A paper featured in last month's Highlights showed the way in which genetics could be used to resolve a longstanding puzzle in signal transduction. This month, genetics provides the puzzle — and (along with some biochemistry) the solution. This is because three papers in Nature explain the counterintuitive observation that an inhibitor of BMP (bone morphogenetic protein) signalling is required for a peak of BMP signalling activity that occurs during development in both invertebrates and vertebrates.

In Drosophila, the dorsoventral gradient of Decapentaplegic (Dpp) — a fly orthologue of vertebrate BMPs — is regulated by several secreted factors, among them Short gastrulation (Sog) and Twisted gastrulation (Tsg). Previous evidence had indicated that both Tsg and Sog act as antagonists of Dpp. However, their mutant phenotypes point to a paradox. A peak in Dpp activity is required to specify a dorsal embryonic structure called the amnioserosa. As Dpp loss-of-function mutants lack this structure, one would expect it to be expanded in the absence of a Dpp antagonist. But surprisingly, Sog and Tsg loss-of-function fly embryos lack the amnioserosa completely.

A solution to this puzzle now emerges in the form of an unusual model for a morphogen gradient, in which Tsg has a dual role. First, Tsg stabilizes the interaction between Sog and Dpp and, thanks to its own diffusability, helps to distribute them throughout the embryo, shaping the gradients of both molecules in doing so. Second, it facilitates Sog's cleavage by a protease called Tolloid (Tld). Near the source of Sog on the embryo's ventral side, Tsg and Tld cannot keep up with the fresh supply of Sog, so here most Dpp is bound in Sog complexes. However, more dorsally, away from the Sog source, Tsg is no longer swamped by high Sog levels and, with the help of Tld, it contributes to Sog cleavage and to the release of active Dpp. These twin functions of Tsg — transport and release of Dpp — are required for the dorsal peak of Dpp activity and for the formation of the amnioserosa.

So what happens in vertebrates? It turns out that vertebrate TSG binds directly to chordin (the vertebrate homologue of Sog) and to BMP4 (Dpp), and increases chordin's affinity for BMP4. Overexpression of tsg mRNA in zebrafish results in a block in BMP signalling, and, as expected, blocking Tsg function results in an expansion of BMP signalling, also seen in chordino (chordin) mutants. Furthermore, ventral injections of BMP antagonists in Xenopus result in an induction of a secondary dorsal axis, although Tsg achieves this only when co-injected with chordin. Tsg's diffusability might be crucial to its function, because its membrane-tethered form alone can induce an extra axis.

The connection between Tsg and Tld that is seen in flies has not been so clearly established in vertebrates, but the cooperative action between Tsg and Sog (chordin) seems highly conserved. This twisting tale provides an interesting example of how a phenotype can lead the unwary astray.