Variation in the intensity of selection on codon bias over time causes contrasting patterns of base composition evolution in Drosophila

Abstract

Four-fold degenerate coding sites form a major component of the genome, and are often used to make inferences about selection and demography, so that understanding their evolution is important. Despite previous efforts, many questions regarding the causes of base composition changes at these sites in Drosophila remain unanswered. To shed further light on this issue, we obtained a new whole-genome polymorphism data set from D. simulans. We analyzed samples from the putatively ancestral range of D. simulans, as well as an existing polymorphism data set from an African population of D. melanogaster. By using D. yakuba as an outgroup, we found clear evidence for selection on 4-fold sites along both lineages over a substantial period, with the intensity of selection increasing with GC content. Based on an explicit model of base composition evolution, we suggest that the observed AT-biased substitution pattern in both lineages is probably due to an ancestral reduction in selection intensity, and is unlikely to be the result of an increase in mutational bias towards AT alone. By using two polymorphism-based methods for estimating selection coefficients over different timescales, we show that the selection intensity on codon usage has been rather stable in D. simulans in the recent past, but the long-term estimates in D. melanogaster are much higher than the short-term ones, indicating a continuing decline in selection intensity, to such an extent that the short-term estimates suggest that selection is only active in the most GC-rich parts of the genome. Finally, we provide evidence for complex evolutionary patterns in the putatively neutral short introns, which cannot be explained by the standard GC-biased gene conversion model. These results reveal a dynamic picture of base composition evolution.

Estimating the parameters of background selection and selective sweeps in Drosophila in the presence of gene conversion

“…This is unexpected, because selection on codon use reduces synonymous site diversity (44), so that a negative relation between Fop and π S would be expected. However, there is little evidence for ongoing selection on codon use at polymorphic synonymous sites in D. melanogaster, in contrast to D. simulans, except for genes with very high codon use bias (45), so that the expected direct effect of selection on codon use on π S in our data is probably negligible. A plausible explanation is that both SSWs and BGS affect π S and Fop in a similar way by reducing the N e for a gene (27,46).…”

Faster‐X evolution: Theory and evidence from Drosophila

“…We first discuss the question of how to obtain data on nearly neutral DNA sequence variants. A recent analysis of selection on GC versus AT variants at third codon positions in a Rwandan population of D. melanogaster, using genomewide polymorphism data from the Drosophila Population Genomics Project (Pool et al., ), suggested that only autosomal genes with a high GC content currently experience selection in favour of GC base pairs (Jackson, Campos, Haddrill, Charlesworth, & Zeng, ); there was no statistically significant evidence for such selection on the X chromosome. (Note that the presence of GC vs. AT base pairs at the 3 rd codon position was used as a proxy for preferred versus unpreferred codons, since D. melanogaster preferred codons end in GC (Akashi, ; Zeng, )).…”

“…However, Jackson et al. () found that short intron (SI) sequences, which are thought to evolve nearly neutrally (Clemente & Vogl, ; Parsch, Novozhilov, Saminadin‐Peter, Wong, & Andolfatto, ), showed statistically significant evidence for a departure from neutrality, favouring GC over AT base pairs on the autosomes in both D. simulans and D. melanogaster , and on the X in D. simulans but not D. melanogaster .…”

“…melanogaster , which is thought to be closely related to the ancestral population of the species (Pool et al., ), the data in Table of Jackson et al. () give X/A ratios of diversity as 1.10 for SI sites and 1.08 for fourfold sites, averaged over regions of the X and A that have nonzero rates of crossing over. This finding is consistent with previous studies that indicated that the mean silent site diversity for the X in African populations of D. melanogaster is similar to, or slightly greater, than that for A (Andolfatto & Przeworski, ; Campos et al., ; Glinka, Ometto, Mousset, Stephan, & De Lorenzo, ; Hutter et al., ; Kauer, Zangerl, Dieringer, & Schloetterer, ; Singh, Macpherson, Jensen, & Petrov, ).…”

“…This departure complicates the interpretation of the results. In contrast, SI sites seem to be approximately in equilibrium with respect to base composition on the D. simulans branch, but not on the D. melanogaster branch (Jackson et al., ).…”