The more complex a pathway, the more the output can be fine-tuned by incoming signals. Unfortunately, this also means that there are more opportunities for things to go wrong, as highlighted by the results of three independent studies led by He, Bienz and Basler.

The Wingless (Wg) — or mammalian Wnt — pathway transduces signals from the cell surface to the nucleus by preventing the glycogen synthase kinase 3 (GSK3)-mediated phosphorylation of Armadillo/β-catenin. When phosphorylated, Armadillo/β-catenin gets degraded. However, in the absence of phosphorylation — that is, in response to Wg or Wnt — Armadillo/β-catenin shuttles to the nucleus to regulate transcription by the T-cell factor, TCF.

Beginning upstream in the Wnt pathway, He and colleagues studied the function of four key amino-terminal β-catenin residues — Ser33, Ser37, Thr41 and Ser45 — which, when phosphorylated, target this protein for degradation by the proteasome–ubiquitin pathway. They showed that phosphorylation of all four residues in a carboxy- to amino-terminal direction was required for β-catenin recognition by β-Trcp, the protein responsible for specifying the demise of phosphorylated β-catenin.

These four residues conform to a consensus glycogen synthase kinase 3β (GSK3β) phosphorylation site, but does GSK3 phosphorylate them all? The authors showed that a separate 'priming' kinase was required to phosphorylate Ser45 before GSK3 could phosphorylate Thr41, Ser37 and Ser33. This 'Ser45 kinase' was identified as casein kinase Iα (CKIα). Depletion of CKIα — but not its close relative CKIɛ — by RNA-mediated interference caused β-catenin to accumulate, which indicates that CKIα is required for Ser45 phosphorylation, the subsequent phosphorylation by GSK3 of Thr41, Ser37 and Ser33 and, ultimately, the degradation of β-catenin. Perhaps it is not surprising, then, that mutations at Ser33, Ser37, Thr41 or Ser45 are associated with colorectal cancer and that CKIα is a candidate tumour suppressor.

Two more genes — legless and pygopus — that function in the Wg/Wnt pathway were identified by Basler's group using genetic screening. Pygopus was also identified independently by Bienz's group. As mutations in either of these genes suppressed the phenotype induced by overexpression of Wg, this indicated that these proteins transduce, rather than produce, the Wg signal. Further genetic analysis showed that Legless and Pygopus function at the bottom of the Wg pathway, downstream of Armadillo/β-catenin. The Legless protein shows three short regions of homology to the mouse and human BCL9 proteins, and evidence that BCL9 is the functional homologue of Legless came from the finding that a BCL9 transgene could rescue the legless-null lethal phenotype.

Legless and Pygopus bind to one another, but Legless was also shown to bind Armadillo/β-catenin. So Pygopus binds to Legless, Legless binds to Armadillo/β-catenin, and Armadillo/β-catenin signalling cannot occur without Legless and Pygopus. What, then, do these two proteins do?

Basler's group showed that Armadillo/β-catenin can bind Legless and TCF simultaneously, hinting that Legless and Pygopus might affect Armadillo/β-catenin-mediated transcription. As Pygopus enhanced TCF-reporter-gene transcription in a tissue-culture assay, this indicates that the principal role of Legless could be to recruit Pygopus to β-catenin in the nucleus, where it influences gene transcription. How it does this is so far unclear, but it is almost certainly mediated by the amino-terminal domain, as the carboxyl terminus is required to bind to Legless. But at least we're moving in the right direction.