during sexual propagation in filamentous ascomycetes and basidiomycetes and protect the meiospores that are generated inside [1,2]. They contain distinct cell types that do not occur during vegetative growth, as described in detail for the filamentous ascomycetes Neurospora crassa and Sordaria macrospora [3,4]. How this differentiation is mediated, and which genes play a role in fruiting body formation is the topic of ongoing studies in several ascomycetes, including Aspergillus nidulans, N. crassa, Podospora anserina and S. macrospora [1]. S. macrospora has served as a model system for fruiting body formation since the 1950s [5][6][7][8][9]. It forms pear-shaped fruiting bodies named perithecia after seven days of growth on a nutrient-rich medium without the need for a mating partner. The lack of aerial hyphae and mitospores allows easy access to perithecia precursor structures and means that only one developmental program is carried out, making data interpretation straightforward. Several mutagenesis approaches led to over a hundred developmental mutants with defects in perithecia formation, which were sorted into different types according to the stage of the developmental block [6]. The "pro" type mutants are blocked after the formation of protoperithecia, which are spherical perithecia precursors formed after three to four days of growth. Complementation, as well as next-generation sequencing (NGS) approaches, have led to the identification of the underlying mutation in ten pro mutants so far reviewed in [5,8]. However, a considerable number of mutants remains uncharacterized. Work on S. macrospora and other fungi has led to the identification of distinct developmental proteins as well as signaling cascades and complexes and transcription factors, among others, that are required for or associated with the fruiting body and ascospore formation for an overview, see [1,5]. Another process that has recently been correlated to sexual reproduction in filamentous ascomycetes is adenosine (A) to inosine (I) mRNA editing. Editing has the potential to cause amino acid changes in proteins and thereby modify or change protein functions [10,11]. Further, several studies have identified molecular factors and pathways with a high energy demand to be linked to development. For example, autophagy is required for fruiting body formation in diverse fungi, probably to maintain high energy levels and sustain the fruiting bodies from the surrounding mycelium [12][13][14][15]. Moreover, several enzymes from primary metabolism have been shown to be required for correct development [16][17][18][19], and mitochondrial respiration has been linked to fruiting body formation by the sterile phenotype of several N. crassa complex I mutants [20,21]. In this study, we aimed to identify new developmental genes important for perithecia formation in S. macrospora in a forward genetics approach. By sequencing the genome of sterile mutant pro34, we identified a large deletion in the pro34 gene encoding a putative respiratory chain assembly factor. We show that PRO34 is required for fast vegetative growth, perithecia and ascospore formation. The pro34 transcript undergoes RNA editing, leading to an amino acid codon change. Loss of the mitochondrial PRO34 leads to respiratory defects due to lack of complex I and incorrect assembly of mitochondrial supercomplexes. Feeding electrons into the respiratory chain thus might proceed via complex II or alternative complexes (i.e., alternative NADH dehydrogenases) which are commonly found in fungi [22]. Although pro34 respires via complexes III and IV of the canonical respiratory chain pathway, surprisingly, concomitant induction of the alternative oxidase (AOX) is observed. Respiration via AOX bypasses complexes III and IV, and therefore we assume that its induction is a consequence of complex I impairment and may represent a mechanism for protection against oxidative damage. Strains and Growth Conditions All S. macrospora strains used in this study are listed in Table S1. Unless stated otherwise, growth conditions were as described previously [23,24]. Transformation was carried out by protoplast formation as described [23], and transformants were selected on medium containing nourseothricin (50 µg/mL). Homokaryotic strains were generated by isolating ascospores from selfed perithecia for fertile strains. Crosses were set up by inoculating agar blocks of two different strains on opposite sides of a corn meal-malt medium (BMM) plate and incubation for 7-14 days in constant light at 20-25 • C. For measuring vegetative growth, strains were pre-cultured on BMM for two days, and standard inoculants were transferred to new BMM plates. The growth front was marked after one and two days, and the distance between these two marks was measured. Data are from two independent experiments with two technical replicates each. For growth tests with AOX inhibitor salicylhydroxamic acid (SHAM, Sigma S607) and complex I inhibitor rotenone (Sigma R8875), three technical replicates per strain were tested on BMM with solvent and/or inhibitor. SHAM and rotenone were dissolved in water and dimethyl sulfoxide (DMSO) to a final concentration of 450 µg/mL and 30 µM, respectively. The growth tests with paraquat (1,1 -dimethyl-4,4 -bipyridinium dichloride; Sigma 856177) were performed with one (wild type, pro34) or two biological replicates. To this end, paraquat was diluted in water to a final concentration of 20, 100, or 200 µM. Hyphal fusion was observed after two days of growth on MMS with cellophane as described using the AxioImager M.1 microscope (Zeiss) [27]. Perithecia formation was assayed on BMM plates after seven days of growth using a Stemi 2000-C stereomicroscope (Zeiss) equipped with an AxioCamERc5s digital camera (Zeiss) and AxioVision software (Zeiss). Ascospore formation was assayed after ten days of growth on BMM plates. Perithecia were cracked open and ascus rosettes were imaged on slides using the AxioImager M.1 microscope (Zeiss). Images were processed with Adobe CS4 and CS6 (Adobe Corp.). DNA Preparation, Illumina Sequencing and Mapping Mutant pro34 from the laboratory collection of General and Molecular Botany at Ruhr-University Bochum was back-crossed several times to the wild type or brown-spored fus [32] ( Figure S1A). Forty sterile strains harboring the pro34 phenotype and generating brown spores (fus background) were collected from a cross of pro34 to fus. Forty black-spored fertile strains, representing the wild type, were collected from three crosses of mutants pro30, pro32, and pro34 to fus [33] ( Figure S1B). Mutants pro30 and pro32 have been described elsewhere [23,33]. DNA was extracted as described previously [32], pooled for samples pro34/fus and wt_3, respectively, and subjected to 50 bp paired-end Illumina/Solexa sequencing with a HiSeq2000 at GATC Biotech (Konstanz, Germany). Cleaning of raw data, mapping to the S. macrospora reference genome [34,35], analysis of sequence variants, and detection of uncovered regions was performed as described [32] using the Burrows Wheeler Alignment tool [36], SAMtools [37], and custom-made Perl scripts. Genome sequencing data have been deposited at the sequence read archive (SRA; acc. no. SRX483576 and SRX502852 for pro34/fus and wild type (wt_3) [23], respectively. Nucleic Acid Isolation, cDNA Synthesis and Verification of RNA Editing RNA isolation from mycelia was performed as described before [38]. For RNA isolation from perithecia, wild type was grown for six days on BMM plates at 27 • C and in constant light. Perithecia were scraped from the plates with a spatula and immediately transferred to 4 • C 100 % ethanol. After two to five days of fixation at 4 • C, perithecia (corresponding to approx. 100 µL volume) were transferred into a tissue grinder (Radnoti) and homogenized for 30 min in 300-500 µL of a mixture of lysis buffer (0.6 M NaCl, 2% (w/v) SDS, 10 mM EDTA, 100 mM Tris-HCl, pH 8.0), phenol and chloroform:isoamyl alcohol (24:1) in a relation of 2:1:1. Phenol extraction and DNA precipitation were performed subsequently as for mycelium samples. cDNA synthesis was performed as described before [39] with the following modifications: After DNase treatment, RNA was combined with 0.5 µg oligo-dT 12-18 primers (Invitrogen) and 0.6 µg of random hexamers (Thermo Scientific). Reverse transcription was performed with MMLV-RT (RNase H − , Promega) according to the manufacturer's protocol for 1h at 42 • C. For verification of pro34 editing, PCR was performed on cDNA using primer pair 2694-04/2694-07 and PCR products were subjected to Sanger sequencing with primer 2694-07. Western Blot Analyses Total protein extraction from S. macrospora and western blot analysis with an anti-GFP antibody to detect PRO34-GFP fusions were performed as described before [40] using JL-8 primary antibody (Living Colors; TaKaRa Bio Europe/Clontech) and an anti-mouse IgG horseradish peroxidase-linked secondary antibody (Cell Signaling). Mitochondrial protein extracts from S. macrospora were isolated according to [42]. To obtain the post-mitochondrial fraction, briefly, the supernatant of the first centrifugation step after the glass wool filtration was used. As this fraction contains large amounts of bovine serum albumin (BSA; as component of the isolation buffer), the protein concentration could not be determined and thus the maximal possible amount (16 µL) was loaded, while 100 µg mitochondrial fraction was used for each lane. After separation of proteins on a 12% SDS polyacrylamide gel, proteins were transferred to PVDF membranes (Immobilon-FL, Millipore) using standard protocols. Blocking and antibody incubation of blotted PVDF membranes were performed according to the LI-COR Odyssey "Western Blot Analysis" handbook. A Sauromatum guttatum antibody raised against the SgAOX (UniProt: P22185) full-length protein (Agrisera, product code AS20 699) was used to detect S. macrospora AOX proteins (UniProt: F7VMK4, F7WA39) (antibody dilution 1:100). The primary antibody to detect SmPORIN was raised against full-length PaPORIN of P. anserina (UniProt: B2B736) (New England peptide, antibody dilution 1:5000). To detect GFP, a GFP antibody of Sigma (G6795) was used (1:10,000). Secondary antibodies conjugated with the infrared dyes IRDye 800CW or IRDye 680 CW (LI-COR) were used (antibody dilution 1:15,000) for detection with the "Odyssey Infrared Imaging System" (LI-COR). BN-PAGE and " In-Gel" Activity Assay BN-PAGE was performed according to the protocol described in detail by [43]. To this end, for each lane, 100 or 200 µg mitochondrial protein extracts were solubilized using a digitonin/protein ratio of 3:1 (w/w). Linear gradient gels (4-13%) overlaid with 3.5% stacking gels were used for the separation of the solubilized samples. The components of the respiratory chain were visualized by Coomassie blue staining and assigned according to [44]. To identify complex IV Coomassie blue staining was omitted, and the gel was incubated in the dark in 50 mM phosphate buffer (pH 7.4) containing 1 mg/mL 3,3diaminobenzidine, 24 U/mL catalase, 1 mg/mL cytochrome c and 75 mg/mL sucrose [45]. To detect complex I, again Coomassie blue staining was omitted. Instead, the gel was treated with 0.1 M Tris-HCl, 0.14 mM NADH, 1.0 mg/mL nitroblue tetrazolium (NBT), pH 7.4 [45] and incubated for at least 15 min in the dark. For "in-gel" SOD activity measurements, mycelial pieces of S. macrospora wild type, pro34, and pro34 transformants were grown for two days on BMM covered with cellophane foil at 27 • C under constant light and subsequently under the same conditions in liquid CM [46] for additional two days. Total S. macrospora protein extracts were isolated as described for P. anserina in [41]. "In-gel" SOD activity was determined with total protein extracts on a native polyacrylamide gel according to [41] with 100 µg total protein extract. As a loading control, an identical gel was stained with Coomassie according to standard protocols. Oxygen Consumption Measurement Mycelial pieces of S. macrospora wild type, pro34, and pro34 transformants were grown for two days on BMM at 27 • C under constant light and subsequently under the same conditions in liquid CM [46] for additional two days. Subsequently, small pieces of mycelium (0.3 to 5 mg) were transferred into the chamber of the "OROBOROS Oxygraph-2k" (OROBOROS INSTRUMENTS) high-resolution respirometer and oxygen consumption rate was measured in liquid CM medium according to the manufacturer's instructions. KCN was added to a final concentration of 1 mM to inhibit respiration via complex IV. Respiration via AOX was inhibited by adding SHAM (final concentration 4 mM). For each condition, at least four different mycelial pieces were analyzed. To illustrate relative oxygen consumption after the addition of the inhibitor, absolute oxygen consumption of the respective strain in the presence of KCN or SHAM was normalized to its total absolute oxygen consumption with no added inhibitors. Sterility in Mutant pro34 Is Caused by a Genomic Deletion in SMAC_02694 In this study, we searched for so far unknown developmental genes that control sexual development in S. macrospora. Therefore, we analyzed non-allelic mutants pro9, pro24, pro34 and pro42 previously generated in a large-scale mutagenesis project for S. macrospora [6,47]. In contrast to the wild-type strain, which generates perithecia after seven days of growth on solid medium, these mutants are unable to generate perithecia and are thus sterile. We first assessed whether the mutation in any of the four mutants was allelic to an already described mutation causing a sterile phenotype in S. macrospora [23,32,33,40,[47][48][49][50][51][52] or a mutation in another yet uncharacterized pro mutant. Therefore, sexual crosses were set up between pro9, pro24, pro34 and pro42 and 19 further sterile mutants with the "pro" phenotype (Table S4). In such crosses, recombinant fertile perithecia are generated in the contact region between two sterile strains in case of non-allelic mutations, while perithecia are never generated in the contact region between strains with allelic mutations. For pro9, pro24 and pro42, several crosses showed no perithecia formation (Table S4), indicating that a defect in an already described developmental gene caused sterility in these mutants. In contrast, all pro34 crosses showed massive production of perithecia in the contact region; thus, the mutation underlying the sterile phenotype in pro34 had not yet been identified. To identify this mutation, the genome of a pool of 40 brown-spored pro34/fus strains was sequenced and analyzed using a previously established pipeline [32]. A summary of the sequencing results is shown in Table S5. Mapping of Illumina reads to the S. macrospora wildtype reference genome version 2 [34,35] revealed no small variant with 100% penetrance in the mutant. However, searching for unmapped regions showed a 1271 bp deletion in contig 2.11, bases 718959-720229, corresponding to bases 323-1593 of the SMAC_02694 ORF. The SMAC_02694 ORF has a length of 2641 bp, contains one intron at position 2379-2472 and codes for a protein of 848 amino acids. We crossed the pro34 strain again to fus to characterize the progeny and to prove that the sterile phenotype co-segregates with the deletion in the SMAC_02694 ORF. PCR analysis of ten fertile and ten sterile ascospore isolates from this cross indeed showed the deletion in all of the sterile strains, but none of the fertile strains ( Figure S2). We renamed SMAC_02694 pro34. Fertility in pro34 Is Restored by Reintroduction of the pro34 Gene In the sterile pro34 mutant, the deletion in the pro34 ORF leads to a frame-shift, which at the protein level results in an altered amino acid sequence from position 108 to a premature stop after 183 amino acids. To analyze the effect on fruiting body formation more closely, strains were subjected to microscopic analysis after seven days of growth. At this time point, the wild type forms pear-shaped perithecia with a diameter of 300-500 µm and a size of 2 mm ( Figure 1A). Unlike the wild type, pro34 was unable to generate mature perithecia and stopped development after the formation of spherical melanized protoperithecia with a diameter of about 110 µm ( Figure 1A). In contrast to wild-type perithecia, which contain asci with eight melanized ascospores each, protoperithecia of pro34 rarely contained discernable asci that sometimes harbored individual immature spores ( Figure 1A, inset). To complement this phenotype, we generated plasmids carrying the complete pro34 gene controlled by the native promoter (pPRO34-NE) or the constitutively active A. nidulans gpd promoter [53] (pPRO34-CE). Transformation of pro34 with each plasmid yielded fertile strains showing perithecia and ascospore formation like the wild type on the BMM fructification medium ( Figure 1A). Thus, the deletion in pro34 indeed causes the sterile phenotype of the mutant. Further, the complementation of the sexual phenotype seems to be independent of the utilized promoter. Notably, a defect in fruiting body formation is coupled to a defect in vegetative hyphal fusion in most described S. macrospora mutants [8]; however, pro34 is capable of vegetative hyphal fusion ( Figure S3). This result indicates that PRO34 is specifically required for fruiting body and ascospore formation. The sexual phenotype of pro34 and complemented strains. Strains were grown for seven and ten days on BMM for assaying perithecia and asci, respectively. (A) The phenotype of wild type, pro34 and transformants with native pro34 gene. Wild type (wt) generates black, pear-shaped fruiting bodies with mature black ascospores, while pro34 generates immature black spherical protoperithecia that rarely contain asci that may contain immature spores (inset, arrowhead). Fruiting body and ascospore formation is restored by integration of wild-type pro34 controlled by the native promoter (NE) and the constitutive A. nidulans gpd promoter (CE). The black scale bar is 100 µ m; the white scale bar is 20 µ m. (B) The phenotype of transformants with PRO34 to GFP fusions. Fruiting body formation is restored by GFP fusions, regardless of the fusion site and the promoter, but ascospore formation is not restored to wild-type levels. The black scale bar is 100 µ m. The pro34 Transcript Undergoes A-to-I mRNA Editing Figure 1. The sexual phenotype of pro34 and complemented strains. Strains were grown for seven and ten days on BMM for assaying perithecia and asci, respectively. (A) The phenotype of wild type, pro34 and transformants with native pro34 gene. Wild type (wt) generates black, pear-shaped fruiting bodies with mature black ascospores, while pro34 generates immature black spherical protoperithecia that rarely contain asci that may contain immature spores (inset, arrowhead). Fruiting body and ascospore formation is restored by integration of wild-type pro34 controlled by the native promoter (NE) and the constitutive A. nidulans gpd promoter (CE). The black scale bar is 100 µm; the white scale bar is 20 µm. (B) The phenotype of transformants with PRO34 to GFP fusions. Fruiting body formation is restored by GFP fusions, regardless of the fusion site and the promoter, but ascospore formation is not restored to wild-type levels. The black scale bar is 100 µm. The pro34 Transcript Undergoes A-to-I mRNA Editing Recently it was described that adenosine (A) to inosine (I) mRNA editing occurs during fruiting body formation in various filamentous ascomycetes reviewed in [10,11]. Editing often leads to amino acid variations of the encoded proteins and tends to affect genes with developmental functions, especially in ascospore formation, and genes that are differentially regulated during development [54][55][56]. We therefore searched for editing sites in the pro34 transcript in published transcriptome datasets. Indeed, a putative A-to-I editing site was detected in the pro34 gene in RNA-seq data from the wild type, but not from two ascospore formation mutants ∆asm2 and ∆asm3 and mutant ∆spt3, lacking a subunit of the SAGA (Spt-Ada-Gcn5 acetyltransferase) transcriptional coactivator complex [57]. We verified the editing site in pro34 by Sanger sequencing, where an exchange of A to guanosine (G) is indicative of an A-to-I editing event [56] (Figure 2). Genomic DNA displayed the expected A at position 2562 of the pro34 transcript. Evidence for editing was detected in a cDNA sample from six days old perithecia, but not in cDNA samples from five-and six-day-old mycelium undergoing sexual differentiation ( Figure 2). These data indicate that editing of the pro34 transcript is correlated with fruiting body formation. In the pro34 transcript, editing of A2562 leads to a codon change, causing an amino acid exchange of tyrosine to cysteine at position 756 of the PRO34 protein. Tyr756 resides within a globular domain predicted by the eukaryotic linear motif (ELM) resource [58], and a comparison with PRO34 homologs searched for at FungiDB [59] showed that this tyrosine is highly conserved ( Figure S4). Thus, transcript editing might have a functional effect on the PRO34 protein. We verified the editing site in pro34 by Sanger sequencing, where an exchange of A to guanosine (G) is indicative of an A-to-I editing event [56] (Figure 2). Genomic DNA displayed the expected A at position 2562 of the pro34 transcript. Evidence for editing was detected in a cDNA sample from six days old perithecia, but not in cDNA samples from five-and six-day-old mycelium undergoing sexual differentiation ( Figure 2). These data indicate that editing of the pro34 transcript is correlated with fruiting body formation. In the pro34 transcript, editing of A2562 leads to a codon change, causing an amino acid exchange of tyrosine to cysteine at position 756 of the PRO34 protein. Tyr756 resides within a globular domain predicted by the eukaryotic linear motif (ELM) resource [58], and a comparison with PRO34 homologs searched for at FungiDB [59] showed that this tyrosine is highly conserved ( Figure S4). Thus, transcript editing might have a functional effect on the PRO34 protein. PRO34 Localizes to Mitochondria The putative PRO34 polypeptide is orthologous to complex I intermediate associated protein 84 (CIA84) from N. crassa ( Figure S4) that has been described to function in mitochondrial complex I assembly [60]. PRO34 and CIA84 show 85 % identity at the amino acid level. Further, MitoFates [61] predicts a mitochondrial targeting peptide (mTP) at the N-terminus of PRO34 that is cleaved after 22 amino acids. To determine whether or not PRO34 localizes to mitochondria in S. macrospora, we generated translational GFP to PRO34 fusions. Plasmids pPRO34-GFP-CE and pGFP- PRO34 Localizes to Mitochondria The putative PRO34 polypeptide is orthologous to complex I intermediate associated protein 84 (CIA84) from N. crassa ( Figure S4) that has been described to function in mitochondrial complex I assembly [60]. PRO34 and CIA84 show 85 % identity at the amino acid level. Further, MitoFates [61] predicts a mitochondrial targeting peptide (mTP) at the N-terminus of PRO34 that is cleaved after 22 amino acids. To determine whether or not PRO34 localizes to mitochondria in S. macrospora, we generated translational GFP to PRO34 fusions. Plasmids pPRO34-GFP-CE and pGFP-PRO34-CE encode C-and N-terminally tagged PRO34, respectively, expressed from the constitutive A. nidulans gpd promoter. Plasmid pPRO34-GFP-NE encodes C-terminally tagged PRO34 expressed from the native pro34 promoter. All three plasmids were transformed into pro34, and transformants showed restored fruiting body formation ( Figure 1B). However, ascospore formation was not restored to wild-type levels, indicating interference of the GFP tag with the function of PRO34 specifically during ascospore formation. Yet, we decided to further analyze these strains, because fruiting body formation per se was fully restored and the GFP tag enabled monitoring of the PRO34 protein in cellular sub-fractions by western blot analysis. Western blot analysis using a GFP antibody showed that full-length GFP-tagged PRO34 with an expected size of 123.8 kDa was only detected in strains carrying C-terminally tagged PRO34, but not in strains carrying N-terminally tagged PRO34, possibly due to cleavage of the GFP ( Figure S5). Three strains with distinct GFP signal were chosen for further microscopic analysis. Strains PRO34-GFP-CE1 and PRO34-GFP-CE2, carrying plasmid pPRO34-GFP-CE, showed fluorescence in mitochondria, revealed by co-staining with MitoTracker ( Figure 3A). Strain GFP-PRO34-CE1 generating N-terminally tagged PRO34 showed mostly cytoplasmic fluorescence ( Figure 3A). Western blot analysis of mitochondrial and post-mitochondrial fractions confirmed this localization ( Figure 3B). The PRO34-GFP fusion protein was present in the mitochondrial fraction of PRO34-GFP-CE1, while the GFP-PRO34 fusion protein in GFP-PRO34-CE1 was not detected in the mitochondrial, but in the post-mitochondrial fraction. This can be explained by the cleavage of the N-terminal GFP tag during mitochondrial import as well as only partial import of this fusion protein. Interestingly, as mentioned above, both N-and C-terminally tagged PRO34 fusion proteins were able to restore perithecia formation in the mutant ( Figure 1B). However, ascospore formation was not restored to wild-type levels in all strains carrying GFPtagged PRO34, independent of the fusion site or the expression level. Yet, differences in complementation effectiveness were evident when assaying vegetative growth. Mutant pro34 showed a reduced growth rate on BMM medium compared to the wild type. Cterminally tagged PRO34 complemented the defect, while N-terminally tagged PRO34 did not ( Figure 3C). In accordance with the localization studies, these results may indicate that mitochondrial import of N-terminally tagged PRO34 is less efficient and that mitochondrial localization of PRO34 is required for complete functionality during vegetative growth. complementation effectiveness were evident when assaying vegetative growth. Mutant pro34 showed a reduced growth rate on BMM medium compared to the wild type. Cterminally tagged PRO34 complemented the defect, while N-terminally tagged PRO34 did not ( Figure 3C). In accordance with the localization studies, these results may indicate that mitochondrial import of N-terminally tagged PRO34 is less efficient and that mitochondrial localization of PRO34 is required for complete functionality during vegetative growth. is not. Anti-PORIN antibody against P. anserina PORIN was used as a control for mitochondrial localization. P. anserina wild type was used as a control strain. (C) Vegetative growth is reduced in pro34 and in GFP-PRO34-CE1. Strains were grown on BMM and growth was measured in a 24h-interval. Data are mean and standard deviation from two biological replicates with two technical replicates each. Mitochondrial Respiratory Complex Assembly Is Altered in pro34 The PRO34 homolog of N. crassa, CIA84, has a role in mitochondrial complex I assembly [60]. We therefore analyzed mitochondrial respiratory complexes in the S. macrospora wild type, which has not been done before, and the pro34 mutant. As a control, the P. anserina wild type was used, since mitochondrial respiratory complexes have been extensively studied in this model organism [41,44,62]. Complex I and complex IV staining was employed after BN-PAGE to identify the mitochondrial respiratory chain complexes as described before [44,45] (Figure 4). While P. anserina predominantly formed mitochondrial supercomplexes S 1 (I 1 III 2 IV 1 ) and S 2 (I 1 III 2 IV 2 ), S. macrospora wild type predominantly formed supercomplex S 0 (I 1 III 2 IV 0 ), but not complex IV-containing S 1 and S 2 . However, other larger complexes with unknown compositions were present in the S. macrospora wild type. The pro34 mutant lacked complex I and supercomplex S 0 . This lack suggests that electrons from NADH have to be fueled into the respiratory chain by alternative proteins. Such proteins are known as internal and external alternative NADH dehydrogenases (aNADH-DHs) in other fungi, and genes encoding putative proteins with NADH dehydrogenase domains predicted by the NCBI BLAST conserved domain search [63] are also found in the genome of S. macrospora. Specifically, the three predicted proteins SMAC_01935, SMAC_02271, and SMAC_07176 correspond to the described aNADH-DHs NDE-2/Pa_7_5390, NDI-1/NDI1 and NDE-1/Pa_1_24200 from N. crassa and P. anserina, respectively ( Figure S6) [35,[64][65][66]. Complex IV was present in the pro34 mutant but seemed to migrate slower than in the wild type ( Figure 4B). Strains carrying GFP-tagged PRO34 showed intermediary results. Both, PRO34-GFP-CE2 and GFP-PRO34-CE1, showed complex I in smaller amounts than in the wild type. Complex IV was also present in both strains, but like in mutant pro34, the complex migrated slower than in the wild type, supposedly indicating an altered composition. as described before [44,45] (Figure 4). While P. anserina predominantly formed mitochondrial supercomplexes S1 (I1III2IV1) and S2 (I1III2IV2), S. macrospora wild type predominantly formed supercomplex S0 (I1III2IV0), but not complex IV-containing S1 and S2. However, other larger complexes with unknown compositions were present in the S. macrospora wild type. The pro34 mutant lacked complex I and supercomplex S0. This lack suggests that electrons from NADH have to be fueled into the respiratory chain by alternative proteins. Such proteins are known as internal and external alternative NADH dehydrogenases (aNADH-DHs) in other fungi, and genes encoding putative proteins with NADH dehydrogenase domains predicted by the NCBI BLAST conserved domain search [63] are also found in the genome of S. macrospora. Specifically, the three predicted proteins SMAC_01935, SMAC_02271, and SMAC_07176 correspond to the described aNADH-DHs NDE-2 / Pa_7_5390, NDI-1 / NDI1 and NDE-1 / Pa_1_24200 from N. crassa and P. anserina, respectively ( Figure S6) [35,[64][65][66]. Complex IV was present in the pro34 mutant but seemed to migrate slower than in the wild type ( Figure 4B). Strains carrying GFP-tagged PRO34 showed intermediary results. Both, PRO34-GFP-CE2 and GFP-PRO34-CE1, showed complex I in smaller amounts than in the wild type. Complex IV was also present in both strains, but like in mutant pro34, the complex migrated slower than in the wild type, supposedly indicating an altered composition. Coomassie blue staining and assigned according to [44]. The Podospora anserina wild-type strain (P.a. wt) was used as a control strain for mitochondrial complex identification. Complex V monomer (V1) and complex III dimer (III2) migrate at the same size. Complex IV staining in (B) and complex I staining in (C) were performed according to [45]. I1, complex I monomer; III2, complex III dimer; IV1, complex IV monomer; V1, complex V monomer; V2, complex V dimer; S0, supercomplex I1III2IV0; S1, supercomplex I1III2IV1; S2, supercomplex I1III2IV2. The components of the respiratory chain were visualized by Coomassie blue staining and assigned according to [44]. The Podospora anserina wild-type strain (P.a. wt) was used as a control strain for mitochondrial complex identification. Complex V monomer (V 1 ) and complex III dimer (III 2 ) migrate at the same size. Complex IV staining in (B) and complex I staining in (C) were performed according to [45]. I 1 , complex I monomer; III 2 , complex III dimer; IV 1 , complex IV monomer; V 1 , complex V monomer; V 2 , complex V dimer; S 0 , supercomplex I 1 III 2 IV 0 ; S 1 , supercomplex I 1 III 2 IV 1 ; S 2 , supercomplex I 1 III 2 IV 2 . Mutant pro34 Has Respiratory Defects Since PRO34 is a mitochondrial protein and the pro34 mutants shows deficiencies in mitochondrial complex assembly, loss of the protein may lead to respiratory impairments. To assess such impairments, the wild type, pro34, PRO34-GFP-CE1 and GFP-PRO34-CE1 were subjected to respiratory measurements. Standard respiration in animals proceeds via complex I (NADH:ubiquinone oxidoreductase, inhibited by rotenone) or complex II (succinate dehydrogenase), complex III (ubiquinol:cytochrome c oxidoreductase), and complex IV (cytochrome c oxidase (COX), inhibited by potassium cyanide, KCN). In fila-mentous fungi several alternative respiratory chain components including the abovementioned aNADH-DHs and an alternative terminal oxidase (AOX) are active under certain conditions [22,67]. In particular, the AOX (inhibited by SHAM) is well characterized. The enzyme allows to bypass of complexes III and IV at the dispense of lower proton pumping. We used KCN and SHAM to analyze the respiration of S. macrospora mycelium. Respiration of the wild type was strongly affected by KCN, indicating that the wild type predominantly respires via the canonical COX-dependent pathway ( Figure 5A). SHAM had only a minor effect on the respiration of the wild type. In contrast, while KCN still affected pro34 respiration, SHAM affected the mutant significantly stronger than the wild type, indicating that pro34 also respires via the AOX pathway. However, although at a lower level, COX (KCN-sensitive)-dependent respiration is still observed in the mutant. Subsequently, we analyzed the phenotypes of the transformants carrying PRO34 to GFP fusion proteins. In both strains, the inhibitory effect by KCN was stronger than by SHAM, like in the wild type. In PRO34-GFP-CE1, however, inhibition by SHAM was significantly lower than in pro34 ( Figure 5A), indicating that respiration is mainly COX-dependent. GFP-PRO34-CE1 showed an intermediary phenotype, and the level of SHAM inhibition was not significantly different from that of the pro34 mutant, indicating that respiration is both, COX-and AOX-dependent. Mutant pro34 Shows Increased Resistance to Oxidative Stress A defect in mitochondrial respiration often leads to increased oxidative stress and thus affects oxidative stress resistance. To test this possibility, the mutant and the transformants were grown for four days on media containing different concentrations of paraquat, an inducer of mitochondrial superoxide production [70,71]. In contrast to wild type, mutant pro34 was able to grow on media with 200 µ M paraquat ( Figure 6). Transformants showed different growth phenotypes, with mainly strains carrying GFP-tagged PRO34 being able to grow on high paraquat concentrations, suggesting again that the GFP tag impairs PRO34 function. To confirm the AOX-dependent respiration in pro34, we performed a western blot analysis with an anti-AOX antibody ( Figure 5B). Two genes coding for AOX proteins can be found in the S. macrospora genome. SMAC_08419 and SMAC_08566 correspond to AOD-3 and AOD-1 from N. crassa [68,69], respectively, and have a deduced size of 34 kDa and 32 kDa, respectively, after cleavage of their mTPs. The western blot ( Figure 5B) indeed shows that AOX was induced in pro34 in comparison to wild type. Comparable amounts of AOX were found in strain GFP-PRO34-CE1. In PRO34-GFP-CE1, only small amounts of the AOX protein were still detectable in a western blot analysis ( Figure 5B). Taken together, the introduction of a GFP-tagged PRO34 into the mutant reverted the respiration phenotype only partially, and the N-terminal tagging of PRO34 seems to either interfere stronger with its function or affect mitochondrial import efficiency. Vegetative growth tests on a medium containing SHAM corroborated these data ( Figure 5C). Growth of pro34 and GFP-PRO34-CE1 was affected on media containing SHAM. Strains PRO34-GFP-CE1 and PRO34-GFP-CE2 showed growth similar to that of the wild type, while strain GFP-PRO34-CE1 was more similar to pro34. We performed further growth tests on media containing complex I inhibitor rotenone with the same strains. Here, DMSO was used as a solvent. Although DMSO itself strongly affected growth rate, Figure 5D shows that growth of all strains except pro34 is reduced by rotenone. This result fits well with the observed lack of complex I in the mutant. PRO34-GFP-CE1, PRO34-GFP-CE2 and GFP-PRO34-CE1 did not show significantly altered growth rates in comparison to the wild type. Taken together, our data show that pro34 respires mainly via the AOX pathway and is less sensitive to complex I inhibition. Mutant pro34 Shows Increased Resistance to Oxidative Stress A defect in mitochondrial respiration often leads to increased oxidative stress and thus affects oxidative stress resistance. To test this possibility, the mutant and the transformants were grown for four days on media containing different concentrations of paraquat, an inducer of mitochondrial superoxide production [70,71]. In contrast to wild type, mutant pro34 was able to grow on media with 200 µM paraquat ( Figure 6). Transformants showed different growth phenotypes, with mainly strains carrying GFP-tagged PRO34 being able to grow on high paraquat concentrations, suggesting again that the GFP tag impairs PRO34 function. The observed resistance to paraquat may be caused by increased levels of superoxide dismutase (SOD), the enzyme that detoxifies superoxide. We, therefore, performed a western blot analysis with antibodies raised against Cu/Zn-SOD, and PaSOD2, an ER-localized / secreted Mn-SOD [72]. The S. macrospora genome encodes six proteins with putative SOD function, SMAC_ 00334, SMAC_00396, SMAC_01384, SMAC_03915, SMAC_05035 and SMAC_05700. Comparison with P. anserina SODs indicates that SMAC_05035 may be the cytosolic Cu/Zn-SOD, SMAC_05700 the mitochondrial Mn-SOD, and SMAC_03915 an ER-localized / extracellular Mn-SOD. Western blot analysis detected one Cu/Zn-SOD and two Mn-SODs in total protein extracts, and "in-gel" SOD activity staining showed the activity of both enzyme types ( Figure S7). However, SOD levels and activity were not altered in the mutant or transformed strains in comparison to the wild type. Thus, enhanced paraquat resistance is not the consequence of elevated superoxide scavenging but rather appears to be due to the lack of complex I and the identified partial respiration via AOX, which does bypass complex III, a major site of superoxide generation The observed resistance to paraquat may be caused by increased levels of superoxide dismutase (SOD), the enzyme that detoxifies superoxide. We, therefore, performed a western blot analysis with antibodies raised against Cu/Zn-SOD, and PaSOD2, an ERlocalized/secreted Mn-SOD [72]. The S. macrospora genome encodes six proteins with putative SOD function, SMAC_ 00334, SMAC_00396, SMAC_01384, SMAC_03915, SMAC_05035 and SMAC_05700. Comparison with P. anserina SODs indicates that SMAC_05035 may be the cytosolic Cu/Zn-SOD, SMAC_05700 the mitochondrial Mn-SOD, and SMAC_03915 an ER-localized/extracellular Mn-SOD. Western blot analysis detected one Cu/Zn-SOD and two Mn-SODs in total protein extracts, and "in-gel" SOD activity staining showed the activity of both enzyme types ( Figure S7). However, SOD levels and activity were not altered in the mutant or transformed strains in comparison to the wild type. Thus, enhanced paraquat resistance is not the consequence of elevated superoxide scavenging but rather appears to be due to the lack of complex I and the identified partial respiration via AOX, which does bypass complex III, a major site of superoxide generation in the standard COX-dependent respiratory chain. In addition, respiration via AOX prevents the over-reduction of ubiquinone which might result in reverse electron transport that generates superoxide. Discussion In this study, we analyzed sterile mutant pro34 from S. macrospora. We found that the mutant has a deletion in the gene pro34, coding for a homolog of mitochondrial complex I chaperone CIA84 from N. crassa [60]. Indeed, mutant pro34 showed impairments in complex I assembly and respiration, indicating a mitochondrial function for PRO34. The mutant was further defective in vegetative growth, fruiting body and ascospore formation. The N. crassa homolog of PRO34, CIA84, has been described as a mitochondrial complex I chaperone [60]. Complex I is an NADH:ubiquinone oxidoreductase complex in the inner mitochondrial membrane. Its function is to couple the electron transfer from NADH to ubiquinone with the translocation of protons across this membrane [73]. Complex I consists of two sub-complexes, a membrane arm and a peripheral arm protruding into the mitochondrial matrix, that forms an L-shaped structure [74,75]. The N. crassa cia84 mutant accumulates the peripheral arm and a small intermediate of the membrane arm, but cannot form the complete membrane arm or the mature complex I [60]. Interestingly, mutant pro34 not only showed complex I deficiency, but also an altered migration of complex IV in a BN-PAGE and a loss of all mitochondrial supercomplexes. To our knowledge, an effect on complex IV has not been described before for a complex I protein, and in N. crassa, complex I was the only affected mitochondrial complex in the cia84 mutant [60,76]. However, in the initial N. crassa study, mitochondrial supercomplexes were not analyzed. In N. crassa, complex I has been described to be required for fruiting body and ascospore formation, but not vegetative growth [20,21,65,77,78]. The cia84 deletion mutant was reported to show reduced vegetative growth and slightly decreased conidiation [60]. Fruiting body formation was not assayed in this study. Our data indicate that in S. macrospora, complex I is required for both, fast vegetative growth as well as the formation of fruiting bodies and ascospores. In general, complex I has been associated with diverse developmental processes in different organisms. Plants and fungi display several bypass options for complex I, III and IV. Nevertheless, they show an impaired redox balance when complex I is dysfunctional [79]. In humans, defects in mitochondrial complex I are associated with diseases of the central nervous system as well as pathologies of the heart and muscle, although the underlying molecular mechanisms remain mostly unclear [80]. Likewise, mitochondrial complex I assembly factors such as CIA30/NDUFAF1 [81,82] have been implicated in human pathologies [83]. Mitochondrial respiration per se has been linked to development before. Besides the above-mentioned N. crassa complex I mutants, many mutants of the aging model P. anserina have been described with defects in mitochondrial respiration [84]. Among these mutants, the ex1 mutant harbors a cox1 deletion [85], thus lacking complex IV (COX) completely, while in the grisea and PaCox17::ble mutants, the delivery of copper to complex IV and thus its function is affected [86,87]. These mutants are either sterile or show a strongly reduced fertility. The developmental phenotypes of respiration mutants may be due to a high energy demand during fruiting body formation, which requires the generation of different cell types as well as the generation of meiotic progeny [1,88]. Indeed, further studies have correlated energy demand with development. For example, half of the genes encoding proteins with sugar transporter domains are differentially regulated during fruiting body development, indicating massive rearrangement of nutrient transport [35]. Further, autophagy is thought to compensate for high energy demands and to redistribute nutrients from the mycelium to the fruiting body. Accordingly, autophagy genes atg-3 and atg-8 from N. crassa are required for protoperithecia development [15], PaAtg1, PaAtg8 and PaAtg24 influence fruiting body formation, ascospore formation as well as ascospore germination in P. anserina [12,13,89], and Smatg4, Smatg8 and Smatg12 are required for fruiting body formation in S. macrospora [14,90]. Enzymes from primary metabolism also play a role in fruiting body formation, among them carbonic anhydrases that catalyze the reversible interconversion between carbon dioxide and bicarbonate, leucine biosynthesis enzyme β-isopropylmalate dehydrogenase, and ATP citrate lyase, which catalyzes the formation of acetyl-CoA and oxaloacetate from CoA [16][17][18][19]. How PRO34 functions at the molecular level during the fruiting body and ascospore formation has to be investigated in future studies. The pro34 transcript undergoes A-to-I RNA editing during fruiting body formation. This type of editing in nuclear protein-coding transcripts has been correlated with fruiting body formation in filamentous ascomycetes [10,11]. Several genes whose transcripts undergo A-to-I editing have been shown to function in ascospore formation, discharge and germination in different fungi. For example, deletion of major facilitator superfamily domain gene amd1 from Fusarium graminearum caused defects in ascus wall formation and ascospore discharge and resulted in ascospores germinating inside the perithecia [91]. Deletion of the serine threonine kinase gene stk-21 from N. crassa resulted in the generation of abnormal asci and a delay in ascospore formation. Here, we describe with pro34 an additional gene whose transcript is affected by RNA editing and which is involved in ascospore formation. For F. graminearum amd1 and N. crassa stk-21 it has been shown that editing is required for proper protein function during fruiting body formation [54,91]. Further studies are necessary to analyze the effect of the single amino acid variation on PRO34 function. In contrast to most other sterile mutants from S. macrospora, pro34 is able to undergo hyphal fusion. S. macrospora mutants with this constellation described so far are the autophagy mutants and mutant spd with the underlying spd4 deletion, lacking a nuclear protein of unknown function [14,90,92]. Interestingly, pro34 seems to have an increased hyphal fusion rate, which might cause the massive production of perithecia in crosses to other sterile mutants, where vegetative hyphal fusion is a prerequisite to generating a competent mycelium. Although the possible mechanism underlying this observation still awaits analysis, pro34 has served as a useful tool for crossing hyphal fusion-deficient strains in the lab (IT, unpublished results). Coming back to the respiratory defect in pro34, the induction of AOX in the mutant is puzzling. The canonical respiration occurs via the highly conserved protein complexes complex I to V, while many fungi, like plants, can induce alternative pathways that bypass one or several protein complexes of the canonical respiratory chain ( Figure 7A) [67]. In case of a lack of complex I, the canonical NADH:ubiquinone oxidoreductase, rotenoneinsensitive aNADH-DHs can overtake its role in feeding electrons from NADH into the respiratory chain by reducing ubiquinone. However, they do not pump protons across the inner mitochondrial membrane [93]. Incidentally, the yeast Saccharomyces cerevisiae lacks complex I completely and by default employs aNADH-DHs for respiration [94,95]. In P. anserina aNADH-DH NDI1 has been shown to rescue a complex I-deficient mutant [66]. As mentioned above, three putative aNADH-DHs are encoded by the S. macrospora genome, including an ortholog of NDI1 ( Figure S6) [35]. Thus, one would expect a complex I mutant to induce these enzymes and further transport electrons via complexes III and IV. Indeed, respiration of mutant pro34 is reduced by KCN-mediated inhibition of complex IV, indicating that the mutant respires via complexes III and IV. We, therefore, propose a model in which the wild type with functional complex I respires via the canonical pathway without any alternative components ( Figure 7B), while pro34 uses aNADH-DHs in combination with complexes III and IV ( Figure 7C). This fine-tuned special setting of the respiratory chain allows the generation of membrane potential at complexes III and VI and leads to reduced ROS and ATP production. The latter may be the key to the observed effects on growth and development. complex IV, indicating that the mutant respires via complexes III and IV. We, therefore, propose a model in which the wild type with functional complex I respires via the canonical pathway without any alternative components ( Figure 7B), while pro34 uses aNADH-DHs in combination with complexes III and IV ( Figure 7C). This fine-tuned special setting of the respiratory chain allows the generation of membrane potential at complexes III and VI and leads to reduced ROS and ATP production. The latter may be the key to the observed effects on growth and development. Main mitochondrial complexes are depicted in blue and numbered. The aNADH-DHs are shown in purple, AOX in red. Cytc, cytochrome c; UQ, ubiquinone. Electrons from NADH are introduced into the respiratory chain by either complex I or aNADH-DHs. Complexes I and III produce superoxide (orange). Proton (yellow) flux is depicted by gray arrows, electron (green) flux by black arrows. The route of electrons in the canonical pathway is shown by solid arrows, the route via alternative components is shown by dashed arrows. Note that only protons transported across the inner mitochondrial membrane and protons used for the generation of water by AOX and complex IV are depicted in the model. Modified from [67,84]. (B) Respiration in wild type follows the canonical route via complexes I through V. (C) Mutant pro34 lacks functional complex I and shows respiration via aNADH-DHs and complexes III and IV. Most probably, complex II and the rotenone-insensitive aNADH-DHs are used for the delivery of reducing equivalents and for feeding electrons to ubiquinone in the mutant. AOX may prevent the reshuffling of electrons to complex I intermediates and thereby prevent oxidative damage. Interestingly, we detected an induction of the AOX enzyme that displays another alternative electron transport route. AOX, found in plants and fungi, is induced to bypass complexes III and IV, and it is able to take electrons from ubiquinone to reduce oxygen to water [67] (Figure 7A,C). Like the aNADH-DH bypass, it does not generate a protonmotive force, and energy is lost as heat [93]. The induction of AOX in pro34 seems to be counterintuitive since the mutant does respire via complexes III and IV. This kind of respiration should provide sufficient protonmotive force and consequently, pro34 is viable and even able to induce sexual development. Interestingly, also in maize plants, AOX was demonstrated to be induced upon complex I deficiency [96]. On the one hand, a possible clue is speculation raised by Maas et al. [97] who observed that in P. anserina neither complex III nor complex IV is required for complex I assembly. This is in sharp contrast to observations in mammals where complex III and complex IV indeed are required for complex I assembly [98][99][100][101]. It seems that in mammals, who do not have an AOX, respiratory supercomplexes which contain complexes III and IV are essential as a kind of scaffold for complex I assembly. It might be possible that in AOX-positive organisms this terminal oxidase can overtake such a function in complex I assembly assistance and therefore be induced in the pro34 mutant that cannot assemble complex I. On the other hand, AOX may prevent the reshuffling of electrons to non-functional complex I intermediates. Such intermediates may be present in mutant pro34 since they have been detected in the N. crassa cia84 mutant [60]. AOX may thereby protect these intermediates from oxidative damage, which itself would lead to more severe phenotypes. Additional studies to validate this scenario are required. The described pro34 mutant certainly is an excellent starting point for such an approach.	v3-fos-license
2018-12-18T01:28:20.675Z	2016-03-28T00:00:00.000	56239999	{ "extfieldsofstudy": [ "Physics" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.epj-conferences.org/articles/epjconf/pdf/2016/09/epjconf_efm2016_02040.pdf", "pdf_hash": "fbc8b43d8c044ac77cc7fc323d59c01747980fcb", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45465", "s2fieldsofstudy": [ "Engineering", "Environmental Science", "Physics" ], "sha1": "fbc8b43d8c044ac77cc7fc323d59c01747980fcb", "year": 2016 }	pes2o/s2orc	The use of wind tunnel facilities to estimate hydrodynamic data Experimental laboratory testing of vortex-induced structural oscillations in flowing water is an expensive and time-consuming procedure, and the testing of high Reynolds number flow regimes is complicated due to the requirement of either a large-scale or high-speed facility. In most cases, Reynolds number scaling effects are unavoidable, and these uncertainties have to be accounted for, usually by means of empirical rules-of-thumb. Instead of performing traditional hydrodynamic measurements, wind tunnel testing in an appropriately designed experimental setup may provide an alternative and much simpler and cheaper framework for estimating the structural behavior under water current and wave loading. Furthermore, the fluid velocities that can be obtained in a wind tunnel are substantially higher than in a water testing facility, thus decreasing the uncertainty from Introduction Vortex-induced vibrations (VIV) of cylinders is a widely studied phenomenon due to its relevance in numerous engineering disciplines.Several reviews have been written on the matter of VIV of circular cylinders [1][2][3], as well as the more complicated cases of interference between two closely separated cylinders [4] and the structural instabilities of rectangular cylinders [5]. A series of wind tunnel test has been performed to investigate the response characteristics of different cylinder cross sections.The objective of the tests is to provide necessary key parameters for the initial design of a submerged floating tube bridge (SFTB) under consideration for the Bjørnafjord crossing south of Bergen, Norway.At the current stage in the design process, the use of wind tunnel testing, as opposed to hydrodynamic testing, is a great advantage. Investigation of vortex-induced structural oscillations in hydrodynamic laboratories is time-consuming and expensive.In addition, the results will have an inherent uncertainty due to Reynolds number scaling effects.Similarity in Reynolds number is exceedingly hard to achieve in a water setup, as it requires either a large-scale setup or very high fluid velocities.The latter situation will also induce very large forces on the model structure.In a wind tunnel, these effects are less prominent, as the fluid velocities that can be obtained are substantially higher than in a waa e-mail: kho@sohansen.dkter testing facility and the inertial forces are much smaller.Wind tunnel experiments allow Reynolds numbers in the range of the resonance wind velocities (10 4 ≤ Re ≤ 10 6 ), which also gives a good basis for predicting the behaviour in full scale.This, along with the relative ease of the setup and the reduction in cost, makes the use of wind tunnel testing to determine hydrodynamic data an attractive alternative to water facility testing. Wind tunnel facilities are widely used in the investigation of vortex dynamics and the related structural issues, and numerous studies of VIV and flow instabilities of cylinders have been carried out in wind tunnels.What these studies have in common, is that they operate with comparatively large mass ratios.For instance, a recent study [6] of VIV on circular cylinders in the post-critical regime refers to a mass ratio of 35 as "quite low with respect to tests in air".A widely sited master's thesis work [7] employs a mass ratio of approximately 250 for wind tunnel experiments.However, for neutrally buoyant structures submerged in water the mass ratio is typically of order 1, and at such low mass ratios the scientific literature does not seem to cover wind tunnel experiments characterising the vortex-induced structural dynamics. Mass and damping have a significant effect on VIV, and in order to obtain a realistic result for the present study, this must be taken into account.The effect of mass and damping on VIV for a circular cylinder has been investigated numerically, and it was found that variations of the mass and damping ratios have almost the same result on the system response.Increasing the mass ratio will reduce the maximum amplitude, and also the velocity range over which lock-in occurs [8].This corresponds to earlier findings [9], where experiments with the same mass-damping parameter were carried out in both air and water.In the present investigation, lightweight helium-filled models have been used for the dynamic tests, in order to get model mass ratios that simulate neutral buoyancy in water. Existing theories and methodologies Vortex-induced vibrations are described mathematically by a number of models of the physics behind vortex formation, the associated oscillating loads and the structural dynamics, incorporating the associated coupling of these phenomena.The physical parameters and principles allow for the formulation of response models used to establish general design criteria primarily related to structural vibrations. Determining parameters and basic principles A number of basic non-dimensional parameters have been established to describe and characterise the main static and dynamic behaviour of structures subjected to in-line and cross-flow vortex-induced vibrations.These parameters depend on relevant structural and flow-specific properties. The specific unique ratio of inertial and viscous forces is given by the Reynolds number, defined by the well-known expression where U is the mean velocity of the undisturbed flow, D is the cross-flow structural dimension and ν is the kinematic viscosity of the fluid. The ratio between the oscillating mass of the structure, being in the model or full-scale configuration, and the mass of the equivalent displaced fluid can be expressed by the mass ratio λ = m osc,structure m osc,fluid . Note that this is a purely structural parameter.In this expression m osc,structure denotes the complete mass of all oscillating parts.For both the numerator and the denominator, the oscillating mass of the surrounding fluid is in this paper approximated by the mass of the displaced fluid.For example, a structure having the same density as the surrounding fluid, will have a mass ratio of 1.A structure with twice the density of the surrounding fluid, will have a mass ratio of 3/2, since the density of the oscillating surrounding fluid reamins unchanged.The Strouhal number is defined by the well-established relation where n s is the frequency of the lateral loads.A similar relationship is defined in oscillatory flow, where the relative orbital motion of fluid particles relative to the characteristic length of the immersed object plays a crucial role.This relationship is expressed by the Keulegan-Carpenter number [10]. For the present analysis, the structural susceptibility to vortex-induced vibrations is characterised by the general mass-damping parameter In the formula above, δ s is the structural damping quantified by the logarithmic damping decrement in still fluid, m e is the equivalent mass per unit length corresponding to the mode considered, ρ is the density of the surrounding fluid, and L is the in-line structural dimension.The presented expression is a generalisation of the Scruton number, taking into account structures having a non-circular cross-sectional geometry [11].Note that in the case of a structure exposed to vortex shedding along its complete extension, the mass-damping parameter is simply a damping term multiplied by the ratio between the density of the oscillating structural mass and the density of the surrounding fluid.The structural response is described in terms of a fundamental relation between the reduced fluid velocity and the reduced amplitude.The reduced fluid velocity is in relation to in-line vibrations defined by where n e is the in-line natural frequency of the structure in still fluid, and the reduced amplitude is defined by where A is the amplitude of the in-line oscillation. It should be noted that the use of the reduced velocity as the parameter characterising the in-line vibrational response may not be optimal, at least for very low mass ratio structures.It seems reasonable that the forcing frequency caused by vortex shedding is dependent on the actual oscillation frequency and not the natural frequency of the structure in a still fluid.For cross-flow oscillations, the velocity normalised by the actual oscillation frequency has been shown to give a better description of the experimental data in some situations [12].However, in the present paper, the reduced velocity is used as the descriptive parameter, since this is in accordance with the majority of the literature and design codes related to in-line oscillations. The basic physical principles of vortex shedding induce an oscillating drag force which drive in-line structural oscillatory vibrations.Two different scenarios can occur, both corresponding to alternating low-pressure vortices on the downstream side of the object.Normal vortex shedding combined with a secondary, symmetric vortex shedding, which occurs as a result of in-line structural motion relative to the fluid, can create downstream pressure variations oscillating with a frequency of approximately three times the Strouhal frequency.This mechanism is responsible for the first fluid instability causing in-line vibrations.The second fluid instability causing in-line vibrations occur for slightly larger fluid velocities, where the forcing is entirely due to normal vortex shedding.Since the shedding is limited to low-pressure vortices on the downstream side of the object, the forcing frequency in the in-line direction is approximately twice the Strouhal frequency.It should also be mentioned, that since vortex-induced in-line vibrations in steady flow occur at relatively low fluid velocities, the oscillation amplitudes are in the order of one magnitude smaller than the corresponding cross-flow vibrations usually experienced at larger flow velocities [10]. In-line amplitude response models Structural vibrations caused by vortex-induced oscillating loads can be described in a simplified fundamental mathematical framework, based on theoretical and experimental research in the oscillating fluid/structure interaction, combined with general accepted experience from the construction and operation of different types of structures susceptible to VIV.To a certain extent, experiments in fluid dynamic testing facilities has also helped to clarify and specify certain aspects of this description. A fundamental description of the in-line reduced amplitude response for submerged cylindrical structures in flowing water can be given in terms of the reduced fluid velocity and a mass-damping parameter.Experiments, such as those performed by King [13], have verified the presence of two fluid instability regions and in broad terms defined the associated in-line response amplitude characteristics. Det Norske Veritas (DNV) has developed applicationbased design criteria and recommendations for free-span submarine pipelines which prescribe in-line response curves based on the reduced velocity and a stability parameter similar to the mass-damping parameter, but taking into account to total modal damping [14].The in-line response model covers both fluid instability regions, corresponding to symmetric and regular, asymmetric vortex shedding for a uniform mode shape.The DNV model is used as a reference guideline for the maximum in-line response of cylindrical low mass ratio structures submerged in steady flowing water; thus, together with the wind tunnel measurements it is used to compare the response characteristic measured for low mass ratio structures in steady air with those expected for similar low mass ratio structures submerged in steady flowing water. In its general description, the maximum reduced in-line amplitude response for neutrally buoyant cylindrical structures submerged in steady flowing water is similar to the principal sketch in figure 1.In King's laboratory testing with an oscillation given by a free-end mode shape, the two instability regions are clearly visible.The DNV specifications are similar, but more general and the response covers are larger area, especially for U r > 3. The DNV model is in figure 1 presented for a stability parameter of 0.2.At higher levels of total damping, the onset reduced velocity is increased slightly and the maximum response is decreased.The reduced velocity above which in-line vibrations are insignificant is also lowered.Note that the response models only describe the response caused by in-line VIV, and other phenomena, such as galloping or torsional structural instabilities, may change the actual response significantly. The ultimate goal of the present analysis is to determine if the known dynamic behaviour of neutrally buoyant structures in steady flowing water, expressed by the presented in-line response models, is similar to the dynamics of neutrally buoyant model structures in air. Figure 1: Principal sketch of the maximum reduced in-line amplitude response as a function of the reduced fluid velocity for low mass ratio cylindrical structures submerged in steady flowing water.The presented general response characteristics are based on published results from laboratory experiments in water for a free-end mode shape [13] (Re = 6 × 10 4 ) and a response model from code specifications for a uniform mode shape, assuming a stability parameter of 0.2 [14]. Experiences and common practice when designing submerged marine structures Vortex-induced vibrations are a key issue in many disciplines of marine engineering design.Currently defined response models, as those presented previously, are used to give overall indications of the onset and intensity of both in-line and cross-flow VIV.Certain structure-specific properties, such as the cross section geometry, inter-structural spacing and the distance to any flow boundaries may, however, influence the response significantly [15].Hydrodynamical tests are as a natural consequence used to investigate the possibility of structural behaviour not covered by existing response models, such as for instance was the case with the initial studies performed at the Norwegian Marine Technology Research Institute (MARINTEK) in connection with the proposed Høgsfjord SFTB crossing and the Ormen Lange pipeline project. The motivation for the present study is to investigate if one can use wind tunnel experiments, as an alternative to hydrodynamic experiments, to obtain information of the hydrodynamic behaviour of cross sections relevant for submerged floating tube bridges.The procedure will also provide valuable indications of how to utilise wind tunnel experiments to link the aerodynamic and hydrodynamic data base closer together and describe a fundamental methodology for the transfer of certain aerodynamic actions to equivalent hydrodynamic actions. Experimental design and setup The performed experimental testing includes measurements of static and dynamic flow-induced model responses performed in wind testing facilities at Svend Ole Hansen ApS, Copenhagen, Denmark and SOH Wind Engineering LLC, Vermont, USA.In the following section, the experimental methodologies of the static and dynamic measurements are outlined and the geometric properties of the analysed cross section models are presented.The analysis considers the following two cross section models: -Circular cylindrical tube -Rectangular 1:3 box For the rectangular box model, the in-line dimension is three times larger than the cross-flow dimension. Static wind tunnel rig To measure the static response in a wind tunnel, a force transducer is placed on the upstream side of the model facilitating measurements of the drag force.Two force transducers are placed at the top to measure the wind-induced lift force and moment.A relatively large mass of the model ensures that the top wires achieve a sufficient pretension such that a positive force is always measured on the top force transducers.The setup is identical on the other side of the wind tunnel, with a similar force transducer placed on the upstream side of the model.A principal sketch of the measurement setup used for static testing is depicted in figure 2. The output from the calibrated force transducers, knowledge of the model geometry and measurements of the flow velocity allow for the estimation of the static coefficients on the fixed structure.These are defined by the following expressions of the mean drag, lift and moment load per unit length acting on a model where q m is the mean wind velocity pressure. Analysing the temporal drag and lift force readings on the nominally non-vibrating model, reveals that the spectral power of the two signals have excess power at certain frequencies.For the lift, this corresponds to the oscillating force caused by vortex shedding, expressed by the Strouhal number for a fixed structure.All results from the static measurements are presented in Sect.4.1.The setup facilitates measurements of the wind-induced drag and lift force and moment.The setup is identical on the other side of the wind tunnel with a force transducer also placed on the upstream side of the model. Dynamic wind tunnel rig To measure the dynamic response, a lightweight model is mounted in the tunnel, constrained in the vertical direction, such that cross-flow deflections are restricted.Initial testing provided evidence that in-line VIV were initialised at wind velocities considerably below those where crossflow VIV are of relevance, thus the restriction in the vertical direction is used only to provide a stable vertical position of the model in the low-turbulent wind tunnel flow.Any resonant cross-flow forcing can easily be identified by a corresponding loss of tension in the top wires, and all measurements performed under such conditions have been discarded.In the horizontal direction, the model is positioned between springs to achieve a desired oscillation frequency.One spring connects to a force transducer via a string that allows the deflection to be read, by a calibration done prior to beginning the experiment.Note that the US wind tunnel is also equipped with lasers to enable direct measurements of the model deflections.A principal sketch of the measurement setup used for dynamic testing is presented in figure 3. The dynamic measurements allows for a characterisation of the model-specific dynamic behaviour, especially linked to vortex-induced in-line vibrations. The wind tunnel is set to run at a series of wind velocities whilst measuring the output of the force transducers.Between each measurement, an appropriate run-in time with constant wind velocity ensures that the oscillations have found a constant level.For each wind velocity, the standard deviations of the measured displacements are determined and by assuming a harmonic motion, the corresponding amplitudes are determined by multiplication by the square root of two. In the dynamic rig, oscillations are measured as functions of different mass ratios, mode shapes and thereby different mass-damping parameters are simulated.Furthermore, decay tests are performed to measure the natural frequency and the damping at zero wind velocities.These data are relevant in the subsequent analysis of the dynamic behaviour of the full-scale structures.Note that the oscil-02040-p.4The results from the dynamic response measurements are presented in Sect.4.2.Furthermore, pressure variations downstream to an in-line oscillating model have been captured using pitot tubes in the US wind tunnel for a single model configurations.This allows for the identification of the different fluid instability regions and therefore also the different physical forcing mechanisms responsible for inline oscillations.These results are presented in Sect.4.3.1. Model design and construction The measurements of the static load coefficients do not introduce mass restrictions on the model and readily available robust high-density materials can be in the model construction.In the dynamic wind tunnel rig, the section models are designed to be approximately neutrally buoyant.This may be established by using a low-mass hollow model construction, and replacing the air inside the model with a low-density gas such as helium.The model including suspension will hereby have a mass close to the mass of the displaced air. The construction of the low-density models required significant engineering and experimentation to meet the structural requirements.The cylindrical models are based on helium-filled cylindrical containers, obtaining additional structural stiffness in part by the internal pressure, but also by using an appropriate chosen shell made by helically wrapped paraffin paper or aluminised polyester.Circular cylindrical models with diameters of 250 mm, 500 mm and 1000 mm have been used for the present analysis. The rectangular model consists of a carbon fiber rod structure, covered with a light-weight paper-type fabric.Due to the sharp-edged geometry, the use of interior helium filled containers, which attain a cylindrical or spherical shape when inflated, were not found to be convenient in this case.A rectangular model with a cross-flow dimension of 200 mm has been used for the present analysis. The width of the wind tunnel models are 1.70 m or 2.40 m, using close to the full width of 1.75 m and 2.50 m for the Copenhagen and Vermont wind tunnels, respectively.The height of the wind tunnels are 1.50 m and 3.00 m, respectively.Resonance wind velocities will occur at Reynolds numbers of approximately 10 4 −10 6 .The possibility of covering both the subcritical and supercritical Reynolds number regime gives a good basis for predicting the behaviour in full scale. Blockage effects A flow occurring in the atmospheric boundary layer or open waters can only to some extent be recreated in an experimental testing facility, due to the finite extent of the generated stream.The spatial limitations may produce several boundary effects which could significantly influence the measurements, i.e. they do not resemble measurements performed in flow conditions without boundary effects.In that sense, it is usually important to consider boundary effects in order to provide a more accurate description of the aerodynamic or hydrodynamic behaviour.However, there is no single established method that takes account for a combination of flow contraction and wall constraints, which are often causing the main blockage effects.In this paper, the data is therefore presented without any correction for blockage, since applying a somewhat heuristic blockage correction expression would produce a data set with a significant level of uncertainty.Also, since most measurements are based on similar model scales, the influence from blockage effects are similar between the measurement series, and the results using different model configurations are therefore often directly comparable.For the 1000 mm diameter circular cylindrical model, the blockage is 33%. Boundaries causing spatial limitations in the cross-flow direction are known to reduce the amplitude of vortexinduced cross-flow vibrations.This effect is not expected to cause a similar reduction for in-line vibrations, since similar spatial limitations do not exist in the main flow direction. Main results The static load coefficients and dynamic properties of several cylindrical models are documented in the following section.These originate from a number of different wind tunnel flow situations, where the in-line dynamics are evaluated exclusively for low mass ratio models.In this setting, the physical flow characteristics causing the two in-line vortex-induced fluid instability regions have been measured and identified using recordings of spatial and temporal downstream fluid pressure variations.Additionally, the risk of galloping and torsional structural instability have been assessed by evaluating the static load coefficients for different inclination angles. Static load coefficients -nominally non-vibrating models Drag, lift and moment coefficients for a circular cylindrical model without end plates have been measured in a wind 02040-p.5The drag coefficient of a circular cylinder without freeend flow in the relevant Reynolds number regime is specified to be C D,0 = 1.2 [16].In the wind tunnel setup, freeend flow exists, but is somewhat limited due to the close proximity of the side walls.The measured drag coefficient is approximately 20% smaller than this value, which is in good agreement with the presence of a limited free-end flow.Note that the drag coefficient of this section model depends on the Reynolds number regime. The drag coefficient of a sharp-cornered rectangular cross section without free-end flow is stated in the Eurocode 1 specifications [17].For a ratio of the cross-flow and in-line dimension of 1:3 a drag coefficient of C D,0 = 1.36 is specified.The measured drag coefficient is approximately 7% smaller than the specified value, which again is in good agreement with the presence of a limited freeend flow.Note that the drag coefficient for a sharp edged section does not depend on the Reynolds number regime. The lift and moment coefficients are theoretically both zero due to the symmetry of the cross sections considered; thus, the measured responses of negligible magnitudes are entirely due to very small fluctuations in the wind flow, model imperfections and minor measurement uncertainties. By analysing the spectral power density function, the lift force signals are found to have distinct peaks caused by vortex shedding.For each cross section model, this procedure allows for the estimation of the Strouhal number for the fixed structure, as listed in table 1. Vortex-induced in-line vibrations in steady flow -vibrating models Vortex-induced in-line vibrations in steady flow have been measured on several experiemental configurations, corresponding to differences in model mass ratio, natural frequency, mode shape and Reynolds number regime. Circular cylindrical model -Mass ratio and natural frequency The dynamic in-line response has been measured on five configurations of a 250 mm diameter cylinder model in flow conditions corresponding to 1 × 10 4 ≤ Re ≤ 3 × 10 4 , see figure 5.For these experiments, the model was set to oscillate in a uniform mode shape and the model was equipped with end plates to limit the disturbance caused by end effects. A precise estimation of the mass-damping parameter is not possible, since the structural damping is not straightforward to determine.However, for these models the massdamping parameter scales with the mass ratio.Assuming a structural damping of δ s = 0.03, the mass-damping parameter is S c G ≈ 0.12 for λ m = 1.1. The onset of the first fluid instability region is located at U r ≈ 1.1 for all five configurations, while the maximum amplitude, being approximately 11% of the cylinder diameter for the mass ratios λ m = 1.1 and λ m = 2.0, is only slightly reduced for the larger mass ratio of λ m = 4.2.The shift between the first and second fluid instability seems to depend on the mass ratio as well.For larger mass ratios, corresponding to a larger stability parameter, the response drops at slightly smaller reduced wind velocities, which is in accordance with the DNV response model. A change in the natural frequency of the model does not seem to influence the characteristics of the response in the first fluid instability region, justifying the use of the reduced fluid velocity as a descriptive parameter.For all configurations, the maximum response occurs in the range 2 < U r < 3, which fits well with the previously presented in-line response models.Also, by increasing the mass ratio, the maximum response seems to occur at slightly smaller reduced wind velocities. The second fluid instability responses for λ m = 2.0 and λ m = 4.2 are similar, being slightly smaller in maximum amplitude for λ m = 4.2.For the models with a mass ratio of λ m = 1.1, the model begins to buckle and deform at reduced wind velocities above U r ≈ 3. When this happens, the tests are stopped because the mode shapes are no longer uniform.As a result, the second fluid instability region is not fully characterised by the models with the lowest mass ratio. Circular cylindrical model -Mode shape For a cylindrical 250 mm diameter model, the influence of the mode shape on the in-line vibrations has been investigated for two model configurations, see figure 6.In these measurements, the flow conditions correspond to 1 × 10 4 ≤ Re ≤ 3 × 10 4 . The cylindrical model is oscillating with an amplitude of around 15% of the cylinder diameter for a free-end mode shape, somewhat larger than the amplitude found for models oscillating in a uniform mode shape, see figure 5.The difference in the amplitude between the mode shapes can likely be explained by a different scaling of positive and negative fluid dynamic damping effects along the structure.This phenomenon is actually well-known for crossflow VIV [18]. Attaching additional mass to the free end of the cylinder increases the mass-damping parameter and decreases the maximum amplitude of the oscillations to around 10% of the cylinder diameter.Also, note that the second fluid instability region (U r ≈ 3) apparently does not give a distinct response in this mass-damping range for a free-end mode shape.The fact that the mass is added to the free end of the model implies that the mass-damping is increased relatively more than the mass ratio listed in the legend on figure 6.Assuming a structural damping of δ s = 0.03, the mass-damping parameters are S c G ≈ 0.17 and S c G ≈ 0.30 for the two presented model configurations. Circular cylindrical model -Reynolds number Dynamic response characteristics in flow conditions corresponding to higher Reynolds numbers, can be obtained in a wind tunnel by increasing the model dimensions or by increasing the wind velocity.Both approaches imply considerable challenges for the model structure.A larger model dimension naturally implies increased sag and bending moments, and the present low mass ratio requirements would often result in a model which is less likely to withstand higher velocities without deformations or similar undesirable effects.Nevertheless, response measurements in even higher Reynolds number regimes have been obtained using a large inflatable scale model, having a diameter of 1000 mm.The responses of four such configurations in high Reynolds number regimes are presented in figure 7.For these four configurations, the mass ratio is held approximately constant and the model was equipped with end plates to limit the disturbance caused by end effects.For Re = 3 × 10 5 , the model is seen to have an in-line response amplitude of up to 9% of the cylinder diameter and two visible fluid instability regions.A single fluid instability region with a much reduced response is observed for Re = 4 × 10 5 and Re = 6 × 10 5 , and for the latter flow regime, non-uniform oscillations were observed for reduced wind velocities larger than U r ≈ 2.2.For the flow regime corresponding to Re = 1 × 10 6 , no clear indication of in-line fluid instabilities is observed and the response is just slowly increasing with wind velocity.The only exception is a single peak at U r ≈ 2.3 corresponding to an amplitude of around 1% of the cylinder diameter. The maximum response, as well as the velocity interval in which oscillations occur, are seen to be significantly reduced for increasing Reynolds numbers.This general trend is most likely not entirely caused by the change in the flow regime characteristics with the increasing Reynolds num- ber, but is also due to related side effects of the increased wind velocity, such as an increased deformation force of the wind tunnel model, and an associated increased energy dissipation in the complete oscillating system.However, it is reasonable to assume that no significant amplification of the overall response would happen as a result of the change in the flow regime for the Reynolds numbers considered.Note also that, as stated previously, wind tunnel blockage is not accounted for in the presented data; thus the reduced wind velocities, and thereby also the Reynold numbers, are slightly underestimated. The physical mechanisms responsible for in-line vibrations is a certain combination of forcing due to vortexshedding and a related change in the aerodynamic damping in the direction of the flow.Exactly the same physical mechanisms working in the cross-flow direction are responsible for cross-flow VIV.In the latter case, a change in aerodynamic damping is responsible for the transition range at which the smaller deflections governed by lift forces changes to larger deflections governed by motion-induced forces.The location of the transition is described by the aerodynamic damping factor for cross-flow amplitudes [11].The presence of large cross-flow deflections governed by motion-induced forces therefore depends on the aerodynamic damping factor which in turn depends on Reynolds number.Since the physical mechanisms responsible for vibrations are similar in the in-line and cross-flow directions, it is expected that the maximum amplitudes of in-line oscillations show a similar dependence on the aerodynamic damping through the Reynolds number.For cross-flow oscillations, the aerodynamic damping factor attains its minimum at Re ≈ 6 × 10 5 according to Eurocode 1 specifications [17], which implies that in this range only small deflections governed by lift forces are to be expected.Apparently, a similar tendency is true for in-line oscillations. Rectangular model In contrast to the previously tested cylindrical models, the geometry of the rectangular cross section requires that the model is constructed using a lightweight skeleton structure, since the sharp-edged geometry cannot be obtained by pressurised balloons alone.This implies that a mass ratio very close to one is very difficult, if not impossible, to obtain with the model dimension of D = 200 mm related to the present wind tunnel testing.Therefore, the smallest mass ratios considered for the rectangular section model are larger than those related to the cylindrical models.A plot of the measured normalised response amplitudes as a function of the reduced wind velocity is presented in figure 8.The flow conditions correspond to Re ≈ 2 × 10 4 .Again, for these models the mass-damping parameter scales with the mass ratio.Assuming a structural damping of δ s = 0.03, the mass-damping parameter is S c G ≈ 0.20 for λ m = 2.2. The presented results show that the rectangular cross section apparently is much less prone to in-line vortexinduced oscillations than the cylindrical cross section.The maximum amplitude is below 2% of the cross-flow dimension, and there is actually a slight increase in amplitude with increasing mass ratio for the four presented model configurations.Note that only a single fluid instability region is visible, located at approximately 3 ≤ U r ≤ 4. The aerodynamic damping for the rectangular cross section is larger than for circular cross sections, simply due to the larger drag coefficient and the fact that a higher reduced wind velocity is necessary to initialise the oscillations.This effect is also relatively more pronounced at small mass ratios at similar reduced wind velocities.It is deemed that this effect is responsible for the limited inline response.The decrease in aerodynamic damping with increasing model mass could also account for the slight increase in maximum response for larger mass ratios. The initialisation of the in-line vortex-induced forcing is related to the Strouhal number for the fixed structure.For the rectangular model and the cylindrical model, the ratio of the measured Strouhal numbers for the fixed structures are approximately 3.6, see table 1, implying an onset of the in-line forcing frequency relative to the structural eigenfrequency at three to four times higher wind velocities for the rectangular model.This is in agreement with the onset of the in-line vortex-induced forcing close to the structural eigenfrequency at U r ≈ 3, which is three times larger than the onset found for the cylindrical models, see Sect.4.2.1. As seen on figure 8, the in-line vibrational behaviour was investigated for reduced wind velocities up to approximately U r = 5.For a similar 1:3 section model, torsional vortex-induced oscillations in air have been reported to have an onset near U r = 6.5 [19].This is approximately twice the reduced velocity at which the in-line oscillations were seen to attain the largest amplitudes, which suggests that the in-line oscillations measured on the rectangular section model are caused by a normal asymmetric vortex shedding pattern. Additional investigations Several experiments have been performed in addition to the investigations of model-specific static and dynamic load 02040-p.8characteristics.In this section some supplementary findings are listed, which confirms or enlightens the previously documented results linking in-line vortex-induced vibrations to basic flow characteristics, and the fluid-dynamical and structural parameters. Identification of fluid instability regions As mentioned in Sect.2, the mechanism responsible for the first fluid instability causing in-line oscillations is regular, alternating vortex shedding combined with a secondary, symmetric vortex shedding pattern which occurs as a result of in-line structural motion relative to the fluid.In the second fluid instability region, the forcing is entirely due to regular vortex shedding from alternating sides of the cylinder.Therefore, the two types of in-line responses are caused by downstream pressure variations oscillating with three and two times the Strouhal frequency, respectively.Measurements of the spatial and temporal downstream pressure variations could therefore be used to identify the forcing mechanism; thus, identifying the corresponding fluid instability regions. In the US wind tunnel, the downstream pressure variations have been captured using two pitot tubes located approximately 750 mm behind a 500 mm diameter cylindrical model oscillating in the in-line direction, having an eigenfrequency of n s ≈ 2.7 Hz.The pitot tubes are vertically aligned with the model top and bottom, see figure 9.The probes are 8 mm diameter NPL type pitot-static tubes with ellipsoidal head complying with ISO 3966.The dynamic pressure sensors have a response time of less than 10 ms, and the setup can easily capture the generated pressure fluctuations.Hot-wire anemometers which are often used to capture high frequency flow fluctuations are therefore not required in the present case. When model oscillations couple with vortex shedding, the eigenfrequency of the model becomes visible in the power spectra of the measured time series of pressure variations at the two downstream pitot tubes.Figures 10 and 11 illustrate such spectral power densities at U r = 2.16 and U r = 3.26, respectively. In the first fluid instability region, the vortex shedding pattern is complex as it is a mix of both alternating and -For relatively low reduced wind velocities (U r ≈ 1), the cross-correlation of the two signals are dominated by correlation peaks of a period T , here defined as the inverse of the model eigenfrequency of approx.2.7 Hz.This indicates that symmetric vortices, which are highly correlated in the wake, are shed with the motion of the model.Concurrently, correlation at zero time lag exists, but maximum correlation is not achieved here due to the occasional alternating vortex pair.By lowering the wind velocity, the vortex shedding becomes increasingly dominated by symmetric pairs, and the model eigenfrequency becomes visible while half the model eigenfrequency disappears from the spectra.Depending of the model natural frequency, the regular asymmetric vortex shedding on a fixed cylinder corresponding to a Strouhal number of S t = 0.18 may be-02040-p.9The second fluid instability region is due to vortex shedding from alternating sides as is familiar on fixed cylinders.This region is characterised by the appearance of peaks in the spectrum at the model eigenfrequency and at half the model eigenfrequency, see figure 11.This suggests that vortices are shed once per model oscillation, as the spectral footprint consists of the frequency by which each single vortex is shed, and also the frequency corresponding to the shedding of a full period of a pair of vortices.If the pitot pressures are subtracted from each other the resulting spectrum reveals no significant contribution at the frequency corresponding to the model eigenfrequency.If the signals are added a clear peak at the model eigenfrequency occurs in the corresponding spectrum.The cross-correlation on figure 13 of this shedding type is approximately sinusoidal with periodic peaks of constant value indicating a long time correlation.The period is 2T , twice the period of the model oscillations, with the first peak correlation at T consolidating the alternating nature of the shedding. Between the two fluid instability regions, the crosscorrelation of the two pitot pressures remains sinusoidal with a period of 2T , but the level of correlation decays quickly.The model eigenfrequency leaves very little footprint in the frequency spectra while half the eigenfrequency remains visible. Measurements of the spatial and temporal downstream pressure variations on an in-line oscillating cylindrical model Structural instability phenomena The angular dependence of the static load coefficients may be used not only to predict fluid-induced forces, but also to investigate if the cross section is susceptible to aeroelastic or hydroelastic structural instability phenomena such as galloping and flutter.For this, the Scruton number, of the more general non-dimensional mass-damping parameter, may be used in predictions of cross-flow galloping, assuming a driving force proportional to the in-line dimension.The rotational structural response can be modelled by a one degree of freedom system with a torsional forcing determined by the fluid-induced moment load.It is a standard approach to consider a linearisation argument to indicate that two types of structural instabilities can occur; one being torsional galloping and one being static divergence [20]. Due to the geometrical symmetry, galloping and torsional structural stability are not relevant for the circular cross section model.For the rectangular cross section, a negative slope of the aerodynamic lift was measured, implying that cross-flow galloping vibrations might exist for small inclinations of the flow due to unstable aerodynamic loading conditions.The aerodynamic moment has a negative slope for small inclinations.This implies that the structure is stable towards flow-induced excessive twist.However, a negative slope is known to increase the risk of torsional galloping, due to negative aerodynamic damping effects. The slope of the moment coefficients for rectangular sections rotating about the geometric center have previously been reported for 1:2 and 1:4 cross sections in the literature, being -0.64 and -18, respectively, for α measured in radians, but using the cross-flow dimension as the characteristic dimension in the definition of the moment load 02040-p.10 per unit length [21].For the analysed 1:3 cross section, the measurements suggest dC M dα \| α=0 ≈ −1.1, corresponding to a value of -9.9 using the cross-flow dimension as the characteristic dimension. Discussion For free-span circular cross-section pipelines submerged in water, the DNV in-line response model gives an empirical description in current dominated full-scale conditions.The presented results indicate that the maximum in-line response of a close to neutrally buoyant circular cylindrical model oscillating in a uniform mode shape in air is welldescribed by the DNV model, both in terms of maximum response and the reduced velocities at which in-line VIV is relevant.The dependence on the stability parameter expressed in the DNV model cannot be confirmed based on the conducted measurements, since a precise estimation of the total damping, or in particular the aerodynamic damping, is not straightforward.However, the overall tendency of a reduction in the maximum response occurring at a lower reduced flow velocity when increasing the stability parameter, or in the present case increasing the mass ratio, is indeed observed in the measured responses. In the performed dynamic testing, models with a mass ratio in the range of, say, 1.0 -2.0 of the surrounding fluid, apparently produce a comparable maximum response for a uniform mode shape.In other words, the related Scruton curve has a small slope near zero.This observation can be used as a guidance for future testing, since it implies that maximum responses might be correctly estimated using models which are only close to being neutrally buoyant. Based on additional measurements in flow conditions corresponding to 10 5 ≤ Re ≤ 10 6 , it is reasonable to assume that no significant amplification of the overall response would happen as a result of the change in the flow regime for the Reynolds numbers considered.A compensation for the slightly lower Reynolds number in the wind tunnel environment compared to full scale conditions, is therefore likely not necessary to be included in the safety requirements.Nevertheless, subsequent wind tunnel measurements at even higher Reynolds numbers could potentially provide additional insight into the importance of this topic. Measuring the spatial and temporal downstream pressure variations on an in-line oscillating cylindrical model allowed for the identification of the first-and second fluid instability region, thus identifying the forcing mechanism causing in-line structural oscillations in steady fluid flow.The investigation verified that the physical phenomena responsible for vortex-induced vibrations in air are similar as those reported in water [10]. The rectangular cross section is apparently much less sensitive to in-line VIV than the cylindrical cross sections, likely caused by a larger fluid-dynamic damping.The onset of in-line VIV happens at larger reduced velocities, probably due to the smaller Strouhal number for the structure. Conclusion The presented results provide a preliminary insight into a fundamental methodology for the transfer of certain aerodynamic actions to equivalent hydrodynamic actions, related to both static response characteristics and in-line vibrations caused by vortex shedding.A consistency between the results obtained in the wind testing facilities at Svend Ole Hansen ApS and SOH Wind Engineering LLC, and results published by others were observed; thus, verifying the understanding of the general dynamic properties of low mass ratio structures and the type of physical phenomena which are of relevance for the in-line structural oscillations in steady fluid flow.This consistency was found on both the qualitative and quantitative levels, giving a response of similar magnitude for experimental configurations corresponding to comparable basic flow characteristics and model parameters.In this sense, the wind tunnel testing has turned out to be a very economical and powerful tool for determining and verifying hydrodynamic data. It is of fundamental interest that wind tunnel experiments have shown that in-line VIV caused by steady winds on model structures with a density comparable to air, and the response prescribed on neutrally buoyant structures in water are of very similar nature.As this is a very wideranging field, the presented work cover basic principles and demonstrates prevailing tendencies.The conversion of model-scale results to full-scale predictions should therefore be assessed accordingly and the interpretation of the presented model-scale data should focus on the fact that the basic physical mechanism generating the in-line vibrations in question are equivalent in model scale and full scale, and in air and water.Nonetheless, the presented model tests have provided clear indications of the static and dynamic properties of specific neutrally buoyant structures and should form a basis for future systematic investigations providing in-depth descriptions on the flow-induced structural behavior. DOI: 10 .1051/ C Owned by the authors, published by EDP Sciences, 3. 1 Testing procedure All conducted model tests are twofold: Tests in the static rig provide data to estimate the drag coefficient C D , and the lift and moment coefficients C L , C M , together with their angular derivatives, which enable a description of fluidinduced forces and the aeroelastic or hydroelastic characteristics of the structure.Spectral analyses of the drag and lift signals provide an additional insight into the oscillating flow mechanisms for the fixed structure.This is described by the Strouhal number for cross-flow vortexinduced oscillations.Tests in the dynamic rig provide data used to predict the risk of flow-induced vibrations, especially vortex-induced in-line vibrations of the structure.Different degrees of spring stiffness are tested, giving resonance wind velocities in the entire wind velocity range of the wind tunnel. Figure 2 : Figure 2: Principal sketch of the measurement setup used for static testing, looking from the side of the wind tunnel.The setup facilitates measurements of the wind-induced drag and lift force and moment.The setup is identical on the other side of the wind tunnel with a force transducer also placed on the upstream side of the model. Figure 3 : Figure 3: Principal sketch of the measurement setup used for dynamic testing, looking from the side of the wind tunnel.The setup facilitates measurements of an oscillating wind-induced in-line force.The US wind tunnel is also equipped with lasers to enable direct measurements of the model deflections. Figure 4 : Figure 4: Sign convention of static forces and moments, here shown for the circular cross section.The flow is from left to right. Table 1 : Calculated static load coefficients of the two section models for a zero degree inclination of the approaching flow in the listed Reynolds number regime.-0.07 -0.01 0.18 4 × 10 4 Rectangular 1.27 0.01 -0.02 0.05 1 × 10 5 environment corresponding to Re ≈ 4 × 10 4 in the Copenhagen wind tunnel.In the wind tunnel in Vermont, similar measurements have been performed on the rectangular 1:3 cross section model in a flow corresponding to Re ≈ 1× 10 5 .The results are listed in table 1, where the static load coefficients are presented according to the definitions stated in Sect.3.2.The diameter of the circular cylindrical model is 250 mm and the cross-flow and in-line dimensions are 500 mm and 1500 mm, respectively, for the rectangular model.The sign convention of static forces and moments is presented in figure 4. Figure 5 : Figure 5: Measured response of five model cylinders configurations in steady air as function of the reduced wind velocity for different model configurations oscillating in a uniform mode shape.The flow conditions correspond to 1 × 10 4 ≤ Re ≤ 3 × 10 4 . Figure 6 : Figure 6: Measured in-line response of a cylindrical model as function of the reduced wind velocity for two different configurations oscillating in a free-end mode shape.The maximum response depends on the mode shape (compare to figure 5) and is reduced when a large mass is added.The flow conditions correspond to 2 × 10 4 ≤ Re ≤ 4 × 10 4 . Figure 7 : Figure 7: Measured in-line response of a 1000 mm diameter model cylinder in steady air as a function of the reduced wind velocity for different Reynold number flow regimes. Figure 8 : Figure 8: Reduced amplitudes of the rectangular section model for different mass ratios.The largest amplitude is below 2% of the cross-flow dimension.For all model configurations, a clear in-line model response is only seen for wind velocities corresponding to approximately 3 ≤ U r ≤ 4. Figure 9 : Figure 9: Sketch of the model setup capturing the downstream pressure variations due to vortex shedding.Two downstream pitot tubes are used to capture the spatial information needed to distinguish between asymmetric and symmetric vortex shedding. Figure 10 : Figure 10: Spectral power density of the top and bottom pitot tube pressures at U r = 2.16, in the first fluid instability region, indicating a clear peak at the models natural frequency n s . Figure 11 : Figure 11: Spectral power density of the top and bottom pitot tube pressures at U r = 3.26, in the second fluid instability region, indicating clear peaks at both the models natural frequency n s and at n s /2. Figure 12 : Figure 12: The cross-correlation between top and bottom pitot pressures at U r = 2.16, indicating a mix of symmetric and regular, asymmetric vortex shedding. Figure 13 : Figure 13: The cross-correlation between top and bottom pitot pressures at U r = 3.26, indicating regular, asymmetric vortex shedding.	v3-fos-license
2018-12-11T21:52:59.595Z	2016-08-25T00:00:00.000	55742649	{ "extfieldsofstudy": [ "Engineering" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "http://downloads.hindawi.com/journals/ijap/2016/7439403.pdf", "pdf_hash": "20f7fe0746e38a2b5861ca059ae4bb148ee93151", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45468", "s2fieldsofstudy": [ "Engineering" ], "sha1": "87249b88b6b95004bb71f09c8381c01b72b9288f", "year": 2016 }	pes2o/s2orc	On-Body Characterization of Planar Differential Antennas for Multiple, Wide, and Narrow Bands This paper reports the results of the on-body experimental tests of a set of four planar differential antennas, originated by design variations of radiating elements with the same shape and characterized by the potential for covering wide and narrow bands. All the antenna designs have been implemented on low-cost FR4 substrate and characterized experimentally through on-body measurements. The results show the impact of the proximity to the human body on antenna performance and the opportunities in terms of potential coverage of wide and narrow bands for future ad hoc designs and implementations through wearable substrates targeting on-body and off-body communication and sensing applications. Introduction UWB radar for contactless detection of respiratory rate of human beings [1][2][3][4] is an active research area requiring innovative transceiver and antenna developments [5,6]. A novel planar differential antenna was proposed recently by our research group [7]. This antenna was designed to meet the design constrains, electromagnetic performance, and physical integration, required by the ultrawideband (UWB) pulse radar sensor operating approximately in the frequency band from 3 to 5 GHz [2][3][4]. The proximity between the radiating elements of the proposed differential antenna has allowed achieving a significant bandwidth enhancement. In addition to original differential antenna development, we also investigated some design variations with the objective of showing how these reflect on the antenna performance, showing that the antennas originating from radiating elements with the same shape can be potentially adopted through an agile design strategy to cover wide and narrow bands of interest for a number of wireless applications, communication, and sensing, as a consequence of variations in size and orientation of the radiating elements with respect to the original planar differential antenna [8]. Numerous studies were carried out about the effects of human body on performance of planar antennas. Concerning nondifferential antennas, many contributions reported the effects on S-parameters and gain patterns, like in [9,10]. Considering differentially fed antennas, in [11], the in-body near-field radiation pattern of an on-body antenna was characterized. In other studies, implantable antennas were considered, like in [12,13], where the in-body radiation patterns and S-parameters were investigated. However, to the authors' knowledge, the experimental characterization of differential antennas on human body, in terms of off-body far-field radiation patterns, which is proposed in the present paper, has received no attention yet in existing literature. Fully differential transceivers are less prone to commonmode noise and interferers; thereby differential antennas could play a significant role in future networks and communications, especially considering the massive amount of devices envisaged for the Internet of Things era [14,15] that 2 International Journal of Antennas and Propagation are supposed to operate in regions of frequency spectrum already crowded by existing applications and services. In this paper, we report for the first time the results of the on-body experimental characterization of the variations of planar differential antenna design described in [8], resulting from measurements on a real human subject, which have not been reported in any of our previous publications. For reasons of space, the details and results of previous publication [8] will be not repeated here. Thereby, for the insights about the design, inherent properties, and free-space (in air) performances of the antennas, we invite the reader to inspect our previous work reported in [8]. In particular, in this work, here, we will show for the first time how the proximity to the human body affects the antenna performance and that the antennas, all originating from radiating elements with the same shape, can be potentially adopted in future ad hoc designs on flexible substrates or textile fabrics to cover multiple, wide, and narrow bands of interest for a number of wearable applications for both communication and sensing [16]. The paper is organized as follows. Section 2 reports the description of the antenna design variations and summarizes, for reasons of self-consistency, the main results obtained from the previous free-space tests, which provide the term of reference for design and performance. Section 3 reports the measurement results obtained from the tests on a real human body. Finally, in Section 4, the conclusions are drawn. Planar Differential Antennas: Designs and Tests Two different sets of planar differential antennas were designed on a low-cost FR4 substrate ( = 4.4, dielectric thickness = 1.6 mm, copper thickness = 35 m, and loss tangent tan = 0.02) for a complete characterization as stand-alone devices. The layout drawing of the first set of antennas, namely, folded planar differential antennas, is shown in Figure 1. The layout drawing of the second set of antennas, being antipodal planar differential antennas, is shown in Figure 2. At first glance, we observe that each radiating element of the antenna consists of a semicircle and a triangle that provides a smooth transition towards the input pins [7]. In particular, in the folded planar differential antenna, the two radiating elements are folded one aside the other. In particular, the design is characterized by the aperture angle of triangle and rotation angle between the symmetry axes of the two radiating elements. The main features of this design approach are that it allows a compact design of both the transmitter and receiver antennas on the same board of the implemented radar sensor based on a system-on-a-chip radar transceiver, still maintaining good performance [2][3][4]. In the antipodal planar differential antenna, the two radiating elements are rotated by 180 ∘ one with respect to the others, so resulting in an antipodal position. In the following subsections, we will summarize the results obtained for the experimental open-space tests. In particular, we report the results of two antenna designs for each set, folded and antipodal, as follows: (A) Folded planar differential antenna with = 45 ∘ and = 3 cm, as shown in Figure 1. All the antenna designs were carried out by means of the electromagnetic (EM) simulator Momentum by Keysight Technologies5, and characterized experimentally in anechoic chamber. Folded Radiating Elements: = 45 ∘ and = 3 cm. The physical implementation is shown in Figure 3. Distance between the two sides of the antenna is equal to 1 mm. Diameter is equal to 3 cm in order to resonate at the frequency of interest (i.e., roughly 3 GHz). The aperture angle of the triangle is = 45 ∘ . The rotation angle of the two radiating elements is = 45 ∘ . Two microstrip feeding lines were added in order to allow the connection of the antenna to a Vector Network Analyzer (VNA) by means of 2.92 mm connectors (horizontal) and carry out the experimental tests. Width of the microstrip line feeding paths ( ) is equal to 3 mm in order to exhibit a characteristic impedance of = 50 Ω. The distance between the two inputs of the antenna is equal to in = 1 cm to allow the placement of two adjacent connectors, as shown in Figure 1. The characteristic sizes of the UWB antenna are reported in Table 1. The simulated and measured 11 parameter and VSWR are reported in Figure 4. The measured 11 exhibits a magnitude lower than −10 dB almost in the whole band of interest, roughly from 3 to 5 GHz [17,18]. Figure 4 reports also the results for the voltage wave standing ratio (VSWR), which is lower than two almost in the whole band of interest. Folded Radiating Elements: = 45 ∘ and = 4 cm. Simulated and measured 11 parameter as a function of frequency of the antenna with folded radiating elements ( = 45 ∘ with = 4 cm) are shown in Figure 5. All the other design parameters are unchanged with respect to the folded antenna with = 3 cm. The measurements show 11 lower than approximately −7.5 dB from 2.4 to 5 GHz. In particular, 11 < −10 dB roughly from 2.3 to 2.7 GHz and from 3.3 to 5.1 GHz. Figure 5 reports also the measurement results for the VSWR which is lower than about 2 over the same bands reported above, for the measured antenna. As expected, the results confirm that the bandwidth, for this design variation with = 4 cm, is extended roughly for about 0.5 GHz towards the lower frequencies with respect to the original design with = 3 cm. This result shows that this design variation has the potential to be compatible with multiband operations for both industrial scientific medical (ISM) narrow band at 2.4 GHz [19,20] and the lower portion of the UWB band from about 3 to 5 GHz [3]. Antipodal Radiating Elements: = 180 ∘ and = 3 cm. The physical implementation of the antipodal planar differential antenna is shown in Figure 6. The design parameters are summarized in Table 2. 11 parameter and VSWR resulting from simulations and measurements are shown in Figure 7. \| 11 \| is lower than −10 dB in the frequency band from about 1.1 to 1.44 GHz. The antenna is thus compatible with operation All the other design parameters are unchanged with respect to those summarized for the previous case in Table 2. 11 parameter and VSWR resulting from simulations and measurements are shown in Figure 8. We get \| 11 \| < −10 dB in the frequency band from 0.8 to 1.06 GHz. It is worth observing how the increase of diameter ( = 4 cm), with respect to the case with = 3 cm, enables the potential coverage of the ISM band at 868 MHz [22,23]. The discrepancies between measured and simulated 11 , for the discussed antenna variations, are the results of a number of factors influencing the impedance matching of the antenna prototype during measurements, which are not taken into account in the simulation model. Namely, the presence of SMA connectors and solders in the real prototype, VSWR VSWR sim. not accounting for in the simulation, can result in deviation of input impedances, with a consequent variation in measured S-parameters with respect to the simulated ones. Measurement cables, although accurately shielded during measurements, can also be partly responsible for such deviations, it is very difficult to completely eliminate coupling between them and the radiating elements. However, the discrepancies can be still considered more than acceptable. On-Body Characterization All four variations of planar differential antennas summarized in the previous section were characterized through onbody experimental tests, in order to investigate the impact of the proximity to the human body on the antenna performance and explore their potentialities for future implementations through ad hoc designs on flexible substrates or textile fabrics [14] and exploitation in wearable applications [21]. The measurements were carried out in anechoic chamber, where the antennas under test were placed on the chest area of a real human subject, with fixed distance of 15 mm between the printed circuit board of the antenna and the body surface, achieved by means of a dielectric spacer placed between them. Each antenna was mounted on the spacer, and all together were placed on the chest of the human body. The radiating element of the antenna is on the opposite side of the interface with the spacer. -axis is oriented away from and perpendicularly to the body surface, and -axis is pointing upwards, as shown in Figures 9 and 10(a). The main physical and dielectric characteristics parameters of the antenna and spacer fabric are reported in Table 3. International Journal of Antennas and Propagation 5 It is worth remarking that the purpose of the experimental on-body characterization was to test a realistic situation, that is, with the antenna placed on a real human body, where slight distance variations occur due to small body movements due to respiration and spacer compression. The scheme in Figure 9 is therefore a simple illustrative representation of the antenna placement on the body, not displaying the abovementioned realistic effects. The reflection coefficient as a function of frequency and the differential gain pattern on azimuth plane ( ), ( ) were determined for all four antennas. The gain pattern was derived by measuring, by means of an Agilent PNAX network analyzer, transmission coefficient 1 between the single output of a standard-gain horn and the differential terminals of the antenna under test, for different azimuth ( ) orientations, then rescaling the measured value by calibrating path loss, gain of the transmitting horn, and mismatch factors at both ends, in order to obtain the gain pattern. For all the measurements, particular care was taken to avoid as much possible interference and scattering as possible from measurement cables which, for this purpose, were covered by radiofrequency absorbing material. In addition, ferrite chokes were placed around the measurement cables in order to furtherly reduce scattering and undesired coupling with the antenna. Folded Radiating Elements: = 45 ∘ and = 3 cm. The measured reflection coefficient and azimuth gain pattern at = 3 GHz, for the smaller antenna with folded radiating elements, on human body, are shown in Figures 11 and 12, respectively. One can see how the reflection coefficient, compared to the open-space measurements, is affected by small variations due to the proximity with human body. However, the reflection coefficient remains more than acceptable, being 11 < −10 dB in the frequency band from 2.85 to 4.53 GHz. The measured differential gain pattern at = 3 GHz (i.e., about the lowest resonance frequency), shown in Figure 12, exhibits a main lobe in the half space oriented away from the human body (i.e., −90 ∘ < < +90 ∘ ). The maximum gain for = 0 ∘ is about 4.14 dBi, which is increased by an amount of about 2.7 dB with respect to the open-space case, for which the maximum gain was about 1.43 dBi. It is worth noting that, in this case, the human body is acting as a reflector, producing an increase of the maximum gain and directivity. The backscattering gain is largely attenuated by the human body which absorbs most of the power radiated by the antenna. Thereby, this antenna is also capable of operating when worn by a human body, for UWB off-body wireless communications in the lower portion of the UWB frequency spectrum (e.g., 3-5 GHz). Folded Radiating Elements: = 45 ∘ and = 4cm. Figures 13 and 14 show the measured differential reflection coefficient and the azimuth gain pattern of the larger version of the antenna with folded elements (i.e., with = 4 cm). Similar to the folded antenna with = 3 cm, the reflection coefficient and consequently the VSWR undergo a slight variation due to the presence of the human body. In particular, we have 11 < −8.3 dB and VSWR < 2.1 in the lower portion of the UWB band from 3 to 5 GHz, which indicates more than acceptable performance of this antenna when worn on human body. Moreover, 11 < −10 dB in the bands from 2.3 to 2.7 GHz and from 3.3 to 5 GHz. Concerning the measured gain pattern of the on-body antenna shown in Figure 14, at = 2.5 GHz, this has a main lobe directed away from the body with a maximum of about 1.5 dBi in broadside direction. The maximum gain is increased of about 3.5 dB when compared to open-space measurements, indicating reflection from the human body. This antenna variation, similar to the antenna with larger folded elements, is potentially suitable for on-body operation in UWB off-body communications in the lower UWB spectrum and 2.4 GHz ISM band. Antipodal Radiating Elements: = 180 ∘ and = 3 cm. The on-body measured reflection coefficient and gain pattern at = 1.2 GHz, compared to the open-space results, for the smaller antipodal antenna are shown in Figures 15 and 16, respectively. The reflection coefficient is affected by slight variations due to the proximity to the human body. In particular, −10 dB 11 band becomes slightly narrower, extending in the frequency band from 1.08 to 1.37 GHz, and the resonance frequency decreases by about 10 MHz to value of = 1.21 GHz. performance with respect to the open-space measurements. In spite of all, the body-worn antenna is still capable of providing a good coverage of the 868 MHz ISM band. As for the gain pattern in plane, Figure 18 shows that the gain pattern at = 868 MHz of the antenna worn on body is affected by a reduction of the maximum gain, which amounts to about −1 dBi, 2 dB lower than the value of about 1 dBi in open space. Moreover, the backscattering of the gain pattern is, similar to all other antennas, severely attenuated by the human body. This antenna variation, in onbody operation, still exhibits the characteristic performance suitable for its potential usage in the Ultra High Frequency (UHF) band at 915 MHz for Radiofrequency Identification (RFID). Conclusions Four different variations of planar differential antennas, for wide and narrow band applications, differing in dimensions and orientation of the radiating elements, were experimentally characterized when worn on the chest of a human body. Measurements were performed of the differential reflection coefficient and horizontal gain pattern of the antenna, with a distance of 15 mm between antenna and human body. Results were compared with those obtained with the antenna in open space and the following observations could be made. Concerning the impact on the performance, the presence of the human body has a clear effect on reflection coefficient and gain patterns. In particular, as for the gain pattern, the proximity to the human body produces, for all antenna variations, a severe attenuation of the backscattering portion of the gain pattern, towards the back side of the wearer, due to absorption of the radiated power by the human body. Moreover, for 3 of the 4 antennas, the maximum gain measured on-body for the lowest resonance frequency, is affected by a small increase with respect to the open-space case, indicating that the human body acts as a reflector. For the larger ( = 4 cm) version of the antenna with antipodal elements only, the maximum gain results in a reduction of about 2 dB with respect to the open-space measurements. Concerning reflection coefficient, the presence of the human body results in small variations of the characteristics, with the most noticeable effect being a slight lowering of the resonance frequencies. The reflection coefficient performance remains more than acceptable for operation at the frequencies of interest. In conclusion, the performance of all antennas remains more than acceptable when they are worn at 15 mm of distance from the chest area of a human body. This proves the potential for future wearable implementations of all considered prototypes, in particular for short range off-body communications, with applications such as monitoring of vital signs or body area network personal communications. Moreover, the four antenna variations allow the coverage of several frequency bands of interests, such as the lower portion of the UWB from 3 to 5 GHz, the 868 MHz and 2.45 GHz ISM bands, the 1.2 GHz band (lower L-band) for the Global Positioning and Navigation Satellite System (GNSS), and the UHF band at 915 MHz for Radiofrequency Identification (RFID).	v3-fos-license
2023-10-14T15:44:29.797Z	2023-06-10T00:00:00.000	264095944	{ "extfieldsofstudy": [], "oa_license": "CCBYSA", "oa_status": "CLOSED", "oa_url": null, "pdf_hash": "3e409395b0862e90ea81728e988a5b831fc986b4", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45470", "s2fieldsofstudy": [], "sha1": "01d4fe98a405d255f30e313d0bcb5ef8e6b0af42", "year": 2023 }	pes2o/s2orc	Technical and tactical evolution of the offensive team sequences in LaLiga between 2008 and 2021. Is Spanish football now a more associative game? The aim of this investigation was to study the technical and tactical evolution of the offensive team sequences in the Spanish football teams from 2008/09 to 2020/21. A comparative analysis including twelve variables related to the development of offensive sequences in 4940 matches was performed from 2008/09 to 2020/21 seasons of the Spanish professional football league (LaLiga). All match observations were recorded using a validated video tracking system. Multilevel linear mixed models were used to examine the differences across seasons, considering the effects of contextual variables. The number of passes per sequence (2.4 [CI: 2.2–2.5] vs 3.2 [CI: 3.0–3.4]; +33.3%), the passing accuracy (72.1 [CI: 70.6–73.5] vs 76.9 [CI: 75.4–78.3]%; +6.8%) and the average duration of the team sequences (6.4 [CI: 5.9–6.8] vs 8.3 [CI: 7.8–8.7] seconds; +25.76%) showed a small increasing trend over the seasons (P < 0.05). In contrast, variables such as the direct speed of progression (2.2 [CI: 2.1–2.3] vs 1.6 [CI: 1.5–1.7] metres/second; -24.5%), key passes (8.1 [CI: 7.6–8.5] vs 6.8 [CI: 6.3–7.2]; -15.8%), and the sequences that ended in the attacking third (64.8 [CI: 62,7–66.8] vs 57.1 [CI: 55.1–59.2]; -11.7%) or in a shot (13.0 [CI: 12.4–13.6] vs 10.2 [CI: 9.6–10.8]; -21.6%) showed a small decreasing trend from 2008/09 to 2020/21 (P < 0.05). Spanish professional football teams slightly evolved technically and tactically towards a more associative style of play that includes longer passing sequences. This evolution also involved a decreasing speed of progression and fewer technical actions such as through balls, key passes and shots. INTRODUCTION Match analysis in football (soccer) has exponentially grown in the last decades [1] and currently football is one of the most scrutinized team sports.The development of multiple technological systems (GPS devices, multi-camera tracking systems, etc.) has led to the collection of a high quantity of data during football matches even in real time [2].This has allowed researchers and practitioners to have more opportunities to evaluate the activity of the players in relation to the technical, tactical, physical, and psychological match performance more accurately [3].In this developing context, recent investigations have explored the technical, tactical and physical evolution of professional football competitions such as the Spanish LaLiga [4,5], the English Premier League [6,7], the German Bundesliga [8] and the UEFA Champions League [9].Mainly, these investigations have been focused on the analysis of the evolution of players' performance indicators such as the distance covered at different speeds and the number of technical actions per match.Considering that some Technical and tactical evolution of the offensive team sequences in LaLiga between 2008 and 2021.Is Spanish football now a more associative game? Therefore, the aim of this investigation was to study the longitudinal technical and tactical evolution of the offensive team sequences in the Spanish LaLiga football teams from the 2008/09 to 2020/21 seasons. Sample The study sample consisted of 4940 match observations in 38 professional football teams that competed in the professional Spanish football league (LaLiga) in any of the thirteen seasons under investigation (from 2008/09 to 2020/21).During this period, data of offensive sequences for a total of 9880 sets of data were obtained (i.e., two sets of data per match, corresponding to the two teams competing).Data were obtained from LaLiga, which authorized the analysis of the variables included in this investigation and the publication of results with a scientific objective.In accordance with the ethical guidelines of LaLiga, this investigation does not include information that identifies football players. Procedure This investigation is a descriptive and comparative analysis to determine the evolution of offensive team sequences in the first division of Spanish football for thirteen seasons.The technical and tactical characteristics of each offensive team sequence were collected by team sequence, which can capture the team technical and tactical behaviour and performance when in possession of the ball [13]. For instance, the analysis of offensive team sequences has been a key field of study since the beginning of football performance analysis [14].Since then, a constant debate has been carried out by researchers about the convenience and offensive effectiveness of long versus short passing sequences [15,16].Furthermore, the analysis of passing sequences has also been explored through other techniques such as network analysis [17] and positional data [18].However, although there has been a constant evolution of training methods [19,20], match strategies [21] and match analysis methods [1,22] in the last decade that have provided insightful scientific literature and practical applications, there is still limited evidence about how the game of football has evolved over the years.Thus, it seems necessary and very interesting to explore how offensive team sequences are evolving technically and tactically to explore the effects of training and competition methods.In this analysis, the evaluation of the offensive moment could capture key technical and tactical variables related to the speed of the ball when progressing to the opponent's goal, the duration and number of skilled actions such as passes, as well as the location of ball possession and player's movements, which are key to evaluate styles of play [23] and to achieve football success [24]. TABLE 1. Description and definition of the technical and tactical variables evaluated in the present study. Dimension Variables Description Characteristics of team sequences Number of sequences Technical variable that quantifies the total number of offensive team sequences registered per team in a match. Sequence width Tactical variable that quantifies the average distance between the leftmost point and the rightmost point reached by the ball in the offensive sequence, in metres. Sequence length Tactical variable that quantifies the average distance that the ball travelled forward during each sequence per team and match, in metres. Sequence time Tactical variable that quantifies the average duration of the offensive team sequences per match, in seconds. Passing performance Total passes Technical variable that quantifies the total number of passes performed per team during the match. Passing accuracy (%) Technical variable that quantifies the percentage of passes that are completed over the total number of passes per match. Passes per sequence Tactical variable that quantifies the average number of passes performed per team during the team offensive sequences. Direct speed Tactical variable that quantifies the average distance that the ball moved towards the opponent goal line during the sequence per second, in metres per second (m/s). Offensive performance Through balls Technical variable that quantifies the total number of passes that penetrated through the opposing defensive line per team during the match. Key passes Technical variable that quantifies the total number of passes that allow the recipient of the ball to directly shoot at goal per team during the match. Sequences that end in the attacking third Tactical variable that quantifies the total number of offensive team sequences that enter and finish in the final third of the field during the match. Sequences that end in a shot Tactical variable that quantifies the total number of offensive team sequences that lead to producing a shot during the match. Offensive team sequences in football Mediacoach, a video tracking system that can validly assess teams' match statistics during match play [25].Mediacoach collects data by using information from OPTA Sportsdata and then organizes the information to facilitate the analysis of variables for professional football teams.Recent investigations have presented the analysis of several of these data such as match statistics [26] and offensive and defensive playing style variables [27].With this system, all the team sequences produced in each LaLiga match are automatically recorded and analyzed.Operationally, an offensive team sequence is defined as a passage of play in possession of the ball that belongs to one team and is ended by defensive actions, stoppages in play or a shot.For each offensive team sequence, a total of twelve technical and tactical variables were analyzed to describe the general characteristics of the team sequences, as well as the passing and the offensive performance (see Table 1).On one hand, this analysis includes collective tactical variables such as number of sequences per match, sequence width, length, duration, passes per sequence, direct speed, sequences that end in the attacking third and sequences that end in a shot.On the other hand, technical variables such as number of passes per match, passing accuracy, key passes and through passes are also included in the study. Statistical analysis The data were transferred from Mediacoach to a .csvdatabase which was organized in Microsoft Excel.All statistical analyses were carried out using the software IBM SPSS Statistics Version 27.0.Due to the hierarchical structure of teams' performance in football (each team has its own tactical style), a multilevel mixed model [28] was performed to cluster the collective performance (level 2) into teams (level 1).With this organization of the data, a generalized linear model was carried out to explore the longitudinal effect of the season (fixed effects) on the different tactical variables evaluated in this study, considering the effect of the team (random effects).Thus, the "team effects" represented unobserved team characteristics that influence the collective performance and account for the non-independence of the data [29].To consider the possible contextual effects [30], the model included as fixed effects other four contextual variables: match location (home versus away), match outcome (win versus draw vs lose), ranking of the opponent in quartiles (first, second, third and fourth) and ranking of the team in quartiles (first, second, third and fourth).The Cohen F 2 statistic was calculated as the effect size of the fixed effects [31].The Cohen F 2 effect size is a measure of the proportion of variance in the outcome explained by the fixed effects included in the model.In this regard, it was considered a trivial effect at a value lower than 0.02, small at a value of 0.02, medium at a value of 0.15 and large at a value of 0.35 [32]. Graphic charts with the predicted means and confidence intervals were displayed to show the longitudinal evolution of the different tactical variables throughout the seasons according to the generalized mixed linear model.Pairwise comparisons of the estimated means were performed through Fisher's least significant test.The significance level was set to P < 0.050. RESULTS Table 2 shows the comparison of data between the 2008/2009 and 2020/2021 seasons.All variables, except for "sequence width" were different between the seasons (P < 0.05).Particularly, the number of sequences largely decreased over the seasons (P < 0.05), while the sequences length moderately decreased (P < 0.05) and the sequence time had a small increase over the years (P < 0.05).Regarding the average duration of teams' sequences, a significant increasing trend was found (P < 0.05), especially from the season 2014/15 onwards, when the average time of team sequences changed from 6.4 seconds to 8.3 seconds (+33.3%). Figure 2 shows that total passes per match (+18.7%),passing accuracy (+6.8%) and the number of passes per sequence (+34.4%)registered an increasing and constant trend over the seasons (P < 0.05).Otherwise, the direct speed of progression showed a decreasing trend, especially from the 2013/2014 season onwards (P < 0.05). Regarding the offensive indicators, the number of key passes significantly decreased from the 2014/15 season (P < 0.05), while the number of through balls started a decreasing process from the 2011/12 season onwards (P < 0.05).Finally, the number of sequences that ended in the attacking third gradually decreased (P < 0.05).Also, Regarding the passing performance, the total number of passes, passing accuracy and passes per sequences showed a small increase in the last years (P < 0.05), while the direct speed increased (small size effect, P < 0.05).As for the offensive indicators, the number of through balls, key passes and sequences that end in the attacking third and in a shot significantly decreased from 2008/2009 to 2020/2021.play that includes longer passing sequences at the cost of decreasing speed of progression during the game.These results also suggest that football teams now produce more elaborate sequences until the opportunity to attack is present, as the number of goals has not changed [4], despite there being a lower number of sequences that end in the attacking third and those ending in a shot. Our study revealed that the number of passes per match, the number of passes per sequence and the average duration of each sequence showed a small increasing trend over the seasons.Although existing literature has reported that the Spanish LaLiga is characterized by a more combinative style of play in comparison to other leagues [10,33], this study confirms that this style is the result of a constant evolution, at least since the 2008/09 season.In line with these results and focusing on the role of specific playing positions, Lago-Peñas et al. [5] found that, in LaLiga, central backs significantly increased the number of total passes and the number of long passes from 2012 to the number of sequences that ended in a shot decreased (p < 0.05) with statistically significant differences between the 2008/09 season and all the seasons after the 2011/12 season (Figure 3). DISCUSSION The aim of this investigation was to study the technical and tactical 2019.This indicates that central backs are now more involved in the build-up phase of the game and suggests that offensive sequences are now formed further from the opposing team's goal.Additionally, the direct speed, a variable that measures the capacity of the team to progress in each sequence, as the average speed of ball movement towards the opponent's goal line during the sequence, has decreased, suggesting that sequences are now more horizontal, reducing the use of direct play game styles. In this regard, Yi et al. [9] recently observed that teams in European leagues are now more focused on the control of match play by increasing passing frequency and accuracy, which coincides with the outcomes of the current investigation.Other studies have also reported an increasing passing performance in English Premier League from 2006 to 2013 [7] and the German Bundesliga from 2014 to 2017 [8].In contrast, Zhou et al. [34] did not find any increase in the passing performance in the Chinese Super League from 2012 to 2017.Although these results reflect the different technical and tactical development of each domestic competition according to cultural, economic, and social dimensions, it seems that there is an international tendency for a higher passing frequency in elite football teams, converting football today into a more combinative sport than it was a decade ago. In the Spanish context, this technical and tactical evolution towards a more associative game seems to have had a small boost during the 2010/11 season.In this season, the number of passes per match increased by 8.4% from the previous season (Figure 2), with 8.0% more passes per sequence.It is interesting to mention that this season took place just after the Spanish National team won the 2010 World Cup tournament in South Africa with a very popular possession-oriented style of play [35,36].Also, the 2010/11 season was played two years after Pep Guardiola took over FC Barcelona in a very successful period that started by winning one Offensive team sequences in football which is probably due to the better protection of key spaces by the defensive teams.These technical and tactical characteristics seem to create a more controlled and predictable offensive context where penetrating into the final third of the pitch or achieving a shot could become more difficult than several years ago. This study is not exempt from limitations.Firstly, it is crucial to mention that this investigation was performed following a static approach [41], where data are collected per match without considering the contextual variables that change throughout the match.Additionally, the current study was carried out with data from a national football league of elite male players and the results of the investigation should not be extrapolated to other leagues, other categories, or to women's football. CONCLUSIONS The current data show that football teams competing in LaLiga have slightly evolved technically and tactically towards a more associative and combinative style of play that includes longer passing sequences both in time and quantity of passes.This technical and tactical evolution also presents a decreasing speed of progression and distance progressed in each sequence, as well as a small decrease in the number of through balls, key passes, and shots. As practical applications for football coaches and sporting directors, it seems that football tactics are in a slight evolution towards a more associative style of play with less offensive verticality and penetration, which requires coaches not only to train players to be accurate in their actions to possess the ball but also to be able to acquire excellent dribbling or passing skills to disrupt the defensive organization of the opposing team.Additionally, as the number of through balls, key passes, and shots has been decreasing in recent years, having players with attributes to produce these actions may be key to having a successful football squad.The knowledge provided by this study encourages football coaches and analysts to reflect on the actual development of football tactics. Disclosure of interest The authors report no conflict of interest. Data availability statement The data that support the findings of this study are available from LaLiga.Restrictions apply to the availability of these data, which were used under license for this study.Data are available from the authors with the permission of LaLiga. Champions League title and two consecutive LaLiga trophies with a style of play characterized by making a great number of passes and creating constant interactions and connections between players [37].Thus, the possible effect of these two successful teams that implemented possession-oriented styles of play may have influenced the rest of the Spanish teams to adopt more associative passing sequences in the next years to try to achieve higher success. Despite the small increase in the number of passes and time per sequence, our study found that the length of the sequences and the speed of progression showed a small descending trend over the seasons.From a practical perspective, these two changes suggest that teams progressed a shorter distance than before, despite using longer sequences.Furthermore, the number of through balls and key passes also showed a descending trend.Although further research is needed to understand the current attributes of passing in football, these results do not indicate that passing tempo or ball speed has decreased, but that teams now decide to produce more elaborate plays with more horizontal and backward passes until a favorable moment to attack and progress is obtained.Additionally, these findings suggest that teams could have strengthened their defensive organization [4], and pressure on the ball, which would make it more difficult to penetrate through the opposing lines, forcing attacking teams to display more passing combinations to overcome the defensive teams.This tactical scenario with both a more offensive combination and defensive organization could reduce the opportunities for the teams to have moments of transitions from defense to attack to recover the ball, which is very effective to create goal-scoring opportunities [38,39].Similar to these results, Konefal et al. [8] suggested that the evolution of football seems to be directed towards more collective and organized tactics, reflecting a better understanding of the tactical roles of players.These technical and tactical features could have a physical impact so that this better collective organization could help players to optimize their physical efforts and save energy to perform higher-intensity actions during the game, which is another new feature of modern football [40].In line with this interpretation, the recent study by Lago-Peñas et al. [5] recorded a reduction in the total distance covered by LaLiga players, while the number of efforts made at high-intensity running increased in recent seasons. As for the offensive performance, our data revealed that those sequences that ended in the attacking third or achieved a shot showed a descending trend over the seasons.Our results support the findings of Errekagorri et al. [4], which demonstrated how Spanish teams produced a smaller number of technical actions such as shots and crosses in recent seasons.These facts are surely aligned with the previous findings that indicated fewer through balls and key passes, Figure 1 FIG. 1 . Figure 1 depicts the longitudinal evolution of the number of sequences per match, sequence width, length, and duration.In comparison to the 2008/09 season, the number of sequences started to decrease in the 2016/17 season (p < 0.05) and remained lower in the last five seasons under investigation.The sequence width remained stable through the seasons, rounding to the average of 34 metres, while the length of the sequences showed a descending trend over the seasons, initiating this decline in the 2014/15 season (p < 0.05). FIG. 3 . FIG. 3.Predicted means and 95% confidence intervals for the variables "key pass", "through balls", "sequences that end in the attacking third" and "sequences that end in a shot" in LaLiga from 2008/09 to 2020/21 according to the generalized mixed linear model.*= Significantly different from the 2008/2009 season (Fisher's least significant test) (p < 0.05).	v3-fos-license
2021-10-22T15:53:50.850Z	2021-09-01T00:00:00.000	239241390	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.mdpi.com/2227-9717/9/9/1563/pdf", "pdf_hash": "1919563b24339d2bb5c721a62dc443dd29215ee1", "pdf_src": "ScienceParsePlus", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45472", "s2fieldsofstudy": [ "Computer Science" ], "sha1": "d75787432819c0a30d34b19397fb27a6c9a60fc1", "year": 2021 }	pes2o/s2orc	Improving Sports Outcome Prediction Process Using Integrating Adaptive Weighted Features and Machine Learning Techniques : Developing an effective sports performance analysis process is an attractive issue in sports team management. This study proposed an improved sports outcome prediction process by integrating adaptive weighted features and machine learning algorithms for basketball game score prediction. The feature engineering method is used to construct designed features based on game-lag information and adaptive weighting of variables in the proposed prediction process. These designed features are then applied to the ﬁve machine learning methods, including classiﬁcation and regression trees (CART), random forest (RF), stochastic gradient boosting (SGB), eXtreme gradient boosting (XGBoost), and extreme learning machine (ELM) for constructing effective prediction models. The empirical results from National Basketball Association (NBA) data revealed that the proposed sports outcome prediction process could generate a promising prediction result compared to the competing models without adaptive weighting features. Our results also showed that the machine learning models with four game-lags information and adaptive weighting of power could generate better prediction performance. Introduction The sports market has proliferated at the beginning of the 21st century with new approaches and techniques such as streaming broadcasting, social media, and the global supply chain. With the growth of sports market value, sports team management attracted considerable attention and became a popular topic from researchers. For example, some studies established decision support methods to assist the teams' decision of rookie draft [1,2]. Some research aimed to analyze the teams' player selection and scouting policy [3,4]. Moreover, developing an effective sports performance analysis process such as determining the factors that affect the results of sports games [2,5] and predicting games or players' performance [6,7] is one of the most attractive issues in sports teams' management. Sports outcome prediction is a significant part of the sports performance analysis process. It influences many sports markets, such as sports team management, betting, and customer service, since the accurate prediction of sports game outcome provides detailed information for managing strategy reference, bookmakers, and attraction for viewers. For example, a team supervisor, such as a general manager (GM), is facing numerous vital decisions on the daily management of the team. The accurate prediction of future games provides the reference of marketing strategy to design specific fans' service activities. Moreover, detail and accurate prediction provide GM and head coach vital information Processes 2021, 9,1563 3 of 16 amount of variables and observations by combining several randomized decision tress and aggregates their predictions by averaging [50][51][52][53][54][55][56][57][58][59][60]. SGB is a hybrid method that merging boosting and bagging techniques, and the input data are selected by random sampling at each stage of the steepest gradient algorithm-based boosting procedure [24,45,52,[61][62][63][64][65][66][67]. XGBoost is a supervised machine learning algorithm developed from a scalable end-to-end gradient tree boosting principle [24,[68][69][70][71][72][73]. ELM is a single-hidden-layer feedforward neural network that randomly selects the input weights and systematically calculates the output weights of the network [24,[73][74][75][76][77]. In the proposed NBA game score prediction process, the most acceptable and accessible statistics of an NBA game are collected and used as initial variables. Then, using these variables as input, the feature engineering method is used to construct the designed feature based on the interaction of game-lag information and adaptive weighting of variables. These designed features are then applied to the five machine learning methods, i.e., CART, RF, SGB, XGBoost, and ELM, to build prediction models. Finally, by evaluating the performance of prediction models, the sufficient selection of game-lag information, and proper adaptive weighting of variables combination on features were identified. This research is organized as follows: Section 1 presents the background and the concept of this study. Section 2 shows the brief introduction of machine learning techniques that were used in this research. Section 3 demonstrates the details of the outcome of the proposed sports prediction process. Section 4 gives the proposed sports outcome prediction process's empirical results, followed by the Conclusions section. Methodology Five machine learning techniques involving CART, RF, SGB, XGBoost, and ELM are utilized in this study. CART is a statistical procedure often used as a regression tool to analyze continuous data. CART can be concluded into three stages. The first stage involves developing the tree using a recursive partitioning technique to filter variables and split data points using a splitting criterion. After a large tree is produced, the second stage uses a pruning procedure that coordinates with a minimal cost complexity measure. The methodology's last stage is identifying the optimal tree by corresponding with a tree yielding the lowest crossvalidated or testing set error rate [42][43][44][45][46][47][48][49]. For modeling the CART model, the "rpart" R package of version 4.1-15 [78] with pruning strategy was used. RF has provided promising performance as a general-purpose regression method. By combining several randomized decision trees and aggregates their predictions by averaging, the approach of RF has shown excellent performance in the dataset with large amount of variables and observations. It is also flexible enough to be implemented to large-scale task, is conveniently adapted to various ad hoc learning problems, and returns measures of variable importance [50][51][52][53][54][55][56][57][58][59][60]. The "randomForest" R package of version 4.6-14 [79] was used to construct the RF model. SGB is a hybrid algorithm that merging boosting and bagging techniques. Data are filtered by random sampling at each phase of the steepest gradient method-based boosting process. SGB develops smaller trees at each stage of the boosting procedure instead of developing a full regression tree. Optimal data fractionation is computed by coordinate with a consequential process, and the residual of every fraction is obtained. The results are comprised to lower the sensitivity of these methods for target data [24,45,52,[61][62][63][64][65][66][67]. The SGB model was constructed by the "gbm" R package of version 2.1.8 [80]. XGBoost commonly uses a tree-based supervised machine learning technique developed from a scalable end-to-end gradient tree boosting principle. A weak learner is developed to be optimistically correlated with the negative gradient of the loss function and refers to the entire model. XGBoost is a generalized gradient boosting decision tree applied by a new tree scanning method that reduces tree building time. XGBoost moderated overfitting to support arbitrary adaptable loss functions by regularization term [24,[68][69][70][71][72][73]. The "xgboost" R package of version 1.3.2.1 [81] is used to construct the XGBoost model. ELM is a single-hidden-layer feedforward neural network that randomly indicates the weighting of input and systematically calculates that weighting of the network's output. ELM has a quicker model building time compared to the traditional feedforward network machine learning methods. ELM reduces common disadvantages found in gradient-based techniques [24,[73][74][75][76][77]. The ELM model was constructed by the "elmNN" package version 1.0 [82]. The default used activation function in this package is radial basis. All R packages were implemented in R software of version 3.6.2 [83]. The default setting of modeling strategy of each package is used. To find the best hyper-parameter set for building an effective prediction model, "caret" R package of version 6.0.84 [84] was implemented for tuning the important hyper-parameters of each machine learning algorithm. Proposed Sports Outcome Prediction Process The adaptive weighting information was integrated into the feature design process in the proposed sports prediction process outcome. Then, the five machine learning algorithms are implemented to predict the final score of the NBA game using the designed features. The flowchart of the proposed process is shown in Figure 1. refers to the entire model. XGBoost is a generalized gradient boosting decision tree applied by a new tree scanning method that reduces tree building time. XGBoost moderated overfitting to support arbitrary adaptable loss functions by regularization term [24,[68][69][70][71][72][73]. The "xgboost" R package of version 1.3.2.1 [81] is used to construct the XGBoost model. ELM is a single-hidden-layer feedforward neural network that randomly indicates the weighting of input and systematically calculates that weighting of the network's output. ELM has a quicker model building time compared to the traditional feedforward network machine learning methods. ELM reduces common disadvantages found in gradientbased techniques [24,[73][74][75][76][77]. The ELM model was constructed by the "elmNN"package version 1.0 [82]. The default used activation function in this package is radial basis. All R packages were implemented in R software of version 3.6.2 [83]. The default setting of modeling strategy of each package is used. To find the best hyper-parameter set for building an effective prediction model, "caret" R package of version 6.0.84 [84] was implemented for tuning the important hyper-parameters of each machine learning algorithm. Proposed Sports Outcome Prediction Process The adaptive weighting information was integrated into the feature design process in the proposed sports prediction process outcome. Then, the five machine learning algorithms are implemented to predict the final score of the NBA game using the designed features. The flowchart of the proposed process is shown in Figure 1. Step 1: Data Collection The first step is data collection. We acquired data from the basketball-reference website (https://www.basketball-reference.com, accessed on 15 March 2021) for every NBA 2018-2019 regular-season games. This NBA season consists of 1230 regular-season games, Step 1: Data Collection The first step is data collection. We acquired data from the basketball-reference website (https://www.basketball-reference.com, accessed on 15 March 2021) for every NBA 2018-2019 regular-season games. This NBA season consists of 1230 regular-season games, and each game has two teams (home team and away team) on the court. Each game presents two data points, one from home and another from the away team. Therefore, 2460 data points were collected and used in this study. A total of 15 variables are collected and utilized in this research. One is the team's final game score, and the remaining 14 are the most widely used statistics of the basketball game, including the team's defensive and offensive performance. Table 1 presents variable Table 1. Variable Description. Variables Abbreviation Description 3-Point Field Goal Percentage of a team in t th game X 5,t FTA Free Throw Attempts of a team in t th game X 6,t FT% Free Throw Percentage of a team in t th game X 7,t ORB Offensive Rebounds of a team in t th game X 8,t DRB Defensive Rebounds of a team in t th game X 9,t AST Assists of a team in t th game X 10,t STL Steals of a team in t th game X 11,t BLK Blocks of a team in t th game X 12,t TOV Turnovers of a team in t th game X 13,t PF Personal Fouls of a team in t th game X 14,t H/A Home or Away game of a team in t th game Y t Score Team Score of a team in t th game Step 2: Data Normalization Data normalization is implemented before feature construction since each variable has different scales. The min-max normalization technique is used to transform a value v of variable V to v in the range [0, 1] by using the following equation as a calculation: where maxX i and minX i are the maximum and minimum values for the attribute X i . Data normalization is implemented to ensure that larger input variable values will not influence smaller input variables values to reduce prediction errors. Step 3: Feature Design based on Adaptive Weighting This step aims to integrate adaptive weighting techniques into feature construction. We design our features for the prediction models based on the normalized variables shown in Table 1. This feature construction proceeds based on the interaction of two dimensions: game-lag information and the exponential power of adaptive weighting. First, we define the game-lag information as "the n th game before game t". Researchers utilized the single game-lag information of up to six games for model construction in recent related research [14][15][16][17][18][19][20][21]. This research considers three to six game-lag information instead of considering only single game-lag information. Moreover, this research considers the moving average of the past statistic of a basketball game in order to be sufficient for evaluating a team's performance. We calculate the mean value of a normalized variable within l game-lags to evaluate a team's performance during a specific duration. Second, it is crucial to understand that the nearest data point theoretically has more influence on the prediction unknown target. Therefore, this research designed the adaptive weighting method based on inverse-distance weighting (IDW) by integrating exponential power d as a weighting control parameter. The designed feature X l,d i,t can be described as follow: Processes 2021, 9, 1563 6 of 16 and the adaptive weighting, AW l n,d , is calculated as follow: i,t is the designed i th predictor variable at the t th game with l game-lags based on d exponential power of adaptive weighting. For instance, for the first normalized variable (i = 1), if we wish to design a feature considering four game-lag's information (l = 3) with weighting control parameter d = 1 for the game No. 25 (or 25th game, t = 25) of a team, the values of the first variable in the previous three games are calculated as the designed feature. That is 16. This instance can be found in Figure 2 as the line of d = 1. It can be observed in Figure 2 that the weighting distribution on the line of d = 0 will be equal with all the variables, which represents the simple average method. The weighting distribution of d = 0 is demonstrated Figure 4 illustrates the weighting distribution of four feature set under l = 5. Figure 5 shows the weighting distribution of four feature set under l = 6. Note that with the higher value of selected game-lag information, the nearest data point (t-1) weighting is not necessarily growing with it. Step 4: Prediction Model Construction In this phase, we construct a prediction model for predicting the final scores of the NBA games by using 14 designed features with five machine learning methods, namely CART, RF, SGB, XGBoost, and ELM. According to step3, each variable can be extended to 16 features which is the combination of four game-lags and four weighting control parameters. For modeling a machine learning prediction model, the 14 designed variables with Most recent research using previous basketball game's statistics to predict the outcome of basketball games construct the feature by using the simple average method [14][15][16][17][18][19][20][21] Processes 2021, 9, 1563 8 of 16 which is equal to setting weighting control parameter (d) = 0 in Equations (2) and (3). The weighting distribution of adaptive weighting method (d = 1, d = 2, and d = 3) allocates the most weighting on the nearest data point, i.e., the data point on t-1. The weighting of t-1 is growing with the higher of a weighting control parameter. On the other hand, the weighting of the farthest data point is allocated with the lowest weighting. The margin between the highest weighting and lowest weighting is increased with the higher weighting control parameter. Step 4: Prediction Model Construction In this phase, we construct a prediction model for predicting the final scores of the NBA games by using 14 designed features with five machine learning methods, namely CART, RF, SGB, XGBoost, and ELM. According to step3, each variable can be extended to 16 features which is the combination of four game-lags and four weighting control parameters. For modeling a machine learning prediction model, the 14 designed variables with a specific game-lag and weighting control parameter are used as predictor variables. That is, for a machine learning method, 16 prediction models will be can be generated for evaluating the effects of different game-lags and different weighting control parameters based on prediction performance. That is, using the designed features with a specific game-lag and weighting control parameter (X l,d i,t ) to predict the final score of a game (Y t ) can be expressed as the following equation: Note that all 14 designed features (1 ≤ i ≤ 14) were used with three to six game-lags' information (3 ≤ l ≤ 6) and with zero to three of the weighting control parameters (0 ≤ d ≤ 3) for each Y t . As aforementioned, this research compares the prediction performance with different weighting control parameters under a different selection of game-lag information. Since we use up to six games' information as our game-lag information, the season's first six games are skipped (7 ≤ t ≤ 82). Step 5: Performance Evaluation This research aims to discover the influence of patterns from previous games to the target game and is a cross-sectional analysis. Therefore, this study utilized the cross-validation method which is commonly used in sports outcome prediction [85][86][87] to estimate the performance of the proposed prediction process. Moreover, according to [13], cross-validation method can be a better validation method in sports outcome prediction. We replicate 10-fold cross-validation 100 times. This study used the root mean square error (RMSE) as the indicator to evaluate the prediction performance of each method as it is one of the most commonly used prediction performance indicators [88,89] and has been used as a standard statistical metric to measure model performance in meteorology, air quality, and climate researches [90][91][92]. Many sports outcome-prediction research used only RMSE as the prediction performance indicator [93][94][95]. Therefore, this research use RMSE as prediction performance indicator. It is calculated as Equation (6): where n is the sample size, and e i represents the error of predictions. Step 6: Final Results In the final phase, after the performance evaluation of the prediction models, the best production process can be obtained. Based on the best prediction process, we shall determine the sufficient selection of game-lag information and practical preference of exponential power on adaptive weighting. Empirical Results In this research, each NBA game's statistics of both competing teams in the 2018-2019 regular season were collected to evaluate the performance of the proposed basketball game score prediction process. Figure 6 shows the prediction performance of each prediction model with different weighting control parameters (d) under game-lag (l) = 3. It can be observed that prediction performance varies from different weighting control parameters. Each prediction model reach its best prediction performance with weighting control parameter(d) where n is the sample size, and represents the error of predictions. Step 6: Final Results In the final phase, after the performance evaluation of the prediction models, the best production process can be obtained. Based on the best prediction process, we shall determine the sufficient selection of game-lag information and practical preference of exponential power on adaptive weighting. Empirical Results In this research, each NBA game's statistics of both competing teams in the 2018-2019 regular season were collected to evaluate the performance of the proposed basketball game score prediction process. Figure 6 shows the prediction performance of each prediction model with different weighting control parameters (d) under game-lag (l) = 3. It can be observed that prediction performance varies from different weighting control parameters. Each prediction model reach its best prediction performance with weighting control parameter(d) Figure 8 presents the prediction performance of prediction models with different weighting control parameters (d) under game-lag (l) = 5. Every prediction model reaches its best prediction performance with weighting control parameter (d) = 1, such as CART (RMSE = 12.4316), RF (RMSE = 12.1525), SGB (RMSE = 12.2448), XGBoost (RMSE = 12.3145), and ELM (RMSE = 12.6748). Note that the prediction performance became worse with the increase of weighting control parameters from 1 to 3. Table 2 summarizes the mean and standard deviation (SD) of RMSE of each machine learning method with different weighting control parameters (d) under a different scenario of game-lag information (l). It can be observed that every model reaches their best prediction performance with weighting control parameter value of level one (d = 1) under the scenario of selecting four as game-lag information (l = 4), such as CART (RMSE = 11.7564), RF (RMSE = 11.6303), SGB (RMSE = 11.5586), XGBoost (RMSE = 11.6941), and ELM (RMSE = 11.8020). Compared to the methods with a simple average weighting feature (d = 0), the models with a weighting control parameter value of level one (d = 1) provide promising effects on the improvement of prediction performance. Note that with the higher value of the weighting control parameter, the prediction performance is not necessarily encouraged. Table 2 summarizes the mean and standard deviation (SD) of RMSE of each machine learning method with different weighting control parameters (d) under a different scenario of game-lag information (l). It can be observed that every model reaches their best prediction performance with weighting control parameter value of level one (d = 1) under the scenario of selecting four as game-lag information (l = 4), such as CART (RMSE = 11.7564), RF (RMSE = 11.6303), SGB (RMSE = 11.5586), XGBoost (RMSE = 11.6941), and ELM (RMSE = 11.8020). Compared to the methods with a simple average weighting feature Processes 2021, 9, 1563 11 of 16 (d = 0), the models with a weighting control parameter value of level one (d = 1) provide promising effects on the improvement of prediction performance. Note that with the higher value of the weighting control parameter, the prediction performance is not necessarily encouraged. However, it also can be observed in Table 2 that the difference between models' results is relatively small. The confident interval (CI) is calculated in order to determine whether the difference between models and feature combinations is significant. Table 3 shows the confident intervals of each machine learning method with different weighting control parameters and different game-lag information. It reveals that the prediction performance of weighting control parameter of one and a game-lag of four significantly outperformed other feature combinations used as an input predictor for the five machine learning algorithms. It also can be observed from Tables 2 and 3 that although the SGB with d = 1 and l = 4 slightly outperforms the competing methods in this feature combination, the difference is not statistically significant. Therefore, these five machine learning algorithms provide a promising prediction performance by using designed features with a weighting control parameter of one and game-lag of four as predictor variables. Moreover, from Tables 2 and 3, it can be seen that the models' prediction performance with weighting control parameter of two and three are not as good as a model with a weighting control parameter of one. The potential cause of this circumstance can be explained as, by observing Figures 2-5, the weighting of each data point is linearly declining with a weighting control parameter of one while the weighting of each data point is nonlinearly declining with a weighting control parameter of two and three. The weighting distribution curve with a linear decline represents a team's performance over the last few games are linearly and stable, referable to the target game t. Contrarily, for the weighting distribution curve with a non-linear decline, a team's performance over the last few games is unstable and is relatively not referable. That is, the influence of the performance in nearer games are enhanced and subsequently too high, while the performance in farther games are declining too fast. However, since NBA is a professional team sport, coach and players are well-trained to adapt themselves by substitute the cooling players with hot-hand players, using time-outs to adjust their condition, or provide and receive assistance with teammates to cover the unstable performance. Therefore, the weighting distribution with a linear decline with a farther data point which represents the stable team performance is a more appropriate distribution for the prediction on outcome of team sports. Conclusions This research integrated the designed features using the adaptive weighted method and CART, RF, SGB, XGBoost, and ELM machine learning methods for constructing an effective sports outcome prediction process. The designed features were based on the interaction of three to six pieces of game-lags information, and four different levels of the adaptive weighting of variables were generated. This study collected data from all the regular-season games of the NBA 2018-2019 seasons as illustrative examples. Empirical results showed that the proposed sports outcome prediction could generate a promising prediction result compared to the competing models without adaptive weighting features. All the five machine learning methods reach their best prediction performance with the weighting control parameter value of level one and four pieces of game-lags information. Although SGB in this feature combination slightly outperforms the competing methods, the difference is not statistically significant. Therefore, these five machine learning algorithms provide promising prediction performance by using feature combination with weighting control parameter of one and game-lag of four to generate the input features. Integrating the machine learning methods with adaptive weighting based on feature selection and combination strategies to generate an improved version of the proposed scheme is worthy further investigated since the exploration of feature selection results is an important task in the implementation of machine learning algorithms in sports outcome prediction [9]. Thus, modifying the proposed scheme to adapt to feature selection techniques could be one of the future research directions. Moreover, exploring the performance of the proposed scheme with more NBA seasons' data could be also a future research direction.	v3-fos-license
2018-04-03T03:08:33.635Z	1998-03-27T00:00:00.000	318538	{ "extfieldsofstudy": [ "Medicine", "Biology" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "http://www.jbc.org/content/273/13/7181.full.pdf", "pdf_hash": "cd1c4c0bf8b27ca0c012954ec644d5acf1917b18", "pdf_src": "Adhoc", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45473", "s2fieldsofstudy": [ "Biology", "Medicine" ], "sha1": "1462bdbe184679f714824eeff933178b75e34ab8", "year": 1998 }	pes2o/s2orc	Regulation of the phosphorylation state of rhodopsin by dopamine. G protein-coupled receptors (GPCRs) are regulated by kinases and phosphatases that control their phosphorylation state. Here, the possibility that the state of GPCR phosphorylation could be affected by paracrine input was explored. We show that dopamine increased the rate of dephosphorylation of rhodopsin, the light receptor, in intact frog retinas. Further, we found that rod outer segments from dopamine-treated retinas contained increased rhodopsin phosphatase activity, indicating that this effect of dopamine on rhodopsin was mediated by stimulation of rhodopsin phosphatase. Dopamine is a ubiquitous neuromodulator and, in the retina, is released from the inner cell layers. Thus, our results identify a pathway for feedback regulation of rhodopsin from the inner retina and illustrate the involvement of dopamine in paracrine regulation of the sensitivity of a GPCR. G protein-coupled receptors (GPCRs) 1 represent a widespread family of proteins that transduce a large variety of signals, such as light, odorants, hormones, and neurotransmitters. They have common structural elements, including seven transmembrane domains, and are regulated by many homologous mechanisms. Understanding these regulatory mechanisms is therefore a central question in signal transduction. Deactivation of a GPCR involves phosphorylation of the receptor, and its subsequent resensitization requires dephosphorylation. Accordingly, the light receptor, rhodopsin, undergoes light-dependent phosphorylation and must be subsequently dephosphorylated (1). The phosphorylation state of GPCRs is regulated typically by GPCR kinases (GRKs) and second messenger-regulated kinases, and, on the other hand, by phosphatases. GRKs preferentially phosphorylate agonist-occupied or activated GPCRs, whereas the second messenger-dependent kinases (cAMP-dependent protein kinase and protein kinase C) may phosphorylate nonactivated receptors (2,3). Phosphatases that regulate GPCRs belong to the phosphatase 2A family or are dependent on Ca 2ϩ (4 -8). The kinases and phosphatases that affect the phosphorylation of GPCRs may in turn be regulated. Most obviously, second messenger-regulated kinases may mediate input from different signal transduction pathways (9 -11). GRKs can also be regulated by other pathways (12). Rhodopsin kinase (GRK1) activity, for example, is inhibited by the Ca 2ϩ -binding protein, recoverin, when Ca 2ϩ levels are high (13). ␤-Adrenergic receptor kinase (GRK2) and GRK5 are both phosphorylated by the second messenger kinase, protein kinase C, resulting in their activation (14,15) and inactivation (16), respectively. Less is known about the dephosphorylation of GPCRs and the regulation of their phosphatases, although rhodopsin dephosphorylation appears to be affected by Ca 2ϩ levels. Bovine rhodopsin can be dephosphorylated by a phosphatase 2A (5,6) and by a Ca 2ϩ -sensitive phosphatase (7), both of which are present in photoreceptor outer segments. In Drosophila, rhodopsin is dephosphorylated by the rdgC protein, which possesses a putative Ca 2ϩ -binding domain in addition to a phosphatase catalytic domain (8,17). Regulation of kinases and phosphatases thus provides upstream mechanisms for modulating GPCRs. The focus of the present study was on whether input from the inner retina could affect the phosphorylation state of rhodopsin. Such input could potentially originate from general light-or dark-adaptive signals or from a circadian oscillator. The most likely candidate for an intercellular messenger is the major catecholamine in retina, dopamine. Amacrine and interplexiform cells in the inner retina release dopamine in response to light and under the control of a circadian clock (18,19). Photoreceptor cells possess dopamine receptors (20 -22), and dopamine has been shown to influence retinomotor movements and phototransductive membrane shedding (23)(24)(25)(26). We demonstrate here that dopamine feedback to the photoreceptor cells affects the kinetics of rhodopsin dephosphorylation in intact frog retinas, indicating that the light receptor can be regulated by paracrine input. EXPERIMENTAL PROCEDURES Materials-Dopamine hydrochloride, R(ϩ)-SCH-23390 hydrochloride, and spiperone hydrochloride were purchased from Research Biochemicals International, Inc. [ 32 P]Orthophosphoric acid (ϳ9,000 Ci/ mmol) was from NEN Life Science Products. All other chemicals were reagent grade. Northern grass frogs (Rana pipiens) weighing 20 -30 g were purchased from Carolina Biological Supply Co. and treated according to NIH and University of California at San Diego animal care guidelines. Incubation of Frog Retinas and Analysis of Rhodopsin Phosphorylation-The procedure for incubation of retinas and analysis of rhodopsin phosphorylation followed that described previously (27). Retinas were removed from dark-adapted animals. Each intact retina was incubated under dim red light in 1 ml of amphibian culture medium (35 mM NaHCO 3 , 75 mM NaCl, 3 mM KCl, 2 mM CaCl 2 , 1 mM MgSO 4 , 10 mM Na-HEPES, pH 7.3, 10 g/ml phenol red, 1 mg/ml casamino acids, 10 mM ␤-D(ϩ)-glucose, 0.1 mg/ml Na-L-ascorbate, and 20 Ci/ml [ 32 P]orthophosphate) for 2 h and then for 10 min with or without receptor ligands. One retina from each frog served as the experimental, and the other served as the control. Retinas were illuminated by a calibrated flash of light that photoexcited 6 Ϯ 3 or 80 Ϯ 6% of the rhodopsin (27) and then incubated up to 1 h under dim red light. All incubations were carried out at 22-23°C. The specific 32 P incorporation into rhodopsin was determined after SDS-PAGE (12% acrylamide) by densitometry of Coomassie Blue-stained and radioactive (PhosphorImager) bands and expressed in relative units per constant amount of rhodopsin. Three methods were used for preparation of rhodopsin sam-ples: Method 1, at the end of the incubation period, 250 l of buffer A (20 mM sodium phosphate, 50 mM Tris-Cl, 10 mM EDTA, 10 mM EGTA, pH 7.3) was added to each retina. Rod outer segments (ROSs) were detached from the rest of the retina by vortexing for 10 s. After allowing the retina remnants to settle for 2 min on ice, the ROS-enriched fraction was removed and centrifuged for 30 s at 16,000 ϫ g. The pellet (crude ROSs) was suspended in 100 l of 1% SDS at room temperature and centrifuged (20,000 ϫ g, 20 min, room temperature) to remove debris, before adding SDS-PAGE sample buffer; Method 2, the crude ROSs from Method 1 were purified further by sucrose gradient centrifugation (28); and Method 3, reactions of rhodopsin phosphorylation in intact retinas were stopped by adding 250 l of 50% trichloroacetic acid, and the total retinal proteins were subjected to SDS-PAGE. In all three methods, each rhodopsin fraction from an individual retina in SDS-PAGE sample buffer was divided into two portions. One was heated at 100°C for 15 min, and the other was kept at room temperature. Heating in SDS promotes the oligomerization of rhodopsin, so that rhodopsin no longer migrates with an apparent molecular mass of ϳ36 kDa in the gel. The heated sample was therefore used to measure protein and radioactivity that was not from rhodopsin in this area (ϳ36 kDa) of the gel. This background was subtracted from the data obtained from lanes with samples that were not heated. Peripherin/rds, for example, is a photoreceptor outer segment phosphoprotein (29) that has a similar apparent molecular mass. We confirmed by Western blot analysis that bovine peripherin/rds does not oligomerize under the conditions used in the present experiments. Assay of Phosphatase Activity in ROSs-32 P phosphorylation of rhodopsin was performed in intact retinas as above. Rod outer segments from dopamine-treated and control retinas were purified as described (Method 2, above), and their cytosolic fractions (0.6 mg/ml) were used to assay phosphatase activity (1 nM rhodopsin; 22°C for 30 min, during which time 32 P release was linear) (7,31). Radioactive products were separated by SDS-PAGE and analyzed by a PhosphorImager. Statistical Analyses-Paired Student's t tests were performed to determine the probability (p) of no significant difference. RESULTS AND DISCUSSION After retinas were exposed to a flash of light, rhodopsin was phosphorylated, reaching maximal phosphorylation level after 10 min (27). The rate of phosphorylation in control retinas and in retinas exposed to 100 M exogenous dopamine was similar (Fig. 1). After 30 min, the level of rhodopsin phosphorylation decreased, with rhodopsin in the dopamine-treated retinas dephosphorylated at a faster rate. By 45 min after the light flash, the level of phosphorylation of rhodopsin in dopamine-treated retinas was only 50% that in control retinas (Fig. 1). A similar result was obtained by three different procedures of sample preparation, as described under "Experimental Procedures"; results from Method 1 are illustrated in Fig. 1. Moreover, this Curves are a result of exponential (0 -20 min) and logistic (20 -60 min) fitting of the data. There was considerable variation in the amount of 32 P incorporated among different animals and among different experiments following the initial ϳ10 min after the flash. Such variation is also evident in published profiles of rhodopsin phosphorylation in intact frog retinas by others (e.g. Fig. 1 in Ref. 44). B, different plot of the same data. Here, the amount of incorporated 32 P (per mg rhodopsin) in each dopamine-treated retina is expressed relative to that in the control retina from the same animal; the rhodopsin phosphorylation level in each control retina was normalized to 100%. The superimposed histograms indicate the mean of the relative 32 Regulation of Rhodopsin by Dopamine 7182 result was found irrespective of whether 80 or 6% of the rhodopsin was photoexcited by the flash (Fig. 2). To test whether or not the effect of dopamine resulted in a general effect on ROS protein phosphorylation, we carried out two tests. First, the amount of [ 32 P]ATP was measured in ROSs following incubation of retinas for 45 min in the presence or the absence of dopamine. PhosphorImager analysis of two-dimensional TLC plates showed that in dopamine-treated retinas the amount of ROS [ 32 P]ATP was similar to that in control retinas (102 Ϯ 7%; n ϭ 12; p ϭ 0.94). Second, we observed that the radioactivity of minor phosphoproteins was unaffected by dopamine (Fig. 3). These results are consistent with dopamine having a specific effect on the phosphorylation state of rhodopsin. Further analysis of the reduction in the level of rhodopsin phosphorylation 45 min after the flash showed that it was effected by nanomolar concentrations of exogenous dopamine (Fig. 4). These concentrations are in the range of reported dissociation constants (K d ) for dopamine receptors in the high affinity state (32). Dopamine receptors fall into two general classes, D1-like and D2-like. D1 receptors act by activating adenylate cyclase. D2 receptors typically act by inhibiting adenylate cyclase (32)(33)(34). To test which class of receptor might be involved in mediating the dopamine effect on rhodopsin phosphorylation, we tested whether antagonists selective for D1-like or D2-like receptors would counter the lowered phosphorylation level found 45 min after the light flash. As illustrated in Fig. 5, SCH-23390, a selective D1 antagonist, did not interfere with the dopamine effect. However, spiperone, a selective D2 antagonist, did; it resulted in a higher level of phosphorylation. This finding is consistent with previous reports identifying D2-like receptors on rod photoreceptors (20,(35)(36)(37). These results indicate that exposure to dopamine and activation of D2 receptors on photoreceptor cells alters the kinetics of rhodopsin dephosphorylation. One explanation is that dopamine leads to activation of rhodopsin phosphatases. Alternatively, dopamine could lead to preferential phosphorylation at a site that is dephosphorylated more rapidly. Protein kinase C phosphorylates a domain that is not a primary phosphorylation site for rhodopsin kinase (38,39), and stimulation of protein kinase C phosphorylation of rhodopsin results in faster dephosphorylation (27). However, altering the relative activities of protein kinase C and rhodopsin kinase results in a different rate of phosphorylation (27), which was not evident in dopamine-treated samples (Fig. 1). In experiments to test whether ROSs from dopamine-treated retinas contained greater phosphatase activity, purified ROS membranes containing 32 P-phosphorylated rhodopsin were incubated with ROS cytosol from control or dopamine-treated retinas (7,31). Fig. 6 illustrates that dopamine-treated ROS cytosol contained significantly more rhodopsin phosphatase activity. Previous work has shown that dopamine may regulate phosphatase-1 via D1 receptors and cyclic AMP-dependent kinase phosphorylation of DARPP-32 (dopamine and cAMP-regulated phosphoprotein), a phosphatase-1 inhibitor (40). However, this is the first report suggesting an effect of dopamine on other phosphatases and on phosphatase activity via D2 receptors. Dephosphorylation of GPCRs has received less attention . The data shown (mean Ϯ S.E.) represent the amount of rhodopsin phosphorylation relative to that in control retinas (no exogenous dopamine), which was normalized to 100%. In each experiment, one retina was used for each concentration, and the experiment was repeated six times. The curve is a result of nonlinear regression analysis. than their phosphorylation. In comparison to our knowledge of kinases that phosphorylate GPCRs, less is known about the phosphatases that dephosphorylate them, and, in particular, how these phosphatases are regulated. However, dephosphorylation of rhodopsin is necessary to complete the rhodopsin cycle following light activation and then deactivation by phos-phorylation and arrestin binding (41)(42)(43). The importance of rhodopsin dephosphorylation is emphasized by Drosophila rdgC mutants. In the absence of rhodopsin phosphatase (the product of the rdgC gene), the phosphorylation state of rhodopsin is abnormally high, termination of the light response is defective, and the photoreceptor cells degenerate (8). The present results demonstrate a role for dopamine in the regulation of rhodopsin dephosphorylation and indeed suggest that it effects stimulation of rhodopsin phosphatase. Because dopamine is normally released by cells in the inner retina, these results identify the potential for a novel means of regulation of rhodopsin: from the inner retina back to the light receptor. FIG. 5. Identification of the dopamine receptor subtype responsible for mediating the effect of dopamine on rhodopsin dephosphorylation. Retinas were incubated for 45 min after the flash of light. They were incubated with 1 nM dopamine Ϯ SCH-23390 (100 nM), a selective D1-like receptor antagonist, or Ϯ spiperone (100 nM), a selective D2-like receptor antagonist. One retina from each frog was treated with dopamine, and the other was treated with dopamine plus antagonist. Spiperone but not SCH-23390 inhibited the effect of dopamine; in these retinas the level of rhodopsin phosphorylation was higher. Each value shown is the mean Ϯ S.E. of seven retinas from four separate experiments, expressed as a percentage of the rhodopsin phosphorylation level in control retinas (no exogenous dopamine, no antagonists) at the same time. FIG. 6. The effect of dopamine on rhodopsin phosphatase activity in rod photoreceptor outer segments. Intact frog retinas were treated with dopamine (1 nM) or spiperone (100 nM, control) and incubated for 45 min after the flash. The cytosolic fraction from purified rod outer segments of these retinas was then used to dephosphorylate frog rhodopsin that had been previously 32 P-phosphorylated in intact retinas. Phosphatase activity was determined by the release of 32 P from rhodopsin (7,31).	v3-fos-license
2020-03-05T10:58:20.015Z	2020-02-28T00:00:00.000	213223930	{ "extfieldsofstudy": [ "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://doi.org/10.18231/j.ijca.2020.035", "pdf_hash": "0477f35b1e4eb620c1ada793829193e90823861f", "pdf_src": "MergedPDFExtraction", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45475", "s2fieldsofstudy": [ "Medicine" ], "sha1": "26055adfc6f2333c9b867ec1d8fc95e88a988e68", "year": 2020 }	pes2o/s2orc	Still’s disease and anaesthetic concerns: A case report Adult-onset Still ’s disease (AOSD), is a chronic systemic inflammatory 1 disorder rarely encountered in clinical practice, described by Sir George Frederick Still in 1897. AOSD is of unknown aetiology with the incidence estimated to be 0.16 per 100,000 persons with articular and extra-articular or systemic manifestations. AOSD has a more acute course than compared to rheumatoid arthritis in adults, often affecting many parts of the body before settling in the various joints. Its diagnosis is made by exclusion. Securing airway mainly tracheal intubation may be difficult due to involvement of cervical spine, temporomandibular joint and laryngeal involvement (crico-arytenoid arthritis). In addition, intermittent disease flare-ups with laryngeal involvement may cause delayed extubation. This case highlights the anaesthetic concerns involved with the Still’s disease. © 2020 Published by Innovative Publication. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by/4.0/) Case Report A 33 -year-old averagely built female patient presented to emergency room of obstetrics and gynaecology department of a tertiary care centre for pain in abdomen and bleeding per vaginum for last 5-6 hours. A quick sonography of abdomen showed ruptured ectopic pregnancy with mild free fluid in the abdomen. Seeing the condition of patient, an emergency laparotomy was planned. Preoperatively on history, she had amenorrhea since 8 weeks with pain abdomen of increasing intensity for last few hours and bleeding per vaginum. She also gave history of joint pain along with swollen painful joints with limited movement in both hands since many years. She also had a history of treatment for pott's spine 3-4 years back. While screening her past medical records, we found that she was diagnosed with AOSD at 26 years of age and was on methotrexate, steroid therapy and NSAIDS for one and half years after which patient herself defaulted the treatment and presently not on any medication for AOSD since last 5 years. General physical examination revealed hyper-pigmented rash over the both malar prominences. Her vitals recorded * Corresponding author. E-mail address: lalit.doc@gmail.com (L. Gupta). as pulse 110/minute, regular with BP of 90/54 mm-Hg and BMI of 24.5 kg/m2. Auscultation of heart and lungs was unremarkable. ECG was normal. Airway examination revealed mouth opening of 3.0 cm, Mallampatti class 3 and restricted neck extension and flexion. Haemoglobin-7.2gm%, and Platelet count-1.2 lakh/mm 3 were only blood reports available. LFT, K FT and coagulation profile could not be done because of urgency of surgery. Instruction to arrange adequate blood products was given. Anaesthesia Conduct General anaesthesia was planned taking into consideration the urgency of operation, severe anaemia and history of pott's spine. Well-informed written high risk consent was taken in view of difficult airway and associated AOSD. After shifting the patient in the operation theatre, all standard ASA monitors were attached (ECG, NIBP, pulse oximeter) and baseline parameters noted (HR=110beats/min, BP=90/ 54mm of Hg, MAP=63 mm of Hg Fiberoptic was not available with us in emergency. ENT surgeon was also called for stand by in OT during induction of general anaesthesia. Inotropic infusions along with compatible blood and blood products were kept ready. Patient was given ranitidine 50mg and ondansetron 4mg intravenously. After preoxygenation with 100% O 2 for 3 minutes, rapid sequence induction was done with thiopentone 5mg/kg i.v and succinylcholine 1.5mg/kg i.v. Laryngoscopy was done with the McGrath video laryngoscope in view of restricted neck movement but the view was very hazy, so direct laryngoscopy with Macintosh blade #3was attempted and the Cormack Lehane grade 3b laryngeal view confirmed. Intubating bougie was inserted across the vocal cords and size 7.0 mm ID cuffed oral PVC endotracheal tube was rail roaded over the bougie, thereafter bougie removed and bilateral air entry confirmed. Airway management was completely atraumatic. Patient attached to mechanical ventilator on volume control mod e with tidal volume of 8 ml/kg. Anaesthesia was maintained with O 2 :N 2 O mixture 50:50 with Isoflurane up to 1 MAC with boluses of Atracurium. Dexamethasone IV 4 mg was given for suspected airway oedema and Fentanyl 2µg/kg body weight IV was given to the patient for analgesia. Intraoperative blood loss was 900 ml, which was replaced with 2 units of packed cell volume and 1.0 litre crystalloid. Urine output intraoperatively was 150 ml. Paracetamol infusion 1 gm IV given for multimodal analgesia 30 minutes before the expected extubation and bilateral TAP block with 20ml 0.25% bupivacaine was given on each side at the end of surgery. Intraoperative course remained uneventful. At the end of surgery, pulse rate was 92/min and BP was 94/56 (MAP 64 mm Hg). ETT cuff deflation demonstrated peritubal air leak signifying that no laryngeal oedema was present. Patient was extubated when fully awake, breathing spontaneously and shifted to post-anesthesia care unit. Postoperative course remained stable. Discussion AOSD is a rare chronic systemic inflammatory disorder, affecting mainly patients between 16 and 35 years of age. Etiopathogenesis remained unknown; however, a genetic component with HLA antigens involvement has been proposed. Adult Onset Still 's Disease Diagnostic Criteria 2,3 helps in reaching a diagnosis for further management. Abnormalities predisposing to a difficult airway include TMJ ankylosis, cervical spine or atlantoaxial joint involvement, and cricoarytenoid arthritis. Acute cricoarytenoiditis flares resulting in marked arytenoids swelling, narrowed glottic aperture, and upper airway obstruction 4 causing symptoms of sore throat, hoarseness, odynophagia or occasional stridor, have been well described. The disease's clinical course has three patterns: (1) self-limiting with remission within a year, (2) intermittent with recurrent disease flare-ups and complete remission, and (3) chronic articular pattern with persistent active disease. AOSD is a diagnosis of exclusion with nonspecific lab results such as significant Leucocytosis, markedly elevated ESR & CRP, with the absence of positive ANA and rheumatoid factor. Treatment with NSAIDs, corticosteroids, and/or methotrexate is the mainstay therapy. 5 Disease-modifying anti-rheumatic drugs (DMARDs) have been used with mixed results. Lately, use of TNF-inhibitor and IL receptor antagonist has proved to be effective. Symptomatic patients with laryngeal edema should initially be treated with racemic epinephrine nebulizations, humidified O 2 , and systemic corticosteroids. If regional anaesthesia is not a reasonable option, then fiberoptic intubation with a small ETT is recommended provided there is no urgency of surgery. Surgical intervention with tracheostomy, arytenoidectomy or, arytenoidopexy may be necessary if the problem persists despite medical treatment. Conclusion Airway management in the presence of an arthritic triad involving cervical spine, temporomandibular joints, and larynx may challenge the expertise of even the most experienced anaesthesiologist. Varying degrees of laryngeal obstruction due to cricoarytenoid arthritis is a well-known but uncommon complication of rheumatologic disorders, and anaesthesiologists should be fully aware of this problem. A thorough preoperative airway assessment, preparedness for potential problems and alternative plans are essential for the successful and safe management of these patients. In our case, anaesthesia team was well prepared for a difficult airway; but patient was intubated orally without any significant problem and had an uneventful anaesthetic course. Postoperatively, laryngeal reflexes must return before extubation, and oversedation must be avoided in order that the poor airway found in these patients is at best preserved. Conflict of interest None. Source of funding None.	v3-fos-license
2018-04-03T05:55:04.429Z	2004-01-23T00:00:00.000	20854966	{ "extfieldsofstudy": [ "Medicine", "Biology" ], "oa_license": "CCBY", "oa_status": "HYBRID", "oa_url": "http://www.jbc.org/content/279/4/2993.full.pdf", "pdf_hash": "55a1ab5fb25c653342276c73c0e6ac1e8594c13b", "pdf_src": "Highwire", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45476", "s2fieldsofstudy": [ "Biology" ], "sha1": "774b9dc2967d0db4b3def60e17fe40a1d80f1eb7", "year": 2004 }	pes2o/s2orc	Erythroblast Transformation by FLI-1 Depends upon Its Specific DNA Binding and Transcriptional Activation Properties* FLI-1 is a transcriptional regulator of the ETS family of proteins. Insertional activation at the FLI-1 locus is an early event in F-murine leukemia virus-induced erythroleukemia. Consistent with its essential role in erythroid transformation, enforced expression of FLI-1 in primary erythroblasts strongly impairs the response of these cells to erythropoietin (Epo), a cytokine essential to erythropoiesis. We show here that point mutations in the ETS domain that abolished FLI-1 binding to specific DNA elements (ETS-binding sites) suppressed the ability of FLI-1 to transform erythroblasts. The exchange of the entire ETS domain (DNA binding domain) of FLI-1 for that of PU.1 changed the DNA binding specificity of FLI-1 for that of PU.1 and impaired FLI-1 transforming properties. In contrast, ETS domain swapping mutants that maintained the DNA binding specificity of FLI-1 did not affect the ability of FLI-1 to transform erythroblasts. Deletion and swapping mutants that failed to inhibit the DNA binding activity of FLI-1 but impaired its transcriptional activation properties were also transformation-defective. Taken together, these results show that both the ability of FLI-1 to inhibit Epo-induced differentiation of erythroblasts and to confer enhanced cell survival in the absence of Epo critically depend upon FLI-1 ETS-binding site-dependent transcriptional activation properties. FLI-1 is a member of the ETS family of transcriptional regulators, which plays an essential role in development and oncogenesis (for review see Ref. 1). FLI-1 shares with other ETS proteins a conserved ETS domain that is responsible for its targeting to the nucleus and specific binding to 10-bp-long DNA sequences centered over a GGA core (ETS-binding sites (EBSs) 1 ) (2). When bound to specific EBSs in the context of cellular, viral, or model promoters/enhancers, FLI-1 most often activates transcription, a property that relies on two activation domains localized on the C-terminal and N-terminal sides of the ETS domain (3)(4)(5). However, in the context of specific promoters, FLI-1 binding results in transcriptional repression through ill defined mechanisms (6,7). In addition to its transcriptional regulatory properties resulting from its tethering to DNA, FLI-1 has been shown to modulate in trans the activity of other unrelated transcriptional regulators through protein-protein interactions (8,9). The human FLI-1 and the highly related ERG genes are rearranged as the result of specific chromosomal translocations in Ͼ95% of the cases of Ewing sarcoma, a pediatric tumor of neuroectodermal origin (10,11). In this disease, the 3Ј-part of FLI-1 or ERG is translocated to the 5Ј-half of EWS, a member of the TET gene family of RNA-binding proteins (12). The resulting fusion gene encodes an EWS-FLI-1 or EWS-ERG fusion protein in which the N-terminal activation domain of FLI-1/ERG is replaced by the potent N-terminal activation domain of EWS, thereby generating an altered regulator of EBS-driven transcription (13)(14)(15)(16). EWS-FLI-1 transforms NIH3T3 cells to anchorage-independent growth and accelerates the tumorigenic potential of these cells following their transplantation in nude mice (14,17). The molecular events involved in EWS-FLI-1 transforming properties remain to be characterized but appear to rely on both DNA binding-dependent and DNA binding-independent properties (18,19). FLI-1 was originally identified as a common proviral integration site in erythroleukemia induced in the newborn mouse by the F-murine leukemia virus component of the Friend virus complex (for review see Refs. 1 and 20). The other component of the Friend virus complex, SFFV, induces erythroleukemia in adult mouse. In this case, the hallmark of the disease is the proviral insertional activation of Spi-1/PU.1, another member of the ETS gene family (21). The early phase of F-murine leukemia virus-induced erythroleukemia is characterized by the expansion of erythroblasts in the spleen of infected animals and by severe anemia. The emergence of proliferating erythroblasts is concomitant with the rearrangement of the FLI-1 locus and the activation of FLI-1 expression (20,22). At that stage, proliferating cells are not immortalized, and additional genetic events are required, including the loss of the p53 tumor suppressor gene, to bypass senescence and induce immortalization (22,23). Consistent with its central role in F-murine leukemia virus-induced erythroleukemia, enforced expression of FLI-1 in a mouse erythroleukemic cell line (7) and primary avian erythroblasts (24) has been shown to strongly interfere with the normal response of these cells to erythropoietin (Epo), a cytokine essential to erythropoiesis (25,26). The mechanisms underlying the transforming properties of FLI-1 in erythroblasts are mostly not understood. They have been proposed to involve FLI-1 interference with molecular events important for erythroid differentiation (7,8) and FLI-1-directed activation of novel genes not normally expressed in erythroblasts (27). We report here the analysis of the DNA binding activity, transcriptional regulatory properties, and erythroblast transforming ability of a series of FLI-1 mutants. Our results show that the transforming properties of FLI-1 in primary erythroblasts are critically dependent upon its binding activity to specific EBSs. In addition, both the survival-inducing and differentiation-inhibiting properties of FLI-1 require a transcriptionally active protein, indicating that they involve the transcriptional activation of specific genes. Cell Culture, Retroviral-mediated Gene Transfer, and Differentiation Assays-HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Invitrogen), 1 mM glutamine, penicillin, and streptomycin. Chicken embryo fibroblasts were grown in the same medium supplemented with 2% chicken serum (Sigma). Infectious avian retroviruses were generated by co-transfection of primary chicken embryo fibroblasts with the replication-competent pRCAS-mEpoR and the different pSFCV retroviral vector derivatives and infected cells selected by their resistance to G418 (Invitrogen). To generate primary erythroblasts, chicken bone marrow cells were doubly infected with the S13 virus, encoding the temperature-sensitive ts-v-Sea protein kinase (34) together with the different SFCV/R-CAS-EpoR viral stocks as described previously (28). Infected bone marrow cells were seeded in CFU-E-methocel supplemented with 100 ng/ml recombinant chicken SCF, 1.4 nM insulin (Novo-Nordik), and 2.7 mg/ml G418. Under these conditions, only G418-resistant and transduced erythroblasts form colonies within 5-6 days at 37°C. Erythroblast colonies were picked and expanded in CFU-E medium containing SCF and insulin. Differentiation analyses were performed as described previously (28). To check for expression of the expected exogenous proteins, Western blots were performed using anti-EpoR (Santa Cruz Biotechnology, sc-697), anti-FLI-1 (Santa Cruz Biotechnology, sc-356), anti-HA (Santa Cruz Biotechnology, sc-805), anti-ERK2 (Santa Cruz Biotechnology, sc-154), and an anti-pan-ETS monoclonal antibody (kindly provided by N. K. Bhat, Frederick, MD). Transient Tranfections, Luciferase Assays, and Immunofluorescence Analyses-For transactivation experiments, 2.5 ϫ 10 5 HeLa cells were plated in 6-well plates and transfected 24 h later with the indicated amount of plasmid DNA using the Lipofectamine Plus reagent (Invitrogen) in the absence of serum, as recommended by the manufacturer. The DNA mixture included the indicated amounts of reporter gene constructs and expression plasmids. The total amount of expression plasmid was kept constant by the addition of empty ⌬EB vector, and the total amount of DNA was kept constant to 1 g by the addition of carrier plasmid DNA. Cell lysates were prepared 24 h after transfection and assayed for luciferase activity by using the luciferase assay system kit (Promega). The results shown are the mean of at least four independent transfections experiments. Immunofluorescence analyses were performed using HeLa cells transfected with 5 g of the indicated expression plasmids. Two days after transfection, cells were fixed with 4% paraformaldehyde and permeabilized by incubation in 0.3% Triton X-100 10 min. After incubation for 1 h with a 1:200 dilution in PBS, 10% fetal calf serum of anti-HA antibody, cells were washed in PBS and incubated with a 1:100 dilution in PBS, 10% fetal calf serum of a fluorescein isothiocyanate-conjugated goat anti-rabbit immunoglobulin (Amersham Biosciences) and nuclei were stained by Hoechst. After several PBS washes and dehydration, coverslips were mounted in Mowiol. Fluorescence was visualized with an epifluorescence microscope. RESULTS FLI-1 can regulate transcription either following its binding to specific DNA elements present in the enhancer/promoter regions of responsive genes or independently of DNA binding, through interactions in trans with other transcriptional regulators. To investigate the relative importance of these mecha-nisms to the transforming properties of FLI-1, a series of FLI-1 mutants were generated and characterized for their DNA binding, transcriptional regulatory properties, and transforming properties in erythroblasts. NMR studies have shown that the DNA binding domain (ETS domain) of FLI-1 is composed of three ␣-helices (␣1-␣3) and a four-stranded ␤-sheet (strands ␤1-␤4) (36); see Fig. 1). Helices ␣2 and ␣3 together with intervening residues form a helix-turn-helix motif that lies over the surface formed by the four-stranded ␤-sheet. Crystallographic studies have shown that residues from both the helix-turn-helix and ␤3-␤4 motifs make direct contact with DNA and determine to a major extent the DNA binding specificity of ETS proteins (37,38). Three point mutants in the ETS domain were constructed (see Fig. 1 for a schematic description of the mutants). In FLI-1[R337,340L], the arginine residues in helix ␣3 that make direct base contacts with the guanine residues of the EBS GGA core were mutated to leucine. This mutation has been shown previously to abolish the specific DNA binding activity of the isolated ETS domain of FLI-1 (36) and of EWS-FLI-1 (13). In FLI-1[I347E], an isoleucine residue at the end of helix ␣3 was changed for a glutamic acid residue, a mutation that inhibits the specific binding to DNA of EWS-FLI-1 (18). We also constructed FLI-1[D344V] in which an aspartate residue immediately C-terminal to the ␣3 recognition helix was mutated into valine, a mutation that has been reported to only minimally affect the DNA binding activity of ETS proteins but broadens their DNA binding specificity (39). The DNA binding activity of these mutants was analyzed by electrophoretic mobility shift assay using in vitro translated proteins ( Fig. 2A, lanes 2 and 7-10) and a 32 P-labeled doublestranded DNA probe corresponding to an optimized FLI-1binding site (probe A, see "Experimental Procedures"). As expected, both FLI-1[R337,340L] and FLI-1[I347E] were defective for their specific binding to DNA, whereas FLI-1[D344V] bound DNA similarly as wtFLI-1 (Fig. 2B, compare lanes 2 and 3; Fig. 2B, compare lanes 5 and 9 -10 1). The ETS domain of FLI-1 is closely related to that of ETS-1 (67% identity) but only distantly related to that of PU.1 (38% identity). Consistent with this, FLI-1 and ETS-1 display a similar DNA binding specificity in vitro (2,40,41), whereas PU.1 binds to EBSs (Pu boxes) with distinct base composition both at the 5Ј-and 3Ј-sides of the GGA core (42). Both FLI-1[EEE] and FLI-1 [PPP] were expressed at similar levels as wtFLI-1 following in vitro transcription/translation in reticulocyte lysates of the corresponding DNA templates ( Fig. 2A, lanes 2-4). The DNA binding activity and specificity of these mutants were compared with that of wtFLI-1 by electrophoretic mobility shift analyses using two additional EBS probes, besides probe A. Probe T is identical to probe A except for the transversion of the last adenine of the GGAA core sequence for a thymine, whereas the Pu probe is a high affinity binding site for PU.1 (43). In vitro translated PU.1 was used as specificity control in these analyses. wtFLI-1 bound probes A and T but failed to bind the Pu probe (lane 2 in each panel of Fig. 2D) whereas PU.1 more efficiently bound the Pu probe than probe A and failed to bind the T probe (Fig. 2E). Similarly to wtFLI-1, FLI-1[EEE] bound probes A and T but failed to bind the Pu probe (lanes 4 in Fig. 2D). In contrast, FLI-1[PPP] displayed higher binding activity toward the Pu probe than to probe A and failed to bind probe T, thereby adopting the DNA binding specificity of PU.1 (lanes 3 in Fig. 2D). This was confirmed in competitive EMSA in which the binding of FLI-1[PPP] to probe A was competed in a dose-dependent manner by the addition of an increasing molar excess of unlabeled oligonucleotides Pu and A but was barely competed by the same excess of oligonucleotide T (Fig. 3A, compare lanes 2-11 in FLI-1(PPP) panel). The same competition pattern was observed for PU.1 (Fig. 3A, lanes 2-11 in PU.1 panel). In contrast, binding of both wtFLI-1 and FLI-1[EEE] to probe A was competed in a dose-dependent manner by unlabeled oligonucleotide A and T but not by the Pu oligonucleotide (Fig. 3A, lanes 2-11 in the wt-FLI-1 and FLI-1(EEE) panels). As expected (Fig. 3A, (lanes 2-12). The binding reaction mixtures included either no competitor oligonucleotide (lanes 1 and 2) or a 10- fold (lanes 3, 6, and 9), 30-fold (lanes 4, 7, and 10), or 90-fold ( lanes 5, 8, 11, and 12) molar excess of unlabeled competitor oligonucleotides as indicated at the top. B, EMSA was carried out with 200 fmol of oligonucleotide T as a probe and either control reticulocyte lysate (lane 1), in vitro expressed wtFLI-1, or the indicated FLI-1 derivatives (lanes 2-9). The binding reaction mixtures included either no competitor oligonucleotide (lanes 1 and 2) or a 10-fold (lanes 3 and 6), 30-fold (lanes 4 and 7), or 90-fold (lanes 5, 8, and 9) molar excess of the indicated unlabeled competitor oligonucleotide. lane 12), none of the protein-DNA complexes were competed by a mutant Am competitor that is identical to oligonucleotide A except for GG-to-CC transversion in the GGA core, a mutation known to abolish specific binding of ETS proteins to DNA. These experiments show that substitution of the ETS domain of FLI-1 by that of ETS-1 generates a protein with a DNA binding specificity similar to that of FLI-1, whereas its substitution for the ETS domain of PU.1 changes the DNA binding specificity of FLI-1 for that of PU.1. The three-dimensional fold of several ETS domain-DNA complexes, including those of FLI-1 and PU.1, are highly superposable (38). Additional swapping mutants were therefore generated in which specific structural elements of the ETS domain of FLI-1 were replaced by the corresponding sub-domain of PU.1. Specifically, FLI-1 mutants PFF and FFP were generated by swapping either FLI-1 ␣1␤1␤2 or ␤3␤4 motifs for the corresponding domains of PU.1 (Fig. 1). Both mutants were stably expressed following in vitro translation in reticulocyte lysate ( Fig. 2A, lanes 5 and 6). Substitution of the ␣1␤1␤2 motif of FLI-1 by the corresponding motifs of PU.1 resulted in a protein with similar DNA binding activity as wtFLI-1, as evidenced by the efficient binding of FLI-1[PFF] to A (Fig. 2B) and T probes (Fig. 2C). Of note, FLI-1[PFF] displayed broadened DNA binding specificity because, unlike wtFLI-1, it was able to bind the Pu probe (Fig. 2F). Competitive EMSA showed that FLI-1[PFF] bound the A and T probes with comparable efficiency but only bound the Pu probe with low affinity because the FLI-1[PFF]probe A complex was inefficiently displaced in the presence of a large molar excess of unlabeled Pu oligonucleotide used as competitor (Fig. 3A, lanes 2-11 in the FLI-1(PFF) panel). Bind-ing of FLI-1[PFF] to the Pu probe is therefore at least an order of magnitude lower as compared with its binding to conventional EBSs. Substitution of the ␤3␤4 motif of FLI-1 by the corresponding region of PU.1 in FLI-1[FFP] resulted in a protein with slightly higher DNA binding activity toward the A and T probes as compared with wtFLI-1 (Fig. 2, B and C) and broadened DNA binding specificity as evidenced by its ability to bind the Pu probe (Fig. 2F). However, FLI-1[FFP] only inefficiently bound to the Pu probe as analyzed by competitive EMSA (Fig. 3A, lanes 2-11, in FLI-1(FFP) FLI-1 mutants were next analyzed for their transcriptional regulatory properties in transient co-transfection assays as described previously (13). Expression plasmids for the protein under study and an EBS-driven luciferase gene reporter construct were co-transfected in HeLa cells, and transactivation was monitored by luciferase assays. Importantly, all FLI-1 mutants analyzed were found to be expressed at levels similar to wtFLI-1 in HeLa cells (Fig. 4A) and to localize to the nucleus of transfected cells (Fig. 4B). We first used the tkD2A-Luc reporter in which a duplicate copy of the ETS-responsive re- ). B, HeLa cells transfected with either control expression plasmid or expression plasmid for HA-wtFLI-1 or the indicated FLI-1 mutants were seeded to collagen-treated coverslips, fixed in 4% paraformaldehyde, permeabilized with 0.3% Triton X-100, and incubated with either anti-HA or anti-FLI-1 antibodies as indicated, followed by a fluorescein isothiocyanate-conjugated rabbit immunoglobulin G. The inset corresponds to Hoechst staining of the same field. gion-1 of the human T-cell leukemia virus type I-long terminal repeat is inserted upstream of the Ϫ55 herpes simplex virus thymidine kinase (tk) promoter. Our previous analyses have shown that transcriptional activation of this promoter by FLI-1 depends upon FLI-1 binding to two specific EBSs in ERR1 (3). Consistent with these results, wtFLI-1 activated luciferase expression from tkD2A-Luc in a dose-dependent manner (Fig. 5A) whereas FLI-1 mutants defective for their specific binding to DNA (FLI-1[R337,340L]; FLI-1[I347E]) were completely inactive (Fig. 5A). Mutant FLI-1[D344V], which binds DNA with the same activity as wtFLI-1, efficiently transactivated tkD2A-Luc (Fig. 5A). The transcriptional activation properties of FLI-1 have been reported to depend upon two activation domains that localize N-terminally (NTAD) and C-terminally (CTAD) with respect to the ETS domain, respectively (5). Both FLI-1[EEE] and FLI-1[FFP] transactivated tkD2A-Luc in a dose-dependent fashion, albeit with reduced efficiency as compared with wtFLI-1 (Fig. 5C). The reduced activity of these mutants correlated with their reduced binding to ERR1 as compared with wtFLI-1. Indeed, competitive EMSA showed that specific DNA binding of wtFLI-1, FLI-1[EEE], and FLI-1 [FFP] was similarly competed by increasing amounts of unlabeled oligonucleotide T (Fig. 3B). In contrast, when ERR1 was used as competitor, it inhibited more efficiently the specific binding of wtFLI-1 than that of either FLI-1[EEE] or FLI-1[FFP] (Fig. 3B). In line with their very low binding activity to ERR1 (data not shown), PU.1 and FLI-1[PPP] failed to transactivate the tkD2A-Luc reporter (Fig. 5C). However, both proteins efficiently transactivated PU 3 -tk81-Luc, a reporter plasmid carrying three copies of a Pu box (see "Experimental Procedures") inserted 5Ј of the Ϫ81 herpes simplex virus tk promoter (Fig. 5E). As expected, none of these proteins were able to transactivate the control tk81-Luc reporter lacking Pu boxes (data not shown). To analyze the transforming activity of FLI-1 mutants and compare them to wtFLI-1, the corresponding cDNAs were introduced by retrovirally mediated gene transfer into primary avian erythroblasts, and the effects of these proteins on Epodependent differentiation and survival were analyzed as described previously (28) (see "Experimental Procedures"). A dozen clones expressing both mEpoR and the respective FLI-1 proteins at similar levels (Fig. 6A) were grown for further study. The results obtained with one representative clone of each combination are reported. In accordance with previous studies (28), mEpoR control erythroblasts differentiated terminally in response to hEpo as evidenced by their exit from the cell cycle 2 days after differentiation induction (Fig. 6B, Co panel), their accumulation over time of high levels hemoglobin (Fig. 6C, Co panel), and their acquisition of the morphology of fully mature erythrocytes (Fig. 7, Co panel, ϩhEpo). In the absence of hEpo, control erythroblasts rapidly die by apoptosis (28) as evidenced by the presence of only cell debris in cultures maintained for 3 days in the absence of hEpo (Fig. 7, Co panel, ϪhEpo). As reported previously (24), expression of wtFLI-1 inhibited Epo-induced differentiation of mEpoR erythroblasts as evidenced by their lack of hemoglobin accumulation (Fig. 6C, wtFLI-1) and their maintenance of an immature morphology (Fig. 7, wtFLI-1). Rather, mEpoR/wtFLI-1 erythroblasts self-renewed for several generations as differentiation-arrested cells under these conditions (Fig. 6B). Consistent with our previous analyses (24,27), cell survival in mEpoR/wtFLI-1 erythroblasts was enhanced as compared with control mEpoR erythroblasts when these cultures were maintained in the absence of hEpo (Fig. 7, compare Co and wtFLI-1 panels, ϪhEpo). FLI-1 [PPP] in which the DNA binding specificity of FLI-1 was switched for that of PU.1 was transformation-defective as mEpoR/FLI-1[PPP] erythroblasts terminally differentiated in response to hEpo and died in its absence (FLI-1(PPP) panels in Fig. 6, B and C, and Fig. 7). In contrast, FLI-1[EEE] and FLI-1[FFP], which displayed a DNA binding specificity similar to that of wtFLI-1, both inhibited hEpo-induced differentiation and induced enhanced survival in the absence of hEpo (Fig. 6, B and C, and Fig. 7). FLI-1[EEE] and FLI-1[FFP] are therefore indistinguishable from wtFLI-1 for their ability to transform erythroblasts. The transcriptionally defective FLI-1[PFF] mutant was unable to transform erythroblasts as mEpoR/FLI-1[PFF] erythroblasts were indistinguishable from control mEpoR erythroblasts in their ability to terminally differentiate in response to hEpo (Fig. 6, B and C, and Fig. 7) and to rapidly die by apoptosis in the absence of hEpo (Fig. 7). Of note, neither the NTAD nor the CTAD proved to be absolutely required in a non-redundant fashion to FLI-1 transforming properties because both mEpoR/FLI-1 and mEpoR/FLI-1[225-452] erythroblasts failed to differentiate in response to hEpo (Fig. 6, B and C, and Fig. 7) and showed prolonged survival in the absence of hEpo (Fig. 7). In contrast, deletion of both activation domains in FLI-1[276 -373] generated a transformation-defective protein (Figs. 6 and 7). Finally, we analyzed the transforming properties of VP16/FLI-1 [276 -373]. As shown in Fig. 7, mEpoR/VP16/FLI-1[276 -373] erythroblasts were only partially affected in their response to hEpo because these cells showed prolonged survival in the absence of hEpo when compared to mEpoR control erythroblasts and underwent only partial morphological differentiation in response to hEpo (Fig. 7). In line with this, VP16/FLI-1[276 -373] erythroblasts accumulated intermediate levels of hemoglobin (Fig. 6D). Taken together, these results show that erythroblast transformation by FLI-1 requires both its specific DNA binding activity and transcriptional activation properties. To investigate whether the transcriptional regulatory properties of FLI-1 and its derived mutants as assessed in reporter gene assays would translate into gene deregulation in vivo, we analyzed the expression of several genes involved in FLI-1-induced erythroblast transformation. Our previous studies (27) have shown that BCL-2 is a direct target of FLI-1 in primary erythroblasts and that up-regulation of BCL-2 expression is involved in the ability of FLI-1 to induce erythroblast survival in the absence of hEpo. Two other genes were identified by differential screening to be strongly up-regulated in FLI-1-transformed erythroblasts: the first encodes SLAP, a SH2-and SH3-domain adaptor, and the second encodes inositol polyphosphate 5Ј-phosphatase ((45) and data not shown). mEpoR control erythroblasts, mEpoR/wtFLI-1, mEpoR/FLI-1[PPP], and mEpoR/FLI-1[PFF], were maintained in the absence of hEpo for 24 h, and expression of BCL-2, SLAP, and IN5PP5P was analyzed by RT-PCR. The results of Fig. 8 show that expression all three genes was up-regulated in wtFLI-1transformed erythroblasts as compared with control cells. In contrast, none of these genes are detectably expressed in erythroblasts expressing the transactivation-defective FLI-1[PFF] and in erythroblasts expressing the transformation-defective FLI-1[PPP]. DISCUSSION The early and recurrent activation of FLI-1 by retroviral insertional mutagenesis in F-murine leukemia virus-induced erythroleukemia suggests that up-regulation of FLI-1 expression is the initiating event in this multistep leukemia model. Consistent with this notion, enforced expression of FLI-1 in mouse erythroleukemic cell lines (see Ref. 7 and this study) and in primary avian erythroblasts (24) was found to inhibit Epoinduced differentiation and to promote cell survival in the absence of Epo. The molecular mechanisms involved in the ability of FLI-1 to modify the normal response of erythroblasts to Epo stimulation or withdrawal are only partially understood. The survival-inducing properties of FLI-1 in erythroblasts result in part from the activation of BCL-2 gene expression and subsequent induction of the anti-apoptotic BCL-2 protein (27,46). The ability of FLI-1 to inhibit erythroblast differentiation has been proposed to involve FLI-1-mediated repression of retinoblastoma gene transcription (7). This may result in inhibition of an intrinsic but non-cell autonomous retinoblastoma function required for terminal erythroid differentiation (47). Alternatively, a direct interference of FLI-1 with other transcriptional regulators important for erythroid differentiation such as retinoid/thyroid nuclear hormone receptors and EKLF has been proposed (8,9). The results of the present study show that the transforming properties of FLI-1 in erythroblasts critically depend upon both its specific binding to DNA and transcriptional activation prop-erties. First, point mutations in the ETS domain of FLI-1 that abolished specific DNA binding to FLI-1-responsive EBSs were found to suppress both the survival-inducing and differentiation inhibitory properties of FLI-1 in erythroblasts. Second, a mutant of FLI-1 retaining the DNA binding specificity of wt-FLI-1 but defective in its ability to activate EBS-driven transcription (FLI-1[PFF]) completely failed to transform erythroblasts. Third, substitution of the ETS domain of FLI-1 by that FIG. 6. Comparison of the transforming properties of wtFLI-1 and FLI-1 mutants in primary erythroblasts. Erythroblast clones expressing either EpoR (Co, control), EpoR ϩ wtFLI-1, or EpoR ϩ FLI-1 derivatives were generated by retrovirally mediated gene transfer as described under "Experimental Procedures." A, expression of exogenous proteins in representative erythroblast clones expressing EpoR (lanes 1 and 2), EpoR/HA-wtFLI-1 (lanes 3, 4, and 13 of the distantly related ETS protein PU.1 resulted in a FLI-1 protein (FLI-1[PPP]) with a DNA binding specificity skewed to that of PU.1 and to the loss of its ability to transform erythroblasts. In contrast, swapping mutants in the ETS domain that maintained the DNA binding specificity of FLI-1 (FLI-1[EEE]; FLI-1[FFP]) resulted in fully transforming proteins. Taken together, these results indicate that the survival-inducing properties of FLI-1 and its ability to inhibit erythroid differentiation both require the activation of specific target genes, the regulation of which is under the control of FLI-1-responsive EBS elements. We have shown recently that the BCL-2 gene is a direct FLI-1 target in erythroblasts and that BCL-2 up-regulation contributes to the enhanced survival of these cells in the absence of Epo (27). However, overexpression of BCL-2 (27), although favoring cell survival, does not interfere with terminal differentiation, suggesting that the block imposed by FLI-1 upon Epo-induced differentiation must result from the activation of additional genes. The search for genes differentially activated in FLI-1-transformed erythroblasts is underway. Evidence gathered so far indicates that FLI-1 induces the de novo up-regulation of several genes in erythroblasts that encode adaptors and effectors of signaling pathways, 2 suggesting that FLI-1 may inhibit terminal differentiation through its ability to specifically interfere with EpoR signaling output. Besides its role in nuclear targeting and DNA binding, the ETS domain also contributes to the specific interaction of ETS proteins with other transcriptional regulators. These interactions can take place independently of DNA binding when ETS proteins interact in trans with specific factors, often through the ␤3/␤4 motif of the ETS domain (48 -52). Other interactions result in the cooperative binding of ETS proteins with unrelated factors to composite cis-acting DNA-response elements. In this case, structural studies have shown that cooperativity results from limited interactions between amino acid residues of each partner and often involves the winged helix-turn-helix region of the ETS domain. These interactions induce a local modification in the conformation of one or both factors that results in a change in specific DNA contacts to provide additional energy for the complex to bind composite DNA elements (53)(54)(55)(56). It cannot be excluded that the difference in activity between wtFLI-1 and transformation-defective mutants like FLI- [PPP] involves, in addition to a change in DNA binding specificity, the disruption of a protein-protein interaction critical to the transforming properties of wtFLI-1. In this scenario, this interaction would be conserved in other mutants like FLI-1[EEE] and would not involve the ␤3/␤4 fold of FLI-1 because substitution of the ␤3/␤4 region of FLI-1 by that of PU.1 in FLI-1[FFP] generated a fully transforming protein. The requirement for specific DNA binding activity and deregulated EBS-driven transcription in order for FLI-1 to transform erythroblasts contrasts with results obtained for the EWS-FLI-1 fusion protein specific for Ewing sarcoma. EWS-FLI-1 is a potent activator of EBS-driven transcription (13), and part of its transforming properties is believed to derive from its ability to deregulate the expression of critical genes either directly (16,57) or indirectly (13, 17, 58 -60). As for wtFLI-1, introduction of the equivalent of the I347E and R337L,R340L mutations in the ETS domain of EWS-FLI-1 results in proteins that are also defective for their specific binding to DNA. However, in the case of EWS-FLI-1, these mutants retain substantial colony forming activity and essentially intact tumorigenic inducing potential (18,19). These results suggest that the transforming properties of EWS-FLI-1 also rely on DNA binding-independent mechanisms that could be related to the ability of EWS-FLI-1 to interfere with the activity of EWS as an adaptor to couple gene transcription to RNA splicing (31,61,62). The notion that the transforming properties of FLI-1 and EWS-FLI-1 involve distinct mechanisms is further substantiated by the fact that although deletion of the C-terminal domain of FLI-1 in FLI-1 does not impair its ability to transform erythroblasts, the same deletion in EWS-FLI-1 severely reduces its anchorage-inducing and tumorigenic properties (17). It should be stressed, however, that these notions mostly derive from the analysis of EWS-FLI-1 transforming properties in NIH3T3 fibroblasts, a cellular background distinct from the neuroectodermal origin of the cells presumably targeted by EWS-FLI-1. It remains to be seen FIG. 7. Comparison of the transforming properties of wtFLI-1 and FLI-1 mutants in erythroblasts, morphological analyses. A representative EpoR erythroblast clone (Co, control) and representative EpoR clones expressing either wtFLI-1 or the different FLI-1 mutants were maintained at 42°C in differentiation medium either in the presence (ϩhEpo) or absence of hEpo (ϪhEpo). Aliquots of cells were cytocentrifuged 3 days after differentiation induction and stained with neutral benzidine and Giemsa. Immature cells stain blue; hemoglobinized cells stain brown. FIG. 8. RT-PCR analysis of FLI-1 target genes in erythroblasts. Two clones of each mEpoR, mEpoR/wtFLI-1, mEpoR/FLI-1[PPP], and mEpoR/FLI-1[PFF] erythroblasts were maintained for 24 h at 42°C in the absence of hEpo; RNA were extracted and reverse-transcribed. Primers specific for the genes encoding BCL-2, SLAP, inositol polyphosphate 5Ј-phosphatase (IN5PP5P), and ribosomal protein S17 (Co, control) were used to amplify the corresponding sequences. Aliquots of 0.5 and 2 l of each RT products were used. PCR-amplified products were analyzed by agarose gel electrophoresis in the presence of ethidium bromide. whether the differences between the determinants involved in the ability of FLI-1 to transform erythroblasts and those involved in EWS-FLI-1 transforming properties will hold when a more appropriate cellular system for the latter becomes available.	v3-fos-license
2022-02-11T16:05:18.554Z	2022-01-01T00:00:00.000	246734980	{ "extfieldsofstudy": [], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.shs-conferences.org/articles/shsconf/pdf/2022/04/shsconf_eac-law2021_00085.pdf", "pdf_hash": "2e663edaee5676aea41c00a630a6061bd7c10e11", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45479", "s2fieldsofstudy": [ "Law" ], "sha1": "a4475d1d568d4dcb0c13f91295f5dbfc9933c1be", "year": 2022 }	pes2o/s2orc	On the question of some tendencies in the unification of national and international jurisdiction of Internet disputes The author analyzes the development of foreign legislative and law enforcement approaches to resolving the issue of Internet disputes jurisdiction. It is concluded that today the development of national and international legal regulation of issues Internet disputes jurisdiction occurs in two directions: firstly, the reception of national jurisdiction legal approaches in international legal acts; secondly, the implementation of international legal acts approaches into national legislation. The author draws attention to the fact that, in the absence of a universe international treaty regulating the issues of Internet disputes international jurisdiction, a detailed development of national law rules, regardless of the particular jurisdiction, taking into account the specifics of Internet relations and Internet disputes cannot solve the problem of jurisdiction conflict in resolving this category of disputes. Introduction The volume growth of online legal relations naturally leads to the growth of disputes arising from them. At the same time, the courts resolving such disputes, regardless of which state they extend their powers to, often face questions related to the interstate nature of the Internet and the technical specifics of its functioning. The development of rules for determining the competent authority for considering Internet disputes is supposed to be one of these issues. The complication of resolving this issue of determining the competent authority for the consideration of Internet disputes is based on a number of features of the Internet functioning and, consequently, Internet legal relations, including: 1) the use of special technical means by the subjects of the legal relations [1]; 2) "anonymization" of subjects of the legal relations, manifested in the possibility of participation in the legal relations without specifying identification features [2]; 3) the remoteness (even on an interstate scale) of the subjects from each other [3]; 4) transboundariness as a natural consequence of the presence of legal entities and third parties in different jurisdictions [4]. Before turning to a direct analytical presentation of the article matter, it is considered possible to note that for the purposes of this article, Internet disputes are understood as the disputes in which the fact of using the Internet affects the determination of the competent authority for the consideration of the dispute, the establishment of the parties to a case, the subject of evidence in the case and permissible means of proof. At the same time, the definition is not unified, since other definitions of Internet disputes can be found in the scientific literature [5]. The complication of legal relations arising on the Internet by a foreign element determines the intention of various countries to legally qualify the actions of their participants. Considering Internet relations specifics and the enlargement of the Internet relations segment in the social space, both at the level of individual states and at the international level, general and casual (i.e. relevant to certain types of Internet disputes (such as domain disputes, etc.)) rules for determining Internet disputes jurisdiction are being developed. Determining the jurisdiction of Internet disputes: the USA experience The United States is the center of concentration of the largest domain zones (.com;. org; .net). Due to it American law has become one of the first to establish special rules for determining jurisdiction of Internet disputes. The development of new rules was concerned with the need of division of the competence of the courts from different states. In determining jurisdiction in Internet disputes, American courts are guided by both traditional rules and criteria customized to the online space: personal jurisdiction, a "sliding scale" test and purposeful subordination. The personal jurisdiction rule is a general jurisdictional rule assumes that a state court has jurisdiction to consider a dispute against a person if he physically presents on its territory [6]. One of the first American cases in which judges paid attention to the extraterritorial nature of the Internet 1996. In that case the extraterritorial nature of the Internet was concerned with the possibility of bringing information to the attention of Network users in an unlimited volume and in an indefinite geography. Considering the dispute, the court came to the conclusion that it is necessary to apply a long-arm statute, which implies the extension of the right beyond the territorial jurisdiction of the court, to advertising posted on the Internet [7]. Further development of the personal jurisdiction rule in relation to Internet disputes is associated with the use of the 'minimum contacts' test in the cases of International Shoe Co v. Washington, 1945 and Maritz, Inc. v. Cybergold, Inc., 1996 [8]. In essence, this is a five-step approach, on the basis of which it is determined that it is possible to extend the jurisdiction of the court of the state in which the dispute is being considered to a non-resident defendant. "Minimum contacts" refer to the actions of the defendant that have such a level of contact with the forum state that makes him to reasonably expect a claim to be filed in this state. The "minimum contacts" test is aimed at determining a number of circumstances: 1) the nature and quality of the contacts with the forum state; 2) the quantity of the contacts; 3) the relation of the cause of action to those contacts; 4) the interest of the forum state in extending its jurisdiction to its residents; 5) the interest of the parties to the dispute [8]. The first three criteria are the main ones, while the last two are of a secondary nature. Another traditional test for determining personal jurisdiction, in addition to "minimum contacts", is "Calder test" (Calder v. Jones, 1984). According to this rule, transformed under the online world, the personal jurisdiction of the court can be extended to a nonresident, if he 1) committed intentional actions, 2) directly aimed at the forum state, 3) with the understanding that the main damage of the actions will be caused in the forum state [9]. The first special jurisdictional rule for Internet disputes was the "sliding scale" rule, or "Zippo test", in the case of Zippo Manufacturing Co. v. Zippo Dot Com, Inc., 1997. It reflects the influence of a website interactivity degree on the possibility of providing personal jurisdiction. When resolving this dispute, the court used an innovative approach. It proposed to determine the extension of the court's jurisdiction to the website by a three-step scale, each step of which is determined by the nature and quality of business activities carried out via the Internet. The court presented the next structure of the scale: -at the one end of the scale there are cases in which the defendant is using a website for doing business with a high degree of evidence, and it is so-called "active website". For example, if the defendant makes deals with residents of another state and they involve the exchange of computer files, then personal jurisdiction is appropriate. -at the opposite end there are cases where the defendant only posts on the Internet some information available to residents of another state. This type of website is supposed to be "passive". Even if the defendant acts more through a passive website than with a simple posting of information, is does not form a personal jurisdiction. -at the center of the scale there are situations with the defendant having an interactive website through which users can interact with the host computer. In such cases, the resolution of the jurisdiction issue depends on the degree of interactivity and commercialization of the information exchange via the website [10]. The development of the scale forced American courts to use the methodology of classifying websites into active and passive. If the passivity of the defendant's website is established, personal jurisdiction over the is abscent (ex. Bensusan Restaurant Corp. v. King, 1996, Cybersell, Inc. v. Cybersell, Inc.) Despite the breadth of applications, American legal society criticizes Zippo test. It has become a kind of "trigger" for the development of additional criteria for personal jurisdictionpurpose, or purposeful activities (Toys «R» Us [11], Inc. v. Step Two, Kindig v. Creative Controls [12]). Further development of the jurisdictional rules in the United States occurred in light of the eBay promotion. The question of whether the forum state jurisdiction can be established in cases where a non-resident defendant uses an intermediary website (eBay) available to residents of the forum state was firstly raised in the case of Boschetto v. Hansing [13]. In this case, the court, rejecting the claim due to the lack of personal jurisdiction, pointed out that the fact of selling only one car through the eBay website in the absence of any other actions cannot entail the extension of the personal jurisdiction of the forum state to a non-resident person. Thus in the case of Attaway v. Omega [14] court established personal jurisdiction not only through the "minimum contacts" test, but also on the basis of the defendant's actions in the form of appearing (even through an agent) in the forum state to receive goods purchased via the Internet. Despite the obvious variety of American approaches developed for determining jurisdiction in Internet disputes, courts rarely use only one criterion when considering particular Imternet disputes cases. On the contrary, in judicial acts, there is often a consistent application of several criteria -from the 'Zippo test' and the 'Calder test' up to relatively innovative methods that arise in the light of E-commerce development. At the same time, the practice of US courts cannot demonstrate the uniformity. For example, some courts adopt a "sliding scale" as the only mechanism for checking the existence of personal jurisdiction but others consider the inadmissibility of the approach. It goes without saying that this thesis can be easily justified by the need for the common law courts to be guided more by actual circumstances than by the desire to "attract" the circumstances of new disputes to successfully developed legal positions. However, it still seems that the formation of judicial practice uniformity cannot and should not be the prerogative of only the courts of the civil legal system, since the mentioned uniformity is a well-known mechanism for ensuring stability and predictability of business. The latter, in the conditions of almost uncontrolled development of the online world, can be guaranteed at least through traditional state and interstate institutions in the form of jurisdictional rules. Both corporations and persons who act in the Internet space should understand which of their actions can affect the forum state, and which can not. Such awareness will either minimize the risks of initiating disputes in a foreign jurisdiction, or exclude a potential "surprise" in the form of a lawsuit in a foreign country. Despite this, it would be absolutely wrong to deny the substantial development of American judicial practice in the jurisdictional aspect of the issue of Internet dispute resolving and exclude the possibility of accepting some of the mechanisms developed by it by Russian procedural law. Determining the jurisdiction of Internet disputes: the EU experience Nowadays there are some international treaties regulating jurisdictional issues are in force in the European Union. For example, in the Brussels I Regulation on jurisdiction and the recognition and enforcement of judgments in civil and commercial matters (as amended by EU Directive No. 1215/2012) [15] the jurisdictional rules are classified as general and special. Under the general rule, persons domiciled in a member state shall, whatever their nationality, be sued in the courts of that member state. Persons domiciled in a member state may be sued in the courts of another member state only by virtue of the rules set out in the Regulation. Special rules are established in Art. 7. Thus, a person may act as a defendant in a court of another EU member state, in particular, in the following cases: 1) in matters relating to a contract, in the courts for the place of performance of the obligation in question; 2) in matters relating to tort, delict or quasi-delict, in the courts for the place where the harmful event occurred or may occur; 3) if he is one of a number of defendants, in the courts for the place where anyone of them is domiciled, provided the claims are so closely connected that it is expedient to hear and determine them together to avoid the risk of irreconcilable judgments resulting from separate proceedings, etc.; The jurisdiction over disputes relating to insurance, as well as those arising from consumer contracts, is also determined in a special way. Regarding the consumer disputes, the Regulation establishes that the consumer has the right to sue both at the state of his domicile and at the state of the defendant's domicile. This includes if the contract was concluded with a party engaged in commercial or other professional activities in the consumer's domicile or otherwise directs his activities to the consumer's domicile or to several member states, including the consumer's domicile, and the contract falls within the scope of this activity. Brussels I Regulation, which entered into force on January 2015, is now also used to determine jurisdiction for disputes with non-EU residents. Despite the fact that the Regulation does not contain special rules on the issues of jurisdiction in relation to Internet disputes, the general rules that exist in it are quite correlated with the issues arising during the resolution of jurisdictional conflicts and allow them to be prevented. The EU's jurisdictional rules are also contained, in particular, in the Lugano Convention of 16 September 1988 on the Jurisdiction and Enforcement of Judgments in Civil and Commercial Matters [16] and Regulation (EU) 2017/1001 of the European Parliament and of the Council of 14 June 2017 on the European Union trade mark [17]. The EUTMR establishes autonomous rules of international jurisdiction over the disputes on infringement of an EU trademark, directly excluding the application of both the jurisdictional rules of the Brussels I Regulation and the national rules of the EU member States applicable to foreigners when considering disputes on infringement of national trademarks. The violation of the EUTMR disputes are subject to consideration in the specialized courts for EU trademarks existing in the each of member states. Thus, the place of consideration of the dispute always exists within the EU. The main question is in which state the dispute should be resolved. Despite the European law is primarily considered to be a civil law, the development of special jurisdictional rules applicable to Internet disputes does not occur through the adoption of additional international acts, but mainly through the development of legal approaches by the European Court of Justice. The European Court pays special attention to the issues of applying the test factor of lex loci delicti, which allows suing at the place of actual or potential occurrence of harm. For example, with regard to copyright, there are currently two most common ways of determination the place of commission of a tort in the Internet: 1) intention to target -the country of whose people of the site is addressed to; 2) accessibility online -the country where the site with the content is available; The first approach is reflected in the decision of October 18, 2012 in the case of Sportradar (C-173/11) [18] -the jurisdiction is held by the court of the country whose users the content of the website is addressed to. Concluding that focusing on the targeting of content does not always meet the general rule of fairness in dispute resolution, the European Court rejected it in favor of the actual availability of content in the case of Pinckney (C-170/12) [19]. With this approach, the court of the state in which the content posted on the website is available, which, according to the plaintiff, violates his copyright, will have the competence to consider a dispute on the recovery of damages caused by copyright infringement. The European Court paid special attention to the issues of jurisdiction in disputes related to the violation of trademark rights on the Internet. The territoriality principle applies to the legal protection of trademarks. It means that the exclusive right to a trademark is valid within the state that has granted legal protection [20]. Consequently, a violation of the exclusive right to a trademark can take place only if it is committed in this state. Thus, in the Wintersteiger case [21], the EU court pointed out that the concept of causing harm for the purposes of establishing jurisdiction may not coincide with the concept of causing harm in accordance with substantive law. The first is determined in accordance with the Brussels I Regulation. The EU Court of Justice confirmed that in the context of online trademark violations, the concept of "places where a tort has occurred or may occur" in Article 7 (2) of the Regulation also covers both the place where the damage was caused and the place of the event that caused the damage. Based on this, a claim for violation of the right to a national trademark can be brought in the courts of the member state in which it is registered, since protection in this case is limited to the territory of that State, and this State will be the place where the damage was allegedly caused. The question of whether the defendant's actions constitute a violation should be assessed by the court of the state of protection of the mark in the light of the applicable substantive law. In addition, the EU Court noted that the territorial restriction of the protection of a national mark does not exclude the international jurisdiction of courts of other states, in addition to where the trademark is registered. This legal position was expressed by the EU Court in the sense that in disputes related to trademark infringement, the place of the event that caused the damage is the place of activation of the ad display process. Since this place, associated with the location of the server of the search engine operator, is often uncertain, in these disputes, the location of the advertiser must be considered the place of commission of the offense [21]. Unification of jurisdictional approaches to the consideration of Internet disputes: the essence and prospects After analyzing the development of national and international legal regulation of issues of Internet disputes jurisdiction, it is possible to distinguish two directions. The first direction consists in the reception of national jurisdiction legal approaches in international legal acts. The USA is a kind of driving force in matters of regulating Internet relations and Internet disputes related to them. Due to the concentration of the largest commercial entities on the US territory, which laid the foundations for the E-commerce development, and specific federal relations, it was necessary for the USA to develop approaches that allow resolving jurisdictional issues between the states. The criterion of purposeful subordination, the 'Zippo test' and the 'Calder test' were developed through American law enforcement practice. Subsequently, these approaches were gradually reflected in the practice of the EU Court of Justice. Although EU law, when determining jurisdiction, adheres to the criterion of the place of commission of a tort or the occurrence of damage, and in American law, on the contrary, the most common criterion is the purpose of the defendant's activity, sometimes the opposite is also found in practice. In addition, the criterion of the place of commission of a tort or the occurrence of damage in Internet disputes is comparable in its content to the 'Calder test'. Thus, the analysis of the previously cited Sportradar case (C-173/11) shows that the European law-enforcer is trying to exercise the powers that the US courts have to adjust the jurisdictional provisions enshrined in the Brussels I Regulation, "expanding" the criterion of the place of damage by the criterion of site availability, that is, the purposeful subordination. The second direction consists in the implementation of international legal acts approaches into national legislation For example, in Article 247 of the Arbitration Procedure Code of the Russian Federation, the criterion of causing harm or the place of occurrence of damage (paragraph 4 of Part 1), and the criterion of purposeful impact on the territory of the Russian Federation (paragraph 9 of Part 1), and the principle of close connection (paragraph 10 of Part 1) are fixed. Similar criteria are contained in Part 3 of Article 402 of the Civil Procedural Code of the Russian Federation. Without the purpose to analyze mainly and exclusively the Russian procedural rules for determining the competence of Russian courts in relation to Internet disputes in this article, it should be briefly noted that to some extent they really cope with the resolution of such issues. Nevertheless, the paradox of the jurisdictional rules, which are best known to the legal community, is that they are aimed at establishing the connection of an Internet dispute with the territory of a state, i.e. they are territorial in nature. In turn, one of the key features of the Internet is its extraterritorial nature, due to which it is not always possible to establish the territorial limits of Internet influence [22]. The degree of influence actually exerted on users via the Internet cannot be quantified. It is not limited by any territorial or personal limits. It is not even limited by chronological limits. Information that gets online on one continent becomes available almost immediately on another continent, no matter what audience it is aimed at. Here Henry Adams' statement seems appropriate: "The difference is slight, to the influence of an author, whether he is read by five hundred readers, or by five hundred thousand; if he can select the five hundred, he reaches the five hundred thousand" [23]. Therefore, it is possible to create highly developed national law rules regulating the Internet disputes jurisdictional issues as much as necessary, but they will act exactly as long as they do not "collide" with the equally perfectly constructed law rules of a foreign legal system. Usually and truly predictably, state courts that implement one of the important components of the state sovereignty, resolving jurisdictional issues, proceed from the intention to establish their competence to resolve an Internet dispute in various ways, while excluding the possibility of establishing jurisdiction over this issue of another state. Thus, the conflict of jurisdictions in the framework of Internet disputes resolving becomes inevitable. The existing international legal regulation is clearly globally insufficient to prevent this conflict: there is no global international treaty regulating the issues of international jurisdiction of Internet disputes. The adoption of European regulations, including those of a special nature, is an important step towards the unification of international and national jurisdiction in the European space. However, its effect is limited to the EU. Conclusion It goes without saying that, given the current political situation, the possibility of concluding an international agreement on cooperation between countries in resolving Internet disputes is significantly limited. At the same time, a departure from the actually widespread paradigm of mediating one thing by another (which dominates something -a question that goes beyond both the research topic and the law) and cooperation in the perception of the best practices for resolving Internet disputes jurisdictional issues that are common in various countries would allow creating a universal international document. That significantly reduces jurisdictional conflicts when considering and resolving Internet disputes.	v3-fos-license
2018-12-12T23:06:42.747Z	2016-01-01T00:00:00.000	54932567	{ "extfieldsofstudy": [ "Physics" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://www.epj-conferences.org/articles/epjconf/pdf/2016/05/epjconf_toet2016_01042.pdf", "pdf_hash": "b40cdffbb727a66ef53eeebee1035814c9478e37", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45480", "s2fieldsofstudy": [ "Physics", "Engineering" ], "sha1": "b40cdffbb727a66ef53eeebee1035814c9478e37", "year": 2016 }	pes2o/s2orc	ANALYSIS OF INFLUENCE OF HEAT INSULATION ON THE THERMAL REGIME OF STORAGE TANKS WITH LIQUEFIED NATURAL GAS Is numerically investigated the process of convective heat transfer in the reservoirs of liquefied natural gas (LNG). The regimes of natural convection in a closed rectangular region with different intensity of heat exchange at the external borders are investigated. Is solved the time-dependent system of energy and Navier-Stokes equations in the dimensionless variables “vorticity – the stream function”. Are obtained distributions of the hydrodynamic parameters and temperatures, that characterize basic regularities of the processes. The special features of the formation of circulation flows are isolated and the analysis of the temperature distribution in the solution region is carried out. Is shown the influence of geometric characteristics and intensity of heat exchange on the outer boundaries of reservoir on the temperature field in the LNG storage. Introduction Natural gas is the important component of the guarantee of global energy needs in view of a number of advantages over other forms of the mineral of fuel.Frequently the transport of liquefied natural gas occurs more preferably in comparison with other methods of its delivery [1,2].Demand on LNG grows, but the logistics processes of production and storage of LNG in the modern environment -not an easy task.For the low-temperature storage LNG are used the constructions with different geometric and insulating characteristics [3,4].The analysis of the influence of such characteristics on the formation of the convective flows of liquefied natural gas in the storage tanks is one of important for their design and operation [4][5][6][7].Accordingly, it is of interest for numerical study of unsteady convective heat transfer in a rectangular LNG storage and analysis of the impact as the geometric dimensions and the influx of heat from the outer limits of the reservoir.Is respectively of interest conducting numerical studies of nonstationary convective heat transfer in a rectangular LNG storage and the analysis of influence both the geometric dimensions and inflows of heat from the outer boundaries of reservoir. Statement of the problem and the method of solution a Corresponding author: elf@tpu.ru 2 Are examined several versions storage LNG in the reservoir of rectangular form with the varied conditions for heat insulation (Fig. 1).The process of heat transfer in the solution field (Fig. 1) was described by the system of the nonstationary two-dimensional convection equations in the Boussinesq approximation 2. On the upper free boundary was assigned the free surface condition, on the restthe second kind.Numerical studies are carried out for the regimes of natural convection, which correspond to the Grashof numbers to 10 7 (laminar regime).Numerical solution of the problem is carried out by the method of finite differences using the algorithm was tested in group decision problems [8][9][10][11][12] of conjugate heat transfer in areas with local energy sources. Results and discussion Numerical analysis is carried out on four fairly typical examples of the size of areas: 1) l = 10 m, h = 5 m; 2) l = 10 m, h = 10 m; 3) l = 10 m, h = 20; 4) l = 20 m, h = 10 m.Heat-flux density on three outer boundaries of region (Fig. 1) it was assigned by the equal: q = 0.05 W/m 2 .The value l is selected as the dimensionless scale. Figure 2 shows the temperature field, which is formed with different relationships of the reservoir dimensions on the x coordinates and y.In all cases, is clearly see stable stationary during the formation of two vortices.Depending on the sizes of the sides of the solution region the vortices take the form, which corresponds to the direction of gravitational forces.With an increase in the size of reservoir along the height grows the gradient of temperatures (Fig. 3,4), which leads to the stratification of liquid in the reservoir.When a temperature difference of altitude of more than 0.1 degrees high probability of spontaneous revolution of different density layers, caused by temperature difference (the phenomenon of "rollover" [4][5][6][7]).The most intensive circulation flows in the liquefied natural gas appear in the wide low reservoirs.The use of tanks with this relationship of sides excludes stratification LNG in them and, correspondingly, the appearance of the phenomenon "of rollover".A change in heat exchange conditions on lower boundary does not exert a substantial influence on the structure of circulation flows and the temperature field in the reservoir (fig.5, 6).The decrease of the value of heat flux on lower boundary leads to reduction in the gradient of temperatures in the reservoir on the height (fig.6) and to the smoothing of temperature profile in the central section of region (X = 0.5,0 ≤ Y ≤1) (fig.7).The basic heating of liquid is achieved from the side of boundary with the large heat flow (fig.8).The prevailing vortex, which intensifies the mixing process of liquid in the region (fig.9).It is established that, a change in the intensity of heat flow on one of the vertical boundaries leads to scale changes in the flow pattern and temperature field in the entire region of the solution. Conclusion Conducted numerical investigations of convective heat transfer in the low-temperature storage of liquefied natural gas with different geometric and thermal conditions give the new information, which not only characterizes regime of convective flow in such reservoirs, but also it can be used for the improvement of the procedures of their design and improvement in the operating conditions.Thermophysical Basis of Energy Technologies 2015 01042-p.7 DOI: 10 .1051/ C Owned by the authors, published by EDP Sciences, Figure 2 .Figure 3 . Figure 2. Contours of stream function in the storage tanks LNG with different geometric dimensions: a) l = 10 m, h = 5 m; b) l = 10 m, h = 10 m; c) l = 10 m, h = 20; d) l = 20 m, h = 10 m.The numerical values of coordinates and temperatures are given in a dimensionless form.The results of numerical studies suggest a significant influence of the geometric dimensions of LNG storage on non-stationary temperature fields and stream lines. Figure 6 . Figure 6.Contours of stream function in the LNG storage tank: a) with the heat insulation on lower boundary (q = 0), b) with the heat flow on lower boundary (q = 0.05 W/m 2 ). Figure 7 .Figure 8 .Figure 9 . Figure 7. Temperature profile in the section X = 0.5, 0 ≤ Y ≤ 1: 1) with the heat insulation on lower boundary (q = 0), 2) with the heat flow on lower boundary (q = 0.05 W/m 2 ).The fields of temperature and flow line with the different values of heat flux on all three boundaries of the solution region are shown on figures 8,9.	v3-fos-license
2018-05-13T02:01:15.259Z	2018-04-23T00:00:00.000	13726027	{ "extfieldsofstudy": [ "Biology", "Medicine" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://openmicrobiologyjournal.com/VOLUME/12/PAGE/71/PDF/", "pdf_hash": "66468f5a2aea8f223e03d2f6843134dd86385ff6", "pdf_src": "PubMedCentral", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45482", "s2fieldsofstudy": [ "Biology" ], "sha1": "66468f5a2aea8f223e03d2f6843134dd86385ff6", "year": 2018 }	pes2o/s2orc	Probiotics Strains Modulate Gut Microbiota and Lipid Metabolism in Mule Ducks Background: Livestock production should respond to societal, environmental and economic changes. Since 2006 and the ban on antibiotics as growth factors in European Union, the use of probiotics has become widespread and has demonstrated the effect of intestinal microbiota on the performance of farm animals. Objective: The aim of this study was to investigate the effect of supplementation with Lactobacillus salivarius (as a probiotics strain or combined with other strains) on zootechnical performance, metabolic and immune gene expression and intestinal microbiota diversity in mule ducks using high-throughput sequencing and real-time PCR. Method: The mule ducks were reared for 79 days and overfed for 12 days with or without probiotics. Samples were collected at 14 (starting period) and 91 days (end of overfeeding period), 3 hours post feeding. Results: Irrespective of digestive content, age, level of feed intake or supplementation with probiotics, Firmicutes, Proteobacteria and Bacteroidetes were the dominant phyla in the bacterial community in mule ducks. At 14 days, both the ileal and cecal samples were dominated by Firmicutes (in particular the Clostridiales order). Overfeeding induced a shift between Clostridiales and Lactobacillales in the ileal samples whereas in the cecal samples, the relative abundance of Firmicutes decreased. Overfeeding also induced hepatic over-expression of Fatty Acid Synthase (FAS) and of the lipid transporter gene Fatty Acid Binding Protein 4 (FABP4). This increase in lipid metabolism genes is associated with a decrease in inflammatory response. Conclusion: Finally, probiotic supplementation had only a slight impact on gene expression and microbiota diversity, both at 14 days and after overfeeding. INTRODUCTION The intestinal microbiota plays an essential role in the host physiology and forms a complex ecosystem [1]. Microbiota can affect gut morphology or nutrient utilization [2], stimulate the immune response [3] and protect against pathogens [4]. In chickens, Lactobacillus spp. is the major genus in the small intestine (duodenum, jejunum and ileum) whereas Clostridium spp. and Bacteroides spp. are dominant in the ceca [5,6]. In ducks, Firmicutes and Bacteroidetes are the dominant phyla in both the ileum and the ceca, as previously described in chickens and mammals [7 -10]. However, Lactobacillus is not dominant, as described in adult chickens [8,9]. In rearing geese, the dominant phyla are Proteobacteria and Firmicutes in intestinal microbiota [10,11]. Previous studies have also shown that the diversity of intestinal microbiota is modified by diet [12]. During overfeeding, waterfowls are exclusively fed with corn which induces a hepatic steatosis called "foie gras", resulting from the storage of fatty acids in the liver [13,14]. Recent studies have shown a strong effect of diet and genetics on microbiota diversity, in particular an increase in Lactobacillus spp. after overfeeding [8 -10]. Interestingly, a link between the ability of ducks to trigger a liver steatosis and the composition of the intestinal microbiota according to the genetic type has also been highlighted [8,9]. Probiotics (live microorganisms) are known to be beneficial to their host [15] and since the ban on the use of antibiotics as growth factors in European Union, their use to improve animal health, welfare and productivity has increased [16]. Various actions have been described; for example, probiotics can reduce the presence of pathogenic bacteria in the intestine, probably through the production of lactic acid, bacteriocins or both [4,17,18]. Competition can also occur for the occupation of ecological niches [19]. Immune resistance has also been modulated by probiotic supplementation [20]. Probiotics have an effect on body weight gain and liver weights in young chicks and ducklings [21]. In broiler chickens, Kabir et al. [22] demonstrated a higher liver weight and better immune to sheep red blood cell immunization when chickens were supplemented with Enterococcus faecium. The Lactobacillus genus has also been regularly cited as an efficient probiotic in poultry [23,24]. Lactobacillus is known to have hydrolytic action and to favor digestion and absorption of nutrients, making them available to the host [25,26]. Lactobacillus inoculation in young chicks and ducklings has an effect on body weight gain and liver weights [21]. It has also been shown in geese that anti-inflammatory immunity mechanisms are linked to the Lactobacillus genus, and partially explain the animal's good health during overfeeding [10]. While many studies have been carried out on the effect of microbiota on poultry performance [27,28], little is known about the role of microbiota in waterfowl during overfeeding. The rapid evolution of the diversity of intestinal microbiota during the early stages of development demonstrated by Best et al. [29] and Rey et al. [30] in ducks and the strong effect of strain inoculation in newborn chickens on the composition of microbiota [31,32] suggest that supplementation in ducklings will be more efficient in modulating microbiota and host physiology immediately after hatching. In the poultry industry, there is an increasing interest in probiotics because in many countries, growers can no longer use antibiotics as growth factors [33]. In this study, we analyzed the effect of probiotic supplementation (Lactobacillus salivarius isolated from ducks at the end of overfeeding or strains isolated from chickens) on growth performance, microbial diversity, metabolism and immune gene expression in ducks. Animals and Experimental Design This trial was performed from February 2 nd 2015 to May 4 th 2015. The animals were cared for in accordance with the animal research guidelines of the French Ministry of Agriculture and the Directive 2010/63/EU. This trial was carried out at the Experimental Station for Waterfowls of INRA (Benquet, France with accreditation number B40-037-1). The schematic diagram of the experimental design is available in Fig. (1). A total of 150 post-hatching ducklings were randomly allocated to three separate pens (18m 2 per pen). The first group received the probiotic isolated by the cultural method (group A). The second received a probiotic product composed of a mix of strains isolated from healthy chickens (group B). Sterilized water was used as a control for the third group (group C). (1). Schematic diagram of the experimental design. During the starter and growing diets, animals from each experimental group were reared in separate pens (18m2 per pen). Heating was provided for the first 14 days, then no heat was provided and the animals had access to the outdoor pen (8m2). All ducks were fed ad libitum, from hatching to 28 days of age, with a starting diet (11.43 MJ/kg; crude proteins: 17.5%) and from 28 to 79 days of age with a growing diet (11.37 MJ/kg; crude proteins: 15.5%). At the age of 56 days, the animals were subjected to hourly rationing in order to prepare the overfeeding. At 80 days of age, these ducks were overfed with corn mixture (13.9 MJ/kg; crude proteins: 8.9%) for 12 days in cages containing four ducks (21 meals). During the overfeeding period, the ducks were allocated to cages according to the experimental group. The composition of the different diets used in this study is listed in (Table 1). During the rearing period (from hatching to 79 days of age), it was decided to mix the feed with adequate probiotics (group A and B) up to a final bacterial load of 2.10 8 CFU / g; or with sterile water (group C). During the overfeeding period (from 80 to 91 days of age), the animals were inoculated with probiotic A or B or sterile water by oral gavage with a gastric tube and syringe before the meal, according to the weight of diet received by the animal. During the overfeeding period, the animals that had been given probiotics during the rearing period were divided into two equal subgroups. One of the two subgroups continued to receive the adequate probiotics (group A+ and B+) and the other did not (group A-and B-), instead receiving sterile water like the control animals (group C). The individual BW (body weight) was registered every 14 days during experimentation. Feed distribution was registered every 2 days during the period from day 1 to day 79 (individual pen measurements). Food intake was measured at each distribution by deducting uneaten food from distributed food. Food intake was registered individually during the overfeeding period (from 80 to 91 days of age). The ducks were killed by exsanguination after electric stunning, 3 h after the last meal to homogenize the filling level of the ducks' digestive tract. Ten animals from each experimental group were selected at 14 days of age and were considered as the control group for the overfeeding effect (starting point, SP point). Ten animals from each experimental group were selected at 78 days of age and were only considered for their performance level (before overfeeding, Bof point). The 30 remaining animals were killed at the end of the overfeeding period (End of Overfeeding, Eof point). Probiotics and Food Preparation Two probiotics were used in this experiment. The first one (A) was a Lactobacillus strain isolated by the cultural method on MRS/Rogosa media (VWR chemicals, Radnor, Penn, USA) from the ileal digestive content of a duck at the end of overfeeding. Ileal digestive contents were an collected and inoculated on MRS/Rogosa agar plate after serial dilution in a Tryptone buffer. Ten characteristics colonies were purified on the MRS/Rogosa. Once isolated and purified, these strains were cultured in liquid MRS medium for 12 h at 37° C. These strains were then identified by sequencing the PCR product of the gene encoding the 16S RNA (Custom DNA Sequencing, Eurofins Genomics, Luxembourg, Luxembourg). Three strains were identified as Lactobacillus salivarius. The cultures of these three strains were then centrifuged to pellet the cells and resuspended in sterile water with a ratio of 0.8-fold the volume of culture. The stability of resuspended cells in the animal food was tested by mixing the resuspended cells with feed to a bacterial load of 2.10 10 CFU / g. The most stable strain was selected. Bacterial quantification decreased from 2.10 10 to 2.10 8 CFU / g during the first 24 h after diet preparation, so diets with probiotics were prepared and distributed 24 h later during the animal experimentation. The second probiotic is a probiotic product in powder form including probiotic bacteria isolated from the small intestinal tract (Enterococcus faecium and Bifidobacterium animalis) and cecum (Pediococcus acidilactici and Lactobacillus salivarius) of healthy adult chickens. The product had a total bacterial count of 1.10 11 CFU / g. To inoculate the animals, it was decided to mix probiotic B with the diet, at the same dose as probiotic A. For the control group, the same volume of sterile water was used. Sampling for Biological Analysis Blood was sampled for plasma analysis. The plasmas were separated by centrifugation at 3000×g for 10 min at 4° C and stored at −20° C. After dissection, the liver, pectoralis major (muscle) and subcutaneous adipose tissues (SAT) were weighed, sampled and stored at −80° C for the study of gene expression. The ileum and ceca were immediately collected and kept on ice. The digestive contents of the ileum and ceca were collected by gently squeezing the organ and stored at −80° C for the study of gut microbiota. Fatty liver Melting Rate Measurement In order to determinate the fatty liver melting rate (or fat loss during cooking), approximately 200g of the fatty liver were weighed and put into a glass jar with salt (12g/kg) and pepper (2g/kg). The jars were then cooked for 1 h in water in an autoclave at 85°C under a pressure of 0.8 bar. The temperature was controlled in water and in control jars equipped with temperature sensors. After 30 min of chilling by circulating cool water in the autoclave, the jars were stored at 4°C. The jars were opened after 2 months and exuded fat during cooking was carefully removed from the liver. RNA Isolation and Reverse Transcription Total RNA was isolated from the frozen tissue according to the TRIZOL method (Invitrogen/Life technologies, Carlsbad, Cal, USA). Total RNA concentration was measured by spectrophotometry using a NanoVuePlus (GE Healthcare, Chicago, Illinois, USA), and all samples were normalized at 500 ng /µl before storage at -80°C until cDNA generation. The integrity of total RNA was analyzed by electrophoresis. cDNA was obtained by reverse transcription using the enzyme Superscript III (Invitrogen/Life technologies, Carlsbad, Cal, USA) and a mix of oligo dT and random primers (Promega, Fitchburg, Wisconsin, USA). 3 µg of total RNA was used, and the absence of contamination by DNA was verified by two negative controls (without RNA and without Superscript). The reaction was carried out in a StepOne (Applied Biosystem, Waltham, Massachusetts, USA) for 25° C / 5 min, 55° C / 60 min, 70° C / 15 min and retention at 4°C until storage at −20°C. Real Time PCR, Primers and Gene Expression Analysis Ten birds per experimental group (with or without probiotics according to sampling point) were used for quantitative PCR analysis, both for metabolism and immune response gene expression. All primer sets are listed in Table (2). The reactions were run in duplicate in a final volume of 15 μL. The PCR mix was made up of 7.5 μL of SybrGreen Universal PCR Master Mix (Quanta Bioscience, Gaithersburg, MD, USA), 5.5 μL of 500 nM specific primers, and 2 μL of template cDNA or the negative controls. Real-time PCR was performed in a StepOne instrument (Applied Biosystem, Waltham, Massachusetts, USA) with an initial denaturation step of 10 min at 95° C, and 35 cycles for denaturation of 15 s at 95° C, annealing/extension for 1 min at a specific temperature for each primer, and 1 final cycle at 70° C for 15 s. Melt curve analyses were done by slowly heating the PCR mixtures from 60 to 95° C, and the cycle threshold (Ct) was determined with the StepOne Applied Biosystem software 2.3. The chosen reference gene was Actin B but all the results were confirmed with glyceraldehyde 3-phosphate dehydrogenase (GAPDH). The standard group was group C at SP point. For several genes, results could only be obtained at Eof point. In this case, the reference group was group C at Eof point. All Cycle thresholds (Ct) were collected. Results are expressed as 2 -∆∆Ct with ∆∆Ct= ((Ct target -Ct ref ) sample)-((Ct target -Ct ref ) standard). DNA Extraction Total genomic DNA from the ileal and cecal samples was extracted by combining mechanical and thermic lysis using an Ultra Turrax Digital Homogenizer IKA T-25 (Fisher Scientific, Illkirch, FR) and a QIAamp Fast DNA Stool Mini Kit (QiagenGmbh, Hilden, DE) according to the manufacturer's instructions with 220mg as starting material. The lysis temperature used was 95°C. The DNA sample was eluted with 50 µl of ATE buffer (Qiagen Gmbh, Hilden, DE) and stored at -20 °C. The quantity and quality of DNA extracted were measured using the NanoVue Plus (GE Healthcare, Chicago, Illinois, USA). 16S rRNA Amplification and Sequencing The PCR for sequencing were realized on the 16S rRNA gene according to the method described by Lluch et al. [34] using MiSeq kit reagents v2 (2x250 bp pair ended reads). Amplicons from the V3-V4 regions of 16SrRNA genes were generated using specific bacterial primers 5'CTTTCCCTACACGACGCTCTTCCGATCTAC GGRAGGCAGCAG 3' and 5'GGAGTTCAGACGTGTGCT CTTCCGATCTTACCAGGGTATCTAATCCT 3'. The preparation of amplicons was performed in a total volume of 50µL containing 1 U TAQ Polymerase and adequate 10 X PCR buffer (MTP Taq DNA Polymerase, Sigma), 200µM of dNTP (Sigma), 0.2µM of each primer and 2µL of DNA template. The amplification program consisted of an initial denaturation step at 94°C for 1 min and 32 cycles of denaturation at 94°C for 1 min, annealing at 63° C for 1 min and elongation at 72° C for 1 min. In the end, a final extension step at 72° C for 10 min was carried out. The quality of PCR products was controlled by electrophoresis. 2µL of PCR product were deposited on an agarose gel (1% / TBE) with load Buffer for 30-40 min at 135 V. Amplicons were then sent to the INRA genomic platform in Toulouse for sequencing. The amplicons were purified briefly using the magnetic beads Agencourt AMPure XP-PCR Purification (Beckman Coulter, Brea, CA, USA) following the 96well format procedure, modified as follows: beads / PCR reactional volume ratio of 0.8 x and final elution volume of 32 μl using Elution Buffer EB (Qiagen). The concentration of the purified amplicons was controlled using Nanodrop 8000 spectrophotometry (Thermo Scientific). Single multiplexing was performed using a homemade 6-bp index, added to reverse primer during a second PCR with 12 cycles using a forward primer (5'AATGATACGGCGACCACCGAGA TCTACACTCTTTCCCTACACGAC 3') and reverse primer (5'CAAGCAGAAGACGGCATACGAGAT-index-GTGACTGGAGTTCAGACGTGT 3'). The resulting PCR products were purified and loaded onto the Illumina MiSeq cartridge according to the manufacturer's instructions. The quality of the run was checked internally using PhiX Illumina, and then each pair-end sequence was assigned to its sample with the help of the previously integrated index. Sequencing Analysis All software for analysis of sequenced data was used in the FROGS pipeline developed by the French National Institute of Agricultural Research (INRA, Toulouse, France) [35]. First the sequences were denoised using the Flash 1.2.11 and Cutadapt 1.7.1 tools [36,37]. Only sequences with sizes between 380 and 500 bp, without ambiguous bases and with the two primers (5' primer ACGGGAGGCAGCAG and 3' primer AGGATTAGATACCCTGGA) were kept. Secondly, the sequences were clustered with Swarm 2.0 [38] in two steps, as recommended by Escudié et al. [35]. Swarm uses an iterative growth process to cluster sequences. In each growth step, the sequence of the previous step was used to find the other sequences with a number of differences inferior or equal to the "Aggregation distance". The first step, or denoising step, served to build very fine clusters with an aggregation distance equal to 1 between the sequences of each crown. After the denoising, a second swarm with an aggregation equal to 3 between seeds from this first clustering was used to delineate OTUs. Next, PCR chimeras were removed using VChime of the Vsearch package [39]. This step was performed after clustering in order to shorten the analysis time. Next, the remaining clusters were filtered. Only clusters present in at least 2 samples and with an abundance greater than 0.005% of the total sequences [40] were kept. Finally, taxonomic affiliation for each OTU was obtained with BLASTn+ [41] by blasting the sequences on the Silva128 database [42]. Statistical Analysis Values given in the text are expressed as means ± SEM. Analyses of the OTU table were performed with the PhyloSeq package [43]. Relative abundance, Chao1 diversity richness, Shannon and InvSimpson richness were determined. Beta-diversity was determined using nMDS with the Bray-Curtis distance method. Sparse Partial Least Squares Discriminant Analysis (s PLS DA) was performed to determine the most discriminant OTU using the mixOmics package [44] with CSS normalization and log transformation of count sequences. A cutoff value of 0.95 was used to select the discriminant OTUs. Statistical analyses were performed with ANOVA analysis of variance using the R software 3.3.1 [45]. Differences were considered as significant when P-value < 0.05. When significant, differences between treatments were compared using Tukey's test and FDR with BH correction [46]. In order to analyze relationships between the bacterial community and bio-chemical plasma parameters and growth parameters, an sPLS (sparse Partial Least Squares) analysis was also carried out using the mixOmics package. The values are expressed as means ± standard error of the mean (SEM). a, b, c, Means in the same row with unlike superscripts differ (P < 0.05). Bird Performance Concerning group C, during the rearing period, BW increased from 508±6 g (SP point) to 4306±75 g (Bof point) (P<0.001). Moreover, BW was statistically different between experimental groups at SP point. In fact, the BW of group C at SP point was higher than those of group A and B at the same point (508±6 g vs 486±5 and 484±6 g; P<0.001). This difference disappeared at 42 days of age (data not shown). During the rearing period, no effect of probiotic addition was observed on liver weight (LW), muscle weight (MW) and subcutaneous fat weight (SFW). In group C, between SP and Eof point, the overfeeding period increased BW (4 306±75 g to 6 261±97 g; P<0.001), LW (93±5 g to 628±21 g; P<0.001) and SFW (57±8 g to 151±5 g; P<0.001). MW was not influenced by overfeeding (310±13 g vs 315±5 g; N.S) Table (3). As previously, no effect of probiotic addition during overfeeding was observed on BW, LW, MW and SFW. Feed consumption during overfeeding was expressed as g of raw feed/bird Table (3). No statistical difference in feed consumption was observed during rearing (data not shown) and the overfeeding period Table (3), with or without probiotic supplementation. The fatty liver melting rate (fat loss during cooking) was used to determinate fatty liver quality. Here, probiotic addition had a significant effect on the fatty liver's melting rate. In fact, the melting rate decreased in group A-at Eof point compared to group C at the same point (19.6±1.2% vs 12.7±1.7%; P<0.05) ( Table 3). However, no significant differences were observed between group A-and the other probiotic groups (A+, B-, B+). Glycemia and Triglyceridemia No effect of probiotic supplementation was observed during SP point on glucose and triglyceride plasma concentration Table (3). Between SP and Eof point, glycemia increased (from 1.99±0.71 g / L to 2.94±0.84 g / L in group C; P<0.005), as did triglyceridemia (from 1.43±0.25 mmol / L to 4.76±1.37 mmol / L in group C; P<0.001). Probiotic supplementation had no impact on glucose concentration at Eof point. Concerning the plasma concentration of triglycerides during Eof point, we observed that group A+ had a higher triglyceridemia than group C (5.86±1.32 vs 4.76±1.37 mmol / L; P<0.05). However, no significant differences were observed between group A+ and the other probiotic groups (A-, B-, B+). Metabolic and Immune Response Gene Expression The expression of gene markers of lipogenesis and lipid uptake were then highlighted Table (4). Metabolic gene expression was measured in the liver. Fatty Acid Synthase (FAS) expression increased strongly between SP and Eof point (717±312-fold in group C; P<0.005). In the same way, Fatty Acid Binding Protein 4 (FABP4) expression increased strongly between SP and Eof point (177±77-fold in group C; P<0.001). However, probiotic supplementation had no effect on metabolic gene expression either at Eof point or at SP, irrespective of the tissues and genes studied Table (4). For Fasting-Induced Adipose Factor / angiopoietin-like protein 4 (Fiaf factor), expression increased more than 3 times between SP and Eof point. Bacterial Community A total of 4,603,890 16s RNA sequences were obtained from the MiSeq sequencing for the 102 samples. After preprocessing with FROGS, a total of 4,029,552 sequences were kept. After clustering with Swarm 2.0, we obtained 356,437 clusters. Then the PCR chimeras (70,343 clusters), representing 190,937 sequences, were removed. The 286,094 remaining clusters were filtered according to several criteria and the clusters present in at least 2 samples were kept as well as the ones representing at least 0.005% of total sequences [40]. Thus, 412 clusters or OTUs (Operational Taxonomic Units) were kept, representing 75.9% of total sequences, with an average of 104±5 OTUs per sample representing an average number of sequences of 33,321±1037 per sample. Finally, 7 Phyla, 99 genera and 412 species were detected in all samples independently of experimental group or sample origin. Ileal Bacterial Community (Core Microbiota in all Samples) In the ileal content, the major phylum, independently of the experimental group, was Firmicutes (98.3±1.4%). Proteobacteria represented around 1.4±0.5% of the population. Finally, Bacteroidetes and Actinobacteria each represented 0.1±0.1% of the population. Other phyla such as Tenericutes and Fusobacteria represented less than 0.1% of the population Fig. (2a). To evaluate the microbiota composition at finer taxonomic levels, order distributions were analyzed Fig. (2). The phylum of Firmicutes was dominated by Clostridiales and Lactobacillales orders (55.6±9.9% and 42.7±7.3% respectively of the total population) and Proteobacteria was dominated by the orders of Burkholderiales (0.6±0.5% of the total population) and Enterobacteriales (0.2±0.1% of the total population). A representation of ileal microbiota on family level according to experimental group is available in Fig. (3). Finally, 95 genera were detected in all ileal samples. Fig. (2). Relative abundance (%) evaluated at the phylum (a) and order (b) levels of main bacterial groups in the ceca before (Bof) and after (Eof) overfeeding according to experimental group. Cecal Bacterial Community (Core Microbiota) In the cecal content, the major phyla, regardless of the experimental group, were Firmicutes (50.6±17.4% of the total population), Bacteroidetes (25.3±10.2% of the total population) and Proteobacteria (23.4±9.4% of the total population). Actinobacteria and Tenericutes represented around 0.6±0.5% of the population. Finally, Fusobacteria and Cyanobacteria represented less than 0.1±0.1% of the population Fig. (4a). At order level, Firmicutes were dominated by Clostridiales (47.7±16.5% of the total population) and Bacteroidetes were dominated by Bacteroidales (25.2±10.2% of the total population). The orders of Enterobacteriales and Desulfovibrionales accounted for 10.6±4.4% and 12.0±5.0% of the total population for the Proteobacteria phylum Fig. (4b). A representation of cecal microbiota on family level according to experimental group is available in Fig. (5). Finally, all 99 genera were detected in all cecal samples. Effect of Overfeeding and Probiotics on Ileal and Cecal Microbial Communities in Mule Ducks In both the ileal and cecal samples, probiotic addition had no effect on richness and diversity at SP point, but at Eof, diversity and richness tended to decrease Table ( Table (5). However, in group C, InvSimpson in ceca tended to decrease at Eof point compared to SP point (17.2±2.7 in to 6.9±3.1; N.S). This difference was statistical between both groups A at SP point and groups A at Eof point (30.0±5.0 to 7.8±2.3 in group A+ and 7.8±2.3 in group A-; P<0.05) and between groups B at SP point and groups B at Eof point (27.0±6.2 to 7.7±2.00 in group B+ and 9.1±2.0 in group B-; P<0.05). Finally, during the rearing period probiotic addition had no effect on either the ileal or the cecal samples ( Figs. 1 and 2). The values are expressed as means ± standard error of the mean (SEM). a, b, c, Means in the same row with unlike superscripts differ (P < 0.05). Effect on Ileum No effect of probiotics or overfeeding was detected at phyla levels in the ileal samples. At order level, in group C, we observed a shift of Clostridiales (97.7±0.3% to 17.6±5.4%; P<0.001) and Lactobacillales (0.5±0.1% to 80.4±5.5%; P<0.001) between SP and Eof point. In the same group, the relative abundance of Bacillales also decreased between SP and Eof point (0.1±0.1% at SP point to 0.0±0.0% at Eof point; P<0.05). In the same way, the probiotic addition had a slight effect on the order level. Group A-at Eof point had higher relative abundance of Burkholderiales (2.0±0.8%) than group A (0.2±0.1%; P<0.05) and group B (0.2±0.0%; P<0.05) at SP point. However, no statistical difference was observed between the other groups. In the ileal samples, the Lactobacillus genus represents 0.1±0.1, 0.2±0.2 and 0.3±0.1% of the total population respectively in group A, B and C at SP point. No statistical difference was observed between experimental groups at SP point. In group C, as in the other experimental groups, the overfeeding period increased the relative abundance of the Lactobacillus genus (from 0.3±0.1 to 78.0±10.3% of total population, P<0.001). Regardless of the experimental group, the most abundant Lactobacillus during the rearing period were L. aviarius, L. salivarius and L. amylovorus with 0.1±0.1% of total relative abundance each. During the overfeeding period, the most abundant Lactobacillus species were L. amylovorus (9.8±7.2% of relative abundance) and L. plantarum (9.3±6.8% of relative abundance). During the overfeeding period, L. salivarius represented 1.5±0.5% of the Lactobacillus genus. Effect on Ceca At phylum and order level, the probiotic addition had no impact on relative abundance at SP point. At phylum level, in group C, Firmicutes decreased from 99.1±0.2% at SP point to 17.2±7.6% of the population at Eof point (P<0.001). In the same way, in group C, Actinobacteria decreased between SP and Eof point (P<0.05). However, this phylum represented less than 0.1% of the total population. The decrease in the Firmicutes and Actinobacteria phyla between SP and Eof point was detected in all experimental groups independently of the supplementation of probiotics On the other hand, Bacteroidetes and Proteobacteria tended to increase (respectively from 0.1±0.1% at SP point to 33.5±10.9% of population at Eof point and 0.6±0.2% to 48.6±16.4% of population, N.S) in group C. The increase in Bacteroidetes between SP and Eof point was significant between group A at SP point (0.1±0.1% of total population) and group A+ (50.3±9.8% of total population) (P<0.05). In the same way, the Proteobacteria phylum increased between group B (0.7±0.2% of total population) and group B+ (46.9±8.6% of total population) (P<0.05). At order level, in group C, the huge decrease in Firmicutes can be explained by the decrease in the Clostridiales order from 89.5±4.5% at SP point to 17.1±7.6% of total population at Eof point (P<0.001). In the same way, the increase in Bacteroidetes between group A at SP point and group A+ at Eof point was related to an increase in Bacteroidales (0.1±0.1% in group A to 50.3±9.8% of total population in group A+; P<0.05). Finally, probiotic addition influenced the relative abundance at Eof point of Campylobacterales only, which decreased in group B+ (0.1±0.0% of total population; P<0.05) and group B-(absence of detection) compared to group C at Eof point (0.3±0.3% of total population; P<0.05). No statistical difference was observed between Eof point of groups B+ and B-. Conversely, an increase in Campylobacterales from 0.3±0.3% to 2.1±0.9% (P<0.005) was observed between group A+ and group C at Eof point. No statistical difference was observed between the other groups. At the species level, Campylobacter jejuni was dominant. Fig. (3). Heatmap representation of ileal microbiota at family level. Samples and OTUs were clustered using nMDS ordination with the Bray-Curtis ecological distance method. The different families are indicated at the level of the order. Samples were indicated according to experimental group. s PLS DA for Ileal Samples Discriminant analysis (s-PLS DA) was performed to identify the most discriminant OTUs between experimental groups. A first s-PLS DA in order to highlight the most discriminant OTUs was performed between SP and Eof point, independently of the probiotics supplementation Fig. (6a). The OTUs enabling to discriminate between these two groups were correlated to the OTUs affiliated to two Lachnospiraceae families (from the Clostridiales order and undetermined species) and to OTU_456 (affiliated to Enterococcus durans 100%). The second s-PLS DA was performed between the experimental groups at SP point Fig. (6b). Group B was the most separated group. The separation between these three groups was also correlated to 4 OTUs affiliated to Lachnospiraceae, 3 OTUs affiliated to Ruminococcaceae families (from the Clostridiales order and undetermined species) and to OTU_456 (affiliated to Enterococcus durans 100%). Finally, a third s-PLS DA was performed between the experimental groups at Eof point Fig. (6c). Groups A+ and B+ were the most variable groups. The separation between these two groups was correlated to 2 OTUs affiliated to Lachnospiraceae (undetermined species), 3 OTUs affiliated to Ruminococcaceae families (undetermined species) and to 3 OTUs affiliated to the Lactobacillus genus (OTU_400: L. curvatus 98.5%, OTU_44: L. vaginalis 100% and OTU_270: L. paralimentarius 99%). However, these OTUs are very minor (less than 0.05% of relative abundance). Fig. (4). Relative abundance (%) evaluated at the phylum (a) and order (b) levels of main bacterial groups in the ileum before (Bof) and after (Eof) overfeeding according to experimental group. sPLS Analysis: Relationship Between Bacterial Taxonomic Profiles and Liver Bio-Chemical Parameters and Growth Performances We studied the relationships between the bio-chemical plasma parameters and growth parameters and the OTU profile of the microbiota by sPLS. According to the leave-one-out process, to compute the Mean Square Error of Prediction, two components were computed and one projection plan (components 1 and 2) was considered. Fig. (7) is a graphical representation of the selected OTUs on the first two sPLS dimensions. The selected OTUs and the biochemical and growth parameters are projected onto correlation circles where highly correlated variates cluster together (within a data set or between the two data sets). The first component did not allow us to group together growth and biochemical parameters. Considering the OTUs, high number of them were negatively correlated with bio-chemical and growth parameters. Most of them were members of Clostridiales, especially Lachnospiracae and Ruminococcae, already highlighted to be the most discriminant OTUs between experimental groups in sPLS-DA. s PLS DA for Cecal Samples Discriminant analysis (s-PLS DA) was also performed to identify the most discriminant OTUs between experimental groups. The first analysis was performed between SP and Eof point independently of the probiotics supplementation Fig. (8a). The OTUs enabling to discriminate between these two groups were at least correlated to OTU_230 affiliated to the Alistipes genus 97.6% (Bacteroidales order and undetermined species) and to OTU_431 (affiliated to Clostridiales vadin BB60 group 97.7%). It is nevertheless noteworthy that the increase in the Alistipes between the two points was significant (from 0.0±0.0 to 6.8±2.3% of total population in group C up to 10.2±3.2% in group A-; P<0.005). A second s-PLS DA was performed at SP point for groups regarding the supplementation of probiotics or not Fig. (8b). Group A is the most separated group. The separation between these three groups was correlated to 3 OTUs affiliated to the Ruminococcaceae family (undetermined species). Finally, a s-PLS DA was performed with Eof point for groups with probiotics or not Fig. (8c). The separation between groups is not so clear even though the cutoff used was lower (cutoff lowered to 0.75 instead of 0.95 for the 2 other ones). However, the separation between these groups was correlated to 2 OTUs affiliated to the Barnesiella genus (OTU_141 affiliated to B. viscericola 96.2% and OTU_195 affiliated to Barnesiella spp. 94.8%, OTU_183 affiliated to Variovorax paradoxus 100%) and to OTU_15 affiliated to Lachnospiraceae 100% (undetermined species). However, the relative abundance of these OTUs is less than 0.01%. Fig. (6). Sparse Partial Least Square Discriminant Analysis (sPLS-DA) of the ileal microbial community (a) before and after overfeeding independently of experimental 2 group, (b) between the three experimental groups during the rearing period, (c) between the five experimental groups after the overfeeding period according to the first two explanatory variables. Control groups before and after overfeeding are represented in black while the experimental groups with probiotic A are represented in green and groups with probiotic B in red. Fig. (7). sPLS variables representation of the liver biochemical parameters and growth parameters (blue) and the selected bacterial OUT in ileal microbiota (red) on the first two sPLS dimensions. Variable lying outside the small correlation circle are highly correlated. Variables that cluster together are correlated. sPLS Analysis: Relationship Between Bacterial Taxonomic Profiles and Liver Bio-Chemical Parameters and Growth Performances As described for ileal samples, we studied the relationships between the bio-chemical plasma parameters and parameters and the OTU profile of the microbiota by sPLS. Fig. (9) is a graphical representation of the selected OTUs on the first two sPLS dimensions. In accordance with Table (3), bio-chemical and growth parameters were grouped together and positively correlated except for glycemia. Considering the OTUs, high number of them were positively correlated with bio-chemical and growth parameters. Most of them were members of Clostridiales, especially Lachnospiracae and Ruminococcae, already highlighted to be the most discriminant OTUs between experimental groups. Furthermore, Lactobacillus and Pediococcus genera of Lactobacillales order were negatively correlated with these parameters. DISCUSSION In this study, high-throughput sequencing was used for the first time to identify the modulation of ileal and cecal microbiota, metabolism gene expression and bird performance after supplementation with Lactobacillus salivarius (isolated from ducks or as a mixture with other lactic acid bacteria) from hatching to the overfeeding period. Lactobacillus strains and other lactic acid bacteria have been described as increasing starch digestibility in chickens and energy harvested from food [28,47]. Furthermore, Lactobacillus salivarius is the dominant species isolated from intestinal content in ducks and in geese in previous studies, as well as in our work [48,49]. The use of Lactobacillus strains as probiotics has been very well described in broiler chickens, especially to stimulate an immune response, digestive health and growth performance [18,24,50]. In ducks, Bacillus subtilis is the strain most commonly used as a probiotic and few studies on the potential role of Lactobacillus as a probiotic have been reported [9,51]. As previously described in ducks, geese and chickens, here, Firmicutes, Proteobacteria and Bacteroidetes are also dominant in all samples independently of sampling point, overfeeding, supplementation of probiotics or digestive contents [8 -11, 27]. In the ileal samples, Firmicutes is the dominant phyla (more than 98% of sequences) and in particular the order Clostridiales, unlike in chickens where Lactobacilliales dominate [52]. Proteobacteria was the second most common phylum in our study, as well in Canada geese [53], graylag geese [10] and Muscovy (Cairana moschata) and mule ducks [8,9]. This observation suggests that the bacterial digestive metabolism of chicken and waterfowl (both ducks and geese) could be quite different in terms of the ability to trigger a hepatic steatosis. In the cecal samples, even though the Firmicutes phylum was also dominant (up to 50%), Bacteroidetes (25%) and Proteobacteria (23%) were more abundant in our work in comparison with previous studies on ducks and geese [8 -10]. Moreover, in chickens, geese and ducks, it has previously been evidenced that the intestinal microbiota is modulated by diet, environment and host genetics, which could partially explain the differences [8 -11]. In our study, strong differences were observed between SP and Eof point in terms of microbial diversity in both the ileal and cecal samples, suggesting that age and overfeeding can modulate intestinal microbiota. Vasai et al. [9] showed little effect of overfeeding on the microbiota at the ceca level but in our study, we compared samples from ducks overfed with younger birds, which can probably explain the greater differences in microbial diversity in the ceca. The decrease in Firmicutes (order Clostridiales) associated with the increase in Bacteroidetes (order Bacteroidiales) and Proteobacteria has already been highlighted in a previous study in the lab [30]. Furthermore, in Pekin ducks, the most abundant bacteria (more than 95%) over the first 36 days were also Firmicutes, in particular, the Clostridiales order, as in our study [29]. Interestingly, in our work, at Eof point, a strong increase in the abundance of Lactobacillus, as described in geese and ducks, was observed but not linked to the supplementation of probiotics [8 -10]. The abundance of Lactobacillus in overfeeding is high so we hypothesize that the supplementation did not yield statistical differences. Overfeeding increased the relative abundance of Lactobacillales (essentially Lactobacillus genus) associated with a decrease in Clostridiales, as previously described in ducks and geese [8 -10]. Lactobacillus strains are very well known as amylolytic bacteria and increase in pigs, rats and cattle with diets containing high levels of starch [54 -56]. Furthermore, the Lactobacillus genus has been identified as increasing amylase activity in the small intestine in chickens [57]. Then, Lachnospiracae and Ruminoccocae were identified as most discriminant OTUs between experimental groups in both ileum and ceca. Moreover, these OTUs were respectively negatively and positively correlated with growth and bio-chemical parameters in ileum and ceca. Interestingly, these families were identified in previous works, as enriched in ceca of chickens with good FCR (food conversion ratio) and increased body weight [58,59]. Then Lactobacillus were also correlated with body weight gain [58]. Overfeeding increased liver weight and fattening levels in the control group, in line with previous studies [8,9,14,60]. However, while the body weight was affected by both probiotic supplementations during the first 28 days, other growth parameters (liver, fat and muscle weights) were not improved. Several works show an improvement in body weight and FCR (feed conversion ratio) and protection against pathogens [50, 61 -63]. But other studies do not show any positive effect on the performance of chickens or ducks [9,64,65]. The differences obtained in these studies could be partially explained by the difference in strains and their concentrations, the methods of supplementation (feed, water, or invasive method) or the genetic strains of birds. Although liver weight was not affected by probiotic supplementation, slight differences in melting rate were observed after overfeeding. Next, the group with the L. salivarius supplementation only during the rearing period had a lower melting rate than the control group, suggesting that this strain could improve melting performance, which is a very important performance parameter for farmers and the industry. Nevertheless, it is now known that although the fatty liver melting rate is directly related to fatty liver weight, other parameters not yet identified are also implicated [66,67]. Next, the plasma concentrations of glucose and triglycerides (TG) increased between SP and Eof point, which is quite consistent with the metabolic state of overfed ducks and geese, where significant changes occur [13,14,68,69]. So as previously described in these studies, the overfeeding period strongly increased the TG and glucose plasma concentration in our study as well. However, probiotic supplementation had no effect on these parameters during both rearing and overfeeding periods. Moreover, at the end of the overfeeding period, the group with L. salivarius supplementation had a higher triglyceride concentration than group C at Eof point. Two known mechanisms during overfeeding may explain this difference. First, during overfeeding, de novo lipids synthesized in the liver are exported via the VLDL to the peripheral tissues via the blood circulation [13,69]. The second mechanism is the lipid re-uptake of the liver at the end of overfeeding, as demonstrated by Tavernier et al. [14]. So the higher blood TG level can be explained either by a higher export or a lower lipid re-uptake, or both. Interestingly, overfeeding had a huge impact on host metabolic gene expression, as previously described [14,60]. Next, genes implicated in de novo lipogenesis (Fasn, DGAT2), in Fatty acid transport (FABP4), increased strongly. A modulation of the energy balance enhanced by intestinal microbiota, as demonstrated by Bäckhed et al. [1], could be also related to the change in plasma TG and glucose concentrations. Several studies show that in mammals, the microbiota triggers the storage of triglycerides through the suppression of the Fiaf factor, a lipoprotein lipase (LPL) circulating inhibitor [1,70,71]. In ducks, the LPL activity correlates positively with a higher storage in peripheral tissues instead of fat storage in the liver [13,72,73]. Interestingly, the expression of the Fiaf factor increases significantly during the overfeeding period according to the decrease in LPL activity during overfeeding in mule and Muscovy ducks [13,72,73]. Changes observed in intestinal microbiota after overfeeding allow us to partially explain metabolic changes in ducks via the FIAF factor expression. Furthermore, in mice, the implication of the intestinal microbiota has been identified in susceptibility to hepatic steatosis, as described in mice regarding NASH susceptibility [74]. Moreover, hepatic steatosis in mice is generally associated with an increase in Firmicutes [74,75] while a significant increase in the Clostridiales order is associated with metabolic protection [75]. Here, Firmicutes remained stable, but Lactobacilliales increased whereas Clostridiales decreased on the order level. Moreover, immune response was also modulated during the overfeeding period, as previously described in geese [10]. These authors showed that the complement system, part of the inflammatory system, was suppressed during overfeeding and partially explained it by an increase in blood lactic acid from enriched Lactobacillus. Here, LITAF (responding to LPS, a component of the gram-negative wall) as well as IL-8 gene expression decreased when PPARγ increased after overfeeding, in line with the decrease in Campylobacter observed. Furthermore, supplementation with L. sakei in ducks leads to a decrease in Enterobacteria [9]. Other probiotic strains such as Enterococcus faecalis are also able to down-regulate PPARγ activity and IL-10 levels in humans [76]. Furthermore, in chickens, Lactobacillus in diet protects birds against coccidiosis by enhancing immune stimulation [23]. In our study, probiotic supplementation did not affect metabolic gene expression during the rearing or overfeeding periods. Probiotic supplementation allowed us to show a slight effect on immune response modulation during the overfeeding period, more specifically for the Litaf gene. The Litaf gene encodes for a transcription factor activated in response to the presence of lipopolysaccharide. As mentioned above, the Lactobacillus and Enterococcus genera are known to modulate the immune response. The presence of these two genera in probiotic B could therefore explain the lower pro-inflammatory response in this group showed by a decrease in Litaf expression. Probiotic supplementation had no statistical effect on microbial population, either in the ileum or the caeca contents at phylum or order levels during the rearing period. In the same way, probiotic supplementation had a slight effect on bacterial communities during the overfeeding period. CONCLUSION To conclude, our results confirm the significant changes to metabolism gene expression and microbial diversity triggered by overfeeding, but not so much by probiotic supplementation. Interestingly, anti-inflammatory response seems to be decreased in overfed ducks and probably explained by changes in microbial composition. This work probably partially explains the tolerance to hepatic steatosis in ducks. The identification of protective factors could also offer therapeutic clues against hepatic steatosis in humans. ETHICS APPROVAL AND CONSENT TO PARTICIPATE The animals were cared for in accordance with the animal research guidelines of the French Ministry of Agriculture and the Directive 2010/63/EU. This trial was carried out at the Experimental Station for Waterfowls of INRA (Benquet, France with accreditation number B40-037-1). We also acknowledge the GetPlage INRA team (in particular Pauline Heuillard) for providing phyloseq script and for helpful suggestions concerning the use of sequence analysis software. The logistic support from INRA Toulouse was deeply appreciated.	v3-fos-license
2018-12-30T09:18:41.840Z	2009-08-05T00:00:00.000	73647583	{ "extfieldsofstudy": [ "Materials Science" ], "oa_license": "CCBY", "oa_status": "GOLD", "oa_url": "https://downloads.hindawi.com/journals/amse/2009/503042.pdf", "pdf_hash": "a9f7f103e7c0f71a64a07749f00af6e6741c49ee", "pdf_src": "Anansi", "provenance": "20241012_202828_00062_s4wg2_08f3ca33-e262-4f97-a832-04b691aafe33.zst:45483", "s2fieldsofstudy": [ "Materials Science", "Physics" ], "sha1": "a9f7f103e7c0f71a64a07749f00af6e6741c49ee", "year": 2009 }	pes2o/s2orc	Polymer Light-Emitting Diodes Efficiency Dependence on Bipolar Charge Traps Concentration The efficiency of light-emitting diodes (LEDs) based on poly[(9,9-dioctylfluorenyl-2,7-diyl)-alt-1,4-benzo-{2,1′-3}-thiadiazole)], F8BT, is optimized upon simultaneous doping with a hole and an electron trapping molecule, namely, N,N′-Bis(3-methylphenyl)N,N′-diphenylbenzidine and 2-(4-biphenylyl)-5-(4-tert-butylphenyl)-1,3,4-oxadiazole, respectively. It is shown that, for devices with poly(3,4-ethylene dioxythiophene) doped with polystyrene sulfonic acid as hole-injection layer material and magnesium cathodes, the efficiency is nearly doubled (from ca. 2.5 to 3.7 cd/A) upon doping with ca. 0.34% by weight of both compounds. Introduction Polymer light-emitting diodes (LEDs) have suffered significant developments since the first report of electroluminescence observation in conjugated polymers in 1990 [1].This technology is now available in the first commercial products incorporating polymer displays.A very successful combination of materials development and device engineering has been at the origin of this tremendous evolution.Doping of the active polymer layer has been one the approaches used to improve devices efficiency (in particular upon use of phosphorescent dopants [2]) and tune the emission colour, upon use of energy accepting dopants [3,4].Studies on LEDs based on this kind of composites have shown that the emission from the dopants is generally amplified, with respect to the emission obtained upon photoexcitation, due to charge trapping and direct recombination at the dopants [3,4].Sainova et al. [5] have shown that doping a blue emitting polyfluorene copolymer with various hole transporting materials (in 3% by weight) leads to an improvement of colour stability.Furthermore, they showed that the use of a dopant with lower ionization potential, acting as hole trapping, leads to a significant efficiency improvement.A similar strategy is widely used in the optimization of LEDs based on sublimed, low-molecular-weight organic materials [6,7].In this paper we explore the use of bipolar doping, that is, using a hole and an electron trapping dopant, to improve the efficiency of poly[(9,9-dioctylfluorenyl-2,7-diyl)-alt-1,4benzo-{2,1 -3}-thiadiazole)] (F8BT) based LEDs.By varying the dopant concentration we find a concentration range leading to improved electroluminescence (EL) efficiency.An increase from ca. 2.5 to 3.7 cd/A was obtained upon doping with ca.0.34% by weight of both compounds. Experimental Solutions of F8BT (from American Dye Source, ADS133YE) in chloroform were doped with N,N -Bis(3-methylphenyl)-N,N -diphenylbenzidine, or TPD, and 2-(4-biphenylyl)-5-(4-tert-butylphenyl)-1,3,4-oxadiazole, or PBD (both from Aldrich).For the preparation of each solution, the same weight of the two dopants was used.Table 1 shows the compositions of the different solutions (and films) used in this study.Assuming a specific gravity of 1 g/cm 3 for the films of neat F8BT and the composites, the concentration number by volume unit of each dopant is also shown in Table 1. Indium-tin-oxide-(ITO)-coated glass substrates were cleaned with acetone and 2-propanol and treated with oxygen plasma.PEDOT: PSS (or PEDOT, for short, Baytron P from Bayer) was spin coated on top and annealed on a hot Results and Discussion The rationale behind the choice of these two dopants is demonstrated in Figure 1. The workfunction of PEDOT:PSS and Mg was taken from [9,10], respectively.Ionization potentials (Ips) and electron affinities (EA) of TPD and PBD were taken from [11] and [12], respectively.Based on cyclic voltammetry measurements on drop-cast films of F8BT, we previously determined Ip = 5.9 eV and EA = 2.42 eV [13].We note that this EA value for F8BT is much lower than the value of 3.2 eV estimated [14] from Ip, considering the optical gap (2.4 eV) and assuming an exciton binding energy of 0.3 eV.In spite of this difference, we consider that the value obtained from CV is closer to the relevant energy for electron transfer, due to the closer similarity of processes. Based on the relative position of the frontier levels (assuming vacuum level alignment) we conclude that TPD acts a hole trap with respect to F8BT, while PBD is an electron trap.As evidenced in Figure 1, the electron trapping energy (difference between EA(F8BT) and EA(PBD)) is ca.0.2 eV, smaller than the hole trapping energy of 0.5 eV (difference between Ip(F8BT) and Ip(TPD)).Supporting evidence comes from studies of LEDs showing that doping F8BT with 2% by weight of TPD causes a much more significant decrease of the current than does a similar doping with PBD (also 2% by weight) with respect to LEDs based on neat F8BT.Furthermore, both PBD and TPD are blue emitters.No energy transfer occurs from F8BT to any of the two dopants.In solid state, the photoluminescence (PL) emission of TPD (with smaller energy gap than PBD) peaks at ca. 420 nm, while the emission of F8BT peaks at about 550 nm.No spectral overlap occurs between F8BT emission and TPD absorption.Any exciton formed at the dopant sites would be, instead, transferred to F8BT.In fact, the electroluminescence spectra of the doped devices are the same as that of the devices based on neat F8BT, without any evidence of dopants interference. For the preparation of the composites, we used the same amount, by weight, of each dopant.Due to the lower molecular weight of PBD, this is always present in higher concentration number per unit volume, as observed in Table 1.However, the effect of this higher concentration with respect to TPD is attenuated by its lower charge trapping energy, as discussed above. Figure 2 shows typical results obtained for the LEDs prepared with neat F8BT and with the composites A to F described in Table 1. Figure 2(a) shows that doping of F8BT leads to a decrease of the current flowing through the devices.Such current decreases upon increase of dopant concentration.Both disorder and charge trapping contribute to this effect, and both explain the observed trend.In view of the relative possition of the frontier levels, we consider that charge trapping is likely the most effective factor, supporting the relative energetic prosition of the frontier levels shown in Figure 1.The luminance-voltage behaviour is also strongly dependent on dopant concentration: the maximum luminance increases from device A (110 cd/m 2 ) to neat F8BT (16200 cd/m 2 ), being this maximum luminance attained at increasing voltages upon increase of dopant concentration.A significantly different luminancevoltage behaviour is observed for the device prepared with the composite having the highest dopants concentration (device A).For the other devices, they all have similar behaviour, with similar values for the light-onset voltage (3 to 3.5 V). The electroluminescence (EL) efficiencies shown in Figure 2(c) are maximized at intermeditate doping levels (compositions C, D, and E).A decrease of dopant concentration leads to efficiencies closer to that of neat F8BT, while higher concentrations (above composition E) lead to an efficiency decrease. The above results are explained in terms of electron trapping at PBD sites and hole trapping at TPD sites.The presence of these bipolar traps improves the charge balance within the emissive layer leading to an efficiency increase.At intermediate dye concentration levels (ca. 10 20 to 10 19 cm −3 ) we find maximum efficiencies.We attribute the decrease of EL efficiency at the too high concentrations to trapped charge-induced luminescence quenching.At the lowest concentrations we observe a decrease of the trapping effect, approaching the behaviour of the devices based on neat F8BT. Conclusions Doping with adequate bipolar charge trapping dyes can lead to an increase of polymer LEDs efficiency, providing their concentration is optimised.For the studied system, composition D is the one giving, on average, the highest efficiency, with an increase of ca.60% with respect to neat F8BT.Furthermore, the use of bipolar dopant concentrations in the range of compositions D to E lead to maximum EL efficiency while allowing for maximum luminances (ca. 10 4 cd/m 2 ) not much different from those of neat F8BTbased LEDs.and the composites indentified in Table 1.The device identification (A to F) is that of the corresponding composite used as active layer. Table 1 : [8]position of the composites used in this study.C for 2 minutes, in air.Its final thickness is 45-50 nm, as measured with a Dektak profilometer.Thin films (thicknesses of ca.85-90 nm) of neat F8BT or dye doped F8BT were then spun coated on top of PEDOT from chloroform solutions.Magnesium cathodes were thermally deposited at a base pressure of ∼10 −5 mbar, defining 4 mm 2 pixels.The devices were characterised as described in[8].	v3-fos-license

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.