Gene Duplication, Gene Conversion and Codon Bias

Abstract

Within many genes, synonymous codons are found in very unequal frequencies. This observation of codon bias is often interpreted as a reflection of natural selection acting to mold the codon “choice” to match the frequency of corresponding tRNAs.1–6 While the observation of codon bias is undeniable, the selective interpretation has been questioned. For instance, the fact that many different codons are biased toward the same nucleotide has led to the suggestion that codon bias may simply be a reflection of mutational bias.7,8 On the other hand, arguments against mutational bias are based on the observation that the level of codon bias can vary dramatically between different genes in a single genome⁵and between coding sequences and adjacent non-coding sequences.10 In this paper we consider the possibility that at least some cases of non-random codon usage are due to a bias in DNA repair. Furthermore, we point out that DNA repair enzymes do not affect all genes in a uniform manner. For instance, the effects of DNA repair are especially pronounced in duplicated genes that are undergoing concerted evolution, and such genes do show an extreme bias in the distribution of synonymous codons. In general, competing theories about the causes of codon bias can be tested based on the predictions they make about the patterns of longterm evolutionary trends at the non-silent codon positions.10–11 Consequently, we will discuss the possible relationship between biased codon usage and the amino acid composition of proteins. We conclude that the interaction of biased DNA repair and mutation may influence, not only the difference in codon bias between species, but also the observed differences between genes within a single genome.