Mode of action of VegT

The amphibian embryo develops into the three germ layers in a predictable fashion, the pigmented animal pole forming the ectoderm, the yolky vegetal pole forming the endoderm and the mesoderm developing from an annulus at the equator. Mesoderm formation is believed to require a TGF-β family signal originating from vegetal cells because a truncated dominant negative activin receptor, promiscuous in inhibiting TGF-β family receptors, inhibits mesoderm formation in the intact embryo (Kessler and Melton, 1994; Slack, 1994; Hemmati-Brivanlou and Melton, 1992; Schulte-Merker et al., 1994). The main candidate for this signal has been Vg1, since its mRNA, as well as its protein is located in the vegetal pole and its processed ligand induces both endoderm and mesoderm (Kessler and Melton, 1994; Henry et al., 1996). Although the processing of the Vg1 precursor has not been detected in the egg or embryo, experiments with dominant negative Vg1 ligands suggested that Vg1 was essential for dorsal mesoderm formation (Joseph and Melton, 1998). Recently, however, the effects of ablating maternal stores of the T-box transcription factor variously known as Antipodean (Stennard et al., 1996), Xombi (Lustig et al., 1996), VegT (Zhang and King, 1996) and Brat (Horb and Thomsen, 1997) were described (Zhang et al., 1998). Late blastula vegetal masses depleted of VegT mRNA were unable to induce mesoderm in wild-type animal caps, suggesting that the mesoderm-inducing signal is predominantly zygotic, and is a target of VegT (although it is possible that a zygotic component is required for the processing of a maternal signal, or that a maternal signal acts earlier than the late blastula stage). The depleted embryos still form mesoderm, but its appearance is very considerably delayed – perhaps because there is a second slower mesoderminducing pathway, or because 5-10% of maternal VegT remains (Zhang et al., 1998). What are the candidates for a zygotic mesoderm-inducing signal? The experiments of Zhang et al. (1998) imply that it is generated by VegT in the vegetal pole, but to date only one TGF-β family member, Derrière, has been shown to be inducible by VegT (Sun et al., 1999). Derrière is present in the vegetal pole, and induces mesoderm and endoderm in the animal cap assay. A dominant inhibitory version severely reduces trunk and tail mesoderm, but leaves heads relatively unaffected (Sun et al., 1999). A cleaved version of Cerberus protein, Cerberus-short, has been shown to inhibit Xnr1 signalling, but not activin or Vg1, and injection of cer-S mRNA eliminates expression of the panmesodermal marker Xbra in whole embryos, implicating Xnr1 as a major component of the signal (Piccolo et al., 1999). Xnr1 has been shown to be transcribed in the blastula vegetal pole (Jones et al., 1995). This would imply a conserved role for nodal-related proteins, since Nodal is needed to make the node and primitive streak in the mouse (Conlon et al., 1994; Zhou et al., 1993) and a double nodal-related mutant in the zebrafish lacks most mesoderm and endodermal gene expression is disrupted (Feldman et al., 1998). However, the specificity of cer-S against other members of the Xnr family and its effect on Derrière have not yet been ascertained, and Hsu et al. (1998) reported that Cerberus also inhibits activin signalling. Finally, experiments with Xenopus follistatin were taken to indicate that activin is needed to form the ventral mesoderm (Marchant et al., 1998). However, this could be an effect of follistatin on BMP4 and it has also been reported that mammalian follistatin does not block the vegetal pole signal (Slack, 1991). Nevertheless, Dyson and Gurdon (1997) primarily detected 4903 Development 126, 4903-4911 (1999) Printed in Great Britain © The Company of Biologists Limited 1999 DEV1456