GENETICS AND FUNCTION OF DNA MISMATCH REPAIR

The overall goal of our research is to uncover the molecular genetic alterations involved in the development of hereditary and sporadic human cancer. Our main objectives are to characterize the genetic changes in cancer cells, and to evaluate normal gene function and pathogenesis of disease. Ultimately, knowledge about the molecular basis of tumor development will facilitate the establishment of more effective ways to prevent, diagnose and treat cancer.

Recently, we have turned our attention to the study of hereditary colorectal cancer (CRC) and its connection to defective DNA mismatch repair. CRC is the second leading cause of death from cancer in Western countries. Every year, approximately 130,000 new cases are diagnosed and 50,000 CRC-related deaths occur in the United States. A conspicuous fraction of these cases (5-15%) are the consequence of a genetic predisposition. The main familial CRC syndrome is hereditary non-polyposis CRC (HNPCC), or Lynch Syndrome. HNPCC is an autosomal dominant disorder characterized by early onset colorectal tumors that are frequently multiple, and may be associated with extracolonic malignancies such as cancers of the endometrium, stomach, ovary, brain, skin and urinary tract.

Patients affected by HNPCC carry a germline mutation in genes involved in DNA mismatch repair (MMR), a specialized system that corrects base-base mismatches, short insertions/deletions and recombination-derived heteroduplexes. DNA MMR contributes to mutational avoidance and genetic stability, thus performing a tumor suppressor function. Tumors from HNPCC patients harbor a genome-wide DNA replication/repair defect, the hallmark of which is length instability of microsatellite repeat sequences. A subset of sporadic colonic and extracolonic cancers also exhibit microsatellite instability (MSI).

Human MMR genes (MSH [mutS homolog] 2, MLH [mutL homolog] 1, MSH3, GTBP [G/T mismatch-binding protein] / MSH6, PMS [postmeiotic segregation] 2 and PMS1) encode homologues of the E. coli proteins MutS and MutL, which belong to the methyl-directed, post-replicative MMR system. In this system, the mismatch is detected by MutS; after interaction with MutL, the single-strand endonuclease, MutH, is activated and incises the DNA strand carrying the mutation. In the bacterial system, MutH has the pivotal role of identifying the newly synthesized strand, i.e. the strand carrying the mutation. MutH identifies and cleaves the new strand by virtue of its transient lack of adenine methylation at GATC sites. Despite its critical function, eukaryotic homologues of MutH, i.e., eukaryotic MMR endonucleases, have not been identified. Furthermore, the molecular determinants of strand discrimination in eukaryotic cells, which lack GATC methylation, have remained elusive. Thus, it is necessary to identify novel eukaryotic MMR genes that may be involved in the process of strand discrimination. Analysis of these genes may further elucidate the genetic basis of HNPCC.

We attempted to identify proteins that interact with known MMR components. By analogy to the MutL-MutH interaction in the bacterial system, a eukaryotic MMR endonuclease would be expected to interact with the MutL homologue MLH1. Using the "yeast interaction trap" with MLH1 as bait, we isolated 22 clones (f1 to f22). Clone f5 strongly interacted with the MLH1 bait and was further characterized. Analysis of the predicted f5 protein sequence revealed a tripartite structure comprising: 1) an amino-terminal methyl-CpG binding domain (MBD) with homology to MeCP (methyl-CpG-binding protein) 2 and PCM (protein containing MBD) 1, which are transcriptional repressors that bind to methylated DNA; 2) a central region with multiple nuclear localization signals; 3) a carboxy-terminal catalytic domain exhibiting homology to several bacterial damage-specific endonucleases with glycosylase/lyase activity, including MutY and endonuclease III from E. coli, mismatch glycosylase Mig.Mth from Methanobacterium thermoautotrophicum, and ultraviolet repair endonuclease from Micrococcus luteus. Because of its domain organization, f5 was named MED1 (methyl-CpG binding endonuclease 1; Figure 1).

The specificity of the MED1-MLH1 interaction was assayed by retransforming virgin yeast cells with f5 plasmid and the MLH1 bait plasmid, as well as with irrelevant bait constructs. The results confirmed specificity of the interaction in yeast cells (Figure 2A). Coimmunoprecipitation experiments revealed that the MED1 interaction with MLH1 also occurs in human cells (Figure 2B). These experiments suggest that MED1 is a novel DNA repair protein that may affect MLH1-regulated processes, including MMR. We are currently defining the biochemical and functional properties of MED1 and its role in human cancer (see below).

BIOCHEMICAL AND FUNCTIONAL CHAR-ACTERIZATION OF MED1. SCHEPIS, CICCHILLITTI, RICCIO, BELLACOSA, in collaboration with MATSUMOTO,^§ YEUNG^§

The domain structure of MED1 (Figure 1) predicts two major biochemical properties, DNA binding and endonuclease activity. The DNA binding properties of MED1 were analyzed by electromobility shift assay (EMSA). The purified MBD of MED1 was incubated with a ³²P-labeled double-stranded oligonucleotide of arbitrary sequence containing four symmetrical methyl-CpG sites (fully methylated oligo). MBD was also incubated with ³²P-labeled double-stranded oligonucleotides of identical sequence in which cytosines replaced methyl-cytosines on one or both strands (hemimethylated and unmethylated oligos, respectively). EMSA results indicated that the MED1 MBD binds to fully-and hemimethylated DNA, but fails to bind to unmethylated DNA (Figure 3). Competition analyses indicated that the binding affinity of MED1 to fully methylated DNA is higher than to hemimethylated DNA (Figure 3). Presently, we are trying to identify the critical residues involved in methyl-CpG recognition.

MED1 endonuclease activity was assayed by evaluating the conversion of supercoiled plasmid DNA into nicked and linear molecules. Incubation of supercoiled DNA with purified recombinant MED1 resulted in a dose- and time-dependent appearance of nicked and linearized molecules. A deletion mutant lacking the putative endonuclease domain failed to produce nicked and linearized DNA molecules, whereas the isolated endonuclease domain also resulted in the appearance of nicked and linearized DNA molecules. This demonstrates that this domain is sufficient for MED1 single- and double-stranded endonuclease activity. It will be important to precisely characterize the MED1 endonuclease activity using defined DNA substrates and determine whether MED1 has N-glycosylase and/or apurinic/apyrimidinic site lyase activity. Also, it is important to determine whether binding to fully methylated or hemimethylated DNA affects the catalytic properties of MED1.

To assess the physiological significance of MED1 in DNA repair, we conducted cell transfection studies. The modular structure of MED1, with an amino-terminal domain involved in binding to methylated DNA and a carboxy-terminal region involved in catalysis and complex formation with MLH1, suggests that deletion of one domain might generate mutants with dominant-negative properties. MMR-proficient, microsatellite-stable, SW480 colon carcinoma cells were transfected with empty vector or expression constructs of wild type MED1 and two deletion mutants, DMBD and Dendo. Stable cell lines (pooled cultures) were generated and were re-transfected with the episomal vector pCAR-OF containing an out-of-frame (CA)₂₈ repeat within the coding region of the b-galactosidase (b-gal) gene. Insertions/deletions in the (CA) repeat may reconstitute the reading frame and restore b-gal activity of this reporter, suggesting a defect in MMR activity. As a control for the efficiency of transfection/expression, cell lines were also transfected with the pCAR-IF vector containing the b-gal gene with an in-frame (CA)₂₇ repeat. Following selection of episome-containing cells, pooled cultures were stained with X-gal to assess b-gal activity (indicated by blue cells). Results indicate that DMBD-transfected cultures contain approximately 30-fold more blue cells than other cultures. This suggests that the DMBD mutant acts in a dominant-negative fashion and possibly perturbs endogenous MMR. To conclusively demonstrate that MED1 is a bona fide MMR protein, we are trying to elucidate its role in repair of mismatch-containing substrates in both in vivo and in vitro reconstituted systems. If MED1 is a MMR protein, it is important to determine whether, like the endonuclease MutH, it is involved in strand discrimination. However, based on its homology to bacterial damage-specific glycosylases/lyases, it is possible that MED1 functions in a pathway of base excision repair.

The domain structure and biochemical properties of MED1 imply a potential role for cytosine methylation in eukaryotic DNA repair, perhaps MMR. Although the transfection experiments with the DMBD mutant of MED1 indicate that the MBD domain is important for MED1 function, the precise role of the recognition of CpG-methylated DNA needs to be established. An attractive hypothesis is that, similar to the bacterial methyl-directed reaction, strand-specificity in human MMR can be determined, at least in part, by the MED1-mediated recognition of transiently hemimethylated CpG sites generated after DNA replication.

Based on the preceding data, which suggest a possible role for MED in MMR, the MED1 gene could be involved in the pathogenesis of HNPCC and sporadic CRCs with MSI. Since only 50 to 70% of HNPCC and sporadic colorectal tumors with MSI carry mutations in the known MMR genes, MED1 mutations may account for some of the remaining cases. We are conducting a mutational analysis in human cancer specimens. Our hypothesis is that MED1 mutations may be relevant to the pathogenesis of human cancer.

AOKI, M., BATISTA, A., BELLACOSA, A., TSICHLIS, P.N., VOGT, P.K. The Akt kinase: molecular determinants of oncogenicity. Proc. Natl. Acad. Sci. USA 95:14950-14955, 1998.

BELLACOSA, A., CHAN, T.O., AHMED, N.N., DATTA, K., MALSTROM, S., STOKOE, D., MCCORMICK, F., FENG, J., TSICHLIS, P.N. Akt activation by growth factors is a multiple-step process: The role of the PH domain. Oncogene 17:313-325, 1998.

BELLACOSA, A., CICCHILLITTI, L., SCHEPIS, F., RICCIO, A., YEUNG, A.T., MATSUMOTO, Y., GOLEMIS, E.A., GENUARDI, M., NERI, G. MED1, a Novel Human Methyl-CpG Binding Endonuclease, Interacts with the DNA Mismatch Repair Protein MLH1. Proc. Natl. Acad. Sci. USA 96:3969-3974, 1999.

CATTANI, P., HOHAUS, S., BELLACOSA, A., GENUARDI, M., CAVALLO, ALMADORI, G., S., CADONI, G., GALLI, J., MAURIZI, M., FADDA, G., NERI, G. Association between cyclin D1 gene amplification and HPV infection in human laryngeal squamous cell carcinoma. Clinical Cancer Res. 4:2585-2589, 1998.

EVES, E.M., XIONG, W., BELLACOSA, A., AHMED, N.N., KENNEDY, S.G., TSICHLIS, P.N., ROSNER, M.R., HAY, N. Akt, a target of phosphatidylinositol 3-kinase, inhibits apoptosis in a differentiating neuronal cell line. Mol. Cell. Biol. 18:2143-2152, 1998.

GENUARDI, M., ANTI, M., CAPOZZI, E., LEONARDI, F., FORNASARIG, M., NOVELLA, E., BELLACOSA, A., VALENTI, A., GASBARRINI, G.B., RONCUCCI, L., BENATTI, P., PERCESEPE, A., PONZ DE LEON, M., COCO, C., DE PAOLI, A., VALENTINI, M., BOIOCCHI, M., NERI, G., VIEL, A. MLH1 and MSH2 constitutional mutations in colorectal cancer families not meeting the standard criteria for hereditary nonpolyposis colorectal cancer. Int. J. Cancer 75:835-839, 1998.

GENUARDI, M., VIEL, A., BONORA, D., CAPOZZI, E., BELLACOSA, A., LEONARDI, F., VALLE, R., VENTURA, A., PEDRONI, M., BOIOCCHI, M., NERI, G. Characterization of MLH1 and MSH2 alternative splicing and its relevance to molecular testing of colorectal cancer susceptibility. Hum. Genet. 102:15-20, 1998.

MARONE, M., SCAMBIA, G., GIANNITELLI, C., FERRANDINA, G., MASCIULLO, V., BELLACOSA, A., BENEDETTI PANICI, P., MANCUSO, S. Analysis of cyclin E and cdk2 in ovarian cancer: gene amplification and RNA overexpression. Int. J. Cancer: 75:34-49, 1998.

MUISE-HELMERICKS, R.C., GRIMES, H.L., BELLACOSA, A., MALSTROM, S.E., TSICHLIS, P.N., ROSEN, N. Cyclin D expression is controlled post-transcriptionally via a phosphatidylinositol 3-kinase/Akt-dependent pathway. J. Biol. Chem. 273:29864-29872, 1998.

^§   Fox Chase researcher

^a   M. Genuardi, G. Neri: Dept. of Medical Genetics, Catholic University Medical School, Rome, Italy

^b   L.A. Aaltonen, A. Loukola: Dept. of Medical Genetics, Haartman Institute, University of Helsinki, Helsinki, Finland

^c   A. Percesepe: Dept. of Internal Medicine, University of Modena, Modena, Italy

Illustrations or unpublished data in these reports should not be used without permission of the author.

Fox Chase Cancer Center

Scientific Report 1998