Clustered DNA lesion sites as a source of mutations during human colorectal tumourigenesis.

The role of gene mutations in tumourigenesis is well understood, however, the mechanism(s) by which they arise are less clear. Indeed, the common assumption that tumourigenic mutations are the result of DNA replication errors is apparently contradicted by the very low division frequency of the cells from which tumours are thought to arise (i.e. deep stem cells). As a potential solution to this paradox, we tested a model whereby clustered DNA lesion sites (CLS) (where several lesions occur within a few base pairs of each other on opposing strands) could give rise to mutations in quiescent cells. We used statistical analyses to search for sets of dinucleotide sequences (designated target sequences) that are present at and in close proximity to mutation sites in four genes associated with human colorectal tumourigenesis (adenomatosis polyposis coli (APC), v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS), phosphoinositide-3-kinase, catalytic, alpha polypeptide (PIK3CA), and tumour protein p53 (TP53)). The dinucleotides CG, AC-GT, TG, and GC were identified as target sequences in at least three of the genes analysed. Consistent with their designation as target sequences, these dinucleotides have all been identified as high probability sites of oxidative damage formation in in vitro studies. Our results strongly suggest a statistical association between the presence of multiple, clustered target sequences and mutational events. We propose that CLS are a major source of mutations during human tumourigenesis.