This guest post is by Ted Morrow, Jessica Abbott, and Will Gilks on their review paper Gilks et al. “The evolution of sex differences in disease genetics”
Our paper forms part of a research project (2Sexes_1Genome, 2012-16) devoted to investigating how sex-specific and sexually antagonistic selection influences the genome, and in particular whether genetic variants that are maintained as a result of these forms of selection could contribute to disease risk. We had three main aims with our paper, which we outline below together with a motivation for each.
Our first aim was to summarise evidence for sex-dependent genetic architecture in complex traits that were otherwise shared between the sexes. We focused particularly on disease phenotypes in humans, although a range of complex traits from diverse taxa were considered. The motivation for this was to establish a baseline for how widespread or rare sex-specific genetic architecture is. An important paper in this respect, published in Nature Reviews Genetics (Ober et al., 2008) specifically addressed the question of sex-specific genetic architecture in human diseases. It reviewed selected examples within the human disease genetics literature for sex-specific effects on a range of phenotypes. They concluded that studies where sex was ignored would miss some important variants that contribute to disease risk. While the Ober et al. (2008) paper makes a robust case for investigating sex as a factor in genetic analyses, several other genome-wide association studies in the primary literature have been published since, suggesting that an up to date review of these would be worthwhile. We did not intend to conduct a full-scale meta-analysis, although that would probably be a very informative exercise given potential problems in terms of reporting bias, non-independence of traits, and selection of traits with known sexual dimorphism. Nonetheless, a clear pattern emerges of widespread evidence of sex-specific genetic architecture based on heritability estimates (see Figure 1 in our paper), eQTLs, gene manipulations, expression studies, and SNPs with sex-by-genotype effects (see Table 1 in our paper). A recently published paper (not included in our review) even reports 10 out of 13 loci reaching genome-wide significance for recombination rate having sex-specific effects (Kong et al., 2013).
The second aim was to show how evolutionary theory could provide ultimate explanations for the origins of sex-specific genetic architecture. In this way, we propose that a deeper understanding of why genes cause disease, and why some common diseases show sexually dimorphic expression, may emerge. The evolutionary theory of why the sexes may differ phenotypically goes back to Darwin’s observations (1871) of how selection acts in males and females. He characterized males as active competitors, engaging in physical battles with rivals or investing in costly signals with which to woo potential mates. Females, on the other hand were characterized as being coy and choosy. There is now good evidence that mate choice is something not only limited to females, and that sexual selection also operates well after copulation (i.e. sperm competition and cryptic female choice). The key point is that fundamental differences between the sexes occur in terms of investment in reproduction, and as a consequence the routes by which males and females may maximize their fitness are often different. In other words, both natural and sexual selection frequently take sex-specific forms in terms of strength and/or direction. The latter possibility that selection acts antagonistically between the sexes is well established in several laboratory and wild populations, including humans. From a human disease perspective, disease may occur as a result of an individual’s phenotypic difference (or departure) from an optimal phenotype (where a particular trait value has the greatest fitness). This difference could be the result of a genetic constraint imposed by an intersexual genetic correlation for that trait, or indirectly (i.e. pleiotropically) though genetic correlations with other traits. Sex-specific or sexually antagonistic selection could therefore maintain genetic variation within a population that is either less favourable or actually deleterious for one sex. A recent model (Morrow & Connallon, 2013) shows how alleles with sex-specific or sexually antagonistic effects will contribute more to genetic variation for disease predisposition than alleles that are deleterious to both sexes in equal measure, and achieve higher allele frequencies. As a result, sexual dimorphism in the genetic architecture of complex polygenic diseases would emerge within the population. This evolutionary model clearly indicates that the search for loci contributing to disease risk in humans would benefit from exploring sex-specific genetic effects.
The final aim was to provide readers with an overview of the analytical options available for detecting sex-specific associations in genome-wide studies of complex diseases and phenotypes. As we show, more studies are investigating and discovering sex-dependent effects using GWAS data, Common strategies are to separate or stratify the samples within case and control groups by sex, or to model sex as a covariate. The first approach reduces the statistical power to detect sex-dependent effects, and thus only strong ones will be detected. The second simply controls for any sex-specific effects, it is not intended to identify them. We instead advocate the inclusion of a genotype-by-sex interaction term in statistical models, available as an option in some of the commonly used analytical platforms such as GenABEL and PLINK.
Overall, we hope our article raises the profile of sex-specific genetic effects, a topic that is already apparently receiving increasing interest judging by the recent crop of sex-specific associations appearing in the GWAS literature. This forms a more general theme within the field of human disease genetics, of exploring the impact of interaction effects, such as genotype-by-environment interactions. The identification of strong main effects has had successes but the debate over the ‘missing heritability’ of complex traits has activated researchers to look beyond to more complex processes such as epistasis and environmental effects. We welcome any comments either here on Haldane’s Sieve or in the comments section of biorXiv where are article is currently posted.
2Sexes_1Genome. 2012-16. Edward H. Morrow. FP7 ERC Starting Grant – Evolutionary, population and environmental biology. http://www.2020-horizon.com/2SEXES-1GENOME-Sex-specific-genetic-effects-on-fitness-and-human-disease(2SEXES-1GENOME)-s2903.html
Darwin, C. 1871. The Descent of Man. Prometheus Books, New York.
Kong, A., Thorleifsson, G., Frigge, M.L., Masson, G., Gudbjartsson, D.F., Villemoes, R., et al. 2013. Common and low-frequency variants associated with genome-wide recombination rate. Nat. Genet. doi:10.1038/ng.2833.
Morrow, E.H. & Connallon, T. 2013. Implications of sex-specific selection for the genetic basis of disease. Evol. Appl. doi:10.1111/eva.12097.
Ober, C., Loisel, D.A. & Gilad, Y. 2008. Sex-specific genetic architecture of human disease. Nat Rev Genet 9: 911–922.