To be or not to be, that is the question faced by cells in an early developing Xenopus embryo. Whether a cell's fate is to become ectoderm, mesoderm or endoderm, is determined by the end of gastrulation. Several zygotically expressed genes are known to have roles in mesoderm and endoderm induction. But so far, the only maternally expressed genes that have been identified show mesoderm-inducing properties. Research has led to the prediction that there might be a gene that prevents the ectodermal cells — fated to become neural or epidermal — from becoming mesoderm. In other words, there seems to be an early acting (maternally expressed) mesoderm inhibitor in the embryo. Until now, there have been no strong candidates. Mesoderm-inhibiting factors, such as Cereberus have only been found to be expressed zygotically at later stages in development. In the journal Development, Bell et al. describe the biological and biochemical characterization of the secreted factor Coco and propose that Coco fulfils the properties of a maternally expressed mesoderm inhibitor.

Xenopus embryos injected with Coco mRNA in one cell of the four-cell stage embryo. The asterisk shows an extra ectopic head. Courtesy of E. Bell, Rockefeller University, New York, USA.

In a developing embryo, cell-fate specification occurs both directly, as expression of genes instructs induction of a cell type, and indirectly, whereby one factor causes inhibition of another factor, thereby preventing the action of the second factor on the induction of specific cell types. Examples include the Nodal gene Vg1, which induce cells to become mesoderm, and Cerberus, which induces neural tissue and ectoderm by blocking Nodal signals and by inhibiting the expression of mesodermal markers (such as Xnr1) and bone morphogenetic proteins (BMPs).

Bell et al. identified Coco in a screen for genes regulated by the TGFβ-inhibitor Smad7, and found that it shows homology to the Cerberus/Dan/Gremlin superfamily of BMP inhibitors. In situ hybridization and RT-PCR analysis showed that Coco is a unique member of this family of proteins as it is maternally expressed and is present in an animal–vegetal gradient in the egg. Complementary to this, the mesoderm-inducers, such as Vg1 and Xnr1, are expressed in a vegetal–animal gradient. As the embryo develops, Coco expression becomes restricted to the animal pole, whereas Vg1 remains in the vegetal pole to induce genes such as brachyury that go on to characterize general mesoderm. So, the expression pattern of Coco has been determined, but what is its function?

In vivo and animal-cap explant injections of Coco mRNA in two-cell stage embryos prevents expression of mesoderm markers brachyury and Fgf8 at the beginning of gastrulation, even when co-injected with BMPs, TGFβs and Wnts. The authors propose that Coco represses these mesoderm inducers by blocking BMP and TGFβ signals in the ectoderm. To elucidate the mechanism behind this effect, they injected Coco mRNA with the Wnt8 and the Bmp4 responsive promoters; co-injected embryos showed complete inhibition of transcriptional activation of both Wnt8 and Bmp4. Coco immunoprecipitated with Bmp4 and Xnr1, further implying that Coco can directly interact with BMP/TGFβ proteins. To test the specificity of this interaction, when Coco was injected in the presence of another signalling molecule, fibroblast growth factor (FGF), it did not prevent mesoderm formation. So, the effect of Coco is not global, but is specific for selected signalling molecules.

The phenotypes of the Xenopus embryos that result from the over-expression of Coco show that it also acts as an ectodermal inducer. Embryos that were injected with Coco mRNA in the animal pole showed extended anterior structures and ectopic cement glands at the tadpole stage. After injection in the vegetal pole, 75% of embryos had extra head-like structures containing forebrain and midbrain tissue, with some even containing a single eye.

Bell et al. have described the biological actions of Coco, together with the mechanisms that might lie behind them. The next step will be to confirm these findings through loss-of-function experiments.