Trophoblast Stem Cells

Abstract

Trophoblast stem cells (TSC) are the precursors of the differentiated cells of the placenta. In the mouse, TSC can be derived from outgrowths of either blastocyst polar trophectoderm (TE) or extraembryonic ectoderm (ExE), which originates from polar TE after implantation. The mouse TSC niche appears to be located within the ExE adjacent to the epiblast, on which it depends for essential growth factors, but whether this cellular architecture is the same in other species remains to be determined. Mouse TSC self-renewal can be sustained by culture on mitotically inactivated feeder cells, which provide one or more factors related to the NODAL pathway, and a medium supplemented with FGF4, heparin, and fetal bovine serum. Repression of the gene network that maintains pluripotency and emergence of the transcription factor pathways that specify a trophoblast (TR) fate enables TSC derivation in vitro and placental formation in vivo. Disrupting the pluripotent network of embryonic stem cells (ESC) causes them to default to a TR ground state. Pluripotent cells that have acquired sublethal chromosomal alterations may be sequestered into TR for similar reasons. The transition from ESC to TSC, which appears to be unidirectional, reveals important aspects of initial fate decisions in mice. TSC have yet to be derived from domestic species in which remarkable TR growth precedes embryogenesis. Recent derivation of TSC from blastocysts of the rhesus monkey suggests that isolation of the human equivalents may be possible and will reveal the extent to which mechanisms uncovered by using animal models are true in our own species.

In the mouse, trophoblast stem cells, the precursors of the differentiated cells of the placenta, can be derived from outgrowths of either blastocyte polar trophectoderm (TE) of the extraembryonic ectoderm (ExE), that originates from polar TE after implantation.

INTRODUCTION

Lineage-specific stem cells are generally recognized as subpopulations, usually small in number, that exist in a unique niche, thereby allowing them to self-renew while simultaneously creating more specialized progeny that ultimately advance to yield fully differentiated cells (Fig. 1). It is assumed that placental stem cells, here termed trophoblast stem cells (TSC), exist in all placental mammals, especially during the early stages of placental development when trophoblast (TR) growth is maximal. Nevertheless, with the exception of the mouse and the rhesus macaque, little is known about either their location or phenotype.

FIG. 1.

Diagram to illustrate the origin and fate of TSC in the mouse. Stem cell precursors in polar TE are hypothesized to give rise to a population of TSC. TSC are located adjacent to the developing epiblast, which arises from the ICM (see Fig. 2). The TSC self-renew but also generate faster-dividing, multipotent progenitors that provide the precursors of the differentiated cell types of the placenta. The information shown is speculative, and the differentiated tissue lineages are minimized for simplicity.

Segregation of trophectoderm (TE), the precursor of placental TR cells, from the inner cell mass (ICM) at the blastocyst stage is the first obvious fate decision during formation of the mammalian conceptus. Many of the subsequent steps in placental development vary widely across species. In hemochorial placentation, which occurs in humans and mice, there is extensive invasion of TR into the uterine wall. Consequently, the placenta is chimeric—comprised of intimately associated cells from both the conceptus and the mother. In other species, invasion of maternal tissue either fails to occur (e.g., swine and some prosimians) or is very limited (e.g., ruminants such as cattle and sheep). As a result, the placenta is either entirely or very largely of conceptus origin [1, 2]. These noninvasive types of placentation are probably a derived state, having evolved from ancestors whose placentas had a hemochorial interface [3]. Hence, in the case of the placenta, simpler does not equate with more primitive.

Despite the fact that placentas from different mammalian groups exhibit a remarkable range of gross morphologies, histological organization, and invasiveness, they perform similar functions. These include exchange (of nutrients, waste products, and dissolved gases), anchoring/mechanical support, immune protection, and the production of hormones that adjust maternal physiology to the needs of the growing offspring [2]. The placentas of eutherian mammals are comprised of several specialized cell types that are derived from lineage-committed precursors whose coordinated actions fulfill these diverse functions. The formation of such terminally differentiated cells probably begins within a group of self-renewing, multipotent TSC. In this review, we make inferences about the mechanisms that give rise to the TR lineage during the early embryonic period, the niche that supports their continued expansion, and the properties of TSC that can be propagated in the laboratory. In common with other aspects of mammalian embryology, much of the available information is derived from the mouse, but, wherever possible, we have attempted to broaden the scope of the discussion to include other species.

ORIGIN OF TR

In the mouse, TE materializes as a distinct epithelial layer of the blastocyst at around the 32-cell stage of development, approximately 84 h postconception [4, 5]. After the blastocyst expands and hatches from the zona pellucida around Day 4.5 postcoitus, the abembryonic mural TE subpopulation attaches the conceptus to the uterine wall and induces a primary decidual response. These cells undergo several rounds of DNA endoreduplication and give rise to the primary giant cell population of the yolk sac placenta that invade further into the uterine wall and play a transport function until midway through pregnancy [6]. Mitotic activity of polar TE in contact with the ICM is retained, however, [7] and these cells are the precursor of all the TR lineages that make up the mature chorioallantoic placenta [8]. In humans, a similar developmental program occurs on a somewhat protracted time scale [9], except polar TE is believed to provide the initial implantation contact. However, as we shall see, attempts to derive TSC from human TE have failed, leaving open questions about the developmental potential of the initial TR population. Relative to the mouse and human, formation of the blastocyst is initiated much earlier in some species, such as at the four-cell stage in the elephant shrew [10], and later in others, such as after the 64-cell stage in cattle [11]. Complicating matters even further, the multicellular blastocysts of the tenrecs (insectivores found on Madagascar and in parts of Africa) are made up of a single epithelial layer that surrounds a fluid-filled blastocoel cavity [10]. A rather similar pattern of development is found in marsupials. Presumably some or all the epithelial cells are totipotent in tenrecs and marsupials, that is, capable of generating both TR and embryonic lineages.

In the mouse, commitment to the TR lineage is irreversible by the blastocyst stage, but the preceding transition is gradual. Cells expressing an early TR marker, the transcription factor CDX2, can usually be detected in several blastomeres by the eight-cell stage of development [12, 13] and sometimes earlier. Nevertheless, the CDX2-positive outer cells of 16-cell stage morulae can contribute to the ICM if they are moved to an internal position in embryo reconstitution experiments. Similarly, inner cells repositioned to the outer surface can form TE. Finally, the ICM can regenerate TE [14]. These data suggest a delicate balance between commitment to the pluripotent ICM and to the lineage-restricted TR precursors, highlighting the regulative nature of the initial stages of fate specification [15]. The extent to which these findings apply in other eutherian species is largely unknown, but differences may exist, as suggested by the fact that events often proceed quite differently after the blastocyst stage, depending on the taxon [10].

THE TSC NICHE

Mouse

Although polar TE cells are mitotically active, a feature dependent on the close proximity of the ICM [7], they are not a self-sustaining population, existing instead for only a short time during the peri-implantation period. In contrast, a great deal of evidence suggests that the polar TE is the immediate precursor of self-renewing TSC from which the mature TR lineages arise [16].

Paracrine signals are likely to be crucial for the survival of both polar TE and its progeny populations. For example, Pou5f1 (Oct4)-deficient embryos, which lack an ICM, do not support TR proliferation [17], probably because ICM-derived EGF and FGF4 signals are important [16–18]. As development proceeds, other factors, notably NODAL, derived from epiblast also act locally [19]. TR cells appear to have evolved mechanisms for modulating these signals. For example, the SMAD4 inhibitor ectodermin (TRIM), an ubiquitin ligase, precisely doses NODAL activity, balancing TR self-renewal and differentiation [20]. Conversely, TR signals support ICM development. For example, the docking protein and fibroblast growth factor receptor substrate FRS2 mediates activation of the extracellular signal-regulated protein kinase (ERK) pathway to enhance FGF-4 induced CDX2 expression in TSC. In turn, CDX2 binds to an FGF4-responsive enhancer element in the promoter region of BMP4, leading to production and secretion of this growth factor, which can rescue the defective growth of Frs2^−/− ICM [21]. Thus, it appears that the growth factor milieu of the polar TE drives these cells to a stem cell fate, and these cells, in turn, give rise to multipotent progenitor cells that form the ectoplacental cone and secondary giant TR cells and the ExE [16, 22, 23] (Fig. 2). Given the dynamic nature of embryonic and ExE development, it is likely that this specialized niche is transient and maintained only for 3–4 days during postimplantation development [22]. Accordingly, in vitro-derived mouse TSC are obtained from either proliferating outgrowths of polar TE or ExE explanted from a region bordering the epiblast [16]. These founder cells are then generally cultured on inactivated mouse embryonic fibroblasts to provide NODAL in a medium supplemented with fetal bovine serum, FGF4, and heparin. These conditions are sufficient to maintain self-renewal, an undifferentiated phenotype, and TR multipotency over multiple passages.

FIG. 2.

Diagram to illustrate the dependence of the polar TE in the Day 3.5 blastocyst (A) and TSC in a conceptus at Day 6.5 (B) on factors produced by the ICM and epiblast, respectively, of the mouse conceptus. For simplicity, the polar TE, because of its transient nature, is regarded as the precursor of the “true” TSC population in its initial niche adjacent to the epiblast. FGF4 and NODAL are assumed to be produced by the ICM and epiblast. The insets in A and B show the regions illustrated in the main figures. The diagrams are based on those of Tanaka et al. [16] with permission from AAAS.

As TR cells move away from the primary niche, they continue to divide but also begin to differentiate into the specialized lineages that form the mature placenta and carry out its numerous functions (Fig. 1). For example, ExE provides the various specialized TR cells of the labyrinth layer and chorion, while the ectoplacental cone forms the spongioTR layer between the outer, invading giant cells and the labyrinth. TSC populations have also been inferred to exist at these secondary sites, especially in the chorionic ectoderm, where their niche may be dependent on factors provided from the ectoplacental cavity rather than from the epiblast [22]. In this context, it is interesting to note that TR giant cells appear to have multiple origins in different functional zones of the mouse placenta [24].

Primates

Cells with some properties of TSC have been derived from outgrowths of rhesus macaque blastocyst TE [25, 26]. They can be maintained for multiple passages in the absence of feeders or growth factors. The rhesus TSC express an interesting set of markers, only a portion of which are shared with their mouse counterparts. These include chorionic gonadotropin (CG; a unique primate TR marker), CD9, KRT7, POU5F1, and EOMES. Curiously, CDX2 expression was not detected. The expression of CG and lack of expression of CDX2, however, tend to imply that these cells may not be TSC but committed progenitor cells, already partially differentiated. On the other hand, when they differentiate further, these putative TSC either form syncytiotrophoblasts or exhibit the invasive behavior of extravillous TR, suggesting that they can adopt multiple fates. These contradictions will need to be resolved before the rhesus cell lines can be justifiably assigned stem cell status.

The derivation of either human TSC or self-renewing progenitors has yet to be reported. Therefore, we can only speculate about the environment in which they reside. Likewise, very little is known about the initial TR populations, their developmental potential, and their rates of differentiation. For example, one of the largest studies published to date suggests that implantation, defined as the appearance of chorionic gonadotropin in maternal urine, usually occurs between Days 8 and 10, but about one-third of those conceptuses that do implant are subsequently lost [27, 28]. Thus, it is very difficult to know whether the limited number of microscopic images of human embryos that have developed in vivo, archived primarily in the Carnegie collection, would have resulted in live births. Nevertheless, a few conclusions are widely accepted. For example, unlike the mouse, it is the polar rather than the mural TE that mediates initial implantation. The mechanisms include the same ligand-receptor interactions that mediate leukocyte rolling and tethering [29]. Specimens in the Carnegie collection suggest that the invasive population is syncytial [30]. On the other hand, most images of blastocysts matured in vitro suggest that the TE is comprised primarily of mononuclear TR [31, 32]. After initial implantation of rhesus monkey conceptuses, however, a polarized unilaminar syncytium does emerge and spreads throughout the trophoblastic plate, forming characteristic intrasyncytial clefts. The rapid development of these cells is posited to create a chamber for maternal blood, which can slowly percolate through the area without coagulating once TR cells breach maternal vessels [33]. Analyses of samples in the Carnegie collection from analogous stages suggest that a similar transition may occur in humans.

At a molecular level, it appears that the initial low oxygen environment, which is maintained for much of the first trimester, favors rapid TR proliferation and the likely maintenance of TSC rather than differentiation and is one reason that placental development outstrips that of the embryo during the first few weeks of human pregnancy [34]. Specifically, TR fate is governed by the transcription factor network, such as hypoxia-inducible factors (HIFs), and protein degradation machinery, such as the Van Hippel Lindau protein/prolyl hydroxylase system, which mediates cellular responses to hypoxia [35]. It is interesting to note that the HIFs play both oxygen-dependent and -independent roles in governing TR differentiation through mechanisms that include epigenetic modifications. Similarly, factors that control blood coagulation also appear to play dual roles, regulating this cascade and, by independent mechanisms, TR growth [36].

Currently, it is unknown whether any of these initial TR populations emerging from the blastocyst correspond to TSC. The fact that efforts to derive lines from blastocyst TE have failed to date could be evidence that this population does not have the developmental potential to give rise to the TR progenitors that form the mature placenta. Instead, these cells may emerge later as the chorionic villi begin to form, a conclusion that is supported by the recent description of clusters of cells expressing TSC markers in regions where cytotrophoblasts initiate invasion into maternal decidua [37]. In either case, specimens in the Carnegie collection provide no evidence that the TR is the source of all ExE tissues, especially extraembryonic mesoderm, which appears to arise from cells that emerge from the primitive streak and is likely to be an important component of the TSC niche [38]. The signals the niche provides are also likely to include human-specific cues. For example, BMP4 induces monkey ESC to assume an endoderm fate [39], whereas this factor triggers hESC to assume TR characteristics [40].

Ungulate Species

A striking feature observed in domestic species such as swine [41], sheep [42], and cattle [43] is that the TR elongates dramatically well before attachment, providing the conceptuses with an enormous area of exposed TE (the chorionic epithelium) and the consequent ability to exploit uterine secretions throughout the uterine lumen for sustained growth and development (Fig. 3A). As pointed out by Rielland et al. [44], polar TE overlying the embryonic disk is soon lost in the domestic species and does not contribute to the placenta, while the TR component associated with elongated outgrowth is the homolog of mouse mural TE (Fig. 3). Consequently, TR and epiblast are in contact only around the embryonic disk. It remains unclear whether the expansion of mural TE is reliant on a small resident stem cell population immediately surrounding the embryonic disc or on stem cells scattered throughout the TE or whether sustained growth is entirely due to proliferation of progenitor cells that are already partially differentiated and possibly committed to particular TR cell types. It seems unlikely that this dramatic period of TR growth is sustained by growth factors produced by the still small embryonic disk. Instead, TR proliferation may be orchestrated by the maternal system and specifically by uterine secretions, which contain a range of growth factors [45].

FIG. 3.

Development of TR of ungulate species from mural trophectoderm. A) Pig conceptus at about Day 13 of pregnancy, at which stage each can achieve a length of 1 m or greater. The position of the relatively undeveloped embryonic disc is arrowed. B) The formation of the embryonic plate in the goat conceptus, illustrating the loss of polar TE during the early stages of conceptus elongation and the growth of mural TE and underlying extraembryonic endoderm. The expanded blastocyst (B1) is at around Day 10 of pregnancy, but within a few days polar trophectoderm disappears (B2), and the embryonic plate is formed (B3). The location of the TSC “niche,” if it exists in such conceptuses, is unknown. The diagrams are redrawn (with permission) from John R. Fuller's illustrations in Austin [149].

Perhaps as a consequence of these complications, authenticated TSC have not been described from any ungulate species, although primary TR cell lines have been derived from sheep and goat [46], pigs [47–50], and cattle [51–53]. Many of these cell lines grow continuously in culture and display signature genes characteristic of TR, but multipotency and other aspects of a stem cell phenotype have not been demonstrated. Conceivably, the lines represent a stage of differentiation beyond the TSC.

TRANSCRIPTIONAL NETWORKS UNDERPINNING TSC STEMNESS

We have considered several criteria to define genes implicated in establishing TSC self-renewal and multipotency. One is whether the gene is linked to early developmental failure of the placenta in the mouse, the best genetic model available. A second is whether TSC differentiate when the expression of a particular gene is silenced. A third is whether forced expression of a gene in ESC diverts these cells toward the TR lineage. One other criterion is whether the expression and epigenetic status of a particular gene is consistent with its being part of the transcriptional networks responsible for TSC stemness, either when ESC are driven toward the TR lineage or when TSC are forced to differentiate by external cues. None of these measures can be considered to provide conclusive proof for stemness status, and not all genes will conform to all four standards. For example, as we shall see, certain genes, likely to be required for TSC maintenance, increase rather than lose expression as TSC differentiate. In addition, the list that follows is not exhaustive and has been chosen subjectively on the basis of what the authors consider to be the most compelling information available. A recent publication [54] describes, in detail, the transcriptional networks implicated in mouse TR differentiation and provides a useful corollary to the sections that follow.

Cdx2

Genetic manipulation of the mouse has indicated that many genes are needed to form a fully functional placenta [54–61]. Only a subset has been linked to developmental failure either before or at implantation, suggesting that such genes are required to control TR emergence and not simply its functional differentiation. The caudal-related transcription factor Cdx2 is one such gene. It is the best-known marker for distinguishing TE from ICM cells in the mouse and a variety of other species [62–65]. Cdx2^−/− mouse conceptuses fail to implant [66], although they do form a rudimentary blastocoel cavity [67, 68]. Moreover, ectopic expression of CDX2 in mouse ESC down-regulates POU5F1 [69] and suppresses the expression of genes that are components of pluripotency networks, thereby causing the cells to adopt a TR phenotype. CDX2 expression is also lost as TSC differentiate. Interestingly, depletion of both maternal and zygotic CDX2 immediately after fertilization leads to developmental arrest at much earlier stages than expected from elimination of only zygotic CDX2. This developmental arrest is associated with defects in cell polarization and, consequently, compaction at the 8- and 16-cell stages. Cells deprived of CDX2 have cell cycle aberrations and reduced expression of the TE-specific genes Gata3 and Eomes [70]. Thus, while CDX2 protein may not be necessary for initial TE specification, it is required for the TR lineage to emerge.

Tead4

Mouse embryos mutant for Tead4 fail to form even a rudimentary blastocyst [71, 72] and cannot maintain CDX2 expression beyond the early morula stage. Instead, the entire conceptus consists of ICM-like cells expressing POU5F1 and NANOG. Importantly, TSC cannot be isolated from either Cdx2^−/− or Tead4^−/− blastocysts, while ES cells can. Moreover, forcing Tead4 transcription in mouse ESC induces CDX2 expression and a TR phenotype [73]. Thus, Tead4 is probably the earliest-known gene that is required for TR specification.

Eomes

The product of the T-box gene Eomes is the earliest-acting transcription factor known to be required for immediate postimplantation lineage commitment steps, as mice lacking the Eomes gene expression fail to exhibit a proper TE-to-TR transition. While they do implant, they arrest at a blastocyst-like stage of development [74], and such conceptuses are unable to form TSC. In other words, Eomes expression may not be necessary for TE specification but is required for subsequent proliferation of polar TE and, potentially, TSC emergence. Consistent with this hypothesis, Eomes is expressed in mouse TSC and is down-regulated on differentiation [75]. However, EOMES is also expressed in the ICM [76], suggesting that it may have a broad role in stem cell self-renewal.

Gata3

Like Cdx2, the Gata3 gene is transcribed in TE but not in the ICM of mouse blastocysts [77] and can force emergence of TR when expressed ectopically in mouse ESC [78]. Nevertheless, GATA3-mediated TR fate is not dependent on CDX2 and vice versa. Although both genes are regulated by TEAD4, they appear to operate semi-independently, specifying TR fate via different pathways and targets [78]. Attempts to prove a specific role for GATA3 in directing TE emergence in embryos by gene knockout have not been successful, possibly because of overlap in transcriptional activity with GATA2, with which it is coexpressed [79, 80]. Finally, GATA3 binds to the promoter sequences of many of the same genes targeted by EOMES—TCFAP2C, SMARCA4, and ETS2—transcriptional factors that have been postulated to be part of a regulatory network maintaining TSC stemness [75].

Tcfap2c

The protein product of this gene (formerly known as AP-2 gamma) localizes to TE in mouse blastocysts. Like GATA3, overexpression in mouse ESC induces a TR cell fate independently of CDX2 [81]. Tcfap2c^−/− embryos implant [82, 83] but die between Days 7 and 9 of pregnancy. The primary reason has been linked to failure of labyrinth precursors to proliferate, suggesting that the TCFAP2C defect is manifested at the level of TSC. In TSC, TCAP2C in a partnership with EOMES (see above) and SMARCA4 (see below) occupies the promoters of genes known to be implicated in ESC self-renewal, including Bmp4, Esrrb, Klf5, Lifr, Stat3, and Zfp42 (Rex1), as well as various DNA methyl transferases and histone deacetylases [75]. Such observations suggest that ESC and TSC share common means of controlling self-renewal and maintaining pluripotency. The same complex also occupies the presumed control sequences of genes that are enriched in TSC relative to ESC, such as Elf5, Fgfr2 and Fgfr4, Hand1, and Tead4. It seems possible that the EOMES/SMARCA4/TCFAP2C triad in TSC could have a somewhat analogous function to POU5F1/SOX2/NANOG in ESC.

Smarca4 (Brg1)

In the mouse, loss of SMARCA4, a transcription factor involved in chromatin remodeling, halts development at the blastocyst stage. SMARCA4 appears to regulate ESC self-renewal, as its knockdown causes differentiation and repression of several pluripotency genes, such as Pou5f1 and Sox2 [84, 85]. Additionally, SMARCA4 has recently been implicated as a crucial factor in maintaining TSC self-renewal [75] by interacting with other transcription factors that are essential to this process (see TCFAP2C above). SMARCA4 may also partner with CDX2 to repress Pou5f1 expression. As discussed in greater detail below, TR emergence requires destabilization of the pluripotency network governed by POU5F1 and its partners.

Ets2

The Ets2 gene is widely expressed and necessary for normal development of the embryo proper, but it also directs crucial events in the placenta. In mice, for example, Ets2 inactivation, like that of Tcfap2c, causes embryonic lethality by Day 8.5 [86] due to failures in ExE development [87] and abnormal TR differentiation [86]. ETS2 is also required for full expression of a wide range of signature TR genes, including CGA and CGB in the human [88, 89] and IFNT in ruminant TR [90, 91]. In TSC, Ets2 silencing slows growth, decreases expression of a number of genes implicated in self-renewal, and initiates differentiation [92, 93]. Mechanisms include the transcriptional regulation of Cdx2, which has a series of ETS2-binding elements in its upstream control regions, with GATA factors playing a role as well [92].

Elf5

ELF5, a second ETS-domain transcription factor, is implicated in early placental development. Elf5^−/− conceptuses implant, form an ectoplacental cone, but lack ExE. TSC cannot be derived from these embryos, suggesting a fundamental defect in their generation or self-renewal [94]. In many respects, the mutant conceptuses have similar features to those null for ETS2. Elf5 is normally methylated and not expressed in ESC and hypomethylated and expressed in TSC [95]. One proposed role for ELF5 is in transcriptional control of Cdx2 and Eomes, with the three transcription factors forming a positive feedback loop that reinforces the early stages of TSC commitment [95]. Interestingly, ELF5/CDX2 double-positive cells have been identified among human villous cytotrophoblasts and, as discussed previously in the section on primate TSC, may define a TSC-like subpopulation [37].

Esrrb

The orphan nuclear receptor Esrrb, whose gene product is structurally related to the estrogen receptor ER-alpha, functions in both ESC and TSC. In the former, it regulates Nanog [96]. Ectopic expression with POU5F1 and SOX2 reprograms fibroblasts into iPSC [97]. In mouse conceptuses, ESRRB localizes to the ExE region [98], and the ablation of its gene triggers precocious differentiation of TR precursors to giant cells. Addition of the synthetic estrogen diethylstilbesterol (DES), which causes dissociation from ESRRB of a necessary coactivator, possibly a steroid, drives differentiation and TR giant cell accumulation. The latter observation, which helped confirm a role for ESRRB in maintaining TSC self-renewal, suggests that elements of early placental development might be under steroid control.

Sox2 and Other Genes That Operate in the Pluripotency Network

As discussed below, destabilization of the transcriptional networks that govern ESC self-renewal results in TSC formation. SOX2 (along with POU5F1 and NANOG) form a trio of transcription factors that are central to self-renewal and pluripotency of mouse and human ESC [99]. Yet, like the gene products of Esrrb (see above), Klf5, Lin28, Rest, Sall4, Stat3, Tbx3, Foxd3, and Tert, SOX2 is also highly expressed in TSC [44, 75]. These observations suggest that there may be cohorts of factors potentiating stemness of both ESC and TSC. SOX2, which in the mouse conceptus is expressed in multipotent cells of ExE as well as epiblast, is down-regulated as the TR lineages differentiate [100]. Although persistence of maternal SOX2 protein through early development complicates the interpretation of standard knockout experiments, crossing heterozygote nulls showed that mutant blastocysts lacked a defined ICM and died soon after implantation without developing either epiblast or ExE [100], a phenotype similar to that observed for Foxd3^−/− conceptuses [101, 102]. Neither Foxd3^−/− nor Sox2-mutant blastocysts generate ESC or TSC, and, normally, SOX2 expression is even higher in TSC than in ESC [75]. Less than a 2-fold increase of SOX2 protein levels in ESC is sufficient to down-regulate Nanog and drive TR and mesodermal and ectodermal differentiation [103]. This finding, along with analogous data showing that POU5F1 levels are crucial to ESC self-renewal [104], highlight the importance of the rheostat mechanisms that control protein generation and turnover during the early stages of embryonic development, a subject about which little is known and that could have equally important ramifications for cell cycle control [105].

Signaling Pathways Implicated in TSC Maintenance

It is clear that derivation and self-renewal of ESC and TSC require different inputs. For example, mouse ESC depend on LIF/STAT3 signaling, whereas mouse TSC require FGF4 and components released by the feeder cells, especially proteins with TGFB/ACTIVIN-like activity [106], to sustain proliferation in an undifferentiated state. Other less well defined factors are components of the fetal bovine serum normally added to the culture medium. In mouse conceptuses, the epiblast is the local source of FGF4 and the TGFB-related protein NODAL, which are needed to sustain proliferation within the neighboring ExE and prevent differentiation [19] (Fig. 2). The NODAL/TGFB pathway requires proteolytic processing of NODAL, activin type II receptors (ACVR2A and ACVR2B), ALK4 and ALK7 signaling through their protein kinase activities, and SMAD 2 and 3 transcription factors, which are substrates for the receptor kinases. FGF4, on the other hand, binds its cognate receptor, the serine/threonine protein kinase FGFR2, and feeds into the RAS/MAPK1(ERK2) signal transduction pathway via a number of intermediary components that include the adapter protein FRS2 [21] and the signaling protein tyrosine phosphatase PTPN11 (SHP2) [107]. NODAL and FGF4 receptors and their downstream signaling networks, which are far more complex than presented in this brief summary, appear central to maintaining the stem cell niche where the balance between self-renewal and differentiation is crucial to normal development.

Although mouse ESC appear to be insensitive to FGF4, activation of the RAS/MAPK1 (ERK2) pathway by other means, such as overexpression of a constitutively active Hras1 gene, up-regulates TSC markers and drives the ESC toward a TR fate, while inhibiting MAPK signaling with PD98059 markedly reduces the ability of mouse blastocysts to form outgrowths and give rise to TSC [13]. These observations could explain why MAPK inhibition aids the derivation of ESC from “difficult” species such as the rat [108] and in reprogramming of somatic cells to iPSC [109].

It should be emphasized that FGF4 dependency may not be a universal feature of TSC. For example, TE-derived TSC from the rhesus monkey do not require endogenous growth factors for their expansion [25]. Instead, their self-renewal is entirely supported by specific constituents of the culture medium, which contains 15% fetal bovine serum. Additionally, well-authenticated TSC derived from blastocyst outgrowths of the common vole (Microtus rossiaemeridionalis) do not require exogenous FGF4 for continued self-renewal in an undifferentiated state. Moreover, the cells did not express Fgfr2, which encodes the mouse TSC surface receptor for FGF4 [110]. Curiously, TSC derived from mouse blastocysts produced by nuclear transfer also showed little or no FGF4 dependency, and their activin requirement was much reduced [111]. The reasons are unclear but might be related to incomplete reprogramming, as evidenced by their increased capacity for self-renewal and the placentomegaly that is routinely observed in cloned mice [112].

WNT signaling may also be operative in mouse TSC. Lymphoid enhance factor-1 (LEF1), a downstream component of the pathway, has been implicated in modulating Cdx2 expression [113]. However, Lef1^−/− conceptuses are born alive, suggesting that the placenta is functional despite the fact that several other organs are defective [114]. Whether other factors can serve compensatory roles for LEF1 is unknown.

BMP4-regulated pathways also seem to play a role. For example, BMP4 family members promote hESC differentiation to TR in two-dimensional cultures [40, 115]. Specifically, addition of BMP4 as low as 10 ng/ml leads to the appearance of larger, more flattened cells at the periphery of the colonies that express TR markers [116]. In the absence of FGF2, which is widely used to culture hESC, this process appears to be unidirectional, with no other lineages emerging [116, 117]. In contrast to spontaneous colony differentiation, which most often leads to cell migration away from the outer edges of the colonies, TR arise from the periphery in response to BMP4 and move inward [40, 116, 118]. Within a few days, multinuclear hCG+ TR cells and a mononuclear HLAG+ TR population emerge. Whether these BMP4-treated colonies acquire a transient TSC population remains unclear since they differentiate so quickly, particularly under normoxic conditions [116]. Finally, it should be emphasized that growth factor effects depend on the culture substratum [119, 120], most likely through interactions with signals transmitted through adhesion receptors, a principle first noted for mature TR lineages [121].

Suppression of Pluripotency and Acquisition of the TR Ground State

Much has been made of the apparent reciprocal relationship between Pou5f1 and Cdx2 and the ability of each to silence the other's function to allow segregation of TE and ICM. Indeed, it is often assumed that at the blastocyst stage of development, POU5F1 is restricted to ICM and absent from TE [122], while CDX2 is expressed only in TE [63, 67, 69, 123]. While the latter statement is correct, the former is not. Even in the mouse, the main source of information on the two transcription factors, POU5F1 is detectable in TE, although it gradually disappears as the blastocyst expands and implantation begins. Similarly, coexpression of POU5F1 and CDX2 occurs in human [124], monkey [62], bovine [125–127], and porcine blastocysts [128]. On the other hand, transcripts for POU5F1 are lost from TE by the blastocyst stage in bovine conceptuses [129], suggesting that the POU expression that remains is the result of persistent protein rather than gene transcription and continued translation. Together these observations suggest that POU5F1 protein is not sufficient to repress TE emergence, although it may simply be a question of relative levels.

Generally, ESC and TSC lines also have opposite POU5F1 and CDX2 expression. POU5F1, which is abundant in ESC, works cooperatively with NANOG, SOX2, and other transcription factors to regulate expression of the gene networks that maintain pluripotency and repress emergence of the specialized lineages [130, 131]. In contrast, there is no evidence that mouse TSC transcribe or translate Pou5f1. The first hint that overexpression of what are now known as pluripotency genes might actually be able to trigger dedifferentiation of somatic cells was the demonstration that forced expression of POU5F1 in choriocarcinoma cells almost completely silenced expression of CGA and CGB [132, 133]. It is now clear that, in ESC, this silencing effect applies to a large number of other lineage-specific genes. Conversely, the pluripotent effects of POU5F1 and its cohorts can be overridden by reducing their expression. Knockdown of Pou5f1 [134–136], Nanog, and Sox2 [137–139] in mouse ESC causes the cells to differentiate predominantly to TR. In human ESC, siRNA silencing of POU5F1 followed by culture without FGF2 also leads to TR differentiation, with up-regulation of CDX2 and other TR marker genes [135, 140, 141]. A similar up-regulation of TR markers has been observed after NANOG silencing [142], although markers of extraembryonic endoderm also increase.

Finally, it is worth noting that some lines of stem cells isolated from rat blastocysts appear to combine features of ESC, TSC, and stem cell precursors of extraembryonic endoderm [143]. These mixed populations, although probably clonal in nature, exist in distinct but interconvertible forms within the culture and have been hypothesized to originate from uncommitted cells present in the ICM of the founder blastocysts. Their existence emphasizes that the networks defining stemness show varying degrees of durability and can easily be switched. Beyond transcriptional regulators, other factors can bias early cell fate decisions. For example, polyploid cells are shunted to the TE and subsequent TR lineages during mouse embryogenesis [144, 145], suggesting that genomic imbalance can also prompt lineage switching. The mechanisms involved, which are not understood, could reveal interesting insights that might have implications for other fields, such as cancer biology.

SUMMARY AND FUTURE DIRECTIONS

What new developments are on the horizon for TSC research? Currently, a great deal is known about the generation of mouse TE and TSC. Much less is understood about the equivalent cell types and their developmental potential in humans and other mammals. Clearly, the stemness networks that maintain human ESC pluripotency, which are relatively fragile and easily overridden, if intact, repress TR emergence. It may be possible to devise cell culture conditions for capturing cells in the early stages of the ESC-to-TSC transition that can self-renew in a relatively undifferentiated state and, on directed differentiation, form TR. At the same time, methods for producing human ESC and iPSCs are improving, and it is likely that new developments will enable the derivation of lines, like the ones from rat ICM [143], that are totipotent rather than pluripotent. Provided that stabilizing culture conditions can be developed, it should then be possible to isolate early-stage TR clones that can be propagated indefinitely. Achieving such a goal will allow investigators to study the ESC-to-TE transition in humans, which is largely a black box. Finally, identification and isolation of the more mature stem cell population or populations that reside in the early gestation human placenta and enable growth of this organ will allow us to study the events downstream of TE allocation. Parallel advancements in analytical technologies occurring alongside the development of the next generation of cell-based tools make feasible the generation of the comprehensive data sets that will reveal the mechanisms that control the entire process of TR differentiation. Given the rapid rate at which this organ evolved [146] and its structural diversity among species [147], it is likely that microRNAs and epigenetic mechanisms are playing pivotal roles. Overall, the development of cell culture models to study human TR differentiation is particularly important for understanding placentation in our own species. Basic knowledge about the processes involved is crucial for advancing our understanding of the origin of pregnancy complications, such as preeclampsia, that are thought to involve defects in the differentiation of specific TR populations [148].