Selection againstdam methylation sites in the genomes of DNA of enterobacteriophages

Summary Postreplicative methylation of adenine inEscherichia coli DNA to produce G^6m ATC (where^6mA is 6-methyladenine) has been associated with preferential daughter-strand repair and possibly regulation of replication. An analysis was undertaken to determine if these, or other, as yet unknown roles of GATC, have had an effect on the frequency of GATC inE. coli or bacteriophage DNA. It was first ascertained that the most accurate predictions of GATC frequency were based on the observed frequencies of GAT and ATC, which would be expected since these predictors take into account preferences in codon usage. The predicted frequencies were compared with observed GATC frequencies in all available bacterial and phage nucleotide sequences. The frequency of GATC was close to the predicted frequency in most genes ofE. coli and its RNA bacteriophages and in the genes of nonenteric bacteria and their bacteriophages. However, for DNA enterobacteriophages the observed frequency of DNA enterobacteriophages the observed frequency of GATC was generally significantly lower than predicted when assessed by the chi square test. No elevation in the rate of mutation of^6mA in GATC relative to other bases was found when pairs of DNA sequences from closely related phages or pairs of homologous genes from enterobacteria were compared, nor was any preferred pathway for mutation of^6mA evident in theE. coli DNA bacteriophages. This situation contrasts with that of 5-methylcytosine, which is hypermutable, with a preferred pathway to thymine. Thus, the low level of GATC in enterobacteriophages is probably due not to^6mA hypermutability, but to selection against GATC in order to bypass a GATC-mediated host function.