HMSH2 Mutations in Hereditary Nonpolyposis Colorectal Cancer Kindreds1

It has recently been shown that hereditary nonpolyposis colorectal cancer (HNPCC) is caused by hereditable defects in DNA mismatch repair genes. However, the fraction of HNPCC due to defects in any one repair gene and the nature of these mutations are not known. We analyzed 29 HNPCC kindreds for mutations in the prototype DNA mismatch repair gene li.MSIIl by a combination of linkage analysis, polymerase chain reaction-based screening, and sequencing of the coding region. The com plete intron/exon structure of the gene was ascertained to facilitate this analysis. The results suggest that at least 40% of classic HNPCC kindreds are associated with germline mutations in HMSH2 and that most of these mutations produce drastic alterations in the predicted protein product. Introduction HNPCC3 was one of the first familial cancer syndromes described in the biomédicaliterature (1) but its molecular pathogenesis has only recently become clear. HNPCC appears to be due to hereditary defects in DNA mismatch repair (2-7). In prokaryotes and Saccharomyces cerevisiae, the products of the DNA mismatch repair genes mutS and mutL participate in the recognition and correction of mismatched base pairs resulting from replication errors (reviewed in Ref. 8). These genes have been highly conserved through evolution; five human genes homologous to those responsible for mismatch repair in uni cellular organisms have been discovered (2-7, 9, 10). Two of the human genes (hMSH2 and DUG) are related to mutS, while the other three (hMLHl, hPMSl, hPMS2) are homologous to mutL. Mutations in HMSH2 or hMLHl have been identified in selected HNPCC kindreds. Cancers that arise in HNPCC patients are genetically unstable (5, 11). It has been shown that nearly all cancers from HNPCC patients (12-14), as well as 12-18% of sporadic colorectal cancers (15-17) and a variable fraction of other tumor types (18-23), have accumu lated multiple mutations at repeated sequences distributed throughout their genome. Cell extracts from the tumors of HNPCC patients lack biochemically definable DNA mismatch repair capacity in vitro, while normal cells from HNPCC patients are mismatch repair proficient (5). On the basis of the biochemical and genetic analyses, it has been suggested that the wild-type alÃ-eleof the relevant mismatch repair Received 6/6/94: accepted 7/21/94. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1Supported by grants from the Clayton Fund, the Academy of Finland. The Finnish cancer Societies, the Sigrid Juselius Foundation, the Folkhalsan Institute of Genetics, the Nebraska Health Department, the Council for Tobacco Research, and NIH grants CA35494, CA47527, CA57345, CA62924, and CA09326. 2 To whom requests for reprints should be addressed, at Johns Hopkins Oncology Center, 424 North Bond Street, Baltimore, MD 21231. 3 The abbreviations used are: HNPCC. hereditary nonpolyposis colorectal cancer; IVSP, in vitro synthesized protein; PCR, polymerase chain reaction; RT. reverse transcriptasc; cDNA, complementary DNA. gene is inactivated during tumor formation, thereby increasing the mutation rate and accelerating tumor progression (4). This scenario is formally equivalent to that proposed by Knudson (24) to explain tumorigenesis associated with inherited mutations of tumor suppres sor genes. Although it is currently believed that all HNPCC patients have hereditary defects in DNA mismatch repair genes, the proportion of kindreds with mutations in any specific gene is not known. Presum ably, mutations in any of the DNA mismatch repair genes could lead to HNPCC. Information on the nature and number of these mutations is critical for designing effective strategies to detect the mutations in families with HNPCC and to provide appropriate genetic counseling. In this study, we have evaluated a cohort of classically defined HNPCC kindreds to determine the prevalence of mutations in the prototype human DNA mismatch repair gene hMSH2. Our results indicate that at least 40% of classic HNPCC kindreds are caused by mutations in this single gene. Materials and Methods Samples. Lymphocytes were obtained from one or more affected members in each kindred. The member(s) chosen for analysis included at least one who was under 50 years of age when diagnosed with colorectal cancer. Two families were from Finland, one was from New Zealand, and the remainder were from North America. RNA and DNA were purified from the fresh lymphocytes or from Epstein-Barr virus-transformed lymphoblastoid cell lines as described previously (25, 26). Analysis of cDNA. cDNA was generated using random hexamers and RT as described (27). The PCR was used to amplify the HMSH2 transcript in two overlapping fragments. Fragment A contained codons 1-628 and fragment B contained codons 250—934. PCR was performed using 35 cycles of 95°C (30 s), 58°C(1 min) and 70°C(2 min, 30 s) in the buffer described by Sidransky et al. (28). The primers used for RT-PCR included signals for transcription by T7 polymerase and in vitro translation at their 5' ends [codons 1-628 (fragment A): 5'-GGATCCTAATACGACTCACTATAGGGAGACCACCATGGCGGTGCAGCCGAAGG-3'4 and 5'-CCTTTCTCCAAAATGGCTGG-3'; codons 250-934 (fragment B): 5'-GGATCCTAATACGACTCACTATAGGGAGACCACCATGGGAGAGCAGATGAATAGTGCTG-3'4 and S'-GCTTATCAATATTACCTTCATTCCATTACTGGG-S']. Thus, the RT-PCR products could be transcribed and translated in vitro, as described by Powell et al. (27), to search for the presence of mutations which resulted in an altered size of the encoded polypeptide. Controls for RT-PCR included a cDNA sample processed identically except for the omission of RT. Intron-Exon Borders. The sequences of most intron-exon junctions were ascertained by direct sequencing of the PI genomic clone M1015 (4) using primers chosen from the cDNA sequence (4). M1015 contained all hMSH2 exons except exon 1. A PI clone containing the first exon was obtained by screening a human PI library (Genome Systems, Inc.) with a PCR product corresponding to nucleotides 12-272 of the cDNA. This clone was then used 4 These primers have sequences required for transcription and translation at their 5' ends.