Adaptive evolution of non-coding DNA in Drosophila

Abstract

Time to junk the term ‘junk DNA’, or to reserve it for DNA of proven uselessness. Geneticists favour the less judgmental term ‘non-coding DNA’ for those parts of the genome not translated into protein, and there is growing evidence that it is important in disease, development and evolution. Despite this, little is known about the evolutionary forces acting on it. Now a new population genetics approach shows that most non-coding DNA in Drosophila melanogaster is subject to adaptive evolution and selection. The big surprise comes from a comparison between Drosophila species: a significant fraction of the divergence between species in non-coding DNA is adaptive, driven by positive selection. In fact, the number of beneficial substitutions in non-coding DNA is an order of magnitude larger than in proteins. Non-coding DNA includes ‘cis-acting’ regulatory sequences, so this finding may reflect the immense importance of regulatory evolution, previously suggested on intuitive grounds. A large fraction of eukaryotic genomes consists of DNA that is not translated into protein sequence, and little is known about its functional significance. Here I show that several classes of non-coding DNA in Drosophila are evolving considerably slower than synonymous sites, and yet show an excess of between-species divergence relative to polymorphism when compared with synonymous sites. The former is a hallmark of selective constraint, but the latter is a signature of adaptive evolution, resembling general patterns of protein evolution in Drosophila1,2. I estimate that about 40–70% of nucleotides in intergenic regions, untranslated portions of mature mRNAs (UTRs) and most intronic DNA are evolutionarily constrained relative to synonymous sites. However, I also use an extension to the McDonald–Kreitman test3 to show that a substantial fraction of the nucleotide divergence in these regions was driven to fixation by positive selection (about 20% for most intronic and intergenic DNA, and 60% for UTRs). On the basis of these observations, I suggest that a large fraction of the non-translated genome is functionally important and subject to both purifying selection and adaptive evolution. These results imply that, although positive selection is clearly an important facet of protein evolution, adaptive changes to non-coding DNA might have been considerably more common in the evolution of D. melanogaster.