Origin of Somatic Mutations in β-Catenin versus Adenomatous Polyposis Coli in Colon Cancer: Random Mutagenesis in Animal Models versus Nonrandom Mutagenesis in Humans.

Abstract

Wnt signaling is compromised early in the development of human colorectal cancer (CRC) due to truncating nonsense mutations in adenomatous polyposis coli (APC). CRC induced by chemical carcinogens, such as heterocyclic aromatic amines and azoxymethane, in mice also involves dysregulation of Wnt signaling but via activating missense mutations in the β-catenin oncogene despite the fact that genetically modified mice harboring an inactive APC allele efficiently develop CRC. In contrast, activating mutations in β-catenin are rarely observed in human CRC. Dysregulation of the Wnt signaling pathway by the two distinct mechanisms reveals insights into the etiology of human CRC. On the basis of calculations related to DNA adduct levels produced in mouse CRC models using mutagens, and the number of stem cells in the mouse colon, we show that two nonsense mutations required for biallelic disruption of APC are statistically unlikely to produce CRC in experiments using small numbers of mice. We calculate that an activating mutation in one allele near the critical GSK3β phosphorylation site on β-catenin is >105-times more likely to produce CRC by random mutagenesis due to chemicals than inactivating two alleles in APC, yet it does not occur in humans. Therefore, the mutagenesis mechanism in human CRC cannot be random. We explain that nonsense APC mutations predominate in human CRC because of deamination at 5-methylcytosine at CGA and CAG codons, coupled with the number of human colonic stem cells and lifespan. Our analyses, including a comparison of mutation type and age at CRC diagnosis in U.S. and Chinese patients, also indicate that APC mutations in CRC are not due to environmental mutagens that randomly damage DNA.