The multivariate analysis of unknown significance variants in mismatch repair genes demonstrates their pathogenicity in Lynch Syndrome

The Hereditary Non-Polyposis Colorectal Cancer (HNPCC), also known as Lynch Syndrome (LS), is an autosomal dominantly inherited cancer syndrome that accounts for about 3-5% of all colorectal cancers (CRCs). It is commonly associated with germline mutations in the mismatch repair (MMR) genes. The loss of function of one or more of these proteins results in a significant genomic instability at somatic level, particularly in repetitive DNA sequences (microsatellites) present in many oncogenes and tumor suppressor genes. This mutator phenotype promotes the tumorigenesis process that justifies the earlier age of onset of the disease (approximately 45 years). LS is characterized by high lifetime risk for tumor development, especially CRC, endometrial cancer and other extracolonic tumors. These extra-colonic malignancies include carcinomas of the small intestine, stomach, pancreas, biliary tract, ovarium, upper urinary tract and brain. The MLH1 and MSH2 mutations account for about 40% and 39% of HNPCC cases, respectively, while mutations in minor MMR genes, MSH6, PMS2, MLH3 and MSH3, justify a total of 21% of cases, in which the MSH6 gene gives a greater contribution (11%). The molecular characterization of patients with a clinical diagnosis of Lynch Syndrome relies on the identification of point mutations by DHPLC and direct sequencing, and large rearrangements by MLPA, in the major MMR genes, MLH1 and MSH2. This strategy does not always provide exaustive information for genetic counseling. Indeed, we analyzed 117 families selected according to international diagnostic criteria (Amsterdam Criteria, Lynch HT et al, ICG-HNPCC, Gastroenterology 1999 and Bethesda Guidelines, Umar et al, J Natl Cancer Inst., 2004). Molecular diagnosis was achieved for 53 families while no MLH1/MSH2 mutations were identified in the remaining 64 families. Moreover, we identified genetic variants of uncertain significance (VUS) (missense, intronic and silent variants) in several other patients. Therefore, in this study we have characterized the VUS that we had identified in the major MMR genes, in order to classify them as pathogenic, using the following approaches: segregation analysis, population studies to exclude the polymorphic nature of the variant, assessment of Microsatellite Instability (MSI), gene expression studies both at mRNA and protein levels and in silico analysis with a variety of bioinformatics tools such as HSF (Human splicing Finder), PolyPhen (Polymorphism Phenotyping), SIFT (Sorting Intolerant From Tolerant), and PredictProtein. According to literature data (Colon Cancer Family Registry in 2009, Insight Variant Interpretatioin Committee 2011), a combination of these strategies has been used for each variant in order to assess the pathogenicity of uncertain variants. We performed phenotypic and functional characterization of some of the most interesting variants of uncertain significance identified (three variants in the MLH1 gene and one in the MSH2 gene) that were associated with severe disease phenotypes. In particular, the study of two mutations detected in the 3'untranslated regions (3'UTR) of the MLH1 and MSH2 genes (c*30_32delTTC in the MLH1 gene and c*226A>G in the MSH2 gene) gave very interesting results. The c*30_32delTTC was reported as a benign variant in the MMR gene variants database (Insight-group database). Our results showed lower levels both at RNA and protein levels. These data are in agreement with recent literature studies reported on this variant (Mao G. et al., J Biol Chem, 2008). In particular, the region in which falls the mutation has been identified as a putative target point of the miR-422a and a mechanism of feedback regulation between MLH1 and this miRNA was also demonstrated (Mao G. et al., Cell Res, 2012). In our study, in more aggressive phenotypes, this mutation has always been identified in association with another variant, the c.454-51t> c. Therefore, a likely pathogenetic additive effect between these two genetic alterations could be proposed. The multivariate analysis of the other 3'UTR variant, the c.*226A>G in MSH2 gene, also provided very interesting data in order to assess the correlation with the disease phenotype. In our study, this variant was found associated with both typical Lynch Syndrome features (colorectal and endometrial tumors) and atypical phenotypes such as Hodgkin Lymphoma. This analysis showed increased mRNA and protein levels, as also confirmed by a functional luciferase assay. In order to clarify the molecular mechanism at the basis of this over-expression, in silico analysis was performed for prediction of miRNA target sites (TargetScan and MiRanda) and transcriptional regulation factor binding sites (TRANSFAC). The region in which falls the mutation is identified as a putative target point of two miRNAs (hsa-miR-137, hsa-miR47953p), and four trans-acting protein factors, ZNF333, POU6F1, CDP and PMX1, known also as transcriptional repressors. Therefore, we hypothesized that this variant could prevent the binding of these factors with the MSH2 3'UTR leading to unregulated expression of the MSH2 gene. In agreement with several literature data showing a deleterious effect derived from overproduction of MMR proteins (H. Zhang et al., Canc Res 1999; Shcherbakova PV. et al., Mol. Cell. Biol., 2001 ), it is conceivable that the variant c.*226A> G in the MSH2 gene has a pathogenic role in the development of the disease. In our study we also analyzed the minor MMR genes, MSH6, PMS2, MLH3 and MSH3, for the presence of germline variants in patients who had been resulted to be negative for germline mutations in the major MMR genes. Recently, literature data suggest that the MMR proteins, including the minor MMR factors, may have other functions in cell cycle and apoptosis regulation, in addition to the post-replicative repair role, that could be highly relevant for the carcinogenesis process (Jiricny J. et al. Mol Cell Biol 2006, Ji et al., BMC Med 2012, Mao G., et al., Cell Res, 2012). According to these data, we undertook a mutational analysis on the minor MMR genes and were able to identify 77 variants, five of which were pathogenic mutations, 5 missense variants already described in the literature as pathogenic mutations and 67 unclassified variants (missense, silent and intronic variants). A detailed phenotypic characterization, as described above for the MLH1 and MSH2, was also carried out for some of these variants, in order to classify them as likely pathogenetic. The most relevant result of this mutational analysis on minor MMR genes was the simultaneous presence of multiple molecular alterations in different genes or in a single gene in several unrelated patients. We proposed that some (or all) of these variants could constitute low penetrance alleles, with an additive effect on the risk of the disease (Duraturo F. et al., Int J Cancer 2011). This hypothesis is confirmed by an exhaustive molecular analysis performed for one of the 64 subjects analyzed in this study. Two literature studies on the yeast genome also demonstrated our hypothesis (Martinez SL. et al. PNAS, 2010, Kumar C. et al. Mol Biol, 2011). The probability of a synergistic effect between low-penetrance alleles could also explain the loss of segregation observed for two pathogenetic variants identified in the MSH6 gene, c.3261_62insC and c.3296_97 delTT. These two mutations have been found associated with variants in other minor MMR genes in some of affected members of the families analyzed. Moreover, we can not exclude the concomitant presence of alterations in other genes involved in carcinogenesis but not investigated in this study. In conclusion, this study allowed to clarify the genotype-phenotype correlations in Lynch syndrome, demonstrating the importance of a multifactorial likelihood analytical approach, also in order to understand the molecular mechanisms that regulate the MMR protein and their putative prognostic and therapeutic implications in Lynch Syndrome. Moreover, the simultaneous presence of molecular alterations in several genes (major and minor MMR genes) could suggest an additive effect of these mutations in cancer predisposition, against the classical monogenic transmission and in favor of a polygenic inheritance, although further studies on other familial cases will be needed to confirm this hypothesis