Pluripotent epiblast stem cells (EpiSCs) derived from postimplantation embryos exhibit properties that are characteristically different when compared with pluripotent embryonic stem cells (ESCs) derived from mouse blastocysts. However, EpiSCs are relatively less well characterised compared with ESCs. In particular, the relationship between EpiSCs and primordial germ cells (PGCs)is unknown, and is worthy of investigation because PGCs originate from postimplantation epiblast cells in vivo. We show that EpiSCs have an infinite capacity for generating PGCs, under conditions that sustain their pluripotency and self-renewal. These PGCs generated in vitro show appropriate transcriptional and epigenetic reprogramming events and are able to develop further into late germ cells. Notably, the PGCs can, in turn, be induced to undergo dedifferentiation into pluripotent embryonic germ cells (EGCs), which resemble ESCs and not the EpiSC from which they are derived. Our observations demonstrate intrinsic reprogramming during specification of PGCs that results in the erasure of epigenetic memory of EpiSCs following reactivation of the X-chromosome, DNA demethylation and re-expression of key pluripotency genes. This study provides novel insights into the nature and properties of EpiSCs,and introduces an in vitro model system that will be useful for investigations on PGC specification and on mechanisms regulating epigenetic reprogramming in germ cells.

The pluripotent state is first established in the inner cell mass (ICM) of blastocysts that develops from the totipotent zygote(Niwa, 2007; Surani et al., 2007). During subsequent postimplantation development, the epiblast cells originating from the ICM commence differentiation into diverse somatic cells, but these cells also give rise to primordial germ cells (PGCs). Furthermore, it is possible to generate pluripotent stem cells in vitro under appropriate culture conditions from the ICM, the postimplantation epiblast and PGCs; these are embryonic stem cells (ESCs), epiblast stem cells (EpiSCs) and embryonic germ cells (EGCs),respectively (Brons et al.,2007; Evans and Kaufman,1981; Matsui et al.,1992; Resnick et al.,1992; Tesar et al.,2007). Although all these pluripotent stem cells share some common properties, such as expression of key pluripotent-associated genes including Oct4, Nanog and Sox2, EpiSCs in particular differ significantly from ESCs and EGCs in their epigenetic state and gene transcription profile (Brons et al.,2007; Guo et al.,2009; Hayashi et al.,2008; Tesar et al.,2007).

Notably, EpiSCs inherit many properties from epiblast cells, including an inactive X chromosome (in female lines) and higher expression levels of genes such as Fgf5, Lefty1 and Cer1, as well as the repression or lower expression of some pluripotency genes. By contrast, the epigenetic state and gene expression profile of EGCs is strikingly similar to that of ESCs because the specification of PGCs involves epigenetic reprogramming that results in the erasure of the epigenetic memory of epiblast cells(Hayashi and Surani, 2009). This includes reactivation of the inactive X chromosome(Chuva de Sousa Lopes et al.,2008; de Napoles et al.,2007; Sugimoto and Abe,2007), repression of the somatic programme, re-expression of many pluripotency-specific genes, and chromatin modifications and DNA demethylation(Ancelin et al., 2006; Hajkova et al., 2008; Ohinata et al., 2005; Seki et al., 2005; Seki et al., 2007). The changes that occur during the dedifferentiation of PGCs to EGCs convert unipotent germ cells to pluripotent stem cells, and follow from the downregulation of Blimp1 and the translocation of Prmt5 from the nucleus to the cytoplasm (Durcova-Hills et al.,2008). Thus, while PGCs represent a unipotent lineage, their epigenetic state is closer to that of the ICM of blastocysts, which explains why EGCs are like ESCs and unlike EpiSCs.

Investigations on many aspects of the underlying mechanism of epigenetic reprogramming during PGC specification would be facilitated by an in vitro system, especially if it is possible to generate large numbers of PGCs. In this respect, EpiSCs are of particular interest, as they are derived from epiblast cells from which PGCs originate in vivo. Here, we analysed EpiSCs carrying Blimp1-GFP or Stella-GFP reporters; Blimp1 (also known as Prdm1) is the key determinant of PGCs in mice, and the earliest known marker of PGC precursors (Ohinata et al.,2005). Stella (Dppa3 - Mouse Genome Informatics) is detected in Blimp1-positive cells at the end of the PGC specification process(Hayashi et al., 2007; Kurimoto et al., 2008; Ohinata et al., 2005). Our analyses revealed that it is possible to generate PGC precursors and fully specified germ cells from EpiSCs, which occurs following appropriate epigenetic reprogramming events and changes in the gene expression profile. Importantly, we found continuous delineation of PGCs from EpiSCs under conditions that retain their capacity for self-renewal and pluripotent state. Thus, EpiSCs are a potential source for a large number of PGCs, and they provide an important in vitro system for studies on the germ cell lineage.

Embryo collection and cell culture

Epiblast tissue was collected on embryonic day (E) 6.5 embryos from 129SvEv females crossed with Blimp1-Gfp-BAC or Stella-Gfp-BAC transgenic males (Ohinata et al.,2005; Payer et al.,2006). EpiSCs were derived from the epiblast as described previously with slight modification (Brons et al., 2007). Briefly, E6.5 epiblasts were stripped from the visceral endoderm (VE), and cultured in chemically defined medium(Brons et al., 2007) on mouse embryonic fibroblasts (MEFs). After several passages, the culture medium was replaced with N2B27 medium supplemented with 20% KSR, 2 ng/ml recombinant human activin A (Peprotech) and 12 ng/ml bFGF (Invitrogen).

Immunofluorescence and FACS analysis

For FACS analysis, single cell suspensions of EpiSCs were analysed and sorted by FACSort with CELLQUEST software (BD Bioscience) and FACSAria (BD Bioscience). For SSEA1 staining, single cells were incubated with PE-conjugated anti-SSEA1 antibody (R&D Systems), according to the manufacturer's instructions. For intracytoplasmic staining, EpiSCs were fixed with 4% paraformaldehyde, permeabilised with 0.1% Triton X-100, and stained with mouse anti-Oct4 mAb (BD Bioscience) followed by Alexa 568-conjugated goat anti-mouse IgG (Invitrogen). To intensify the signal, these samples were stained again with Alexa568-conjugated donkey anti-goat IgG (Invitrogen). For immunofluorescence analysis, cells were fixed with 4% paraformaldehyde, washed with PBS and then stained with appropriate primary antibodies: rabbit polyclonal anti-Stella antibody, which was raised against a synthetic peptide;rat anti-GFP monoclonal antibody (mAb; Nacalai tesque); mouse anti-Oct4 mAb(BD Bioscience); rabbit anti-Mvh polyclonal antibody (Abcam) and rabbit anti-Sycp3 polyclonal antibody (kindly provided from Dr Chuma, Kyoto University, Japan). Secondary antibodies conjugated with Alexa488, Alexa568 or Alexa647 were used to detect the binding of the primary antibodies to specific cells. DAPI (4,6′-diamidino-2-phenylindole) was used for staining nuclei. The stained cells were analysed by a BioRad (Hercules, CA, USA)Radiance 2000 confocal microscope.

PGC induction and EG cell derivation

To promote differentiation of EpiSCs, BMP4 (R&D Systems) was added to the culture medium at 500 ng/ml. The cells were washed after 24 hours and cultured with Noggin (R&D Systems) at 250 ng/ml and chordin (R&D Systems) at 1.25 μg/ml. All culture media was supplemented with activin A and bFGF. Cells were analysed at 48 hours after the addition of BMP4. EGC derivation from Stella+ EpiSCs was performed as described previously, with slight modifications(Durcova-Hills and Surani,2008). Briefly, FACS-sorted Stella+ EpiSCs (sorted at a rate of 3000 cells/well) were cultured on mitotically inactive Sl4-m220 in DMEM supplemented with 15% FCS, nonessential amino acids, 2 mM L-glutamine, penicillin/streptomycin, 1 mM sodium pyruvate, 50μM 2-mercaptoethanol, 1000 U/ml LIF (Chemicon International) and 25 ng/ml bFGF. After 5-6 days of culture, the resulting EGC-like colonies were dissociated by trypsin treatment and spread on MEF. After derivation, EGCs were cultured on MEFs in the medium without bFGF.

Co-culture of PGCs with gonadal cells

For the culture of in vitro-derived PGCs with gonadal cells, female gonads were collected from E12.5 embryos. A small number of the collected gonads were cut into small pieces and put on transwell inserts (Falcon) in DMEM supplemented with 20% fetal calf serum (FCS), 2 mM L-glutamine, and penicillin/streptomycin. In addition, 10-12 gonads were dissociated by treatment with trypsin and mixed with the FACS-sorted 3000 Stella-GFP+ PGCs. After centrifugation, the cell suspension was transferred into a small drop of medium on a petri dish. The cells were brought together to the centre of the dish by gently rocking the culture dish for 1 minute by hand. The `centralised' aggregate of cells was aspirated with a minimum volume of medium using a glass pipette, and transferred through the gaps present between the small pieces of gonads on the transwell inserts. The culture medium was subsequently changed every day.

RT-PCR analysis

For quantitative PCR (Q-PCR) analysis, cDNA was synthesised from total RNA using Superscript II and oligo d(T) primers. Single-cell cDNA preparation was performed as described previously(Kurimoto et al., 2007). cDNA was used for Q-PCR reactions using Sybr Green master mix (Qiagen) and specific primers. Primer sequences used in this study are shown in Table S1 in the supplementary material.

We first derived three separate EpiSC lines carrying the Blimp1-BAC-GFP reporter (henceforth called Blimp1-EpiSC)(Ohinata et al., 2005), as this gene is the key determinant and the earliest marker of PGC precursors. Blimp1-EpiSC lines were similar to non-transgenic EpiSCs, and showed the capacity for extensive self-renewal while retaining expression of the pluripotent genes for at least 60 passages (data not shown). We first detected Blimp1-GFP expression at the start of the derivation process in almost all epiblast cells after 2 days of culture, but this expression became gradually downregulated (data not shown). However, even after the completion of the derivation process, we detected Blimp1-GFP expression in a significant proportion (10 to 50%) of the self-renewing Blimp1-EpiSCs(Fig. 1A); some of these GFP-positive cells were also Oct4 positive(Fig. 1B). We observed Blimp1-GFP expressing cells in all the established Blimp1-EpiSC lines. However, the proportion of Blimp1-GFP+ cells was variable, which was partly due to their spontaneous differentiation, as the frequency of the Blimp1-GFP+ cells was proportional to the morphological state of the colonies, which apparently increased whenever we observed morphologically differentiated colonies (see below). Notably, quantitative PCR (Q-PCR)analysis of FACS-sorted Blimp1-GFP+ cells revealed the expression of key markers of early PGCs, including Nanos3, Prdm14 and Dnd1 (Fig. 1C). Furthermore, we observed repression of Hoxa1 in the Blimp1-GFP+ cells, which is also one of the key features of PGC specification in vivo (Saitou et al.,2002). More significantly, we also detected transcripts of Stella (also known as Dppa3), which is a definitive marker of PGCs, exclusively in the Blimp1-GFP+ cells(Saitou et al., 2002; Sato et al., 2002)(Fig. 1C). These observations suggest that, amongst the Blimp1-GFP+ cells, we were detecting early PGCs originating from the EpiSCs in culture.

Additional analysis of the Blimp1-GFP+ cells, however, revealed expression of Gata4 and Gata6(Fig. 1C), which are markers of the visceral endoderm (VE) at early postimplantation stage. This is perhaps not surprising as Blimp1-GFP expression is also detected in the VE of E6.25 embryos (Ohinata et al.,2005). This suggests that the Blimp1-GFP+ cells we observed in vitro are composed of at least the PGC precursors and VE cells that are generated spontaneously in EpiSC culture. We could use Oct4 and/or SSEA1 expression as additional markers to distinguish between PGC precursors and VE, as PGC precursors would express these markers, whereas the VE would instead be expected to show expression of Gata4/Gata6. Indeed, immunostaining and single cell Q-PCR revealed mutually exclusive expression of Oct4 and Gata4/Gata6 in the Blimp1-GFP+ cells (see Fig. S1 in the supplementary material). In additional immunostaining studies, we also found that the SSEA1+/Blimp1-GFP+ cells were positive for the PGC markers, and lacked expression of Gata4/Gata6 (see Fig. S2 in the supplementary material). FACS analysis revealed that between 6 and 10% of the cells in Blimp1-EpiSC cultures were Oct4+/Blimp1-GFP+putative PGC precursors (Fig. S1 in the supplementary material). Thus, a combination of Blimp1-GFP with either Oct4 or SSEA1 can be used as markers to detect and isolate PGC precursors. Notably, these germ cell precursors are yet to undergo the full process of specification before being irreversibly committed as founder PGCs. The final commitment to PGC fate is evident by the expression of Stella, a PGC-specific marker.

To follow the final events associated with the establishment of germ cells from PGC precursors originating from EpiSCs, we used Stella as a reporter by generating four EpiSC lines from Stella-BAC-GFP reporter mice(Stella-EpiSCs) (Payer et al.,2006). We detected Stella-GFP expression in a small proportion(0-1.5%) of cells in cultures of Stella-EpiSCs(Fig. 1D); these cells were also positive for the endogenous Stella protein(Fig. 1E). We first observed Stella-GFP+ cells as early as the third passage of Stella-EpiSC cultures. Such Stella-GFP+ cells were consistently observed for at least 42 passages in all the Stella-EpiSC lines. The proportion of cells that were detected as Stella-GFP+ consistently ranged between 0-1.5% by FACS analysis regardless of the passage number Stella-EpiSCs. Q-PCR analysis using single cells confirmed that Stella-GFP+ cells were positive for other markers of early germ cells, including Blimp1, Dnd1, Nanos3and Prdm14, the latter in particular is another key regulator of PGC specification (Fig. 1F)(Yamaji et al., 2008). These cells, as expected, were confined within the Blimp1+ population, as confirmed by immunostaining in which all the Stella+ cells(120/120) were also positive for Blimp1-GFP and Oct4(Fig. 1G; data not shown). These results demonstrate that we could detect germ cells at distinct stages of development: from the initial Blimp1+ PGC-precursors to the Stella-GFP+ cells detected at the completion of specification process.

Next, we investigated the origin of the PGC population in more detail,beginning with the appearance of Blimp1-GFP+ cells in EpiSC cultures. We therefore started with Blimp1-GFP- cells isolated by FACS-sorting, and monitored the appearance of Blimp1-GFP+ PGC precursor cells. Indeed, although there were only a few Blimp1-GFP+cells after day 1 of culture, we found a progressive increase in the appearance of Blimp1+/Oct4+ cells on days 3 and 5 of culture (Fig. 2A,B). As expected, the Stella-expressing cells were detected at a later time, at least 5 days after the start of culture (Fig. 2A,B). These results demonstrate that Stella+ cells in EpiSC culture were derived from the Blimp1-GFP- population via the Blimp1-GFP+/Oct4+ PGC precursor population; this sequence of events resembles PGC specification in vivo(Hayashi et al., 2007; Kurimoto et al., 2008; Ohinata et al., 2005).

Our observations so far indicate that EpiSCs maintained under culture conditions that promote their self-renewal without compromising their pluripotency (as judged by the expression of Oct4) can continually generate a population of PGCs, which raised the question of how the balance between different subpopulations is maintained. According to some recently suggested models, it appears that stem cells can fluctuate between different epigenetic and phenotypical states while undergoing self-renewal in culture(Enver et al., 2009; Graf and Stadtfeld, 2008; Hayashi et al., 2008). To address this issue concerning the emergence of PGCs from EpiSCs, we examined the fate of FACS-sorted Blimp1-GFP+/SSEA1+ and Stella-GFP+ cells. We found that the culture of isolated Blimp1-GFP-/Oct4+ cells gradually showed an increase in the number of colonies at days 3-5 of culture(Fig. 2C,D). By contrast, the Blimp1-GFP+/SSEA1+ cells showed poor colony formation(Fig. 2E); indeed many of these cells remained as single cells (Fig. 2F). This result demonstrates the existence of a dynamic equilibrium in EpiSCs, in which some Blimp1-GFP+/SSEA1+cells maintain the potential to revert to Blimp1-GFP-/Oct4+ cells. Considering our observation that differentiation of PGCs from EpiSCs is accomplished in a stepwise manner(Fig. 1F), it is likely that the reversible population of Blimp1-GFP+/SSEA1+ cells is at an early stage of their progression towards PGCs, and lack expression of later markers of germ cells, such as Prdm14 and Dnd1. At day 3 and day 5 of culture of Blimp1-GFP+/SSEA1+ cells,∼20% showed strong expression of Stella(Fig. 2C,D), while others with dense nuclei were presumably dying. Counts of living single cells and colonies at day 3 of culture showed that approximately 90% (268/297) of the sorted Blimp1-GFP+/SSEA1+ cells stayed as single cells, while around 10% (29/297) of the cells formed colonies and reverted to Blimp1-/Oct4+ cells. By contrast, around 90% (100/114)of Blimp1-GFP-/SSEA1+ cells formed colonies(Fig. 2F). Furthermore,differentiated colonies containing Blimp1+/Oct4- cells,which represent the VE cells, were more frequently seen in cultures of Blimp1-GFP-/SSEA1+ cells(Fig. 2F). Notably, unlike Blimp1+/SSEA1+ cells, Stella-GFP+ cells could not form colonies, and eventually died under EpiSC culture conditions(Fig. 2E).

Based on the combined observations, it appears that EpiSCs are in a state of dynamic equilibrium in which the delineation of PGC precursors and specified PGCs occurs continually. In EpiSCs, the Blimp1-GFP- cells represent the pluripotent cells, which, in part, differentiate into Blimp1+ PGC precursors, followed by the formation of Stella+ PGCs. By contrast, Blimp1-GFP+/SSEA1+PGC precursors are predominantly poised towards the germ cell fate, and some of these cells go through to complete the specification process, whereas others undergo apoptosis or revert back to Blimp1-/Oct4+cells. Consistent with this interpretation, we found that Myc (also known as c-myc) and Eras were dramatically downregulated in the Blimp1+/Oct4+ cells (see Fig. S3 in the supplementary material). Myc is a putative target of Blimp1 and its downregulation is thought to result in a slowing down of the cell cycle of PGCs (Lin et al., 1997; McLaren and Lawson, 2005). Eras is a known regulator of ESC proliferation(Takahashi et al., 2003). Thus, downregulation of these genes might result in poor proliferation and colony formation of Blimp1+/Oct4+ cells.

Because Bmp4 is a key signal for PGC specification(Lawson et al., 1999), we sought to determine whether this signalling molecule could influence PGC formation from EpiSCs. It should be noted that although we had not added any recombinant BMP4 to our culture medium, EpiSCs themselves express Bmp4. We therefore added Dorsomorphin, an inhibitor of BMP signalling(Yu et al., 2008) to EpiSC cultures to monitor the effect on PGC derivation. Indeed, addition of Dorsomorphin to EpiSCs in culture reduced the proportion of both the Stella+ and the Blimp1+/SSEA1+ cells(Fig. 2G). Conversely, addition of recombinant BMP4 induced a small but significant increase in the proportion of Stella-GFP+ cells, as well as of the Blimp1+/SSEA1+ cells(Fig. 2G). However, we did not detect significant PGC induction when BMP8b was added simultaneously with BMP4 in the culture (data not shown). These results demonstrate that BMP4 is at least one of the key cytokines present in the culture, and may promote PGC formation from EpiSCs. It is possible that localised activation of BMP-Smad signalling in EpiSCs produces an appropriate microenvironment in which PGC precursors might be delineated in response to the signalling. This is likely as in human ES cells (hESCs), which resemble EpiSCs, the levels of phosphorylated Smad1 are variable in individual colonies, and this causes their heterogeneous response to differentiation(Peerani et al., 2007).

We next investigated whether appropriate epigenetic reprogramming events occur in the PGC derived from EpiSCs. First, we determined the DNA methylation status of the pluripotent genes Stella, Rex1 (also known as Zfp42) and Fbxo15, all of which are hypermethylated in EpiSCs (Fig. 3A; see also Fig. S4 in the supplementary material), but not in ESCs(Hayashi et al., 2008; Imamura et al., 2006). Strikingly, the Stella locus becomes hypomethylated during the transition from Blmp1+/Oct4+ to Stella+ cells(Fig. 3A); this was also the case with Rex1 and Fbxo15 loci (see Fig. S4 in the supplementary material), demonstrating that PGC specification from EpiSCs is coupled with DNA demethylation of these loci. This epigenetic change is an indicator of an important reprogramming event(Seki et al., 2005), which helps to erase the epigenetic memory of EpiSCs. Concomitantly, the de novo DNA methyltransferase Dnmt3b was downregulated in Stella+ EpiSCs (see Fig. S3 in the supplementary material), which is consistent with similar observations on nascent PGCs in vivo(Yabuta et al., 2006). Although the precise effect of loss of Dnmt3b on reprogramming remains to be elucidated, this may be an essential step towards later epigenetic changes in germ cells.

Considering that the onset of Stella expression in PGCs in EpiSC cultures is associated with an irreversible cell fate decision towards PGCs, we expected other key epigenetic changes in these cells, including reactivation of the inactive X chromosome in emerging EpiSCs. Indeed, most of the Blimp1-/Oct4+ and Blimp1+/Oct4+cells exhibited accumulation of tri-methyl histone H3 lysine 27 (H3K27me3),which is diagnostic for the inactive X chromosome(Fig. 3B, arrows). By contrast,H3K27me3 spot was detectable in only about 30% of Stella+ cells,indicating the commencement of X-reactivation. In addition, global levels of H3K27me3 increased (Fig. 3B,arrowhead), but the levels of histone H3 lysine 9 (H3K9me2) decreased in the Stella+ cells (Fig. 3C), as is the case in nascent PGCs in vivo(Hajkova et al., 2008; Seki et al., 2005). Consistently, the levels of Glp transcripts were downregulated[Glp encodes a protein that together with G9a is essential for H3K9me2 methylation (Tachibana et al.,2005); also known as Ehmt1 - Mouse Genome Informatics;see Fig. S3 in the supplementary material]. These changes are amongst the first steps towards extensive epigenetic reprogramming events in germ cells.

Pluripotent EGCs, which exhibit striking similarities to ESCs, can be derived from PGCs (Matsui et al.,1992; Resnick et al.,1992). Notably, PGCs acquire the potential to dedifferentiate into EGCs after E8.5, when the appropriate epigenetic reprogramming process is completed (Hayashi and Surani,2009). To evaluate whether PGCs from EpiSCs can undergo such reprogramming, we cultured FACS-sorted Stella-GFP+ cells under conditions that promote the derivation of EGCs (see Materials and methods). We performed four independent culture experiments using 3000 Stella-GFP+ cells in each experiment, which resulted in the derivation of three EGC lines from independent Stella-EpiSC clones. Compared with EGC derivation from PGCs isolated from embryos in vivo(Matsui et al., 1992; Resnick et al., 1992), the derivation frequency from Stella+ cells in vitro was less efficient(data not shown). Nevertheless, these EGCs derived from EpiSCs showed appropriate gene expression and epigenetic changes. For example, we detected re-expression of Stella and Pecam1, which are not expressed in EpiSCs(Fig. 3D). Q-PCR analysis of other marker genes further confirmed that the EGCs from Stella-GFP+cells were similar to ESCs and clearly distinguishable from EpiSCs(Fig. 3E). Furthermore, we could no longer detect accumulation of H3K27me3 in these EGCs(Fig. 3F). The signalling pathway involved in the self-renewal of EGCs switched to LIF/Bmp4 (as is the case for ESC) from the bFGF/Activin that is required for EpiSCs(Fig. 3G). Thus, cumulative evidence demonstrates that intrinsic reprogramming events during the formation of PGCs in vitro converts the epigenetic state of EpiSCs to that of EGCs,which is equivalent to ESCs. We exclude the possibility, however unlikely,that EpiSCs could directly dedifferentiate into ESC-like cells under these culture conditions, as bisulphite sequence analysis of EGCs revealed the erasure of allelic DNA methylation associated with the imprinted gene loci,supporting the case for their origin from PGCs(Fig. 3H).

Finally, we went on to investigate whether the Stella+ cells can be induced to proliferate and undergo further development in vitro, which is restricted under EpiSC culture conditions. To investigate this possibility, we co-cultured Stella-GFP+ cells with E12.5 female gonadal cells,which also contained endogenous germ cells. This showed that the in vitro generated PGCs were able to proliferate and develop further(Fig. 4A,B). We found that clumps of Stella-GFP+ cells, which probably originated from single cells, were composed of between two and ten cells after 7 days of culture(Fig. 4B,C), indicating that they had undergone up to four cell divisions. After 7-10 days of culture, we detected expression of the mouse vasa homologue Mvh (Ddx4 - Mouse Genome Informatics; Fig. 4C), as well as of Sycp3, a marker of meiosis, in the Stella-GFP+ cells(Fig. 4D). We also observed a progressive reduction in the levels of Stella-GFP expression in some of the Sycp3+ cells (Fig. 4D), suggesting that these cells had undergone further differentiation, similar to the events decribed in vivo(Sato et al., 2002). After about 40 days in culture, we observed oocyte-like cells with strong expression of Stella-GFP and a characteristically large nucleus(Fig. 4E). However, the efficiency of the formation oocyte-like cells was very low (0.6±0.8 oocyte-like cells/3000 Stella-GFP+ cells; n=6), although these cells were morphologically indistinguishable from oocytes that developed from the endogenous germ cells present in the same co-cultures(Fig. 4F). These results demonstrate that Stella+ cells originating from EpiSCs have the key hallmarks of the germ cell lineage.

Our study shows that self-renewing EpiSCs, unlike ESCs, produce a continuous stream of PGCs in vitro (Fig. 4G) that can potentially generate an infinite number of germ cells. The emergence of PGCs from EpiSCs in vitro is accompanied by appropriate epigenetic reprogramming events. Notably, the intrinsic reprogramming associated with PGC specification results in the erasure of the epigenetic memory of EpiSCs. This is evident because EGCs derived from PGCs resemble ESCs and not EpiSCs, suggesting that they acquire an earlier ICM-like epigenetic state. This in vitro model for the derivation of PGCs from EpiSCs will be particularly useful for investigations on the underlying mechanisms of epigenetic reprogramming events in early germ cells. The fact that EpiSCs resemble hESCs also raises a possibility that our investigation could be extended to hESCs, with the prospect of studying the earliest stages of the human germ cell lineage.

We thank Drs Mitinori Saitou and Shinichiro Chuma for Blimp1-GFP mice and the Sycp3 antibody, respectively. This work was supported by the Japan Society for Promotion of Science to K.H. and the Wellcome Trust(062801). Deposited in PMC for release after 6 months.

Ancelin, K., Lange, U. C., Hajkova, P., Schneider, R.,Bannister, A. J., Kouzarides, T. and Surani, M. A. (
2006
). Blimp1 associates with Prmt5 and directs histone arginine methylation in mouse germ cells.
Nat. Cell Biol.
8
,
623
-630.
Brons, I. G., Smithers, L. E., Trotter, M. W., Rugg-Gunn, P.,Sun, B., Chuva de Sousa Lopes, S. M., Howlett, S. K., Clarkson, A.,Ahrlund-Richter, L., Pedersen, R. A. et al. (
2007
). Derivation of pluripotent epiblast stem cells from mammalian embryos.
Nature
448
,
191
-195.
Chuva de Sousa Lopes, S. M., Hayashi, K., Shovlin, T. C.,Mifsud, W., Surani, M. A. and McLaren, A. (
2008
). X chromosome activity in mouse XX primordial germ cells.
PLoS Genet.
4
,
e30
.
de Napoles, M., Nesterova, T. and Brockdorff, N.(
2007
). Early loss of Xist RNA expression and inactive X chromosome associated chromatin modification in developing primordial germ cells.
PLoS ONE
2
,
e860
.
Durcova-Hills, G. and Surani, A. (
2008
). Reprogramming primordial germ cells (PGC) to embryonic germ (EG) cells.
Curr. Protoc. Stem Cell Biol.
Chapter 1,Unit1A.3.
Durcova-Hills, G., Tang, F., Doody, G., Tooze, R. and Surani, M. A. (
2008
). Reprogramming primordial germ cells into pluripotent stem cells.
PLoS ONE
3
,
e3531
.
Enver, T., Pera, M., Peterson, C. and Andrews, P. W.(
2009
). Stem cell states, fates, and the rules of attraction.
Cell Stem Cell
4
,
387
-397.
Evans, M. J. and Kaufman, M. H. (
1981
). Establishment in culture of pluripotential cells from mouse embryos.
Nature
292
,
154
-156.
Graf, T. and Stadtfeld, M. (
2008
). Heterogeneity of embryonic and adult stem cells.
Cell Stem Cell
3
,
480
-483.
Guo, G., Yang, J., Nichols, J., Hall, J. S., Eyres, I.,Mansfield, W. and Smith, A. (
2009
). Klf4 reverts developmentally programmed restriction of ground state pluripotency.
Development
136
,
1063
-1069.
Hajkova, P., Ancelin, K., Waldmann, T., Lacoste, N., Lange, U. C., Cesari, F., Lee, C., Almouzni, G., Schneider, R. and Surani, M. A.(
2008
). Chromatin dynamics during epigenetic reprogramming in the mouse germ line.
Nature
452
,
877
-881.
Hayashi, K. and Surani, M. A. (
2009
). Resetting the epigenome beyond pluripotency in the germ line.
Cell Stem Cell
4
,
493
-498.
Hayashi, K., de Sousa Lopes, S. M. and Surani, M. A.(
2007
). Germ cell specification in mice.
Science
316
,
394
-396.
Hayashi, K., de Sousa Lopes, S. M., Tang, F. and Surani, A.(
2008
). Dynamic equilibrium and heterogeneity of mouse pluripotent stem cells with distinct functional and epigenetic states.
Cell Stem Cell
3
,
391
-401.
Imamura, M., Miura, K., Iwabuchi, K., Ichisaka, T., Nakagawa,M., Lee, J., Kanatsu-Shinohara, M., Shinohara, T. and Yamanaka, S.(
2006
). Transcriptional repression and DNA hypermethylation of a small set of ES cell marker genes in male germline stem cells.
BMC Dev. Biol.
6
,
34
.
Kurimoto, K., Yabuta, Y., Ohinata, Y. and Saitou, M.(
2007
). Global single-cell cDNA amplification to provide a template for representative high-density oligonucleotide microarray analysis.
Nat. Protoc.
2
,
739
-752.
Kurimoto, K., Yabuta, Y., Ohinata, Y., Shigeta, M., Yamanaka, K. and Saitou, M. (
2008
). Complex genome-wide transcription dynamics orchestrated by Blimp1 for the specification of the germ cell lineage in mice.
Genes Dev.
22
,
1617
-1635.
Lawson, K. A., Dunn, N. R., Roelen, B. A., Zeinstra, L. M.,Davis, A. M., Wright, C. V., Korving, J. P. and Hogan, B. L.(
1999
). Bmp4 is required for the generation of primordial germ cells in the mouse embryo.
Genes Dev.
13
,
424
-436.
Lin, Y., Wong, K. and Calame, K. (
1997
). Repression of c-myc transcription by Blimp-1, an inducer of terminal B cell differentiation.
Science
276
,
596
-599.
Lucifero, D., Mertineit, C., Clarke, H. J., Bestor, T. H. and Trasler, J. M. (
2002
). Methylation dynamics of imprinted genes in mouse germ cells.
Genomics
79
,
530
-538.
Matsui, Y., Zsebo, K. and Hogan, B. L. (
1992
). Derivation of pluripotential embryonic stem cells from murine primordial germ cells in culture.
Cell
70
,
841
-847.
McLaren, A. and Lawson, K. A. (
2005
). How is the mouse germ-cell lineage established?
Differentiation
73
,
435
-437.
Niwa, H. (
2007
). How is pluripotency determined and maintained?
Development
134
,
635
-646.
Ohinata, Y., Payer, B., O'Carroll, D., Ancelin, K., Ono, Y.,Sano, M., Barton, S. C., Obukhanych, T., Nussenzweig, M., Tarakhovsky, A. et al. (
2005
). Blimp1 is a critical determinant of the germ cell lineage in mice.
Nature
436
,
207
-213.
Payer, B., Chuva de Sousa Lopes, S. M., Barton, S. C., Lee, C.,Saitou, M. and Surani, M. A. (
2006
). Generation of stella-GFP transgenic mice: A novel tool to study germ cell development.
Genesis
44
,
75
-83.
Peerani, R., Rao, B. M., Bauwens, C., Yin, T., Wood, G. A.,Nagy, A., Kumacheva, E. and Zandstra, P. W. (
2007
). Niche-mediated control of human embryonic stem cell self-renewal and differentiation.
EMBO J.
26
,
4744
-4755.
Resnick, J. L., Bixler, L. S., Cheng, L. and Donovan, P. J.(
1992
). Long-term proliferation of mouse primordial germ cells in culture.
Nature
359
,
550
-551.
Saitou, M., Barton, S. C. and Surani, M. A.(
2002
). A molecular programme for the specification of germ cell fate in mice.
Nature
418
,
293
-300.
Sato, M., Kimura, T., Kurokawa, K., Fujita, Y., Abe, K.,Masuhara, M., Yasunaga, T., Ryo, A., Yamamoto, M. and Nakano, T.(
2002
). Identification of PGC7, a new gene expressed specifically in preimplantation embryos and germ cells.
Mech. Dev.
113
,
91
-94.
Seki, Y., Hayashi, K., Itoh, K., Mizugaki, M., Saitou, M. and Matsui, Y. (
2005
). Extensive and orderly reprogramming of genome-wide chromatin modifications associated with specification and early development of germ cells in mice.
Dev. Biol.
278
,
440
-458.
Seki, Y., Yamaji, M., Yabuta, Y., Sano, M., Shigeta, M., Matsui,Y., Saga, Y., Tachibana, M., Shinkai, Y. and Saitou, M.(
2007
). Cellular dynamics associated with the genome-wide epigenetic reprogramming in migrating primordial germ cells in mice.
Development
134
,
2627
-2638.
Sugimoto, M. and Abe, K. (
2007
). X chromosome reactivation initiates in nascent primordial germ cells in mice.
PLoS Genet.
3
,
e116
.
Surani, M. A., Hayashi, K. and Hajkova, P.(
2007
). Genetic and epigenetic regulators of pluripotency.
Cell
128
,
747
-762.
Tachibana, M., Ueda, J., Fukuda, M., Takeda, N., Ohta, T.,Iwanari, H., Sakihama, T., Kodama, T., Hamakubo, T. and Shinkai, Y.(
2005
). Histone methyltransferases G9a and GLP form heteromeric complexes and are both crucial for methylation of euchromatin at H3-K9.
Genes Dev.
19
,
815
-826.
Takahashi, K., Mitsui, K. and Yamanaka, S.(
2003
). Role of ERas in promoting tumour-like properties in mouse embryonic stem cells.
Nature
423
,
541
-545.
Tesar, P. J., Chenoweth, J. G., Brook, F. A., Davies, T. J.,Evans, E. P., Mack, D. L., Gardner, R. L. and McKay, R. D.(
2007
). New cell lines from mouse epiblast share defining features with human embryonic stem cells.
Nature
448
,
196
-199.
Yabuta, Y., Kurimoto, K., Ohinata, Y., Seki, Y. and Saitou,M. (
2006
). Gene expression dynamics during germline specification in mice identified by quantitative single-cell gene expression profiling.
Biol. Reprod.
75
,
705
-716.
Yamaji, M., Seki, Y., Kurimoto, K., Yabuta, Y., Yuasa, M.,Shigeta, M., Yamanaka, K., Ohinata, Y. and Saitou, M. (
2008
). Critical function of Prdm14 for the establishment of the germ cell lineage in mice.
Nat. Genet.
40
,
1016
-1022.
Yu, P. B., Hong, C. C., Sachidanandan, C., Babitt, J. L., Deng,D. Y., Hoyng, S. A., Lin, H. Y., Bloch, K. D. and Peterson, R. T.(
2008
). Dorsomorphin inhibits BMP signals required for embryogenesis and iron metabolism.
Nat. Chem. Biol.
4
,
33
-41.

Supplementary information