| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Minireview |
Department of Cell and Developmental Biology, University of Michigan Medical School, Ann Arbor, Michigan 48109-0616
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONPLURIPOTENT CELLS OF THE...STEM CELL CHARACTERISTICSLINEAGE SEGREGATION IN THE...DIFFERENTIATION PARADIGMSLESSONS FOR THE EMBRYO:...CONCLUSIONSREFERENCES |
|---|
central nervous system, developmental biology, early development, ectoderm, embryo, endoderm, epiblast, germ cell, growth factors, inner cell mass, mesoderm, pluripotency
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONPLURIPOTENT CELLS OF THE...STEM CELL CHARACTERISTICSLINEAGE SEGREGATION IN THE...DIFFERENTIATION PARADIGMSLESSONS FOR THE EMBRYO:...CONCLUSIONSREFERENCES |
|---|
|
PLURIPOTENT CELLS OF THE EARLY EMBRYO |
|---|
|
TOPABSTRACTINTRODUCTIONPLURIPOTENT CELLS OF THE...STEM CELL CHARACTERISTICSLINEAGE SEGREGATION IN THE...DIFFERENTIATION PARADIGMSLESSONS FOR THE EMBRYO:...CONCLUSIONSREFERENCES |
|---|
|
Until about E5.5, individual epiblast cells remain totipotent as judged by their ability to integrate following transplantation into the ICM of E3.5 blastocysts, and like the ICM, continue to express alkaline phosphatase, E-cadherin, SSEA1, and OCT4 [12]. Cells of the epiblast gradually lose pluripotency, losing the ability to colonize the ICM at E6.5–E7.5, with the genome-wide methylation of CpG islands in somatic cells that occurs at this stage [10]. As late as E7.5, the epiblast retains the ability to form teratocarcinomas in nude mice, however. Epiblast cell number increases significantly before and during gastrulation; by E5.5, there are approximately 120 cells in the epiblast, increasing to approximately 800 at the onset of gastrulation [13].
Despite the striking regulative capacity of cleavage staged embryos, there may be subtle genetic differences between blastomeres as early as the two-cell stage, but whether there is an unequal allocation of maternal factors at the first cleavage can only be inferred. Although it has been assumed that, before compaction, individual blastomeres are equivalent, essential positional information determined at the two-cell stage must somehow be inherited [14]. To date, gene profiling has identified many differentially expressed transcripts in the two cell types in the blastocyst [e.g., 15]. As the ICM differentiates to form the hypoblast and then epiblast, there are major changes in gene expression that restrict the developmental potential of the embryo [e.g., 16], although the epiblast appears to be largely unpatterned until early streak stages. Restrictions in gene expression in the epiblast have been invoked to explain the inability to derive ES cells from epiblast, unlike the ICM, and also to explain variations in differentiation potential of the various ES cell lines. There is also considerable interest in identifying genes expressed differentially between ICM and epiblast, which might be involved in maintaining pluripotency or promoting differentiation.
Although gene targeting experiments have produced many preimplantation lethal phenotypes, these studies have identified few genes involved in maintaining pluripotency of the ICM; most lethality appears to be due to roles of the mutated genes in basic cellular functions. Of the genes that produce selective effects on the ICM rather than trophoblast, many, such as integrin beta 1 (Itgb1), are essential in maintaining the organization of the early embryo [17]. Others, such as taube nuss (Tbn), suppress apoptosis in the epiblast [18], and deletion of other genes expressed in the ICM or epiblast cause premature differentiation, e.g., Smad2 null embryos [19]. A threshold number of epiblast cells appear to be required for the initiation of gastrulation, so that genes that interfere with epiblast cell number also produce gastrulation phenotypes. Because they express transcripts typical of the ICM and are totipotent, ES cells are increasingly being employed to tease out the roles of genes involved in maintaining pluripotency vs. differentiation of the early embryo. Several have expression patterns and knock out phenotypes that support an essential role in maintaining pluripotency.
|
STEM CELL CHARACTERISTICS |
|---|
|
TOPABSTRACTINTRODUCTIONPLURIPOTENT CELLS OF THE...STEM CELL CHARACTERISTICSLINEAGE SEGREGATION IN THE...DIFFERENTIATION PARADIGMSLESSONS FOR THE EMBRYO:...CONCLUSIONSREFERENCES |
|---|
Interestingly, LIF is only able to sustain ES cells in the presence of serum, suggesting that additional factors are required. A number of unique factors that play a role in maintaining ES cell self-renewal and pluripotency have been identified. These would include the essential POU homeodomain protein OCT4 [24], the variant homeodomain containing protein NANOG [25–27], the SRY family member Sox2 [28], Foxd3 (previously Genesis) a member of the forkhead winged-helix family [29], and possibly Wnt signaling [30]. Somewhat surprisingly, bone morphogenetic protein 4 (BMP4) can replace serum in cultures containing LIF by inducing the expression of Id (inhibitors of differentiation) genes [31]. In general, factors that maintain pluripotency are expressed in the future ICM at compaction of the morula, in the ICM of the blastocyst, and are downregulated with lineage differentiation at gastrulation. Expression profiling studies comparing undifferentiated and differentiated ES cells are identifying many additional genes that are downregulated with differentiation [32], so the list of players is likely to increase.
The best studied of the transcription factors involved in maintaining pluripotency of ES cells and the epiblast is the POU homeodomain containing transcription factor OCT4, which is encoded by the Pou5f1 locus. Maternal OCT4 protein is present in unfertilized oocytes, remaining in two-cell to eight-cell embryos [33, 34]. Oct4 mRNA is expressed in the cells in the morula destined to form ICM, but at compaction Oct4 is downregulated in trophectoderm. In the early blastocyst, however, there is a transient upregulation of Oct4 in the primitive endoderm followed by a final downregulation with differentiation [34]. Expression remains in the epiblast of the egg cylinder, with OCT4 protein becoming further restricted after gastrulation to primordial germ cells [34, 35]; Oct4 is also expressed at high levels in germ cell tumors [36]. Gene-targeting experiments have demonstrated that Oct4 is essential for the formation of the inner cell mass and the epiblast because homozygous null embryos die between implantation and egg cylinder stages of development [35]. OCT4 is required for self-renewal of ES cells, but unlike STAT3, OCT4 and the NANOG protein also appear to inhibit differentiation. Forced overexpression of Oct4 does not render ES cells free of their dependence on LIF. Rather, conditional up- or downregulation of Oct4 expression in ES cells demonstrates that downregulation produces trophoblast-like cells [37–39], while increased levels of expression produce primitive mesoderm and endoderm [37], reminiscent of the transient upregulation as primitive endoderm initially differentiates in the blastocyst [34]. Several downstream targets of Oct4 have been identified [40], including the extracellular matrix protein osteopontin [41], which may promote the lateral migration of primitive endoderm to line the trophoblast as well as initiate expression of endoderm lineage-specific gene expression [37]. Other targets include Hand1 and Fgf4 [42], which are expressed in early trophectoderm [37]; Fbx15, which is expressed in ES cells and later in testis [43]; and Zfp42, also known as Rex1 [44]. Interestingly, OCT4 binding of octamer DNA sequences can either repress or activate transcription depending on flanking sequence [24], consistent with its critical role in maintaining pluripotency and controlling differentiation.
The Oct4 promoter contains a retinoic acid (RA) -responsive repressor element [45], such that exposure of ES cells to the morphogen/teratogen retinoic acid inhibits Oct4 expression and allows expression of lineage-specific transcription factors present in ES cells. The Oct4 promoter also contains a proximal element that drives Oct4 expression in the epiblast that is downregulated at gastrulation, and a germ cell-specific distal enhancer that is not active in the epiblast [46]. Germ cell nuclear factor, an orphan nuclear receptor, binds to the distal enhancer to restrict expression to the germ cell lineage [47]. By using this distal promoter to drive expression of EGFP to ES cells, then selecting those cells, it has recently been possible to differentiate oocytes from mouse ES cells in vitro [48].
OCT4 interacts with a number of other factors, including the SRY-related HMG box containing transcription factor SOX2, E1A protein, the transcriptional coactivator Utf1 [49], and the Forkhead box protein FOXD3. Interestingly, because OCT4 and FOXD3 bind the same sequence, the presence of Oct4 early in development appears to repress transcriptional activation of Foxd3, suggesting a mechanism by which lineage-specific gene expression (in the case of Foxd3 endoderm lineage specification) can be controlled before gastrulation, when OCT4 levels decrease [50]. Despite the widespread belief that downregulation of Oct4 is required for the differentiation of somatic lineages, recent studies have demonstrated a role for Oct4 in neuronal differentiation of ES cells [51] and in the ability of isthmic neuroepithelial cells to respond to FGF8 signals [52].
The recently identified pluripotency factor Nanog is a member of the homeobox family of DNA-binding transcription factors [25, 26]. Two human genes and a single mouse gene with three alternative splice variants have now been identified [27]. Nanog is first expressed in the compacted morula, in the ICM, then in the proximal epiblast at the location of the future primitive streak. Nanog is downregulated at gastrulation in mesoderm and endoderm, remaining in the epiblast to E8. Nanog is present in ES cells, primordial germ, and EG cells [25–27] and can be detected by reverse transcription-PCR in adult tissues [27] as well. Interestingly, NANOG, STAT3, and OCT4 appear to affect both overlapping and independent events. Overexpression of Nanog in ES cells renders them independent of LIF [25, 26] and of BMP [25] and suppresses their differentiation capability [26], possibly by sustaining Id gene expression. However, Nanog is expressed in Oct4 –/– embryos [25], suggesting that they act in parallel pathways. Nanog null embryos and ES cells form extraembryonic endoderm at the expense of epiblast [26]. Thus, like Oct4, Nanog appears to regulate both self-renewal and inhibit differentiation; however, Nanog appears to be critical at slightly later stages of development because deletion does not affect early stages of trophoblast differentiation (as is the case for Oct4) and it is expressed in the epiblast until E8, remaining in some adult tissues [27]. Like Oct4, it appears that Nanog may regulate differentiation by transcriptional repression of genes that promote differentiation. In the case of Nanog, the enhancers of both Gata6 and Rex1/Zfp42 [26] contain NANOG binding sites. Interestingly, NANOG contains an amino terminal region of homology to SMAD4, suggesting that NANOG in the epiblast may play a role in controlling TGFß signaling [27]. Smad4 has previously been shown to be critical at gastrulation; Smad4 –/– embryos die at E6.5 due to failure of proliferation in the epiblast [53].
Foxd3 and Sox2 appear to play a slightly later role in maintaining the epiblast and ES cells. Foxd3 (also known as Genesis), a member of the Forkhead box (Fox) family of transcription factors, with a winged-helix DNA binding structure, was initially characterized as being restricted to ES cells and teratomas [54]. More detailed analyses have determined that it is first expressed in the blastocyst, persisting in the primitive ectoderm at E6.5, then becoming largely restricted to neural crest cells [55]. It has not been possible to derive ES cells from embryos null for Foxd3 nor to form teratocarcinomas from them. Although homozygous null embryos appear to form a normal ICM (that expresses Oct4, Sox2, and Fgf4), they die at gastrulation on E6.5 lacking epiblast, with an increase in extraembryonic ectoderm and endoderm [29]. Chimera rescue studies have indicated that the requirement for Foxd3 in the ICM is noncell autonomous, suggesting that it may regulate a required cell surface component or secreted molecule. Foxd3 and Oct4 appear to cooperatively regulate expression of some target genes, such as osteopontin (Spp1), with Oct4 acting as a corepressor of Foxd3-induced endodermal lineage transcription factors.
The SRY family member Sox2, which is expressed in the ICM and in ES cells, is a coactivator of Oct4 activity [56], and gene targeting experiments suggest that Sox2 is required to maintain the epiblast, in a cell autonomous manner [28]. Like OCT4, maternal SOX2 protein is present in preovulatory oocytes, then, unlike most maternal proteins that are destroyed by the eight-cell stage, maternal SOX2 protein is maintained in the ICM and trophoblast. Zygotic expression of Sox2 is first detected at the morula stage (E2.5), then it is expressed not only in the inner cell mass (E3.5) and the epiblast at E6.5, but also in extraembryonic lineages, then throughout the neural plate, becoming restricted to stem cells; neural, gut, and germ cells [57]. Likely due to the persistence of maternal SOX2 protein in the ICM, Sox2 –/– embryos appeared morphologically normal at the blastocyst stage of development but died shortly after implantation at E6.0, with no egg cylinder, although trophoblast and primitive endoderm were present. Outgrowths of blastocysts grown in vitro were also initially normal, but formed only trophectoderm and primitive endoderm, suggesting that SOX2 is required to maintain the epiblast and that maternal protein may have masked an early lethal phenotype in null embryos.
Interestingly, forced expression of Sox2 does not cause differentiation of ES cells or render them OCT4 or LIF independent, but results in cell death [26]. OCT4 and SOX2 cooperate in transcriptional activation of several genes, including osteopontin. They cooperatively bind an Fgf4 enhancer element [42] to control Fgf4 expression in the ICM and epiblast. ES cells cannot be derived from Fgf4 –/– blastocysts unless FGF4 protein is added to the medium. FGF4 from the epiblast also appears to play an important role in patterning extraembryonic and trophoblast lineages in the early embryo [58].
|
LINEAGE SEGREGATION IN THE EPIBLAST |
|---|
|
TOPABSTRACTINTRODUCTIONPLURIPOTENT CELLS OF THE...STEM CELL CHARACTERISTICSLINEAGE SEGREGATION IN THE...DIFFERENTIATION PARADIGMSLESSONS FOR THE EMBRYO:...CONCLUSIONSREFERENCES |
|---|
Germ cells are the first embryonic lineage to differentiate from the epiblast, at about E5.5. Germ cells migrate from their origin in the proximal epiblast to the extraembryonic mesoderm in the base of the allantois, where primordial germ cells (PGC) escape somatic cell DNA methylation, then migrate back into the embryo and eventually into the genital ridges. Factors present in the extraembryonic ectoderm, particularly BMP4 and BMP8b, are required in vivo for germ cell determination, although direct exposure of mouse ES cells in monolayer culture to BMP4 produces mesoderm [64].
Oocytes have been obtained by sorting ES cells expressing EGFP under the control of the distal (germ cell-restricted) Oct4 enhancer [48]. After 26 days in vitro, oocytes grew to 50–70 µm, then divided to form blastocyst-like structures that expressed oocyte-specific markers, including Zp2, but not Zp1, so were very fragile. PGCs have also been obtained from the rare SSEA1+, OCT4+ ES cell that remains in differentiating embryoid bodies [65]. These cells were mitotically active in response to retinoic acid (a property unique to PGCs as ES cells differentiate in the presence of RA), expressed markers of the male germ line, erased methylation marks on imprinted genes, completed meiosis, and, following injection into haploid oocytes, reconstituted the diploid chromosome content and supported development to the blastocyst stage. An interesting difference between the two studies is that male germ cells were obtained only from three-dimensional culture (as embryoid bodies), while female germ cells were derived in two-dimensional cultures (although many lifted from the culture dish). Three-dimensional culture was also required for male germ cell differentiation from ES cells sorted based on expression of the mouse homolog of the vasa gene (Mvh/Ddx4) driving eGfp or lacZ [66]. Mvh-expressing ES cells were cocultured with cells expressing BMP4, followed by additional coculture with dissociated E12.5 male gonad cells, then were transplanted under the testis capsule, where differentiating ES cells undergo meiotic divisions and produce sperm. These studies set the stage to carefully study the germ cell niche and present an important opportunity to examine the sequential steps in the development of the germ cell lineage.
At gastrulation, the pluripotent epiblast is gradually transformed into the three primary germ cell layers: ectoderm, mesoderm, and endoderm, as well as forming extraembryonic mesoderm. At about E6.5, the primitive streak begins to form in the proximal epiblast near the junction with extraembryonic tissues, marking the posterior region of the embryo. Cells in the epiblast divide rapidly and delaminate from the epithelial sheet, forming a middle layer of mesoderm. Additional cells move from the anterior streak into the visceral endoderm, replacing it with definitive (embryonic) endoderm first in the anterior region of the embryo, and, at later stages of gastrulation, in the posterior portion of the embryo. Signals from the transient anterior visceral endoderm influence patterning in the epiblast [67].
As gastrulation progresses, the streak proliferates rapidly and extends down the posterior side of the embryo. Cells retain pluripotentiality, and in the human embryo, remnants of the primitive streak may form sacrococcygeal teratomas, which contain tissues from all germ layers [68]. The anterior-most approximately 20 cells of the primitive streak lack a visceral endoderm layer [69] and comprise a transient signaling center, the node, which secretes signaling factors that induce the ectoderm to form neural ectoderm. Cells migrating from the epiblast through the node form the notochord, which secretes molecules that pattern the neural ectoderm slightly later in development.
ES cells have been employed extensively to study the signaling pathways involved in the differentiation of the three primary germ layers, and it has often been suggested that, because they express many genes typical of the ICM, ES cells may be the functional equivalent of the dorsal ectoderm animal cap [70] that has been so useful in studying gene expression and lineage segregation in Amphibian development. A corollary to this postulate is that the differentiation of ES cells requires an obligate step first to endoderm, mesoderm, or ectoderm. For example, as ES cells differentiate to mature derivatives, such as a pancreatic beta cell, the ES cell may first form endoderm and is not capable of direct differentiation to a beta cell. Alternatively, ES cells may simply have a limited differentiation repertoire that reflects that of the inner cell mass: to primitive endoderm, trophectoderm, or epiblast, which may be required to acquire competency for further lineage-specific differentiation [71]. Thus, the variability in differentiation potential between ES lines as well as the tendency of certain lines to form particular differentiated derivatives might be explained by the stage of development attained by the embryo at the time of explantation into culture to derive ES cells.
At gastrulation, there is an allocation of epiblast between the germ layers, making it difficult to tease out the specific roles of the many factors that can affect lineage specification. Because gastrulation fails if the epiblast is disorganized or if a threshold number of epiblast cells is not present, it is often not possible to distinguish between genes that affect the epiblast and those that control lineage specification. ES cells therefore present an important opportunity to identify genes involved in directly controlling lineage differentiation in the early embryo.
Mesoderm differentiation at gastrulation fails in a number of gene-targeted embryos: Bmp4 null embryos fail to form mesoderm [72], as do Smad4 deficient mice [53], and Smad4 null ES cells differentiate preferentially into neuronal cells [73]. Mesoderm-like cells can differentiate from ES cells exposed to BMP4, upregulating T expression in a concentration-dependent manner [64, 70, 74], while BMP4 treatment of slightly more differentiated cells promotes surface ectoderm differentiation at the expense of neural ectoderm [61, 75].
At gastrulation, promesoderm genes often act at the expense of ectoderm, and vice versa. Thus, Wnt3a promotes differentiation of the epiblast to a mesodermal fate, Wnt3a –/– embryos form ectopic neural tubes rather than mesoderm at gastrulation [76], and T (brachyury) mutant embryos form neuroepithelial rosettes at the expense of mesoderm [77]. In chimeric embryos, Fgfr1 mutant epiblast also forms multiple neural tubes rather than mesoderm [78] and is required for expression of mesoderm-restricted genes T and Tbx6 [79]. FGF4 is required for survival and proliferation of the ICM and ES cells [80], and exposure of isolated epiblast to FGF2 induces mesoderm [81]. Other candidates for mesoderm determination genes would include Mixl1, a paired class homeobox gene, which is expressed in primitive streak and early mesoderm [82], which, when deleted, produces lethal defects of axial mesoderm [83].
Embryonic endoderm forms as cells migrate through the primitive streak to replace visceral endoderm, and it is clear that signaling from the visceral endoderm is required for induction and patterning of both mesoderm and neural ectoderm [67]. When the outer layer of differentiated endodermal cells is removed from ES cell aggregates (embryoid bodies), differentiation fails and cells remain ES like, further supporting a requirement for endoderm-derived signals for differentiation [84]. This is reminiscent of the debate regarding whether the primitive endoderm rind that forms in late blastocyst stages is inhibitory to ES cell derivation [see, e.g., 12].
Nanog expression in the early epiblast appears to inhibit endoderm differentiation by repressing Gata6, and extraembryonic endoderm has been produced from ES cells by the forced expression of Gata4 or Gata6 [71]. Early high levels of OCT4 appear to repress transcription of Foxd3 to control the early differentiation of the endoderm lineage [50] and overexpression of Foxa2 produces endoderm from ES cells [85]. BMP signals also appear to be involved in visceral endoderm differentiation and cavitation of the embryoid body [86], and Smad2 null ES fail to contribute to embryonic endoderm. Loss of Sox17 from the early embryo decreases, but does not inhibit formation of embryonic endoderm, although Sox17 –/– ES cells do not colonize gut endoderm in chimeras [87]. Beta-catenin (Catnb) null embryos also fail to form embryonic endoderm [88], suggesting a requirement for Wnt signaling in endoderm differentiation as well.
In the gastrulating embryo, induction of neural ectoderm from ectoderm is thought to result from inhibition of BMP signal transduction in the ectoderm by factors secreted by the node, particularly NOGGIN, CHORDIN, and FOLLISTATIN. Exposure of mouse ES cells to BMP4 consistently inhibits neural differentiation and induces mesoderm from ES cells or surface ectoderm from slightly later EPL cells. Primitive neural cells differentiate from ES cells in response to NOGGIN protein [73] or following transfection with noggin (nog) expression vectors. Interestingly, in addition to neural differentiation, chordin (Chrd) overexpression in ES cells produced mesoderm-like cells [89]. Interfering with the BMP signal transduction machinery in ES cells (Smad 4 –/– ES cells) also results in widespread neuronal differentiation [73], confirming the role of this signaling cascade in neural differentiation.
Wnt signaling appears to be involved in the proliferation of ES cells [30] and inhibition of the Wnt signaling pathway at different levels [90–92] has demonstrated that Wnt signals are involved in neural differentiation, but the particular players have not yet been identified. Interestingly, inhibition of Wnt signaling (inactivation of Apc or introduction of a d/n ß-catenin) in ES cells inhibited neural differentiation and upregulated expression of BMP4 and BMP7, specific inhibitors of neural differentiation [91]. NOGGIN protein overrode aspects of this inhibition, suggesting a molecular mechanism that bridges induction, withdrawal from cell cycle, and neural differentiation [92].
FGFs are also involved in cell-fate decisions at gastrulation and in ES cells. FGF4 is expressed throughout the epiblast [93], and Fgf4 transcription is coregulated by OCT4 and SOX2 binding to an Fgf4-enhancer element [42]. Fgf4 –/– embryos die at gastrulation and Fgf4 –/– ES cells are unable to proliferate in vitro [94], but can be derived if FGF4 is added to the culture medium [93]. Treatment of differentiating ES cells with FGF4 induces the expression of the pan-neural Sox1 gene in monolayer culture [95]. Recently, the churchill (chch) gene has been identified in the chick embryo and shown to be induced by low, sustained FGF signaling [96]. Chch induces expression of the Smad interacting protein 1 (Fhx1b gene) that directly inhibits brachyury (T) expression, thereby inhibiting mesoderm differentiation and sensitizing the ectoderm to neural inducers. (However, null embryos only exhibit neural crest defects; [97]). Like geminin (Gmnn), which is strongly induced by the BMP inhibitors NOGGIN and CHORDIN [98], chch marks the neuroepithelial boundary with the surface ectoderm and may represent an elusive intermediate between induction and adoption of a neural fate by the ectoderm.
|
DIFFERENTIATION PARADIGMS |
|---|
|
TOPABSTRACTINTRODUCTIONPLURIPOTENT CELLS OF THE...STEM CELL CHARACTERISTICSLINEAGE SEGREGATION IN THE...DIFFERENTIATION PARADIGMSLESSONS FOR THE EMBRYO:...CONCLUSIONSREFERENCES |
|---|
When grown in suspension culture, ES cells form small aggregates of cells surrounded by an outer layer of visceral endoderm [99]. Because their size, differentiation capacity, and gene expression profile resemble the early postimplantation embryo, these aggregates have been termed embryoid bodies "EB" and are often employed as models of differentiation and gene expression in early development.
Because the pattern of differentiation of ES cells as EB parallels that of the early embryos, most differentiation paradigms [but see 87, 88, 95] begin with a several-day aggregation step in which ES cells are grown in suspension in the absence of LIF (± retinoic acid) to form embryoid bodies [100, 101]. EBs have been considered to be the developmental equivalent of the egg cylinder-staged mouse embryo, with an outer layer of endoderm and a core of differentiating cells, often surrounding epithelial-lined cavities, cystic EBs. RNA in situ hybridization analyses have demonstrated that derivatives of all germ layers differentiate in EBs, in some ways recapitulating in vivo gene-expression patterns [102]. While it is clear that cell-cell interactions promoted by aggregation can initiate differentiation, several recent studies have demonstrated a requirement for three-dimensional culture to obtain lineage differentiation, particularly of germ cells [65, 66], and FGF effects on differentiation vary depending on if exposure occurs in monolayer [95] or three-dimensional culture [93].
One explanation for these observations is that only the outer layer of endoderm of the EB is exposed to added growth factors, and cells in the interior of an EB differentiate in response to signals produced by the differentiating endoderm. There is some evidence for this, as when initial differentiated cells were removed from EBs (non-Oct4-expressing cells were killed by antibiotic selection), sequential differentiation was inhibited [81], implicating secreted signaling factors from the early differentiating endoderm in the subsequent differentiation within the EB. Incomplete differentiation of ES cells, commonly observed in embryoid bodies, may also be due to the fact that differentiated cells, particularly mesoderm, produce LIF [103].
As in the embryo, the endoderm secretes BMP2 and INDIAN HEDGEHOG; the cells in the interior of the aggregate, like the epiblast, express BMP4 [83]. Considerable evidence based on EB formation using gene-targeted ES cells has demonstrated that signaling between the outer endoderm to the core is responsible for differentiation of visceral endoderm and additional signaling by molecules produced by the visceral endoderm (as in the early embryo) pattern the ectoderm forming in the core of the aggregate. The hedgehog signaling family has recently been shown to play a role in this lineage differentiation in EB. Indian hedgehog (Ihh) expression in the visceral endoderm increases with differentiation of the EB, where it appears to promote embryonic ectoderm differentiation from the core cells of the EB [104]. Consistent with these observations, interference with hedgehog signaling resulted in the conversion of visceral endoderm to parietal endoderm and failure of ectoderm to differentiate into neural ectoderm.
Clearly, both cell number and cell-cell interactions are critical in lineage differentiation of the early embryo, and EB differentiation can offer much to the analysis of these pathways if differentiation is standardized. Hanging drop cultures to control cell number, careful analysis of the pattern of differentiation within EBs using lineage-marked ES cells (discussion below) with confocal microscopy could be employed to carefully analyze the spatiotemporal patterns of differentiation within the EB. Lineage-specific ablation and killing of differentiated cells at later time periods will help elucidate the critical periods and eventually the signaling molecules involved. The Notch-Delta signaling pathway [see, e.g., 105], cell-cell adhesion molecules, cadherins are obvious targets for these analyses. Carefully controlled studies could also elucidate whether tissue types differentiate in proximity to each other as they do in the early embryo, each presumably producing growth and differentiation factors that promote differentiation of the other.
Coculture
When the precise growth factor combinations that promote cell-type-specific differentiation are not known, ES cells have been cocultured with target tissues to promote differentiation. Consistent with their differentiation near the septum transversum in the early embryo, coculture of ES cells with explants of chick cardiac mesoderm has been efficient in producing hepatic tissue [106]. Culture of ES cells on primary stromal cells derived from the aorta-gonad-mesonephros region of the embryo or on the S17 cell line promoted neural differentiation [107], while culture on the PA6 cell line [74] or the MS5 line [107] produced dopaminergic neurons from ES cells.
If the desired cell/tissue type can be identified in mixed cultures based on a clear cell morphology such as striated muscle, beating heart cells, neuronal cells with long neurites, this approach can be very successful. However, very often, EB cultures have been shown to express markers of the desired cell type, but other cells present in the differentiating mass have not been identified. Highly differentiated derivatives (including many neuronal lineages) have been obtained by aggregation as EBs followed by sequential exposure to growth and differentiation factors known to be expressed during the development of the region [see, e.g., 107, 108]. As more is known of the molecular embryology of particular tissues, it will be possible to more reliably obtain highly differentiated cell types from ES cells.
Promoter/Trap: Cell Trapping
Promoter trapping has produced important insights into gene expression and function in development [109], and has allowed the dissection of the role cis/trans-acting factors in individual gene promoters/enhancers. Promoterless markers have been inserted into ES cells to identify novel genes [110, 111] and growth factor responsive genes [112]. Recently, modifications to this approach have created ES cells marked by constitutive expression (often driven by the ubiquitous ß-actin promoter) of the jellyfish green fluorescent protein or the yellow fluorescent protein [see, e.g., 113]. Fluorochrome-marked ES cells have been useful in analyzing aggregation chimeras and have allowed differentiated progeny to be traced following implantation [114].
Because differentiation as EB has not proven to be a particularly efficient method of obtaining purified populations of differentiated cells, promoters/enhancers from genes restricted to particular cell types have been employed to identify differentiating cells based on fluorescence and enrichment using flow sorting or magnetic bead isolation. Modifications of this technique would include the use of cell type-restricted promoters driving a neomycin phosphotransferase (neo) cassette so that cells not differentiating along the desired pathway can be killed by high levels of antibiotic. Multiple cell lines must be carefully characterized to ensure that random insertion has not disrupted important functional genes, or the marker can be inserted into the homologous chromosomal region [115].
To examine the very early lineage segregation events that occur at gastrulation, the role of growth factors and signaling molecules in lineage differentiation, several indicator ES cell lines have been developed to monitor the kinetics of differentiation, such as Oct4-eGfp. To identify primitive mesoderm, an ES cell line in which EGFP expression is controlled by the mesoderm-restricted gene brachyury (T) was created [116]. This cell line has been employed to investigate the early differentiation of mesodermal lineages, endothelium, and hematopoietic cells. The Tau (Mapt) gene [117; Fig. 2] and the central nervous system-restricted nestin (Nes) enhancer [118] have been similarly employed to drive expression of EGFP to identify primitive neurons and neural cells. These indicator lines provide the opportunity to study the growth factor responsiveness of ES cells, monitor their differentiation, and mark cells for implantation. Unfortunately, a similar early endoderm line has not yet been developed.
|
Highly differentiated derivatives have also been derived using lineage selection. Klug [119] initially employed this approach (the myosin heavy chain gene drove expression of neo) to derive cardiomyocytes from EB. A similar lineage-selection approach was employed by Li et al. [115] using ES cells in which ß-geo was knocked into the endogenous Sox2 locus, to derive neurons by the antibiotic killing of other lineages.
Lineage selection has been quite successful in producing highly enriched populations of cells, but because relatively few cells within an EB differentiate into the desired lineage, this technique has not been a practical way to derive large numbers of differentiated cells. Lineage selection can be particularly successful when the trapped gene is restricted to a single lineage, and when it is expressed in terminally differentiated populations, rather than being transiently expressed during differentiation. It has also been quite effective in deriving cells producing a neurotransmitter, growth factor or hormone such as insulin [120].
Ultimately, combined approaches will be required to obtain large numbers of highly pure, differentiated cells, such as myelinating oligodendroglia or motor neurons. For example, when the Olig2 promoter was employed to express EGFP, a progenitor population poised to differentiate into spinal cord neurons and oligodendroglia was developed. Despite its name, Olig2 is unfortunately not restricted in expression to the oligodendrocyte lineage, so additional manipulations are required to produce myelinating cells [121]. To obtain motor neurons, ES cells were differentiated as EB, exposed sequentially to retinoic acid to first produce posterior (spinal cord) rather than brain fates, then to sonic hedgehog protein to cause them to adopt a ventral fate. The starting ES population was derived from a transgenic mouse strain in which EGFP is expressed under the control of the HB9 gene (restricted to the motor neurons), so that, following differentiation, a motor neuron (EGFP+) population could be sorted [108]. This combined approach is successful when the marked gene is known to be expressed uniquely in the region/cells of interest and when much is known regarding the region-specific growth factor milieu.
Forced Gene Expression
ES cell lines have also been developed in which expression of a putative developmental control gene can be controlled. A number of approaches, including expression via an episomal construct [122] or insertion into a ROSA locus [123], have allowed consistent expression without the concern of interrupting essential genes. Expression constructs that can be induced, by e.g., doxycycline or tamoxifen, so that varying levels of expression or pulses of gene expression can be attained will be particularly important in analyzing developmental cascades.
Ultimately, combinations of positive and negative selection, growth factor-prompted differentiation ± RNA inhibition to block nondesired lineages may be required to derive cell types of interest. Importantly, introduction of suicide or proapoptotic genes into ES cells [e.g., 124] will allow investigators to remove cells following implantation if they reenter cycle to prevent tumor formation or to remove some of the implanted cells should they overproduce the desired factor.
|
LESSONS FOR THE EMBRYO: NEW GENES, NEW FUNCTIONS |
|---|
|
TOPABSTRACTINTRODUCTIONPLURIPOTENT CELLS OF THE...STEM CELL CHARACTERISTICSLINEAGE SEGREGATION IN THE...DIFFERENTIATION PARADIGMSLESSONS FOR THE EMBRYO:...CONCLUSIONSREFERENCES |
|---|
ES cells are increasingly being used to identify new genes involved in promoting and inhibiting differentiation using widespread mutagenesis [126], gene profiling, and arraying technologies [32, 127]. In combination with rapid gene knockdown using siRNAs [38, 39, 128], rapid assessment of their role(s) in differentiation can be accomplished.
ES cells are also being employed as powerful cell biological models of disease. Nuclear transplantation to create models of human disease, overexpression of abnormal proteins such as ß-amyloid [129], proteins containing expanded trinucleotide repeat (CAG) tracts [130] will be important in both understanding disease progression and providing a platform to identify and test novel pharmacological interventions.
In addition to traditional gene targeting approaches [131], ES cells have been employed to introduce subtle modifications into the nuclear genome, including single base alterations, to introduce mutations into mitochondrial DNA and then the germline [132], to construct transchromosomal mice as models of human genetic disease such as Trisomy 21 [133], to study methylation during X chromosome inactivation [134], providing an important tool to understanding the complex interplay of genes in the differentiation of the early embryo.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONPLURIPOTENT CELLS OF THE...STEM CELL CHARACTERISTICSLINEAGE SEGREGATION IN THE...DIFFERENTIATION PARADIGMSLESSONS FOR THE EMBRYO:...CONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Correspondence. FAX: 734 763 1166; oshea{at}umich.edu
Received: 2 February 2004.
First decision: 26 February 2004.
Accepted: 23 August 2004.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONPLURIPOTENT CELLS OF THE...STEM CELL CHARACTERISTICSLINEAGE SEGREGATION IN THE...DIFFERENTIATION PARADIGMSLESSONS FOR THE EMBRYO:...CONCLUSIONSREFERENCES |
|---|
