Epistasis, Pleiotropy, and the Evolution of Modularity

By James Cheverud.



Epistasis occurs when the measured effect of a gene at one locus varies depending on the genotype present at one or more other loci. Epistasis is fully reciprocal and designations of major and modifying loci depend on allele frequencies in the population under consideration not on intrinsic characteristics of the loci themselves. Epistasis has been sought and measured in two different ways, first by using the similarity among various kinds of relatives to obtain a measure of the level of epistatic variance in a population and, second, by directly measuring additive, dominance, and epistatic genotypic values at a locus conditional on genotypes at a second locus. Epistasis is very difficult to detect by measuring the epistatic population variance because special kinds of designs are needed and because only a portion of epistasis' effects on population variance components are included in the epistatic variance term. Genotypic epistasis also produces additive and dominance variance in populations. It is through its effect on additive variance, that epistasis plays an important evolutionary role. Models relating genotypic epistasis to additive, dominance, and epistatic population variance components are presented along with the variance patterns produced by epistasis at different population allele frequencies. Examples of epistasis for obesity in a cross of inbred mouse lines will be presented. One way in which epistasis plays a role in evolution is through its interaction with pleiotropy. Pleiotropy occurs when alleles at a locus affect multiple traits. Pleiotropy is thought to be ubiquitous in that all genes are expected to affect multiple traits. The pleiotropic range is the set of traits affected by variation at that locus. QTL studies of a variety of phenotypic systems, including growth, body composition, and skeletal features, indicate that pleiotropy tends to be modular, with pleiotropic ranges restricted to subsets of functionally and developmentally related traits. The question arises of how the observed pleiotropic ranges become limited so that only subsets of phenotypic features are affected by variation at a given locus. In order for the pleiotropic range of a locus to evolve, it must be genetically variable. One way in which variation in pleiotropic effects can arise is through differential epistasis, when epistatic interactions between loci differ for different phenotypes. Under these conditions, the pleiotropic range displayed by one locus depends on the genotype present at other loci. Genotypic variation at these other loci results in variation in pleiotropic range at the target locus. Classical examples of this phenomenon have been described in Drosophila (abnormal abdomen) and maize (opaque-2).

We have developed a genomic mapping test to locate quantitative trait loci that show variation in their pleiotropic effects because they affect the relationship between traits. These loci are referred to as relationship QTLs (rQTLs) because they affect the relationship between traits. Such loci may or may not have a direct affect on the trait involved. We have mapped rQTLs for mandibular allometry, the relationship between local measures of mandibular morphology and total mandibular length. Such loci are common and found throughout the genome. In the rQTL analysis itself, the loci that it interacts with are not directly specified. However, we can perform a genome scan for loci that interact with the specified rQTL to identify partner loci that, together with the rQTL, have different epistatic interactions for the different traits involved. Thus, we have shown that there is substantial genetic variation in the pleiotropic effects of loci due to differential epistasis. Such variation is necessary for the evolution of modular pleiotropic effects, like those discussed earlier. While it is clear that genetic variation in pleiotropy is fairly common, it is not clear how natural selection would act on this variation to produce modular pleiotropy. It is possible that non-uniform stabilizing selection would favor pleiotropy for traits that interact in their effects on fitness but modularization of trait sets that do not interact in this way. Also, it is possible that correlated directional selection on traits would favor their being affected by common pleiotropic loci with pleiotropy including other traits is disfavored because of potential fitness-reducing correlated responses to selection. These ideas and others need to be considered in future theoretical work.

Other information

The CEES seminar room has a coffee-machine – it is therefore recommended that you come a bit earlier and get yourself a good cup of coffee (for the price of 3 NOK).


Published Feb. 3, 2012 1:35 PM - Last modified Feb. 3, 2012 1:38 PM