Sex-Biased Gene Expression Under Sexually Antagonistic and Sex-Limited Selection

Abstract

Sex differences in gene expression are ubiquitous, evolve quickly, and are expected to underlie phenotypic sexual dimorphism (SD). Despite long-standing interest, the impact of sex-specific selection on the transcriptome remains poorly understood. Here, we test fundamental questions on the role of constraints on gene expression evolution arising from the mode of selection and genetic architecture. We also test the relationship between sex-biased expression and evolved SD. We assess these using body size selection lines in the seed beetle, Callosobruchus maculatus, that have evolved variation in SD in response to either sex-limited (SL) or sexually antagonistic (SA). We find that sex differences in the phenotypic responses and expression changes are generally well aligned. SL selection, despite a phenotypic response similar to SA selection in males, but not in females, resulted in a more extensive expression differentiation and increase of sex-biased expression than SA selection. These patterns show that SA selection imposes a transcriptomic constraint and is not required for sex-bias to evolve. Sex-biased transcripts show lower cross-sex correlations in expression changes than unbiased transcripts, suggesting greater sex differences in their underlying genetic architecture. Although male-biased transcripts are disproportionately affected when selection targeted males, we find no support for a transcriptome-wide association between sex-bias and SD. In the light of these unique experimental insights into how sex-specific selection on size changes adult transcription, our findings have important implications for inferring selection history and mode from patterns of sex-biased gene expression in natural populations.

Keywords: RNA-Seq; experimental evolution; intra-locus sexual conflict; sex-biased expression; sexual selection.

Fig. 1.

Left column: variation in SSD across the selection lines sequenced in this study (top), and the body mass distributions of males and females across the same lines (bottom). Right column: schematic illustrations of the predictions a) of a positive association between the number/proportion of sex-biased (SB) genes and phenotypic SD, b) that ancestrally sex-biased genes are disproportionately affected by SL and SA selection and c) show lower cross-sex correlations in expression evolution compared with unbiased (UB) genes. SA selection should lead to fewer, but more sex-specific expression changes than SL selection. d) Phenotypic and expression changes are associated in terms of the identity of transcripts.

Fig. 2.

The proportion of a) male-biased and b) female-biased transcripts as a function of SD across selection line treatments. Inset text above each panel gives the result of a Spearman rank correlation test for the association between the proportion of transcripts and SD. Inset text next to each point give the absolute number of transcripts in each case. The data are split by chromosome category into autosomes—“A”, the X-chromsome—“X”.

Fig. 3.

logFC expression between SL_m↓ and SL_m↑, or SA and SL_m↑, for males and females. Data are split by whether males show a unique response in the SL_m↓ versus. SL_m↑ contrast, a unique response in the SA versus. SL_m↑ contrast, or a parallel response in both, and by sex-bias status. Points are colored by whether both of the sexes show the same significant expression difference (white) or only one of them (gray). Inset text gives results of Pearson's correlation tests as well as correlation coefficients for each panel. Note that the axis labels are different for the final column of panels. The logFC values for males and females of the parallel transcript class in the SA versus. SL_m↑ contrasts is given in the supplementary materials, Supplementary Material online (supplementary fig. S6, Supplementary Material online).