編者案：今天出版的Nature還報(bào)道了用胚胎干細(xì)胞建立體外精子和卵子結(jié)合成配體的模型，這是利用胚胎干細(xì)胞研究?jī)尚陨车拈_(kāi)端,！值得關(guān)注（www.bioon.com)

Nature 427, 148 - 154 (08 January 2004); doi:10.1038/nature02247
Nature AOP, published online 10 December 2003

 
 
 
Derivation of embryonic germ cells and male gametes from embryonic stem cells

NIELS GEIJSEN1,2, MELISSA HOROSCHAK1,3, KITAI KIM1,3, JOOST GRIBNAU1, KEVIN EGGAN4 & GEORGE Q. DALEY1,3

Egg and sperm cells (gametes) of the mouse are derived from a founder population of primordial germ cells that are set aside early in embryogenesis. Primordial germ cells arise from the proximal epiblast, a region of the early mouse embryo that also contributes to the first blood lineages of the embryonic yolk sac1. Embryonic stem cells differentiate in vitro into cystic structures called embryoid bodies consisting of tissue lineages typical of the early mouse embryo2, 3. Because embryoid bodies sustain blood development, we reasoned that they might also support primordial germ cell formation. Here we isolate primordial germ cells from embryoid bodies, and derive continuously growing lines of embryonic germ cells. Embryonic germ cells show erasure of the methylation markers (imprints) of the Igf2r and H19 genes, a property characteristic of the germ lineage. We show that embryoid bodies support maturation of the primordial germ cells into haploid male gametes, which when injected into oocytes restore the somatic diploid chromosome complement and develop into blastocysts. Our ability to derive germ cells from embryonic stem cells provides an accessible in vitro model system for studies of germline epigenetic modification and mammalian gametogenesis.

We differentiated embryonic stem (ES) cells according to our standard methods4, and isolated messenger RNA from whole embryoid bodies (EBs) at several time points. We used polymerase chain reaction with reverse transcription (RT–PCR) to detect expression of genes implicated in ES cell pluripotency (the POU domain transcription factor Oct4) and germ cell development, including stella and fragilis (Fgls)5, and a set of genes that are exclusively expressed in the germ line and are absent from somatic tissues (Dazl, Piwil2, Rnf17, Rnh2, Tdrd1 and Tex14)6-9. All of these genes were expressed in undifferentiated ES cells, and a subset underwent rapid extinction with EB formation (Fig. 1). Rnh2, Tdrd1 and Tex14 decreased to undetectable levels very early in EB development (day 3–4), suggesting efficient differentiation and commitment to distinct cell fates. Stella and Fgls expression declined immediately on EB formation, but low levels persisted over the course of EB differentiation.

 Figure 1 Expression of germ-cell-specific genes during EB development.   Full legend
 
High resolution image and legend (38k)
 

Expression of the surface antigen SSEA1, a marker of pluripotent ES cells, also wanes on EB development, but rare SSEA1 positive (SSEA1+) cells persist in differentiated EBs10. We differentiated ES cells carrying an Oct4-promoter-driven green fluorescent protein (GFP) reporter gene11, and observed a similar overall decrease in GFP+ cell populations upon EB differentiation, as well as a persistent rare population of GFP+ cells (data not shown). Although these might represent residual undifferentiated ES cells, SSEA1 and Oct4 expression are also features of primordial germ cells (PGCs). We thus sought to determine whether these SSEA1+/Oct4+ cells within EBs represented residual undifferentiated ES cells or true PGCs.

There is a lack of markers that can suitably distinguish between ES cells and PGCs; however, retinoic acid acts to rapidly differentiate ES cells while stimulating proliferation of PGCs, and can therefore be used to distinguish between these two cell populations12. We plated ES cells or cells derived from EBs onto a mouse embryonic feeder (MEF) cell layer and cultured the cells for 7 days in the presence of 2 µM retinoic acid (Fig. 2a). We then quantified the residual SSEA1+ cells. In retinoic acid-treated ES cells, greater than 99% of cells extinguished SSEA1 expression. In contrast, in retinoic acid-treated cultures of EB-derived cells, a significant percentage of SSEA1+ cells persisted and expanded modestly in culture, with an apparent wave of formation peaking at around day 5 (Fig. 2b). Cells that differentiated for 5 days in retinoic acid were fixed and stained for alkaline phosphatase. Very few cells positive for alkaline phosphatase persisted in the retinoic acid-treated ES cell cultures (Fig. 2c, left panel). In contrast, large colonies of cells positive for alkaline phosphatase surrounded by motile cells that resembled migratory PGCs were abundant in retinoic acid-treated cultures of EB-derived cells (Fig. 2c, right panel). This differential effect of retinoic acid strongly suggested that the SSEA1+ population of cells from EBs were PGCs.

 Figure 2 Development of primordial germ cells in the differentiating EB.   Full legend
 
High resolution image and legend (86k)
 

To obtain conclusive evidence that the retinoic acid-resistant ES-like colonies were indeed PGCs, we analysed whether these cells manifested erasure of epigenetic imprints; this is a unique property of PGCs that is maintained in PGC-derived embryonic germ cells13. Imprinting of the Igf2r gene is determined by parental origin, with expression only from the maternal allele14. A specific region of the Igf2r gene has been identified, differentially methylated region 2 (DMR2), which is hypermethylated only on the maternally inherited allele15. We isolated SSEA1+ cells from EBs at different time points and cultured the cells for 7 days in the presence of retinoic acid. Individual retinoic acid-resistant colonies were isolated and expanded on gelatinized tissue culture plastic in the presence of leukaemia inhibitory factor (LIF), stem cell factor (SCF) and basic fibroblast growth factor; these are conditions that support the derivation of embryonic germ cells16, 17. We analysed the methylation status of DMR2 in independent clones of the parental ES cell line and in candidate embryonic germ clones by restriction digestion of the genomic DNA with PvuII and the methylation-sensitive enzyme MluI. Independent ES cell clones demonstrated a somatic methylation profile in which only one allele was digested (Fig. 2d, top panel, lanes 1–5). Most of the day 4 EB-derived embryonic germ cell clones displayed a similar somatic imprinting profile (Fig. 2d, top panel, lanes 6–9), with the exception of one clone in which methylation was erased (Fig. 2d, top panel, lane 10), as demonstrated by digestion of both alleles. Notably, at day 7 of EB development, 6 of 7 embryonic germ-like cells showed an unmethylated pattern (Fig. 2d, top panel, lanes 11–16), and at day 10 of EB development all embryonic germ clones had lost imprinting of the Igf2r gene. Similar results were obtained showing erasure of the methylation marker at the H19/Igf2 locus (Fig. 2d, bottom panel). These data demonstrate that the EB-derived PGCs display phenotypic and biological properties of PGCs developing in vivo.

We then used immunomagnetic bead sorting to isolate SSEA1+ candidate PGCs from differentiated EBs, and used RT–PCR to analyse expression of germ-cell-specific markers. Oct4 was expressed in the SSEA1+ population throughout EB development (Fig. 3a, left panel). Oct4 expression was almost undetectable in the SSEA1-negative (SSEA1-) fraction, demonstrating the effectiveness of the SSEA1 selection (Fig. 3a, right panel). Tex14 and Rnh2, which demonstrated a rapid downregulation on EB differentiation in the whole EB population, became undetectable in the purified SSEA1+ fraction of early EBs (days 3, 4; Fig. 3b). At day 5 expression levels of these genes rose, and by day 6 they reached a level comparable to ES cells. Although less marked than Tex14 and Rnh2, expression of Piwil2 and Dazl follows a similar pattern of increased expression over time in SSEA1+ differentiated EB-derived cell populations. This suggests that the developing EB supports a cellular environment similar to the early embryonic microenvironment in which PGCs are found.

 Figure 3 RT–PCR detection of germ-cell-specific genes in SSEA1+ cells isolated from developing EBs.   Full legend
 
High resolution image and legend (36k)
 

To determine whether the PGCs arise in a defined region of the EB, we used immunohistochemistry to simultaneously visualize CD41+ haematopoietic cells18 and SSEA1+ germ cells in cryosections of 7-day-old EBs (Fig. 3c). Similar to the developing embryo, the SSEA1+ PGCs in the developing EB exist in close juxtaposition to the cells of the developing haematopoietic system. The co-localization of nascent blood and germ cell populations in the peripheral zones of the developing EB raises interesting questions about their clonal origins and the microenvironment that specifies their distinct cell fates.

We next investigated whether the EB-derived PGCs undergo further differentiation into functional gametes. We analysed the expression of Sry, a male gene that determines germ cell fate. Sry expression was first detected in day 5 EBs, heralding initiation of a male germ cell developmental programme (Fig. 4a). Germ cell nuclear factor (Gcnf) has a role in confining Oct4 expression to the germ line19. We observed transient expression of Gcnf starting at day 7 and peaking at around day 11, suggesting a temporal window during which the germ cells become fully specified. At about day 11 of EB differentiation we found a strong upregulation of acrosin and haprin, genes tightly associated with male germ cell development (Fig. 4b). Acrosin is part of the acrosomal complex transcribed as proacrosin in the diploid germ cell population20, whereas haprin is a member of the RING finger–B box–coiled-coil family of transcription factors, which have a role in spermatogenesis and the formation of germ cell tumours21-23. Our observation of upregulation of genes associated with male germ cell maturation prompted us to investigate the presence of stromal supporter cells. Indeed, we detected the message for the luteinizing hormone/gonadotropin receptor (LH-R) as well as müllerian inhibiting substance (MIS); these are markers of Leydig and Sertoli cells, respectively. We failed to detect expression of zona pellucida proteins Zp1 and Zp2, whereas Zp3 is expressed in ES cells and early EBs (Fig. 4b), suggesting that within the context of EB differentiation, the default programme of female gametogenesis is suppressed. Indeed when comparing the expression of two genes specific for male germ cell differentiation, AZ1 and ret finger protein (Rfp), we observed exclusive expression of these genes in EBs derived from male (XY) but not female (XX) ES cells (Fig. 4c).

 Figure 4 EBs support differentiation of haploid male germ cells that support fertilization of oocytes.   Full legend
 
High resolution image and legend (108k)
 

To investigate whether male germ cells undergo meiosis in the context of the EB, we immunostained cell populations with an antibody that specifically recognizes male meiotic germ cells (FE-J124), and analysed these cells for DNA content using the fluorescent DNA-binding dye Hoechst 33342. In the testes of adult mice, both FE-J1+ and haploid (1C) cells are readily identified (Fig. 4d, left column of top and middle panels). Analysing the FE-J1+ cell population for DNA content shows that this marker of male germ cells recognizes a minor diploid (2C) population of primary spermatocytes and a predominant population of cells with a haploid (1C) DNA complement, as reported24 (Fig. 4d, bottom panel, left). In the EB cell suspension, haploid (1C) cells are discernable but obscured by a range of apoptotic cells with sub-2C DNA content (Fig. 4d, middle panel, right). However, when only the cells positive for the FE-J1 antibody are analysed, a distinct haploid (1C) population is observed, representing the predominant class of FE-J1-staining cells (Fig. 4d, bottom panel, right). The low proportion of antibody-positive cells (0.01%) and the relatively high ratio between diploid (2C) primary spermatocytes and haploid (1C) cells suggests that meiosis is highly inefficient in EBs. This experiment, however, demonstrates that the EB microenvironment is permissive for male germ cell development and meiotic maturation.

We further analysed the FE-J1+ cells by immunofluorescence microscopy. The FE-J1 antibody recognizes the anterior acrosome on early and late pachytene spermatocytes and on round spermatids24. As can be seen in Fig. 4e, FE-J1+ cells isolated from testes (top panel) or from day 20 EBs (bottom panel) demonstrate polarized apical perinuclear staining, indicating that the EB-derived cells have a similar morphology to testis-derived haploid cells, and possibly represent round spermatids.

Finally, we investigated the biological function of the EB-derived haploid cells and assayed their capacity to fertilize oocytes. We isolated FE-J1+/GFP+ haploid cells from day 20 EBs by flow cytometry, and performed intracytoplasmic injection into recipient oocytes. Five separate microinjection experiments performed in two independent laboratories produced comparable results. Approximately 50% of the injected oocytes supported cleavage to the 2-cell stage (n = 125), with 20% displaying progression to blastocysts. Figure 4f shows four representative blastocysts expressing the GFP transgene, indicating successful complementation of the oocyte genome by the EB-derived haploid cells. Furthermore, fluorescence in situ hybridization (FISH) using probes directed against an autosomal locus and markers on the X and Y chromosome demonstrate a normal diploid chromosome complement and expected ratios of male (XY) and female (XX) embryos (Fig. 4g). Efforts are underway to determine whether embryos arising from fertilization with EB-derived male gametes will develop normally after uterine transfer.

Germ cell development remains a largely unexplored but fascinating process of cell fate specification. Germ cells represent a privileged class of cells, given responsibility for perpetuating pluripotency and ensuring propagation of the gene pool. The genetic mechanisms that account for maintenance of pluripotency and restrict somatic differentiation are beginning to yield to molecular analyses5, but an in vitro model system that recapitulates germ cell specification will greatly facilitate these studies.

Recently, Schöler and colleagues reported the generation of oocytes from mouse ES cells in culture25. The reported differentiation happened spontaneously over a period of nearly 50 days. In contrast to our method of ES cell differentiation into EBs, Schöler and colleagues used bulk two-dimensional differentiation on tissue culture plastic, in which both male and female lines of ES cells yielded oocytes. Given our observation of male germ cell development, we speculate that EBs may preserve more of the tissue organization reflective of the embryonic gonadal ridge, thereby enabling male germ lineage specification. In support of this notion, we detected expression of the müllerian inhibiting substance in EBs by RT–PCR. While this manuscript was under review, Noce and colleagues reported the successful derivation of male lineage germ cells from ES cells in vitro26. Our work corroborates and extends this report by demonstrating the erasure of methylation markers at imprinted loci and the successful fertilization of oocytes by the ES-derived male gametes.

Although EBs may not reflect the precise temporal and spatial features of embryonic development, their ready derivation from ES cells in vitro has proven valuable for genetic studies of tissue differentiation, by linking gene deletions or ectopic transgene expression to specific cellular phenotypes. We have demonstrated the in vitro differentiation of ES cells into primordial germ cells, which proliferate in response to retinoic acid and give rise to embryonic germ-cell-like clones that undergo erasure of imprints. We are currently using this in vitro system to investigate whether EB differentiation sustains a male germ cell niche that enables the proper restoration of male imprints, thereby affording an in vitro system to address the genetic and biochemical mechanisms of this fundamental epigenetic modification. As demonstrated by molecular analysis, the EBs support a programme of male germ cell differentiation, culminating in the formation of haploid cells that manifest the morphology and fertilization potential of male haploid germ cells of the round spermatid stage. Our report, together with the recent demonstration of oocyte and sperm generation from ES cells25, 26, signals a new realm of possibilities for investigating germ cell development, epigenetic reprogramming and germline gene modification.

Methods
Mouse strains and ES cells C57BL/6-TGN(ACTbEGFP) mice were from Jackson Laboratories27. 129SvEv mice were from Taconic. ES cells were derived from an F1 cross between C57BL/6-TGN(ACTbEGFP)1Osb and 129SvEv.

Cell culture ES cells were maintained on irradiated MEFs in DME/15% IFS, 0.1 mM non-essential amino acids (GIBCO), 2 mM glutamine, penicillin/streptomycin (GIBCO), 0.1 mM -mercaptoethanol, and 1,000 U ml-1 LIF (Invitrogen). For EB differentiation, ES cells were digested with trypsin, collected in EB medium (IMDM/15% IFS, 200 µg ml-1 iron-saturated transferrin (Sigma), 4.5 mM monothiolglycerol (Sigma), 50 µg ml-1 ascorbic acid (Sigma) and 2 mM glutamine) and plated for 45 min to allow MEFs to adhere. Non-adherent cells were collected and plated in hanging drops at 200 cells per 30 µl droplet in an inverted bacterial Petri dish. EBs were collected from the hanging drops at day 3 and transferred into 10 ml EB medium in slowly rotating 10 cm Petri dishes. At day 4, EBs were fed by exchanging half of their spent medium. Cells were collected by collagenase treatment and re-suspension in cell dissociation buffer (Invitrogen).

RT–PCR RT–PCR amplifications were titrated to be within a linear range of amplification. Primers used are: Oct4(f) 5'-GTGGATTCTCGAACCTGGCT-3', Oct4(r) 5'-GTCTCCAGACTCCACCTCAC-3'; stella(f) 5'-CAGCCGTACCTGTGGAGAACAAGAG-3', stella(r) 5'-AGCCCTGGGCCTCACAGCTT-3'; Fgls(f) 5'-TTGCTCCGCACCATGAACCA-3', Fgls(r) 5'-TGAAGCACTTCAGGACCGGA-3'; Dazl(f) 5'-GCCAGCACTCAGTCTTCATC-3', Dazl(r) 5'-GTTGGAGGCTGCATGTAAGT-3'; Piwil2(f) 5'-CCGTCATGAAGGAGAGCTCG-3', Piwil2(r) 5'-GGAACGACTCTGTGCTGGAT-3'; Rnf17(f) 5'-GACACACAGTCTAACAGAGG-3', Rnf17(r) 5'-AGGACAGCAGCATCTACCTT-3'; Rnh2(f) 5'-CATAAGTGGCAACGAAGAGC-3', Rnh2(r) 5'-GTTACAGGCTGCTACCATCA-3'; Tdrd1(f) 5'-GCAGTTCTGCTCTGTCAAGG-3', Tdrd1(r) 5'-CAGAGCGTGGAATCACATGG-3'; Tex14(f) 5'-GAAGCTTGAGCAGGAGGTAG-3', Tex14(r) 5'-TTCAGAAGACACAGACGCCA-3'; Gcnf(f) 5'-GTGGAAGACCAGGACGACGA-3', Gcnf(r) 5'-CCTACTGGATGATAGTGTGG-3'; acrosin(f) 5'-CGGAGTCTACACAGCCACCT-3', acrosin(r) 5'-GCATGAGTGATGAGGAGGTT-3'; haprin(f) 5'-CCAGAACATGAGACAGAGAG-3', haprin(r) 5'-AGCAACTTCCTGAGCATACC-3'; Hprt(f) 5'-GCTGGTGAAAAGGACCTCT-3', Hprt(r) 5'-CACAGGACTAGAACACCTGC-3'; Sry(f) 5'-TTACAGCCTGCAGTTGCCTC-3', Sry(r) 5'-GGTCATAGAACTGCTGTTGC-3'; MIS(f) 5'-TTGGTGCTAACCGTGGACTT-3', MIS(r) 5'-GCAGAGCACGAACCAAGCGA-3'; LH-R(f) 5'-TGCAACCTCCTCAATCTGTC-3', LH-R(r) 5'-AGCGTGGCAACCAGTAGGCT-3'; Zp1(f) 5'-GAGTGACTGTGTTGCCATAG-3', Zp1(r) 5'-GCCACACTGGTCTCACTACG-3'; Zp2(f) 5'-GCTACACACATGACTCTCAC-3', Zp2(r) 5'-GGTGACTCACAGTGGCACTC-3'; Zp3(f) 5'-TTGAGCAGAAGCAGTCCAGC-3', Zp3(r) 5'-CGGTTGCCTTGTGGATGGTC-3'; actin(f) 5'-ACCAACTGGGACGATATGGAGAAGA-3', actin(r) 5'-CTCTTTGATGTCACGCACGATTTC-3'.

Immunomagnetic isolation of SSEA1+ cells Cells collected from EBs were incubated (30 min) with a monoclonal antibody against SSEA1 (Hybridoma bank) at 4 °C in PBS/0.5% BSA. Cells were washed twice with ice-cold PBS/0.5% BSA before addition of immunomagnetic rat anti-mouse IgM beads (Dynal), and were incubated for 1 h at 4 °C with slow rotation. Magnetic separation of SSEA1+ beads associated with the cells was performed according to the manufacturer's protocol.

Southern analysis Genomic DNA was prepared from individual clones of the parental ES cell line or EB-derived embryonic germ cells. For the embryonic germ cell clones, SSEA1+ cells were isolated at different days of EB development and cells were grown in the presence of retinoic acid for 7 days followed by 2 days of culture without retinoic acid. Individual clones were isolated and expanded on gelatinized tissue culture plastic in the presence of LIF to remove feeder cells. Genomic DNA was isolated and digested with PvuII and MluI for the detection of Igf2r methylation, or with SacI and HhaI for the analysis of H19 imprints. DNA was separated on a 0.7% agarose gel and Southern blots were generated by standard methods. Filters were hybridized with a probe covering region 2 of the Igf2r receptor (pPP4).

FACS analysis Testicular and EB-derived cell suspensions were obtained by enzymatic digest of the tissue. Briefly, cells were incubated at 37 °C for 15 min with digest buffer (0.1% collagenase IV, 0.2% hyaluronidase and 50 U ml-1 DNase (all Sigma)). Cells were then dissociated using cell dissociation buffer (Invitrogen), collected by centrifugation and a second digest was performed. Cell clumps were removed using a 70 µM strainer and cells were re-suspended in ice-cold RPMI plus 0.5% FBS. Cells were incubated with FE-J1, a haploid male germ-cell-specific antibody (Hybridoma bank24), for 30 min at 4 °C. The cells were washed twice with RPMI/0.5% FBS and incubated with phycoerythrin (PE)-conjugated rat anti-mouse IgM for 30 min at 4 °C. Cells were washed twice with RPMI/0.5% FBS, re-suspended in RPMI/0.5% FBS containing 2.5 µg ml-1 cytochalasin B (Sigma) and sorted on a Becton-Dickinson FACSCalibur.

Oocyte injections Eight–ten-week-old B6D2F1/J mice (Jackson Laboratories) were used as oocyte donors. EB-derived donor cells were re-suspended in KSOM (Speciality Medium) containing 10% (w/v) polyvinyl alcohol (Sigma) and 0.01% (w/v) bovine serum albumin (Sigma). Intracytoplasmic injection into cumulus-free oocytes was carried out in H-KSOM containing 5 µg ml-1 cytochalasin B (Sigma) and 3% (w/v) sucrose at room temperature using a Nikon Eclipse TE300 microscope equipped with Narashige hydraulic micromanipulators and Hoffman modulation contrast. The injected oocytes were washed five times in KSOM to remove the cytochalasin. Reconstructed embryos were activated either in KSOM containing 10 µM calcium ionophore A23187 (Sigma) for 5 min, followed by 2 mM 6-dimethylaminopurine (Sigma) in KSOM at 37 °C in 5% CO2 for 3 h, or for 5 h in Ca2+-free medium containing 10 mM SrCl2. Embryos were then washed five times in KSOM. Embryos were cultured in KSOM at 37 °C in 5% CO2.

DNA-FISH DNA-FISH was performed as described28 with minor modifications. The zona was removed from embryos using acid tyrodase, and the morulae were incubated in a small drop of 0.075 M KCl on a slide for 5 min. Embryos were fixed by replacing the KCl with 3:1 methanol:acetic acid. Cells were permeabilized in 0.5% triton/PBS for 10 min, washed twice in PBS and dehydrated in 70%, 90% and 100% ethanol. Bacterial artificial chromosome (BAC) clones used for FISH analysis were from BACPAC resources and from Invitrogen. Autosomal sequences were detected with BAC RP23-20P21, which is specific for the  light chain locus of immunoglobulin. X chromosomal sequences were detected with BAC CT7-228O4, which is specific for the Irak1 locus. Y chromosomal sequences were detected with BAC RP24-507D23, which is specific for Y chromosomal repeats.