Abstract: The evolutionary suppression of synonymous codon sites is a controversial topic. Although some studies have indicated that synonymous substitution is under positive selection, most of these studies relied on comparison of homologous genes and/or a limited number of sequences. In the present work, we compared the selection strength at synonymous sites for two types of protein-encoding genes: genes encoding enzymes and protein genes encoding non-enzymes. Our method does not require assumptions concerning, for example, evolutionary equilibrium or population size. We compared ∼70 000 genes from the fully sequenced mammalian Homo sapiens, Pan troglodytes, Mus musculus, and Rattus norvegicus genomes and found that the percentage of C and G in the third position of a codon positively correlates with the percentage of the G/C content within ± 20 000 nucleotides of the gene. More interestingly, we found that synonymous sites in mammalian genes encoding enzymes have undergone stronger selection than did such sites in genes encoding proteins that are not enzymes.

Key words: codon bias, GC content, mammal, selection, synonymous substitution