Dorsal spine evolution in threespine sticklebacks via a splicing change in MSX2A

Background: Dorsal spine reduction in threespine sticklebacks (Gasterosteus aculeatus) is a classic example of recurrent skeletal evolution in nature. Sticklebacks in marine environments typically have long spines that form part of their skeletal armor. Many derived freshwater populations have evolved shorter spines. Changes in spine length are controlled in part by a quantitative trait locus (QTL) previously mapped to chromosome 4, but the causative gene and mutations underlying the repeated evolution of this interesting skeletal trait have not been identified. Results: Refined mapping of the spine length QTL shows that it lies near the MSX2A transcription factor gene. MSX2A is expressed in developing spines. In F1 marine x freshwater fish, the marine allele is preferentially expressed. Differences in expression can be attributed to splicing regulation. Due to the use of an alternative 5' splice site within the first exon, the freshwater allele produces greater amounts of a shortened, non-functional transcript and makes less of the full-length transcript. Sequence changes in the MSX2A region are shared by many freshwater fish, suggesting that repeated evolution occurs by reuse of a spine-reduction variant. To demonstrate the effect of full-length MSX2A on spine length, we produced transgenic freshwater fish expressing a copy of marine MSX2A. The spines of the transgenic fish were significantly longer on average than those of their non-transgenic siblings, partially reversing the reduced spine lengths that have evolved in freshwater populations. Conclusions: MSX2A is a major gene underlying dorsal spine reduction in freshwater sticklebacks. The gene is linked to a separate gene controlling bony plate loss, helping explain the concerted effects of chromosome 4 on multiple armor-reduction traits. The nature of the molecular changes provides an interesting example of morphological evolution occurring not through a simple amino acid change, nor through a change only in gene expression levels, but through a change in the ratio of splice products encoding both normal and truncated proteins.