Strong evidence for lineage and sequence specificity of substitution rates and patterns in Drosophila

Abstract

Rates of single nucleotide substitution in Drosophila are highly variable within the genome, and several examples illustrate that evolutionary rates differ among Drosophila species as well. Here, we use a maximum likelihood method to quantify lineage-specific substitutional patterns and apply this method to 4-fold degenerate synonymous sites and introns from more than 8,000 genes aligned in the Drosophila melanogaster group. We find that within species, different classes of sequence evolve at different rates, with long introns evolving most slowly and short introns evolving most rapidly. Relative rates of individual single nucleotide substitutions vary approximately 3-fold among lineages, yielding patterns of substitution that are comparatively less GC-biased in the melanogaster species complex relative to Drosophila yakuba and Drosophila erecta. These results are consistent with a model coupling a mutational shift toward reduced GC content, or a shift in mutation-selection balance, in the D. melanogaster species complex, with variation in selective constraint among different classes of DNA sequence. Finally, base composition of coding and intronic sequences is not at equilibrium with respect to substitutional patterns, which primarily reflects the slow rate of the substitutional process. These results thus support the view that mutational and/or selective processes are labile on an evolutionary timescale and that if the process is indeed selection driven, then the distribution of selective constraint is variable across the genome.

FIG. 1.—

Rates of each of the six complementary pairs of single nucleotide substitutions in the five species of the melanogaster subgroup in (a) 4-fold degenerate synonymous sites on the autosomes, (b) long autosomal introns, (c) short autosomal introns, (d) 4-fold degenerate synonymous sites on the X chromosome, (e) long introns on the X chromosome, and (f) short introns on the X chromosome. Error bars denote standard error due to sampling.

FIG. 2.—

Expected stationary GC content at 4-fold degenerate synonymous sites (leftmost pie chart), long introns (middle pie chart), and short introns (rightmost pie chart) on the (a) autosomes and (b) X chromosome in extant species of melanogaster subgroup as well as at three internal nodes on the subgroup phylogeny. Note that substitutional rates and stationary GC content could not be estimated on the branch leading to Drosophila ananassae and the branch preceding split of the melanogaster species complex and the Drosophila erecta/Drosophila yakuba lineages because these branches are connected to the root.

FIG. 3.—

Total per-lineage substitution rate (per site) in synonymous and intronic sequences in the melanogaster subgroup. Error bars denote standard error due to sampling.

FIG. 4.—

Comparison of observed GC content compared with expected stationary GC content given the stationary distributions of the estimated lineage-specific transition matrices in (a) Drosophila melanogaster, (b) Drosophila sechellia, (c) Drosophila simulans, (d) Drosophila erecta, and (e) Drosophila yakuba. Expected stationary GC content presented here is equivalent to those data presented in figure 2.