Lack of MSH2 and MSH6 Characterizes Endometrial but not Colon Carcinomas in Hereditary Nonpolyposis Colorectal Cancer¹

HNPCC³ is characterized by susceptibility to CRC as well as a variety of extracolonic cancers, notably EC. The incidence of EC even exceeds that of CRC (60% and 54%, respectively) in female mutation carriers (1). The syndrome is linked to germline mutations in DNA mismatch repair genes, mainly MLH1 (∼50%) or MSH2 (∼40%; Ref. 2).⁴ In a recent investigation, the presence of EC was found to be an independent predictor of a germline mutation of MSH2 or MLH1 in kindreds with colorectal cancer clustering (3). The risk for EC may vary according to the predisposing mutation. For example, it has been suggested that the lifetime risk of EC is higher in MSH2 mutation carriers as compared with MLH1 mutation carriers (61% versus 42%), whereas the CRC risk is similar (4). Moreover, ∼10% of mutations affect MSH6, and these families often display atypical hyperplastic lesions and carcinomas of the endometrium (5, 6).⁵ The significance of MSH6 in EC development is emphasized further by observations that a majority of epithelial tumors in Msh6-deficient mice originate from the uterus (and the skin) and only rarely from the intestine (7). Sporadic tumors with the MSI(+) phenotype have provided an alternative approach to evaluate the involvement of different MMR genes in colorectal and endometrial tumorigenesis. Thus, immunohistochemical studies have shown that most sporadic MSI(+) colorectal cancers arise from MLH1 inactivation, which is mainly attributable to promoter hypermethylation (8, 9). Whereas MLH1 promoter hypermethylation and the concomitant lack of expression of this protein occurs in some 15% of sporadic endometrial cancers as well (10), we are not aware of any large-scale immunohistochemical analyses evaluating the relative involvement of different MMR proteins in sporadic EC tumors.

To clarify the role of MLH1, MSH2, and MSH6 protein expression as possible factors contributing to the HNPCC tumor spectrum, we used immunohistochemistry to compare the expression patterns of these proteins in ECs and CRCs derived from either the same HNPCC patients or close relatives carrying the same mutations. We provide evidence that the development of EC in HNPCC is selectively associated with the MSH2/MSH6 protein complex deficiency.

The tumors were collected from a series of well-characterized HNPCC families segregating germline mutations in either MLH1 or MSH2 (11, 12). The mutations and their present designations are as follows. (a) MLH1 mutation 1, 3.5-kb genomic deletion affecting codons 578–632 of exon 16 and flanking intron sequences; mutation 2, g→a at 454−1 at the splice acceptor of exon 6; mutation 3, G→C at 1975 (codon 659) of exon 17; mutation 4, g→c at 1409+1 at splice donor of exon 12; mutation 5, T→G at 320 (codon 107) of exon 4; mutation 6, g→a at 1039−1 at the splice acceptor of exon 12; mutation 7, g→t at 1559−1 at the splice acceptor of exon 14; and mutation 8, C→T at 1975 (codon 659) in exon 17; and (b) MSH2 mutation 9, del CA at 1550 (codon 518) of exon 10. The analysis included 42 endometrial cancers (including two cases of atypical hyperplasia, a precursor lesion to endometrial cancer) and 35 colorectal cancers from patients carrying the same predisposing mutations.

Immunohistochemical staining for MLH1, MSH2, and MSH6 was performed as described previously (13). Primary anti-MLH1 antibody (13271A; PharMingen) was incubated with the sections for 1 h, anti-MSH2 antibody (NA26; Calbiochem) for 24 h, and anti-MSH6 antibody (2D4; Serotec; Ref. 14) for 2 h at room temperature at concentrations of 1.25 μg/ml, 3 μg/ml, and 5 μg/ml antibody diluent, respectively. The MLH1 and the MSH6 antibodies are mouse monoclonal antibodies that were prepared with full-length human proteins, whereas the MSH2 antibody is a mouse monoclonal antibody generated with an amino terminal fragment of the human MSH2 protein. Normal mucosa showing strong nuclear staining for all three proteins was used as a positive control.

Fisher’s exact test was used to assess differences between the groups.

In this study, we focused on the expression of MLH1, MSH2, and MSH6 proteins as determinants of the HNPCC tumor spectrum. A particular advantage of the present investigation lies in the fact that it allows a pairwise comparison of colorectal and endometrial cancers from carriers of eight different MLH1 mutations and one MSH2 mutation and includes cases with both cancers originating from the same patients. Detailed results of immunohistochemical analysis are shown in Table 1. As expected, the gene that was mutated in the germline was always inactivated in the respective tumors. Moreover, among MSH2 mutation carriers (mutation 9) all three CRCs as well as two ECs showed loss of MSH6 (complete or partial), whereas MLH1 was always expressed. Remarkably, among MLH1 mutation carriers (mutations 1–8), loss of MSH2 and/or MSH6 protein (complete or partial) occurred in 19 of 37 cases of endometrial cancer (51%) and in 2 of 2 cases of atypical hyperplasia (100%) as compared with none of 32 colorectal cancers (P = 1.3 × 10⁻⁷; two-tailed Fisher’s exact test). Moreover, among eight pairs of endometrial and colorectal cancers, both originating from the same patients, five ECs but none of the CRCs showed deficient MSH2 and/or MSH6 expression (Table 1 and Fig. 1).

The MSH2 and MSH6 proteins together form a heterodimer, MutSα, that is a mismatch recognition factor. In vivo studies in mice (7) as well as in vitro studies in human cells (14, 15) have shown that the MSH6 protein (Msh6 in mouse) is unstable in the absence of its partner MSH2 (Msh2). Furthermore, endometrial tumor cells from MSH6 mutation carriers as well as Msh6^−/− mouse cells have been found to display reduced levels of the MSH2 and Msh2 protein, respectively (7, 16). Our finding that both MSH2 and MSH6 were often affected simultaneously in the present tumor series is in agreement with these previous observations and suggests the impairment of the MutSα complex as a likely mechanism.

The association between MSH2 and MSH6 explains the equal involvement of MSH6 in EC versus CRC from MSH2 germline mutation carriers, whereas the expression patterns in EC and CRC from MLH1 mutation carriers are strikingly different. Our results show that lack of MSH2 and/or MSH6 is associated with the development of endometrial but not colon cancers in carriers of MLH1 mutations, the most frequent cause of HNPCC predisposition (2). The results in endometrial tumors are broadly consistent with a recent study on a smaller series of MLH1 mutation carriers in which only 5 of 11 endometrial carcinomas and hyperplasias retained both MSH2 and MSH6 (16). Interestingly, three of the tumors showed a complete loss of MSH6, whereas only one showed a complete loss of MLH1. There are several possible reasons why, in that study, in contrast to ours, the expression of the MLH1 gene that was mutated in the germline was not unequivocally lost in the respective tumors. These include a different anti-MLH1 antibody and a different immunohistochemical method used, as well as different germline mutations studied (the individual mutations were not specified in Ref. 16). Because of these essential differences, the results in these two studies are only partly comparable.

Mechanisms that in the tumors from MLH1 germline mutation carriers lead to the lack of MSH2, or to the lack of MSH6 even in the presence of its partner MSH2, remain unknown. As discussed above, unsuccessful heterodimerization may result in at least MSH6 degradation. Additionally, inactivation of MLH1 may induce secondary mutations in MSH2 and, as is well established, in a coding mononucleotide repeat within MSH6 in particular (17). However, the present investigation showed a poor correlation between the latter mutations and protein expression.⁶ Alternatively, because sporadic cancers with the MSI(+) phenotype often display promoter hypermethylation affecting MLH1 and several other genes (10, 18), epigenetic silencing of the individual MMR genes in the somatic target tissues remains as a possibility. Considering the significantly lower frequency of MLH1 promoter hypermethylation in tumors from HNPCC patients as compared with sporadic tumors, as reported previously by us (9) and others (19), a methylation mechanism may not be highly likely. Moreover, in contrast to MLH1, in vivo studies of human tumors have shown that MSH2 (19) or MSH6 (20) promoters are not generally prone to hypermethylation. The possibility of epigenetic inactivation of MSH6 cannot be totally excluded, however, because—despite unsuccessful attempts to determine directly the methylation status of the MSH6 promoter—a recent in vitro study provided experimental evidence that this gene can be transcriptionally silenced by cytosine methylation (21).

Taken together, we demonstrate, in a carefully designed comparative analysis, that the lack of MSH2 and/or MSH6 is frequently seen in ECs from MLH1 mutation carriers but is absent in CRCs derived from either the same patients or their close relatives carrying the same mutations. Our findings suggest that the development of the HNPCC-related endometrial tumors is selectively associated with the human MutSα deficiency. Further studies are necessary to establish the exact mechanisms of the expression changes and to explore whether, beyond CRC and EC, similar differences characterize other tumors of the HNPCC spectrum.

We thank Ritva Haider and Saila Saarinen for expert technical assistance and Siv Lindroos for sample collection.