In mammals, the Cs of most CpG base pairs are methylated and hypermutable, as methyl-C is spontaneously deaminated, producing a T:G mismatch. Consequently, methyl-CpGs have a mutation rate that is 10–50 times greater than Cs in the context of other bases, and CpGs that are not subject to selection are replaced by TpG/CpA and eliminated from the genome. Walser and Furano took advantage of these properties of CpGs and compared the CpG contents of LINE-1 non-long-terminal-repeat retrotransposons in human and chimpanzee genomes. As these retrotransposon sequences are not under selection, 'old' sequences should have a lower CpG content than 'younger' ones.
Intriguingly, previous work has shown that the correlation between CpG content and neutral mutation rate holds for non-CpG-containing sites as well as for CpG sequences. How could this effect be explained? Two main explanations have been put forward: the chromosomal environment could influence the non-CpG mutation rate and CpG content or, alternatively, CpGs could directly affect flanking non-CpG sequences. Walser and Furano found that the correlation between non-CpG mutation rates and CpG content is best fitted by a sigmoidal function, suggesting that a threshold CpG content is needed to affect the non-CpG mutation rate and that a 'saturation point' can also be reached, after which further increases in CpG content have no effect on mutation rate. In addition, they found that the type of mutation is affected by the non-CpG mutation rate — there were changes in the ratio of transition mutations (purine to purine or pyramidine to pyramidine interchanges) to transversion mutations (purine to pyramidine interchanges) that correlated with the changes in non-CpG mutation rate. As different underlying mechanisms produce transitions and transversions, the authors suggest that their findings are more consistent with the hypothesis that CpGs have a direct mutational effect on surrounding non-CpG DNA.
This is a preview of subscription content, access via your institution