Concept of a Mesendodermal Progenitor Population
It is unclear when the mesoderm and endoderm lineages of the mouse segregate and whether there is an epiblast-derived mesendodermal progenitor that is competent to differentiate into either endoderm or mesoderm but not ectoderm. In Caenorhabditis elegans and sea urchins, a mesendodermal progenitor can be identified at the 4-cell and 16-cell stage, respectively. Activity of signaling cascades leads to allocation of the mesendodermal progenitors and then their differentiation into mesoderm and endoderm (Angerer and Angerer,2000; Davidson et al.,2002; Maduro and Rothman,2002). Single-cell fate mapping in zebrafish at the mid-blastula stage localizes a mesendoderm population distinct from that which solely produces mesoderm (Warga and Nusslein-Volhard,1999). Molecular findings reveal the coexpression of the mesodermal gene Brachyury (T) and the endodermal gene Gata5 in cells that are fated to be mesendoderm, whereas cells expressing Brachyury only are localized to the region that forms mesoderm exclusively (Rodaway et al.,1999).
Evidence for a mesendodermal progenitor in amniotes, and specifically in the mouse, is more limited. Recently, several studies on embryonic stem (ES) cell differentiation in vitro point to the existence of a common mesendodermal precursor. Overexpression of Nodal in mouse embryonic stem cells leads to an up-regulation of both endodermal (Hnf4α, Ihh, Afp, Alb) and mesodermal genes (T, Bmp5, Meox1, Flk1) and a down-regulation of neurectodermal (Sox2, Otx2, Ncam) genes (Pfendler et al.,2005). The differentiation of endoderm (Foxa2, Sox17) occurs in cells derived from the T-positive population, which can also differentiate into mesoderm, but not from the T-negative population, implicating the former as bipotential mesendodermal precursors (Kubo et al.,2004). Of interest, the ability to form endoderm and mesoderm can be influenced by the level of either Nodal or activin, which share a common downstream signaling cascade, with higher levels favoring the formation of endoderm. However, only specific types of mesoderm are induced together with the endoderm (Kubo et al.,2004), raising the possibility that not all mesoderm is formed by means of the mesendoderm route. When induced, Goosecoid–green fluorescent protein (GFP; Gsc-GFP) -positive ES cells express organizer markers (Foxa2, Lhx1, and Chrd) and preferentially form mesoderm (that expresses Vegfr2 and Pdgfrb) and endoderm (expressing Foxa2). Endodermal cells derived from Gsc-GFP+ cells are Gsc+/Sox17+ and do not express Sox7, Hnf4, or Pthr1, suggesting they are unlikely to be visceral endoderm. Under conditions that generate the definitive endoderm, a Gsc+/Sox17− mesodermal population is also generated, further supporting the presence of a mesendodermal precursor in vitro (Yasunaga et al.,2005). However, no Gsc-GFP positive cells have been shown to form endodermal cells only, indicating that the transition through the mesendodermal phase is a prerequisite for endoderm formation in vitro (Tada et al.,2005).
In vivo, lineage analysis of the epiblast cells of the early- and midstreak embryo demonstrates that cell fate is regionalized in the epiblast such that precursors of definitive endoderm are found adjacent to the anterior segment of the primitive streak (Figs. 1A, 2A). Fate maps of the early streak epiblast generated from these experiments show this region to be in the posterior–distal region of the epiblast, and this area lies within the mesoderm forming region (Lawson et al.,1991). However, the clonal descendants of these epiblast cells are not restricted to one germ layer (Lawson et al.,1991). Transplantation studies have confirmed that this population contributes to both anterior mesoderm and endoderm (Kinder et al.,2001) and is characterized by the expression of T, Gsc (Fig. 2B), Foxa2, Eomes, Wnt3, and Mixl1 (Robb and Tam,2004). Taken together, the gene expression and cell fate data highlight the anterior region of the primitive streak as being the location of the putative mesendodermal progenitor in the mouse embryo.
Figure 2. A: Fate map of the midstreak epiblast showing the endoderm forming region (green) and its overlap with the regions forming mesoderm (pink) and ectoderm (blue; Modified from Lawson, 1999). B: Expression of Gsc in the midgastrula organizer and Brachyury in the primitive streak at the midstreak stage. Their expression overlaps at the anterior end of the primitive streak, where the endodermal precursors lie. C: Schematic of Nodal mutants and their impact on the formation of the mesoderm (red) and endoderm (green) showing the correlation of the varying levels of nodal signaling and specification of the endodermal tissues. PrChP, prechordal plate.
Download figure to PowerPoint
Molecular evidence for a mesendodermal progenitor population in the mouse is found in mouse mutants displaying defects in germ layer formation. In the mouse, Nodal signaling is required generally for gastrulation, making it difficult to establish a specific role in mesendoderm formation. Mutants with dramatically curtailed Nodal signaling activity (Zhou et al.,1993; Ding et al.,1998; Weinstein et al.,1998) fail to form a primitive streak and, consequently, lack both mesoderm and endoderm. Hypomorphic mutants with moderate levels of Nodal signaling activity can form some endoderm, whereas a further reduced level leads to a loss of endoderm, although not mesoderm. Of interest, the Nodal dosage influences the type of endoderm and mesoderm that may be formed, with anterior endoderm requiring stronger signaling than posterior endoderm (Lowe et al.,2001; Liu et al.,2004). Elevated Nodal signaling caused by loss of the repressor DRAP1 results in an expanded primitive streak and production of excessive mesoderm (Iratni et al.,2002). These data implicate the level of Nodal signaling in regulating the selective allocation of mesoderm and endoderm (Fig. 2C) and are consistent with, but not sufficient to prove, the presence of a common progenitor for mesoderm and endoderm.
The signaling pathway that generates definitive endoderm in zebrafish and Xenopus involves key factors, including Nodal, Mix-type, Sox, and Gata factors (Stainier,2002; Tam et al.,2003). Although the interactions are much less well defined in the mouse, the phenotypes of mouse mutants suggest that the molecular pathways may be broadly conserved. One mix-like gene, Mixl1, has been identified in mouse, whereas there are three in zebrafish and seven in Xenopus (Pearce and Evans,1999; Robb et al.,2000). Mouse embryos lacking Mixl1 are deficient of definitive endoderm but contain excess T- and Nodal-expressing tissues. Consistent with an impaired endodermal potency, Mixl1-null ES cells are less able to colonize the embryonic gut of chimeric embryos (Hart et al.,2002). Furthermore, promoter analysis suggests that Mixl1 is a downstream target of Nodal signaling (Hart et al.,2005), which may act to curtail Nodal and T activity, thus facilitating endoderm differentiation.
Evidence for the role of wingless-type MMTV integration site (WNT) signaling in endoderm formation in the mouse embryo is less substantial. In embryos that lack β-catenin activity specifically in the endodermal derivatives, cells that are reminiscent of cardiac mesoderm are found among the definitive endoderm. Chimera analysis suggests that β-catenin–deficient cells lose the ability to form endoderm and adopt a mesodermal fate instead (Lickert et al.,2002). This phenotype may represent a WNT-mediated lineage choice by a common mesendodermal progenitor, but the possibility of a fate switch of endodermal cells to mesoderm cannot be excluded. A potential downstream target of β-catenin is Sox17, which plays an essential role in endoderm formation in Xenopus and interacts with β-catenin to regulate the transcription of other endodermal genes (Sinner et al.,2004).
A role for Sox17 as a downstream transcription factor in endoderm formation is partially conserved in vertebrates with the mouse mutant deficient of definitive endoderm and Sox17−/− ES cells excluded from the gut of the chimera (Kanai-Azuma et al.,2002). However, the details differ or remain unclear, for example, there is no mouse ortholog of the sox factor Casanova, which is essential for endoderm formation in zebrafish (Alexander et al.,1999). Similarly, although GATA factors are required for endoderm differentiation in zebrafish and Xenopus, a role in the mouse has not been definitively established, which may be because first, these genes are expressed and required in tissues such as the visceral endoderm and heart as well as the definitive endoderm, and second, they are likely to function redundantly (Molkentin,2000; Stainier,2002; Watt et al.,2004; Zhao et al.,2005).
In the final analysis, although the precursors of the mesoderm and endoderm are colocalized in the epiblast and cells of the two germ layers emerge together during gastrulation and their formation is regulated by common molecular activities, the evidence for a mesendodermal progenitor in the mouse remains inconclusive. Although studies in stem cells are suggestive of a mesendodermal precursor, in vitro studies are artificial and do not replicate the interactions that occur among heterogeneous cell types in the embryo. The spectrum of defects in Nodal-related mutants highlights the temporal influence and gene dosage effects on lineage choices. Thus, if there is a mesendodermal progenitor, its lineage potency might not remain constant but could vary depending on the time of lineage allocation during gastrulation.
Regionalization of Cell Fates and Organ-Forming Potency in the Embryonic Gut
The initial regionalization of the definitive endoderm occurs concurrently with the formation of the definitive endoderm as it exits the primitive streak. The timing of cell migration from the primitive streak influences the final destination in the anterior–posterior body axis, such that cells recruited earlier into the primitive streak will form the foregut, and those recruited later will contribute to gut in progressively more posterior regions (Lawson et al.,1986; Lawson and Pedersen,1987; Tam and Beddington,1992). There is a paucity of pan-tissue markers for the definitive endoderm, and this finding, to some degree, reflects the regionalization of the definitive endoderm from the time of its allocation and/or its functional similarity to the visceral endoderm. Two of the earliest markers are Cerl and Sox17, although neither is restricted to the definitive endoderm nor are they uniformly expressed by the definitive endoderm (Fig. 1B). Cerl expression in the definitive endoderm is transient and restricted to the anterior definitive endoderm (Belo et al.,1997). Sox17 expression is not restricted permanently to any segment of the gut: expression begins in the anterior definitive endoderm (prospective foregut endoderm) and is progressively up-regulated successively in the mid- and hindgut and down-regulated in anterior segments (Kanai-Azuma et al.,2002). The early restriction of gene expression in the definitive endoderm suggests an early regionalization of cell fate along the anterior–posterior length of the embryonic gut.
Further evidence of segmental specialization of cell fates arises from the study of chimeric mice in which Sox17-null cells are incapable of colonizing the mid- and hindgut but do contribute to the foregut, but at a reduced frequency. Similar results were observed for Mixl1 (Hart et al.,2002) and Smad2 (Tremblay et al.,2000). In contrast, Foxa2 (Hnf3β) null cells were found to colonize the hindgut but not the fore- or midgut (Dufort et al.,1998). Again, this finding demonstrates molecular heterogeneity of the definitive endoderm along its anterior–posterior axis. Nodal signaling impacts on cell fate decisions, but the mouse mutants suggest that graded Nodal signaling governs endodermal fate decision as well as segmental contribution to the gut, with higher Nodal signaling specifying more anterior fates and vice versa (Fig. 2C). As a result, the foregut endoderm is more sensitive to a reduction in Nodal signaling than the mid- or hindgut. This finding suggests that initial regionalization of cell fates is mediated by Nodal signaling and occurs concurrently with endoderm specification.
During the morphogenesis of the embryonic gut, the expression of early genes such as Cerberus-like (Cerl) and Lhx1 is down-regulated, whereas other genes such as Hhex and Dickkopf-1 (Dkk1) become restricted to specific regions where organ primordia are formed (Table 1). The definitive endoderm is regionalized through reciprocal interactions with surrounding tissues that delineate sites of organ formation. Many of the genes that display regionalized expression are members of the Hox and ParaHox families and are reputed to confer anterior–posterior identity. However, many Hox genes are expressed in both the endoderm and the adjacent mesoderm and expression in the endoderm is more limited than in the mesoderm, so their specific role in the endoderm is unclear (Beck et al.,2000). Whereas the pattern of gene expression may reflect the underlying regionalization of the embryonic gut, it offers no clues as to the mechanism that establishes this tissue's pattern or the degree of determination of tissue fate. In vitro culture studies show that the molecular identity of the endoderm at the late-streak stage may be influenced by the presence of ectoderm/mesoderm from specific regions along the anterior–posterior axis, which release soluble factors such as fibroblast growth factor-4 (FGF4) to pattern the endoderm (Wells and Melton,2000); nevertheless, the early stages of regionalization are poorly defined. An interaction with surrounding tissues continues to be critical for endoderm differentiation, as fate becomes progressively more restricted, and such interactions govern the emergence of organ primordia (Wells and Melton,1999). By E9.0, several genes are expressed in well-defined restricted domains of the endoderm. Some such as Hhex, Pax1, and Pdx1, which are expressed initially in a regionalized pattern remain actively transcribed during the development of specific visceral organs and, thus, may have later organ-specific roles in addition to a role in the establishment of the organ primordium (Grapin-Botton and Melton,2000).
Table 1. Selection of Genes Expressed in the Endoderm at the Late-Streak and Early-Somite Stage
| ||Gene||Endoderm expression||Other tissues||Reference|
|Late-streak||Dkk1||Anterior-most foregut||Anterior-most AVE||Mukhopadhyay et al.,2001|
| ||Cerl||Anterior endoderm||AVE||Belo et al.,1997|
| ||Sox17||Anterior endoderm||Extraembryonic and anterior visceral endoderm||Kanai-Azuma et al.,2002|
|Early Somite||Irx3||Anterior foregut||Midbrain, hindbrain, and anterior spinal cord||Bosse et al.,1997|
| ||Irx1||Foregut||Midbrain||Bosse et al.,1997|
| ||Pax1||Mediolateral anterior endoderm||Somites||Wallin et al.,1996|
| ||Dkk1||Anterior and ventral foregut and posterior hindgut||Primitive streak and pre-somitic mesenchyme, midbrain neuroepithelium||Mukhopadhyay et al.,2001|
| ||Hex||AIP (liver forming region)||Extraembryonic blood islands||Martinez-Barbera and Beddington,2001|
| ||Hoxb1||Gut caudal to headfolds||Paraxial mesoderm and somites caudal to headfolds and rhombomere four||Huang et al.,1998|
| ||Sox17||Mid- and hindgut endoderm||None||Kanai-Azuma et al.,2002|
| ||Cdx2||Hindgut||Posterior mesoderm, tail bud, caudal neural tube||Beck et al.,1995|