The Evolution of Mutation Rates: Lessons from X-Linked and Imprinted Genes

Laurence Hurst
Department of Biology and Biochemistry; University of Bath

There exists considerable variation both between organisms and between genes within organisms in the rate of mutation. How can we account for this variation? There are three theories for the evolution of mutation rates: The adaptive hypothesis: selection favors mutation, for, without mutations adaptive evolution could not occur. The constraint hypothesis: selection favors as low a mutation rate as possible but due to physicochernical limitations the rate cannot be reduced further than it is. The trade-off hypothesis: selection favors a low mutation rate, but the rate is determined by a trade-off between the costs of keeping it low and the benefits of having it low. In this lecture, I shall test the third of these hypotheses and show that it is broadly consistent with observed patterns of mutation rate variation. First, it correctly predicts much of the between organism variation in mutation rates: as genome sizes get larger the per base pair mutation rate gets lower, at least in single celled organisms. Second, it is consistent with considerable amount of the within genome variation.

This variation in rates within a genome is typically assessed by the rate of silent site evolution of the genes. According to the neutral theory of evolution the rate of silent/neutral substitutions is the mutation rate. As mutations in hemizygously expressed genes are very much more costly to an organism than masked recessive ones in autosomal genes, the trade-off theory can be extended to predict that genes that are hemizygously expressed should have lower silent site substitution rates than normal autosomal genes. We test this by comparison of orthologous mouse-rat genes.

We report that the silent site rate of evolution of X-linked genes (N=37) is, as predicted, very low, approximately 2/3 that of autosomal genes (N=297). This low rate, additionally, cannot be explained exclusively by a male bias to the mutation rate, nor by selection acting on silent sites (there is, for example no correlation between codon usage bias and substitution rate). Additionally, we report that imprinted genes (N=15) have an equally low rate of silent site evolution. Put another way, the problems that organisms face as a result of imprinting, as regards the exposure of deleterious recessive mutations, appear to have been minimized by the reduction in the rate of such mutations.

The mechanistic basis for these differences is obscure, although we find that removal of methylation induced changes does not alter the conclusions. Putative differences in methylation pattern of the different gene classes do not therefore fully explain the variation in silent site substitution rates. Both of the findings point to the existence of modifiers acting on the mutation rate. But at what level do they act-can they act, for example, on a gene by gene basis? The fact that silent site evolutionary rates are repeatable Within genes (e.g. Igf2r), suggests that some deterministic force is affecting silent site substitution rate at even very low scales. However, our evidence suggests that modifiers are most likely to be acting at a cluster level: imprinted genes in clusters have low rates of silent site evolution but non-clustered ones appear not to. Sample sizes are too low at present to be sure and firm conclusions must wait for farther analysis.