Gene Silencing in Phenomena Related to DNA Repair

Gene silencing in phenomena related to DNA repair

Tsunehiro Mukai*,1 and Mutsuo Sekiguchi2

1Division of Molecular Biology and Genetics, Department of Biomolecular Sciences, Saga Medical School, 5-1-1 Nabeshima, Saga 849-8501, Japan; 2Biomolecular Engineering Research Institute, 6-2-3 Furuedai, Suita, Osaka 565-0874, Japan

DNA methylation is essential for embryonic development and histone methylation, has recently been recognized. and important for transcriptional repression, as observed Current studies suggest that DNA methylation is in several biological phenomena. These include genomic indirectly linked with histone methylation (Tamaru imprinting, X-inactivation and carcinogenesis. The basic and Selker, 2001; Jackson et al., 2002). mechanism by which DNA methlyation silences tran- In cancer, many critical genes are aberrantly silenced by scription is generally understood, but there is still much DNA methylation. Recent studies have shown that to be learned about how DNA methyltransferase is silencing of certain DNA repair genes by DNA methyla- targeted to a specific region of the gene. Silencing by tion might be related to the occurrence of tumorigenic DNA methylation occurs at an early stage of carcino- mutations. Among mismatch repair genes, whose defects genesis, when the DNA repair genes, MGMT and would cause tumor-prone genetic diseases, MLH1 hMLH1, are frequently inactivated, resulting in muta- frequently loses its potential through lowered levels of tions in key cancer-related genes in cells. Mice defective gene expression, due to hypermethylation of its promoter in Mgmt and/or Mlh1 gave clear evidence of the region. Epigenetic silencing is also observed with the Mgmt significant roles of these proteins in carcinogenesis. gene, whose product repairs the promutagenic DNA lesion Recently, it has been demonstrated that DNA methyla- O6-methylguanine. These alterations have been widely tion is linked to histone methylation in fungi and plants, observed in human tumors (Esteller et al., 2000a, 2001a; although it remains unknown whether this mechanism Jones and Baylin, 2002; Baylin et al., 2001). Thus, arriving occurs in mammalian systems. at a solid understanding of the mechanisms underlying Oncogene (2002) 21, 9033 – 9042. doi:10.1038/sj.onc. gene silencing is of utmost importance in carcinogenesis as 1206095 well as mutagenesis. Here we describe first the general features of gene Keywords: DNA methylation; histone methylation; silencing through DNA methylation, and then proceed cancer; Mgmt; Mlh1; animal model to the problems related to loss of expression of DNA repair genes.

Mechanism of gene silencing by DNA methylation Introduction DNA methylation in various biological phenomena Epigenetics is the study of mitotically and/or meiotically heritable changes in gene function that cannot be Differential expression takes place through normal explained by changes in DNA sequence (Russo et al., development, and a part of the process is related to 1996). For convenience, epigenetics can be classified into tissue-specific expression after differentiation, imprint- two types: one is dependent on DNA methylation and ing, and X-inactivation. It is well known that DNA the other is not. DNA methylation is the major methylation is involved in these phenomena. In epigenetic modification of the mammalian genome, addition, aberrant DNA methylation is frequently now widely accepted as being important in the observed in cancer, in which expression of certain regulation of gene expression in mammals as well as in genes is greatly altered. While global DNA hypo- higher plants. The major modified base in DNA methylation occurs in wide areas of chromosomes, methylation is 5-methylcytosine (m5C) within a CpG inactivation of genes in tumor cells occurs regionally in dinucleotide. The CpG dinucleotide is mainly localized non-random fashion in certain localized areas of in the CpG islands, which are distributed in promoter chromosomes and contributes to establishing cancer regions of certain genes. DNA methylation is associated phenotypes. Another biological role of DNA methyla- with changes in chromatin structure; consequently, gene tion is prevention of expression of parasitic DNA silencing is achieved through interference in transcrip- elements, such as retrotransposons, retroviruses, and tion. The significance of DNA methylation-independent some repetitive elements, limiting their spread through epigenetics, such as histone acetylation/deacetylation the genome. Here we will discuss general features of gene expression, including tissue-specific expression, imprinting, and X-inactivation and then proceed to *Correspondence: T Mukai, E-mail: [email protected] silencing of DNA repair genes. Gene silencing in DNA repair T Mukai and M Sekiguchi 9034 Tissue-specific expression has long been speculated How does DNA methylation repress the target gene? to be controlled by DNA methylation. However, no genes responsible for the DNA methylation process CpG islands rich in CpG sequences are found in have been identified. Although some candidate genes promoter regions of half of the genes in the were isolated, gene targeting experiments for DNA mammalian genome and are usually unmethylated in methyltransferase have failed to support their roles in normal cells, except for genes observed in some regulation of tissue-specific genes (Walsh and Bestor, biological phenomena related to gene silencing. The 1999). Recently, some other candidate genes have been direct correlation between CpG island methylation and described (de Smet et al., 1999; Futscher et al., 2002), silencing of gene transcription is as follows (Santini et and further studies are awaited. al., 2001): (1) cells in which silencing occurs are usually Genomic imprinting is a functional non-equivalence transcriptionally competent for the affected gene if this of the parental genome and results in the expression of gene is unmethylated; (2) demethylation by pharmaco- gene from only one of the two parental chromosomes logic agents results in reactivation of gene expression (Reik and Walter, 2001). This phenomenon is an (Jones, 1985); and (3) in vitro methylation substantially epigenetic marking by which expression of imprinted reduces gene expression in transfection assay. genes becomes dependent on their parental origin. The process of inhibition of transcription was Many of these genes play key roles in growth, initially thought to be simply due to physical differentiation and behavior. The primary mechanism interference when the methyl group protruded into of genomic imprinting is thought to be CpG methyla- the major groove and in consequence lowered levels of tion (Reik and Walter, 2001). Since only one of the association with the transcriptional apparatus (Tate parental alleles is methylated, this type of methylation and Bird, 1993). However, it was revealed that the is allele specific. The region for differential methylation inhibitory mechanism exerts its effect through the is usually referred to as ‘DMR’ (differentially methy- binding of specific proteins to the methylated DNA lated region), which is a hallmark of imprinted genes sequences. The responsible protein, MeCP2, was (Li et al., 1993). If methylation in DMR originates identified (Lewis et al., 1992), and subsequently an from a germ cell, this region is called an imprinting additional four proteins were found in mammals. center (IC) (Ben-Porath and Cedar, 2000). Targeting These proteins commonly possess a methylCpG- experiments of candidate ICs, derived from the mouse binding domain, and were named MBD1 through counterpart corresponding to human imprinted region, MBD4 (Hendrich and Bird, 1998). MethylCpG-binding showed loss of imprinting of those regions (Wutz et al., proteins (MBDs), MeCP2 and MBD1 through MBD4, 1997; Thorvaldsen et al., 1998; Yang et al., 1998). belong to MBD family members. MeCP2 and MBD1 Recently, it was revealed that histone modifications are have a transcriptional repressor domain (TRD), to important to determine whether the DNA methylation which the corerepressor mSin3A would bind. This occurs (Tamaru et al., 2001; Jackson et al., 2002). It is corerepressor protein constitutes the core of a multi- proposed that histone methylation at lysine 9 is a protein complex, in which histone deacetylase is gametic imprint in the imprinting center PWS-IC (Xin included. It turned out that the mechanism by which et al., 2001), while cytosine methylation of the PWS-IC DNA methylation could inhibit transcription was due occurs only after fertilization (El-Maarri et al., 2001). to sterical blocking of transcription factors by binding X-chromosome inactivation in mammals is the the MBDs. means for dosage compensation of X-linked genes, The process of transcriptional repression by DNA which is achieved by transcriptional silencing of genes methylation is illustrated in Figure 1. DNMT recruited on one of the two X chromosomes in the female by unknown DNA binding protein methylates DNA, (Avner and Heard, 2001). Xist RNA, which is while this DNMT interacts with HDACs (Figure 1a) specifically transcribed in inactive X, is produced after (Robertson et al., 2000; Fuks et al., 2000; Rountree et differentiation. This RNA product is spread and coats al., 2000). DNMT can thereby act as a repressor by its own chromosome to establish the inactive state. To methylating CpG sites and simultaneously by being achieve this, histone deacethylation and DNA methyla- associated with histone deacethylation. Then, one of tion of X-linked genes, as well as the recruitment of the MBDs binds tightly to chromosomes in a methylation- histone variant macroH2A, occur and presumably dependent manner. TDR, one of the binding domains of assist in transforming the Xist RNA-coated chromo- MBDs, associates with histone deacetylase to form some into a stably inactive and condensed chromatin deacetylated histone (Jones et al., 1998; Nan et al., 1998) state. Recent studies show that methylation of histone, (Figure 1b). This deacetylation of histone (and/or other particularly of H3, is important to mark the inactive X; proteins) is an essential component of this repression namely, H3 methylation provides an epigenetic imprint mechanism, since transcriptional repression in vivo is or a nucleation center for Xist RNA on the inactive X relieved by a deacetylase inhibitor, tricostatin A. (Heard et al., 2001). Furthermore, the inactive X chromosome is enriched for H3 methylated at lysine 9 How does DNMT target some specific regions of the (H3-K9) (Boggs et al., 2002; Peters et al., 2002). It has chromosome? also been demonstrated that histone methylation occurs prior to DNA methylation of CpG island DNA methylation does not occur randomly in a (Brockdorff, 2002). genome DNA. Methylation can be targeted in vivo

Oncogene Gene silencing in DNA repair T Mukai and M Sekiguchi 9035

Figure 1 Mechanism of transcriptional repression by DNA methylation. (a) DNMT bound to unknown protein X methylates 5-CpG islands of DNA, while HDAC interacting with DNMT deacetylates acetylated histone. (b) After methylation of DNA, Figure 2 Mechanism of recruitment of DNMT to the RARb2 one of MBDs (methyCpG binding proteins) binds to 5-methyl- gene in APL. (a) PML – RAR fusion protein, which binds to CpG. HDAC interacting with MBD executes histone deacetyl- the DNA element of RARb2, interacts with DNMT, thereby re- ation, resulting in deacetylation of histone and in consequence cruiting DNMT complex to the binding site of RARb2 and transcriptional repression. DNMT, DNA methyltransferase; methylating the RARb2 region. HDAC-bound DNMT then dea- HDAC, histone deacetylase; MBD, methylCpG-binding protein; cetylates histone, resulting in repression of transcription. (b) Bind- cross-hatched large circle, acetylated histone; open large circle, ing of RA to its receptor along with the addition of a deacetylated histone; closed small circle with bar, methylated demethylating agent releases the DNMT/HDAC complex from CpG; open small circle with bar, non-methylated CpG PML – RAR fusion protein, resulting in DNA demethylation, histone acetylation and in consequence transcriptional reactivation. RA, retinoic acid; open large circle, deacetylated histone; cross- hatched large circle, acetylated histone within certain regions of the chromosomes where responsible for CpG methylation of its target genes and repetitive DNA elements, centromeres, and imprinted whether aberrant methylation is relevant for its genes exist. In tumor cells, this methylation is biological activity. When applied to the RA receptor aberrantly targeted to impair expression of tumor RARb2, which contains a CpG island and is suppressor genes. For methylation, DNA methyltrans- considered to be a putative tumor suppressor, PML – ferase (DNMT) has to be recruited to a targeting site RAR bound to the RARb2 promoter and induced its on the gene promoter. transcriptional silencing. Methylated sites on the APL DNMT1-containing fraction with deacetylase activ- samples were located near the promoter and exon 1 ity, isolated from HeLa cell, contained HDAC1, the regions, and physical association between PML – RAR tumor-suppressor protein Rb and the sequence-specific and endogenous DNMT1 and DNMT3a was evident. transcriptional activator E2F1 (Robertson et al., 2000). Furthermore, DNMTs were enriched at the RARb2 This complex is capable of repressing transcription promoter in the presence of PML – RAR, demonstrat- from promoters containing E2F1-binding sites. The ing that PML – RAR and DNMTs form a stable role of this complex in targeting methylation to a complex on the RARb2 promoter (Figure 2a). specific region of the gene was established by the Treatment with RA and 5-Aza-dC induced promoter following experiment (Figure 2). demethylation, thereby resulting in reactivation of the PML – RAR, an oncogenic transcription factor gene expression, and, consequently, reversion of the found in acute promyelocytic leukemias (APL), is a transformed phenotype (Figure 2b). This is the first fusion protein of the promyelocytic leukemia (PML) case demonstrating targeting methylation to a specified protein with retinoic acid receptor (RAR) and sequence. There is another example to show that functions as a transcriptional regulator of retinoic acid DNMT3a can be targeted to promoter through its (RA)-responsible genes. Using this PML – RAR, di association with a sequence-specific transcription Croce et al. (2002) examined whether PML – RAR is factor, RP58 (Fuks et al., 2001). This RP58 is a

Oncogene Gene silencing in DNA repair T Mukai and M Sekiguchi 9036 DNA-binding transcriptional repressor protein found repression. Although interaction between HP1 and at transcriptionally silent heterochromatin. HMTase was initially found in the heterochromatic region, this interaction has also been observed in the euchromatic region (Nielsen et al., 2001). It seems that Does K9 methylation in H3 control DNA methylation? HMTase enzymes are able to discriminate between the The core histone tails are susceptible to a variety of heterochromatic and the euchromatic region. At least covalent modifications, including acetylation, phos- four different types of HMTases, SUV39H1/Suv39h1, phorylation, methylation, and ubiquitination. These G9a, SETDB1/ESET and Eu-HMTase1, are presently modifications may be critical for transcriptional and known (Rea et al., 2000; Tachibana et al., 2001; translational regulation (Zhang and Reinberg, 2001). Schultz et al., 2002; Yang et al., 2002; Ogawa et al., Based on this, ‘histone code’ hypothesis has been 2002). SUV39H1/Suv39h1 enzyme might act on proposed (Strahl and Allis, 2000; Jenuwein and Allis, heterochromatin to repress the pericentric heterochro- 2001). matin region, while the remaining three might target Methylation of histone H3 and H4 is important in transcriptionally active sites of the euchromatin this process, and we focus here on methylation of the regions. lysine 9 residue in histone H3 (H3-K9), since it could Further insight on this process was recently reported be linked to DNA methylation. SUV39H1/Suv39h1 with fungi as well as with plants. Tamaru et al. (2001) gene encodes histone H3-specific methyltransferase demonstrated that replacement of H3-K9 with either (HMTase) that methylates lysine 9 of the amino leucine or arginine residue induced a markedly reduced terminus region of histone H3 (Rea et al., 2000). H3- level of DNA methylation in Neurospora crassa, K9 was selectively recognized by the chromo domain indicating that DNA methylation depends on histone of HP1 (heterochromatin protein 1) (Bannister et al., methylation (Figure 3b) (Tamaru et al., 2001). A 2001; Lachner et al., 2001) (Figure 3a). HP1 is then similar observation was made with Arabidopsis thaliana localized in the heterochromatin region, where gene (Jackson et al., 2002). However, how histone methyla- silencing occurs. Thus, the chromo domain of HP1 is tion regulates CpG methylation is not known. necessary for both targeting H3-K9 and transcriptional These results indicate that a causal relationship between histone methylation and DNA methylation is conserved throughout eukaryotes. Recruitment of DNA methyltransferase to chromatin by HP1 could be a general eukaryotic phenomenon, as evidenced by a recent study with mouse. Expression of the mouse Mage-a2 gene is controlled by DNA methylation and usually turned off in most tissues, except for testis (de Smet et al., 1999). In G9+/+ ES cells, H3-K9 is enriched in the chromatin containing the Mage-a2 promoter sequence, while its content is severely reduced in G9a7/7 ES cells, suggesting that DNA methylation will be lost by depletion of G9a HMTase in G9a7/7 (Tachibana et al., 2002).

Silencing of DNA repair genes

General aspects of gene silencing by DNA methylation in cancer Two hits are required for the full inactivation of a tumor-suppressor gene (Knudson, 1971), and this is usually achieved by a mutation in one allele and loss of heterozygosity in another allele. Jones and Laird (1999) pointed out that many tumor-suppressor genes are inactivated by DNA methylation at considerable frequencies, and methylation of one allele with coordination of mutation or LOH in another-allele, or methylation of both alleles, would cause the same Figure 3 Model for molecular mechanism of DNA methylation through histone methylation. (a) Methylated histone (H3-K9) is effect. recognized by HP1. HP1 interacts with HMTase, resulting in ad- Genes affected by DNA methylation in cancer are ditional methylation of histone for self-maintenance. (b) HP1 presented in ‘Genes affected by promoter CpG island bound to methylated histone recruits DNMT, resulting in DNA methylation in aging and/or cancer’ (www3.mdanderson. methylation. HP1, heterochromatin protein 1; HMTase, histone H3 methyltransferase; DNMT, DNA methyltransferase; open org/leukemia/methylation/cgi.html). Approximately 70 large circle, non-methylated histone; marked large circle, methy- genes are listed, which include genes related to cell- lated histone; closed small circle with bar, methylated CpG cycle control, DNA repair, apoptosis, and metastatic

Oncogene Gene silencing in DNA repair T Mukai and M Sekiguchi 9037 potential. Most but not all tumor-suppressor genes are their lack or possession of a very low level of O6- included. methylguanine-DNA methyltransferase activity. This How widespread is DNA hypermethylation in tumor enzyme repairs promutagenic DNA lesion, O6-methyl- cells? The restriction landmark genome scanning guanine and O4-methylthymine, which account for (RLGS) is a suitable method for such global analysis approximately 5 and 0.1% of the total alkylated (Hatada et al., 1991). It was estimated that approxi- DNA adducts, respectively (Koike et al., 1990). Despite mately 600 CpG islands (13%) of the 45 000 in the the lack of enzyme activity, Mer- cells show no gross genome are aberrantly methylated in tumors (Ante- alteration in coding and promoter regions of the Mgmt quera and Bird, 1993; Costello et al., 2000). The gene, encoding methyltransferase protein; therefore, the patterns of CpG island methylation are non-random decreased levels of enzyme activity may be due to and show distinct tumor-type specificities, suggesting transcriptional silencing (Harris et al., 1991; Nakatsu et that each tumor may display a distinguishable al., 1993). methylation subtype in the genome. Cultivation of Mer- cells in the presence of certain Gene hypermethylation profiles for a total of 12 low levels of alkylating agents yields alkylation- genes in human cancer have been provided, which were resistant cells. Most of these cells are still deficient in derived from 600 specimens of primary tumors of 15 methyltransferase activity, and the second alteration tissues (Esteller et al., 2001b). From this analysis, the may have occurred to suppress the original high following features of human tumors have been derived. susceptibility to alkylating agents. Although molecular First, profiles of promoter hypermethylation differ for mechanisms underlying this phenomenon are not fully each cancer type, providing tumor-type and gene- understood, most of these phenotypic revertants are specific profiles. This is consistent with the result devoid of mismatch repair activity (Branch et al., 1993; obtained by the RLGS method (Costello et al., 2000). Kat et al., 1993). Among several genes involved in Second, epigenetic inactivation may affect all of the mismatch repair, epigenetic silencing of the Mlh1 gene molecular pathways involved in cell immortalization occurs most frequently. There is hypermethylation of and transformation. Third, these epigenetic changes CpG islands in promoter regions of Mgmt or Mlh1 occur in the absence of a genetic lesion and also genes in several cancer cell lines as well as in human biallelically if coding sequences are wild type (Herman carcinoma (Watts et al., 1997; Qian and Brent, 1997; et al., 1995; Esteller et al., 2000b; Veigl et al., 1998). Veigl et al., 1998; Deng et al., 1999; Danam et al., Finally, it seems that epigenetic changes are one of 1999; Esteller et al., 2000a). several early steps in carcinogenesis (Belinsky et al., To clarify the roles of MGMT and MLH1 protein in 1998; Esteller et al., 1999, 2000b; Fleisher et al., 2000). carcinogenesis, mouse lines defective in either one or Of genes related to DNA repair, hMLH1 and both of these genes were constructed using gene MGMT draw a special attention concerning aberrant targeting techniques. Mice defective in the Mgmt gene DNA methylation in cancer. Human DNA mismatch are hypersensitive to the killing effect of alkylating repair genes (hMLH1, hMSH2, hPMS1 and hPMS2) agents (Tsuzuki et al., 1996). In addition, a large are known as the cause of HNPCC (hereditary non- number of tumors occurred in Mgmt7/7 mice exposed polyposis colorectal cancer). Among them, hMLH1 to relatively low doses of MNU and dimethylnitrosoa- exhibits a frequent occurrence of DNA hypermethyla- mine, whereas no or few tumors occurred in normal tion, which is related to genetic instability- mice treated in the same manner (Sakumi et al., 1997; microsatellite instability (Kane et al., 1997; Herman Iwakuma et al., 1997). Death of Mgmt7/7 mice after et al., 1998; Esteller et al., 1998). As a consequence of MNU administration was closely related to bone loss of this gene function, mutations would accumulate marrow damage and dysplastic mucosae of intestines in the genome. MGMT, on the other hand, is involved together with crypt abscesses. This severe myelosup- in repair of alkylated bases in DNA. Since O6- pression led to a drastic decrease in the number of methylguanine, which can be repaired by the MGMT peripheral leukocytes and platelets (Tsuzuki et al., function, causes mutations leading to tumor induction, 1996). Thus, methyltransferase has a vital role in preservation of the function of this gene is important protecting these organs from toxic effects of alkylating for preventing induction of tumors. These situations agents. It is notable that Mgmt7/7 mice are consider- have been shown in an animal model, as described in ably more sensitive to chemotherapeutic alkylating the following section. drugs, presently in clinical use, than are wild type mice (Glassner et al., 1999; Shiraishi et al., 2000). The killing and the tumorigenic effects of alkylating Animal models for lack of DNA repair genes agents can be dissociated by introduction of a Some human and rodent cell lines, particularly those mismatch repair defect into methytransferase-deficient derived from tumors, show an increased sensitivity to mice (Kawate et al., 1998). Mice with mutations in simple alkylating agents, such as N-methyl-N-nitro- both alleles of the Mgmt and the Mlh1 genes are as sourea (MNU) and N-methyl-N’-nitro-N-nitrosoguani- resistant to MNU as are wild type mice, in terms of dine (MNNG), and yield more mutations when treated survival, but do have numerous tumors after receiving with relatively low doses of such agents (Day et al., MNU (Figure 4). In contrast to Mgmt7/7 Mlh1+/+ 1979; Sklar and Strauss, 1981). These cell lines were mice with smaller thymus and hypocellular bone termed Mer- or Mex- and were further characterized by marrow after MNU administration, no conspicuous

Oncogene Gene silencing in DNA repair T Mukai and M Sekiguchi 9038 change was found in Mgmt7/7 Mlh17/7 mice treated and initiation of repair reaction (Modrich and Lahhue, in the same manner. Thus, introduction of a mismatch 1996; Papadopoulos and Lindblom, 1997; Jiricny, repair gene defect renders methyltransferase-deficient 1998). It is generally assumed that a defect in any mice resistant to the lethal action of alkylating agents, one of these genes would lead to defects in mismatch still maintaining the high susceptibility to tumor repair. In fact, mutations in these genes have been formation. Such mice may be useful for evaluating found in human HNPCC patients, though the carcinogenic effects of various substances, including frequencies of mutations in each of the genes were those for therapeutic application. significantly different (Leach et al., 1993; Bronner et In this context, there is a problem that these doubly al., 1994; Papadopoulos et al., 1994; Kolodner et al., deficient mice had a small but significant number of 1995). Thus, there is a possibility that mutations in tumors even without exposure to MNU (see Figure 4). genes other than Mlh1 might lead to a similar This phenomenon is common to mice defective in phenotype observed with Mlh1+/7 and Mlh17/7 mice. mismatch repair (de Wind et al., 1998; Prolla et al., However, it was reported that after treatment with O6- 1998), and such a high incidence of tumors is benzylguanine, an inhibitor of O6-methylguanine characteristic of HNPCC patients defective in one of methyltransferase, Msh2+/7 cells were as sensitive as the mismatch repair genes (Leach et al., 1993; Bronner Msh2+/+ cells to alkylating agents (de Wind et al., et al., 1994; Papadopoulos et al., 1994; Kolodner et al., 1995). This apparent difference may be due to different 1995). levels of expression of the genes or to different modes This complexity was resolved by introducing the of action of these proteins. Mlh1+/7 mutation, instead of Mlh17/7,inthe Mlh1 expression in tumor tissues of Mgmt7/7 methyltransferase-deficient mice (Kawate et al., 2000). Mlh1+/7 mice exposed to MNU is heterozygous; Mgmt7/7 Mlh1+/7 mice, with about half the amount four of 13 lymphoma samples analysed showed of MLH1 protein as Mgmt7/7 Mlh1+/+ mice, were complete absence of MLH1 protein, whereas the resistant to the killing action of MNU, up to the level remaining lymphoma cells contained half of the of 30 mg/kg body weight. LD50 values were estimated normal amount of MLH1 protein (Kawate et al., as 20, 120, and 280 mg/kg body weight for Mgmt7/7 2000). PCR analyses of the DNA from those MLH1- Mlh1+/+, Mgmt7/7 Mlh1+/7 and Mgmt7/7 Mlh17/7 deficient lymphomas revealed no deletion or large mice, respectively. Eight weeks after exposure to 30 mg/ alteration in the wild-type Mlh1 allele. It may be that kg of MNU, 40% of MNU-treated Mgmt7/7 Mlh1+/7 a base substitution was induced by MNU or that mice had tumors, and there were no tumors in those mice some modification occurred in a certain region of the not given the treatment (Figure 4). It seems that the gene, such as hypermethylation in the promoter cellular content of MLH1 protein is a critical factor for region. Whatever the cause, this loss of Mlh1 determining if damaged cells enter into the pathway expression might be related to transformation of the leading to mutation induction or to apoptotic cell death. cell. A complete deficiency of MLH1 function would This situation is illustrated in a model shown in Figure 5. lead to an increased frequency of errors during DNA At an early step in mismatch repair, four of the replication. It has been shown that some gene products of at least five genes, Msh2, Msh3, Msh6, expression is altered during tumor development and Mlh1 and Pms2, form complexes and play an such alterations in the genome would further accel- important role in recognition of mismatched pairs erate progression of the tumor.

Figure 4 Effects of MNU on tumor induction in mice with various genetic backgrounds. Mice (6 weeks old) were given 30 mg/kg of body weight of MNU or phosphate-buffered saline (control) i.p. and killed 8 weeks later for examination. Data were taken from Kawate et al. (1998, 2000)

Oncogene Gene silencing in DNA repair T Mukai and M Sekiguchi 9039

Figure 5 Anti-carcinogenic mechanism. Alkylating agents produce O6-methylguanine and other alkylated bases in DNA. During DNA replication, O6-methylguanine can pair with thymine as well as cytosine, leading to G : C?A : T transition mutations, which eventually cause tumor induction. This mutagenic event is prevented by two consecutive processes: ﬁrst, repair of O6-methylguanine by transfer of a methyl group to MGMT protein; and second, elimination of cells carrying methylated lesions through apoptosis, in which MLH1 and other mismatch repair proteins are involved

Most HNPCC patients have mutations in only one transition mutations were found frequently in both K- allele of certain mismatch repair genes, but frequently ras and p53 genes of various cancers. Of colorectal produce tumors (Leach et al., 1993; Bronner et al., cancer samples with transition mutations in the K-ras 1994; Papadopoulos et al., 1994; Kolodner et al., gene, 71% showed hypermethylation of MGMT 1995). Some of these tumors are devoid of mismatch (Esteller et al., 2000a). In the case of p53, transition repair protein, a phenomenon similar to that observed mutations were 64, 71, and 92% in non-small cell lung in Mlh1+/7 mice (Kawate et al., 2000). Thus, cancer, colorectal cancer, and astrocytomas (secondary alterations in the level of Mlh1 expression may be an glioblastomas), respectively (Esteller et al., 2001a; Wolf important factor in determining genetic stability as well et al., 2001; Nakamura et al., 2001). Recently, as susceptibility to carcinogens. Rajagopalan et al. (2002) found that in colorectal cancers, Braf mutations occur only in tumors that do not carry mutations in the K-ras. There was a striking Epigenetic silencing observed with clinical samples difference in the frequency of Braf mutations between Among DNA repair genes, hMHH1, MGMT and cancers with and without mismatch repair deficiency: BRCA1 most frequently exhibit epigenetic silencing. mismatch repair-deficient tumors had a very high Thus, levels and extents of DNA methylation of these incidence of Braf mutations and a lower incidence of genes have been examined with special reference to K-ras mutations as compared with mismatch repair- gene and tumor types (Esteller et al., 2001b). Tissues proficient colorectal cancers. showing more than 20% of methylation in these genes Referring to the results of Mgmt and Mlh1 double are colon, lung, head and neck, lymphoma, brain, and knockout mice (Iwakuma et al., 1997; Sakumi et al., esophagus in MGMT; colon, uterus, and stomach in 1997; Kawate et al., 1998, 2000), it is of interest to hMLH1. In addition, tumor types are different, note the expression levels of both MGMT and MLH1 depending on the genes affected. The epigenetic lesion genes in human cancers. We examined 46 samples of is often an early event. Both MGMT and hMLH1 hepatocellular carcinomas by Western blot analysis to showed DNA methylation even in precancerous tissues: see MGMT and MLH1 expression deficiencies MGMT in colorectal adenoma (Esteller et al., 2000b) (Matsukura et al., 2003). Twenty per cent of tumor and hMLH1 in endometrial hyperplasia (Esteller et al., samples showed hMLH1 deficiency, while 60% of 1999) and ulcerative colitis (Fleisher et al., 2000). them lacked MGMT expression. Of interest is the In tumors of mice induced by DMNA (dimethylni- observation that 10% of these samples lacked both trosamine) and MNU, G : C to A : T transitions occur MGMT and MLH1 expression, implying that defec- frequently in the K-ras gene (You et al., 1989; Chen et tive expression of these two genes occurs al., 1994). Many clinical samples, derived from independently. This may be related to the observation MGMT-deficient or MLH1-deficient cancers, were that in colorectal cancers, MGMT and hMLH1 genes examined for such mutations in tumor-related genes. are inactivated at high incidence by hypermethylation In the samples whose MGMT gene is hypermethylated, (Esteller et al., 2001b).

Oncogene Gene silencing in DNA repair T Mukai and M Sekiguchi 9040 Future prospect human. Targeting of the HMTase gene, G9a, also indicates that DNA methylation could be lost in Mage- Recent works in epigenetics has raised an important a2, as described earlier. All these results strongly issue for DNA methylation. One of the major suggest that H3-K9 methylation is an epigenetic questions is whether DNA methylation is a primary imprint, which could cause DNA methylation as a cause of certain biological phenomena or a conse- secondary event. quence of certain preceding events. It was naturally It is important to know whether DNA methylation is thought in the case of genomic imprinting that DNA closely linked to histone methylation in cancer tissues. methylation is a gametic imprint, since targeting of the If this pathway is aberrantly regulated in tumor cells, Dnmt1 gene resulted in failure of imprinting, due to we may observe a different relationship between these loss of maintenance methylation (Li et al., 1992). two processes. Further studies are clearly necessary to However, recent studies in fungus and plant revealed understand nature of DNA and histone modification in that the histone H3-K9 mark is necessary for DNA carcinogenesis and to develop therapeutic agents to methylation, indicating that histone methylation could intervene in these processes. be an upstream event for DNA methylation. In mammals, H3-K9 methylation is observed as an earliest known chromatin change during X-inactivation Acknowledgments and is also required for genomic imprinting. In We thank Drs S Matsukura, Y Satoh, T Nakagawachi and genomic imprinting, H3-K9 methylation occurs speci- H Yatsuki for helping us to collect data. We also thank fically on the inactive allele of imprinted gene in Drs H Soejima and S Matsukura for useful discussions.

References

Antequera F and Bird A. (1993). Proc. Natl. Acad. Sci. USA, Deng G, Chen A, Hong J, Chae HS and Kim YS. (1999). 90, 11995 – 11999. Cancer Res., 59, 2029 – 2033. Avner P and Heard E. (2001). Nat. Rev. Genet., 2, 59 – 67. Di Croce L, Raker VA, Corsaro M, Fazi F, Fanelli M, Bannister AJ, Zegerman P, Partridge JF, Miska EA, Thomas FarettaM,FuksF,LoCocoF,KouzaridesT,NerviC, JO, Allshire RC and Kouzarides T. (2001). Nature, 410, Minucci S and Pelicci PG. (2002). Science, 295, 1079 – 120 – 124. 1082. Baylin SB, Esteller M, Rountree MR, Bachman KE, El-Maarri O, Buiting K, Peery EG, Kroisel PM, Balaban B, Schuebel K and Herman JG. (2001). Hum. Mol. Genet., WagnerK,UrmanB,HeydJ,LichC,BrannanCI,Walter 10, 687 – 692. J and Horsthemke B. (2001). Nat. Genet., 27, 341 – 344. Belinsky SA, Nikula KJ, Palmisano WA, Michels R, Esteller M, Levine R, Baylin SB, Ellenson LH and Herman Saccomanno G, Gabrielson E, Baylin SB and Herman JG. (1998). Oncogene, 17, 2413 – 2417. JG. (1998). Proc. Natl. Acad. Sci. USA, 95, 11891 – 11896. Esteller M, Catasus L, Matias-Guiu X, Mutter GL, Prat J, Ben-Porath I and Cedar H. (2000). Curr.Opin.Genet.Dev., Baylin SB and Herman JG. (1999). Am. J. Pathol., 155, 10, 550 – 554. 1767 – 1772. Boggs BA, Cheung P, Heard E, Spector DL, Chinault AC EstellerM,ToyotaM,Sanchez-CespedesM,CapellaG, and Allis CD. (2002). Nat. Genet., 30, 73 – 76. Peinado MA, Watkins DN, Issa JP, Sidransky D, Baylin Branch P, Aquilina G, Bignami M and Karren P. (1993). SB and Herman JG. (2000a). Cancer Res., 60, 2368 – 2371. Nature, 362, 652 – 654. EstellerM,SparksA,ToyotaM,Sanchez-CespedesM, Brockdorff N. (2002). Trends Genet., 18, 352 – 358. Capella G, Peinado MA, Gonzalez S, Tarafa G, Sidransky Bronner CE, Baker SM, Morrison PT, Warren G, Smith LG, D, Meltzer SJ, Baylin SB and Herman JG. (2000b). Cancer LescoeMK,KaneM,EarabinoC,LipfordJand Res., 60, 4366 – 4371. Lindblom A et al. (1994). Nature, 368, 258 – 261. Esteller M, Risques RA, Toyota M, Capella G, Moreno V, Chen B, You L, Wang Y, Stoner GD and You M. (1994). Peinado MA, Baylin SB and Herman JG. (2001a). Cancer Carcinogenesis, 15, 2031 – 2035. Res., 61, 4689 – 4692. Costello JF, Fruhwald MC, Smiraglia DJ, Rush LJ, Esteller M, Corn PG, Baylin SB and Herman JG. (2001b). Robertson GP, Gao X, Wright FA, Feramisco JD, Cancer Res., 61, 3225 – 3229. Peltomaki P, Lang JC, Schuller DE, Yu L, Bloomﬁeld Fleisher AS, Esteller M, Harpaz N, Leytin A, Rashid A, Xu CD,CaligiuriMA,YatesA,NishikawaR,SuHuangH, Y,LiangJ,StineOC,YinJ,ZouTT,AbrahamJM,Kong Petrelli NJ, Zhang X, O’Dorisio MS, Held WA, Cavenee D, Wilson KT, James SP, Herman JG and Meltzer SJ. WK and Plass C. (2000). Nat. Genet., 24, 132 – 138. (2000). Cancer Res., 60, 4864 – 4868. Danam RP, Qian XC, Howell SR and Brent TP. (1999). Mol. Fuks F, Burgers WA, Brehm A, Hughes-Davies L and Carcinog., 24, 85 – 89. Kouzarides T. (2000). Nat. Genet., 24, 88 – 91. Day III RS, Ziolkowski CH, Scudiero DA, Meyer SA and Fuks F, Burgers WA, Godin N, Kasai M and Kouzarides T. Mattern MR. (1979). Nature, 279, 797 – 799. (2001). EMBO J., 20, 2536 – 2544. De Smet C, Lurquin C, Lethe B, Martelange V and Boon T. Futscher BW, Oshiro MM, Wozniak RJ, Holtan N, Hanigan (1999). Mol. Cell. Biol., 19, 7327 – 7335. CL, Duan H and Domann FE. (2002). Nat. Genet., 31, De Wind N, Dekker M, Berns A, Radman M and te Riele H. 175 – 179. (1995). Cell, 82, 321 – 330. De Wind N, Dekker M, van Rossum A, van der Valk M and te Riele H. (1998). Cancer Res., 58, 248 – 255.

Oncogene Gene silencing in DNA repair T Mukai and M Sekiguchi 9041 Glassner BJ, Weeda G, Allan JM, Broekhof JL, Carls NH, Modrich P and Lahhue R. (1996). Annu.Rev.Biochem.,65, Donker I, Engelward BP, Hampson RJ, Hersmus R, Hickman 101 – 133. MJ, Roth RB, Warren HB, Wu MM, Hoeijmakers JH and Nakamura M, Watanabe T, Yonekawa Y, Kleihues P and Samson LD. (1999). Mutagenesis, 14, 339 – 347. Ohgaki H. (2001). Carcinogenesis, 22, 1715 – 1719. Harris LC, Potter PM, Tano K, Shiota S, Mitra S and Brent Nakatsu Y, Hattori K, Hayakawa H, Shimizu K and TP. (1991). Nucleic Acids Res., 19, 6163 – 6167. Sekiguchi M. (1993). Mutat. Res., 293, 119 – 132. Hatada I, Hayashizaki Y, Hirotsune S, Komatsubara H and Nan X, Ng HH, Johnson CA, Laherty CD, Turner BM, Mukai T. (1991). Proc. Natl. Acad. Sci. USA, 88, 9523 – Eisenman RN and Bird A. (1998). Nature, 393, 386 – 389. 9527. Nielsen SJ, Schneider R, Bauer UM, Bannister AJ, Morrison Heard E, Rougeulle C, Arnaud D, Avner P, Allis CD and A, O’Carroll D, Firestein R, Cleary M, Jenuwein T, Spector DL. (2001). Cell, 107, 727 – 738. Herrera RE and Kouzarides T. (2001). Nature, 412, 561 – Hendrich B and Bird A. (1998). Mol. Cell. Biol., 18, 6538 – 565. 6547. Ogawa H, Ishiguro K, Gaubatz S, Livingston DM and Herman JG, Merlo A, Mao L, Lapidus RG, Issa JP, Nakatani Y. (2002). Science, 296, 1132 – 1136. Davidson NE, Sidransky D and Baylin SB. (1995). Cancer Papadopoulos N, Nicolaides NC, Wei YF, Rubern SM, Res., 55, 4525 – 4530. Carter KC, Rosen CA, Haseltine WA, Fleischmann RD, Herman JG, Umar A, Polyak K, Graff JR, Ahuja N, Issa JP, FraserCM,AdamsMD,VenterJC,HamiltonSR, Markowitz S, Willson JK, Hamilton SR, Kinzler KW, Petersen GM, Watson P, Lynch HT, Peltomaki P, Mecklin Kane MF, Kolodner RD, Vogelstein B, Kunkel TA and J, Chapelle A, Kinzler KW and Vogelstein B. (1994). Baylin SB. (1998). Proc. Natl. Acad. Sci. USA, 95, 6870 – Science, 263, 1625 – 1629. 6875. Papadopoulos N and Lindblom A. (1997). Human Mutat., Iwakuma T, Sakumi K, Nakatsuru Y, Kawate H, Igarashi H, 10, 89 – 99. Shiraishi A, Tsuzuki T, Ishikawa T and Sekiguchi M. Peters AH, Mermoud JE, O’Carroll D, Pagani M, Schweizer (1997). Carcinogenesis, 18, 1631 – 1635. D, Brockdorff N and Jenuwein T. (2002). Nat. Genet., 30, Jackson JP, Lindroth AM, Cao X and Jacobsen SE. (2002). 77 – 80. Nature, 416, 556 – 560. Prolla TA, Baker SM, Harris AC, Tsao JL, Yao X, Bronner Jenuwein T and Allis CD. (2001). Science, 293, 1074 – 1080. CE, Zheng B, Gordon M, Reneker J, Arnheim N, Shibata Jiricny J. (1998). Mutat. Res., 409, 107 – 121. D, Bradley A and Liskay RM. (1998). Nature Genet., 18, Jones PA. (1985). Cell, 40, 485 – 486. 276 – 279. Jones PA and Laird PW. (1999). Nat. Genet., 21, 163 – 167. Qian XC and Brent TP. (1997). Cancer Res., 57, 3672 – 3677. Jones PA and Baylin SB. (2002). Nat. Rev. Genet., 3, 415 – Rajagopalan H, Bardelli A, Lengauer C, Kinzler KW, 428. Vogelstein B and Velculescu VE. (2002). Nature, 418,934. Jones PL, Veenstra GJ, Wade PA, Vermaak D, Kass SU, Rea S, Eisenhaber F, O’Carroll D, Strahl BD, Sun ZW, Landsberger N, Strouboulis J and Wolffe AP. (1998). Nat. Schmid M, Opravil S, Mechtler K, Ponting CP, Allis CD Genet., 19, 187 – 191. and Jenuwein T. (2000). Nature, 406, 593 – 599. KaneMF,LodaM,GaidaGM,LipmanJ,MishraR, Reik W and Walter J. (2001). Nat. Rev. Genet., 2, 21 – 32. Goldman H, Jessup JM and Kolodner R. (1997). Cancer Rountree MR, Bachman KE and Baylin SB. (2000). Nat. Res., 57, 808 – 811. Genet., 25, 269 – 277. Kat A, Thilly WG, Fang WH, Longley MJ, Li GM and Robertson KD, Ait-Si-Ali S, Yokochi T, Wade PA, Jones PL Modrich P. (1993). Proc. Natl. Acad. Sci. USA, 90, 6424 – and Wolffe AP. (2000). Nat. Genet., 25, 338 – 342. 6428. Russo VEA, Martienssen RA and Riggs AD. (eds) (1996). Kawate H, Sakumi K, Tsuzuki T, Nakatsuru Y, Ishikawa T, Epigenetic Mechanisms of Gene Regulation.NewYork: Takahashi S, Takano H, Noda T and Sekiguchi M. (1998). Cold Spring Harbor Laboratory Press. Proc.Natl.Acad.Sci.USA,95, 5116 – 5120. Sakumi K, Shiraishi A, Shimizu S, Tsuzuki T, Ishikawa T Kawate H, Itoh R, Sakumi K, Nakabeppu Y, Tsuzuki T, Ide and Sekiguchi M. (1997). Cancer Res., 57, 2415 – 2418. F, Ishikawa T, Noda T, Nawata H and Sekiguchi M. Santini V, Kantarjian HM and Issa JP. (2001). Ann. Intern. (2000). Carcinogenesis, 21, 301 – 305. Med., 134, 573 – 586. Knudson AG. (1971). Proc. Natl. Acad. Sci. USA, 68, 820 – Schultz DC, Ayyanathan K, Negorev D, Maul GG and 823. Rauscher III FJ. (2002). Genes Dev., 16, 919 – 932. Koike G, Maki H, Takeya H, Hayakawa H and Sekiguchi M. Shiraishi A, Sakumi K and Sekiguchi M. (2000). Carcinogen- (1990). J. Biol. Chem., 265, 14754 – 14762. esis, 21, 1879 – 1883. Kolodner RD, Hall NR, Lipford J, Kane MF, Mrrison PT, Sklar R and Strauss B. (1981). Nature, 289, 417 – 420. Finan PJ, Burn J, Chapman P, Earabino C, Merchant E Strahl BD and Allis CD. (2000). Nature, 403, 41 – 45. and Bishop T. (1995). Cancer Res., 55, 242 – 248. Tachibana M, Sugimoto K, Fukushima T and Shinkai Y. Lachner M, O’Carroll D, Rea S, Mechtler K and Jenuwein T. (2001). J. Biol. Chem., 276, 25309 – 25317. (2001). Nature, 410, 116 – 120. Tachibana M, Sugimoto K, Nozaki M, Ueda J, Ohta T, Ohki Leach FS, Nicolaides NC, Papadopoulos N, Liu B, Jen J, M, Fukuda M, Takeda N, Niida H, Kato H and Shinkai Y. Parsons R, Peltma¨ ki P, Sistoner P, Aaltoner LA and (2002). Genes Dev., 16, 1779 – 1791. Nystom-Lahti M. (1993). Cell, 75, 1215 – 1225. Tamaru H and Selker EU. (2001). Nature, 414, 277 – 283. Lewis JD, Meehan RR, Henzel WJ, Maurer-Fogy I, Tate PH and Bird AP. (1993). Curr. Opin. Genet. Dev., 3, Jeppesen P, Klein F and Bird A. (1992). Cell, 69, 905 – 914. 226 – 231. Li E, Bestor TH and Jaenisch R. (1992). Cell, 69, 915 – 926. Thorvaldsen JL, Duran KL and Bartolomei MS. (1998). Li E, Beard C and Jaenisch R. (1993). Nature, 366, 362 – 365. Genes Dev., 12, 3693 – 3702. Matsukara S, Soejima H, Nakagawachi T, Yakushiji H, Tsuzuki T, Sakumi K, Shiraishi A, Kawate H, Igarashi H, Ogawa A, Fukuhara M, Miyazaki K, Nakabeppu Y, Iwakuma T, Tominaga Y, Zhang S, Shimizu S, Ishikawa Sekiguchi M and Mukai T. (2003). Brit. J. Cancer, (in T, Nakamura K, Nakao K, Katsuki M and Sekiguchi M. press). (1996). Carcinogenesis, 17, 1215 – 1220.

Oncogene Gene silencing in DNA repair T Mukai and M Sekiguchi 9042 Yang L, Xia L, Wu DY, Wang H, Chansky HA, Schubach Wolf P, Hu YC, Doffek K, Sidransky D and Ahrendt SA. WH, Hickstein DD and Zhang Y. (2002). Oncogene, 21, (2001). Cancer Res., 61, 8113 – 8117. 148 – 152. Walsh CP and Bestor TH. (1999). Genes Dev., 13, 26 – 34. Yang T, Adamson TE, Resnick JL, Leff S, Wevrick R, Watts GS, Pieper RO, Costello JF, Peng YM, Dalton WS Francke U, Jenkins NA, Copeland NG and Brannan CI. and Futscher BW. (1997). Mol. Cell. Biol., 17, 5612 – 5619. (1998). Nat. Genet., 19, 25 – 31. Wutz A, Smrzka OW, Schweifer N, Schellander K, Wagner You M, Candrian U, Maronpot RR, Stoner GD and EF and Barlow DP. (1997). Nature, 389, 745 – 749. Anderson MW. (1989). Proc. Natl. Acad. Sci. USA, 86, Xin Z, Allis CD and Wagstaff J. (2001). Am.J.Hum.Genet., 3070 – 3074. 69, 1389 – 1394. VeiglML,KasturiL,OlechnowiczJ,MaAH,Lutterbaugh Zhang Y and Reinberg D. (2001). Genes Dev., 15, 2343 – JD, Periyasamy S, Li GM, Drummond J, Modrich PL, 2360. Sedwick WD and Markowitz SD. (1998). Proc. Natl. Acad. Sci. USA, 95, 8698 – 8702.

Oncogene