The First Lineage Decisions in Mouse and Chick Embryos Are Regulated Differently
In the mouse, Eomes is maternally expressed and thus can be detected in the oocyte and during the first cleavage stages (McConnell et al., 2005). At the late blastocyst stage Eomes expression, under the regulation of Cdx2 (Strumpf et al., 2005), is restricted to the TE and, after implantation, to its derivative, the extraembryonic ectoderm (Ciruna and Rossant, 1999; Russ et al., 2000). In the gastrulating embryo at E6.5, Eomes is also expressed on the posterior side of the embryo—a pattern similar to that seen in Xenopus, zebrafish, and chick—where it later plays a role in mesoderm specification and epithelium-to-mesenchyme transition (Arnold et al., 2008).
Analysis of the early chick embryo by in situ hybridization detected strong expression of Eomes at the earliest stage we could examine, corresponding to the freshly laid egg (EGK-X; Fig. 1A–A″; Eyal-Giladi and Kochav, 1976). At this stage one can distinguish the extraembryonic AO, and the epiblast and hypoblast of the AP. Eomes is expressed in the AO and the hypoblast but is excluded from the epiblast (Fig. 1A′–A″). We begin to detect expression in epiblast in the region of the Köller's sicke as it forms at the posterior side of the embryo at EGK-XI-XII stage (Fig. 1B–B″), as well as in the previous mentioned territories. At Hamburger and Hamilton stage 1 (HH1; Bellairs and Osmond, 2005), the expression in the AO and hypoblast persist (Fig. 1C–C′), but Eomes is also detected in the emerging primitive streak (Fig. 1C″). Expression in the AO and hypoblast decays by HH2, when we detected Eomes mRNA only in the primitive streak (Fig. 1D–D″).
Figure 1. Expression of trophoblast markers in prestreak chick embryos. A–D: Eomes expression is detected from stage EGK-X to the beginning of primitive streak formation. A: At EGK-X (A) Eomes is expressed in the area opaca (A′) and the hypoblast of the area pellucida, whereas it is excluded from the epiblast (A″). B,C: From EGK-XII to HH1 (B,C) the expression is maintained in the area opaca (B′,C′) and hypoblast of the area pellucida (C′), and it also appears in the region of the Köller's sickle (B,B″) and the emerging primitive streak (C,C″). D: With the extension of the primitive streak (HH2, D) Eomes expression disappears from the area opaca (D′) and becomes restricted to the primitive streak (D″). E–H: Cdx2 expression is not detected until EGK-XII (F), when it begins to be weakly expressed in the area opaca (F′; F″ section corresponds to Köller's sickle region). Cdx2 only becomes strong in the area opaca by HH1–HH2 (G–G′,H–H′); at these stages there is no expression in embryonic regions. The darker region in the primitive streak observed in whole-mount at Hamburger and Hamilton stage (HH) 2 (H) does not correspond to true hybridization signal, as is confirmed by the complete lack of signal seen in sections of stage-matched embryos (H″). I–L: Bmp4 and Fgfr2 are not expressed in extraembryonic structures at prestreak stages (I,K) and they are first detected at HH4 (J,L). AO, area opaca; AP, area pellucida; Epi, epiblast; Hypo, hypoblast; KS, Köller's sickle; PS, primitive streak. Horizontal lines indicate the plane of the sections shown in panels below of the same embryos (A–F) or in stage-matched specimens (G,H).
Download figure to PowerPoint
Expression of Eomes in the chick hypoblast might be related to the role of this gene in endoderm specification in both zebrafish and mouse. However, mouse Eomes is not expressed in the PE, the equivalent of the chick hypoblast (O'Farrell et al., 2004), but instead in the anterior primitive streak, from where the definitive endoderm will form (Arnold et al., 2008). Nevertheless, there is a transient expression of Eomes in the visceral endoderm (Ciruna and Rossant, 1999; Kwon and Hadjantonakis, 2007), a PE derivative, that could relate to the expression we observe in the chick hypoblast. In zebrafish, it is the maternal Eomes product that is localized to the marginal blastomeres, where it induces formation of the endoderm (Bjornson et al., 2005). Maternal expression of Eomes has also been reported in mouse (McConnell et al., 2005), although with no spatial restriction. We cannot rule out that the high amounts of Eomes mRNA detected at EGK-X might represent maternal transcripts, because this stage coincides with the transition from maternal to zygotic gene expression in the chick embryo (Zagris et al., 1998; Elis et al., 2008). Further studies will be required to determine whether a role for Eomes in endoderm specification is conserved in the chick embryo and if this is related to its expression in the hypoblast. In this regard, it is interesting to note that recent results show that, contrary to established views, the extraembryonic endoderm contributes to the definitive endoderm in the mouse (Kwon et al., 2008).
The early expression of chick Eomes in the AO suggests that the similarities between extraembryonic tissues in mouse and chick may extend beyond mere morphological features. We, therefore, examined the expression of orthologues of other genes that are TE-specific in mouse. Cdx2 is the first gene known to be specifically expressed in the mouse blastocyst and has been shown to be upstream of Eomes during TE specification. Cdx2-deficient embryos manifest an earlier TE phenotype than Eomes-deficient embryos, and while Eomes expression is lost in Cdx2 knockouts, Cdx2 expression is intact in Eomes knockouts (Strumpf et al., 2005; Ralston and Rossant, 2008). However, in the early chick embryo, and contrary to what we see for Eomes, we did not detect Cdx2 expression at stage EGK-X (Fig. 1E–E″), and only by EGK-XII we began to detect weak expression in the AO (Fig. 1F–F″). By HH1 Cdx2 expression was found only in the AO with no expression in the nascent primitive streak (Fig. 1G–G′). Finally, strong expression of Cdx2 restricted to the AO was seen at HH2, when Eomes expression is clearly down-regulated in this domain (Fig. 1H–H″). Apparent expression in the primitive streak at this stage is due to background, as no staining is observed in sections. At later stages Cdx2 expression is restricted to the caudal part of the primitive streak (data not shown), in a pattern identical to that previously reported (Marom et al., 1997). According to these results, Eomes and Cdx2 expression domains only overlap during a restricted period of time in the extraembryonic region of the chick embryo at pregastrulation stages. By HH1, when Cdx2 begins to be expressed in the AO, Eomes expression is about to decay in this territory. This differs markedly from what happens in the mouse blastocyst, and makes it difficult to position Cdx2 upstream of Eomes in the chick embryo. These results thus argue against an overall conservation in amniotes of the gene regulatory networks that control extraembryonic fate.
Given that Cdx2 is absent from the early extraembryonic lineages of the chick embryo, we examined whether the same was also the case for other mouse TE-specific genes. Bmp4 and Fgfr2 are expressed in the mouse TE and its derivative the extraembryonic ectoderm (Haffner-Krausz et al., 1999; Lawson et al., 1999; Rossant and Cross, 2001). These factors form part of the trophoblast-epiblast crosstalk network: Fgfr2 expressed on trophoblast cells binds Fgf4 from the epiblast, activating signals that maintain the trophoblast (reviewed in Rossant and Cross, 2001), while trophoblast-expressed Bmp4 maintains the patterning of the epiblast (Fujiwara et al., 2002; Di-Gregorio et al., 2007). Whole-mount in situ hybridization in early chick embryos revealed that neither Bmp4 (Fig. 1I,J) nor Fgfr2 (Fig. 1K,L) are expressed until stage HH4, confirming previous reports (Chapman et al., 2002; Lunn et al., 2007): these factors show no expression whatsoever in chick extraembryonic domains. Similarly, we found no expression in pregastrulation stages of Tead4 (data not shown), a gene that in mouse has been found to be upstream of Cdx2 in specification of the trophoectoderm (Yagi et al., 2007; Nishioka et al., 2008), in response to the Hippo signaling pathway (Nishioka et al., 2009).
In both chick and mouse, expression of Eomes is initially restricted to extraembryonic lineages, what could indicate a similar regulation of the first lineage decisions in avians and mammals. However, other factors involved in the specification and maintenance of mammalian TE are not expressed in a comparable manner in the early chick extraembryonic domains, and the later onset of Cdx2 expression in chick appears to indicate that Eomes expression in extraembryonic lineages is not regulated in the same manner as in mouse. Moreover, the absence of Bmp4 and Fgfr2 suggests that the extraembryonic–epiblast communication network is a novel acquisition in mammals.
Eomes Is Expressed in the Primordial Germ Cells of the Chick but Not the Mouse Embryo
Bulfone et al. (1999) have previously described the expression pattern of chick Eomes from stages HH3 to HH28. According to this report, at HH3 Eomes is expressed in the anterior-most part of the AP, in the hypoblast, and ectoderm anterior and lateral to the PS. This expression decays by HH5 with the regression of the PS, and disappears by HH6. Later in development, at HH25, Eomes is expressed in the telencephalic pallium of the developing brain. This pattern, with expression in the gastrulating embryo and later in the central nervous system, is widely conserved in Xenopus, zebrafish, and mouse (Bulfone et al., 1999; Ciruna and Rossant, 1999; Kimura et al., 1999; Mione et al., 2001; Bachy et al., 2002). In embryos from HH19 to HH22, we detected Eomes expression in the previously described territories, such as the telencephalon (Fig. 2D,F), but also detected Eomes mRNA in a novel and nonconserved domain: the PGC (Fig. 2E,G).
Figure 2. Eomes is expressed in primordial germ cells in the chick. A–C: At HH4, Eomes expression is detected in the anterior half of the embryo (A) and in scattered cells (primordial germ cells [PGC]) in the boundary between the anterior area opaca and area pellucida (A′: magnification of the square in A). By Hamburger and Hamilton stage (HH) 4+ (B) the expression in the epiblast is restricted to the PGC and the node (B′: magnification of the square in B; arrowhead indicates PGC). With the regression of the primitive streak (HH5, C) epiblast expression is seen only around the node and in the head fold, and the signal in the PGC along the germinal crescent is very apparent (C′: magnification of the square in C). D–F: At later stages Eomes is expressed in the forebrain of the chick (black arrow in D; F) and along the genital ridges (red arrow in D; E: magnification of the chick in D). G: The PGC expressing Eomes along the genital ridge of a HH22 chick are shown in this section. H–J: In sections of mouse embryos, Eomes is detected in a conserved pattern in the forebrain of the E10.5 embryo (black arrow in H) but not in the genital ridges (red arrow in H; I: magnification of the embryo in H) or in the gonads at E13.5 (J). K,L: The PGC marker Pou5f1 is clearly detected in these territories (K, expression in PGC along the genital ridge of the E10.5 embryo; L, expression in E13.5 gonads). Ovaries are shown on the left and testis on the right (J, L).
Download figure to PowerPoint
Eomes expression in PGC becomes apparent by HH4 in the germinal crescent, located at the anterior boundary of the AO and AP (Fig. 2A,A′). PGC are large cells that can be found dispersed in the germinal crescent and have a very characteristic rounded shape, making them easily distinguishable from other cell populations (Bellairs and Osmond, 2005). At HH4+, Eomes expression in the primitive streak decreases and becomes restricted to the node (Fig. 2B), and by HH5 expression is detected anterior to the node along the regressing primitive streak and in the head process and forming head folds (Fig. 2C). At these stages, the PGC in the germinal crescent remain positive for Eomes (Fig. 2B′,C′). Starting at HH10, PGC migrate from the germinal crescent by means of the vascular system toward the gonads that they reach by HH16 (Bellairs and Osmond, 2005). We detected Eomes expression along the genital ridges in whole-mount embryos from HH19 to HH24 (Fig. 2D,E; and data not shown), and restriction of this expression to PGC was confirmed in tissue sections of HH22 embryos (Fig. 2G).
Expression of Eomes in PGC has not been reported in any other vertebrate. We, therefore, re-examined the expression pattern of Eomes in the mouse embryo, focusing on the genital ridges and gonads. PGC are specified in mouse at around E7.0, in the region where the mesoderm is being determined (reviewed in Matsui and Okamura, 2005). At this time, Eomes is highly expressed in mesoderm, making it difficult to ascertain if it is expressed in the PGC. At E10.5, we detected Eomes in the previously described domain in the developing brain, but found no expression in the genital ridges (Fig. 2H,I). Conversely, at this stage, we found clear expression of Pou5f1, a well characterized PGC marker (Fig. 2K; Scholer, 1991). Similarly, no expression of Eomes was detected in the gonads of E13.5 embryos (Fig. 2J), contrasting with the gonadal expression of Pouf51 at this stage (Fig. 2L). The lack of expression of mouse Eomes in gonads was confirmed by reverse transcriptase-polymerase chain reaction (RT-PCR; data not shown). PGC, therefore, represent a novel and nonconserved expression domain of Eomes in chick.
Several genes are expressed in developing PGC in chick, such as Vasa, Dead end, Nanog, and the Pou5f1-related gene PouV (Tsunekawa et al., 2000; Stebler et al., 2004; Canon et al., 2006; Lavial et al., 2007). Of special interest are Nanog and Pou5f1, because of their pivotal role in specifying the early lineages of the mouse blastocyst and in maintaining embryonic stem cell pluripotency within the epiblast at this stage (Boiani and Scholer, 2005; Niwa, 2007). This places them as repressors of extraembryonic fate in mouse, contrary to Eomes, which contributes to TE formation and maintains trophoblast stem cells. Furthermore, mouse Oct4 and Nanog bind to genomic regions in the vicinity of Eomes, and are thought to directly repress its transcription (Loh et al., 2006). Accordingly, Nanog and Pou5f1, but not Eomes, are expressed in PGC in mouse (Scholer, 1991; Hatano et al., 2005; Yamaguchi et al., 2005). In contrast, in chick, we find Eomes expressed together with Nanog and PouV in this stem cell population. The role of Eomes in PGC and its relation to Nanog and PouV in the chick remain to be elucidated.
Conserved Elements in the Genomic Region of the Mouse and Chick Eomes Genes Are Unable to Drive TE-Specific Reporter Expression in Blastocysts
Given the common expression of Eomes in extraembryonic domains of the mouse blastocyst and the prestreak chick embryo, we reasoned that these domains might be controlled by conserved regulatory mechanisms. We, therefore, examined whether noncoding regions surrounding the mouse Eomes gene and conserved with chick were able to direct restricted reporter expression in the TE of transgenic mouse blastocysts.
We first compared the Eomes genomic regions in human, mouse, opossum, and chick, using the gene annotations provided by the Ensembl Genome Browser server (www.ensembl.org) (Fig. 3A). Human EOMES is located on chromosome 3, flanked by two intergenic regions of 519 kb (upstream) and 260 kb (downstream). Murine Eomes is located on chromosome 9, opossum Eomes on chromosome 8, and the chick Eomes orthologue on chromosome 2. There is conserved synteny among the upstream regions of all species. However, the downstream region found in human, opossum, and chick is absent in mouse. This is likely due to an intra-chromosomal rearrangement, resulting in the region downstream of mouse Eomes being syntenic with a region located 10 Mb upstream of human EOMES. The fact that the downstream region of mouse Eomes has suffered a rearrangement, suggesting a major reduction in the intergenic noncoding region, strongly indicates that most Eomes regulatory elements are likely to be located in the upstream region. Recently, a BAC transgenic mouse strain has been described (Tg(Eomes::GFP)) that recapitulates most aspects of Eomes expression in the mouse. Green fluorescent protein (GFP) expression is detected in the TE of the late blastocyst and in the extraembryonic ectoderm at early streak stages, but is not maintained in the chorion at late streak stages (Kwon and Hadjantonakis, 2007). This bacterial artificial chromosome (BAC) construct covers ∼186 kb upstream and ∼18 kb downstream of Eomes, confirming that most if not all regulatory elements are located upstream of Eomes, and for that reason we restricted our analysis to that region (Fig. 3A,B). Furthermore, a homozygous translocation involving a breakpoint upstream of EOMES has been found in humans (Baala et al., 2007), which maps inside the region covered by the BAC construct described above (Fig. 3B). This translocation results in microcephaly and other neuroanatomical defects, but obviously does not affect the earlier role of EOMES in TE development, what would argue that all the regulatory information required for proper expression of EOMES at preimplantation stages has not been disconnected from the gene by this translocation.
Figure 3. Analysis of Eomes regulation in the mouse blastocyst. A: The genomic region containing Eomes is conserved between mammals and chick with the exception of the region downstream of mouse Eomes, which has undergone rearrangement and lost synteny with respect to other vertebrates. The region covered by the bacterial artificial chromosome (BAC) used to create the mouse transgenic line Tg(Eomes::GFP), which recapitulates Eomes expression (Kwon and Hadjantonakis, 2007), is indicated. B: This region was used in a global multiple alignment analysis, with the mouse as base organism, to identify elements conserved in all species examined (highly conserved elements [HCE]). The five HCE identified (Eo1 to Eo5) are highlighted blue. The position of the breakpoint involved in a homozygous translocation described in humans that results in late neurodevelopmental defects (Baala et al., 2007) is shown by a vertical green line. C: Eo1 to Eo5 were tested for their enhancer activity in blastocysts, and none was able to drive expression of lacZ to a level comparable to that of the previously described Pou5f1 enhancer (Pou5f1DE). Scattered dots were detected inside cells of HCE transgenic and control blastocysts, in a pattern like that shown for Eo1. The table shows the percentage of lacZ positive blastocysts obtained for each construct. D: Conversely we could detect enhancer activity for Eo5, that drives expression of the transgene to the forebrain (arrowhead), an evolutionarily conserved domain of Eomes expression among vertebrates.
Download figure to PowerPoint
Multiple-alignment of sequences from diverse organisms corresponding to the same genomic region is a powerful tool for identifying cis-regulatory elements (Boffelli et al., 2004; de la Calle-Mustienes et al., 2005). This approach is based on the assumption that a high level of conservation in a noncoding region must be due to a functional constraint. We performed multiple-alignment analysis of the noncoding regions flanking Eomes in mouse, human, opossum, and chick, and selected highly conserved elements (HCE) present in all four species. Five HCE (Eo1 to Eo5) were selected for testing of their enhancer activity in a mouse transient transgenic assay (Fig. 3B). None of the HCE tested was able to drive expression of the reporter gene lacZ in blastocysts to the level induced by a previously described Pou5f1 enhancer element, used as positive control (Fig. 3C; Yeom et al., 1996). Instead, all the HCE drove low-level punctuated expression indistinguishable from a negative control (Fig. 3C; see Experimental Procedures). It is interesting that HCE Eo5 lies within one of the regions detected in a genome-wide chromatin immunoprecipitation assay with p300 designed to identify forebrain enhancers with extremely high accuracy (Visel et al., 2009). When we tested this HCE for its enhancer activity in transgenic embryos at E10.5, we could detect specific expression of the reporter gene lacZ in the telencephalon (Fig. 3D). This element also drove expression of the reporter consistently in the midbrain, although this is not a site of endogenous Eomes expression in mouse or chick. This suggests that elements conserved between mouse and chick do indeed control conserved aspects of Eomes expression, such as in the forebrain.
The results of our transgenic analysis suggest that the enhancer elements responsible for Eomes expression in the TE of the mouse blastocyst are not located in regions highly conserved in chick. This supports the idea that Eomes is regulated differently in the extraembryonic tissues of mouse and chick embryos. Nevertheless, we cannot rule out the possibility that conserved binding sites might exist in less conserved regions that fall below the threshold used in our analysis, or that other conserved regions located outside of the 200-kb region analyzed would drive expression in extraembryonic tissues.