Although sequence similarity between genes is a good indicator of their evolutionary relatedness, you wouldn't necessarily expect their encoded proteins to have similar activities — particularly if they belong to different species. But Wang et al. recently found an interesting exception: complete functional overlap between the fruitfly atonal (ato) gene and its mouse homologue, Math1 . Although these two transcription factor genes have only partial sequence similarity and have different functions in the two species, they substitute for each other seamlessly. This work is the first to find an invertebrate gene sequence that can completely rescue the mutant phenotype of its vertebrate counterpart.

The Drosophila gene atonal has a well-characterized function in specifying cell fates in the peripheral nervous system (PNS) and is required for axonal branching in the central nervous system (CNS). By contrast, the mouse homologue Math1 specifies cell identities in both the PNS and the CNS and, oddly enough, in certain gut cell lineages. But just how different are these two genes? Not at all, it seems. The expression of the Math1 coding sequence under the control of the ato enhancer rescues the ato−/− fly embryonic phenotype, which includes missing photoreceptors and stretch receptors. The reciprocal experiment gave an equally clear-cut result: Math1−/− mice that expressed only one copy of ato were indistiguishable from their wild-type littermates. What is surprising about this work is that Ato and Math1, although similar enough to be considered homologous, are only 68% identical in a crucial domain and not at all similar elsewhere.

The differences between Ato and Math1 probably do not arise from new protein domains but from cis-regulatory changes that cause the two genes to be expressed in different tissues. Ato is not expressed in the fly gut, for example, but it substitutes perfectly for Math1 in this tissue in the mouse. Full functional conservation is commonplace between paralogous genes — those derived from gene duplication (see the review on p827 of this issue) — but has never before been documented in genes such as Math1 and ato that are related through speciation.