The germ cell lineage is a specified cell population that passes through a series of differentiation steps before giving rise, eventually, to either eggs or sperm. We have investigated the manner in which primordial germ cells (PGCs) are reprogrammed in vitro to form pluripotent stem cells in response to exogenous fibroblast growth factor-2 (FGF-2). The response is dependent on time of exposure and concentration of FGF-2. PGCs isolated in culture show a motile phenotype and lose any expression of a characteristic germ cell marker, mouse vasa homolog. Subsequently, some but not all of the cells show further changes of phenotype, accompanied by changes in expression of endogenous FGF-2 and up-regulation of its receptor, fibroblast growth factor receptor-3, in the nucleus. We propose that it is from this reprogrammed component of the now heterogeneous PGC population that pluripotent stem cells arise.
Pluripotent cells may be defined pragmatically as cells that contribute to multiple lineages if introduced into host blastocysts. Thus, cleavage-stage blastomeres and cells from the blastocyst inner cell mass (ICM) are pluripotent. Embryonic stem (ES) cells derived from the ICM/epiblast are not only pluripotent but, if leukemia inhibitory factor (LIF) is present in the culture medium, they are capable of proliferating indefinitely in an undifferentiated state. If LIF is removed, ES cells give rise to a range of differentiated cell types, including primordial germ cells that can develop into oocytes and sperm [1–3]. PGCs can also give rise to pluripotent stem cells, termed embryonic germ (EG) cells.
PGCs constitute a specified cell lineage; hence, the transition from PGCs to EG cells involves dedifferentiation. The basis for this is obscure, although the derivation of EG cells from PGCs is known to be associated with signaling pathways induced by growth factors and their receptors . PGCs isolated from mouse embryos of different gestational stages and cocultured with feeder cells will proliferate for several days but then cease division [5–7]. Growth factors added to the culture medium were found to extend the proliferation and survival of PGCs to some degree [4, 8, 9], but only when stem cell factor (SCF), LIF, and fibroblast growth factor-2 (FGF-2) were combined together did PGCs continue to proliferate beyond their in vivo fate [10, 11]. Once the multicellular colonies were formed, FGF-2 was not required in the culture medium for their further subculture and expansion to produce pluripotent cell lines. EG cells have been derived from PGCs soon after their initial specification, as well as during their migratory period and after arrival in the genital ridge. Pluripotent stem cell lines have recently been derived also from spermatogonial cells taken from newborn mouse testes, representing a still further stage of germ cell differentiation .
FGFs exert their biological effects in an autocrine or paracrine fashion, whereby they interact with transmembrane tyrosine kinase fibroblast growth factor receptors (FGFRs) and promote their dimerization and activation . Although many studies have revealed general roles for FGF signaling in specific developmental events, there is a lack of information about the role of FGF signaling during the reprogramming of PGCs into EG cells. In this study, we investigated under what conditions and duration of culture PGCs dedifferentiate, lose their germ-cell characteristics, and acquire the developmental pluripotency of EG cells. In particular, we examined the expression of FGF-2 and FGFRs in PGCs and EG cells, as well as in cultured PGCs exposed to exogeneous FGF-2, FGF-5, FGF-9, and FGF-10.
We showed that after 9–10 days of PGC culture, multicellular PGC colonies contain cells that can make chimeric fetuses in vivo, thus demonstrating their pluripotency. Freshly isolated PGCs did not make chimeras. Exposure to a sufficient concentration of exogenous FGF-2 for just the first 24 hours of PGC culture proved to be critical: in its presence, the PGCs underwent morphological changes, endogenous FGF-2 was upregulated, and high expression of FGFR-3 was seen in PGC nuclei as well as in the cytoplasm. In the absence of FGF-2, none of these changes were observed, and addition of FGF-2 after an interval of 24 hours did not restore the capacity to form pluripotent stem cells. Our in situ immunofluorescence results on cultured PGCs suggest that PGCs are first dedifferentiated, and subsequently a small proportion of them is reprogrammed into EG cells.
Materials and Methods
MF1 females mated with homozygous Rosa26 males (mixed background)  were used to provide fetuses. For aggregation and blastocyst injection experiments, MF1 or (C57BL/6 × CBA) F1 embryos were recovered at 2.5 or 3.5 days postcoitum (dpc). The morning of the vaginal plug was counted as 0.5 dpc.
Dissection of PGC-containing tissues and the culture conditions for derivation of EG cells were as described . Briefly, PGC suspension was seeded onto mitotically inactive Sl4-m220 cells (producing SCF) in Dulbecco's modified Eagle's medium (DMEM) containing nonessential amino acids, l-glutamine, penicillin/streptomycin (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), pyruvic acid (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), and 15% fetal calf serum (Sigma-Aldrich). The medium during primary culture was supplemented with both LIF (1,200 IU/ml; Chemicon, Temecula, CA, http://www.chemicon.com) and FGF-2 (human recombinant 25 ng/ml; Invitrogen). After 9–10 days of primary culture, colonies with an embryonic germ cell (EGC)-like phenotype were observed. Some were used for either aggregation or blastocyst injection experiments. Others were subcultured, and EG cell lines were derived. FGF-2, FGF-5, FGF-9, and FGF-10 were obtained from R&D Systems Inc. (Minneapolis, http://www.rndsystems.com). SU5402 (final concentration, 10 or 25 μM; Calbiochem, San Diego, http://www.emdbiosciences.com) was added to the culture medium immediately or after 1 day of culture. Control cultures were cultured with an equal volume of the vehicle dimethyl sulfoxide. For pulse-treated culture experiments, 8.5- and 11.5-dpc PGC cultures were treated with FGF-2 for 1, 3, or 5 days (6 days in the case of 11.5-dpc PGCs) and then washed with phosphate-buffered saline (PBS) and cultured with medium containing only LIF. After 10 days, cultures were stained for tissue nonspecific alkaline phosphatase (TNAP) activity , and EGC-like colonies were counted. In reverse cultures, PGCs were cultured for 1 or 3 days in medium containing only LIF, and then FGF-2 was added. Some EGC-like colonies from pulse-treated cultures were subcultured, and EG cell lines were derived. Differentiation of EG cells was induced as described .
Polymerase Chain Reaction for FGFR-1, -2, -3, and -4
We designed primers to the transmembrane (TM) domains of the four FGF receptors (supplemental online Table 1). The specificity of the primers was confirmed by using cloned regions of FGFR-1, FGFR-2, FGFR-3, and FGFR-4 as templates for polymerase chain reaction (PCR). Total RNA was made from 11.5-dpc urogenital ridges, 12.5-dpc male and female genital ridges, and Sl4-m220 cells, using TRI reagent (Sigma-Aldrich). Twenty microliters of cDNA was prepared from 1 μg of total RNA by oligo(dT) priming (Superscript II; Invitrogen). cDNAs made from purified 11.5 and 12.5 PGCs were described previously .
The PCR conditions for the FGFRs were as follows: 15 minutes at 94°C; 35 or 40 cycles of 30 seconds at 94°C, 30 seconds at 65°C, and 40 seconds at 72°C; followed by 5 minutes at 72°C. The 50-μl reaction contained 1× Q buffer (Qiagen, Hilden, Germany, http://www1.qiagen.com), supplemented with 200 μM dNTPs, 3 μM of each primer, 2 μl of cDNA, and 0.125 μl (1 μl per five units) of HotStarTaq DNA polymerase (Qiagen). DNA sequencing confirmed the identity of PCR products.
Cultured 11.5-dpc PGCs and EB outgrowths were cultured for 6–10 days on coverslips and then stained for SSEA-1, TROMA-1, fibronectin, and α-internexin . PGC-containing tissues or 8.5-dpc PGCs cultured in Lab-tek chambers (Nunc) were fixed with 2% paraformaldehyde, permeabilized with methanol, and blocked with 10% bovine serum albumin in PBS. Primary antibodies (anti-FGFR-1, -2, -3, or -4 or FGF-2, 1:300) were incubated overnight at 4°C and then washed with PBS and incubated with appropriate secondary antibodies (Molecular Probes Inc., Eugene, OR, http://probes.invitrogen.com). Nuclei were counterstained with Toto-3 (Molecular Probes) and mounted. Samples were observed by confocal microscopy (Bio-Rad 1024 [Bio-Rad, Hercules, CA, http://www.bio-rad.com] attached to a Nikon Eclipse 800 microscope).
Outgrowth of Aggregates, Blastocyst Injections, LacZ Staining, and Histology
Freshly isolated PGCs  or cells from colonies formed after 30, 25, 13, or 9 days of culture (11.5-dpc PGCs) and 24 days of culture (12.5-dpc PGCs) were aggregated with zona-free morulae. After 48 hours, blastocysts were transferred into culture dishes, and after 3–4 days, the outgrowths were stained for LacZ. The purity of isolated PGCs was about 80% as judged by TNAP activity.
Freshly isolated PGCs, or cells from EGC-like colonies from 10-day cultures of male (Ube1x)  11.5-dpc PGCs were injected into blastocysts, which were transferred to pseudopregnant F1 females. Fetuses (8.5, 9.5, and 13.5 dpc) were processed for LacZ .
Characterization of Primary Multicellular Colonies Formed by Dedifferentiated PGCs
Earlier studies showed that PGCs are reprogrammed into pluripotent EG cells by exposing them to FGF-2. Once PGCs form multicellular colonies during primary culture, the presence of exogenous FGF-2 is not required. First, we examined whether cells in such colonies still express markers of differentiated PGCs or whether they have acquired markers of pluripotent stem cells. We found that all cells within primary colonies are positive for TNAP, Oct-4, germ cell nuclear antigen-1, and E-cadherin (data not shown) and did not express MVH (data not shown), a protein expressed by PGC at 11.5 dpc onward. Then, we examined whether cells from these primary multicellular colonies, without subculture, already have ability to form chimeras, or whether this ability only emerges with subculture or in later passages. For this purpose, we devised a rapid in vitro method based on the differences in the developmental potency of PGCs and established EG cells. When LacZ-marked freshly isolated PGCs or EG cells were aggregated with morulae, the resulting blastocysts attached to the substrate, with inner cell mass growing upwards from the trophoblast to form an epiblast. Chimeric epiblasts were identified with X-gal staining for LacZ. We found that EG cells made chimeric epiblasts with high frequency, but freshly isolated PGCs never integrated into the host epiblasts (Table 1). Cultured PGCs at 11.5 or 12.5 dpc formed primary colonies resembling ES cell/EG cell colonies in their morphology (Fig. 1A, arrow) and high TNAP activity (Fig. 1B, red staining) after about 9 days of culture. With these cells, we observed one chimeric epiblast out of 10 (Fig. 1C, blue staining; Table 1). As the period of PGC culture increased to 3 weeks or more, the proportion of chimeric epiblasts increased (Table 1), and the number of LacZ-marked PGCs incorporated in the host ICM increased from just a few, with no evidence of proliferation (Fig. 1D, arrow), to clumps of at least 20 to 40 cells. No LacZ-stained cells were observed in any of the trophoblast outgrowths.
Additional evidence that cells from primary colonies are pluripotent stem cells was provided by their ability to make embryoid bodies in vitro and to make chimeras in vivo. Cells from a few colonies were used to make embryoid bodies that gave rise to endoderm, ectoderm, and mesoderm cell types (data not shown). Other cells from primary colonies of 11.5-dpc male LacZ-marked PGCs cultured for 10 days were injected into host blastocysts. The incidence of chimerism, 41% (7 of 17) in fetuses examined at 13.5 dpc and 56% (13 of 23) in fetuses at 17.5 dpc, appeared higher than we had observed in vitro, probably because the number of cells used per embryo was greater. The 13.5-dpc fetuses were processed for histology: LacZ-marked cells were observed in many tissues (supplemental online Fig. 1). At 17.5 dpc, the chimeric fetuses were weighed and processed for skeletal analysis. Chimeras made with EG cells derived from 11.5- and 12.5-dpc PGCs have previously been reported to show skeletal and growth abnormalities [21, 22]. Our results confirmed these findings (supplemental online Fig. 2A, 2B). When freshly isolated male and female 11.5-dpc PGCs were injected into blastocysts, we did not observe any incorporation of LacZ-marked cells in 39 stained embryos.
Taken together, our results suggest that our primary multicellular colonies already contained pluripotent stem cells derived from reprogrammed PGCs. Cells from primary colonies could be propagated in subsequent passages while keeping their pluripotent stem cell status. We termed the primary multicellular colonies EGC-like colonies.
The Effects of FGF Signaling in the Reprogramming of PGCs to EG Cells
To investigate the role of exogenous FGF-2 in the reprogramming of PGCs into EG cells, we first needed to determine the distribution of endogenous FGF-2. Using immunofluorescence, we examined whether endogenous FGF-2 is present in freshly isolated PGCs. At 8.5 dpc, FGF-2 was predominantly detected in cytoplasm of PGCs (distinguished from somatic cells by green staining with Oct-4 antibody), but we could also detect a weak signal in the nuclei (Fig. 2A, arrow). Expression of FGF-2 was detected in cytoplasm and nuclei of 11.5-dpc and 12.5-dpc PGCs (Fig. 2B–2D) and strongly in both cytoplasm and nuclei of EG cells (Fig. 2E) and cytoplasm of surrounding somatic cells and feeder cells, Sl4-m220 (Fig. 2F). This in vivo expression of FGF-2 in somatic cells and feeder cells was downregulated in vitro under our culture conditions.
The widespread expression of FGF-2 in and around PGCs suggests that the requirement of additional exogenous FGF-2 for EG-cell derivation may have a quantitative basis. We therefore investigated the effect of different concentrations on cultured PGCs. Cultures were treated with 5, 10, or 25 ng of FGF-2 per ml of medium, and TNAP-positive EGC-like colonies were counted after 10 days. Cultures treated with 25 or 10 ng/ml of FGF-2 gave rise to similar numbers of colonies, but cultures treated with 5 ng/ml of FGF-2 generated only few colonies (Fig. 3A). When PGCs were cultured without the addition of FGF-2, no EGC-like colonies were observed (data not shown). Since heparin is known to increase the affinity of FGFs for their receptor , we cultured 8.5-dpc PGCs in the presence of FGF-2 and heparin (10 ng/ml), but no further increase of colonies was observed (Fig. 3A).
Knockout data suggest that functions of many FGFs are redundant or that other FGF members can compensate for FGF-2. We therefore tested whether FGF-5, FGF-9, and FGF-10 at 20 ng/ml would trigger the reprogramming of 8.5-dpc and 11.5-dpc PGCs into EG cells. After 10 days of culture, TNAP colonies were counted. The efficiency of reprogramming of PGCs is very different at 8.5 and 11.5 dpc, since the number of PGCs per embryo is estimated at approximately 150 cells and 5,000 cells, respectively. These estimates are subject to much variation. We therefore expressed our results in terms of number of colonies, standardizing the amount of starting material to allow valid comparisons among the variables in any one experiment. Comparisons between different experiments may be misleading. We found that FGF-2 (which binds all FGFRs) and FGF-5 (which binds FGFR-1 and FGFR-2) were more efficient in generating EGC-like colonies than FGF-9 (which binds FGFR-2, FGFR-3, or FGFR-4) or FGF-10 (which binds FGFR-1 and FGFR-2) (Fig. 3B). We also examined whether combining FGFs would induce a synergistic effect in PGC cultures, but no additive effect was observed (data not shown). Some EGC-like colonies from cultures with FGF-5, FGF-9, and FGF-10 were further expanded, and EG cell lines were derived. All tested lines expressed Oct-4, SSEA-1, and TNAP and made embryoid bodies that differentiated into three germ layers (data not shown).
Exogenous FGF-2 Exhibits Time-Dependent Effects on EGC-Like Colony Formation
Next, we examined the time of exposure to FGF-2 required to give rise to EGC-like colonies. Cultures of 8.5-dpc or 11.5-dpc PGCs were treated with FGF-2 for 1, 3, or 6 days and then cultured in medium supplemented only with LIF. After 10 days, fixed cultures were stained for TNAP activity, and the colonies were counted. Exposure to FGF-2 beyond the first 24 hours did not increase the number of colonies (Fig. 3C). To show that EGC-like colonies arising under those conditions are pluripotent, we subcultured EGC-like colonies, leading to EG cell lines that differentiated into three germ layers (data not shown). To explore the significance of the first 24 hours, we allowed an interval of 24 hours to elapse before FGF-2 was added. In that event, no EGC-like colonies were observed (data not shown), suggesting the possibility that the expression of relevant FGF receptors was modified according to whether or not exogenous FGF-2 was present in the culture medium.
We therefore examined by reverse transcription (RT)-PCR the expression in PGCs of the four FGF receptors  to which FGF-2 can bind. We designed specific primers for the four FGFRs to identify the transcripts. These primers would not have recognized splice variants; however, we were consistently able to detect the transcripts of all four FGFRs in 11.5-dpc urogenital ridges and in 12.5- and 13.5-dpc male and female genital ridges (data not shown). When we analyzed cDNA made from purified 11.5- or 12.5-dpc male or female PGCs (90% purity) , we detected transcripts of FGFR-1 and FGFR-3 in 11.5-dpc PGCs; FGFR-1, -3, and -4 in 12.5-dpc male or female PGCs; and all four FGFRs in EG cells and Sl4-m220 cells (Fig. 4).
Since our RT-PCR data suggest that FGFR-1 and FGFR-3 are expressed in 11.5-dpc PGCs, we examined expression of the FGFR-1 and FGFR-3 proteins by immunofluorescence. At 8.5 dpc, we observed the low expression of FGFR-1 (supplemental online Fig. 3A) and FGFR-3 (Fig. 5A–5C) in PGCs, whereas high expression was observed in the surrounding somatic cells (Fig. 5B, arrow). At 11.5 dpc, FGFR-1 and FGFR-3 were detected in both cytoplasm and nuclei of PGCs (supplemental online Fig. 3B, 3F). At 12.5 dpc, we found FGFR-1 only in cytoplasm and FGFR-3 at high levels in both nuclei and cytoplasm of male PGCs (supplemental online Fig. 3D, 3H), but both FGFR-1 and FGFR-3 were detected in the nuclei and cytoplasm of female PGCs (supplemental online Fig. 3C, 3G). In EG cells, the FGFR-1 protein was found in nuclei and at a low level in cytoplasm (supplemental online Fig. 3E), but the FGFR-3 protein was detected in both nuclei and cytoplasm (Fig. 5M–5O). A few somatic cells isolated in 11.5-dpc urogenital ridge and 12.5-dpc male genital ridge showed FGFR-2 protein, but no FGFR-4 protein was detected (data not shown).
Preliminary data on adding SU5402, a synthetic inhibitor of FGFR-1 tyrosine kinase activity, to the medium at the beginning of PGC culture and counting the number of EGC-like colonies after 10 days suggested that it is FGFR-3 rather than FGFR-1 that transduces the signal(s) triggering the reprogramming of PGCs into EGC-like colonies.
PGCs Exhibit Heterogeneity in Response to Standard Culture Conditions
In vivo, PGC differentiation involves changes in both cell morphology and gene expression. To investigate the impact of exogenously added FGF-2 both on the morphology of cultured PGCs and on the expression of FGF-2 and FGFR-3 proteins, we performed in situ immunofluorescence on the cultures. Cultures were fixed, and PGCs were identified with TG-1 antibody. This recognizes SSEA-1 surface antigen, identifying the PGCs and demonstrating their morphology. We also examined the expression of MVH protein, which in vivo is only expressed in PGCs from 11.5 dpc onward.
Based on SSEA-1 cell-surface staining (green), most 11.5-dpc PGCs were found after 1 day of culture as single cells or two-cell groups, possessing characteristics typical of motile cells (Fig. 6A, 6B). The PGCs were elongated with pseudopodia and fine filopodia extensions (Fig. 6A, inset). Some PGCs expressed MVH at high levels (Fig. 6A, arrows) and some at low levels (Fig. 6B, arrow). After 3 days of culture, we observed many small colonies comprising three to four cells. We classified them into three phenotypes: PGC-like colonies, only motile phenotype (Fig. 6C, arrow), 50%; EGC-like, only nonmotile phenotype (Fig. 6D), 15%; mixed motile and nonmotile phenotypes (Fig. 6D, inset), 20%. The remaining 15% were unclassifiable because of fragmented SSEA-1 expression. We did not detect MVH protein in 3-day-old cultures. After 6 days of culture, we observed colonies of mixed shapes (49%; Fig. 6E); large EGC-like colonies (19%; Fig. 6F) of round or elongated shape containing cells showing strong, homogeneous expression of SSEA-1, and colonies with fragmented expression of SSEA-1 (32%). These results suggest that shortly after putting them into culture, PGCs are heterogeneous with respect to both cell morphology and expression of SSEA-1 and MVH proteins.
Then we examined expression of FGF-2 and FGFR-3 in cultured PGCs. At 8.5 dpc, PGCs were identified either with Oct-4 or SSEA-1 antibodies. Again, most analyzed PGCs were found as single cells exhibiting motile phenotype after 24 hours of FGF-2 stimulation. We detected no FGF-2 in these motile cells, but in the nonmotile cells, FGF-2 staining was spread over the entire cell membrane of PGCs with the most intense signal located in the cytoplasm (data not shown). When cultures were stained with antibodies to FGFR-3, PGCs cultured without exogenous FGF-2 expressed FGFR-3 only in the cytoplasm (Fig. 5D–5F, arrow). In contrast, when FGF-2 was present in the culture medium, we detected high expression of FGFR-3 in both nuclei and cytoplasm of a few PGCs (Fig. 5G–5I, arrow), although in most, the expression in nuclei was low (data not shown). FGFR-3 is weakly expressed on both feeder and somatic cells.
After 4–5 days of FGF-2 stimulation, we observed many small colonies comprising three to four cells, as well as some PGCs remaining as two-cell aggregates or single motile cells. FGFR-3 was detected in both cytoplasm and nuclei of PGCs in small colonies (Fig. 5J–5L). FGF-2 was present in the cytoplasm of PGCs in small colonies, but not in motile cells (data not shown). These results once again confirmed that cultured PGCs are heterogeneous. We detected no FGF-2 protein on the feeder cells or somatic cells, although FGF-2 was present in the cytoplasm of some somatic cells in 11.5- and 12.5-dpc genital ridges and in both cytoplasm and nuclei of proliferating Sl4-m220 cells. This suggests that FGF-2 was downregulated in nonproliferating Sl4-m220 cells and somatic cells under the culture conditions required for reprogramming of PGCs into EG cells.
In this study, we have investigated the reprogramming of PGCs into pluripotent stem cells induced by exogenous FGF-2. We found that FGF-2 is needed for no longer than 24 hours for the reprogramming of 8.5-dpc or 11.5-dpc PGCs to pluripotency in culture, but in its absence, no reprogramming takes place. Furthermore, the responsiveness of PGCs to FGF-2 is lost in culture over 24 hours. Recently, expression of several FGFRs as well as FGFs has been reported in mouse PGCs [24 –26]. Our immunofluorescence data on FGFR expression in 11.5-dpc PGCs are in general agreement with those of Schmahl et al. . When PGCs were placed in culture in the presence of exogenous FGF-2 and LIF, a few showed high expression of FGFR-3 in the nucleus as well as in the cytoplasm. FGF-2 was expressed only by cells that adopted a rounded, nonmotile phenotype: the subsequent large, multicellular colonies were made up entirely of these rounded, nonmotile cells. In contrast, PGCs cultured with LIF alone (a condition that does not allow the transition to pluripotency) did not form colonies, they showed little expression of FGF-2, and they expressed FGFR-3 only in the cytoplasm. We suggest, therefore, that the conversion of PGCs into EGCs does not involve all the cultured PGCs, but only those that upregulate FGF-2 and drive expression of FGFR-3, or at least one of its splice variants, in the nucleus (Fig. 5B).
When we aggregated cells from 13-day cultures of PGCs with host embryos, LacZ staining revealed that in two out of three chimeric blastocyst outgrowths, the cultured cells became incorporated but did not proliferate in the host ICM (Table 1; Fig. 1D). This again suggests that PGCs are heterogeneous. Furthermore, some of the EGC-like colonies formed early in the course of primary culture of PGCs were not homogeneous: we found mixed colonies, containing cells of both motile and non-motile phenotypes. We may also conclude that reprogramming is not a single-step process. We do not know whether each colony derives from a single cell or from more than one cell. Some cells within a colony may be fully reprogrammed into EG cells, but others may be in an intermediate state. Some of the cultured PGCs may be able to colonize the ICM but may fail to proliferate; others may do better, or less well. To identify an intermediate status of PGCs within EGC-like colonies, we stained multicellular colonies for TNAP activity (Fig. 1B) and anti-SSEA-1 antibodies (Fig. 6F). In each case, staining was homogeneous, suggesting that neither TNAP nor SSEA-1 can distinguish between an intermediate status of PGCs and reprogrammed EG cells. Kawase et al.  reported that PGCs under culture conditions differing from ours were converted into what appeared to be an intermediate state between PGCs and EG cells, but these cells did not survive subculture. Other cells again may be similar to PGCs, unchanged by culture, since we showed that freshly isolated PGCs aggregated with morulae did not make any chimeric blastocyst outgrowths (Table 1). Evidently, the culture environment affects cell fate, but not all the PGCs react to it in the same way (Fig. 6C, 6D).
We propose that reprogramming of PGCs in culture is a complex process that requires both dedifferentiation and acquisition of a pluripotent stem cell phenotype. The success rate of this process presumably involves a combination of transcriptional networks within the PGCs and external factors, including culture conditions. This is supported by our unpublished observations on reprogramming efficiency, which declines as PGC development progresses, until at 12.5 dpc, PGCs become very difficult to reprogram under standard culture conditions (; unpublished observations). The expression pattern of FGFR-3 in nuclei and cytoplasm of 12.5-dpc PGCs in vivo and in cultured PGCs (supplemental online Fig. 3G, 3H; Fig. 5H) is similar; however, the expression levels are higher in cultured PGCs. This implies that for reprogramming PGCs, not only is the translocation of FGFR-3 into the nucleus important, but also its increased expression level is important.
The reprogramming event can also be triggered by other factors such as forskolin, which increases the intracellular cyclic AMP (cAMP) levels, or all-trans retinoic acid, which interacts with nuclear retinoic acid receptors, suggesting that several signaling pathways in collaboration with LIF can induce the reprogramming of cultured PGCs.
A variety of growth factors/cytokines have been shown to affect PGC numbers in a dose-dependent manner in vitro . Resnick et al.  showed that FGF-2 stimulates dose-dependent proliferation of PGCs with a peak response at 1 ng/ml during 3-day culture. We also found a dose-response relationship of FGF-2 on the reprogramming process, increasing from 5 ng/ml to a plateau at 10 ng/ml (Fig. 3A). In an attempt to increase the efficiency of the reprogramming process, we added heparin to the cultures since heparin is known to increase the affinity and half-life of the FGF-FGFR complex. However, we observed no increase in numbers of EGC-like colonies compared with controls (Fig. 3A), suggesting that heparin did not increase the binding affinity for the binding of FGF-2 to its receptors in our cultures.
Functions of many FGFs may be redundant, so that other FGF family members can compensate for their loss. FGF-2-deficient mice have no defect in the germ cell lineage . Recently, expression of FGF-4, FGF-8, and FGF-17 in 11.5-dpc PGCs has been reported . One explanation for a possible lack of effect on PGCs in FGF-2-null mice is substitution by other members of the FGF family in vivo. Here, we have reported that FGF-5, FGF-9, or FGF-10 can substitute for FGF-2 in 8.5-dpc or 11.5-dpc PGC cultures in supporting the derivation of pluripotent EG cell lines in vitro; however, the results also suggested that FGF-9 and FGF-10 proved less effective than FGF-2 and FGF-5, owing perhaps to their differential affinity for FGFRs.
What prevents the formation of EGC-like cells in vivo? PGCs placed into culture lose their contacts with somatic cells of the embryo. They no longer have any direct contact with surrounding somatic cells, and at the same time they are bathed in ligand solutions, whereas PGCs in the embryo are in intact tissues and are both stimulated and repressed locally, to retain and progress their differentiated status. In particular, the level of FGF seems crucial if PGCs are to dedifferentiate into EGC-like colonies. In vivo, negative regulators of FGF signaling presumably regulate the level of FGF expression, but in vitro, this system will be disturbed. In vitro FGF-2 was not expressed either by feeder cells or by surrounding somatic cells, and we never observed EGC-like cells in medium containing LIF alone. To trigger the conversion, 5 ng/ml or more of exogenous FGF-2 was required. As a result, expression of the FGF-2 receptor FGFR-3 was upregulated in the nucleus, either by translocation from the cytoplasm, or by increasing the expression of an FGFR-3 splice variant predominantly found in the nucleus . This upregulation only occurred in a proportion of PGCs: we propose that it is these PGCs that are enable to be reprogrammed into EG cells. FGF-2 may be required for the nuclear localization of its receptor in EG cell precursors, just as FGF-9 is necessary (although not sufficient) for the nuclear localization of FGFR-2 in Sertoli cell precursors .
Growth factor tyrosine kinase receptors, including FGFRs, carry out their role in signal transduction at the cell surface; however, many of these transmembrane proteins also translocate to the nucleus after ligand stimulation [29, 30]. Recently, it has been demonstrated that nuclear translocation of FGFR-1 is involved in the regulation of cell proliferation . However, in many cases, little is known about the mechanism of nuclear import of the receptor. Furthermore, no definitive function for nuclear localization of receptors or their splice variants has yet been elucidated.
Table Table 1.. Frequency of incorporation into blastocyst outgrowths in vitro
The authors indicate no potential conflicts of interest.
We are grateful to the Wellcome Trust (M.A.S. and A.M.), the Leverhulme Foundation, and the Isaac Newton Trust (A.M.) for financial support. G.D-H. is a Biotechnology and Biological Sciences Research Council-funded MRC Stem Cell Fellow. I.R.A. is a Lister Fellow. Current address for I.R.A.: MRC Human Genetics Unit, Western General Hospital, Crewe Road, Edinburgh, U.K.