text
string | predicted_class
string | confidence
float16 |
|---|---|---|
Insufficient pre-DSB coalignment in the hybrids may enhance the probability of entanglements between chromosomes and increase the occurrence of ectopic recombinational intermediates between nearly homologous sites of nonhomologous chromosomes. Because the sequence homology matching in these intermediates is apparently low, they should be less stable. The unwanted DNA connections tend to be eliminated by the mismatch repair system59 and active chromosome movements62. We propose that the balance in stability between “wanted” and “unwanted” connections is impaired in hybrids, due to a decrease of sequence homology between the homologous regions of the parental species. This leads to variability between oocytes in the number and size of regions involved in multivalents.
|
study
| 100.0 |
These requirements of normal synapsis between homologous chromosomes are apparently not met in the hybrids. Multiple heterozygosity for chromosomal rearrangements hinders preliminary DSB-independent pairing between homologous chromosomes and increases the incidence of non-homologous associations. Divergence between parental genomes decreases homology at the sequence level, the stability of homologous heteroduplexes, and affects the efficiency of discrimination between correct and ectopic interhomologue interactions. In hybrids, the wide variation between genetically and chromosomally identical oocytes in the ratio of homologously paired to non-homologously paired regions indicates that the choice between homologous and non-homologous synapsis at each of these steps is random. Detailed molecular mechanisms of these processes remain to be elucidated, and interspecies hybrids, such as those reported here, provide an excellent model for future studies.
|
study
| 100.0 |
The sex difference observed in this study can be categorised as an example of “graduated steps of sterility”30 from advanced in females to complete in males. Genetic and chromosomal incompatibility is probably amplified in the male hybrids by the well known vulnerability of spermatogenesis to pairing aberrations28. We observed stochastic variation in “degree of sterility” even between oocytes of the same F1 genotype. Those containing “correctly” paired configurations are probably able to produce viable balanced oocytes. The more asynapsed regions an oocyte contain, the larger part of its genome undergoes meiotic silencing of unsynapsed chromatin (MSUC)6465, and the higher the chance for the oocyte to be directed to apoptosis. The results of this study indicate that reproductive isolation based on hybrid sterility may be built up in a gradual mode. A gradual genetic divergence and the sequential fixation of different chromosome rearrangements in isolated populations increase the probability of pairing errors followed by MSUC and apoptosis in the hybrid gametocytes.
|
study
| 100.0 |
Seven adult male and 12 newborn female hybrids between M. arvalis (dams) and M. levis (sires) were examined, as well as three adult male M. arvalis, four adult male and three newborn female M. levis. Captive-bred colonies of the parental species were established from individuals trapped in Leningrad district (M. arvalis) and Novosibirsk district (M. levis) and maintained in the animal house of the Institute of Cytology and Genetics. Maintenance, handling and euthanasia of animals followed protocols approved by the Animal Care and Use Committee of the Institute of Cytology and Genetics. Experiments described in this manuscript were carried out in accordance with the approved national guidelines for the care and use of laboratory animals.
|
study
| 99.94 |
Chromosome spreads were prepared from spermatocytes or embryonic oocytes according to Peters et al.66. Cell spreads were treated as described in Anderson et al.44 using rabbit polyclonal anti-SYCP3 (1:500; Abcam), mouse monoclonal anti-SYCP3 (1:100; Abcam), rabbit polyclonal anti-SYCP1 (1:500; Abcam), mouse monoclonal anti-MLH1 (1:50; Abcam), rabbit polyclonal anti-RAD51 (1:200; Calbiochem), mouse monoclonal anti-γH2A.X (1:500; Abcam), rabbit polyclonal anti-γH2A.X (1:500; Abcam) and human anticentromere (ACA) (1:100; Antibodies Inc) primary antibodies. The secondary antibodies used were Cy3-conjugated goat anti-rabbit (1:500; Jackson ImmunoResearch), Alexa450-conjugated goat anti-rabbit (1:100; Invitrogen), FITC-conjugated donkey anti-rabbit (1:200; Jackson ImmunoResearch), FITC-conjugated goat anti-mouse (1:50; Jackson ImmunoResearch), AMCA-conjugated donkey anti-human (1:100; Jackson ImmunoResearch), and Cy3-conjugated goat anti-human (1:100; Jackson ImmunoResearch) antibodies. Antibodies were diluted in PBT (3% bovine serum albumin and 0.05% Tween 20 in phosphate-buffered saline). A solution of 10% PBT was used for blocking. Primary antibody incubations were performed overnight in a humid chamber at 37 °C; and secondary antibody incubations, for 1 h at 37 °C. Slides were mounted in Vectashield antifade mounting medium (Vector Laboratories) to reduce fluorescence fading.
|
study
| 99.94 |
Centromeres were identified by ACA foci. MLH1 signals were scored only if they were localised on the SC. In the parental species the length of the SC of all bivalents was measured in micrometers using MicroMeasure 3.367. To estimate the total SC length in the hybrids, whose pachytene cells contained partially or completely unpaired chromosomes, we measured the lateral elements of SC and then divided the sum by two.
|
study
| 100.0 |
Chemomechanical preparation is the first step to eliminate microorganisms in the root canal system, but it alone is not sufficient to clean the root canal. Ex vivo and clinical studies have indicated that intact areas remain on root canal walls during mechanical preparation, and therefore, it is important to perform irrigation in addition to mechanical preparation15,23. For this purpose, several different irrigation solutions, medicaments and techniques have been used10,11,14.
|
review
| 99.6 |
Calcium hydroxide (CH) is widely used as an intracanal medicament between appointments to increase the number of canals free from bacteria because of its antibacterial, therapeutic, biocompatible, and regenerative properties4,26. However, the remnant CH hinders the penetration of disinfectants and sealers into dentinal tubules and compromises the seal of the canal filling17,22. Therefore, residual CH must be removed before permanent root canal obturation is completed8. In most cases, the residual CH was removed following copious irrigation with sodium hypochlorite (NaOCl) and ethylenediaminetetraacetic acid (EDTA) in combination with the use of master apical file (MAF) or gutta-percha up to working length (WL)24. Different irrigation techniques and devices were used to activate and improve the effectiveness of irrigation solutions7,10,14,25.
|
review
| 97.56 |
A novel irrigation instrument, the XP-endo Finisher, has been introduced by FKG, Dentaire SA (La Chaux-de-Fonds, Switzerland). This instrument’s design is similar to an ISO size #25, 0.00 taper NiTi file. According to manufactures, this file improves the penetration of irrigation solutions to the irregular area of root canal system by expanding its reach 6 mm in diameter3,29.
|
other
| 99.8 |
Although previous studies demonstrated efficacy of sonic and ultrasonic systems on debris and CH removal8,10,28, sufficient information is not yet available in the literature concerning XP-endo Finisher31 and laser-activated irrigation (LAI). Therefore, the aim of this study was to compare the effect of XP-endo Finisher, sonic, ultrasonic and LAI and conventional syringe irrigation techniques in removing CH from simulated lateral irregularities on the root canal wall. The null hypotheses were that the removal of CH was not affected by irrigation technique or the section of root canal (third).
|
study
| 99.94 |
Ethical approval was obtained from the Clinical Research Ethics Committee of the Faculty of Medicine of Gaziosmanpasa University (project number: 15-KAEK-229). One hundred and five human maxillary incisors were used in this study. Mesiodistal and bucco-palatal direction radiographs were taken from the teeth to confirm the presence of a single canal. The teeth were decoronated using straight diamond burs (Komet, Gebr. Brasseler GmbH & Co. KG, Germany) in a conventional high-speed handpiece under water cooling so that each root had a standardized length of 15 mm. The WL was determined to be 1 mm short of the apex using a #10 K file (VDW GmbH, Munich, Germany). The WL was established at 14 mm. Roots were prepared with Reciproc rotary files up to size R40 at WL (VDW GmbH, Germany) and irrigation was performed with 10 mL 2.5% NaOCl. Next, roots were placed in Eppendorf tubes (Labosel, İstanbul, Turkey) filled with a silicone material (Zetaplus soft; Zhermack Clinical, Badia Polesine, Italy). After removal of the roots from the impression material, a longitudinal groove was prepared on the buccal and lingual surface using a narrow diamond bur without cutting the canal wall. A spatula was used to split longitudinally. A number 1S cavitron tip (Aceton, Merignac, France) was modified and inserted into an ultrasonic handpiece (Newtron P5; Satelec, Acteongroup, France) to create artificial standardized grooves. Two of the three standard grooves were created in the coronal and apical part of one segment, and another in the middle part of a second segment. The dimensions of grooves were 0.2 mm in width, 3 mm in length, and 0.5 mm in depth (Figure 1). The dimension of grooves was checked under a stereomicroscope (Zeiss Stemi 2000-C, Carl Zeiss MicroImaging, Göttingen, Germany) at 20X magnification. The root halves and grooves were irrigated with 5 mL of 17% EDTA for 1 min. and 5 mL 2.5% NaOCl for 1 min. while being activated with a toothbrush to remove debris and the smear layer. The standardized grooves were filled with CH paste (Ammdent, Punjab, India), and the root halves were reassembled. The root canals were fully filled with CH paste using a Lentulo spiral. Two radiographs (mesiodistal and bucco-palatal direction) were taken to confirm complete filling of the canals with CH paste. Roots were remounted and placed into Eppendorf tubes. Then samples were divided randomly into 7 groups, each containing 15 teeth. Seven different color stickers were pasted on the caps of the Eppendorf tubes to indicate each of the 7 groups. The access cavities were sealed with temporary filling material (Cavit, 3M ESPE, Seefeld, Germany) and stored in 37°C at 100% relative humidity for 2 weeks.
|
study
| 100.0 |
Group 3- CanalBrush: The root canals were irrigated with 5 mL 2.5% NaOCl and then brushed with a medium size CanalBrush (Coltene/Whaledent GmbHCo. KG, Langenau, Germany) at 600 rpm for 1 min. A final flush was done with 5 mL 2.5% NaOCl. CanalBrushes were inserted 1 mm short of the WL and moved with small vertical movements.
|
other
| 99.9 |
Group 4- XP-endo Finisher: Irrigation protocol was same as Group 3 with the exception that the XP-endo Finisher (FKG, Dentaire Sa, La Chaux-de- Fonds, Switzerland) was used instead of the CanalBrush at 800 rpm with 1 Ncm for 1 min. The file tip was placed 1 mm short of the WL.
|
other
| 99.9 |
Group 5- Sonic Irrigation (Vibringe): A 10 mL 2.5% NaOCl was delivered and sonically activated via the Vibringe system (Vibringe B. V. Corp, Amsterdam, Netherlands). The needle tip was placed 1 mm short of the WL without touching the canal walls, enabling it to vibrate freely for 2 min.
|
other
| 99.9 |
Group 6- Passive ultrasonic irrigation (PUI): Irrigation protocol was the same as Group 3, with the exception that the passive ultrasonic activation was performed using an Irrisafe ultrasonic tip (size 25, 0.02 taper) (Satelec Acteongroup, France) that was placed 1 mm short of the WL. A power setting of 9 was used for duration of 1 min. A 10 mL 2.5% NaOCl solution continuously delivered at a flow rate of approximately 0.16 mL s-1through the unit.
|
other
| 99.7 |
Group 7- Er:YAG laser-activated irrigation: The irrigation solution was activated with the same protocol as in Group 3, with the exception that an Er:YAG laser (Kavo Key 3+, KaVo, Biberach, Germany) with a 2940 nm wavelength for 1 min. was used with endodontic tips (a 28 mm long and diameter ISO 30) in place of a CanalBrush. Laser parameters were 1 W, 10 Hz, 100 mJ, and an energy density of 142.8 J/cm2. The laser tip was inserted into canal at 1 mm short of the WL. When the root canal irrigant dropped or vaporized, the canal space was filled with 2.5% NaOCl.
|
study
| 82.4 |
For each specimen, 11 mL of 2.5% NaOCl was used as irrigation solution and was delivered at a flow rate of approximately 0.08 mL s-1 except for G6 (PUI). The irrigant was delivered into the canal with a double side-vented needle (i dental, Lithuania) except for Group 1 (beveled needle). The amount of remaining CH in the grooves was evaluated under a stereomicroscope (Zeiss Stemi 2000-C, Germany) at 20× magnification and equipped with a digital camera (AxioCam ERc5s, Germany) by two calibrated endodontists using an numeric evaluation scale described by van der Sluis, et al.30 (2007). The scoring system was as follows: score 0, the groove is entirely empty; score 1, CH is present in less than 50% of the groove; score 2, CH is present in more than 50% of the groove, but not completely; and score 3, the groove is completely covered with CH (Figure 2). Evaluation was conducted based on the color codes by two endodontists blinded to the group number. Before scoring, the two endodontists assessed 50 randomly selected specimens simultaneously for calibration purposes. In the case of discrepant scores, a consensus was reached by discussion.
|
study
| 100.0 |
Kruskal-Wallis test was used to compare the non-normal data among groups. For multiple comparisons between the pair-wise groups, Bonferroni-Correction Mann Whitney U test was used. A p-value <.05 was considered significant. The kappa coefficient was used to determine interexaminer agreement. Analyses were performed using SPSS 19 (IBM SPSS Statistics 19, SPSS Inc., Somers, NY).
|
study
| 99.94 |
Results of the two examiners were in good agreement (kappa value=0.897). Comparisons between the groups are presented in Tables 1 and 2. Elimination of CH was more difficult from the apical region. None of the irrigation protocols could completely remove all remnant of CH in all three root regions. Beveled needle irrigation (Group 1) and double side-vented needle irrigation (Group 2) were the significantly least efficient on the elimination of CH from the grooves (P<.001). PUI (Group 6) and Er:YAG laser-activated irrigation (Group 7) removed more CH than the other protocols in all thirds of the root; however, no significant differences were found between these two groups (P>.05). No significant differences were found between XP-endo Finisher (Group 4) and PUI (Group 6) at the coronal and middle regions (P>.05). The Kruskal-Wallis test showed significant differences between the groups for coronal, middle, and apical thirds (P<.05), except for the CanalBrush (Group 3) (P>.05).
|
study
| 100.0 |
Table 1Multiple comparisons according to root canal regions Kruskal-Wallis statistical analysisScores 0123Median[IQR]PGroup 1 (Beveled Needle)Coronal Middle Apical- - -- - -8 5 17 10 142a 3ab 3b 0.023Group 2 (Double Side - Needle)Coronal Middle Apical- - --10 5 25 10 132a 3ab 3b 0.011Group 3 (CanalBrush)Coronal Middle Apical- - -- 3 114 9 71 3 72 2 20.098Group 4 (XP-endo FinisherCoronal Middle Apical- - -4 4 -11 9 7- 2 82a 2a 3b 0.001Group 5 (Vibringe)Coronal Middle Apical- - -2 1 -13 12 8- 2 72a 2ab 2b 0.004Group 6 (PUI)Coronal Middle Apical7 7 15 5 73 3 7- - -1a 1a 1b 0.029Group 7 (LAI)Coronal Middle Apical6 6 26 8 53 1 8- - -1ab 1a 2b 0.022
|
study
| 89.0 |
Table 2Multiple comparisons between groups Kruskal-Wallis statistical analysisScores 0123Median[IQR]PCoronalBeveled Needle--872ac <.001Double Side - Needle--1052ac CanalBrush--1412ac XP-endo Finisher-411-2bc Vibringe-213-2ac PUI753-1b LAI663-1b MiddleBeveled Needle--5103a <.001Double Side - Needle--5103a CanalBrush-3932a XP-endo Finisher-4922ac Vibringe-11222a PUI753-1bc LAI681-1b ApicalBeveled Needle--1143a <.001Double Side - Needle--2133a CanalBrush-1772a XP-endo Finisher--783a Vibringe--872a PUI177-1b LAI258-2b
|
other
| 98.25 |
Irrigation has an important role in controlling endodontic infection and debridement of the root canal system. CH is a widely used intracanal medicament as it creates a physical barrier from microorganisms and avoids the development of reinfections. However, previous studies have shown that removal of CH before the completion of the root canal obturation increased the sealer penetration into the dentinal tubules, and thus provided a good seal and also strengthens the bond between dentine and sealer6,11.
|
study
| 99.94 |
In the practice of endodontics, many different irrigation methods are used for this purpose. In this study, the effectiveness of different irrigation techniques on CH removal was evaluated. In the literature, calculation of the residual amount of CH remaining in the root canal were made by calculating the area of the remnant on dentin wall, by scoring, SEM analysis, volume analysis with spiral CT, and by using a micro-CT5,25,30,32. We preferred the scoring method described in the study by van der Sluis, et al.30 (2007) because it is a simple and easily accessible technique and used in many previous studies1,8.
|
study
| 100.0 |
CH powder was mixed with liquid and used in paste form in this study. Lambrianidis, et al.18 (1999) used CH medications at 42% and 95% concentrations in their study and reported that CH content in the paste had no effect on removal from the root canal wall.
|
study
| 100.0 |
A commonly suggested method for removal of CH is the irrigation with NaOCl and EDTA accompanied by very light instrumentation and the use of MAF19,25. Many techniques have been proposed to increase the efficiency of the irrigation solution. Mechanical agitation with a handpiece, gutta-percha or plastic tools, sonic and ultrasonic activation are also suggested techniques. In addition, LAI is another efficient and current method13,21. Kenee, et al.16 (2006) removed CH by four different procedures and demonstrated that the use of hand files and irrigation solution alone was not very effective on the removal of CH.
|
review
| 99.6 |
Capar, et al.8 (2014) compared CH removal efficiency of EndoVac, Self-Adjusting File (SAF), PUI and conventional irrigation techniques from artificial standardized grooves. They reported that PUI was more effective than other groups when NaOCl was used. Ahmetoglu, et al.1 (2013) compared CH removal efficiency of PUI, SAF and conventional irrigation methods by SEM. Researchers found that PUI was more effective than SAF and traditional irrigation. These results support our research. PUI technique is based on the transmission of acoustic energy to an irrigation solution. The agitation increases the penetration of irrigant to the irregular canal areas and the CH removal capacity of the irrigation solution. Similarly, van der Sluis, et al.30 (2007) investigated the efficiency of CH removal of various irrigation processes and found that PUI with NaOCl was more effective than PUI with water and syringe irrigation with NaOCl.
|
study
| 99.94 |
In this study, the effectiveness of different activation protocols of irrigation solutions were compared between themselves and with traditional irrigation methods. The results are consistent with the findings of the above-mentioned authors. According to our results, LAI and PUI were more effective in the elimination of CH in all 3 regions of the root canal. Therefore, the null hypothesis that no differences would occur among the different irrigation techniques in terms of CH removal was rejected. In addition, no difference was found between PUI and XP-endo Finisher groups in the coronal and middle third regions. Although CanalBrush and sonic-activated irrigation exhibited lower scores than needle irrigation groups, the difference was not statistically significant. Results of our study indicate that there was a significant difference between sections of the root canal in terms of CH removal except for the CanalBrush group; therefore, the second null hypothesis is also rejected.
|
study
| 100.0 |
Despite results by Balvedi, et al.5 (2010), who stated PUI was more effective in terms of CH removal than syringe irrigation in the coronal and middle third regions, no statistically significant difference was found between the two irrigation regimes in the apical third. In the aforementioned study, researchers mixed CH with different liquids and preferred saline solution to the irrigant. In this study, CH was mixed with only distilled water and 2.5% NaOCl was used as the irrigation solution. The reason for the differences between the studies may be the various solutions and CH vehicle used. Similar to our findings, a recent systematic review showed the superiority of PUI over syringe irrigation on the removal of CH from the apical part of root canals10.
|
review
| 99.7 |
Wiseman, et al.32 (2011) evaluated the efficacy of CH removal with sonic irrigation and ultrasonic irrigation in the mesial root canal of mandibular molars by using the same volume irrigant (6% NaOCl+14% EDTA) and same time interval. Authors found PUI repeated 3 times with 20 sec. intervals was more effective than sonic irrigation. This study’s results are similar to the findings of Wiseman, et al.32 (2011), who found PUI was superior to sonic irrigation, although a different duration and irrigation regime was applied.
|
study
| 99.25 |
Tasdemir, et al.27 (2011) evaluated the use of NaOCl and NaOCl+EDTA with different agitation techniques (CanalBrush, PUI, MAF) on the removal of CH. Although the type of solution did not influence the CH removal, CanalBrush and PUI were found statistically more effective than other techniques. Our results were not consistent with the findings of the aforementioned study. Although CanalBrush removed more CH than the needle groups, this difference was not statistically significant. Another difference is that the PUI removed significantly more CH than CanalBrush in all three root regions. Tasdemir, et al.27 (2011) used field measurement methods as opposed to our study that utilized a preferred scoring method with a PUI time interval of 1 min.
|
study
| 99.94 |
Alturaiki, et al.2 (2015) compared the CH removal capacity of EndoVac, sonic, ultrasonic activated irrigation techniques with conventional irrigation. Researchers reported that no statistically significant difference was found between sonic and ultrasonic irrigation in the coronal and middle third regions, but found that sonic irrigation was more efficient in the apical region. Our results do not coincide with the results of these authors. The differences between this outcome and our results may reflect different variables in the study design, such as (i) the sonic device used and (ii) the MAF size. Researchers used EndoActivator (Dentsply Tulsa Dental Specialties, Tulsa, OK), NaOCl and EDTA in combination and prepared root canals up to #45 MAF. In this study, root canals were prepared up to #40 MAF and only 2.5% NaOCl was activated via Vibringe system. However, Bolles, et al.7 (2013) stated that no significant differences were found between EndoActivator and Vibringe systems on the sealer penetration into dentinal tubules. Further investigations are needed to clarify the effectiveness of these devices in removing medicaments.
|
study
| 99.94 |
Li, et al.21 (2015) compared CH removal efficacy of conventional needle irrigation, sonic, ultrasonic activated irrigation, and Photon-Induced Photoacoustic Streaming methods in root canals and isthmuses. Study results showed that laser and ultrasonic activated irrigation protocols were the most effective methods on the elimination of CH in both apical region and isthmuses. The findings of Li, et al.21 (2015) support the results of our research. According to our results, PUI was the most effective method for removal of CH than needle irrigation techniques in all three root regions. In addition, no statistically significant difference was found between XP-endo Finisher and PUI groups in the coronal and middle regions of root canals. However, Leoni, et al.20 (2016) stated that no significant differences were found between PUI and XP-endo Finisher in removing debris from root canal surface.
|
study
| 99.94 |
The XP-endo Finisher is newly introduced NiTi file to improve efficacy of final irrigation procedure. The canal must be shaped at least ISO 25 file to use XP-endo Finisher. The instrument is stabile in martensite form at room temperature and can be bent to the desired shape. The instrument changes its phase to austenite when the temperature reaches the body temperature3,29. We believe that only one study31 compared the effectiveness of the XP-endo Finisher on CH removal. In contrast to our findings, Wigler, et al.31 (2016) reported that no significant difference was found between PUI and XP-endo Finisher on the removal of CH from simulated irregularities in the apical third of root canals. However, previous studies showed its superior debris removal ability from the root canal walls3,20.
|
study
| 99.94 |
The apical third exhibited higher amounts of residual CH than the coronal and middle thirds in all experimental groups, except for LAI. This finding is in line with the results of previous studies12,21. This observation may be related to accumulation and transfer of residual CH to the apical region, which has smaller canal area and smaller volume of irrigation solution, as well as the anatomic complexity of apical third9,12. The action and circulation of irrigants may therefore be hindered.
|
study
| 100.0 |
The activation of NaOCl with different instruments enhanced CH removal. Nevertheless, none of the investigated protocols were able to completely remove the CH from all three root regions. LAI and PUI methods removed more CH than the other protocols from artificial grooves in all thirds of the root canal.
|
study
| 99.9 |
Induced pluripotent stem cells (iPSCs), which are created from an adult cell that has been reprogrammed, enable the development of an unlimited source of any type of human cells needed for drug discovery and clinical applications . iPSCs are able to help track the earliest disease-causing events in cells and can be used as sources of various cell-based therapies. Because a healthy quality of undifferentiated iPSCs is an essential requisite for further experimental and therapeutic approaches, the rapid and robust estimation of iPSC quality is very important to meet growing demands [2–4]. The morphological structure of a healthy or good-quality iPSC colony commonly has tightly compacted round cells and an explicit boundary, whereas unhealthy or bad-quality colonies show a different morphology . The present approach of evaluating the quality of iPSCs on the basis of colony morphology is predominantly subjective and can strongly differ according to individual skills. Therefore, a quantitative system for the rapid and accurate segmentation and estimation of colony quality is essential in order to reduce classification errors. Furthermore, removal of the use of fluorescent labeling or other chemical reagents would be helpful in preparing the iPSCs for additional research experiments.
|
review
| 99.6 |
Automated segmentation of stem cell colonies for phase contrast imaging is challenging and requires specialized algorithms to handle the problems of halo artifacts and overlapping of the colony edges with the feeder cells . The currently available image analysis techniques to achieve stem cell colony selection are based on morphological operations, thresholding, and watershed transformation. A combination of these techniques is designed to examine the status of the colony in each individual research [7–9]. Alternatively, other approaches have adopted commercial software tools that basically use filtering, automatic thresholding, and Voronoi algorithms for stem cell segmentation and tracking [1, 5, 10, 11]. In addition, the morphological categories of colonies based on commercial program require manual interpretation to locate the colony area for feature measurement . Because these aforementioned image analysis techniques are quite problem specific and rely strictly on parametric settings, they lack controllability to manipulate variations among the stem cell heterogeneity on a large scale.
|
review
| 99.8 |
Recently, several supervised machine learning approaches have also been developed and their significance in distinguishing stem cell colonies confirmed [12–15]. The approaches designed for the selection of colonies, based on k-nearest neighbor searching with and without error correction output codes , ensemble support vector machine (SVM) , and random forest methods , acquired local features from the patches of original images. In addition, the development of the learning set for discrimination of cells in the colony has to be done manually. However, a high degree of reliability and cost-effectiveness is very important in clinical applications. Recently, many researchers have focused on implementing the convolutional neural network (CNN) for various medical imaging modalities, and its high reliability and validity for object segmentation and detection applications have been revealed [16, 17]. In addition, CNNs have been successfully applied to microscopic cell imaging data, with robust decisions on ambiguous cell classifications found, making the process suitable for clinical applications [18–20]. A deep CNN method for identifying mitosis in the cell nucleus reportedly had higher satisfactory performance than estimation with conventional methods . A deep multiple instance learning-based CNN approach effectively segmented mammalian and yeast microscopy images with remarkable accuracy . However, the above-mentioned CNN approaches on biomedical imaging data differ from our study, as they have adopted deep neural networks for image segmentation tasks. In this study, our target was to train a CNN model to classify colonies on the basis of obtained features of the segmented colony. Most of the aforementioned studies tested colony morphology for estimating colony categories. However, apart from certain quantitative morphological features, textural features are the most important, as they describe the spatial intensity variations of the colony image. Furthermore, textural features are closely connected with cellular characteristics . Hence, this study considered both colony morphological and textural features for the evaluation of iPSCs.
|
study
| 99.94 |
The objectives of this study were 1) to determine, whether or not the proposed feature vector-based convolutional neural network (V-CNN) is the most suitable and best model of colony quality recognition from the morphological and textural features of a segmented colony; 2) to confirm the promising results of colony quality recognition through use of an accurate cross-validation process; 3) to demonstrate the superiority of the proposed deep V-CNN learning approach over the SVM classification system.
|
study
| 100.0 |
The iPSCs were maintained as described previously . For inactive murine embryonic fibroblasts (MEFs) isolation, we used day 13.5 embryos. After the removal of the head, visceral tissues, and gonads, the remaining bodies were washed and dissociated with 0.25% trypsin-EDTA (Sigma-Aldrich, Saint Louis, MI, USA). Ten-million cells were plated on each gelatin-coated 100- mm dish and incubated at 37°C with 5% CO2. The next day, floating cells were removed by washing with PBS. In this study, MEFs were used within passage 4 to avoid replicative senescence. Normal iPSC line (HPS0063) was obtained from the RIKEN Bioresource Center . The harvested colonies were triturated to generate medium-sized small fragments, which were then seeded on new plates together with the mitomycin C-treated MEFs in complete ES medium composed of DMEM (Sigma-Aldrich) supplemented with 20% knockout serum replacement, 5 ng ml−1 recombinant human basic fibroblast growth factor (Peprotech), 20 mM HEPES buffer (pH 7.3), 0.1 mM 2-mercaptoethanol, 0.1 mM non-essential amino acids, 2 mM l-glutamine and 100 U ml−1 penicillin/streptomycin (all other materials were from Gibco). All images were prepared under the 100× objective of the phase contrast microscope in the BioStation CT system, using automatic Z-focus with a resolution of 1360 × 1024 pixels.
|
study
| 100.0 |
In addition this study analyzes the variations of marker expression among iPSC samples using TRA-1-60 and TRA-1-81 antibody (mouse, 1:100, Chemicon, Billerica, MA, http://www.chemicon.com/). The cells were analysed with a laser scanning confocal microscope equipped with Fluoview SV1000 imaging software (Olympus FV1000, Olympus, Tokyo, Japan). It expressed only on the established healthy iPSCs and not on unhealthy iPSCs which is presented in S1 Fig. Hence the variations of marker expression among healthy and unhealthy colonies and morphology were used to label the iPSC samples used in this study.
|
study
| 100.0 |
A block diagram of the proposed automated system is shown in Fig 1. As mentioned above, the system interfaces image analysis methods with the V-CNN model for segmenting colonies in order to compute their morphological and textural features for use in their classification by deep learning architecture. Robust segmentation of the colony region prior to classification is beneficial for automating pluripotency. However, the computerized segmentation of colony regions with feeder cells included is more challenging for the subsequent measurement of stem cell characteristics . In this study, the entire colony image was used for the segmentation of the colony region. In the beginning of the process, median filtering was used as a preprocessing step to reduce the background noise and further to preserve the edges of the stem cell regions. This works on the original image object, replacing the center value of the window with the median value of all neighboring pixel values. A median mask size value of 9 × 9 pixels was applied to the original image.
|
study
| 100.0 |
After the preprocessing step, we used an iterative multiple thresholding algorithm to separate the image pixels into the foreground and background, where the threshold estimation depends on maximization of the between-class variances of the pixel values . This estimates the thresholds iteratively and returns two optimal thresholds. The iteration continues until the errors become small or the thresholds no longer change. We initialized the thresholds t1 and t2 as I/3 and 2I/3, respectively, where I indicates the intensity range of the image. The error functions are represented as ε1(t1,t2)=[m(0,t1)+m(t1,t2)]/2−t1(1) and ε2(t1,t2)=[m(t1,t2)+m(t2,∞)]/2−t2(2) where m(tm,tn)=1tm+tn+1∑h=tmtnw(h).h(3) where w(h) indicates the image histogram. The thresholds t1 and t2 were updated to force the errors ε1 and ε2 toward zero. The updated thresholds are represented as t1′=t1+ε1(4) t2′=t2+ε2(5)
|
study
| 100.0 |
The resultant binary image was then processed using morphological closing and opening operations. It was closed using a disk-shaped structuring element with a radius of 2, and opened using a diamond-shaped structuring element with a distance of 19. The resultant connected components were then filled, and the contours of the objects were smoothed with morphological erosion and hole-filling operations. Furthermore, the unwanted cells around the colony regions, which are smaller than the user-specified threshold, were removed using a size filtering method. We noticed that a size of 9000 pixels was suitable for removing the other regions that surround the colony area. Finally, the resultant segmented colony region was evaluated for further quantitative feature measurements by connected component labeling with eight-neighbor connectivity. This method is generally used to estimate adjacent pixels that share the same set of intensity values . The segmentation results of the healthy and unhealthy colony image are shown in Figs 2 and 3, respectively.
|
study
| 100.0 |
Each colony region was estimated for ten morphological features; namely, area, perimeter, centroid, equivalent diameter, eccentricity, solidity, major axis, minor axis, extent, and orientation. The definitions of these features are described in S1 Table. The textural features adopted in this study can be explained in terms of a gray-level co-occurrence matrix of 13 features (for details, see ), which reveals the different combinations of pixel intensity values in a specific spatial displacement. The most relevant features of colony categories are identified using the feature selection technique. In this study, Fisher score analysis was applied to determine the most relevant features for the subsequent classification task, while excluding the irrelevant ones. Fisher scores were automatically computed for each feature in the feature sets of training data (Fig 4) and used to select the informative features by which the within-class distance is minimized and the between-class distance is maximized . Specifically, given the selected f features, the input data matrix X ∊ Ra×n reduces to Q ∊ Rm×n. Hence, the Fisher score is represented as follows: argmaxtrQ{V˜t−1V˜b}(6) where V~t and V~b are defined as V˜t=∑i=1n(Qi−μ~)(Qi−μ~)L,V˜b=∑k=1cnk(μ~k−μ~)(μ~k−μ~)L(7) where μ~k and nk are the mean vector and size of the kth class, respectively, in the reduced data space; that is, Qμ˜=∑k=1cnkμ˜k is the overall mean vector of the reduced data.
|
study
| 100.0 |
Cen, centroid; Are, area; Ecc, eccentricity; Per, perimeter; Ori, orientation; Maj, major axis; Min, minor axis; Dia, equivalent diameter; Sol, solidity; Ext, extent; D_V, difference variance; Hom, homogeneity; Ene, energy; D_E, difference entropy; Con, contrast; Cor, correlation; Inf_1, information measure of correlation_1; S_A, sum average; Inf_2, information measure of correlation_2; S_E, sum entropy; Ent, entropy; S_V, sum variance; Var, variance.
|
other
| 99.94 |
CNNs are a branch of neural networks that have been implemented successfully in image recognition and classification [16–21]. Although the CNN has been applied for various medical imaging segmentations, it has not been used previously for the input of feature vector-based classifications for colony quality. The selected features of colony morphology and textures obtained from the segmented colony were entered into the V-CNN model in order for the classification task to identify the colony quality. However, input feature vectors cannot be entered directly into the typical CNN. Hence, we added a transfer function from the feature vectors to the virtual image at the front of the CNN model organization. In addition, the parameters of the mapping function were trained to obtain an adequate transfer function for a target classification task of the CNN framework. Hereafter, we briefly explain the mathematical framework of V-CNN and the process of training to implement the classification task. The V-CNN architecture used in this study is arranged by stacking a set of convolutional, transfer function, and pooling layers in an alternate way, as shown in Fig 5. The main work of the convolutional layer is to estimate local conjunctions of features from the input feature vectors and to map their occurrence to a feature map. As a result of convolution in neuronal networks, the feature vectors are partitioned into perceptrons, generating local flexible fields and finally trampling the perceptron into feature maps of size n1 × n2. In each layer, there is a bank of n filters that detect features at every location of the input. The output Ya(q) of layer q consists of n(q) feature maps of size n1(q)×n2(q). The ath feature map, indicated as Ya(q), is computed as Ya(q)=Ga(q)+∑b=1n(q−1)Va,b(q)*Yb(q−1)(8) where Ga(q) is a bias matrix and Va,b(q) is the filter of size 2t1(q)+1×2t2(q)+1, connecting the bth feature map in layer (q − 1) with the ath feature map in the layer. The weights of these filters and their values are altered throughout the training to reduce the classification error on any training data. The next operation is to apply the transfer function to produce a set of feature maps. This helps the classifier to build nonlinear decision boundaries. The selection of an activation function has a strong influence on the computational costs of both training and validation performances. Hence, in this study, we chose the rectified linear unit (ReLU) as a transfer function, which is many times faster than the other activation functions. It is defined as Ya(q)=max(0,Ya(q−1))(9)
|
study
| 100.0 |
The third operation is the max-pooling layer, which partitions the input feature vectors into a set of non-overlapping rectangles and returns the maximum value of each such rectangular feature set. Furthermore, it significantly reduces the input size and number of network parameters, hence controlling CNN overfitting. It is usually implemented after multiple stages of convolutional and nonlinearity layers in order to minimize both the computational requirements as well as the likelihood of overfitting. The max-pooling layer q has two hyper parameters: the spatial extent of the filter F(q), and the step size S(q). The pooling layer describes a window of size F(q) × F(q) and minimizes the data within this window to a single value. Similarly to the convolutional layer, the window is moved by S(q) positions after each operation. The minimization of the data is repeated at each position of the window until the entire activation volume is spatially reduced. In this study, we evaluated max pooling with a 2 × 2 window, using a step size of 2. The output feature maps of these operations can then be fed as the input to another round of the same three operations (convolutional, ReLU, max-pooling layers). The last operation for final classification is to create fully connected layers of mono-dimensional features, which are designed to map the activation volume from the fusion of previous different layers to a class probability distribution. If (q − 1) is a fully connected layer, then it is defined as ya(q)=f(za(q))(10) where za(q)=∑b=1n(q−1)wa,b(q)ya(q−1)(11)
|
study
| 100.0 |
The purpose of the complete fully connected structure is to tune the weight parameters wa,b(q) to produce a stochastic likelihood representation of each class found on the activation maps created by the combination of convolutional, ReLU, and pooling layers. The repeated implementation of these two operations generates an output vector of class scores, which assists as the classification prediction. In addition, a cost function is employed to reduce the classification error. In this study, we applied a soft-max cost function in order to generate a probability output in the range of 0 to 1 that can automatically be converted to class values. We implemented training and testing of the V-CNN in Python, using the Keras, TensorFlow, NumPy, SciPy, and Scikit-learn Python packages [30–33]. The training data were partitioned into fixed batch sizes (10) of the input feature vectors. All the batches of input feature vectors were evaluated in 20 epochs, which means that the procedure ran 20 times on the entire training data set. The V-CNN was trained with the TensorFlow framework at a learning rate of 0.001, using the Adam optimizer for cross-entropy minimization. The scale of the samples were evaluated using train_test_split () function and random_state parameter with different seed values to ensure the reproducibility of the classification performance of the model. The accuracy and loss values were evaluated to show the fitness of the model. The performance of the proposed V-CNN model for classifying the colony quality was estimated using the morphological feature, textural feature, and combined morphological and textural features (hereafter referred simply to as “combined features”).
|
study
| 100.0 |
The most relevant features of colony morphology and textures determined by Fisher scores using training data (40) were involved in the classification of the iPSC colony quality. On the basis of the Fisher scores, the features with the lowest score were not considered as the most relevant for quality estimation in this study. The Fisher score of each feature of colony morphology and textures for discriminating the quality is presented in S2 and S3 Tables, respectively. Only features that had a Fisher score higher than a certain threshold (e.g., 0.450) were kept, whereas others that showed a less discriminant effect on the classifier were removed. Under the colony morphology feature, the equivalent diameter, minor axis, major axis, solidity, and extent were selected as being the most relevant features, showing higher Fisher scores than the other features. Under textures, the sum variance, variance, sum entropy, entropy, sum average, and information measures of correlation_2 were selected as the most relevant features. The ranges of values of these features for the healthy colonies were higher than those for the unhealthy colonies are depicted in Fig 6. The potential of each individual morphological and textural feature in distinguishing the iPSC colonies was examined from the area under the curve (AUC), using receiver operating characteristic curve analysis (NCSS 11 Statistical Software, Kaysville, UT, USA). The performance of each individual morphological and textural feature (estimated through AUC values) in distinguishing the colonies is summarized in Table 1. Among the morphological features, solidity outperformed the rest with an AUC value of 0.878 ± 0.03 and a confidence interval (CI) of 0.747–0.933 (Fig 7). Among the textures, variance (AUC = 0.859 ± 0.03, CI = 0.741–0.926) slightly outperformed the sum entropy and sum variance in distinguishing the colonies (Fig 8).
|
study
| 100.0 |
In order to analyze the classification performance, the selected features were inserted into the V-CNN model to distinguish the healthy and unhealthy colonies of iPSCs. Since feature vectors cannot be entered directly into the CNN network, we added a transfer function from the feature vectors (11) to the virtual image at the front of the CNN organization. We found that (28 × 28) size of the virtual image was optimum by experiment to produce better results in this study. This was then entered through a stack of two-dimensional convolutional layers of 32 filters with convolution kernel sizes of 3 × 3 throughout the operation of the network. The training process of the model was performed using a labeled dataset of input feature vectors. The testing size of 0.33 and random_state with seed value of 9 was observed to produce best performance of the proposed model. The total dataset was then divided into 60 for training and 30 for testing the model. The V-CNN model had a higher capacity in classifying the quality of colonies, as indicated by its high accuracy and low loss values (Fig 9). The estimated loss values for the morphological, textural, and combined features were low (0.209, 0.285, and 0.202, respectively), implying the behavior of the model after 20 iterations of optimization. In addition, the performance of the V-CNN model was further compared with that of the SVM classifier, using the radial basic function kernel. Quadratic programming was applied to optimize the parameters of the SVM model and the program used in the experiments was implemented using Scikit-learn toolkit . The hyper-parameters γ that control the capacity of the kernel and C, the regularization parameter [34, 35] were determined by using cross-validation. The combination of parameters were observed to produce better results at γ = 2 and C = 1.
|
study
| 100.0 |
The performance of the V-CNN model in determining the quality of iPSC colonies on the basis of various feature sets is presented in Table 2. The accuracy of the morphological features in assessing the colony quality was slightly higher than that of the textural features. In addition, with the V-CNN model, the accuracy of the morphological (95.5%), textural (91.0%), and combined (93.2%) features in determining the quality of colonies was higher than that of those features (86.7%, 83.3%, and 83.4%, respectively) used in the SVM classifier. Furthermore, to validate the performance of the proposed model, precision, recall, and F-measures were used . Precision demonstrates the number of positive predictions divided by the total number of positive class values predicted, and is defined as Precision=∑i=1c(TPi)∑i=1c(TPi+FPi)(12)
|
study
| 100.0 |
F-measure conveys the weighted harmonic mean of the precision and recall, and is defined as F−measure=2×Precision×RecallPrecision+Recall(14) where c is the number of classes, and TP, FP, and FN represent the number of true positives, false positives, and false negatives, respectively. TP indicates when the model predicts the ith class label as “(healthy colony)” and the ith ground truth class label is likewise “(healthy colony).” FP indicates when the model predicts the ith class label as “(healthy colony)” but the ith ground truth class label is “(unhealthy colony).” FN indicates when the model predicts the ith class label as “(unhealthy colony)” but the ith ground truth class label is “(healthy colony).”
|
other
| 69.56 |
The precision, recall, and F-measure values generated in the current study with the V-CNN model were high that indicated the fewer numbers of false positives and false negatives than those generated by the competing SVM classifier (Table 2). Furthermore, the reliability and generalization of the proposed V-CNN model were investigated using a five-fold cross-validation method [30, 31], which functions by partitioning the datasets into k parts (k = 5). The partitioned data are represented as a fold. The method was trained on k–1 folds with one held back, and tested on the held back fold. This was continued five times separately, applying different members of the training and testing data that possess compositions different from those of the other experiment. The mean value of these five different compositions of classification performance was evaluated and considered as the overall accuracy of the model. The experimental results of the five-fold cross-validation of the performance of the V-CNN and SVM classifiers in determining the quality of iPSC colonies on the basis of various feature sets are presented in Table 3. The overall accuracy results using five-fold cross-validation for the V-CNN were much higher than those of the SVM, which was more than 10%. Similarly, five-fold cross-validation of the performance of the two tested models in evaluating the precision, recall, and F-measures produced high values in the range of 85–89% for V-CNN and very low values in the range of 68–83% for SVM, indicating the robustness and effectiveness of the proposed V-CNN approach in determining the quality of colonies.
|
study
| 100.0 |
This study has proposed a new automatic system that interfaces image analysis methods with the V-CNN model for the segmentation and classification of phase contrast microscopy images, using the morphological and textural features of iPSC colonies. To our best knowledge, this is the first study to have designed a vector-based deep CNN for classification of iPSC colony quality. The motivation for implementing the V-CNN in this study was to investigate the suitability of this classifier model in accomplishing the classification of feature vectors of healthy and unhealthy colonies, which is the main contribution of this study. The V-CNN model classifier had a higher discriminant ability with colony morphologies than with colony textures. The model revealed highly acceptable classification accuracy (95.5%) compared with the SVM classifier (86.7%) that was implemented for similar feature vectors in this study. The overall accuracy using five-fold cross-validation with V-CNN and SVM was <90% and 75–77%, respectively. Furthermore, the precision, recall, and F-measure values of the V-CNN model were much higher than those of the SVM classifier. The differences in performance could be due to the transfer of the prior knowledge of feature maps of the V-CNN model across the stacked layers of the network capable of minimizing the classification error. Furthermore, the local shared weight capacity of the neurons drastically reduced the network complexity and number of parameters, thus contributing to the robustness of the performance in classifying the stem cell colonies.
|
study
| 100.0 |
There are several machine learning techniques that are preferred for training algorithms to classify the colonies of stem cells [7–15]. However, unstable classification performances have been demonstrated on colonies that exhibit deformable and changeable morphologies. In other studies using k-nearest neighbor classifier, multiclass quality evaluations of iPSCs based on local features of the colony reportedly had the highest classification accuracy of 62.4% . Other detection methods based on the colony morphology of the stem cells, computed on overlapping blocks of images, revealed moderate accuracy (80%) by means of the linear SVM classifier . A method developed for embryonic stem cell colony segmentation and tracking, using dynamic and morphological features on time data by various machine learning algorithms, revealed both low and high accuracies . However, our current study with deep V-CNN executing batch-based feature vectors of colony morphology has reduced the computational complexity and produced a stable classification performance with an accuracy of 95.5% in discriminating the colonies. In addition, the cross-validated accuracy of the V-CNN classifier (92.4%), as evaluated on the basis of morphological features, was much higher than that of the SVM model (75.2%) in this study. The difference between these results is reasonable, because the proposed system with V-CNN architecture has the advantage of a broader integrated structural complexity to effectively handle a certain degree of variations of features of stem cells and hence produce excellent performance in classifying the colonies, unlike other methods.
|
study
| 100.0 |
Several recent studies have demonstrated the highly significant contributions of CNN toward microscopic cell image segmentation compared with other conventional methods [19, 37]. A deep CNN for bacterial colony segmentation demonstrated its superior performance over the SVM method, similar to our study . Bacterial colony enumeration using CNN obtained improved precision and recall values that were almost similar to those obtained in the current study . However, these aforementioned studies used CNN for the segmentation of image objects, whereas our current study incorporated CNN for the classification of feature vectors of images, and hence the results are strictly speaking not directly comparable. Most of the previous studies considered colony morphology to be the most important criterion for estimating colony categories [5, 8, 38]. In the study by Zahedi et al. , area was the best individual feature for colony discrimination. Contrary to that study, another report found the shape-based solidity feature to have the strongest ability in classifying stem cells, similar to what was found in our current study . However, as the clinical target, feature descriptors, and classifier models of those studies were different from ours, the studies might not be directly comparable. The limitation of our study was the small number of training data used to build the classifier model. Further studies with a much larger number of training data, with various objectives (10×, 20×, and 40×) of iPSC colony images as well as live cell imaging evaluation, should be used to evaluate the performance of the proposed V-CNN model. Furthermore, the present model analyzed gray-level co-occurrence matrix-based texture features for quality determination. In future, the usefulness of other textural features, such as discrete wavelet and geometric moment based analysis, could be considered to enhance and generalize the proposed model.
|
study
| 99.94 |
In conclusion, our newly proposed framework of interfacing image processing methods with the V-CNN model produced encouraging results in determining the iPSC colony quality. Although the CNN has been applied before for microscopic cell image segmentation, this is the first implementation for input feature vector classification of colony quality. The suitability of the V-CNN model for addressing the classification task has been successfully shown, revealing it to have higher classification accuracy than that of the competitive SVM classifier. The proposed V-CNN-based colony identification system has been experimentally tested and cross-validated to be the most optimal model. Additionally, the proposed approach does not require much computational resources, and reduces on architectural and computational complexities, and thus it can be implemented as a valuable tracking technique in a real-time classification system. Overall, our experimental results indicated that the proposed deep V-CNN approach can allow the accurate, rapid detection of colony quality, outperforms the state-of-the-arts, and thus it can be a promising decision support model for clinical applications.
|
study
| 100.0 |
The iridoid glucoside antirrhinoside (Fig. 1) makes up several percent of the dry weight of the common ornamental plant snapdragon (Antirrhinum majus) (1–3). In general, iridoids such as antirrhinoside mediate important plant–insect and insect–insect interactions. Plants appear to harness iridoid glucosides to deter herbivores. The herbivore is affected by the toxic dialdehydes liberated by deglycosylation of the iridoid in the injured plant tissue or insect gut (4, 5). Some herbivorous insects can sequester iridoid glucosides and exploit the toxic effect for their own defensive systems (6). Additionally, many iridoid glucosides are believed to have beneficial health properties for humans. Foods such as olives may owe some of their health-promoting properties to iridoid ingredients with anti-inflammatory (7), antimicrobial, and anticancer effects (8).
|
review
| 99.9 |
Proposed epi-ISY step in antirrhinoside biosynthesis. The biosynthesis of antirrhinoside has been elucidated based on deuterium labeling studies (15–18). Antirrhinoside biosynthesis requires a configuration of the nepetalactol precursor (blue arrow) different from that found previously with CrISY (red arrows). We hypothesize that an epi-ISY performs the synthesis of 7-epi-nepetalactol in A. majus. The CrISY reaction involves hydride transfer from NADPH to generate an enolate intermediate that then cyclizes to various configurational isomers of nepetalactol and iridodial. For the major cis—trans nepetalactol and cis—trans iridodial product of CrISY, the stereochemical nomenclature and atom numbering are shown in purple.
|
study
| 100.0 |
Because iridoid glucosides are apparently not directed against a specific molecular target, we hypothesize that no specific selection pressure acts to limit the structural diversity of iridoids. On the contrary, in a race of arms with herbivore β-glucosidases evolving away from toxic iridoid glucoside specificity, structural diversity of the iridoid glucoside protoxin may be strongly favored. Accordingly, the usual scope of action of plant secondary metabolism, hydroxylations, acylations, and glycosylations, gives rise to chemotaxonomic variability of the iridoid scaffold down to the species level (9, 10). Additional structural diversity originates from configurational variations of the iridoid core scaffold, which has a fused five- and six-membered ring with multiple stereocenters (Fig. 1).
|
study
| 100.0 |
There has been extensive structural and mechanistic investigation of iridoid synthase (ISY),3 the enzyme that creates this core bicyclic scaffold (Fig. 1), from the medicinal plant Madagascar periwinkle (Catharanthus roseus). These studies have revealed how the iridoid core cyclizes after transfer of a hydride from NADPH to the linear precursor 8-oxogeranial (11–13). In Catharanthus, only iridoids with the stereocenter C7 fixed in the S configuration and the ring fusion in the configuration commonly referred to as “cis–trans” (Fig. 1; hydrogens cis at C4a and C7a and trans at C7a and C7) are observed. The hydride transfer step catalyzed by iridoid synthase accounts for the configuration at C7. “Epi-iridoids” with an inverted methyl group at C7 (e.g. antirrhinoside, catalpol, epi-loganic acid, penstemoside) are common in Plantaginaceae (10, 14). A detailed biosynthetic hypothesis for 7-epi-iridoids has been developed in Plantaginaceae based on deuterium labeling. Only deuterated 7-epi-deoxyloganic acid, but not deoxyloganic acid, was incorporated into iridoids in Scrophularia racemosa, Plantago major, and Buddleja davidii. Although the C7 stereocenter is removed and reinstalled in later biosynthetic steps, these studies strongly suggest that 7-epi-nepetalactol (C7-R) is the productive iridoid intermediate (Fig. 1) (15–18). We hypothesize that an epi-iridoid synthase that reduces C7 of 8-oxogeranial with R preference is involved in the biosynthetic pathway. We searched for a homologue of iridoid synthase that performs the R-selective reduction of the iridoid precursor 8-oxogeranial. Here we identify the iridoid synthase from A. majus (AmISY), which displays epi-iridoid synthase activity.
|
study
| 100.0 |
Candidates for AmISY were identified based on sequence homology to ISY from C. roseus (CrISY) in a genome sequence of the JI7 inbred line of A. majus (http://snapdragon.genomics.org.cn/).4 The protein sequence of CrISY was used in a BLAST search against proteins predicted from the genome sequence to yield four hits with amino acid sequence identities between 39% and 66% (Fig. 2a). Candidate Am18679 showed the highest amino acid similarity to CrISY, with 66% identity and 79% similarity. For overexpression in Escherichia coli, all four genes were cloned from cDNA of A. majus flower and leaf tissue and successfully purified via nickel affinity chromatography. Enzyme reactions containing 8-oxogeranial and NADPH as substrates were analyzed by GC-MS. Only protein derived from candidate Am18679 (Fig. 2b), the candidate most similar to CrISY, yielded sizeable quantities of cyclized iridoid product (supplemental Fig. S1). Therefore, candidate Am18679 was named AmISY. Only trace amounts of substrate were consumed, and negligible products were detectable with the more distantly related candidates.
|
study
| 100.0 |
Identification of AmISY. a, phylogenetic tree of iridoid synthase homologues in A. majus (Am), Olea europaea (Oe), C. roseus (Cr), Nepeta cataria (Nc), and N. mussinii (Nm). The neighbor joining tree was built from a MuscleWS alignment using the BLOSUM62 similarity matrix in Jalview 2.10.1 (24). Numbers next to the nodes indicate evolutionary distances. Proteins with proven iridoid synthase activity are highlighted in bold font. One of the A. majus homologues (AmISY or Am18679) groups closely with the iridoid synthases from O. europaea and C. roseus. b, SDS-PAGE of nickel affinity- and gel filtration chromatography–purified AmISY. c, the 8-oxogeranial–dependent NADPH consumption of AmISY showed catalytic parameters close to those of CrISY at a fixed NADPH concentration of 50 μm (AmISY: kcat = 0.72 ± 0.02 s−1, Km = 1.1 ± 0.1 μm; CrISY: kcat = 1.6 ± 0.1 s−1, Km = 4.5 ± 0.2 μm). Values are given as the mean ± S.D. of two independent measurements with different batches of protein. d, qRT-PCR shows tissue-dependent expression of ISY homologues in A. majus. Abundance of the Am29566 transcript was too low for quantification in all tissues. Expression values are given as the mean ± S.D. (four reactions). Each gene was separately normalized to the tissue with the highest expression level. Two replicates each were analyzed for two independent samples of cDNA.
|
study
| 100.0 |
Efforts to directly test the physiological relevance of AmISY were not successful because silencing systems in A. majus using virus-induced gene silencing are highly inefficient (19, 20). However, our hypothesis that AmISY is the physiologically relevant iridoid synthase in A. majus is corroborated by the steady-state kinetic parameters (Fig. 2c; kcat = 0.72 ± 0.02 s−1 and Km = 1.1 ± 0.1 μm), which are similar to those measured for CrISY, for which the physiological role has been confirmed by gene silencing (11). qRT-PCR of AmISY (Fig. 2d) with cDNA from A majus root, leaves, and flowers further indicated that AmISY is highly expressed in leaves and not expressed in roots and flowers. Although antirrhinoside is found in all A. majus tissues, the compound could be synthesized exclusively in leaves and then distributed throughout the plant. Phloem mobility of antirrhinoside has been demonstrated (2).
|
study
| 100.0 |
To investigate the stereoselectivity of the hydride transfer catalyzed by CrISY and AmISY, we initially analyzed reactions with model substrates lacking the 8-oxo group. These substrates can undergo enzymatic reduction, but the missing aldehyde moiety prevents subsequent cyclization (supplemental Fig. S2). With commercially available citral, a mixture of geranial (E-isomer) and neral (Z-isomer), as a substrate, CrISY yielded exclusively S-citronellal in a stereoconvergent fashion. In contrast, AmISY produced a 6:4 mix of R- and S-citronellal (supplemental Fig. S2c). To more closely reflect the structure of the physiological ISY substrate 8-oxogeranial, we synthesized geranial with low neral content (2.5%, supplemental Fig. S2a) by oxidation of geraniol with Dess-Martin periodinane. With this substrate, AmISY showed high stereoselectivity (supplemental Fig. S2d, 89% R). The R selectivity of AmISY observed here strongly supports the proposed biosynthesis of antirrhinoside via R-selective reduction of 8-oxogeranial. AmISY therefore appears to be the first example of an epi-iridoid synthase.
|
study
| 100.0 |
To more rigorously assess the stereoselectivity of the two enzymes, AmISY and CrISY were assayed with the physiological substrate 8-oxogeranial. Analysis of this reaction is complicated by the fact that the product profile consists of a mixture of products. In in vitro assays, both nepetalactol and the open-form iridodials were observed, plus reduced, uncyclized product (Fig. 3). Additionally, small amounts of unidentified compounds were also observed. Before analysis of the stereoselectivity of the AmISY reaction with 8-oxogeranial, a method for resolving all components of the enzymatic reaction was developed, and all minor components of the ISY reaction were identified.
|
study
| 100.0 |
Chiral GC-MS analysis of CrISY and AmISY products. a, synthesis of nepetalactol and iridodial standards by DIBAL reduction of nepetalactones (11, 21, 24). b, reaction products of CrISY and AmISY with 8-oxogeranial substrate were analyzed on a chiral GC-MS column. The intensity axis of all chromatograms was normalized to the tallest peak. In the CrISY reaction, ten products could be identified (red) in comparison with authentic standards. AmISY products, which have the opposite chirality (blue), were matched to the CrISY products based on electron impact fragmentation spectra (supplemental Fig. S3). Product 8′ is presumably hidden under a larger peak in the AmISY chromatogram. The matching of CrISY and AmISY spectra (left inset) and the matching of CrISY and standard spectra (right inset) was verified by calculating pairwise similarity scores for all combinations, where a score of 1 signifies identity. The presence of similarity scores close to one on the diagonal confirms the peak assignment. c, circular dichroism spectra of 7S-cis-trans nepetalactol standard (black, 2 mm in hexane) and of the extracted CrISY (red) and AmISY (blue) reaction products from a reaction conducted in water without buffer. Water without buffer was used because buffer resulted in an attenuated CD signal.
|
study
| 100.0 |
To do this, authentic standards of the side products needed to be prepared. Mass spectrometry data strongly suggested that the minor products of the CrISY reaction were alternative stereoisomers of nepetalactol and iridodial. Standards of 7S-nepetalactol isomers can be obtained via diisobutylaluminum hydride (DIBAL-H) reduction of the cis–cis, cis–trans, trans–cis, and trans—trans nepetalactone isomers (Fig. 3a) (11, 21–23).5 However, the trans–trans isomer could not be isolated in sufficient quantities from plants and was instead generated by base-catalyzed isomerization of the cis–cis isomer, an inefficient uphill process with only 10% yield (21). Both nepetalactones with the ring fusion in trans configuration open directly to the corresponding iridodials (21) because of instability of the strained ring. Iridodials in cis–cis and cis–trans configuration were obtained by incubating the respective nepetalactols in 100 mm HCl overnight. The stereocenter at C1 in nepetalactol, which equilibrates in aqueous solution (21), and the stereocenter forming at C4 upon conversion to iridodial were not resolved. Each of these standards could be separated on achiral and chiral GC-MS columns.
|
study
| 100.0 |
With an analytical method and authentic standards for the 7S stereoisomers in hand, the product profile of CrISY was assigned. In addition to cis–trans nepetalactols and cis—trans iridodials, the expected on-pathway intermediates for iridoid biosynthesis in Catharanthus, a number of other nepetalactol and iridodial diastereomers were observed. According to integrals of GC-MS peaks, combined cis–trans, trans–trans, trans—cis, and cis–cis species make up ∼69%, 21%, 5%, and 5% of the cyclized reaction products, respectively, under these in vitro assay conditions (Fig. 3b). Additionally, a substantial percentage of the entire product mix is reduced, non-cyclized S-8-oxocitronellal, as reported earlier (23%) (11).
|
study
| 100.0 |
Having assigned all components of the enzymatic reaction, the spectra of AmISY and CrISY were compared. AmISY and CrISY reactions analyzed by GC-MS using a standard achiral column gave virtually identical chromatograms (supplemental Fig. S1). Chiral GC-MS, however, revealed substantial differences between the CrISY and AmISY product profiles (Fig. 3b).
|
study
| 100.0 |
Because enantiomers should have identical mass spectra, the diastereomers that were structurally identified in the GC-MS analysis of CrISY products could be matched to the corresponding AmISY enantiomers via the characteristic EI fragmentation spectra (Fig. 3b and supplemental Fig. S3). These spectra strongly suggest that CrISY and AmISY both generate a mixture of diastereomers but that the products of AmISY are exact mirror images of the CrISY products (Fig. 3b).
|
study
| 100.0 |
To further substantiate this hypothesis, CD spectra were obtained for the enzymatic product of CrISY and AmISY. As we predicted, the spectrum of the CrISY product showed an opposite sign compared with the AmISY spectrum, providing further support for the hypothesis that the AmISY product is enantiomeric to the CrISY product.
|
study
| 100.0 |
The analysis of the CrISY diastereomeric profile, as described above, revealed that the majority of the product forms the cis–trans isomer, or 4aS,7S,7aR. The stereochemistry of this isomer matches that of the downstream iridoid products in C. roseus. The major isomer found in the AmISY product profile must then correspond to 4aR,7R,7aS (cis–trans), which is the enantiomer of 4aS,7S,7aR. However, downstream A. majus iridoids are derived from the 4aR,7S,7aR isomer (cis–cis), which is found in only ∼5% of the cyclic AmISY products. Therefore, although AmISY generates the correct stereochemistry at the C7 position, the required cis–cis isomer is not the major product.
|
study
| 100.0 |
To rationalize how AmISY generates the opposite stereocenter at C7, a homology model of AmISY was constructed. The homology model was calculated on the SwissModel server based on the CrISY structure in complex with geranic acid (PDB code 5DF1). There is a high level of amino acid similarity (79%) between AmISY and CrISY, so it is likely that the model accurately reflects the AmISY active site structure.
|
study
| 99.94 |
The Lys-146 and Tyr-178 residues (Fig. 4) that are conserved in ISY homologues (25) and other short-chain dehydrogenases (26) are also present in AmISY. However, compared with CrISY, AmISY shows several large mutations in the 8-oxogeranial binding pocket, most notably A246W and F342L. Previously investigated iridoid synthases, OeISY from Olea (27), NISY from Nepeta,5 and Catharanthus homologues (28), resemble CrISY at these positions (supplemental Table S1), suggesting that these amino acids are at least partially responsible for the altered stereoselectivity in AmISY. However, we note that there are large phylogenetic distances covered by these enzymes.
|
study
| 100.0 |
Shown are structural differences in the active sites of CrISY (A) and AmISY (B). A homology model of AmISY was constructed using the previously reported crystal structure of CrISY (PDB code 5DF1, Ref. 25) in complex with geranic acid (GEA, pink sticks) and NADP+ (cyan sticks). All binding pocket residues mutated in an attempt to invert the specificity of CrISY (supplemental Table S2) and the catalytic, conserved Tyr-178 are shown as sticks. The protein backbone is shown as a light gray tube.
|
study
| 100.0 |
In an attempt to graft the R selectivity of AmISY onto CrISY, we produced a series of CrISY mutants incorporating these sequence changes of AmISY (supplemental Table S2). In the construct CrISY-R1, two mutations, A246W and F342L, increased R-citronellal production from not detectable to 1%. Substitution of two additional residues (I345V and A346V) increased the fraction of R-citronellal 17-fold. Another mutation added to CrISY-R1 (F149W) achieved a 7-fold increase in R product. Either way, the high stereoselectivity of AmISY was not attained. Additional sequence changes, perhaps at second-shell residues, must be required to guarantee selective hydride transfer.
|
study
| 100.0 |
Epi-iridoids are pervasive in the Plantaginaceae as well as numerous other plant families (such as Lamiaceae (29), Rubiaceae (30), Orobanchaceae (31), and Paulowniaceae (29)). With several hundred epi-iridoid–derived structures reported, AmISY may become a reference point for the identification of epi-ISY enzymes in these pathways. We predict that these epi-synthases can be identified by inspection of the distinct active site residues (Trp-149, Trp-246, Leu-342, Val-345, and Val-346) identified by the AmISY homology model and by mutation.
|
study
| 100.0 |
The generation of the S configuration at C7 is well explained by the preference of CrISY to transfer a hydride from the pro-S face to 8-oxogeranial, as demonstrated in biochemical assays (11), and can be structurally rationalized by the crystal structure of the enzyme bound to a geranic acid inhibitor (25). We hypothesize that the active site residues of AmISY are at least partially responsible for binding the 8-oxogeranial substrate in the opposite orientation (Fig. 4), thereby changing the stereochemistry of hydride addition to generate the R stereocenter. The newly discovered R-selective cyclization by AmISY may offer valuable biocatalytic access to a larger range of poorly accessible iridodial- and nepetalactol-related synthons (32).
|
study
| 100.0 |
Enzymes typically exert tight control over the course of a reaction by embedding the transition states and intermediates in a precisely tailored binding pocket that allows no other than the desired orientation of the reactive groups. However, in a few notable cases, reactions in nature are (partially) uncatalyzed (33). For example, in certain cationic cyclizations of terpenes, the role of cyclases has been argued to be limited to generation and protection of the cationic intermediates without full control of the cyclization process (34). Given the mixture of products that result from CrISY and AmISY under in vitro reaction conditions, the cyclization half-reaction of iridoid synthase may be another such example of an uncatalyzed reaction. We hypothesize that the high selectivity of the reduction step and the relatively poor selectivity of the cyclization step in iridoid synthases are mechanistically best explained by enzymatic, stereoselective reduction followed by cyclization in an achiral environment outside the active site.
|
study
| 100.0 |
Two observations support this scenario. First, the cyclization step is not only relatively poorly controlled, but the product ratio is also insensitive to mutations with large impact on the overall NADPH consumption rate (see the supplemental information of Ref. 25). Second, the enantiomeric enolate intermediates produced by CrISY and AmISY (Fig. 1) constitute a sensitive probe for the environment in which cyclization happens. Any selectivity imposed on the cyclization by a chiral protein binding pocket would be highly unlikely to affect these enantiomeric enolates and the transition states for the observed product range in the same fashion. However, the products of both CrISY and AmISY appear to be exact mirror images, as shown by the superimposable achiral GC-MS chromatograms and CD signals of opposite sign.
|
study
| 100.0 |
The metabolic role of AmISY in the plant is supported by the high expression levels in leaves, the physiologically relevant catalytic parameters, and the R selectivity that is consistent with earlier feeding experiments. Moreover, no other iridoid synthase homologs from A. majus were highly active in vitro. However, if we make the reasonable assumption that AmISY is the metabolically relevant enzyme, then the mechanism by which the diverse AmISY products are channeled into the pathway of the abundant natural product antirrhinoside remains an unsolved problem. Biosynthesis of antirrhinoside requires the C7-R stereochemistry, which is indeed produced by AmISY. However, antirrhinoside also requires the cis—cis diastereomer. In contrast, the major product of AmISY is C7-R-cis-trans, whereas C7-R-cis-cis makes up only a few percent of the product.
|
study
| 100.0 |
It is likely that additional enzymes are required to isomerize the major AmISY product so that it can be diverted into the antirrhinoside biosynthetic pathway. For example, the trans—trans iridodial could be converted into the more stable cis—cis iridodial by an epimerase that abstracts the labile C7a-H proton next to the C1-carbonyl (Fig. 1). Alternatively, iridoid synthase could utilize a helper protein to control the stereochemistry of cyclization. This question is not unique to the antirrhinoside pathway. In Nepeta species, a variety of nepetalactones with varying stereochemistry at the C4a and C7a carbons are observed. A species of Nepeta mussinii that exclusively produces the trans–cis iridoid isomer as a final product has an iridoid synthase that predominantly produces the cis–trans isomer. This mismatch also strongly suggests that additional enzymes are required to set the stereochemistry at the iridoid bridgehead carbons in Nepeta.5
|
study
| 100.0 |
Iridoids show a vast array of stereochemical variation in their core scaffold. This stereochemical variation is essential for the structural diversity and range of biological activities found in this class of compounds. How iridoid stereochemistry is controlled during the course of biosynthesis still remains cryptic. However, the discovery of AmISY clearly demonstrates that an alternative class of iridoid synthase is responsible for setting the stereochemistry of C7 by controlling the reduction of the 8-oxogeranial substrate. The discovery of AmISY provides insight into how nature controls the stereochemistry of this important class of compounds.
|
study
| 99.94 |
RNA was isolated from A. majus tissue and purified using the RNeasy Plant Minikit (Qiagen) before reverse transcription using the SuperScript III reverse transcriptase kit (Thermo Fisher Scientific), following the protocols of the suppliers. The candidate genes were PCR-amplified from the cDNA using gene-specific oligonucleotides (supplemental Table S3). In a second PCR reaction, the fragments were endowed with complementary overhangs for subsequent InFusion cloning (Clontech Laboratories) into the pOPINF expression vector (Addgene, 26042) (35).
|
study
| 99.94 |
The candidates were expressed in N-terminally His-tagged form using soluBL21 (DE3) (Genlantis) cells transformed with pOPINF plasmids carrying the desired construct and purified similar to a procedure published previously (25). A flask containing 1 liter of yeast extract and Tryptone medium and 50 μg/ml carbenicillin was inoculated with 1 ml of overnight culture of the expression strain and incubated at 37 °C until the A600 reached 0.6–0.8. The temperature was then reduced to 18 °C, protein production was induced by adding 0.25 mm isopropyl 1-thio-β-d-galactopyranoside, and incubation was continued for 16–20 h. Cells were harvested by centrifugation at 4000 × g for 20 min at 4 °C and resuspended in 50 ml of buffer A (50 mm Tris-HCl, pH 8.0, 50 mm glycine, 5% v/v glycerol, 500 mm NaCl, 20 mm imidazole, and 1 mm β-mercaptoethanol) containing 0.5 mg/ml lysozyme and one tablet of Complete EDTA free protease inhibitor (Roche). Cells were disrupted by sonication on ice for 7 min (2-s sonication, 3-s break). Cell debris was removed from the lysate by centrifugation at 35,000 × g for 20 min at 4 °C. The supernatant was injected on a His-Trap 5-ml nickel affinity column attached to an Äkta purifier (GE Healthcare). Protein was eluted with buffer A containing 500 mm imidazole. Iridoid synthase–containing fractions were pooled, concentrated, and washed with size exclusion buffer B (20 mm HEPES, pH 7.5, 150 mm NaCl, and 1 mm β-mercaptoethanol) in an Amicon centrifugal filter (Millipore) with 30-kDa molecular mass cutoff. For further purification, the protein was loaded onto a HiLoad 16/600 Superdex 200 pg (GE Healthcare) size exclusion column and eluted with buffer B. Protein concentration was determined in triplicate on a Nanodrop spectrophotometer (Thermo Fisher Scientific) using absorbance at 280 nm and calculated extinction coefficients (ExPASy ProtParam; Am18679, 99,350 m−1 cm−1; Am18685, 93,390 m−1 cm−1; Am26155, 101,870 m−1 cm−1; Am29566, 81,360 m−1 cm−1). Protein was flash-frozen in liquid nitrogen and stored at −20 °C until further assays were performed.
|
study
| 100.0 |
The catalytic parameters kcat and Km of AmISY were determined by spectrophotometrically measuring the initial rate of NADPH consumption at 340 nm and 25 °C on a Lambda35 (PerkinElmer Life Sciences) spectrophotometer. Reactions were conducted in plastic cuvettes with 1-cm path length and contained 20 nm AmISY in buffer C (200 mm MOPS, pH 7.0, and 100 mm NaCl), 50 μm NADPH (Sigma, N7505), 0.66–5 μm 8-oxogeranial substrate, and 1% THF in a total volume of 800 μl. The substrate 8-oxogeranial was synthesized as described previously from geranyl acetate (11), stored as a 50 mm stock solution in inhibitor-free tetrahydrofuran at −80 °C, and diluted to the appropriate concentration in water. Reactions were started by addition of enzyme. The background rate before addition of enzyme (2.52 10−6 optical density/s) was subtracted, and initial velocities were calculated using the extinction coefficient of NADPH (6220 m−1 cm−1). Catalytic parameters were calculated in Kaleidagraph 4.0 by nonlinearly fitting a plot of the initial velocities versus substrate concentration to the Michaelis–Menten equation.
|
study
| 100.0 |
To quantify expression levels, RNA was isolated from samples of leaf, flower, and root tissue of two A. majus plants using the RNeasy Plant Minikit (Qiagen). cDNA was prepared from 1 μg of total RNA using the iScript cDNA synthesis kit (Bio-Rad). qRT-PCR was performed on a CFX96 real-time PCR detection system (Bio-Rad) using SSO Advanced SYBR Green Supermix (Bio-Rad). For each of the four candidates, gene-specific oligonucleotides were designed (supplemental Table S3) to amplify a 100-bp long section of the open reading frame, and their individual efficiency was tested. For comparative analysis of the expression of each gene in leaf, flower, and root, the detected transcript levels were compared with the tissue with the highest expression level using the Δ CT method (36).
|
study
| 100.0 |
A protocol for achiral GC-MS analysis of ISY reactions was adapted from procedures published previously (11, 25). Reactions were conducted in a total volume of 50 μl of buffer C containing 0.5 mg/ml nickel affinity-purified enzyme, 0.8 mm NADPH, and 0.6 mm 8-oxogeranial. After 30 min at 30 °C, products were extracted with 100 μl of ethyl acetate in a 400-μl flat-bottom glass insert (Agilent, 5181-3377) in a GC-MS vial closed with a polytetrafluoroethylene septum. Phase separation was improved by centrifugation of the glass insert in a 2-ml plastic tube at 2000 × g for 2 min. A volume of 3 μl of the clear supernatant was injected in splitless mode on a Hewlett Packard 6890 GC-MS equipped with an Agilent HP-5MS 5% phenylmethylsiloxane column (30 m × 250 μm, 0.25-μm film thickness), a 5973 mass selective detector, and an Agilent 7683B series injector and autosampler. The front inlet temperature was set to 220 °C. After an initial hold at 60 °C for 5 min, a thermal gradient was run from 60° to 150 °C at 20 K/min, from 150 °C to 280 °C at 45 K/min, with a final hold of 4 min at a helium flow rate of 37 cm/s and 1 ml/min. After a solvent delay of 5 min, electron impact fragmentation spectra from 50–300 m/z were collected at a fragmentation energy of 70 eV.
|
study
| 100.0 |
This protocol was adapted for chiral analysis of citral and geranial reduction. Reactions were run for 180 min at 30 °C and contained 0.5 μm enzyme, 1 mm citral (Sigma-Aldrich, C83007) or geranial, 2% tetrahydrofuran as a co-solvent, 1 mm NADPH, and buffer C up to a volume of 300 μl. Commercial rac-citronellal (Sigma-Aldrich, 27470) and S-citronellal (TCI, C1454) were used as standards. From 100 μl of ethyl acetate extract, 1 μl was injected at a 10-fold split ratio into a Restek SKY Liner with wool for split injection. The chiral separation was performed on a Supelco β-DEX225 column (30 m × 250 μm, 0.25-μm film thickness) with an isothermal gradient at 93 °C for 33 min at an average velocity of 26 cm/s. Runs were concluded with a temperature gradient up to 220 °C at a rate of 40 K/min and a final hold time of 4 min.
|
study
| 100.0 |
Further modifications were made to the analytical protocol to allow separation of the products of 8-oxogeranial cyclization. Concentrations of 8-oxogeranial and NADPH in the enzyme reaction were set to 0.5 and 1 mm, respectively. The injection volume was set to 1 μl and the split ratio at the GC-MS injector to 6-fold, and, after an initial hold of 5 min at 105 °C, a thermal gradient was run from 105–150 °C at a rate of 1.5 K/min and from 150–220 °C at 60 K/min with a final hold of 4 min. For quantitative comparison of spectra, they were integrated across the entire peak, and background was subtracted in AMDIS-32. Similarity was calculated with the SpectrumSimilarity function of the OrgMassSpecR package in R version 3.3.3 as the cosine of the angle between the intensity vectors.
|
study
| 100.0 |
Enzyme reactions for CD spectroscopy were conducted for 5 h at 30 °C in water with 0.4 mm 8-oxogeranial and 1 mm NADPH as substrates and 0.5 μm enzyme. Enzyme was diluted at least 150-fold from a buffered solution. Products were extracted with ethyl acetate. The extract from a 1.6-ml reaction was evaporated and taken up in 200 μl of hexane. Completeness of the reaction was verified by GC-MS. Spectra were recorded in 1-nm steps with 0.5-s averaging time on a Chirascan Plus spectropolarimeter (Applied Photophysics) at 20 °C in a 1-mm cuvette. Three measurements were averaged, and background with only hexane was subtracted.
|
study
| 100.0 |
All compounds except trans—trans iridodial have been described previously (11, 21).5 The identity and purity of compounds were verified based on NMR spectra recorded on a Bruker 400-MHz/54-mm UltraShield Plus long hold time automated spectrometer at 400 MHz (1H NMR) and 100 MHz (13C NMR). The residual solvent peak of chloroform was adjusted to δ 7.26 (1H NMR) and 77.16 (13C NMR). Assignment of peaks was aided by two-dimensional 1H-COSY and 1H-13C-HSQC data (supplemental information).
|
study
| 100.0 |
Geranial with a low level of neral contamination was synthesized by oxidizing 100 mg of geraniol (0.65 mmol, 1 eq) in a suspension of 327 mg of sodium bicarbonate (3.9 mmol, 6 eq) in 40 ml of dichloromethane with 330 mg of Dess-Martin periodinane (0.78 mmol, 1.2 eq). The reaction was stirred on ice for 90 min and worked up by filtration over a 0.5-cm glass column packed with 6 cm of silica gel on top of 1 cm of anhydrous sodium bicarbonate. The column was washed with 50 ml of 50% diethylether in hexane, and the product was eluted with 100 ml of diethylether. The solvent was evaporated to dryness, the residue was taken up in 4 ml of hexane and filtered over PTFE, and the solvent was evaporated to yield 74 mg of clear oil (0.49 mmol). The product was identified as geranial based on GC-MS analysis and comparison with the National Institute of Standards and Technology (NIST) library and commercial citral (supplemental Fig. S2a).
|
study
| 100.0 |
Trans–cis nepetalactone was isolated from catnip oil by silica flash chromatography as described by Sherden et al.5 Cis—trans nepetalactone synthesis from the same product via base-catalyzed isomerization has been described by Geu-Flores et al. (11). Reduction to the corresponding trans—cis nepetalactol and cis—trans iridodial has been described previously (11).5
|
study
| 99.94 |
Cis–cis nepetalactone was isolated from a Nepeta variety. To identify a plant containing the correct isomer, Nepeta plants were obtained from plant nurseries (Crocus Ltd., Windlesham, UK; Burncoose Nurseries, Gwennap, Cornwall, UK; Herbal Haven, Saffron Walden, UK; Hardy's Cottage Garden Plants, Hampshire, UK). For methanol extraction, 30–50 mg of fresh leaves were frozen in liquid nitrogen and ground to fine powder in 2-ml Safe-Seal plastic tubes with tungsten beads in a ball mill. After addition of 300 μl of methanol to the cold tube, the tube was vortexed. The resulting slurry was transferred to a 2-ml glass vial with a screw cap, and 600 μl of HPLC-grade hexane was added. After vortexing for 10 s, a green hexane layer on top separated from a lighter yellow methanol layer with bleached particles. The hexane phase containing nepetalactones was transferred to a solid-phase extraction column (Phenomenex Strata SI-1 Silica, 55 μm, 70 Å, 100 mg/1 ml) with a Pasteur pipette. Nepetalactones were eluted with 500 μl of 20% ethyl acetate in hexane. For identification of diastereomers, a volume of 2 μl was injected in split mode (50-fold) on the GC-MS instrument described above. Separation was performed on a Phenomenex Zebron ZB5-HT column (5% polyphenylmethylsiloxane; length, 30 m; diameter, 250 μm; film thickness, 0.10 μm) with a 5-m guard column. Helium was used as mobile phase at a constant flow rate of 7.4 ml/min and average velocity of 100 cm/s. After 5 min at 80 °C, the column temperature was increased to 110 °C at a rate of 2.5 K/min, then to 280 °C at 120 K/min, and kept at 280 °C for 4 min. Nepetalactones eluted in the sequence trans–trans (14.16 min), cis–trans (14.48 min), trans–cis (15.71 min), and cis–cis (15.99 min).
|
study
| 100.0 |
To isolate preparative quantities of cis—cis nepetalactone, all green from a flowerless N. mussinii 'Snowflake' plant (Burncoose Nurseries) was cut off a few centimeters above the soil (approximately 40 g wet weight). The tissue was thoroughly blended in a kitchen blender together with 160 ml of water. Water was added up to approximately 500 ml, and organic compounds were extracted with 5 × 100 ml of dichloromethane. The combined fractions were filtered over paper and washed with 200 ml of brine in a separation funnel. The organic phase was dried by adding anhydrous sodium sulfate, and the solvent was evaporated under reduced pressure. The solid residue was taken up in 10 ml of hexane and separated by silica flash chromatography on a 3 × 25 cm column packed in hexane. Compounds were eluted with a gradient from 10–20% ethyl acetate in hexane in steps of 2% (200 ml each). Elution fractions were checked for diastereomeric purity by GC-MS (see above), and pure fractions were pooled and evaporated, yielding 120 mg of yellow oil. The compound was identified as cis—cis nepetalactone in comparison with published 1H NMR spectra (21).
|
study
| 100.0 |
Cis—cis nepetalactol ([4aR,7S,7aS]-4,7-dimethyl-1,4a,5,6,7,7a-hexahydrocyclopenta[c]pyran-1-ol) was obtained by reducing 92 mg (0.55 mmol, 1 eq) of cis–cis-nepetalactone with 95 mg of DIBAL (0.66 mmol, 1.2 eq). Under dry conditions and a nitrogen atmosphere, 710 μl of DIBAL dissolved in hexane was added dropwise during 20 min to a dry ice/acetone–cooled flask containing cis—cis nepetalactone in 5 ml of hexane while stirring. After stirring for another hour, 770 mg of Bäckstrøm reagent (sodium sulfate decahydrate:celite, 1:1, v/v) was added, and the reaction was stirred for another hour on ice. Solid particles were removed by filtration on a glass frit, which was washed with diethyl ether. The residue obtained after removal of solvent under reduced pressure was purified by silica flash chromatography (1.5 × 21 cm column, eluted with up to 20% ethyl acetate in hexane), yielding 35 mg of product (0.21 mmol, 38% yield) as a 70:30 mix of C1 anomers according to NMR. 1H NMR (major anomer): δ 6.02 (1H, dq, J = 1.4/1.4 Hz, C3), 5.01 (1H, dd, J = 6.2/5.4 Hz, C1), 2.72 (1H, d, J = 6.3 Hz, O1), 2.47 (1H, broad ddd, J = 7/7/7 Hz, C4a), 2.26–2.20 (1H, m, C7), 2.16–2.08 (1H, m, C7a), 1.87–1.79 (1H, m, C5), 1.82–1.72 (1H, m, C6), 1.59–1.51 (1H, m, C5), 1.55 (3H, dd, J = 1.1/1.4 Hz, C8), 1.35–1.25 (1H, m, C6), 1.09 (3H, d, J = 7.1 Hz, C9); 13C NMR (major anomer): δ 134.58 (C3), 113.81 (C4), 92.85 (C1), 45.53 (C7a), 38.51 (C4a), 36.01 (C7), 32.54 (C6), 29.70 (C5), 16.73 (C9), 16.47 (C8); 1H NMR (minor anomer): δ 5.98 (1H, dq, J = 1.5/1.5 Hz, C3), 5.33 (1H, dd, J = 5.0/3.3 Hz, C1), 2.52 (1H, dd, J = 5.0/1.3 Hz, O1), 2.38–2.20 (1H, m, C7), 2.36–2.29 (1H, m, C4a), 2.20–2.12 (1H, m, C7a), 2.19–2.10 (1H, m, C5), 1.64–1.53 (1H, m, C5), 1.60 (3H, dd, J = 0.9/1.4 Hz, C8), 1.15 (3H, d, J = 7.0, C9), C6 is not assigned; 13C NMR (minor anomer): δ 132.55 (C3), 92.75 (C1), 44.46 (C7a), 38.99 (C4a), 36.71 (C7), 30.97 (C5), 17.24 (C8), 15.29 (C9), C4 and C6 are not assigned.
|
study
| 100.0 |
Trans—trans nepetalactone was not found in any Nepeta plant in sufficient quantities and had to be synthesized by epimerization of cis—cis nepetalactone under basic conditions (21). In a 50-ml flask equipped with a reflux condenser, 500 mg of cis—cis nepetalactone (1 eq, 3.0 mmol) was dissolved in toluene and refluxed. Progress of the reaction was controlled by GC-MS, and the reaction was stopped when the equilibrium was reached at a 9:1 cis–cis:trans–trans ratio after 6 h. The reaction mix was evaporated under reduced pressure and separated by silica flash chromatography as described for cis—cis nepetalactone. Fractions were checked for the nepetalactone diastereomers by TLC (anisaldehyde stain, 20% ethyl acetate in hexane as eluent), where trans–trans (Rf = 0.65) and cis–cis nepetalactone (Rf = 0.59) were well separated. The 1H NMR was identical to a spectrum published previously (21).
|
study
| 100.0 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.