Homologues of the Drosophila Caudal, Cdx family of homeodomain proteins (Cdx1, 2, 4), have been established as key regulators of posterior development in all vertebrates. In zebrafish, Xenopus, chick, and mouse embryos, different members of the Cdx family are also expressed in the presumptive spinal cord region at gastrula stages, playing a crucial role in its specification (Beck et al., 1995; Gamer and Wright, 1993; Isaacs et al., 1998; Joly et al., 1992; Marom et al., 1997; Meyer and Gruss, 1993; Morales et al., 1996; Nordstrom et al., 2006; Reece-Hoyes et al., 2002). In Xenopus embryos, overexpression of Cdx4 or Cad-VP16 activator chimeric protein results in a posteriorized phenotype with anterior expansion of posterior structures and a concomitant reduction in anterior ones (Faas and Isaacs, 2009; Isaacs et al., 1998). Cdx4 and Cad-VP16 expanded and induced expression of posterior Hox genes of PGs 6–9 in embryos and AC explants, respectively, while suppressing expression of the anterior Otx2 gene in embryos (Isaacs et al., 1998). Notably, hindbrain formation and expression of PG1 and 3 Hox genes were not affected, suggesting that Cdx activity specifically regulates spinal cord development. In addition, overexpression of Xenopus Cad-VP16 in chick embryos induced anterior ectopic expression of HoxB9 (Bel-Vialar et al., 2002). Cdx function is essential for spinal cord development. Various compound MO knockdowns of different combinations of Cdx1, 2, and 4 in Xenopus and zebrafish have revealed at least partial redundancy between these genes, as they exhibited the same effects but to differing extents (Faas and Isaacs, 2009; Shimizu et al., 2005). These compound knockdowns, similarly to overexpression of the Cad-En antimorph protein (Isaacs et al., 1998), resulted in anteriorized embryos that either lack or have poorly developed posterior and tail structures. The compound loss-of-function of Cdx1, 2, and 4 in Xenopus, and Cdx1a and in zebrafish reduced levels of most spinal cord Hox genes of PGs 5–11, with residual expression being pushed further posterior (Faas and Isaacs, 2009; Shimizu et al., 2005). Regulation of posterior Hox gene expression by Cdx proteins was also demonstrated in mouse embryos, where, HoxA7, HoxB8, and HoxC8 were shown to be direct transcriptional targets of Cdx protein (Charite et al., 1998; Deschamps et al., 1999; Subramanian et al., 1995; Taylor et al., 1997). Therefore, Cdx genes are crucial regulators of spinal cord development that induce expression of the further downstream acting, region-specific, PG5–11 Hox genes.

Upstream to this Cdx-Hox hierarchy, the Wnt/β-catenin pathway regulates Cdx family gene expression. Mouse embryos having the Vestigial tail mutation, a Wnt3a hypomorph, exhibited markedly reduced Cdx1 expression (Figure 6.1A) (Prinos et al., 2001). Zebrafish embryos knocked down for both Wnt3a and Wnt8 ligands, show reduced levels of both Cdx1a and Cdx4 expression, as well as a subsequent reduction in HoxB9a expression (Figure 6.1B) (Shimizu et al., 2005). Application of Frz8CRD protein to chick neural plate explants eliminated early expression of both CdxA and CdxB, ex vivo, and this was also correlated with eliminated sequential expression of later spinal cord Hox genes (Figure 6.1C) (Nordstrom et al., 2006). Similarly, in Xenopus embryos, overexpression of the Wnt antagonist, SFRP3 (see Chapter 9, “Anti-Wnt Anterior Determinants”), eliminated expression of all three Cdx genes (Keenan et al., 2006). Wnt/β-catenin also suffices for activating Cdx expression. Overexpression of Wnt8 or Wnt3a induced ectopic anterior expression of Cdx1 in Xenopus embryos (Keenan et al., 2006). Exogenous Wnt3a also induced ectopic anterior expression of Cdx1a and Cdx1 in zebrafish and mouse embryos, respectively (Prinos et al., 2001; Shimizu et al., 2005). Ex vivo, application of Wnt3a (together with FGF4) to chick neural plate explants strongly induced early expression of both CdxA and CdxB, and in Xenopus tropicalis AC explants, Wnt3a overexpression strongly induced Cdx1 expression (Faas and Isaacs, 2009; Nordstrom et al., 2006). This regulation of Cdx expression by Wnt/β-catenin is direct. In the mouse Cdx1 promoter, within 3.6kb from the transcription start site, four Tcf/Lef sites are found. Transgenic promoter reporter analysis confirmed their importance for promoter function, and EMSA assays in COS cells showed binding of Tcf/β-catenin complexes to all these sites (Lickert and Kemler, 2002; Pilon et al., 2007; Prinos et al., 2001). Similar binding sites were also found in the mouse Cdx2 gene (Wang and Shashikant, 2007). The Xenopus Cdx4 gene also has Tcf/Lef sites in its regulatory elements, in juxtaposition to an FGF response element (which also directly regulates Cdx expression, see below). EMSA analysis showed that these sites were also bound by the Tcf protein, and coexpression of β-catenin boosted reporter expression driven by these fragments in vivo (Haremaki et al., 2003). Thus, Cdx genes are direct targets of the Wnt/β-catenin pathway acting downstream to mediate spinal cord-specific Hox gene expression.

FIGURE 6.1

Wnt/β-catenin activity in spinal cord specification. (A) Loss of Wnt/β-catenin activity suppresses expression of the spinal cord inducing gene, Cdx1, in mice Vestigiall tail (vt) embryos, which is a Wnt3a hypomorph (Prinos et al., 2001). (more...)

Conceptually similar to hindbrain induction, Cdx acts in the core of a gene regulatory network that drives and controls spinal cord formation (Figure 5.1). In Xenopus and mouse, Cdx expression is also under direct regulation of FGF/MAPK signaling, and in these species, as well as in the chick, Wnt and FGF signals synergize to robustly induce their expression (Haremaki et al., 2003; Keenan et al., 2006; Nordstrom et al., 2006; Prinos et al., 2001; Shimizu et al., 2005; Wang and Shashikant, 2007). Furthermore, inhibition of FGF signaling on the background of Wnt3a/Wnt8 MO in zebrafish embryos more severely inhibited Cdx1a expression than Wnt3a/Wnt8 MO alone (Shimizu et al., 2005). In mouse embryos, RA signaling also directly regulates Cdx expression, but this seems to represent a slightly later phase of expression maintenance (Lickert and Kemler, 2002; Pilon et al., 2007). Cad-En antimorph proteins blocked induction of HoxA7 and HoxB9 expression by eFGF in Xenopus AC explants (Isaacs et al., 1998). In chick embryos electroporated with Xenopus Cad-En antimorph protein, FGF2-mediated induction of HoxB9 expression was also inhibited (Bel-Vialar et al., 2002). Finally, lost expression of HoxA9 in Cdx1a/Cdx4 morphant zebrafish embryos could not be restored by either Wnt3a or FGF8 overexpression (Shimizu et al., 2005). These observations suggest that Cdx proteins act downstream of Wnt and FGF signaling, and thus Cdx genes integrate signaling by three morphogens to induce posterior Hox gene expression.

These signaling pathways continue to act further downstream of Cdx as well. HoxA9 expression was also lost in zebrafish embryos in which FGF signaling was inhibited, and interestingly, Cdx1 overexpression could not rescue its expression, suggesting that FGF signaling also acts downstream of, or in parallel with, Cdx (Shimizu et al., 2006). In Xenopus tropicalis embryos, a feedback loop between Cdx and Wnt3a was demonstrated. Overexpression of Cdx1 and Cad-VP16 induced Wnt3a expression in AC explants and dramatically expanded its expression in embryos, respectively, while combined Cdx1/Cdx2/Cdx4 MO knockdown also weakly inhibited embryonic Wnt3a expression (Faas and Isaacs, 2009). In agreement with such a loop, Cdx1 protein was shown to positively regulates its own expression in mouse embryos, and this involves a later Wnt signal (Beland et al., 2004; Prinos et al., 2001). Such a loop may contribute to the maintenance or amplification of the entire regulatory network with its Hox expression output, or alternatively, act locally to fine-tune different individual Hox gene expression patterns and levels in distinct regions of the spinal cord. Playing such a crucial role in the core of the regulatory network, connecting signaling events to Hox expression, it is not surprising that such a loop has evolved that focuses on controlling Cdx family gene expression.