Mammalian DNA repair methyltransferases shield O4MeT from ...

Carcinogenesis vol.18 no.5 pp.919–924, 1997

Mammalian DNA repair methyltransferases shield O4MeT from nucleotide excision repair

Leona Samson1, Song Han, John C.Marquis and Lene Juel Rasmussen2 Department of Molecular and Cellular Toxicology, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115, USA 2Present

address: Department of Chemistry and Life Sciences, Roskilde University, PO Box 260, DK-4000 Roskilde, Denmark 1To

whom correspondence should be addressed

O6-Methylguanine (O6MeG) and O4-methylthymine (O4MeT) are potentially mutagenic DNA lesions that cause G:C→A:T and A:T→G:C transition mutations by mispairing during DNA replication, and the repair of O6MeG and O4MeT by DNA repair methyltransferases (MTases) is therefore expected to prevent methylation-induced transitions. The efficiency of O6MeG and O4MeT repair by different MTases can vary by several hundred-fold and the aim of this study was to establish the biological consequences of such differences in the efficiency of repair. The ability of three microbial and two mammalian MTases to prevent methylation-induced G:C→A:T and A:T→G:C transitions is taken as a measure of their ability to repair O6MeG and O4MeT in vivo respectively. All five MTases give complete protection against G:C→A:T transitions. However, while the microbial MTases give complete protection against A:T→G:C transitions, the mammalian MTases actually sensitize cells to A:T→G:C transitions. We hypothesize that the mammalian MTases bind O4MeT lesions in vivo but that, because they are extremely slow at subsequent methyl transfer, binding shields O4MeT from repair by the nucleotide excision repair pathway. Results are presented to support this hypothesis. Introduction Simple methylating agents produce at least a dozen different DNA lesions (1). Only two of these lesions have been definitively shown to induce mutations, namely guanine methylated at the O6 position (O6-methylguanine, O6MeG*) and thymine methylated at the O4 position (O4-methylthymine, O4MeT) (2,3). Consequently, O6MeG and O4MeT are considered to be the lesions most likely responsible for methylationinduced cancers. During replication O6MeG can pair with thymine and O4MeT can pair with guanine, generating G:C→A:T and A:T→G:C transition mutations respectively (2–4). DNA repair methyltransferases (MTases) that repair O6MeG and O4MeT have been found in both prokaryotic and eukaryotic organisms; their action protects against the cytotoxic, mutagenic and carcinogenic effects of alkylating agents (5–12). All known MTases repair O6MeG lesions efficiently, but the efficiency of O4MeT repair varies dramatically between MTases (5,6,13–15). In this study we explore O6MeG,

O6-methylguanine;

O4MeT,

O4-methylthymine;

*Abbreviations: MTases, DNA repair methyltransferases; NER, nucleotide excision repair; MNNG, N-methyl-N9-nitro-N9-nitrosoguanidine. © Oxford University Press

and discuss the biological consequences of repairing O4MeT at such different rates. In Escherichia coli O6MeG and O4MeT lesions are repaired by two MTases, encoded by the ada and ogt genes (16). Ada and Ogt transfer methyl groups from O6MeG or O4MeT to an active site cysteine residue in the MTase protein itself, to form S-methylcysteine (6). In vitro evidence indicates that Ada displays a striking preference for the repair of O6MeG compared with O4MeT lesions and that this preference is reversed for Ogt (13,17,18). Eukaryotic cells also express DNA repair MTases and, like their bacterial counterparts, they transfer methyl groups from DNA to active site cysteine residues (7– 9,16,19,20). The yeast and mammalian MTases have been shown to recognize both O6MeG and O4MeT in vitro (13–15) but, at least for the mammalian MTases, the rate of methyl transfer appears to be several hundred times slower for O4MeT than for O6MeG (14,15,21). Here we determine whether the slow repair of O4MeT by eukaryotic MTases is biologically significant. In MTase-deficient E.coli the nucleotide excision repair (NER) pathway (initiated by the UvrABC exinuclease) recognizes and repairs O6-alkylguanine and O4-alkylthymine DNA lesions and protects E.coli from alkylation-induced mutation (22,23). This finding suggests that NER and MTases compete with each other for the in vivo repair of O-alkylated bases, although most O6MeG lesions are known to be repaired by MTase if it is available in the cell (22). However, when O6MeG is located at one particular site in the φX174 bacteriophage genome it appears that the slower acting UvrABC exinuclease can actually shield O6MeG from repair by MTase, because the presence of an active NER pathway increases O6MeG mutagenicity (24,25). It has also been suggested that the mammalian NER pathway can recognize and repair O6MeG and O4MeT DNA lesions (26–29). In this study we investigated the ability of five different MTases to repair O6MeG and O4MeT in vivo: three microbial DNA repair MTases (Ada and Ogt from E.coli and Mgt1 from Saccharomyces cerevisiae) and two mammalian MTases (MGMT from humans and Mgmt from mice) were examined. Their ability to prevent methylation-induced G:C→A:T and A:T→ G:C transition mutations in E.coli was used as a measure of their in vivo ability to repair O6MeG and O4MeT respectively. All five MTases carried out efficient in vivo O6MeG repair, but only the microbial MTases carried out efficient O4MeT repair. To our surprise, the human and mouse MTases appeared to inhibit O4MeT repair and we present evidence to support the idea that while mammalian MTases specifically bind O4MeT lesions, such binding actually increases mutation by shielding O4MeT from removal via the NER pathway. Materials and methods Bacterial strains The bacterial strains used in this study are described in Table I. The strains CC102 and CC106 are derivatives of FC218 and FC326 respectively, carrying

919

L.Samson et al.

Table I. Bacterial strains Strain

Chromosomal genotype

Episomal genotype

FC36 FC218 FC326 CJM1 CJM2

ara d(lac proB)XIII thiA RifR As FC36 but ogt-1::Kanr d(ada-25)::Camr As FC36 but ogt-1::Kanr d(ada-25)::Camr As FC326 but uvrB5 Tn10 As FC218 but uvrB5 Tn10

F– lacI:lacZ lacI:lacZ lacI:lacZ lacI:lacZ

lacY1 lacY1 lacY1 lacY1

pro1; pro1; pro1; pro1;

Reference CC102 CC106 CC106 CC102

G:C→A:T A:T→G:C A:T→G:C G:C→A:T

52,53 30 30 This study This study

a F’lac episome. The lacZ mutant alleles in strains CC102 and CC106 can be reverted to the wild-type via G:C→A:T and A:T→G:C transition mutations respectively. The strains CJM1 and CJM2 were constructed by transferring a uvrB5 allele linked to a Tn10 transposon into FC218 and FC326 by phage P1 transduction, selecting for tetracycline resistance. The strains were screened for UV sensitivity to confirm the presence of the uvrB5 mutation in the chromosome. Plasmids Plasmids pUC19 and pALTER-1 were purchased from New England Biolabs and Promega respectively. Construction of pUCogt (pOgt) and pSV2ada (pAda) have been described previously (30,31); these plasmids express the ogt and ada genes under the control of their own promoters. Plasmid pAM1 (pMgt1) was constructed by cloning a 700 bp SalI fragment, which contains the full-length MGT1 coding region, into the SalI site of pALTER-1 in front of the lacZ promoter. Plasmid pSS600 (pMgmt) expressing the mouse MTase from the lac promoter was kindly provided by Dr Sankar Mitra (The University of Texas Medical Branch, Galveston, TX). Plasmid pWX1023 (pMGMT) expressing the human MTase gene from the lac promoter was constructed by cloning a BamHI and HindIII digested PCR fragment containing the MGMT gene into pUC18. The human MGMT cDNA was kindly provided by Dr Mutsuo Sekiguchi (Kyushu University, Japan). Media Bacterial cells were grown in either LB medium or M9 minimal medium. Minimal plates contained either 0.025% glucose or 0.025% lactose, 0.025% thiamine and 40 µg/ml methionine, as previously described (30). N-Methyl-N9-nitro-N9-nitrosoguanidine (MNNG)-induced mutagenesis assay MNNG-induced mutations were measured by growing the cells in LB medium to a cell density of 23108 cells/ml for strains FC218 and CJM2 and 53108 cells/ml for strains FC326 and CJM1. The cells were then exposed to several doses of MNNG (Sigma, St Louis, MO) for 15 min, washed in M9 salts and plated on minimal plates containing either glucose (to estimate the number of viable cells) or lactose (to estimate the number of LacZ1 revertants). The plates were incubated for 2 days and counted. [Note that these experiments do not monitor adaptive mutagenesis as previously described in these strains (30).] Mutation frequencies are expressed as the number of induced LacZ1 revertants per 108 surviving cells for strains FC218 and CJM2 and per 109 surviving cells for strains FC326 and CJM1. Strains containing the E.coli and mammalian MTases were grown at 37°C, while strains containing the yeast MTase were grown at 30°C owing to the temperature sensitivity of the yeast protein. Note that while MNNG-induced mutation may vary from experiment to experiment (because of the relative instability of the compound), within each experiment the strains received MNNG at the same time from the same MNNG stock solution and can thus be directly compared; further, each experiment was performed between three and six times and the results were consistent. Band shift assays The oligonucleotides used in this study had the following sequences: 594 CCGCTAGCGGGTACCGAGCTCGAAT-39 (control); 59-CCGCTO MeAGC6 GGGTACCGAGCTCGAAT-39 (O4MeT); 59-CCGCTAGO MeCGGGTACCGAGCTCGAAT-39 (O6MeG); 39-GGCGATCGCCCATGGCTCGAGCTTA-59 (complementary strand). These oligonucleotides were a kind gift from Manjit Dosanjh and John Essigmann (Massachusetts Institute of Technology) as described in Sassanfar et al. (13). To prepare the labeled oligonucleotide duplexes the complementary strand was labeled with T4 polynucleotide kinase and [γ-32P]ATP and mixed with either unlabeled control, O4MeT or O6MeG oliginucleotides in annealing buffer (20 mM Tris–HCl, pH 7.6, 0.1 mM dithiothreitol, 0.01 mM EDTA). The mixture was heated at 70°C for 10 min and allowed to anneal at room temperature for 30 min. Aliquots of 0.15 pmol labeled oligonucleotide duplex, 3 pmol MGMT protein and 0.3 µg poly(dI·dC) were mixed in MTase buffer (50 mM HEPES–KOH, 5% glycerol, 1 mM EDTA, 10 mM dithiothreitol, pH 7.8) in a volume of 20 µl. After incubation for 30 min at 0°C, 50% (w/v) sucrose solution was added to a final concentration

920

Fig. 1. Microbial MTases prevent MNNG-induced G:C→A:T and A:T→G:C transition mutations. (A) FC218/pUC19 (s) and FC218/pAda (d). (B) FC218/pUC19 (,) and FC218/pOgt (.). (C) FC218/pALTER-1 (u) and FC218/pMgt1 (j). (D) FC326/pUC19 (s) and FC326/pAda (d). (E) FC326/pUC19 (,) and FC326/pOgt (.). (F) FC326/pALTER-1 (u) and FC326/pMgt1 (j).

of 36%. The protein–DNA complexes were resolved by electrophoresis through an 8% non-denaturing polyacrylamide gel at 4°C and then visualized by autoradiography. Purified MGMT protein was a kind gift from Dr Anthony Pegg (University Pennsylvania, Hershey, PA).

Results Microbial MTases repair O6MeG and O4MeT lesions in vivo It was previously shown that the E.coli Ada and Ogt and S.cerevisiae Mgt1 MTases can recognize O6MeG and O4MeT lesions in vitro (6,9,13,32,33). To take these studies one step further we carried out quantitative mutagenesis assays to measure the ability of the microbial MTases to repair O6MeG and O4MeT in vivo, by measuring their ability to prevent methylation-induced G:C→A:T transitions (driven by O6MeG) and A:T→G:C transitions (driven by O4MeT). Mutagenesis was estimated in two ada– ogt– E.coli strains, one containing a lacZ– allele that only reverts to Lac1 via a G:C→A:T transition (FC218) and the other containing a lacZ– allele that only reverts to Lac1 via an A:T→G:C transition (FC326) (Table I; 34). The five MTases were expressed from high copy number plasmids and the influence of each MTase on G:C→A:T and A:T→G:C transition mutations (in FC218 and FC326 respectively) was measured upon exposure to MNNG, a methylating agent known to produce both O6MeG and O4MeT DNA lesions (1). Figure 1 shows that all three microbial MTases conferred substantial protection against MNNGinduced G:C→A:T and A:T→G:C transitions, indicating that

Repair methyltransferases shield O4MeT

Fig. 2. Mammalian MTases prevent MNNG-induced G:C→A:T but not A:T→G:C transition mutations. (A) FC218/pUC19 (s) and FC218/pMGMT (d). (B) FC326/pUC19 (s) and FC326/pMGMT (d). (C) FC218/pUC19 (,) and FC218/pMgmt (.). (D) FC326/pUC19 (,) and FC326/pMgmt (.).

all three MTases efficiently repair both O6MeG and O4MeT in vivo. Expression of mammalian MTases protects against methylationinduced G:C→A:T but not A:T→G:C transition mutations In vitro experiments have shown that the mammalian MTases repair O4MeT lesions several hundred-fold less efficiently than O6MeG lesions (15). We therefore set out to determine what effect such vastly different repair rates have in vivo, and the results were quite surprising. Figure 2 shows that both the human and mouse MTases provide complete protection against MNNG-induced G:C→A:T transitions, presumably via efficient O6MeG repair. In stark contrast, the mammalian MTases did not provide any protection against MNNG-induced A:T→G:C transitions, indicating that they do not carry out efficient O4MeT repair in vivo. Moreover, expression of the human (MGMT) and mouse (Mgmt) MTases actually made the cells more sensitive to methylation-induced A:T→G:C transition mutations (Figure 2). We postulated that the mammalian MTases can bind O4MeT lesions in vivo but cannot carry out efficient methyl transfer and, further, that the bound MGMT and Mgmt proteins shield the O4MeT lesions from repair by an alternative pathway, most likely the NER pathway. The nucleotide excision repair pathway prevents methylationinduced G:C→A:T and A:T→G:C transition mutations In vivo DNA repair assays, using monoclonal antibodies to detect specific alkylated bases, have shown that in the absence of Ada and Ogt, the NER pathway can remove O6MeG, O6-

ethylguanine and O4-ethylthymine from the E.coli genome (22). However, to our knowledge, there is no in vivo or in vitro evidence to show that the NER pathway can repair O4MeT DNA lesions in E.coli. If O4MeT lesions are subject to NER then one would predict that Uvr– E.coli would be more sensitive than Uvr1 to methylation-induced A:T→G:C transition mutations. Figure 3 shows that the introduction of a uvrB5 mutant allele into ada– ogt– E.coli made the cells more sensitive to MNNG-induced A:T→G:C transition mutations (Figure 3A). The uvrB5 allele also sensitized ada– ogt– E.coli to G:C→A:T transitions, as would be expected from previous measurements of O6MeG repair in NER-deficient cells (Figure 3B) (22). These results provide in vivo evidence that the E.coli NER pathway removes O4MeT and O6MeG lesions at biologically significant rates. Human MTase (MGMT) binds O4MeT DNA lesions in vitro As mentioned, the mammalian MTases carry out the in vitro repair of O4MeT lesions very inefficiently (15). Such inefficient repair could result from the following: (i) poor O4MeT recognition; (ii) poor methyl transfer; (iii) both. To test whether mammalian MTases can bind an O4MeT lesion in DNA we studied the interaction of human MGMT with a doublestranded oligonucleotide containing a single O4MeT lesion (13). Figure 4 shows that MGMT binds the O4MeT-containing substrate and the O6MeG-containing substrate, but binds poorly to the oligonucleotide containing perfectly complementary bases. These qualitative results demonstrate that MGMT does indeed bind O4MeT DNA lesions and suggests that inefficient O4MeT repair may be due primarily to inefficient methyl transfer. Mammalian MTase expression renders cells functionally Uvr– for methylation-induced A:T→G:C transition mutations If the mammalian MTases effectively shield O4MeT lesions from repair via the NER pathway (initiated by the UvrABC excinuclease) then MGMT expression should have the following effects: (i) when expressed in wild-type cells, MGMT should produce a Uvr– phenotype for methylation-induced A:T→G:C mutation; (ii) when expressed in Uvr– cells, MGMT should have no effect on methylation-induced A:T→G:C transitions (unless MTase binding also shields O4MeT against repair by other pathways). Figure 5Ashows that MGMT expression does indeed induce a Uvr– phenotype and Figure 5B shows that MGMT expression in Uvr– cells has no influence on the induction of A:T→G:C transitions. Taken together our results provide strong evidence to support the model shown in Figure 6. Here we suggest that the human and mouse DNA repair MTases efficiently bind O4MeT lesions in vivo but, since methyl transfer is inefficient, such binding shields the lesions from excision repair and consequently increases the chance that O4MeT lesions will be replicated, thus increasing the frequency of methylation-induced A:T→G:C transition mutations. In contrast, mammalian MTases rapidly transfer methyl groups from O6MeG lesions, thus reducing the chance of O6MeG lesions inducing G:C→A:T transitions upon DNA replication. Discussion The observation that a purified DNA repair protein is or is not capable of repairing a particular DNA lesion in vitro often helps in determining the biological role of that protein, but this information alone cannot provide a complete picture of 921

L.Samson et al.

Fig. 3. Evidence for the repair of O4MeT DNA lesions by the NER pathway. MNNG-induced A:T→G:C and G:C→A:T mutations were measured in the following strains. (A) FC326/pUC19 (,) and CJM1 (uvrB)/ pUC19 (.). (B) FC218/pUC19 (s) and CJM2 (uvrB)/pUC19 (d).

Fig. 5. The increase in A:T→G:C transition mutations caused by mammalian MTases is dependent on a functional NER pathway. MNNGinduced A:T→G:C mutations were measured in the following strains. (A) FC326/pUC19 (s), FC326/pMGMT (d) and CJM1 (uvrB)/pUC19 (u). (B) FC326/pUC19 (s), CJM1 (uvrB)/pMGMT (m) and CJM1 (uvrB)/ pUC19 (u).

Fig. 4. The human DNA repair MTase (MGMT) binds O4MeT-containing DNA. 32P-Labeled oligonucleotide duplex (0.15 pmol) containing either no modified bases (lane 1), O4MeT (lane 2) or O6MeG (lane 3) DNA lesions were incubated with human MGMT protein (3 pmol) and the resulting DNA–protein complexes were visualized after separation on an 8% polyacrylamide gel.

its in vivo role. DNA repair assays have been so well developed that they may detect in vitro activities that have little physiological relevance or they may detect extremely inefficient activities that produce unexpected physiological effects, as demonstrated in this study. In addition, many DNA lesions may be acted upon by more than one DNA repair pathway and it has become clear that we need to establish the biological consequences of different repair pathways competing for the same lesions. Moreover, gene therapy studies are currently underway on the introduction of DNA repair genes into specific cells and tissues (10,11,35–47). Such gene therapy will be used in an attempt to correct specific DNA repair defects or simply to boost the endogenous DNA repair capacity. Our understanding of which DNA repair pathways recognize 922

Fig. 6. A model for processing of O4MeT DNA lesions in vivo. Mammalian MTase (MGMT/Mgmt) binds O4MeT DNA lesions (T*) and shields these from NER. Upon replication the O4MeT containing strand has the opportunity to mispair with guanine to ultimately introduce a transition mutation in some of the progeny.

and repair particular DNA lesions, and how these pathways interact in vivo, has therefore become extremely important. It is well established that the microbial and mammalian DNA repair MTases can recognize both O6MeG and O4MeT lesions in vitro (5–12) and we set out to compare the ability of each protein to repair these lesions in vivo; their expression in E.coli provided a single physiological environment in which each protein could be compared under the same conditions. Several studies have also shown that NER can also participate in the repair of O6MeG in E.coli and mammalian cells (22,23,26–29), and it has been suggested that O4MeT lesions

Repair methyltransferases shield O4MeT

are recognized by the mammalian NER pathway (29). Here we show that inactivation of NER sensitized E.coli to methylationinduced G:C→A:T and A:T→G:C mutations, confirming that O6MeG lesions are subject to NER and establishing that O4MeT lesions are also a NER substrate. Our results also showed however, that in some instances the binding of a DNA repair MTase to its substrate could interfere with NER. Perhaps the first example of one DNA repair pathway influencing the activity of another was the demonstration that the E.coli photolyase repair enzyme stimulates repair of cyclobutane pyrimidine dimers even in the absence of the light required for photolyase activation; the evidence suggests that dimer-bound photolyase facilitates dimer repair via the E.coli NER pathway (48–51). Here we present an example where the non-productive binding of a DNA repair protein to a lesion hinders its repair by NER. In vivo binding of the human and mouse DNA repair MTases to O4MeT lesions in E.coli appears to hinder O4MeT repair by UvrABC-initiated excision repair and so instead of protecting cells, these repair proteins sensitize cells to the induction of A:T→G:C transition mutations by methylating agents. The mammalian MTases did not sensitize cells that lack NER and from this we infer that MTase binding does not hinder any other O4MeT repair pathway. It is now extremely important to establish whether DNA repair MTases interfere with the in vivo repair of O4MeT in human and mouse cells, and to determine whether the production of excess MTase (e.g. from retroviral expression vectors) in mammalian tissues has any unexpected effects on spontaneous mutation and on the susceptibility of those cells to mutations induced by simple methylating agents, particularly those agents used for the chemotherapeutic treatment of cancer patients. Acknowledgements We thank Dr Sankar Mitra for giving us the mouse MGMT cDNA. This study was supported by National Institute of Environmental Health Sciences grant PO1-ES03926 and National Cancer Institute grant CA 55042. S.H. and J.C.M. were supported by a pre-doctoral training grant (ES07155). L.S. is a Burroughs Wellcome Toxicology Scholar.

References 1. Singer,B. and Grunberger,D. (1983) Molecular Biology of Mutagens and Carcinogens. Plenum Press, New York, NY, pp. 55–78. 2. Loechler,E.L., Green,C.L. and Essigmann,J. (1984) In vivo mutagenesis by O6-methylguanine built into a unique site in a viral genome. Proc. Natl Acad. Sci. USA, 81, 6271–6275. 3. Preston,B.D., Singer,B. and Loeb,L.A. (1986) Mutagenic potential of O4methylthymine in vivo determined by an enzymatic approach to sitespecific mutagenesis. Proc. Natl Acad. Sci. USA, 83, 8501–8505. 4. Singer,B. and Dosanjh,M.K. (1990) Site-directed mutagenesis for quantitation of base–base interactions at defined sites. Mutat. Res., 233, 45–51. 5. Koike,G., Maki,H., Takeya,H., Hayakawa,H. and Sekiguchi,M. (1990) Purification, structure and biochemical properties of human O6methylguanine-DNA-methyltransferase. J. Biol. Chem., 265, 14754–14762. 6. Lindahl,T., Sedgwich,B., Sekiguchi,M. and Nakabeppu,Y. (1988) Regulation and expression of the adaptive response to alkylating agents. Annu. Rev. Biochem., 57, 133–157. 7. Sakumi,K., Shirashi,A., Hayakawa,H. and Sekiguchi,M. (1991) Cloning and expression of cDNA for rat O6-methylguanine-DNA methyltransferase. Nucleic Acids Res., 19, 5597–5601. 8. Shiraishi,A., Sakumi,K., Nakatsu,Y., Haykawa,H. and Sekiguchi,M. (1992) Isolation and characterization of cDNA and genomic sequences for mouse O6-methylguanine-DNA-methyltransferase. Carcinogenesis, 13, 289–296. 9. Xiao,W., Derfler,B., Chen,J. and Samson,L. (1991) Primary sequence and biological functions of a Saccharomyces cerevisiae O6-methylguanine/ O4-methylthymine DNA repair methyltransferase gene. EMBO J., 10, 2179–2186. 10. Dumenco,L.L., Allay,E., Norton,K. and Gerson,S.L. (1993) The prevention

of thymic lymphomas in transgenic mice by human O6-alkylguanine-DNA alkyltransferase. Science, 259, 219–222. 11. Zaidi,N.H., Allay,E., Ayi,T.C., Li,B.F.L., Dumenco,L.L., Sy,M.S. and Gerson,S.L. (1995) The immature thymocyte is protected from Nmethylnitrosourea-induced lymphomas by the human MGMT-CD2 transgene. Carcinogenesis, 16, 1047–1053. 12. Tsuzuki,T. et al. (1996) Targeted disruption of the DNA repair methyltransferase gene renders mice hypersensitive to alkylating agent. Carcinogenesis, 17, 1215–1220. 13. Sassanfar,M., Dosanjh,M.K., Essigmann,J.M. and Samson,L. (1991) Relative efficiencies of the bacterial, yeast and human DNA methyltransferases for the repair of O6-methylguanine and O4methylthymine. J. Biol. Chem., 266, 2767–2771. 14. Graves,R.J., Li,B.F. and Swann,P.F. (1987) Repair of synthetic oligonucleotides containing O6-methylguanine, O6-ethylguanine and O4methylthymine, by O6-alkylguanine-DNA alkyltransferase. IARC Sci. Publ., 84, 41–43. 15. Zak,P., Kleibl,K. and Laval,F. (1994) Repair of O6-methylguanine and O4methylthymine by the human and rat O6-methylguanine-DNAmethyltransferases. J. Biol. Chem., 269, 730–733. 16. Friedberg,E.C., Walker,G.C. and Siede,W. (1995) DNA Repair and Mutagenesis. ASM Press, Washington, DC. 17. Wilkinson,M.C., Potter,P.M., Cawkevell,L., Georiadis,P., Patel,D., Swann,P.F. and Margison,G.P. (1989) Purification of the E.coli ogt gene product to homogenity and its rate of action on O6-methylguanine, O6ethylguanine and O4-methylthymine in deoxyribonucleotides. Nucleic Acids Res., 17, 8475–8484. 18. Graves,R.J., Li,B.F. and Swann,P.F. (1989) Repair of O6-methylguanine, O6-ethylguanine, O6-isopropylguanine and O4-methylthymine in synthetic oligodeoxynucleotides by Escherichia coli ada gene O6-alkylguanineDNA-alkyltransferase. Carcinogenesis, 10, 661–666. 19. Hayakawa,H., Koike,G. and Sekiguchi,M. (1990) Expression and cloning of complementary DNA for a human enzyme that repairs O6-methylguanine in DNA. J. Mol. Biol., 213, 739–747. 20. Potter,P.M., Rafferty,J.A., Cawkwell,L., Wilkinson,M.C., Cooper,D.P., O’Connor,D.J. and Margison,G.P. (1991) Isolation and cDNA cloning of a rat O6-alkylguanine-DNA alkyltransferase gene, molecular analysis of expression in rat liver. Carcinogenesis, 12, 727–733. 21. Dolan,M.E. and Pegg,A.E. (1985) Extent of formation of O4-methylthymine in calf thymus DNA methylated by N-methyl-N-nitrosourea and lack of repair of this product by rat liver O6-alkylguanine-DNA-alkyltransferase. Carcinogenesis, 6, 1611–1614. 22. Samson,L., Thomale,J. and Rajewsky,M.F. (1988) Alternative pathways for the in vivo repair of O6-alkylguanine and O4-alkylthymine in Escherichia coli: the adaptive response and nucleotide excision repair. EMBO J., 7, 2261–2267. 23. Voigt,J.M., Housen,B.V., Sancar,A. and Topal,M.D. (1989) Repair of O6methylguanine by ABC excinuclease of Escherichia coli in vitro. J. Biol. Chem., 264, 5172–5176. 24. Chambers,R.W., Sledziewska-Gojska,E., Hirani-Hojatti,S. and BorowyBorowski,H. (1985) uvrA and recA mutations inhibit a site-specific transition produced by a single O6-methylguanine in gene G of bacteriophage phiX174. Proc. Natl Acad. Sci. USA, 82, 7173–7177. 25. Rossi,S.C., Cornad,M., Voigt,J.M. and Topal,M.D. (1989) Excision repair of O6-methylguanine synthesized at the rat H-ras N-methyl-N-nitrosourea activation site and introduced into Escherichia coli. Carcinogenesis, 10, 373–377. 26. Bronstein,S.M., Cochrane,J.E., Craft,T.R., Swenberg,J.A. and Skopek,T.R. (1991) Toxicity, mutagenicity and mutational spectra of N-ethyl-Nnitrosourea in human cell lines with different DNA repair phenotypes. Cancer Res., 51, 5188–5197. 27. Bronstein,S.M., Skopek,T.R. and Swenberg,J.A. (1992) Efficient repair of O6-ethylguanine, but not O4-ethylthymine or O2-ethylthymine, is dependent upon O6-alkylguanine-DNA alkyltransferase and nucleotide excision repair activities in human cells. Cancer Res., 52, 2008–2011. 28. Huang,J.C., Hsu,D.S., Kazantsev,A. and Sancar,A. (1994) Substrate spectrum of human excinuclease: repair of abasic sites, methylated bases, mismatches and bulky adducts. Proc. Natl Acad. Sci. USA, 91, 12213–12217. 29. Klein,J.C. et al. (1994) Role of nucleotide excision repair in processing of O4-alkylthymines in human cells. J. Biol. Chem., 269, 25521–25528. 30. Mackay,W.J., Han,S. and Samson,L. (1994) DNA alkylation repair limits spontaneous base substitution mutations in Escherichia coli. J. Bacteriol., 176, 3224–3230. 31. Rebeck,G.W. and Samson,L. (1991) Induced spontaneous mutation and alkylation sensitivity of Escherichia coli strains lacking the ogt O6-

923

L.Samson et al. methylguanine DNA repair methyltransferase. J. Bacteriol., 173, 2068– 2076. 32. Rebeck,G.W., Smith,C.M., Goad,D.L. and Samson,L. (1989) Characterization of the major DNA repair methyltransferase activity in unadapted Escherichia coli and identification of a similar activity in Salmonella typhimurium. J. Bacteriol., 171, 4563–4568. 33. Sassanfar,M. and Samson,L. (1990) Identification and preliminary characterization of an O6-methylguanine DNA repair methyltransferase in the yeast Saccharomyces cerevisiae. J. Biol. Chem., 265, 20–25. 34. Cupples,C. and Miller,J. (1989) A set of lacZ mutations in Escherichia coli that allow rapid detection of each of the six base substitutions. Proc. Natl Acad. Sci. USA, 86, 5354–5349. 35. Allay,J.A., Dumenco,L.L., Koc,O.N., Liu,L. and Gerson,S.L. (1995) Retroviral transduction and expression of the human alkyltransferase cDNA provides nitrosourea resistance to hematopoietic cells. Blood, 85, 3342–3351. 36. Carreau,M., Quilliet,X., Eveno,E., Salvetti,A., Danos,O., Heard,J.M., Mezzina,M. and Sarasin,A. (1995) Functional retroviral vector for gene therapy of xeroderma pigmentosum group D patients. Hum. Gene Ther., 6, 1307–1315. 37. Dumenco,L.L., Donovan,C., Warman,B., Clapp,D.W., Lim,I.K., Yun,J., Wagner,T., Hanson,R.W. and Gerson,S.L. (1990) Transgenic mice expressing the bacterial O6-alkylguanine-DNA alkyltransferase gene: an animal model to study the role of DNA repair in the carcinogenesis of Nnitroso compounds. Prog. Clin. Biol. Res., 340A, 369–378. 38. Dumenco,L.L., Arce,C., Norton,K., Yun,J., Wagner,T. and Gerson,S.L. (1991) Enhanced repair of O6-methylguanine DNA adducts in the liver of transgenic mice expressing the ada gene. Cancer Res., 51, 3391–3398. 39. Harris,L., Marathi,U., Edwards,C., Houghton,P., Srivastava,D., Vanin,E., Sorrentino,B. and Brent,T. (1995) Retroviral transfer of a bacterial alkyltransferase gene into murine bone marrow protects against chloroethylnitrosourea cytotoxicity. Clin. Cancer Res., 1, 1359–1368. 40. Moritz,T., MacKay,W., Glassner,B.J., Williams,D.A. and Samson,L. (1995) Retrovirus-mediated expression of a DNA repair protein in bone marrow protects hematopoietic cells from nitrosourea-induced toxicity in vitro and in vivo. Cancer Res., 55, 2608–2614. 41. Maze,R., Carney,J.P., Kelley,M.R., Glassner,B.J., Williams,D.A. and Samson,L. (1996) Increasing DNA repair methyltransferase levels via bone marrow stem cell transduction rescues mice from the toxic effects of 1,3-bis(2-chloroethyl)-1-nitrosourea, a chemotherapeutic alkylating agent. Proc. Natl Acad. Sci. USA, 93, 206–210. 42. Nakatsuru,Y., Matsukuma,S., Sekiguchi,M. and Ishikawa,T. (1991) Characterization of O6-methylguanine-DNA methyltransferase in transgenic mice introduced with the E. coli ada gene. Mutat. Res., 254, 225–230. 43. Nakatsuru,Y., Matsukuma,S., Nemoto,N., Sugano,H., Sekiguchi,M. and Ishikawa,T. (1993) O6-Methylguanine-DNA methyltransferase protects against nitrosamine-induced hepatocarcinogenesis. Proc. Natl Acad. Sci. USA, 90, 6468–6472. 44. Walsh,C.E., Nienhuis,A.W., Samulski,R.J., Brown,M.G., Miller,J.L., Young,N.S. and Liu,J.M. (1994) Phenotypic correction of Fanconi anemia in human hematopoietic cells with a recombinant adeno-associated virus vector. J. Clin. Invest., 94, 1440–1448. 45. Walter,C.A. et al. (1993) Brain and liver targeted overexpression of O6methylguanine DNA methyltransferase in transgenic mice. Carcinogenesis, 14, 1537–1543. 46. Wang,G., Weiss,C., Sheng,P.H. and Bresnick,E. (1996) Retrovirus-mediated transfer of the human O6-methylguanine-DNA methyltransferase gene into a murine hematopoietic cell line and resistance to the toxic effects of certain alkylating agents. Biochem. Pharmacol., 51, 1221–1228. 47. Zaidi,N.H., Pretlow,T.P., O’Riordan,M.A., Dumenco,L.L., Allay,E. and Gerson,S.L. (1995) Transgenic expression of human MGMT protects against azoxymethane-induced aberrant crypt foci and G to A mutations in the K-ras oncogene of mouse colon. Carcinogenesis, 16, 451–456. 48. Hayes,J.B., Martin,S.J. and Bhatia,K. (1985) Repair of nonreplicating UVirradiated DNA: cooperative dark repair by Escherichia coli uvr and phr functions. J. Bacteriol., 161, 602–608. 49. Sancar,G.B. and Smith,F.W. (1989) Interactions between yeast photolyase and nucleotide excision repair proteins in Saccharomyces cerevisiae and Escherichia coli. Mol. Cell. Biol., 9, 4767–4776. 50. Yamamoto,K., Fujiwara,Y. and Shinagawa,H. (1983) Evidence that the phr1 gene enhances the ultraviolet resistance of Escherichia coli recA strains in the dark. Mol. Gen. Genet., 192, 282–284. 51. Yamamoto,K., Fujiwara,Y. and Shinagawa,H. (1984) A multicopy phrplasmid increases the ultraviolet resistance of a recA strain of Escherichia coli. Mutat. Res., 131, 11–18.

924

52. Cairns,J. and Foster,P.L. (1991) Adaptive reversion of a frameshift mutation in Escherichia coli. Genetics, 128, 695–701. 53. Coulondre,C. and Miller,J.H. (1977) Genetic studies of the lac repressor. III. Additional correlation of mutational sites with specific amino acid residues. J. Mol. Biol., 117, 525–567. Received on December 2, 1996; revised on January 10, 1997; accepted on January 13, 1997