Differential selection within the Drosophila retinal determination network and evidence for functional divergence between paralog pairs

Abstract

The retinal determination (RD) network in Drosophila comprises 14 known nuclear proteins that include DNA-binding proteins, transcriptional coactivators, kinases, and phosphatases. The composition of the network varies considerably throughout the animal kingdom, with the network in several basal insects having fewer members and with vertebrates having potentially significantly higher numbers of RD genes. One important contributing factor for the variation in gene number within the network is gene duplication. For example, 10 members of the RD network in Drosophila are derived from duplication events. Here we present an analysis of the coding regions of the five pairs of duplicate genes from within the RD network of several different Drosophila species. We demonstrate that there is differential selection across the coding regions of all RD genes. Additionally, some of the most significant differences in ratios of non-silent-to-silent site substitutions (d(N)/d(S)) between paralog pairs are found within regions that have no ascribed function. Previous structure/function analyses of several duplicate genes have identified areas within one gene that contain novel activities when compared with its paralog. The evolutionary analysis presented here identifies these same areas in the paralogs as being under high levels of relaxed selection. We suggest that sequence divergence between paralogs and selection signatures can be used as a reasonable predictor of functional changes in rapidly evolving motifs.