The population genetics of modifier alleles

Recently, Weinreich's lab has begun an entirely distinct and novel line of work on the population genetics of modifiers, a diverse class of genes that influence the fidelity with which genetic material elsewhere in the genome is translated into phenotype within individuals, as well as the fidelity with which it is transmitted over space and time. Modifier mutations are to be contrasted with so-called directly selected mutations, those that directly affect their carrier’s lifetime reproductive success such as those in the β-lactamases described here. Unlike these, natural selection acts on modifier mutations only via persistent statistical associations with directly selected mutations, and recently, Weinreich's group and others have discovered a fascinating and largely unappreciated population genetic characteristic of all such mutations. Moreover, he now hypothesizes that this insight may offer a first opportunity to synthesize the population genetics of this diverse and perplexing class of mutations. Beyond our intrinsic, theoretical interest in these effects, modifier mutations are thought to be widespread in many natural microbial populations, including those constituting metazoan microbiota and those responsible for infectious disease, as well as in populations of cancer cells.

This line of work began with an extraordinary discovery of a postdoc, Dr. Eugene Raynes (PhD in biology from the University of Pennsylvania). Eugene’s thesis research examined the evolution of mutators, mutations that increase their carrier’s mutation rate. In asexual populations, mutators cause genetically linked beneficial and deleterious mutations, and a mutator’s evolutionary fate is largely determined by selection acting on these linked, directly selected mutations. Through a combination of experiments using asexual, replicate lab populations of the brewer's yeast Saccharomyces cerevisiae, individual-based computer simulations, and analytic modeling, Eugene showed that mutators are selectively favored in sufficiently large populations, but selectively disfavored below some critical population size. This phenomenon is called sign inversion, and these results are detailed in Raynes et al. 2018.

Sign inversion stands in sharp contrast to the classical understanding of the population-size dependence of selection acting on beneficial and deleterious mutations, wherein selection becomes less efficient as population size declines. But critically, in the classical framing the direction of selection is independent of population size. More recently, Eugene has extended this work to consider mutator evolution in subdivided populations (Raynes et al 2019), and to show that mutators are not under frequency-dependent selection (Raynes and Weinreich in review).

More recently, Professor Weinreich's group realized that theirs was not the only case of sign inversion: Whitlock et al 2016 demonstrated sign inversion for modifiers of recombination, Nowak et al 2004 demonstrated sign inversion for tit-for-tat cooperators, and Gillespie 1974 demonstrated sign inversion for a mutant with elevated variance in offspring number. The implications of these findings are now an area of active research in the group.