Decoupled differentiation of gene expression and coding sequence among Drosophila populations

Owing to the relevance to evolutionary theories of genotypic and phenotypic evolution, the correspondence of differentiation among natural populations in complex phenotypic traits and genetic markers has been studied extensively, and generally found to be poor. In contrast, the correspondence of differentiation among natural populations in gene expression, now often considered a genomic era proxy for the phenotype, and genetic markers, remains largely unexplored. Here, an analysis of expression and nucleotide sequence polymorphism of 106 genes in Drosophila melanogaster strains of the Cosmopolitan (M) and Zimbabwe, Africa (Z) mating races showed that differentiation of gene expression and of coding sequences, measured as Q(ST) and G(ST), respectively, were uncorrelated and, generally, Q(ST) > G(ST). However, an exploratory analysis showed that GST of the 5 prime sequences of genes was correlated with QST calculated from expression data, while GST of the coding sequences remained uncorrelated with Q(ST). This scenario is consistent with the population differentiation at cis-regulatory regions that is decoupled from differentiation of the coding regions. However, despite evidence for selection on global levels of gene expression (deduced from Q(ST) > G(ST)), 5 prime sequence polymorphisms generally were compatible with selective neutrality, suggesting differentiation in cis-regulated gene expression for these genes has been promoted by drift or selection too weak or too long ago to be detected, or higher organizational levels underlying the genetic architecture of expression are targets of selection. In all, this raises the question how selection on the expression changes (i.e. the phenotype) can be so obvious yet elusive at the level of the nucleotide sequence. Our contrasts between genetic differentiation of populations in expression and sequences revealed that even when genotype and phenotype can be connected the sources of variation that are the target of selection remain to be identified.