DNA mismatch repair (MMR) is a highly conserved biological pathway that plays a key role in maintaining genomic stability. The specificity of MMR is primarily for base-base mismatches and insertion/deletion mispairs generated during DNA replication and recombination. MMR also suppresses homeologous recombination and was recently shown to play a role in DNA damage signaling in eukaryotic cells. Escherichia coli MutS and MutL and their eukaryotic homologs, MutSα and MutLα, respectively, are key players in MMR-associated genome maintenance. Many other protein components that participate in various DNA metabolic pathways, such as PCNA and RPA, are also essential for MMR. Defects in MMR are associated with genome-wide instability, predisposition to certain types of cancer including hereditary non-polyposis colorectal cancer, resistance to certain chemotherapeutic agents, and abnormalities in meiosis and sterility in mammalian systems.

A subset of sporadic colorectal tumors and most tumors developing in hereditary nonpolyposis colorectal cancer patients display frequent alterations in microsatellite sequences. Such tumors have been thought to manifest replication errors (RER+), but the basis for the alterations has remained conjectural. We demonstrate that the mutation rate of (CA)n repeats in RER+ tumor cells is at least 100-fold that in RER- tumor cells and show by in vitro assay that increased mutability of RER+ cells is associated with a profound defect in strand-specific mismatch repair. This deficiency was observed with microsatellite heteroduplexes as well as with heteroduplexes containing single base-base mismatches and affected an early step in the repair pathway. Thus, a true mutator phenotype exists in a subset of tumor cells, the responsible defect is likely to cause transitions and transversions in addition to microsatellite alterations, and a biochemical basis for this phenotype has been identified.

Mutations of DNA mismatch repair genes, including the hMLH1 gene, have been linked to human colon and other cancers in which defective DNA repair is evidenced by the associated instability of DNA microsatellite sequences (MSI). Germ-line hMLH1 mutations are causally associated with inherited MSI colon cancer, and somatic mutations are causally associated with sporadic MSI colon cancer. Previously however, we demonstrated that in many sporadic MSI colon cancers hMLH1 and all other DNA mismatch repair genes are wild type. To investigate this class of tumors further, we examined a group of MSI cancer cell lines, most of which were documented as established from antecedent MSI-positive malignant tumors. In five of six such cases we found that hMLH1 protein was absent, even though hMLH1 -coding sequences were wild type. In each such case, absence of hMLH1 protein was associated with the methylation of the hMLH1 gene promoter. Furthermore, in each case, treatment with the demethylating agent 5-azacytidine induced expression of the absent hMLH1 protein. Moreover, in single cell clones, hMLH1 expression could be turned on, off, and on again by 5-azacytidine exposure, washout, and reexposure. This epigenetic inactivation of hMLH1 additionally accounted for the silencing of both maternal and paternal tumor hMLH1 alleles, both of which could be reactivated by 5-azacytidine. In summary, substantial numbers of human MSI cancers appear to arise by hMLH1 silencing via an epigenetic mechanism that can inactivate both of the hMLH1 alleles. Promoter methylation is intimately associated with this epigenetic silencing mechanism.

A mismatch-binding heterodimer of hMSH2 and a 160-kilodalton polypeptide has been isolated from HeLa cells by virtue of its ability to restore mismatch repair to nuclear extracts of hMSH2-deficient LoVo colorectal tumor cells. This heterodimer, designated hMutSα, also restores mismatch repair to extracts of alkylation-tolerant MT1 lymphoblastoid cells and HCT-15 colorectal tumor cells, which are selectively defective in the repair of base-base and single-nucleotide insertion-deletion mismatches. Because HCT-15 cells appear to be free of hMSH2 mutations, this selective repair defect is likely a result of a deficiency of the hMutSα 160-kilodalton subunit, and mutations in the corresponding gene may confer hypermutability and cancer predisposition.