Genomes differ systematically in Guanine-plus-Cytosine (GC) content. GC-rich genomes encode proteins enriched in amino acids with GC-rich codons, such as Glycine (with codons GGT, GGC, GGA, GGG), while AT-rich genomes are enriched in amino acids with AT-rich codons, such as Phenylalanine (TTC, TTA).Such a correlation seems utterly natural, but what, precisely, is the evolutionary cause? Some suggest that a GC:AT mutation bias acts through neutral evolution to distort nucleotide composition and, in turn, protein composition. But others argue that so much neutral evolution is unlikely, and that the biases must be due to natural selection acting on genome composition.A possibility that has been ignored, both in this specific case, and in evolutionary theory more generally, is that evolution by natural selection may be biased systematically by arbitrary biases in mutation. In physics, for instance, we would have no trouble conceiving of a particle subject to two distinct forces, each of which imposes some component of direction on the movement of the particle. In evolutionary biology it is much more difficult to develop a comparable understanding, because the multi-dimensional space in which an evolving system moves (genotype space) is not well understood, even though much is known about rules of local movement. Largely for historical reasons, evolutionary biologists tend to assume that selection and mutation must be opposed, and that mutation could not cause an evolutionary trend that affects functional features. Previous theoretical work (Yampolsky and Stoltzfus, 2001) showed that this kind of thinking is misleading, at least in terms of local movement in genotype space. Local movement can be both i) adaptive and ii) mutation-biased, at the same time.The present work addresses the same issue on a larger scale: can mutation bias superimpose a long-term trend on adaptation? Here it is shown that, in a model of protein adaptation on a fitness landscape, a modest GC:AT mutation bias can alter amino acid composition substantially, yielding biases in composition like those observed in natural proteins.This work aims at a fundamental improvement in the interpretive framework of comparative analysis, the approach of making functional inferences by interpreting similarities and differences among species (the comparative approach may be contrasted with the a priori approach, which basically fails in biology for various reasons). Comparative analysis has many practical applications in finding disease-associated genes, identifying drug targets, and so on. In the past, comparative analysis has been based on the idea that selection is the only cause of non-randomness in evolution. Understanding how mutation contributes to non-randomness will increase the accuracy of functional inferences made via comparative analysis.
Citation: Journal of Heredity
Pub Type: Journals