Msh2 DNA Mismatch Repair Gene Deficiency and the Food-Borne Mutagen 2-Amino-1-Methy1-6-Phenolimidazo

Msh2 DNA mismatch repair gene de®ciency and the food-borne mutagen 2-amino-1-methy1-6-phenolimidazo [4,5-b] pyridine (PhIP) synergistically aect mutagenesis in mouse colon

Shulin Zhang1, Ruth Lloyd1, Greg Bowden1, Barry W Glickman1 and Johan G de Boer*,1

1Centre for Environmental Health, Department of Biology, University of Victoria, Victoria, BC, V8W 3N5, Canada

Msh2 de®ciency and food-borne carcinogen PhIP have nolimidazo[4,5-b] pyridine (PhIP) is one of the most been implicated as genetic and environmental factors, abundant heterocyclic amines. PhIP induces colon respectively, in human colon carcinogenesis. It is not tumors in male rats, induces predominantly mammary clear whether loss of one or both alleles of Msh2 gene tumors in female F344 rats (Ito et al., 1991), and increases the mutational sensitivity in colon when aberrant crypt foci (ACF), a pre-neoplastic lesion in exposed to environmental carcinogens. In the current the colon of rats as well as in mice (Takahashi et al., study, Msh2+/7/lacI and Msh27/7/lacI double transgenic 1991; Tudek et al., 1989; Sorensen et al., 1997). PhIP mice were treated with PhIP and mutations in the lacI has been shown to be a potent colon mutagen in mice gene were studied in the colon. The spontaneous and rats; it induces signi®cantly elevated mutation mutation frequency (MF) is approximately eightfold frequency and a characteristic mutational spectrum higher in Msh27/7 mice than in Msh2+/+ mice, while (Okonogi et al., 1997a,b). Msh2+/7 mice display similar levels of spontaneous Genetic defects have been suspected to be involved mutation as the Msh2 wild type mice. PhIP induced a in colorectal cancer for several decades (Lynch et al., signi®cant increase in MF in all genotypes of mice. 1985). In recent years, hereditary non-polyposis color- However, induced MF is much higher in Msh27/7 mice ectal cancer (HNPCC) has been attributed to a compared to Msh2+/+ and Msh2+/7 mice. Msh2+/7 mice de®ciency in DNA mismatch repair genes, mainly the displayed an increased level of G:C4T:A transversions Msh2 and Mlh1 (Fishel et al., 1994; Bronner et al., and 71 frameshifts upon PhIP treatment. In contrast, 1994; Leach et al., 1993). A de®ciency in mismatch loss of both Msh2 alleles mainly results in increased repair function has also been implicated in the frequency of G:C4A:T transitions when exposed to pathogenesis of sporadic colorectal cancers (Fornasarig PhIP. These results suggest that a defect in mismatch et al., 2000). When Msh2 nullizygous mice were treated repair may result in an enhanced sensitivity from with the DNA methylating agent N-methyl-N-nitro- exposure to a dietary carcinogen. It also provides insight sourea (MNU), hypermutability was detected in small into interaction between genetic and environmental intestine and thymus (Andrew et al., 1998), but no factors in human carcinogenesis. Oncogene (2001) 20, increased mutability was detected in heterozygous 6066 ± 6072. mice. Recently, Glaab et al. (2000) addressed the cytotoxic and mutagenic response and mutational Keywords: mismatch repair; dietary mutagen; muta- speci®city in human cancer cell lines defective in tional spectra; lacI; cancer Msh2, Mlh1 and Msh6 genes after PhIP treatment (Glaab and Skopek, 1999). It is not known at present whether Msh2 de®ciency confers an increased suscept- Introduction ibility when exposed to speci®c colon mutagens or carcinogens in mammals. Meanwhile, it is important to Colorectal cancer is a prevalent malignancy in the determine whether a population with germline or western world. It accounts for 20 000 deaths in the UK somatic mutations in DNA mismatch repair genes is per year (Toft and Arends, 1998) and is the second at increased risk of developing colon cancers when most common cause of death from cancer in the US exposed to dietary colonic carcinogens. (Boring et al., 1994). Dietary factors, especially Mice harboring the Eschericia coli lacI gene have heterocyclic amines present in cooked meat, have been been used as a mammalian mutation assay for several implicated in the etiology of colon cancer in humans years (Andrew et al., 1996; de Boer and Glickman, (Felton et al., 1997; Gooderham et al., 1997). Among 1998; Dycaico et al., 1994). In this study we used the food-borne carcinogens, 2-amino-1-methyl-6-phe- Msh2/lacI double transgenic mice carrying a dierent functional status of the Msh2 gene as well as the lacI mutational reporter gene. The role of the Msh2 gene in mutagenesis could thus be evaluated by determining *Correspondence: JG de Boer; E-mail: [email protected] Received 13 April 2001; revised 5 June 2001; accepted 14 June mutations in the bacterial lacI gene (Andrew et al., 2001 1998). Effect of mismatch repair on PhIP induced mutation S Zhang et al 6067 PhIP signi®cantly increased the MF of G:C4T:A Results transversions from 0.9+0.361075 to 3.1+0.561075 (P=0.030) as well as 71 frameshifts from Mutation frequency 0.8+0.361075 to 2.2+0.361075 (P=0.039). PhIP Table 1 shows the mutation frequency (MF(+s.e.m.)) in caused a signi®cant increase in G:C4A:T transition three genotypes of mice in both control and PhIP treated in the Msh2 nullizygous mice, from 20.1+3.061075 to groups. In Msh2 pro®cient mice, the spontaneous MF in 33.4+3.061075 (P=0.022). the colon is 5.3+0.361075. PhIP increased the MF to The PhIP induced MF (MF in PhIP treatment group 14.0+2.061075 (P=0.027). Msh2 heterozygous mice 7MF in control group) of G:C4A:T transitions, demonstrate similar levels of spontaneous and PhIP G:C4T:A transversions, and 71 frameshifts in the induced MF as in wild type mice, 6.8+0.861075 and three genotypes are shown in Figure 1. The induced 12.6+0.861075 in the control and PhIP treated group, frequency of G:C4A:T transitions in Msh27/7 mice is respectively, similarly, PhIP caused a signi®cant increase signi®cantly higher than in the other two genotypes, in MF in the Msh2+/7 mice (P=0.029). The spontaneous 12.9+2.861075 vs 2.4+0.861075 and 1.9+0.761075 MF in the Msh2 homozygous de®cient mice was in the wild type and heterozygous mice, respectively 44.0+4.061075, signi®cantly higher than in the wild (P50.010). However, PhIP seems to induce equal type and heterozygous mice. This frequency increased to levels of G:C4T:A transversions in all three genotypes 70.6+3.661075 upon exposure to PhIP (P=0.016). of mice, the induced frequencies being 3.4+1.061075 To measure the mutagenicity caused by PhIP in in Msh2+/+ mice, 2.0+0.561075 in Msh2+/7 mice and dierent mutational backgrounds, the induced muta- 2.2+0.661075 in Msh27/7 mice. PhIP induced fre- tion frequency (InMF) (MF in the treatment group7 quency of 71 frameshifts is also higher in Msh27/7 MF in the control group) was used. This induction is mice (3.4+1.561075) than in the wild type and 8.6+2.161075 and 5.7+0.961075 in the Msh2 heterozygous mice (0.6+0.261075 and 1.4+0.361075 pro®cient and heterozygous mice respectively, however, respectively), however, due to its large variation, this it signi®cantly increased to 26.0+4.161075 in the increase is not signi®cantly dierent. Msh27/7 mice (P50.009, Bonferonni correction) (Table 1). Discussion Mutational specificity The Msh2 gene is one of several genes involved in Tables 2 ± 4 show the mutational speci®city of human DNA mismatch repair (Modrich, 1997; Marra spontaneous and PhIP treated groups in three dierent and Boland, 1996; Eshleman and Markowitz, 1996). Its Msh2 backgrounds. In Msh2+/+ mice, G:C4A:T product forms a heterodimer with Msh3 or Msh6 transitions and G:C4T:A transversions are the two (Acharya et al., 1996; Marsischky et al., 1996). The major spontaneous events. PhIP caused a signi®cant Msh2 protein plays an important role in the recogni- increase in both type of substitutions, increasing them tion of various mismatched base pairs and initiates the from 2.4+0.261075 and 1.4+0.161075 to mismatch repair process (Fishel and Wilson, 1997). A 4.8+0.761075 (P=0.046) and 4.9+1.061075 de®ciency in this gene has been identi®ed as an (P=0.034), respectively. In Msh2 heterozygous mice, important genetic factor involved in the pathogenesis

Table 1 Mutation frequency (MF) in the colon of dierent genotypes of mice No. of Control PhIP PhIP induced Genotype animal pfua No. of mutantsb MF (61075) pfu No. of Mutant MF (61075)MF(61075)c

Msh2+/+ 1 1634000 91 5.6 1286200 254 19.8 2 935600 55 5.9 1194100 114 9.6 3 1154300 51 4.4 1104500 113 10.2 4 943500 140 14.8 5 957200 146 15.3 mean+s.e.md 5.3+0.3 14.0+2.0 8.6+2.1 Msh2+/7 1 860500 57 6.6 747900 79 10.6 2 1025700 48 4.7 797600 100 12.5 3 1231400 105 8.5 901100 129 14.3 4 827500 58 7.0 mean+s.e.m. 6.8+0.8 12.6+0.8 5.7+0.9 Msh27/7 1 1040400 586 56.3 1298200 1044 80.4 2 964000 427 44.3 888100 629 70.8 3 1022600 383 37.5 1308300 897 68.6 4 923800 342 37.0 978400 590 60.3 mean+s.e.m. 44.0+4.0 70.6+3.6 26.0+4.1 apfu: plaque forming unit. bNumber of mutants was corrected for potential clonal expansion (identical mutations identi®ed from same animal were considered to result from clonal expansion). cInduced MF de®ned as the dierence of MF in PhIP treatment group and control group. s.e.m.: dstandard error of mean

Oncogene Effect of mismatch repair on PhIP induced mutation S Zhang et al 6068 Table 2 Mutational speci®city of Msh2+/+ mice Control group PhIP treatment Mutation type No. of mutantsa %MF+s.e.m. (61075)b No. of mutantsa %MF+s.e.m. (61075)

Transitions G:C4A : T 74 45.4 2.4+0.2 105 36.8 4.8+0.7 A:T4G : C 8 4.9 0.3+0.1 11 3.9 0.6+0.2 Transversions G:C4T : A 44 27.0 1.4+0.1 90 31.6 4.9+1.0 G:C4C : G 12 7.4 0.4+0.1 24 8.4 1.2+0.5 A:T4T : A 2 1.2 0.1+0.1 2 0.7 0.1+0.1 A:T4C4G 3 1.8 0.1+0.1 9 3.2 0.5+0.2 Others +1 frameshifts 1 0.6 0.1+0.1 3 1.1 0.2+0.1 71 frameshifts 5 3.1 0.1+0.1 17 6.0 0.8+0.3 Deletions 7 4.3 0.2+0.1 11 3.9 0.5+0.1 Insertions 5 3.1 0.2+0.1 7 2.5 0.3+0.1 Complex changes 0 0.0 0.0 2 0.7 0.1+0.1 Double substitutions 2 1.2 0.1+0.1 4 1.4 0.3+0.1 Total 163 100 285 100

aNumber of mutants was corrected for potential clonal expansion. bMutation frequency in each class of mutation=overall MF6fraction of a speci®c class of mutation

Table 3 Mutational speci®city of Msh2+/7 mice Control group PhIP treatment Mutation type No. of mutantsa %MF+s.e.m. (61075) No. of mutantsa %MF+s.e.m. (61075)

Transitions G:C4A : T 52 38.0 2.7+0.2 29 35.8 4.5+0.8 A:T4G C 12 8.8 0.6+0.1 4 4.9 0.6+0.6 Transversions G:C4T : A 19 13.9 0.9+0.3 20 24.7 3.1+0.5 G:C4C : G 5 3.6 0.2+0.1 6 7.4 0.9+0.1 A:T4T : A 3 2.2 0.2+0.1 2 2.5 0.3+0.1 A:T4C4G 8 5.8 0.4+0.1 3 3.7 0.4+0.2 Others +1 frameshifts 1 0.7 0.1+0.1 0 0.0 0.0 71 frameshifts 18 13.1 0.8+0.3 13 16.0 2.2+0.3 Deletions 11 8.0 0.5+0.2 2 2.5 0.3+0.1 Insertions 3 2.2 0.2+0.1 1 1.2 0.2+0.1 Complex changes 1 0.7 0.1+0.1 0 0.0 0.0 Double substitutions 4 2.9 0.2+0.1 1 1.2 0.2+0.1 Total 137 100 81 100

aNumber of mutants was corrected for potential clonal expansion

Table 4 Mutational speci®city of Msh27/7 mice Control group PhIP treatment Mutation type No. of mutantsa %MF+s.e.m. (61075) No. of mutantsa %MF+s.e.m. (61075)

Transitions G:C4A : T 109 45.6 20.1+3.0 139 47.3 33.4+3.0 A:T4G : C 26 10.9 4.7+0.9 19 6.5 4.1+0.3 Transversions G:C4T : A 15 6.3 2.8+0.7 22 7.5 4.9+1.0 G:C4C : G 6 2.5 0.9+0.3 6 2.0 1.9+0.6 A:T4T : A 5 2.1 0.8+0.4 12 4.1 2.9+0.9 A:T4C4G 7 2.9 1.4+0.4 13 4.4 2.7+0.6 Others +1 frameshifts 11 4.6 1.8+0.4 10 3.4 2.8+0.5 71 frameshifts 40 16.7 7.8+1.8 46 15.6 10.7+1.9 Deletions 13 5.4 2.4+0.3 16 5.4 3.2+0.7 Insertions 0 0.0 0.0 1 0.3 0.2+0.1 Complex changes 6 2.5 1.1+0.4 2 0.7 0.5+0.2 Double substitutions 1 0.4 0.2+0.1 8 2.7 2.6+0.9 Total 239 100 294 100

aNumber of mutants was corrected for potential clonal expansion

Oncogene Effect of mismatch repair on PhIP induced mutation S Zhang et al 6069 study, PhIP induced a signi®cant increase in the frequency of G:C4T:A transversions in wild type mice, consistent with previous reports (Endo et al., 1994; Glaab et al., 2000). In addition, our study shows that the frequency of G:C4A:T transitions also signi®cantly increased after PhIP treatment. A characteristic mutation recovered in rodents after PhIP treatment is 71 frameshifts found at G:C base pairs (Stuart et al., 2000; Zhang et al., 1996). In Msh2+/+ mice, a total of 17 independent 71 frameshifts were identi®ed after PhIP treatment (Table 5), 16 of which occurred at G:C base pairs. In the control group, ®ve 71 frameshift mutations were recovered with only two occurring at G:C base pairs. The Figure 1 The induced mutation frequency of G:C4A:T, portion of 71 frameshifts occurring at G:C base pairs G:C4T:A and 71 frameshifts in three genotypes of mice (error is signi®cantly higher than at A:T base pairs after bars indicate the standard error of mean) PhIP treatment (P=0.023). In Msh2 heterozygous mice, the portion of 71 G:C base pair frameshifts after PhIP treatment is lower than in the Msh2+/+ of HNPCC and other sporadic colorectal cancers PhIP treatment group. However, PhIP increases the (Fishel et al., 1994; Fornasarig et al., 2000; Godard frequency of 71 frameshifts occurring at G:C base et al., 1999). As one of the most abundant and most pairs, from 4/14 in the control group, to 5/8 in the potent heterocyclic amines in our daily diet, PhIP has treatment group, although this increase is marginally been recognized as a possible human colonic mutagen signi®cant (P=0.059). In the Msh27/7 group, the and carcinogen (Ito et al., 1991; Felton et al., 1986; distribution of 71 frameshifts occurring at G:C base Stuart et al., 2000). It is important to address whether pairs and A:T base pairs is similar before and after genetic defects in a population result in an increased PhIP treatment (12/25 and 12/28 in the control and susceptibility to tumorigenesis when individuals are treatment group respectively) (Table 5). One possible exposed to dietary and other environmental carcino- explanation of the dierence between Msh2+/+ and gens. The current study addressed whether a mismatch Msh27/7 mice is that the Msh2 protein repairs mainly repair de®ciency causes hypermutability in mammals insertion/deletion loops formed by adenine or thy- when exposed to a food-borne mutagen. mine. Therefore, in Msh2 de®cient mice, the 71 Our results demonstrate that PhIP is a strong mutagen frameshifts spectrum is a combination of the result of in Msh2 wild type, heterozygous and nullizygous mice. both Msh2 gene de®ciency (resulting in A:T site Msh2 wild type mice displayed a similar level of frameshifts) and PhIP induced frameshifts (at G:C spontaneous mutation frequency (5.3+0.361075)as base pairs). the heterozygous counterpart (6.8+0.861075); the An important issue in understanding HNPCC and spontaneous mutation frequency in the Msh2 nullizy- other sporadic colorectal tumors is whether carriers of gous mice is 44.0+4.061075. After PhIP treatment, the DNA mismatch repair genes have an increased risk of MF signi®cantly increased to 14.0+2.061075, developing cancers. In the current study, the hetero- 12.6+0.861075 and 70.6+3.661075, respectively. PhIP zygous mice did not display an elevated frequency of induced mutation frequencies (InMF) in Msh2+/+ and G:C4A:T transitions or G:C4T:A transversions Msh2+/7 are at similar levels, i.e., 8.6+2.161075 and compared to the wild type group. However, if we 5.7+0.961075, respectively. However, the InMF in- compare the frequency of 71 frameshifts in Msh2 wild creased to 26.0+4.161075 in Msh2 nullizygous mice, a type mice and Msh2 heterozygous mice after PhIP 3 ± 5-fold higher increase compared to that found in the treatment (Tables 2 and 3, Figure 2), the MF in wild type and heterozygous mice. If this environmentally Msh2+/7 mice is signi®cantly higher than in the Msh2+/+ imposed factor causes the same level of increased mice (P=0.025). It is important to note, however, that mutation frequency in the ®ve genes required for the spontaneous frequency of 71 frameshifts in colorectal tumorigenesis (Fearon and Vogelstein, 1990), Msh2+/7 mice is already eightfold higher than in the and the increased mutation frequency does relate to Msh2+/+ mice (0.861075 vs 0.161075). If we use the tumorigenesis as proposed by Loeb (1991), then the induced MF, which eliminates the background fre- Msh2 nullizygous de®ciency will confer a 35 ±55 or 243 to quency, to compare the eect of PhIP in these two 3125 times increase in risk compared to Msh2 pro®cient genotypes of mice, the induced frequency of 71 and heterozygous mice when exposed to PhIP at frameshifts is still twofold higher in heterozygous mice concentrations used in this study. than in wild type mice (1.4+0.361075 vs PhIP induced mutagenicity is mainly caused by the 0.7+0.261075). Recently, Planck et al. (2000) reported bulky adduct N2-(2'-deoxyguanosin-8-yl)-PhIP (dG-C8- that somatic frameshift mutations in mononucleotide PhIP) (Frandsen et al., 1992). This C8 modi®ed repeats have been found in the coding regions of guanine mispairs with adenine resulting in G:C4T:A several genes involved in growth control, apoptosis and transversions (Carothers et al., 1994). In the current DNA repair in HNPCC families. The prevalence of

Oncogene Effect of mismatch repair on PhIP induced mutation S Zhang et al 6070 Table 5 Distribution of 71 frameshifts at C : C/A : T base pairs in single nucleotide insertion/deletion mismatches (Achar- dierent genotypes of mice ya et al., 1996). To a lesser extent, Mutb (Msh2/ G : C/A : T in G : C/A : T in G : C/A : T in Msh3) could repair of 2 ± 8 nucleotides loops (Acharya Msh2+/+ mice Msh2+/7 mice Msh27/7 mice et al., 1996; Genschel et al., 1998; Kolodner and Control group 2/3 4/14 12/25 Marsischky, 1999). Interestingly, some CT dinucleo- PhIP treatment 16/1 8/5 18/28 tide deletions (13 among 239 mutants) at a CTCTCT sequence repeat (position 947/952) were found in Msh27/7 mice, but not in wild type and Msh2+/7 mice. This is consistent with the repair speci®city of the Mutb complex. It could thus be inferred that the defect in the Msh2 gene, a major player in mismatch repair, will induce a wide range of mutations. This may account for the elevation of mutation frequencies in other classes of mutations in addition to G:C4A:T transitions in this study. In vivo mutation frequencies and mutational speci®city in the colon of Msh2 wild type, heterozygous, and nullizygous mice have been determined in this study. Msh2 nullizygous mice have a higher induced mutational frequency after treatment with PhIP Compared to wild type mice, the Msh2+/7 mice demonstrated an increased frequency of 71 frameshifts in both spontaneous and PhIP treatment groups. Based on the prevalence of 71 frameshifts Figure 2 Mutation frequency of G:C4A:T, G:C4T:A and 71 frameshifts after PhIP treatment in Msh2+/+ and Msh2+/7 mice in HNPCC tumors, the current results suggest that the (error bars indicate the standard error of mean) Msh2 carriers are possibly at an increased risk of developing cancers after PhIP exposure. This may provide an animal model for study of pathogenesis frameshifts mutation in various HNPCC populations and prevention of HNPCC. suggests the importance of this kind of mutation in the pathogenesis of HNPCC. We note that the spontaneous mutation frequency in Materials and methods Msh27/7 mice is even higher than the MF in Msh2 pro®cient and heterozygous mice after PhIP treatment. The Msh2/lacI double transgenic mice were kindly provided An Msh2 de®cient cell line also displayed a high MF by Dr Frank R Jirik (University of British Columbia, Vancouver, BC, Canada). These mice are a cross between compared to the wild type cell line (Glaab and Skopek, the BC-1/lacI transgenic and Msh2+/7 mice (Andrew et al., 1999). This may indicate that a de®ciency of the Msh2 1996). According to the commonly used protocol for the gene itself could impose a signi®cant mutational lacI transgenic assay, three to ®ve mice were assigned to burden to the organism. When de®cient organisms each group. At the age of 8 ± 12 weeks, mice from the are exposed to environmental mutagenic compounds, Msh2+/+, Msh2+/7 and Msh27/7 group were treated such as PhIP, these two factors synergistically aect intraperitoneally (i.p) with PhIP (Toronto Research Chemi- mutagenesis in the colon. cals, Toronto, ON, Canada) at a dose of 50 mg/kg, once a As shown in Table 4, the mutational spectrum of week for 4 consecutive weeks. PhIP was dissolved in an Msh27/7 mice displayed a large elevation of the equal amount of dimethyl sulfoxide (DMSO) (Sigma- frequency of G:C4A:T transitions, consistent with a Aldrich) and 0.9% NaCl solution. The mice in control groups were injected with vehicle only. Based on the ®nding high repair eciency of G:T mispairs by Muta (Msh2/ that the same metabolites of PhIP are formed in mice Msh6) in eukaryotes (Kolodner and Marsischky, treated by i.p or per oral (p.o) (Turteltaub et al., 1989) and 1999). However, the frequency of other classes of that the administered dose could be more precisely managed mutation also increased from 2 ± 8-fold compared to by i.p than p.o., we chose intraperitoneal injection as the wild type mice. It was demonstrated (Radman et al., route of exposure. 1985) in E. coli that among eight base-base mis- Twelve weeks after the last treatment, the mice were matches, only C-C is resistant to methyl-directed sacri®ced by carbon dioxide inhalation and cervical mismatch repair. G-T, A-C, A-A and G-G mis- dislocation. Colons were removed, rinsed with phosphate- matches could be repaired eciently, while other buered saline (PBS), ¯ash frozen in liquid nitrogen and mismatches including T-T, T-C or A-G are corrected stored at 7808C. Genomic DNA was isolated by a dialysis puri®cation procedure (Suri et al., 1996). LacI transgenes with poor to good eciency depending on the were recovered from puri®ed mice chromosomal DNA by sequence context (Dohet et al., 1986). Short inser- in vitro l packaging (Stratagene, La Jolla, CA, USA). tion-deletion loops are also good substrates for Packaged phage were plated on an SCS-8 bacterial lawn in mismatch repair (Modrich, 1991). In mammalian cells, the presence of 5-bromo-4-chloro-3-indolyl-b-D-galactopyr- Muta also displayed a wide range of mispair binding anoside (X-gal). Mutant phage appeared as blue plaques, abilities, including all eight base-base mismatches and while the wild type phage were colorless plaques (Rogers et

Oncogene Effect of mismatch repair on PhIP induced mutation S Zhang et al 6071 al., 1995). Mutants were picked and puri®ed by re-plating more than two groups. Mutation frequencies are given with at low density. A 1.5 kb lacI-containing fragment was standard error of mean (s.e.m.). ampli®ed by PCR (Er¯e et al., 1996) and the mutation spectrum was obtained by direct sequencing using LICOR automatic DNA sequencer (Lincoln, NE, USA). Mutation Acknowledgments frequency is calculated as the ratio of total number of The authors thank Dave Walsh, Ken Sojonky, Jana independent mutants and total number of plaques screened. Kangas, and James Holcroft for their technical help; and MF data were analysed using COCHARM (created by Drs Moyra Brackely and Jiangheng He for statistical Troy Johnson, Procter & Gamble, Cincinnati, OH, USA), a analysis and Pauline Tymchuk for editorial assistance. We computer program that executes the General Cochran- also thank Dr Frank R Jirik, Dr Susan E Andrew, and Armitage test (Cariello and Gorelick, 1996). Fisher's exact Agnes Francis for generously providing the experimental test was used for the comparison of mutational changes, and animals. This work was supported by NCIC and the the Boferonni correction was used when comparing data of Cancer Research Society Inc.

References

Acharya S, Wilson T, Gradia S, Kane MF, Guerrette S, Glaab WE, Kort KL and Skopek TR. (2000). Cancer Res., Marsischky GT, Kolodner R and Fishel R. (1996). Proc. 60, 4921 ± 4925. Natl. Acad. Sci. USA, 93, 13629 ± 13634. Glaab WE and Skopek TR. (1999). Carcinogenesis, 20, 391 ± Andrew SE, McKinnon M, Cheng BS, Francis A, Penney J, 394. Reitmair AH, Mak TW and Jirik FR. (1998). Proc. Natl. Godard V, Coulet F, Bernaudin JF, Housset M and Soubrier Acad. Sci. USA, 95, 1126 ± 1130. F. (1999). Ann. Dermatol. Venereol., 126, 600 ± 603. Andrew SE, Pownall S, Fox J, Hsiao L, Hambleton J, Penney Gooderham NJ, Murray S, Lynch AM, Yadollahi-Farsani JE, Kohler SW and Jirik FR. (1996). Mutat. Res., 357, M, Zhao K, Rich K, Boobis AR and Davies DS. (1997). 57 ± 66. Mutat. Res., 376, 53 ± 60. Boring CC, Squires TS, Tong T and Montgomery S. (1994). Ito N, Hasegawa R, Sano M, Tamano S, Esumi H, CA Cancer J. Clin., 44, 7 ± 26. Takayama S and Sugimura T. (1991). Carcinogenesis, 12, Bronner CE, Baker SM, Morrison PT, Warren G, Smith LG, 1503 ± 1506. Lescoe MK, Kane M, Earabino C, Lipford J and Kolodner RD and Marsischky GT. (1999). Curr. Opin. Lindblom A. (1994). Nature, 368, 258 ± 261. Genet. Dev., 9, 89 ± 96. Cariello NF and Gorelick NJ. (1996). Environ. Mol. Leach FS, Nicolaides NC, Papadopoulos N, Liu B, Jen J, Mutagen., 28, 397 ± 404. Parsons R, Peltomaki P, Sistonen P, Aaltonen LA and Carothers AM, Yuan W, Hingerty BE, Broyde S, Grunber- Nystrom-Lahti M. (1993). Cell, 75, 1215 ± 1225. ger D and Snyderwine EG. (1994). Chem. Res. Toxicol., 7, Loeb LA. (1991). Cancer Res., 51, 3075 ± 3079. 209 ± 218. Lynch HT, Schuelke GS, Kimberling WJ, Albano WA, de Boer JG and Glickman BW. (1998). Genetics, 148, 1441 ± Lynch JF, Biscone KA, Lipkin ML, Deschner EE, Mikol 1451. YB and Sandberg AA. (1985). Cancer, 56, 939 ± 951. Dohet C, Wagner R and Radman M. (1986). Proc. Natl. Marra G and Boland CR. (1996). Gastroenterol. Clin. North Acad. Sci. USA, 83, 3395 ± 3397. Am., 25, 755 ± 772. Dycaico MJ, Provost GS, Kretz PL, Ransom SL, Moores JC Marsischky GT, Filosi N, Kane MF and Kolodner R. (1996). and Short JM. (1994). Mutat. Res., 307, 461 ± 478. Genes Dev., 10, 407 ± 420. Endo H, Schut HA and Snyderwine EG. (1994). Cancer Res., Modrich P. (1991). Annu.Rev.Genet.,25, 229 ± 253. 54, 3745 ± 3751. Modrich P. (1997). J. Biol. Chem., 272, 24727 ± 24730. Er¯e HL, Walsh DF, Holcroft J, Hague N, de Boer JG and Okonogi H, Stuart GR, Okochi E, Ushijima T, Sugimura T, Glickman BW. (1996). Environ. Mol. Mutagen., 28, 393 ± Glickman BW and Nagao M. (1997a). Mutat. Res., 395, 396. 93 ± 99. Eshleman JR and Markowitz SD. (1996). Hum. Mol. Genet., Okonogi H, Ushijima T, Zhang XB, Heddle JA, Suzuki T, 5, 1489 ± 1494. Sofuni T, Felton JS, Tucker JD, Sugimura T and Nagao Fearon ER and Vogelstein B. (1990). Cell, 61, 759 ± 767. M. (1997b). Carcinogenesis, 18, 745 ± 748. Felton JS, Knize MG, Shen NH, Lewis PR, Andresen BD, Planck M, Wenngren E, Borg A, Olsson H and Nilbert M. Happe J and Hatch FT. (1986). Carcinogenesis, 7, 1081 ± (2000). Genes Chromosomes Cancer, 29, 33 ± 39. 1086. Radman M, Dohet C, Jones M, Doutriaux MP, Laengle- FeltonJS,MalfattiMA,KnizeMG,SalmonCP,Hopmans Rouault F, Maenhaut-Michel G and Wagner R. (1985). EC and Wu RW. (1997). Mutat. Res., 376, 37 ± 41. Biochimie, 67, 745 ± 752. Fishel R, Lescoe MK, Rao MR, Copeland NG, Jenkins NA, Rogers BJ, Provost GS, Young RR, Putman DL and Short Garber J, Kane M and Kolodner R. (1994). Cell, 77, 167. JM. (1995). Mutat. Res., 327, 57 ± 66. Fishel R and Wilson T. (1997). Curr. Opin. Genet. Dev., 7, Sorensen IK, Kristiansen E, Mortensen A, van Kranen H, 105 ± 113. van Kreijl C, Fodde R and Thorgeirsson SS. (1997). Fornasarig M, Viel A, Valentini M, Capozzi E, Sigon R, De Carcinogenesis, 18, 777 ± 781. Paoli A, Della PL and Boiocchi M. (2000). Oncol. Rep., 7, Stuart GR, Thorleifson E, Okochi E, de Boer JG, Ushijima 39 ± 43. T, Nagao M and Glickman BW. (2000). Mutat. Res., 452, Frandsen H, Grivas S, Andersson R, Dragsted L and Larsen 101 ± 121. JC. (1992). Carcinogenesis, 13, 629 ± 635. Suri A, deBoer J, Kusser W and Glickman BW. (1996). Genschel J, Littman SJ, Drummond JT and Modrich P. Mutat. Res., 372, 23 ± 31. (1998). J. Biol. Chem., 273, 19895 ± 19901.

Oncogene Effect of mismatch repair on PhIP induced mutation S Zhang et al 6072 TakahashiS,OgawaK,OhshimaH,EsumiH,ItoNand Turteltaub KW, Knize MG, Healy SK, Tucker JD and Sugimura T. (1991). Jpn. J. Cancer Res., 82, 135 ± 137. Felton JS. (1989). Food Chem. Toxicol., 27, 667 ± 673. Toft NJ and Arends MJ. (1998). J. Pathol., 185, 123 ± 129. Zhang XB, Felton JS, Tucker JD, Urlando C and Heddle JA. Tudek B, Bird RP and Bruce WR. (1989). Cancer Res., 49, (1996). Carcinogenesis, 17, 2259 ± 2265. 1236 ± 1240.

Oncogene