Special Reviews

Gametes from Embryonic Stem Cells: A Cup Half Empty or Half Full?

See allHide authors and affiliations

Science  20 Apr 2007:
Vol. 316, Issue 5823, pp. 409-410
DOI: 10.1126/science.1138772


When first reported 4 years ago, gametogenesis from embryonic stem (ES) cells promised an accessible in vitro model to facilitate molecular analysis of the germ lineage. Formation of primordial germ cells is robust, but terminal gametogenesis remains inefficient and doubts about gamete function persist. Although useful for research, clinical use of ES cell–derived gametes appears a distant prospect.

Both human and mouse embryonic stem (ES) cells have pluripotency, the remarkable capacity to form all tissues in the body. When mouse ES cells are transplanted into early embryos, ES cells can sustain the entire somatic and germline development of a mouse. Although this proves that ES cells form gametes as part of the complex choreography of embryonic development, deriving gametes solely in vitro seems a daunting challenge. Gametes mature in a niche with specialized supportive cells and a highly regulated hormonal milieu, and it seems implausible that such a structure should self-assemble in vitro. How could the most specialized of cells—the egg and sperm—arise in a petri dish? A critical review of the literature shows that ES cells support the earliest stages of germ lineage formation in cell culture but do not robustly generate functional haploid gametes, making clinical applications in infertility untenable at this time. Nevertheless, the immense value to research that would follow from an in vitro system for gamete production, and the potential for applications in animal breeding and somatic cell reprogramming, makes it fruitful to consider where the field has been and where it is going. Is the cup half empty or half full?

Interest in gametogenesis from ES cells was motivated by a need for an inexhaustible supply of oocytes. Oocytes serve as the target cell for nuclear reprogramming by somatic cell nuclear transfer (SCNT), a method that can be used to derive ES cells from animals with specific genetic features or from specific human patients. SCNT-ES cells could be used to study disease or potentially to create rejection-proof tissues for therapy; however, a major bottleneck for human applications is a source of oocytes. If ES cells could be directed to generate oocytes, a major barrier to SCNT would be eliminated. This requires, however, that ES cells can be prodded to form oocytes that are mature enough to support effective reprogramming of the transferred somatic nucleus.

ES cells differentiate spontaneously in vitro unless cultured in leukemia inhibitory factor (LIF), a cytokine that maintains ES cell self-renewal. After LIF removal, ES cells can differentiate on the surface of the petri dish or aggregate to form 3-dimensional (3D) structures called embryoid bodies (EBs). To varying degrees, such 2D and 3D approaches yield disorganized tissue masses that loosely recapitulate the tissue transformations of the gastrulating embryo and harbor elements of the three primitive embryonic germ layers—ectoderm, endoderm, and mesoderm. Current efforts in numerous labs are aimed at defining which morphogens, cytokines, chemicals, and culture conditions can bias or enhance the formation of a specific tissue of interest. Despite these efforts, directed differentiation remains more hope than reality. Hence, most laboratories pursue a combined direction-selection approach; the cells are first prodded to differentiate with as much direction as possible, then the desired cells are removed from a heterogeneous culture using antibodies against target-cell surface antigens, or by virtue of reporter genes expressed from tissue-specific promoters.

Three independent laboratories originally reported the derivation of gametes from ES cells in vitro using direction-selection strategies (Fig. 1) (13). All three exploited markers or properties that distinguish germ cells from undifferentiated ES cells, a challenge given that germ cells share many features with ES cells (4). In the first case, Scholer and colleagues engineered a truncated promoter for Oct4, a gene central to pluripotency, to express green fluorescent protein (GFP) in germ cells (5). Scholer's group differentiated ES cells harboring the reporter gene in 2D cultures and detected prevalent GFP+ cell clusters, suggestive of primordial germ cell formation. They collected aggregates of GFP+ cells, cultured them for a further several weeks, and observed a remarkable transformation: Complex multicellular structures developed that resembled ovarian follicles, and from these emerged oocyte-like cells surrounded by a zona pellucida–like coating. Extensive molecular, immunohistochemical, and ultrastructural data indicated that these cells were indeed oocytes. Even more surprising, cultures contained cystic structures that resembled blastocysts that the authors speculated arose from spontaneous parthenogenetic activation of the oocytes. Methods for oocyte formation from ES cells have not yet been widely exploited by the community, but Trounson's group has independently reported differentiation of ES cells into oocyte-like cells using testicular cell–conditioned medium, which presumably contains factors that stimulate both egg and sperm development (6). Although tantalizing, these results leave one wondering whether such ES-derived oocytes are functional. Can they be fertilized? Can they serve as recipients in nuclear transfer to support somatic cell reprogramming?

Fig. 1.

In vitro gametogenesis from ES cells. (Top) A 2D culture yields oocyte-like cells and parthenogenetic blastocysts (2). (Middle) A male germ cell niche within EBs supports formation of primitive sperm (1, 3, 13). Oocyte injection with sperm precursors has been reported to generate offspring (13). (Bottom) Primordial germ cells can be isolated from EBs and cultured with retinoic acid and cytokines to generate embryonic germ cells (1).

Two other laboratories reported deriving male gametes from ES cells differentiated into 3D EBs (1, 3). Noce and colleagues exploited a reporter for the germ cell gene vasa and stimulated germ cell formation with bone morphogenetic protein-4 (BMP4). They isolated vasa-GFP+ cells, mixed in tissue from embryonic gonads, and transplanted testes of germ cell–depleted mice. They found marker-positive cells in reconstituted seminiferous tubules, although no sperm function was documented (3). Follow-up studies from the same group found that media glucose concentration influences germ cell formation, suggesting that work remains to optimize differentiation (7). Although initially seeking oocyte development, our laboratory independently observed male gametogenesis from ES cells (1). We selected cells expressing the germ cell antigen SSEA-1 and treated them with retinoic acid, which induces germ cell proliferation. Then, using conditions established for isolating embryonic germ (EG) cells from embryos (8), we cultured EG-like cells that had erased their methylation marks at key imprinted gene loci, a diagnostic feature of primordial germ cells from the embryonic gonadal ridge. From EBs cultured for an additional 3 to 4 weeks, we used an antibody directed against primitive spermatocytes to isolate rare haploid cells that resembled round spermatids. When injected into oocytes, these cells generated diploid blastocysts, but never supported full mouse development.

These early reports all document robust formation of primordial germ cells, but limitations in achieving functional gametes. Using a germ cell–specific stella reporter, Surani's lab has independently confirmed formation of primordial germ cells from ES cells in vitro (9). In human ES cultures, germ cell–specific gene expression has been observed in an appropriate temporal sequence (10) and shown to be enhanced by incubation with bone morphogenetic proteins (11). Nonetheless, gametes have not yet been isolated from human cultures.

Primordial germ cells lack sexual dimorphism. Segregation of germ cell fates into oogonia or spermatagonia occurs within the developing tissues of the fetal gonad, and terminal gametogenesis requires tight hormonal control. The 2D ES culture condition employed by Scholer'sgroup appears to stimulate oogenesis, whereas the 3D EB condition used by Noce's group and our own appears to favor spermatogenesis, for unclear reasons. Germ cell maturation appears to follow oogenesis by default, unless suppressed by male-specific factors like Müllerian inhibiting substance (MIS). Indeed, we detected MIS expression in our EBs, perhaps signaling the formation of a supportive germ cell niche for male gametogenesis (1).

The gold standard for gamete function is the production of offspring. All of the early studies lacked this crucial achievement, but Engel and colleagues recently reported success. Using a GFP reporter for Stra8 (stimulated by retinoic acid-8), a gene expressed in premeiotic germ cells, the Engel group differentiated the pluripotent F9 teratocarcinoma line for 2 months and identified GFP+ cells, that when transplanted into germ cell–depleted testes, produced mature sperm. Although the sperm manifest structural abnormalities and reduced motility, they supported embryo formation after being injected into oocytes (12). Again, the lack of live offspring suggests that functional gametes were not formed. Even more remarkable, however, is their recent claim to have generated live mice from ES-derived gametes. In this case, the group sequentially selected Stra8-GFP+ cells, followed by cells expressing a dsRed reporter driven by the protamine promoter, thus identifying cells at a later stage in spermiogenesis (13). Of 65 embryos formed when the protamine-dsRed+ cells were injected into oocytes, 12 animals were born, and a few harbored the transgene by Southern blot analysis. Although these data are provocative, genome-wide polymorphism analysis would have been helpful to prove that these offspring carry a full haploid contribution from the ES-derived sperm.

Given the inefficiency of terminal gametogenesis from ES cells in vitro, and the paucity of data documenting reproductive function, employing ES cells as a source of gametes to treat infertility seems a distant prospect. For those who see the cup as half full, any gametogenesis is a marvel. Differentiation of ES cells into primordial germ cells appears robust and amenable to experimentation. ES cells thus provide an accessible tool to study genes that specify germ cells, the pathways that control germ cell migration, and the molecular machinery of imprint erasure that acts when germ cells arrive at the embryonic gonadal ridge. For those who view the cup as half empty, much work remains. Evidence of meiosis and terminal gametogenesis is lacking. It remains to be seen whether terminal spermiogenesis can occur in a petri dish and whether oogenesis can be optimized to provide a ready supply of oocytes for reprogramming studies. If additional laboratories replicate the remarkable achievement of living offspring from ES-derived gametes, then many new avenues for germ cell research are possible. Prospects for treating infertility become a more plausible prospect, and the cup will indeed seem more than half full.

References and Notes

Stay Connected to Science

Navigate This Article