Implications of sex-specific selection for the genetic basis of disease

Abstract

Mutation and selection are thought to shape the underlying genetic basis of many common human diseases. However, both processes depend on the context in which they occur, such as environment, genetic background, or sex. Sex has widely known effects on phenotypic expression of genotype, but an analysis of how it influences the evolutionary dynamics of disease-causing variants has not yet been explored. We develop a simple population genetic model of disease susceptibility and evaluate it using a biologically plausible empirically based distribution of fitness effects among contributing mutations. The model predicts that alleles under sex-differential selection, including sexually antagonistic alleles, will disproportionately contribute to genetic variation for disease predisposition, thereby generating substantial sexual dimorphism in the genetic architecture of complex (polygenic) diseases. This is because such alleles evolve into higher population frequencies for a given effect size, relative to alleles experiencing equally strong purifying selection in both sexes. Our results provide a theoretical justification for expecting a sexually dimorphic genetic basis for variation in complex traits such as disease. Moreover, they suggest that such dimorphism is interesting - not merely something to control for - because it reflects the action of natural selection in molding the evolution of common disease phenotypes.

Keywords: contemporary evolution; ecological genetics; evolutionary medicine; evolutionary theory; sexual selection.

Figure 1

Equilibrium genetic diversity at a locus that harbors alleles with sex-specific fitness effects. Gray curves follow the frequency of an allele that is costly to sex 1 (each copy of the allele reduces fitness by amount s_i; see the text for details). Black curves depict heterozygosity at the locus. The fitness effect on the other sex (sex 2) is also negative when t_i > 0 (i.e., t_i/s_i > 0). Its effect is positive, and the allele is sexually antagonistic, when t_i < 0 (i.e., t_i/s_i < 0). Results are based on numerical evaluation of the roots of Δq_i = 0 [see eqn (1) and Supporting Information], with u = 10⁻⁶.

Figure 2

Relative contributions of asymmetrically selected alleles to sex-specific fitness variance. The term t/s represents the degree of asymmetry in selection between the sexes, with t/s = 1 representing completely symmetric effects. Each column shows the relative contribution of specified allele classes (a < t/s < b) to the fitness variance in sex 1. The parameter space t/s < 1 reflects the range of interest, where a deleterious allele in our focal sex (sex 1, which suffers a fitness cost of s) is less costly to fitness in the other sex (sex 2). Results are based on simulated data sets (1 000 000 mutations randomly sampled per column), with selection parameters drawn from a bivariate gamma distribution with equal marginals (shape and scale parameters, k and θ, with E(s) = E(t) = kθ = 0.02), and between-sex correlation of r_st (see Supporting Information for details).