Mutagenesis of theN-(Deoxyguanosin-8-yl)-2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine DNA Adduct in Mammalian Cells

Site-specifically modified oligodeoxynucleotides were used to investigate the mutagenic properties of a major cooked food mutagen-derived DNA adduct,N-(deoxyguanosin-8-yl)-2-amino-1-methyl-6-phenylimidazo[4, 5-b]pyridine (dG-C8-PhIP). dG-C8-PhIP-modified oligodeoxynucleotides were prepared by reacting an oligodeoxynucleotide containing a single dG (5′-TCCTCCTX GCCTCTC, where X = C, A, G, or T) with N-acetoxy-PhIP. The unmodified and dG-C8-PhIP-modified oligomers were inserted into single-stranded phagemid vectors. These single-stranded vectors were transfected into simian kidney (COS-7) cells. The progeny plasmid obtained was used to transform Escherichia coli DH10B. When dC was at the 5′-flanking position to dG-C8-PhIP, preferential incorporation of dCMP, the correct base, was observed opposite the dG-C8-PhIP. Targeted G → T transversions were detected, along with lesser amounts of G → A transitions and G → C transversions. No mutations were detected for the unmodified vector. The influence of sequence context on the dG-C8-PhIP mutation frequency and spectrum was also explored. When the dC 5′-flanking base was replaced by dT, dA, or dG, the mutational spectra were similar to that observed with dC-flanking base. Higher mutational frequencies (28–30%) were observed when dC or dG was 5′ to dG-C8-PhIP. A lower mutational frequency (13%) was observed when dA was at the 5′ to the lesion. Single-base deletions were detected only when dG or dT flanked the adduct. We conclude that dG-C8-PhIP is mutagenic, generating primarily G → T transversions in mammalian cells. The mutational frequency and specificity of dG-C8-PhIP vary depending on the neighboring sequence context.

smoke condensate (13) and is a major mutagenic component recovered from the urine of cigarette smokers (14). This mutagen induces lymphomas in mice (15) and mammary, prostate, and colon carcinomas in rats (16,17). Epidemiological studies indicate that heterocyclic amines, including PhIP, are linked to human colon cancer (18 -20).
Mutational events generated by PhIP have been observed primarily on GC pairs in the dhfr (37) and aprt (38) genes in Chinese hamster ovary cells, in the hgprt gene in Chinese hamster V79 cells (39), and in human lymphoblastoid cells (40). GC 3 TA transversions predominated, followed by GC 3 AT, GC 3 CG, and frameshift (deletion) mutations. Similar mutational spectra were detected in the lacI gene of the Big Blue mouse treated with PhIP (41) and in the APC gene in rat colon tumors induced by PhIP (42).
In this study, oligodeoxynucleotides containing a single dG-C8-PhIP were prepared post-synthetically and inserted into a single-stranded shuttle vector to minimize adduct repair (43). This vector was used to establish the mutagenic specificity and frequency of dG-C8-PhIP in simian kidney (COS-7) cells. In addition, to explore the effect of neighboring bases, mutational specificities were determined with dG-C8-PhIP embedded in different sequence contexts. Our results show that the dG-C8-PhIP-DNA adduct has significant mutagenic potential, generating primarily G 3 T transversions in mammalian cells. The mutational spectra and frequencies vary depending on the neighboring sequence context.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP (specific activity, Ͼ6000 Ci/mmol) was obtained from Amersham Pharmacia Biotech. Escherichia coli DH10B was purchased from Life Technologies, Inc. COS-7 cell line was obtained from the tissue culture facility at SUNY at Stony Brook. EcoRI restriction endonuclease (100 units/l) and T4 DNA ligase (400 units/ l) were obtained from New England Bio Labs.
Preparation of dG-C8-PhIP-N-Hydroxy-PhIP was synthesized by an established method (36). 1 mg of N-hydroxy-PhIP was dissolved in 420 l of Me 2 SO/ethanol (v/v 4:1) and incubated for 5 min at 0°C with 25 l of acetic anhydride for the preparation of N-acetoxy-PhIP. dG (1.5 mg) was reacted at 37°C for 1 h with 30 l of the preparation of N-acetoxy-PhIP in 500 l of 10 mM potassium citrate buffer, pH 6.8. The reaction mixture was evaporated to dryness and subjected to HPLC. A Supelcosil LC-18S column (0.46 ϫ 25 cm, Supelco, Inc.) was eluted with 50 mM ammonium formate, pH 3.5, containing 20% acetonitrile over 3 min, 20 -50% over 5 min, 50% over 5 min, and subsequently 50 -100% over 7 min at a flow rate of 0.8 ml/min. A Waters 990 HPLC instrument equipped with a photodiode array detector was used for isolation of dG-C8-PhIP.
Synthesis of Oligodeoxynucleotides-Unmodified 15-mer oligodeoxynucleotides (5Ј-TCCTCCTXGCCTCTC, where X ϭ C, A, T, or G) were prepared by solid-state synthesis, using an automated DNA synthesizer (44). A 15-mer oligodeoxynucleotide (150 g) was incubated for 1 h at 37°C with 50 l of N-acetoxy-PhIP dissolved in 500 l of 10 mM potassium citrate buffer, pH 6.8, concentrated on Centricon 3, and evaporated to dryness. The dG-C8-PhIP-modified oligomer was isolated from the unmodified oligomer on a Waters reverse-phase Bondapak C 18 column (0.39 ϫ 30 cm), using a linear gradient of 0.05 M triethylammonium acetate, pH 7.0, containing 10 -20% or 10 -30% acetonitrile with an elution time of 60 min and a flow rate of 1.0 ml/min, as described elsewhere (45). These oligomers were further purified by electrophoresis on 20% polyacrylamide gel in the presence of 7 M urea (35 ϫ 42 ϫ 0.04 cm) (46). The oligomers recovered from polyacrylamide The upper strand is a part of ss pMS2 sequence; X represents dG-C8-PhIP. The underlined L13 and R13 probes were used to detect the correct insertion. The underlined S13 of 61-mer scaffold (bottom strand) was used to determine the concentration of ss DNA construct. The probes listed were used for oligodeoxynucleotide hybridization to determine mutation specificity of dG-C8-PhIP. gel electrophoresis were again subjected to HPLC to remove urea.
Site-specific Mutagenesis in COS-7 Cells-SV40-transformed simian kidney cell lines COS-7 and a single-strand (ss) shuttle vector, pMS2 that confers neomycin (Neo R ) and ampicillin resistance (43) were used to establish the mutagenic specificity of dG-C8-PhIP. Construction of a circular ss DNA containing a single dG-C8-PhIP followed procedures established previously in this laboratory (43). pMS2 DNA was annealed to a 61-mer and then digested with EcoRV to create a 15-mer gap (Fig.  2). An unmodified, dG-C8-PhIP-modified 15-mer was ligated to the gapped vector. The ligation mixture was incubated for 2 h with T4 DNA polymerase (1 unit/pmol of DNA) to digest the hybridized 61-mer and then treated with EcoRV and SalI to cleave residual ss pMS2. The reaction mixture was extracted twice with phenol/chloroform (1:1 v/v) and twice with chloroform. Following ethanol precipitation, the DNA was dissolved in distilled water. A portion of the ligation mixture and known amounts of ss pMS2 were subjected to electrophoresis on a 0.9% agarose gel to separate closed circular and linear ss DNA. DNA was transferred to a nylon membrane and hybridized to a 32 P-labeled S13 probe complementary to DNA containing the 15-mer insert. The absolute amount of closed circular ss DNA was established by comparing the radioactivity in the sample with that in known amounts of ss DNA. COS-7 cells were transfected with ss DNA (100 fmol) over 18 h using Lipofectin (47), after which the cells were grown for 2 days in Dulbecco's modified Eagle's medium/10% fetal calf serum. Progeny plasmids were recovered by the method described by Hirt (48), treated with S1 nuclease to digest input ss DNA, and used to transform E. coli DH10B. Transformants were analyzed for mutations by oligodeoxynucleotide hybridization (49,50). The oligodeoxynucleotide probes used to identify progeny phagemids are shown in Fig. 2. Probes L13 and R13 were used to select phagemids containing the correct insert. Transformants that failed to react with L13 and R13 were omitted from the analysis. When L13/R13-positive transformants failed to hybridize to the probes designed to detect events targeted to the lesion site, double-strand DNA was prepared and subjected to dideoxynucleotide sequencing analysis (51).
Preparation of Vectors Containing a dG-C8-PhIP-Unmodified and dG-C8-PhIP-modified oligodeoxynucleotides were ligated into a gapped single-strand vector (Fig. 2). When a part of the ligation mixture was cleaved with BanI and HaeIII restriction enzymes and labeled with 32 P, a 40-mer product was detected on 12% denaturing polyacrylamide gel electrophoresis (Fig. 5). The migration of 40-mers containing a dG-C8-PhIP was slower than that of the unmodified oligomer, as similarly observed for the dG-C8-PhIP-modified 15-mers. The final concentration of ss DNA vector was quantified by Southern blot hybridization. The S13 probe was hybridized to the ligation site of the ss vector (Fig. 2). Using a ␤-phosphorimager, the net product of the closed circular DNA of each constructs was estimated by comparison with variable amount of pMS2 DNA standards. The concentration of the closed circular vector sample was 5.8 -15 ng/l for the unmodified oligomer and 6.7-18 ng/l for the dG-C8-PhIP-modified oligomers, respectively.
The vector modified with dG-C8-PhIP was transfected into simian kidney (COS-7) cells. Progeny phagemid was used to transform E. coli DH10B. Transformants were analyzed for mutations by oligodeoxynucleotide hybridization and by dideoxynucleotide sequencing analysis.

FIG. 4. Enzymatic digestion of dG-C8-PhIP-modified 15-mer.
A, a standard mixture of dC, dG, dT, dA, and dG-C8-PhIP was passed through a Supelcosil LC-18S column. The column was eluted with distilled water over 3 min with 50 mM ammonium acetate, pH 4.5, containing 0 -10% acetonitrile over 15 min and subsequently 10 -100% over 15 min at a flow rate of 1.0 ml/min, as described under "Experimental Procedures." B, 2 g of dG-C8-PhIP-modified 15-mer (5Ј-TC-CTCCTCG PhIP CCTCTC) was digested with nuclease P1 and alkaline phosphatase. Metanol extract of the digested sample was evaporated to dryness and analyzed by HPLC as described under "Experimental Procedures." tioned 5Ј to dG-C8-PhIP, dG-C8-PhIP promoted preferential incorporation of dCMP (72.5%), the correct base, opposite the lesion (Table I). Targeted G 3 T transversions (24.4%) were detected, along with small number of G 3 A transitions (2.6%) and G 3 C transversions (0.5%). In addition, significant numbers of nontargeted mutations representing C 3 T transitions were observed opposite dC 5Ј to the dG-C8-PhIP lesion (Table  I). No mutations were observed with the unmodified vector.
The influence of sequence context on the mutational specificity and frequency of dG-C8-PhIP was explored. When the 5Ј-flanking base was dT, targeted G 3 T base substitutions (14.4%) were detected, accompanied by small number of G 3 A (4.3%) and G 3 C (1.4%) mutations (Table I). In addition, a single-base deletion was detected. The mutational specificity was similar, and the mutational frequency was slightly less under these conditions than when dC was 5Ј to dG-C8-PhIP. When dA was the 5Ј-flanking base, the mutational frequency was 2.2 times less than that observed when dC flanked dG-C8-PhIP (Table I). Preferential G 3 T mutations (11.4%) were observed along with G 3 A transitions (1.3%). No G 3 C mutations were detected.
Oligomers containing a dG-C8-PhIP lesion positioned at codon 60 or 61 of the noncoding strand of the human c-Ha-ras 1 gene was inserted into single-stranded vectors. When dG 5Ј   5. Construction and analysis of shuttle vector. A portion of the vector annealed to the 61-mer scaffold was digested with BanI and HaeIII followed by exchange of the terminal phosphate residue using [␥-32 P]ATP and T4 polynucleotide kinase and subjected to 12% denaturing polyacrylamide gel electrophoresis as described under "Experimental Procedures." site of contiguous dG (5Ј-G PhIP G-) was modified by PhIP, small amounts of G 3 T mutations (7.8%) were detected (Table II). In contrast, when the 3Ј dG (5Ј-GG PhIP -) was modified by PhIP, the mutational frequency increased 3.0 times (Table II). Targeted mutations representing G 3 T transversions (17.4%) were preferentially observed, accompanied by lesser amounts of G 3 A transitions (6.0%) and G 3 C transversions (3.4%). Surprisingly, significant numbers of single-base deletions (3.4%) also were observed.

DISCUSSION
A single-strand plasmid vector was used to explore the mutagenic property of dG-C8-PhIP replicating in COS-7 cells. Targeted G 3 T transversions were detected, along with a small number of G 3 A transitions and/or G 3 C transversion. Our mutational spectra are consistent with those observed in the dhfr gene (37), the aprt gene (38), and hgprt gene (39,40) in mammalian cells exposed to PhIP as well as in the lacI gene of the Big Blue mouse treated with PhIP (41) and in the APC gene in rat colon tumors induced by PhIP (42). Thus, mutations occurring at GC pairs in cells of animals treated by PhIP are most likely due to the major DNA adduct, dG-C8-PhIP.
The mutational potential of dG-C8-PhIP also was determined by replacing the 5Ј-flanking base to the lesion (Fig. 6). Higher mutational frequencies (28 -30%) were observed when dG or dC was at the 5Ј-flanking position to the dG-C8-PhIP, whereas a lower frequency (13%) was observed when dA was 5Ј to the lesion. G 3 T mutations predominate in all sequence contexts, along with lesser amounts of G 3 A mutations. Only when A was the 5Ј-flanking base, the frequency of G 3 A mutations (1.3%) was 2-4.6 times less than that of other flanking bases (2.6 -6.0%), and no G 3 C mutations were observed. Thus, the 5Ј neighboring base influences the mutational frequency and specificity of dG-C8-PhIP.
We also determined the mutational properties of dG-C8-PhIP having the same 5Ј-flanking base but a different 3Јflanking base (5Ј-TG PhIP C-and 5Ј-TG PhIP G-). In both cases, G 3 T mutations predominated, along with a small number of G 3 C mutations and single-base deletions. However, no G 3 A transitions were produced when G 3Ј-flanking base was used. In addition, the mutational frequency of dG-C8-PhIP having C 3Ј position was two times higher than that having G 3Ј position. This indicates that the 3Ј-flanking sequence context also affects the mutational properties of dG-C8-PhIP.
A single-base frameshift (deletion) was observed in dGdC sequences, but not in dAdT sequences (37,39,40). Single-base deletions were frequently detected on the contiguous dGdC sequences of the APC gene (5Ј-GGGA-) in rat colon tumors induced by PhIP (42)   Big Blue mouse treated with PhIP (41). These contiguous dGdC sequences were recognized as mutation hot spots for PhIP (41). In our in vivo studies using 5Ј-CG PhIP C-or 5Ј-AG PhIP C-sequence, no deletions were observed. However, when G was at the 5Јflanking base (5Ј-GG PhIP C-), significant amounts of single-base deletions were detected (Fig. 6). In addition, when T flanked 5Ј to the dG-C8-PhIP (5Ј-TG PhIP C-or 5Ј-TG PhIP G-), a single-base deletion also was observed. Thus, the formation of single-base deletions may be influenced by the neighboring sequence context of dG-C8-PhIP. We have proposed a general frameshift deletion mechanism for chemical carcinogens, including dG-C8-AAF (53). Formation of single-base deletion induced by dG-C8-PhIP is consistent with this in vitro model. During DNA synthesis catalyzed by DNA polymerases on a dG-C8-PhIP-modified template, dCMP was preferentially incorporated opposite the lesion. 2 dCMP inserted opposite dG-C8-PhIP could pair with dG 5Ј to the lesion, forming single-base deletions (Fig. 7, left panel). When dAMP is inserted opposite the lesion, dAMP could pair with dT 5Ј to the lesion, forming single-base deletion (Fig. 7, right panel). The same deletion mechanism may be involved in the formation of single-base deletions observed in the Apc (42) and Iac I (41) genes. The mutagenic specificity of N-(deoxyguanosin-8-yl)-2-aminofluorene (dG-C8-AF) has been established in vivo using a similar sequence context (5Ј-CG AF C-) for dG-C8-PhIP (54). dCMP was incorporated preferentially opposite dG-C8-AF. A small number of targeted incorporation of dAMP (2.0%) and dTMP (1.0%) was also observed. This mutational spectrum is similar to that observed with dG-C8-PhIP. Structural studies show that dG-C8-AF can adopt interconverting anti and syn alignments to pair with an anti-dCMP and an anti-dAMP, respectively (55,56). Based on computational study of the lowest energy structure, dG-C8-PhIP resides in the B-DNA minor groove and adopts a syn conformation to pair with an anti-dAMP (37). Thus, dG-C8-PhIP generally resembles dG-C8-AF (57). However, the mutational frequency of dG-C8-PhIP (27.5%) was nine times higher than that observed in a similar sequence with dG-C8-AF. The nature of structure binding to the dG-C8 position may influence the mutational frequency.
We conclude from this study that dG-C8-PhIP, a major PhIPinduced DNA adduct, is mutagenic lesion in mammalian cells. The mutagenic frequency varies depending on the neighboring sequence context to the lesion. This lesion may be involved in the development of human cancers including colon and breast cancers.