Introduction: Hereditary non-polyposis colorectal cancer (HNPCC) is one of the most common familial cancer syndrome caused by germline mutations in mismatch repair (MMR) genes, such as hMLH1, hMSH2, hMSH6 and hPMS2. Since germline missense hMLH1 mutations are frequently detected in HNPCC, functional analysis has been needed to predict whether these missense mutations lead to the onset of HNPCC. In the present study, we examined the larger number of hMLH1 mutations than our previous study and other studies, in the yeast MMR assay and the in vitro MMR assay, to predict the pathogenecity of these missense mutations and to elucidate the genotype-phenotype relationships as well as function-structure relationships of the mutants. Methods: We chose 99 missense, one 3-bp deletion and one nonsense hMLH1 mutants found as germline mutations in individuals with HNPCC or sporadic colorectal cancers that were documented mainly in the InSight database or the Swiss-Prot database. These mutations were constructed by using the site-directed mutagenesis with the yeast gap-repair assay. These mutants were examined in the three types of functional assays. First, by the yeast assays using three different reporters respectively, each mutant phenotype was determined to retain the dominant mutator effect of the expressed human MLH1 to interfere the yeast MMR system as reported previously. Second, by in-vitro MMR assay, the ability for the expressed hMLH1 mutant to restore MMR activity of MMR-deficient cell extracts was calculated. Finally, the level of the hMLH1 protein expressed in the human colorectal cancer cell line HCT116 was quantified by densitometry. Results: Among these mutants, 28 mutants lost or reduced both the dominant mutator effect in the yeast assay and the MMR activity in the in-vitro assay, while 35 mutants retained both the dominant mutator effect and the MMR activity. The MMR activity of the wild-type MLH1 was 79.7+7.8 % (average + SD) whereas the MLH1-null was 0 + 1.4%. Each hMLH1 mutant had a different ability to restore MMR activity in in-vitro MMR assay. Although the expression level differed among hMLH1 mutants, no correlation was seen between the MMR function and the protein levels. Many mutations in the ATPase domain and the PMS2-interactive domain, especially around an ATP-binding pocket and a heterodimerization surface, showed the loss of function phenotype. Variants found in families with meeting Amsterdam criteria tended to show loss of function phenotype compared with those found in families without meeting Amsterdam criteria. Discussion: We examined 101 hMLH1 mutations using two functional assays and characterized the pathogenicity of hMLH1 mutations. We propose that this functional evaluation of the larger number of mutations can provide more insights into the function-structure relationships and the genotype-phenotype relationships.

[Proc Amer Assoc Cancer Res, Volume 47, 2006]