To clarify how the VE contributed to the development of PGCs, we assessed their formation in explanted Blimp1:gfp E5.75–E6.0 embryos, from which the VE (-VE) was surgically removed (Fig. 1A). In this transgenic model, membrane-tagged GFP is driven by the Blimp1 promoter 2. At E5.75–E6.0, Blimp1(GFP)-positive PGC precursor cells have not yet emerged in the proximal epiblast (asterisks in Fig. 1B), although Blimp1(GFP) is expressed in the VE (white arrows in Fig. 1B). The expression of the transgenic explants was monitored after 1 and 3 days of culture. In -VE explants, no Blimp1(GFP)-positive cells (n = 0/9) were observed after 1 day (Fig. 1C) or 3 days of culture (Fig. 1D). In whole embryo explants, Blimp1(GFP)-positive cells were still observed in the VE after 1 day of culture (white arrow in Fig. 1E), but after 3 days of culture a clear cluster of Blimp1(GFP)-positive PGC precursors was observed at the boundary with the extraembryonic region (n = 4/5) and Blimp1(GFP) expression was downregulated in the VE (Fig. 1F). Together, these observations suggested an important role of the VE in the induction of the Blimp1-positive PGC precursors. Note that the small size of the -VE explants after 3 days of culture should not hamper the specification of PGCs (Fig. 3).

We also studied whether the ExE was involved in formation of PGC precursors or in the subsequent commitment of PGC precursors to PGCs. For this we examined the number of PGC (and PGC precursors) in E5.75–E6.0 Blimp1:gfp, Stella:gfp and ΔPEOct4:gfp explanted embryos cultured in the absence of ExE (Fig. 1A,G–J). After 1 day of culture, similar to whole Blimp1:gfp embryo explants, Blimp1(GFP)-positive cells were observed only in the VE of -ExE explants (white arrow in Fig. 1E,G). However, after 3 days of culture of -ExE explants (n = 13/16), Blimp1(GFP) was also downregulated in the VE, but the epiblast-derivative was entirely Blimp1(GFP)-positive (red arrow in Fig. 1H). Note that the dimensions of the epiblast-derivative in both whole and -ExE explants are similar (cartoon in Fig. 1F,H).

When Stella:gfp or ΔPEOct4:gfp E5.75–E6.0 embryos were cultured for 3 days, no GFP expression (indicative of specified PGCs 89) was observed in whole (Fig. 1I,J) and -ExE explants (bottom left inserts in Fig. 1I,J), suggesting that specification of PGCs had not yet occurred in vitro. After 4 days of culture, the Blimp1(GFP)-positive cluster in the explanted whole embryos gave rise to TNAP-positive PGCs (Fig. 1K), whereas the Blimp1(GFP)-expressing cells in -ExE explants remained TNAP-negative (data not shown). This suggested that the Blimp1(GFP)-positive cells that populated the epiblast in the -ExE explants are either no genuine PGC precursors, or if they are PGC precursors, they may need additional signals to commit to the PGC lineage. In any case, these experiments demonstrate that the ExE is (directly or indirectly) important for the restriction of Blimp1(GFP) induction in the epiblast by the VE.

The absence of Blimp1(GFP)-positive PGC precursor cells in E5.75–E6.0 Blimp1:gfp -VE explants could be a secondary effect due to defects in the initiation of morphogenetic movements associated with gastrulation. Therefore, we cultured slightly older gastrulating embryos (E6.75–E7.0), which have a distinct Blimp1(GFP)-positive cluster of ±20 PGC precursors (Fig. 2A and 2) after having removed the still Blimp1(GFP)-positive VE (-VE). After 3 days in culture, the number of PGCs (Stella-positive and TNAP-positive) observed in Blimp1:gfp E6.75–E7.0 -VE explants was reduced compared to explants from whole embryos, but more importantly, the number of Blimp1(GFP)-expressing cells was also correspondingly reduced by half (±40 cells in the whole embryo explant compared to ±20 cells in the -VE explant) (Fig. 2B–J). Note that some of the Blimp1(GFP)-positive cells were small and probably entering apoptosis (contained fragmented DAPI-positive nucleus). These smaller cells failed to upregulate Stella, but were positive for TNAP (including a TNAP-positive cytoplasmic spot) (Fig. 2B,E,F,I,J). In agreement, we observed higher levels of apoptosis using the TUNEL-assay in the -VE explants when compared to whole explants (data not shown).

Notably, the PGC precursors present in the -VE explants undergo specification to PGCs, even though the PGC precursor cluster was rather reduced in size (Fig. 2B–J), indicating that the size of the cluster is not important for specification. Also unexpected was the fact that some of those PGCs acquired a migratory phenotype and were moving out of the PGC-cluster, similarly to PGCs generated from whole E6.5 embryo-explants (asterisks in Fig. 2C,F) and therefore do not seem to require signals from the VE to do so.

The ectopic expression of Blimp(GFP) in the -ExE explants derived from E5.75–E6.0 Blimp1:gfp embryos raised the possibility that the ExE may be critical for regionalizing the embryo at E5.75–E6.0, so that the Blimp1-positive PGC precursors can only emerge posteriorly. To investigate this in more detail, we cut E5.75–E6.0 transgenic Cerl:gfp embryos 10 longitudinally though the long left-right (L-R) axis (Fig. 3A,B). Cerberus-like (Cerl) is an anterior visceral endoderm (AVE) marker 11 and therefore in Cerl:gfp transgenic embryos GFP expression is present exclusively in the anterior region, allowing clear identification of the embryonic axes (Fig. 3B,C).

The two embryo-halves, each containing epiblast, VE and ExE, were cultured separately for 4 days. After culture, both embryo halves were stained for alkaline phosphatase-activity to assess the presence of TNAP-positive specified PGCs. Notably, the explants-pairs that contained PGCs only exhibited them in one of the explant-halves (Fig. 3D,E) invariably the posterior part (Cerl(GFP)-negative). Only one explant-pair (n = 1/30) showed PGCs in both explant-halves (Fig. 3E). Our data indicates that the anterior VE and anterior ExE were not able to support the formation of PGCs and may even play an active role in repressing Blimp1 expression.

To directly test whether the VE and ExE can continuously recruit PGC precursors from the proximal epiblast cells between E6.25–E7.25, we investigated the capacity for the de novo formation of PGCs in E6.75–E7.0 Stella:gfp embryos from which the original cluster of PGC precursors had been removed. To do this, we either removed the entire VE (which has the form of a cup) and introduced inside a small piece of distal epiblast together with a small piece of ExE (-cluster) (Fig. 4A), or alternatively we removed the entire region of the embryo containing both the proximal epiblast and surrounding proximal VE, and brought the distal embryonic part in contact with the extraembryonic part (-cluster -proximal VE) (Fig. 4A). The reconstituted-embryos were analysed after 3 days of culture. In '-cluster -proximal VE' explants, we observed no induction of PGCs (n = 0/12), however in '-cluster' explants we were able to generate PGCs (n = 18/40, containing 1–38 PGCs) expressing both Stella(GFP) and TNAP (Fig. 4B–G). Together, the data clearly show that the combination of proximal VE and ExE is not only necessary, but also sufficient to induce PGCs in distal epiblast cells, which usually do not give rise to PGCs in the embryo.

To exclude the possibility that the induced PGCs resulted from contamination of proximal epiblast cells present in the ExE, we made reconstituted-embryos using ROSA26:LacZ distal epiblast and wild type VE and ExE. These reconstituted-embryos showed that the newly induced TNAP-positive PGCs are LacZ-positive, demonstrating that the PGCs indeed originated from the ROSA26:LacZ distal epiblast (Fig. 4H–J).

In has been shown in a previous report that a group of E6.5 proximal posterior cells transplanted into a distal position in the embryo were unable to specify as PGCs 12. However, the existence of the Blimp1-positive cluster of PGC precursors was then unknown. Therefore, it is possible that the group of proximal-posterior cells transplanted into the distal part of the embryo did not contain any such Blimp1-positive PGC-precursors. To unambiguously investigate whether Blimp1-expressing PGC precursors could specify to PGCs when placed in the distal epiblast, we transferred small groups of cells including Blimp1(GFP)-positive cells from Blimp1:gfp ROSA26:LacZ E6.5 embryos to the distal epiblast of E6.5 wild type embryos (n = 4) (Fig. 5A). These experiments demonstrated that the (LacZ-positive and initially Blimp1(GFP)-positive) cells could no longer give rise to TNAP-positive PGCs (Fig. 5B) after 3 days of culture. Instead, these cells were incorporated into morphologically more anterior structures faraway from the PGC region in the explant (Fig. 5B). By contrast, transplantation of groups of cells from a proximal-posterior (n = 4) location of E6.5 Blimp1:gfp ROSA26:LacZ embryos into a position close to the location of the PGC-precursor cluster in E6.5 wild type embryos (Fig. 5A), resulted in the formation of a LacZ-positive cluster in the posterior region of the embryo where a few double LacZ-positive and TNAP-positive PGCs were observed next to wild type PGCs (Fig. 5C). These results are in agreement with the previous report from 12 and together demonstrate that Blimp1-expressing PGC precursors still require local signals present exclusively in the proximal-posterior part of the embryo before they can undergo specification to PGCs.

At E5.5, the anterior and posterior ExE exhibit already different characteristics important for the correct patterning (and migration) of the AVE and the subsequent patterning (and migration) of the epiblast 1314. At E5.75, the ExE may no longer influence the migration of the AVE, but it still plays a crucial role in regionalizing the epiblast, allowing gastrulation to take place. Indeed, Elf5 mutants, which contain no ExE, showed defects in embryonic patterning and failure to undergo gastrulation, despite the fact that the AVE is well specified 15. In agreement, our data also suggests that ExE plays a critical role at E5.75–E6.0 in regionalizing the epiblast with respect to where the PGC precursors can emerge.

The anterior ExE (and AVE) does not support development of PGCs and may even inhibit the formation of Blimp1-positive cells. However, it is well known that cells lying throughout the perimeter of the proximal epiblast at E6.0 can potentially give rise 40 hour later to PGCs in the posterior part of the embryo 3. These proximal cells move to the posterior part of the embryo due to the characteristic morphogenetic movements occurring during gastrulation. This directional movement allows some of these cells either to avoid signals from the anterior part of the embryo or to receive the necessary signals in the posterior part to become PGC progenitors.

The VE at E5.75 shows asymmetric expression of several secreted factors, including Cerl and Lefty1 in the AVE, important for the formation of the anteroposterior axis 1617. By contrast, there is little evidence for a direct role of the posterior part of the VE in embryonic patterning.

Our data suggests that the proximal-posterior VE plays a role in the generation of the full population of PGC precursors in the posterior part of the embryo. The role of the proximal-posterior VE could be that of promoting proliferation of the initial pool of PGC progenitors and therefore removing the VE would block cell division of the PGC progenitors, resulting in a smaller PGC population. Alternatively, as the VE is required for general cell survival in culture, we cannot exclude the possibility that normal numbers of PGC precursors were formed but died and could not be traced. However, we favor the possibility that when the VE is removed no more PGC precursors emerge but the ones allocated are able to undergo specification, although due to limitation of the culture system used some degenerate. Therefore, we propose that the proximal-posterior VE (together with the ExE) is actively recruiting additional PGC precursors from epiblast cells between E6.25–E7.25. This model also explains our observations that an E6.5 embryo is able to generate a PGC cluster de novo, when its original cluster is removed and that E6.5 distal epiblast cells when placed proximally (near a PGC precursor existing cluster) are still able to adopt a PGC fate (data not shown).

The results presented show that cross talk between the posterior epiblast, the proximal VE and ExE is needed for the emergence of Blimp1-positive PGC precursors. In addition, we also show that Blimp1-positive PGC precursors do not autonomously undergo specification to PGCs. It is likely that extra inductive signals present locally trigger the specification of Blimp1-positive PGC precursors. The proposed extra inductive signals most probably originate from the extraembryonic mesoderm (ExM) surrounding the PGC progenitor cluster and not the VE and ExE. In contrast to our ideas, Yoshimizu and colleagues (2001) observed that E6.75 dissociated epiblast cells could give rise to PGCs 6, however they cultured those epiblast cells in the presence of feeder cells, which could be providing the necessary signals for specification.

One of the candidate secreted molecules exclusively expressed by the ExM, and not by PGCs, at E7.5 is BMP4 18. In addition, a necessary event during PGC specification is the downregulation of E-cadherin 19, followed by the acquisition of a motile phenotype. Both FGF and WNT signalling pathways are known to regulate gastrulation and can downregulate E-cadherin 2021222324. Further research will demonstrate whether the BMP, WNT or FGF signalling pathways are directly involved in PGC specification.

All experiments involving animals were approved by the Home Office, UK. Wild type mice and transgenic Blimp1:gfp, Stella:gfp, ΔPEOct4:gfp and Cerl:gfp mice were on a mixed C57BL/6 and CBA background, ROSA26:LacZ mice were C57BL/6. Noon of the day of vaginal plug was designated E0.5. Embryos were isolated at E5.75–E6.0 and E6.75–E7.0 in cold Dulbecco's minimal medium (DMEM, Invitrogen) supplemented with 7.5% fetal calf serum (FCS) and 10 mM HEPES. Blimp1:gfp embryos were selected from non-transgenic littermates under an Olympus IX71 inverted microscope (London, UK). If necessary, the VE was removed as described previously 5. Embryos were cut (1) longitudinally through the long L-R axis; (2) transversely at several heights in the presence of absence of the VE; or (3) parts of one or several different embryos were recombined using tungsten needles. Thereafter, they were cultured in DMEM+15% FCS at 37°C and 5% CO₂ on air either on fibronectin (20 μg/ml) coated glass coverslips in 4 well culture plates (NUNC) or in suspension in 3 cm bacteriologic dishes. The two methods were used to obtain explants attached to a coverslip or to maintain the explants in suspension, respectively. The two methods gave similar results in terms of PGC development. Explants derived from E5.75–E6.0 embryos were cultured for maximal 4 days and explants derived from E6.75–E7.0 embryos for maximal 3 days. Fresh embryos and explants were imaged using an Olympus IX71 (London, UK) inverted microscope.

Immunofluorescence staining of explants on coverslips was performed essentially as described 5. The primary antibodies used were rat anti-GFP (1:500, Nacalai Tesque) and rabbit anti-PGC7/Stella (1:2000); and the secondary antibodies used were from Molecular Probes. Explants were imaged using an Olympus IX71 (London, UK) inverted microscope or a Bio-Rad Radiance 2000 confocal microscope (CA, USA).

Some of the explants (grown in suspension or on coverslips) were used for detection of alkaline phosphatase-activity. Briefly, the explants were washed three times in PBS, kept 1 hour in 70% EtOH at 4°C, washed three times with distilled water and treated with an aqueous solution containing 1% veronal (Sigma), 0.12% MgCl₂, 0.02 mg/ml α-naphtyl phosphate (Sigma) and 1 mg/ml Fast Red TR (Sigma).

Recombinates of wild type and ROSA26:LacZ embryos were fixed in 4% paraformaldehyde in PBS, stained for LacZ using standard procedures, embedded in Technovit 8100 (Heraeus Kulzer), sectioned (6 μm) and stained for alkaline phosphatase-activity using ASMX/Fast Red TR (Sigma) following manufacturer's instructions. Explants were imaged using an Olympus IX71 (London, UK) inverted microscope.