The segments that give rise to the vertebrate spine and muscles of the trunk — the somites — are produced sequentially with each oscillation of the segmentation clock. A new microarray study shows that the molecular mechanism that underlies the clock is a large network of genes, transcribed in two groups that are exactly out of phase with each other.

The mRNA levels of a few genes are known to oscillate with the segmentation clock. Most of these are members of the Notch signalling pathway, although a component of the Wnt pathway, Axin2 , is known to oscillate out of phase with them. Olivier Pourquié and colleagues used microarrays to obtain a more comprehensive list of the molecular components of the clock. They used the expression pattern of one known oscillator, Lfng (lunatic fringe), to select a series of 17 mouse embryos that covered an entire oscillation cycle. Microarray data from these samples confirmed the oscillations of known clock genes, and a statistical algorithm was applied to detect other oscillating transcripts.

The transcripts clustered into two groups that were directly out of phase. The known Notch-pathway components clustered with further Notch components, including the Wnt inhibitor Nkd1 (naked cuticle 1 homologue), and also with fibroblast growth factor (FGF)-pathway components. Analysis of a mutant that is defective in Notch signalling revealed that FGF components oscillate in parallel with Notch rather than being dependent on it. The other cluster consisted mainly of Wnt components and their downstream targets, although only a subset of Wnt targets was involved.

The authors anticipate that technical improvements — in microarray coverage and transcript amplification — will increase the number of identified oscillating targets from their 29 to between 50 and 100. Nevertheless, these results are a significant advance on models that involved only a few genes, and they suggest a model in which Notch/FGF and Wnt components mutually inhibit each other. It will be interesting to see whether the half-lives of the proteins are as short as those of the transcripts, or whether mutual inhibition is required to maintain the oscillations at the protein level.