Codon usage bias and effective population sizes on the X chromosome versus the autosomes in Drosophila melanogaster

Abstract

Codon usage bias (CUB) in Drosophila is higher for X-linked genes than for autosomal genes. One possible explanation is that the higher effective recombination rate for genes on the X chromosome compared with the autosomes reduces their susceptibility to Hill-Robertson effects, and thus enhances the efficacy of selection on codon usage. The genome sequence of D. melanogaster was used to test this hypothesis. Contrary to expectation, it was found that, after correcting for the effective recombination rate, CUB remained higher on the X than on the autosomes. In contrast, an analysis of polymorphism data from a Rwandan population showed that mean nucleotide site diversity at 4-fold degenerate sites for genes on the X is approximately three-quarters of the autosomal value after correcting for the effective recombination rate, compared with approximate equality before correction. In addition, these data show that selection for preferred versus unpreferred synonymous variants is stronger on the X than the autosomes, which accounts for the higher CUB of genes on the X chromosome. This difference in the strength of selection does not appear to reflect the effects of dominance of mutations affecting codon usage, differences in gene expression levels between X and autosomes, or differences in mutational bias. Its cause therefore remains unexplained. The stronger selection on CUB on the X chromosome leads to a lower rate of synonymous site divergence compared with the autosomes; this will cause a stronger upward bias for X than A in estimates of the proportion of nonsynonymous mutations fixed by positive selection, for methods based on the McDonald-Kreitman test.

Fig. 1.

Pairwise relationships between several genomic variables. The variables considered are CUB (Fop), effective recombination rate (Rec), CDS length, overall gene expression, and GC content in short introns (GC_I). The relationships between these variables are investigated in four different data sets: oA, autosomal genes in the overlap region; oX, X-linked genes in the overlap region; A, autosomal genes in the full data set which spans the full range of effective recombination rates; and X, X-linked genes in the full data set. We plot the Loess regression lines for each data set and pairwise comparison. We show the Spearman’s rank correlation coefficients and their significance (***P < 0.001; **P < 0.01; *P < 0.05).

Fig. 2.

Effective recombination rate versus 4-fold synonymous diversity (π₄) for the autosomes and 4-fold synonymous diversity multiplied by 4/3 (π₄ corrected) for the X chromosome. Bold lines represent Loess regression lines, in green for the autosomal genes and in red for the X chromosome genes. Dashed lines represent the CIs for the lines. The two vertical lines indicate the lower and upper ends of the overlap region.

Fig. 3.

Frequency spectra at polymorphic synonymous sites for the overlap regions of the X chromosome (oX) and the autosomes (oA).