In recent years the multipotent extraembryonic endoderm (XEN) stem cells have been the center of much attention. In vivo, XEN cells contribute to the formation of the extraembryonic endoderm, visceral and parietal endoderm and later on, the yolk sac. Recent data have shown that the distinction between embryonic and extraembryonic endoderm is not as strict as previously thought due to the integration, and not the displacement, of the visceral endoderm into the definitive embryonic endoderm. Therefore, cells from the extraembryonic endoderm also contribute to definitive endoderm. Many research groups focused on unraveling the potential and ability of XEN cells to both support differentiation and/or differentiate into endoderm-like tissues as an alternative to embryonic stem (ES) cells. Moreover, the conversion of ES to XEN cells, shown recently without genetic manipulations, uncovers significant and novel molecular mechanisms involved in extraembryonic endoderm and definitive endoderm development. XEN cell lines provide a unique model for an early mammalian lineage that complements the established ES and trophoblast stem cell lines. Through the study of essential genes and signaling requirements for XEN cells in vitro, insights will be gained about the developmental program of the extraembryonic and embryonic endodermal lineage in vivo. This review will provide an overview on the current literature focusing on XEN cells as a model for primitive endoderm and possibly definitive endoderm as well as the potential of using these cells for therapeutic applications.
The mature mouse blastocyst (4.5 days post-coitum (dpc)) consists of three distinct cell types: the trophectoderm, which gives rise to the trophoblast and extraembryonic ectoderm (ExEc), the pluripotent cells of the epiblast, and the primitive or extraembryonic endoderm (ExEn), an epithelial layer of cells on the surface of the epiblast. The primitive endoderm gives rise to: (i) visceral endoderm (VE) that surrounds the epiblast and the ExEc; and (ii) parietal endoderm (PE) that interacts with the trophoblast giant cell layer. PE cells migrate along the inner surface of the trophectoderm and together with trophoblast giant cells form the parietal yolk sac (Hogan et al. 1980). The PE, as well as the VE, mediates nutrient-waste exchange for the developing embryo. Initially the VE overlays only the epiblast, but as the ExEc increases in size the VE quickly expands to also cover the ExEc. The VE overlaying the epiblast becomes molecularly and morphologically distinct from the VE in contact with the ExEc around 5.0 dpc, representing the embryonic VE (emVE) and extraembryonic VE (exVE), respectively. The cells from the exVE are columnar and cuboidal, while the emVE cells are flatter and more epithelial in shape (Takito & Al-Awqati 2004).
Around 5.5 dpc a group of cells at the distal tip of the epiblast differentiates into a morphologically distinguishable subset of emVE, the distal visceral endoderm (DVE) (Rivera-Perez et al. 2003; Srinivas et al. 2004). This marks the formation of the first axis of the body, the distal–proximal axis. Within 4–5 h (between approximately 5.75 and 6.0 dpc) the DVE migrates proximally as a continuous epithelial sheet to the prospective anterior pole of the embryo. The underlying mechanism of this migration is yet to be fully characterized and both active migration and differences in the proliferation rate of the anterior versus posterior epiblast have been suggested (Srinivas et al. 2004; Migeotte et al. 2010; Stuckey et al. 2011; Trichas et al. 2011). The unilateral movement of the DVE changes the distal–proximal axis into the anterior–posterior axis of the embryo and the DVE is now called anterior visceral endoderm (AVE).
The VE and its derivatives, play critical roles in organization and differentiation of the epiblast. The VE is the first site of hematopoiesis (Toles et al. 1989; McGrath & Palis 2005) and induces through the expression of Indian hedgehog and vascular endothelial growth factor the formation of blood islands and endothelial cells (Dyer et al. 2001; Byrd et al. 2002; Damert et al. 2002). In addition, the proximal VE was shown to be involved in early primordial germ cell differentiation (de Sousa Lopes et al. 2004, 2007). Finally, microsurgical removal of AVE resulted in anterior neural structures truncations (Thomas & Beddington 1996) and its derived BMP2 signals have been shown to take part in heart positioning and foregut invagination (Madabhushi & Lacy 2011).
Interestingly, the lineage distinction between embryonic and ExEn tissue was marked by the assumption that the VE that surrounds the epiblast was displaced by the definitive endoderm. Recently it has been shown that cells from the VE persist within the definitive endoderm layer of the embryo and contribute to the early gut tube. This suggests that the distinction between extraembryonic and embryonic tissues is not as strict as believed and the lineage that was previously considered to be exclusively embryonic has extraembryonic contributions (Kwon et al. 2008). An interesting question is whether, within the definitive endoderm and potentially its derivative tissues, there are molecular and functional differences between the ExEn- and epiblast-derived cells that may be studied by in vitro culturing and manipulating cells representative of the ExEn.
Derivation, properties and applications of ExEn-derived stem cells
Stem cells can be derived from each of the primary lineages of the mammalian embryo (Fig. 1). ES cells from the inner cell mass (ICM) or early epiblast, trophoblast stem (TS) cells from the trophectoderm layer and XEN stem cells from the primitive endoderm (or ExEn). Most importantly each one of these stem cell systems are capable of indefinite self-renewal in culture and once reintroduced into the mouse embryo will display lineage restricted contributions in the resulting chimeric embryos that are consistent with their lineage of origin (Beddington & Robertson 1989; Tanaka et al. 1998; Kunath et al. 2005). Interestingly, in vivo, XEN cells can only repopulate the PE, rarely the VE (Kunath et al. 2005; Kruithof-de Julio et al. 2011).
Initially, parietal endoderm cell (PEC) lines were isolated by Fowler et al. (1990). In vitro studies have shown that these cells have characteristics of PE, closely resembling the basement membrane matrix of Reichert's membrane. However, their chimera contribution potential was not assessed (Fowler et al. 1990). PECs morphologically resemble the more recently isolated XEN cells (Kunath et al. 2005; reviewed in Rossant 2007) being round and refractile or stellate or epithelial-like cells. Several methods of isolation have been proposed for XEN cell derivation, which lead all to cells with similar morphological characteristics; however, their response to growth factors seems to be influenced by the derivation process. In this context, it is essential to appreciate that primitive endoderm already exhibits heterogeneous expression of some of the DVE/AVE markers, such as Cer1 and Lefty1 (Takaoka et al. 2006, 2011; Torres-Padilla et al. 2007). Furthermore, Nodal/Activin signaling can be differentially perceived within the primitive endoderm (Granier et al. 2011). This would imply that, at the time of signaling pathway maturation, primitive endoderm cells are sensitive to even minor changes in signaling intensity. Interestingly, primitive endoderm-progenitors expressing Oct4 exhibit greater developmental plasticity than Oct4-expressing epiblast progenitors at a similar stage (Grabarek et al. 2012) which could reflect the in vitro heterogeneous nature of XEN cells.
Extensive microarray analysis on XEN cells performed by several groups all agree in the expression of primitive endoderm markers SOX7, GATA4, GATA6 and the VE markers hHEX and DKK1. Furthermore, XEN cells are characterized by the lack of AFP expression as well as the absence of definitive endoderm markers (Kunath et al. 2005; Brown et al. 2010b; Kruithof-de Julio et al. 2011). Given their reproducible derivation from the ExEn and their expression profile similar to this extraembryonic tissue, XEN cells can be a powerful tool to study inductive effects attributed to the AVE. Brown et al. (2010a,b) have undertaken an extensive array analysis of three cell lines that are similar to the heart-inducing AVE: two embryonal carcinoma-derived (END2 and PYS2) and the XEN cells. By comparing the gene expression profiles they have identified a discrete set of genes that could support myocardial differentiation. In addition, by using XEN cell-derived conditioned media on embryoid bodies they were able to enhance cardiogenesis (Brown et al. 2010a). These are interesting observations with potential therapeutic properties. In order to obtain the appropriate cell products (in this context, cardiomyocytes for cardiac repair), it is of importance to stepwise direct the fate of the stem cell of interest with the proper instructive signals in vitro. During development, the VE interacts with the nascent mesoderm to induce the cardiac fate. Therefore, XEN cells could be used to recapitulate this developmental process in vitro and induce and accelerate differentiation of the susceptible stem cells into the cardiac lineage.
Although derived from primitive endoderm, XEN cells contribute efficiently in chimeras to PE but not to VE (Kunath et al. 2005; Kruithof-de Julio et al. 2011). This lack of contribution to the VE lineage could be caused by many factors, including the preferential interaction with the mural trophectoderm (Artus et al. 2012; Kruithof-de Julio and Shen, unpubl. data, 2009), the alteration of a signaling pathway in the establishment of the cell line based on the derivation process (XEN cells derived in the presence of LIF poorly respond to growth factor stimulation) or the derivation of a “committed” cell that has lost potency. In support of the latter, extraembryonic endoderm precursor (XEN-P) stem cells have been derived from rat blastocyst (Debeb et al. 2009; Galat et al. 2009). These cells are characterized by a less “endoderm” defined gene expression profile; they express the ES markers OCT4, REX1, AP, and SSEA1 and contribute to the PE as well as the VE lineages in chimeras. The authors suggest they are precursor endoderm cells as they could represent the first committed step of the ExEn.
Finally, with respect to the plasticity or commitment of stem cells, recently Cho et al. (2012) was able to derive XEN cells from mouse ES cells via the addition of exogenous retinoic acid and Activin. These XEN cells are indistinguishable from embryo-derived XEN cells, including their differentiation capacity (Cho et al. 2012). These data show a high degree of plasticity within the ICM and provide the possibility of deriving XEN cell lines from the various mutant ES cell lines available, thereby shedding light on the factors required for XEN cell derivation and the development of ExEn.
Signaling pathways in XEN cells
Characteristic of XEN cells are the transcription factors GATA4 and GATA6, specifically expressed in the ExEn. GATA6 is present from 3.5 dpc in the ICM in a salt and pepper distribution with NANOG and then restricted to the primitive endoderm (Chazaud et al. 2006). Gata6-deficient mice are embryonically lethal and Gata6-null ES cells fail to specify VE in vivo and in vitro, suggesting a key role for GATA6 in both VE and PE differentiation (Morrisey et al. 1998). Forced and maintained expression of the Gata4 and Gata6 in mouse ES cells mimic XEN cell characteristics both in vivo and in vitro (Fujikura et al. 2002; Shimosato et al. 2007), implying that GATA factors play a crucial role in XEN cell specification. Recently, an essential stemness factor that has been identified in XEN cells is SALL4. In fact, XEN cells cannot be derived from Sall4-null mice. Given it lies upstream of GATA4 and GATA6, SALL4 seems to play a role as an activator of key lineage-defining genes in the ExEn. This partially explains why loss of Gata4 or Gata6 leads to a VE defect and not to a broader loss of primitive endoderm (Lim et al. 2008).
Interestingly, human ES (hES) cells can also mimic ExEn and definitive endoderm by constitutively expressing SOX7 or SOX17, respectively (Seguin et al. 2008). In addition, differentiation of these cells by either BMP4 treatment in SOX7 overexpressing hES cells increased expression of ExEn markers, whereas Activin A treatment in SOX17 overexpressing hES cells increased expression of definitive endoderm markers as expected for hES cells. Both cell lines, however, maintained expression of both NANOG and OCT4 in their undifferentiated state. This suggests a similarity of the SOX7 overexpressing hES to the XEN-P cells rather than to the mouse XEN cells and implies a role for the SOX transcription factors in maintaining a precursor state in the cells preventing them from differentiation. Sox17 is expressed in mouse-derived XEN cells and plays a role in their establishment since XEN cells cannot be derived from Sox17-null embryos. However, in Sox17 mutant mice no significant effect on ExEn development was observed, which the authors attributed to the highly regulative environment of the mouse embryo and the continuous expression of Gata6 and Sox7 in Sox17-deficient embryos (Niakan et al. 2010).
Endogenous non-coding small RNAs (miRNAs) play a role in the pluripotency regulatory network of ES, TS and XEN cells (Marson et al. 2008; Xu et al. 2009; Spruce et al. 2010). Interestingly, Dicer mutants do not correctly pattern the VE. In addition, depletion of miRNAs in XEN cells leads to a loss of multipotency, which seems to be mediated via the modulation of the ERK1/2 signaling pathway and is contrary to what occurs in embryonic-derived cells. This suggests, for the extraembryonic lineage, a crucial role for miRNAs in establishment and maintenance of self-renewal capability (Spruce et al. 2010). Furthermore, extraembryonic tissues have been reported to be hypomethylated when compared to embryonic tissue (Chapman et al. 1984; Monk et al. 1987; Gardner & Davies 1992). Similarly, XEN cells have been recently shown to express low levels of repressive chromatin modifications, such as H3K27me3 (Rugg-Gunn et al. 2010). Low methylation of ExEn could potentially give this tissue a significant competence to easily undergo differentiation or transdifferentiation.
The lack of XEN cell contribution to VE has interested several groups and lead to the findings that Nodal and BMP4, both members of the transforming growth factor (TGF)β superfamily, treatments direct XEN cells toward a VE phenotype in vivo and in vitro (Kruithof-de Julio et al. 2011; Artus et al. 2012; Paca et al. 2012). Nodal binds to type I (ALK4) or type II receptors in the presence of the co-receptor Cripto or Cryptic (members of the EGF-CFC family). Its signal is propagated to the nucleus via phosphorylation of SMAD2 and is tightly regulated by inhibitors including Lefty and Cerberus. XEN cells, treated with either Nodal or Cripto can be differentiated into a VE phenotype in vitro, and in chimeras they contribute more efficiently to VE. This contribution, however, is not VE exclusive, which could be due to the highly heterogeneous population of XEN cells. In XEN cells, Nodal signals solely via the ALK4 receptor and the EGF-CFC Cryptic, as both SB431542 (as inhibitor of the ALK4 receptor) treated XEN cells and XEN cells derived from Cryptic-null mice do not respond to Nodal. Cripto is, at least partially, Nodal-independent as its function cannot be inhibited by SB431542. The authors propose two possible mechanistic scenarios: an alternative low affinity receptor or a non-canonical pathway through an unknown signal transducer (Kruithof-de Julio et al. 2011). Interestingly, similar conclusions have been independently reached by Clements et al. (2011). The authors focus on the cross talk between the Nodal/Activin signaling pathway and the MAPK p38 in XEN cells revealing a novel role for p38 in regulating Nodal thresholds. They observed that the activation of p38 is not ALK4, 5 or 7 dependent as it cannot be inhibited by SB431542; however, it can be triggered by Cripto. This suggests that the non-canonical pathway by which Cripto functions may be p38 dependent (Clements et al. 2011).
As previously mentioned, BMP4 treatment also directs XEN cells to a VE phenotype. In this case through the canonical BMP receptors, as the chemical inhibitor Dorsomorphin prevented the VE induction mediated by BMP4. Interestingly, the mainly induced VE subtype resembles the VE adjacent to ExEc, and not emVE (Artus et al. 2012; Paca et al. 2012). This observation fits with the known expression of BMP4 in the early embryo, which is highest in the ExEc closest to the proximal epiblast (Lawson et al. 1999). BMP4 was also capable of inducing VE differentiation of PE, showing that this cell type is not terminally differentiated, and retains the ability to form VE (Artus et al. 2012; Paca et al. 2012).
Multipotent stem cells have the potential to develop into different cell types in the body during early life and growth. They are distinguished from other cell types by two important characteristics. First, they can self-renew and second, under certain physiologic or experimental conditions, they can be induced to become tissue- or organ-specific cells with special functions.
In order to fulfill the desire to generate a cell-based therapy, it is necessary to manipulate the cells, in vitro, to successfully differentiate them towards the cell-type of interest. The meticulous application of developmental principals to stem cell culture systems is the basis of this kind of research. The major limitation is the very low frequency of differentiated cells identified and the cellular heterogeneity. Only when large numbers of highly enriched progenitors are accessible, methods can be defined for their maturation and their functional capacity rigorously tested in animal models. XEN cells might give the missing link in this differentiation process.
Extraembryonic and embryonic tissues interact to specify each other's commitment. By using XEN cells as an inductive feeder layer for ES cells, epiblast stem cells (EpiSC) or their multipotent progeny, a particular developmental or differentiation fate can be induced (Kruithof-de Julio, Moerkamp and Goumans, unpubl. data, 2012). In this context, XEN cells may not only be a model to understand ExEn development, but also to elucidate the inductive role of the ExEn in embryonic tissue specification, for example, the formation of heart and blood lineages (Brown et al. 2010a; Artus et al. 2012). Once established, their roles in these inductive processes, mutant lines can be derived to pin point the specific genes involved. Thereby, XEN cells may be a powerful in vitro tool (Fig. 2), by either the use of conditioned medium or co-culture, for providing an overview of the factors that support the differentiation during developmental processes, like cardiogenesis and hematopoiesis. The identification of the exogenous differentiation stimuli may be translated into clinical practice for stem cell-based approaches to efficiently obtain a high number of the differentiated cell type of interest. For cell-based therapies, XEN cells are preferably derived from human embryos. However, the derivation from human embryos was only partly supplemented by the SOX7 overexpressing hES cells and remains subject to future studies. Alternatively, for the identification of developmental stimuli, XEN cells could be derived from the mouse postimplantation epiblast. Mouse EpiSC are highly similar to hES cells in their pluripotent state and growth factor requirements and therefore its derived XEN cells may resemble human ExEn development more closely. Cho et al. (2012) was unable to derive XEN cells from mouse EpiSC by using the same protocol as for the derivation of XEN cells from mouse ES cells (Cho et al. 2012). In the case of EpiSC, due to their epigenetic state, XEN cell derivation may require different exogenous stimuli. However, there remains the possibility that EpiSC are indeed unable to convert their developmental fate. Finally, although it has not been shown to date, XEN cells may be derived from somatic cells via direct reprogramming, thereby circumventing the requirement of human embryos.
Although one must keep in mind the highly heterogeneous population and the existence of species-specific differences, which may be due to the stage and derivation protocols, XEN cells can also be a model to study extraembryonic contribution to definitive endoderm and its differentiated derivatives. This raises the question as to whether XEN cells per se can indeed contribute to the primitive gut tube or must they be prior differentiated towards a definitive endodermal lineage. In later stage chimeras derived from Nodal, Cripto or untreated XEN cells contribution to definitive endoderm has never been observed (Kruithof-de Julio and Shen, unpubl. data, 2010), only the contribution to the visceral yolk sac.
In summary, XEN cells provide a valuable cell culture model to dissect the extraembryonic endodermal lineage and its potential contribution to the embryonic one. Thereby, XEN cells may increase our understanding of the molecular mechanism behind endoderm behavior and function during development. Furthermore, XEN cells may give an insight into the inductive factors underlying mesoderm specification, like cardiogenesis and hematopoiesis, which may be reflected into stem cell-based therapeutic approaches.