Early embryonic development and spatiotemporal localization of mammalian primordial germ cell-associated proteins in the basal rodent Lagostomus maximus

The gene network controlling primordial germ cell (PGC) specification in eutherian mammals has been exhaustively investigated in mice. The egg-cylinder morphology of the mouse embryo is the key event enabling inductive signals from the extra-embryonic ectoderm (ExE) to specify epiblast cells as PGCs early on. We investigated the embryonic development and the spatiotemporal localization of PGC-associated proteins in the basal Hystricognathi rodent Lagostomus maximus. L. maximus develops through a flat-disc epiblast far apart from the ExE. In the primitive streak stage, OCT4-positive cells are detected in the posterior pole of the embryo disc in the mesoderm of the proximal epiblast. In the neural plate stage, a reduced 8 to 12 OCT4-positive cell population transiently expresses FRAGILIS, STELLA and SOX17 in the posterior streak. Soon after translocation to the hindgut, pluripotent OCT4 cells start expressing VASA, and then, STELLA and FRAGILIS are turned on during migration toward the genital ridge. L. maximus shows a spatiotemporal pattern of PGC-associated markers divergent from the early PGC restriction model seen in mice. This pattern conforms to alternative models that are based on a pluripotent population in the embryonic axis, where PGCs are specified later during development.

The molecular machinery of primordial germ cell (PGC) specification has been studied in laboratory mice and it has erected as the paradigm for germline development in eutherian mammals 1 . The appearance of PGCs in mice depends on the sequential and overlapping expression of a group of molecules responsible for the specification and differentiation of the germ line, making cells competent to receive specific signals and preventing them from retaining the characteristics of somatic cells [2][3][4][5][6] . These cells are thought to originate from the most proximal epiblast by induction from the extraembryonic ectoderm (ExE) and the visceral endoderm (VEn). Both extraembryonic tissues surround the epiblast cells of the post-implantation egg cylinder embryo from approximately E5.0 to E6.0. The ExE and VEn release the bone morphogenetic proteins (Bmp) 4, 8b and 2 to the surrounding tissue to instruct a small number of pluripotent proximal epiblast cells to become competent to be PGCs, suppressing the somatic program that is adopted by neighboring cells. The high levels of Bmp activate the expression of Fragilis and competent cells acquire the ability to form PGCs. Within these Fragilis-positive cells, approximately 6 cells begin to express Blimp1 and Prdm14 at around E6. 25 7,8 . Blimp1 has a critical role in the foundation of the mouse germ cell lineage since its disruption causes an early block in the process of PGC formation 5,8 . Blimp1-expressing cells increase in number from 6 to 16 at approximately E6.5. By E6.75 to E7, 20-28 cells move posteriorly and develop alkaline phosphatase activity and Stella expression. During early gastrulation (E7. 25), the PGCs form a cluster of approximately 40 cells at the base of the incipient allantois in the extraembryonic mesoderm (ExM) 9,10 . From E7 to E7.75, they regain expression of pluripotency-associated genes, such as Oct4, Sox2, Nanog and Stella, and suppress the expression of genes involved in mesodermal specification. Subsequently and concomitant with an increase in their number, PGCs start to migrate one by one through the developing hindgut endoderm. They then exit the endoderm to reach the mesentery, and at approximately E10.5, they colonize the genital ridges, where they proliferate and differentiate into oogonia or spermatogonia 11 .
Whether the mouse regulatory pathway is the established mechanism of germ line formation in other (or all) mammals remains largely unexplored 1 . There are some key embryological differences between mice and other mammals, especially at the epiblast stage, when PGCs are thought to be specified. In mice, the sequential gene expression of PGC specification depends on the topographical proximity between cells of different embryonic tissues, due to the fact that the epiblast forms a cup-shaped egg cylinder, allowing the most proximal epiblast cells to be in close contact with the ExE. However, the early embryo morphology that most mammals have is a flat disc-like epiblast, the mammotypical early embryo morphology 12 , which clearly departs from the mouse egg cylinder. In the flat disc of non-rodent embryos, such as humans, pigs and rabbits, the epiblast contacts with the VEn, and ExE is absent. Moreover, rodents from taxonomic groups other than murids also display a flat disc embryo. This is the case in guinea-pigs (Cavia porcellus) in which the embryo is cup-shaped similar to in mice but the ExE does not contact the epiblast before or after grastrulation 13 . This topographical embryo difference between mice and other mammals, including non-muroid rodents, makes the search of a general mechanism for PGC specification based on the well-established mouse molecular regulatory path difficult.
The mouse pathway complies with the currently accepted model of PGC formation as an early lineage-restricted cluster of cells in the base of the allantois. Nevertheless, no definitive proof demonstrating the continuity of those early-specified PGC and germ cells in gonads has so far been provided, as reviewed by Mikedis and Downs 14 . These authors propose an alternative model in which the alleged PGCs in the base of the allantois are instead a pool of pluripotent progenitor cells in the posterior end of the primitive streak that builds the fetal-placental interface. The pluripotent cell pool condenses into the allantoic core domain (ACD), which extends the body axis posteriorly through the allantoic midline 15 . ACD pluripotent cells express all PGC markers and contribute to both embryonic and extraembryonic tissues 14 . From this pluripotent population, it is suggested that PGC could be segregated later, once evolutionarily conserved genes of germline development, such as VASA, Dazl and Nanos, begin to be expressed 14 . Although this alternative explanation is proposed for the mouse egg cylinder, it may well apply in flat embryos where the ExE is absent or far apart from the epiblast.
The South American plains vizcacha, Lagostomus maximus, is a New World Hystricognathi (Caviomorpha) fossorial rodent and a close evolutionary relative of Cavia porcellus, inhabiting the southern region of the Neotropics. In this study, we analyzed the post-implantation development in L. maximus embryos and showed that they develop through a flat embryonic disc, in which no contact between epiblast and the ExE occurs before or after gastrulation. Moreover, we show that the sequential expression of germ line markers diverges from that in mice before and after gastrulation and during migration toward the developing gonads and colonization of the genital ridges. The spatiotemporal pattern of germ line markers in L. maximus conforms to the proposal of a pluripotent cell population within the embryonic axis, from which PGCs may become restricted at later stages during migration 14 .

Results
Overview of the implantation development of the plains vizcacha embryo. L. maximus embryo has a cup-shaped morphology with a flat disc-like epiblast. The external features of the embryo during the whole gestation period (147 ± 5days) were analyzed and classified (Table S1). Embryo stages classified according to the external morphology and showing the changes that occur in the topographical appearance of the embryo from the pre-somite to the >60 pairs of somite stages, are shown and described in Figure S1.
Early post-implantation development. Implantation site. At 22-26 days of gestation, 6 to 12 implantation sites (IS) distributed in both uterine horns were normally found ( Figure S2A). IS were visible externally as spherical swellings, measuring approximately 4 cm in diameter (Table S1 and Figure S2A). The pattern of implantation was interstitial, with the embryo disc implanted in an anti-mesometrial orientation ( Figure S2B). The embryo was a double-layered structure composed of an ICM of ectoderm cells (the epiblast) and an outer layer of endoderm cells, the visceral endoderm (VEn). The VEn was in close contact with the ICM, elongating and separating the embryo from the trophoblast. The trophectoderm was in contact with the VEn (Fig. 1A).
Differentiation of the embryo. At 26-32 days of gestation, the embryo enlarged considerably and bulged cone-like into the segmentation cavity (Fig. 1B). In this pre-streak stage, evidence of the embryonic axis and the distinction between the embryonic and extraembryonic tissue (ExE) first appeared (Fig. 1B,C). Cone-like structures consisted of unilaminar VEn and supported the ICM. The single epithelium of the VEn extended from beneath the ICM to the ectoplacental trophoblast (Fig. 1B,C). The ExE was far apart from the ICM, separated by the VEn, resulting in a cup-shaped structure ( Fig. 1B-E). The exocoelomic cavity formed in the elongated VEn, and the ICM was flattened (Fig. 1B,C). The embryo sac was composed of the ectoplacental trophoblast, which enclosed the epamniotic cavity (Fig. 1E). While the outer lamina bounding this cavity was a simple epithelium, the inner lamina appeared to vary from two to three cells in thickness (Fig. 1B,E). The ectoplacental cavity formed in the ectoplacental trophoblast and was filled with extravasated maternal blood ( Fig. 1B-D). No ectoplacental cone (EPC) was identified; the embryo was a bilaminar disc (Fig. 1D) with a rim of mesometrial situated ectoplacental giant cells (EPGC) (Figs 1C,D and S1D,E).
Cavitation, gastrulation and formation of the amniotic cavity. In the advanced stage of 32-39 gestation days, the embryo was a cluster of undifferentiated cells, with a cleft in the center forming an incipient amniotic cavity (Fig. 1F). The amniotic cavity formed by cavitation, and the epiblast cells died by apoptosis (Fig. 1F). The cavitated amnioembryonic mass was not in contact with the ExE; the central lumen, or the amniotic cavity, appeared before gastrulation (Fig. 1D), and the embryo was a typical bilaminar disc (Fig. 1D,F). Before the appearance of the allantoic bud, the embryo developed the ectoplacental, amniotic and exocoelomic cavities (Fig. 1D). After cavitation, at the primitive streak stage, there was no significant evidence of differentiation among the cells in the embryonic disc, but mesoderm cells differentiated in the posterior part of the embryo, indicating the beginning of gastrulation (Fig. 1G). At the neural plate stage, a marked swelling resulted in the allantois at one edge of the disc and head folding at the opposite edge. A trilaminar embryonic disc, including the three primary germ layers, ectoderm, mesoderm and endoderm, was established ( Fig. 1H). At this time-point, the primitive streak, the notochord and the neural tube were formed. The mesoderm appeared and spread outward between the ectoderm and the endoderm of the embryonic disc. It covered the amniotic ectoderm with somatic mesoderm and extended over the visceral endoderm (yolk sac) with splanchnic mesoderm (ExM), and the ectodermal tissue thickened and flattened into the neural plate ( Fig. 1I). At the early head fold pre-somite stage, a hindgut invagination was observed, the allantois extended and the head folded ( Fig. 1J and S1F,G).
Gene expression during germ cell specification and migration. The expression of OCT4 was uniformly observed in all epiblast cells (embryonic ectoderm) of the bilaminar embryonic disc in the pre-streak stage (Figs 2B and 6A,D). Except for a few positive cells, the amnion tissue was mostly negative. Interestingly, during this period, BLIMP1, STELLA, FRAGILIS, and SOX2 proteins were undetectable; however, SOX17-immunoreactive cells were found in the VEn (Fig. 2C-G).
As the embryonic development progressed to gastrulation in primitive streak, OCT4-positive ectoderm cells migrated to the posterior pole of the embryo disc to form mesoderm. The number of OCT4-positive cells decreased dramatically to a minimum of 8 to 12 cells at the neural plate stage (Table 1) and was observed to increase again at the late head-fold stage and especially after translocation to the hindgut.  BLIMP1 (data not shown) and SOX2 expression was undetectable during migration and colonization of the genital ridges (Fig. 5E,K,Q and W). Interestingly, SOX17 expression was not detected during migration (Fig. 5F,L and R), but when the genital ridges were colonized, a low SOX17-positive signal was observed (Fig. 5X).
Germ cell quantification during migration and genital ridge colonization. The number of OCT4-expressing cells was estimated from the embryo disc until fetal ovary colonization. Estimates based on OCT4-expressing cells are shown in Table 1; other detectable markers for each developmental stage are also indicated.
At the embryo disc and primitive streak stages, OCT4-positive cells were abundant, but no other marker was detected at this developmental time-point (Fig. 6A,D). At the neural plate stage, a small group of 8-12 OCT4-expressing cells was identified in the mesoderm. From the late head-fold stage, approximately 25 cells retaining OCT4 expression started to migrate increasing rapidly in number to a hundred cells located at the gut-mesentery (Fig. 6B,E). More than 1,000 cells were detected by the end of migration rapidly increasing to a total of approximately 55,000 cells once fetal ovary colonization was achieved (Fig. 6G).

Discussion
The basal rodent L. maximus develops through the mammotypical embryo disc. In this study, we show that the basal Hystricognathi rodent L. maximus develops through a flat disc, where no contact between the ExE and the epiblast exists. The topological organization of the pre-gastrulating embryo determines a scenario for PGC formation divergent from the mouse model. At the beginning of implantation, the pre-gastrulating L. maximus embryo shows an ICM ball positioned at the distal part of the conceptus, which then begins to transform into a flat-disc epiblast, far apart from the trophoblast located at the proximal part of the conceptus. This topology departs from that found in the rodent embryo model based on laboratory mice, in which the epiblast makes close contact with the ExE and both tissues are enveloped by the VEn, defining the typical mouse egg-cylinder morphology 16 . In L. maximus the epiblast and the amniotic cavity develop from a bilaminar disc by cavitation rather than by folding. The flat disc epiblast morphology of the L. maximus embryo is shared with non-rodent mammals that have been studied so far, such as rabbits, pigs, cows, and humans among others 12 , and it is considered the mammotypical embryo morphology. It is worth noting that L. maximus epiblast morphology resembles that of human embryos, where the epiblast does not contact the trophoblast and remains separated from it by the pro-amniotic cavity before and during gastrulation, whereas in the majority of non-rodent Cells expressing mammalian PGC-associated proteins originate from mesoderm progenitors in L. maximus. OCT4 protein seems to play an essential role in the establishment and maintenance of the germ line. OCT4 expression in the pre-gastrulating embryo was uniform in the epiblast cells, but after the primitive streak stage, OCT4 was mostly down-regulated, and its expression only persisted in a group of cells that was later restricted to the mesoderm of the posterior end of the embryo. It seems likely that OCT4 expression is required for maintaining pluripotency, helping to epigenetically reprogram cells for PGC development that will be specified at a later stage (see below). Pluripotent genes, such as Nanog, Sox2 and Oct4, are restored at E7 to E7.75 in the mouse embryo during early gastrulation, suppressing expression of genes involved in mesodermal specification [17][18][19] .
In mice, Blimp1 orchestrates somatic gene repression and promotes pluripotency genes 20,21 . Despite using two different antibodies and validating them through protein recognition in BLIMP1-expressing tissues of L. maximus, we could not detect its expression during early gastrulation or later stages of development. Because  failure to detect the protein is not evidence of its absence, especially as we did not carry out experiments to define the threshold of protein detection under our conditions of immunohistochemistry, further analysis is still needed to assess when and where BLIMP1 is expressed in L. maximus or if it is even expressed at all. The essential role played by Blimp1 in mouse germ cell specification has also been reported in rabbit embryos 22 , but its role has been considered unlikely in other flat embryos, such as those of pigs 23 . In humans, SOX17 rather than BLIMP1 seems to be the key regulator of PGC destination 24 , although recent in vitro studies also suggest a role for BLIMP1 in human PGC-like cells derived from induced pluripotent stem cells 25 . As in the human embryo, we did detect SOX17 instead of BLIMP1 in the flat disc of L. maximus; when mesodermal OCT4-positive cells become restricted in number, they begin to express SOX17. This advocates for a comparable situation in the regulation of PGC fate in L. maximus and human embryos. In early mouse PGC development, SOX2 instead of SOX17 protein seems to accomplish this essential role; similar to human PGC specification, we did not find SOX2 expression in L. maximus 24,26 . It is likely that the temporal co-localization of SOX17 and OCT4 proteins in L. maximus PGCs plays a major role in inhibiting somatic genes and maintaining pluripotency because alternative SOX/OCT4 pairings target specific loci to regulate pluripotent cell fate and differentiation programs 27 . During migration, SOX17 is down-regulated, and its expression is restored in oogonia after the colonization of the genital ridges. It seems reasonable to consider that SOX17 could act as a regulator of proliferation and the cell cycle 28 , contributing to the continuous rise of healthy germ cells, which characterizes the ovaries of L. maximus throughout fetal life, in the presence of a minimal rate of apoptosis-driven, germ cell attrition 29 .
At the neural plate stage, we detected transient expression of FRAGILIS, STELLA and SOX17 in the proximal epiblast. In mice, the expression of both proteins seems to be necessary for the foundation of PGCs 4 . Fragilis expresses around the most proximal epiblast cells, and its expression intensifies in the posterior extra-embryonic mesoderm. Stella begins to express specifically in Fragilis-expressing cells in the extra-embryonic mesoderm and continues to be expressed in migrating PGCs. In contrast, in L. maximus, transiently expressed FRAGILIS, STELLA and SOX17 turned on again at a later stage (35-39 somites) during migration.
The essential germ line marker VASA was expressed early in L. maximus during the translocation of OCT4-positive cells to the hindgut (8-12 somites). Thereafter, VASA-expressing cells were detected throughout the migration toward the genital ridges (cf. Figs 5 and 6). The expression of VASA was not observed before translocation to the hindgut (data not shown). VASA seems to be a major determinant of the germ cell for this species due to the fact that it is expressed very early in germ line development and continues to be expressed both in fetal and adult ovaries 30,31 . In mice, the expression of VASA protein becomes detectable in PGCs at the late migrating stage in the gut mesentery of 9.5-10.5 dpc embryos 32 . This is also the case in the human embryo, where VASA expression is seen in PGCs migrating near the genital ridges 33,34 .
The spatiotemporal pattern of expression of germ line markers found in L. maximus differs in many aspects from that described in mice, as discussed above. Moreover, it is difficult to understand this expression pattern in light of the currently accepted model on the origin of PGCs as a lineage-restricted cluster of cells in the base of the allantois, specified early just before, or during, gastrulation. In contrast, our results better accommodate an alternative model proposed by Mikedis and Downs 14 , in which PGCs are specified later from a pluripotent progenitor population within the embryonic axis. Before and during gastrulation, the L. maximus embryo showed a population of cells expressing the pluripotent protein OCT4 in the posteriorly extending embryo axis. Early on, at the 8-12 somites stage, these OCT4-positive cells translocated to the hindgut, the universal germ line marker VASA was expressed (cf. Table 1 and Fig. 5), and then, STELLA and FRAGILIS proteins were identified in between the 25-30 somites and 35-39 somites stages (cf. Fig. 5J and Table 1). It seems reasonable to speculate that OCT4/ STELLA/FRAGILIS-expressing cells within the migrating pluripotent population finally are restricted to form PGCs once the evolutionarily conserved germline-specific VASA protein is expressed 14 .
From a few 8-12 OCT4-positive cells at the neural plate stage, a cluster of approximately 25-30 cells was found at the beginning of migration in L. maximus, which is comparable to what has been described in mice and humans. As these cells proliferate during migration, they increase to approximately 1,200 when entering the genital ridges, a comparable number of colonizing PGCs to that seen in humans 35 . Embryo stage  OCT4  VASA  FRAGILIS  STELLA  SOX2  SOX17  BLIMP1 Putative PGC/Germ cell* (means ± S.D.)   Evolutionary considerations. The mouse egg-cylinder morphology has been assumed as the typical early post-implantation rodent embryo; however, both L. maximus, as well as Cavia porcellus, depart from this generalized morphology. Together with Sciuromorpha, Hystricognathi are the first offshoots in the rodent phylogeny 36,37 after the separation of Rodentia and Lagomorpha 38 , whereas Myomorpha, especially the mouse-related clade Muroidea, are the most recently evolved rodents 39 . The planar epiblast morphology in the basal L. maximus supports the idea that the egg-cylinder in which the ExE comes in close contact with the epiblast is an innovation of more recently evolved rodents such as murids rather than a characteristic of the entire order 20 . The appearance of the ExE, deriving from the persistent polar trophoblast, and the germ layer inversion that is provoked as the embryo sinks into the yolk-cavity are the landmarks of the early implantation rodent embryo compared with those of other mammals. This phenomenon occurs in different ways in the few species belonging to the different suborders of Rodentia that have been analyzed so far, and the ExE can be recognized in all cases 40 . The extensive embryo invagination into the yolk-cavity seen in mice, which enables close contact of the ExE and epiblast, does not occur in L. maximus, as shown in this study, or in its close evolutionary relative C. porcellus 13 . Moreover, this seems to be the case in other basal rodents, such as the Sciuromorpha Spermophilus tridecemlineatus, and even in species from basal offshoots of the Myomorpha clade, such as Geomys bursarius 40 .

Embryo disc
Our observation that PGCs are specified late in L. maximus is supported by the recent proposal by Johnson and Alberio 23 that the precocious (before gastrulation) specification of the germ line seen in mice enables an accelerated development of somatic innovations favoring speciation and the characters of typical r-strategists, i.e., short gestations, large litters, small body-size, and a short life-span. The key for early germ line restriction in mice resides in the appearance of the ExE and its capacity to produce germ line inducing proteins 23 . This may well apply to the Muroidea clade but not to other clades in Rodentia, such as Hystricognathi. In L. maximus, as well as in C. porcellus, the ExE remains far apart from the epiblast before and after gastrulation, indicating that it would not be involved in germ line inducing signaling.
Finally, it is worth highlighting that the pattern of expression of PGC-associated markers in L. maximus described in this report diverges from that in mice and might be relevant to other non-rodent mammals with planar embryo morphology, including humans. The early implantation embryo of L. maximus resembles the human embryo in that SOX17 rather than BLIMP1 seems to play a central role in the fate of PGCs 21 , SOX2 is not involved in PGC specification 24 and both embryos have a comparable planar morphology differing from other embryo-disc developing mammals 12 .  Sample collection. A total of 198 embryos/fetuses were recovered and analyzed. Only the embryo located nearest the cervix from each uterine horn was included in this analysis, due to the fact that they are the only ones that develop to term; the remaining anterior-implanted embryos stop developing at early post-implantation stages and are selectively aborted 41,42 . Embryo development of L. maximus was analyzed from the beginning of implantation up to the >60 pairs of somites stage. Post implantation embryo development was rated per week and placed in a relative chronological sequence. Organization and classification of the embryonic stages were established on a comparative basis with mice and close evolutionarily related chinchillas and guinea pigs [43][44][45][46] using Theiler Stages 43 . Developmental data were supplemented with the capture-time, length and width of the implantation site, gross embryo morphology, somite number, and crown-heel length, width and weight in more developed fetuses.

Methods
Tissue collection and histological preparation. Pregnant females were anaesthetized by the intramuscular administration of 13.5 mg/kg body weight of ketamine chlorhydrate (Holliday Scott S.A., Buenos Aires, Argentina) or 0.6 mg/kg body weight xylazine chlorhydrate (Richmond Laboratories, Veterinary Division, Buenos Aires, Argentina) and euthanized using an intracardiac injection of Euthanyl (0.5 ml/kg body weight, sodium pentobarbital, sodium diphenylhydantoin, Brouwer S.A., Buenos Aires, Argentina). Uterine horns were exposed, removed and thoroughly rinsed in PBS, pH 7.4. The IS and the embryos/fetuses were removed from the horns, measured and fixed in cold 4% neutral-buffered para-formaldehyde (PFA) for 24 h. PFA-fixed tissues were dehydrated through a graded series of ethanol (50%, 70%, 95% and 100%), embedded in paraffin, serially sectioned at 6 μm, and mounted onto cleaned coated slides. Sections were dewaxed in xylene (Sigma Chemical Co., St. Louis, MO, USA) and re-hydrated through a series of decreasing concentrations of ethanol. At least 3 to 5 slides of each specimen were stained with haematoxilyn-eosin for general histology inspection. The remaining consecutive serial-sectioned slides were stored at room temperature until used for immunohistochemistry.
Embryo whole mount immunohistochemistry. Whole embryos were PFA-fixed for 4 h at 4 °C and then washed three times in PBT (0.2% Tween-20 in PBS) for 15 min at 4 °C. Embryos were subjected to permeabilization through a series of methanol (25%, 50%, 75%, 90% and 100%) in PBT for 10 min at 4 °C. Embryos were kept at −20 °C overnight and then were subjected to rehydration through 10 min washes in a decreasing series of methanol (90%, 75%, 50% and 25%) in PBT at 4 °C, followed by a final 10 min wash in PBT. Embryos were blocked for endogenous peroxidase activity with 5% H 2 O 2 in methanol for 5 h at room temperature and then incubated for 1 h in PBS-MT blocking solution (PBT + 2% w/v nonfat dry milk) at room temperature on a shaker. Immunoreactivity was achieved by incubating the embryos overnight at 4 °C with a specific anti-OCT4 IgG (1:250, ab19857, Abcam, Cambridge, UK) diluted in PBS-MT on a shaker and washed 5 times for 1 h in PBS-MT. The immune reaction was revealed with biotinylated anti-rabbit IgG overnight with rotation at 4 °C, washed five times for 1 h in PBS-MT, incubated with an avidin-biotin complex (ABC Vectastain Elite Kit, Vector Laboratories, Burlingame, CA, USA) with rotation for 30 min and washed five times for 1 h in PBS-MT. The reaction was visualized with DAB (SK-4100, DAB Kit, Vector Laboratories, Burlingame, CA, USA). Finally, the treated sections were washed three times in PBS-MT once the reaction and staining had reached the desired intensity. Samples were observed and photographed on excavated slides in PBT. They were then preserved in glycine until used for sectioning.
Germ cell quantification and statistical analysis. The total numbers of OCT4-positive cells during gastrulation, migration and genital ridge colonization were quantified. Each embryo was completely cut into 6 μm-serial sections. All the embryo sections were counted for each embryo. A cell count was performed in 3 whole embryos for each developmental stage from neural plate to <30 somites. Embryos with >30 somites were quantified using a stereoscopic method, as previously reported 26 . Cell counting was expressed as the mean ± S.D.
GraphPad Prism software (version 5.0 for Windows GraphPad Software, San Diego, CA, USA) was used for a one-way analysis of variance. A Newman-Keuls test was used when differences between more than two groups were compared. A p-value of less than 0.05 was considered statistically significant.