Body-methylated genes in Arabidopsis thaliana are functionally important and evolve slowly.

DNA methylation of coding regions, known as gene body methylation, is conserved across eukaryotic lineages. The function of body methylation is not known, but it may either prevent aberrant expression from intragenic promoters or enhance the accuracy of splicing. Given these putative functions, we hypothesized that body-methylated genes would be both longer and more functionally important than unmethylated genes. To test these hypotheses, we reanalyzed single-base resolution bisulfite sequence data from Arabidopsis thaliana to differentiate body-methylated genes from unmethylated genes using a probabilistic approach. Contrasting genic characteristics between the two groups, we found that body-methylated genes tend to be longer and to be more functionally important, as measured by phenotypic effects of insertional mutants and by gene expression, than unmethylated genes. We also found that methylated genes evolve more slowly than unmethylated genes, despite the potential for increased mutation rates in methylated CpG dinucleotides. We propose that slower rates in body-methylated genes are a function of higher selective constraint, lower nucleosome occupancy, and a lower proportion of CpG dinucleotides.