|Home | About | Journals | Submit | Contact Us | Français|
Tobacco-specific nitrosamines, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) and N′-nitrosonornicotine (NNN), are considered to be human carcinogens. Both compounds are metabolized to pyridyloxobutylating intermediates that react with DNA to form adducts such as 7-[4-(3-pyridyl)-4-oxobut-1-yl]-guanine (7-pobG), O2-[4-(3-pyridyl)-4-oxobut-1-yl]-cytosine (O2-pobC), O2-[4-(3-pyridyl)-4-oxobut-1-yl]-2′-deoxythymidine (O2-pobdT), O6-[4-(3-pyridyl)-4-oxobut-1-yl]-2′-deoxyguanosine (O6-pobdG) and 4-hydroxy-1-(3-pyridyl)-1-butanone (HPB)-releasing adducts. The role of specific DNA adducts in the overall genotoxic activity of the pyridyloxobutylation pathway is not known. One adduct, O6-pobdG, is mutagenic. To characterize the mutagenic and cytotoxic properties of pyridyloxobutyl DNA adducts, the impact of DNA repair pathways on the cytotoxic and mutagenic properties of the model pyridyloxobutylating agent, 4-(acetoxymethylnitrosamino)-1-(3-pyridyl)-1-butanone (NNKOAc) was investigated in Chinese hamster ovary (CHO) cell lines proficient or deficient in O6-alkylguanine DNA alkyltransferase (AGT), deficient in both AGT and base excision repair (BER), or deficient in both AGT and nucleotide excision repair (NER). The repair of the four pyridyloxobutyl DNA adducts was determined in the same cell lines via sensitive LC-MS/MS methods. NNKOAc was more cytotoxic in the cell lines lacking AGT, BER and NER repair pathways. It also induced more mutations in the hprt gene in the BER and NER deficient cell lines. However, AGT expression did not influence NNKOAc’s mutagenicity despite efficient repair of O6-pobdG. Analysis of the hprt mutational spectra indicated that NNKOAc primarily caused point mutations at AT base pairs. GC to AT transition mutations were a minor contributor to the overall mutation spectrum, providing a rationale for the observation that AGT does not protect against the overall mutagenic properties of NNKOAc in this model system. The only adduct affected by the absence of effective NER was O2-pobdT. Slower repair of O2-pobdT in NER deficient cells was associated with increased AT to TA transversion mutations, supporting the hypothesis that these mutations are caused by O2-pobdT. Together, these data support a hypothesis that the pyridyloxobutylation pathway generates multiple mutagenic and toxic adducts.
The tobacco-specific nitrosamines, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) and N′-nitrosonornicotine (NNN), are potent carcinogens in laboratory animals, generating tumors at sites comparable to those observed in smokers (1). NNK is a powerful lung carcinogen that induces lung adenocarcinomas in rodents at lifetime doses similar to those experienced by smokers (1). NNN is carcinogenic to the esophagus, nasal cavity and respiratory tract (1). This nitrosamine is present in higher amounts than any other esophageal carcinogen in tobacco smoke. As a result of these and related mechanistic studies, NNK and NNN are considered to be human carcinogens (1,2).
Both NNK and NNN require metabolic activation to form DNA adducts. NNK is metabolized to either a methylating agent or a pyridyloxobutylating agent whereas NNN is metabolized to a pyridyloxobutylating agent (Figure 1). While the contribution of O6-methylguanine (O6-mG) to the carcinogenic properties of NNK has been well studied (1), the role of specific adducts in the mutagenic and carcinogenic properties of the pyridyloxobutylation pathway has not been fully investigated. The majority of the pyridyloxobutyl adducts are unstable to DNA hydrolysis conditions and decompose to release 4-hydroxy-1-(3-pyridyl)-1-butanone (HPB) (3,4). Four pyridyloxobutyl DNA adducts have been recently characterized (Figure 1): 7-[4-(3-pyridyl)-4-oxobut-1-yl]-guanine (7-pobG) (5), O2-[4-(3-pyridyl)-4-oxobut-1-yl]-cytosine (O2-pobC) (6), O2-[4-(3-pyridyl)-4-oxobut-1-yl]-2′-deoxy-thymidine (O2-pobdT) (6), and O6-[4-(3-pyridyl)-4-oxobut-1-yl]-2′-deoxyguanosine (O6-pobdG) (5–7). These adducts have been detected in DNA from NNK- or NNN-treated rodents (8–10).
Several lines of evidence indicate that pyridyloxobutyl DNA adducts likely contribute to the carcinogenic properties of these nitrosamines. First, model compounds that only pyridyloxobutylate DNA are potent mutagens in Salmonella tymphimurium (11,12). Second, the model pyridyloxobutylating agent, 4-(acetoxymethylnitrosamino)-1-(3-pyridyl)-1-butanone (NNKOAc), is a lung tumorigen in A/J mice under chronic dosing conditions (13). Third, pyridyloxobutyl DNA adducts accumulate in pulmonary Type II cells and correlate with lung tumor formation in NNK-treated rats (14). Finally, pyridyloxobutyl DNA adducts also build up in normal lung tissues of lung cancer patients (15), indicating that the formation and persistence of these adducts may be important in human lung cancer associated with tobacco use.
The role of specific pyridyloxobutyl DNA adducts in the overall mutagenic and carcinogenic properties of pyridyloxobutylating nitrosamines is unknown. We have demonstrated that O6-pobdG is mutagenic in both bacteria and human cells (16). This adduct produced only GC to AT transition mutations in bacteria whereas in human cells a small number of GC to TA transversions and more complex mutations were also observed. Consistent with the potential importance of this adduct to the mutagenic activity of pyridyloxobutylating agents, GC to AT transitions and GC to TA transversions were observed in the K-ras oncogene of tumors induced by NNKOAc in A/J mouse lungs (13). The mutagenic properties of the other pyridyloxobutyl DNA adducts is largely unknown. A preliminary report describing in vitro polymerase reactions with O2-pobdT indicates that this adduct likely miscodes during DNA replication (17). However, the mutagenic properties of O2-pobdT have not been studied in a cell-based assay. Together, these data indicate that multiple pyridyloxobutyl DNA adducts likely contribute to the overall genotoxic properties of the pyridyloxobutylating metabolites of NNK and NNN.
DNA repair pathways protect the cell against the toxic and mutagenic effects of DNA alkylating agents. One repair protein, O6-alkylguanine DNA alkyltransferase (AGT), is important in the repair of O6-pobdG, reducing the mutagenicity of this adduct (16,18). Pyridyloxobutyl DNA adducts are also repaired by nucleotide excision repair (NER) (19), however, the specific adducts removed by this pathway have not been determined. Base excision repair (BER) is a third potential repair pathway for pyridyloxobutyl DNA adducts, however, there is no information on its importance in the overall repair of these DNA adducts.
In this report, we explore the involvement of AGT, NER and BER in protecting a cell from the mutagenic and cytotoxic properties of NNKOAc. In the presence of esterase, NNKOAc decomposes to the same pyridyloxobutylating agent formed upon methyl hydroxylation of NNK or 2′-hydroxylation of NNN (4,20). These studies were performed with Chinese hamster ovary (CHO) cells in which the frequency of point mutations can be readily determined in the X-linked hypoxanthine-guanine phosphoribosyl transferase (hprt) loci by growing the treated cells in the presence of 6-thio-guanine. Wild-type cells generate the toxic 6-thioguanine monophosphate and are subsequently killed off. This assay allows for the quantitative measurement of mutation frequency as well as the isolation of mutant colonies for mutation spectrum determination. We determined the cytotoxic and mutagenic properties of NNKOAc in five Chinese hamster ovary (CHO) cell lines - AA8, UV5, EM9, CHOpcDNA3 and CHOAGT (Table 1). The parental cell line, AA8, does not express the repair protein AGT (21). UV5 has a mutation in the ERCC-2 gene (XPD) and, as a result, is deficient in NER (22). EM9 lacks XRCC1 and is compromised in BER (23). CHOpcDNA3 cells are derived from AA8 cells and are stably transfected with the empty pCMV-neo-Bam vector (21). CHOAGT cells are stably transfected with the pCMV-hAGT expressing human AGT (21). These cell lines have been used to determine the importance of DNA repair pathways in protecting cells against the cytotoxic and genotoxic effects of a variety of DNA damaging agents (21,24–30). We measured the influence of the repair proteins on the mutational spectrum of NNKOAc as well as the formation and repair of the known pyridyloxobutyl DNA adducts, 7-pobG, O6-pobdG, O2-pobC and O2-pobdT. Collectively, these studies shed new insights into the mutagenic and toxic properties of pyridyloxobutyl DNA adducts.
Caution: NNKOAc is toxic and mutagenic in cells and is carcinogenic in laboratory animals. It should be handled with proper safety equipment and precautions.
NNKOAc, [pyridine-2H4]O2-[4-(3-pyridyl)-4-oxobut-1-yl]cytosine, [pyridine-2H4]7-[4-(3-pyridyl)-4-oxobut-1-yl]guanine, [1,2,2-2H3]O6-[oxo-4-(3-pyridyl)butyl]-2′-deoxyguanosine and [pyridine-2H4]O2-[4-(3-pyridyl)-4-oxobut-1-yl]thymidine were synthesized as previously described (20,31,32). Except where indicated, reagents were from Sigma (St. Louis, MO).
CHO cell lines AA8 (wild-type), UV5 (XPD defects) and EM9 (XRCC1 defects) were obtained from American Type Culture Collection (Manassas, Virginia). These experiments were also performed with AA8 cells stably transfected with the pCMV-neo-Bam vector without an inserted cDNA sequence (CHOpcDNA3) or the same plasmid expressing human AGT (CHOAGT) (21). Western blot analysis confirmed the expression of AGT in CHOAGT cells (21). The AGT activity in the cells was approximately 450 fmol/mg protein (21). The cell lines were maintained in Minimum Essential Medium Alpha Modification (αMEM) supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and 1% penicillin/streptomycin at 37°C in a humidified 5% CO2 incubator. Pre-existing hprt mutants were eliminated by growing the cells in complete medium supplemented with 100 μM hypoxanthine, 0.4 μM aminopterin and 16 μM thymidine as previously described (25).
Cytotoxicity and mutation frequency at the hprt locus was determined according to published methods (33,34). Six-well plates were seeded with 2 × 105 exponentially growing cells 18–24 h prior to treatment. The cells were exposed to 0–48 μM NNKOAc for 1 h. Each treatment was performed in duplicate. The cells were washed with PBS and grown in fresh medium for 16–24 h. The cells were then harvested, counted, and replated for measurement of colony forming ability and mutant frequency measurements. Each experiment was repeated a minimum of three times.
Colony forming ability was determined by plating 100 cells from each treatment into 6-well plates (33). Two wells were plated per treatment. The plating efficiencies were similar for all five cell lines. After 6–7 days growth in normal medium, colonies were stained with 0.05% crystal violet and were counted with an aCOLyte colony counter (Synbiosis, Frederick, MD). Cell survival was normalized to the number of colonies from untreated cells.
For the mutagenesis assay, each treated culture was plated into 60 mm dishes and grown in normal medium, with subculturing, to allow phenotypic expression of induced mutants. After 7 days, the cells were harvested, counted and plated for determination of cloning efficiency and mutant frequency. The cloning efficiency was determined by seeding 200 cells into 60 mm dishes in normal medium. Colonies were stained with crystal violet after 6–7 days growth and counted. The mutant frequency was determined by plating 100,000 cells into 100 mm dishes in medium containing 5 μg/ml 6-thioguanine to select for mutant colonies. Ten dishes per treatment were plated. Colonies were stained after 8 days of growth in the selection medium and counted with an aCOLyte colony counter. The mutant frequency was expressed as the 6-thioguanine resistant colonies per 100,000 viable cells.
The effect of cell type on cytotoxicity was determined using analysis of variance at each concentration of NNKOAc. Since the F-test for the simple effect was significant, pair wise comparisons of UV5 and EM9 with AA8 were conducted. At each step of the analysis, the level of significance was 0.05; the calculated p-values were unadjusted for multiple comparisons. The two-sample t-test was employed to compare mutant frequency between cell lines at each concentration of NNKOAc.
CHO cells (1×106) were seeded into 100 mm dishes 16–24 h prior to treatment. The cells were exposed to 24 μM NNKOAc for 1 h, washed once with PBS and further grown in drug-free fresh medium for 16–24 h. The cells were then harvested and split into 10 × T25 flasks. The cells were grown in normal complete medium for 7 days to allow the phenotypic expression of mutants, with two subcultures during the 7 day period to avoid confluence. Cells were then harvested, counted, and plated into 100 mm dishes (100,000 cells/dish) in medium containing 5 μg/mL 6-thioguanine (selection medium). Two plates were generated from each T25 flask. After 9 days growth in the selection medium, 2 mutant colonies from each plate were picked and transferred to 24 well plates for continued growth in selection medium. The cells in the 24-well plates were harvested and transferred into 100 mm dishes respectively 3 days later, and continued growth in selection medium. Once the cells were 80–90% confluent, they were harvested, counted, and aliquoted into 50,000 cell portions for RT-PCR analysis.
A total of 36 (EM9), 39 (UV5, AA8), 57 (CHOpcDNA3) or 69 (CHOAGT) mutant colonies were picked and expanded for molecular characterization. Control and mutant hprt mRNA was reversed transcribed and amplified for sequence analysis. If two clones from the same flask displayed identical mutations, one of these clones was removed from our analysis since they were likely sibling clones. In samples where the PCR product was absent, truncated or produced multiple amplicons, PCR analysis of the nine hprt exons was performed with genomic DNA (25,26). These latter studies allowed us to determine if the irregularities in the initial reverse transcript PCR reactions resulted from large deletions or mutations resulting in splicing defects.
cDNA was prepared from lysates of mutant clones (5 × 104 cells) with a Cells-to-cDNA II Kit (Ambion, Austin, TX) according to manufacturer instructions. The cDNA was amplified using hprt-specific primers and Platinum Taq DNA Polymerase High Fidelity (Invitrogen). In a first round of PCR, a 789 bp product was generated using primers 1 (hprt positions 23–42: 5′-CTCGGCGCCTCCTCTGCGGG) and 2 (hprt positions 792–811: 5′-GGTAATTTTACTGGGAAC). The PCR cycling conditions were as follows: initial melting step (94 °C, 2 min), 35 cycles of denaturation (94 °C, 40 sec). annealing (55 °C, 40 sec) and extension (72 °C, 1 min) followed by a final extension step at 72 °C for 7 min. PCR products were analyzed on an agarose gel and if a band at 789 bp was present, then the PCR band was purified for sequencing using the QIAquick PCR Purification Kit (Qiagen, Valencia, CA). If the 789 bp band was absent or weak, then a nested PCR was performed with primers 3 (hprt positions 47–66: 5′-CTCCTCACACCGCTCTTCGC) and 4 (hprt positions 764–783: 5′-CTCCTCGTGTTTGCAGATTC). The PCR reactions were performed as described above with the exception that the annealing was performed at 60 °C. The corresponding 737 bp product was excised from the gel and purified using the QIAquick Gel Extraction Kit (Qiagen). All PCR products were sequenced at the Biomedical Genomics Center, University of Minnesota.
In samples where the cDNA PCR product was absent, truncated or produced multiple amplicons, the hprt exons 1–9 were amplified with genomic DNA as the template according to a published method (26). Genomic DNA was isolated from 2–5 × 106 cells with DNeasy Blood and Tissue isolation kit (Qiagen) according to the manufacturer instructions. Multiplex PCR reactions were performed exactly as described (26). The reaction products were visualized on 1.3% agarose gels. These PCR products were not sequenced.
Chi-square analysis and Fisher’s exact test were used for the comparisons of the mutational spectra between cell lines. The p-values were unadjusted for multiple comparisons.
Cells (~1.5 × 106) were seeded into 150 mm dishes. Three days later, the cells were exposed to 0–32 μM NNKOAc for 1 h. Each condition was performed in triplicate. At 0–48 h, cells were harvested and the contents of three plates were pooled for DNA isolation using Nucleobond AX-G DNA isolation kits (Clontech, Palo Alto, CA). Adduct levels were determined with LC-MS/MS according to previously published methods (31,32,35). Levels of 7-pobG, O6-pobdG, O2-pobC and O2-pobdT were measured in enzyme hydrolysates and expressed relative to the 2′-deoxyguanosine concentration in the hydrolysates (31,32). In addition, levels of the nucleobase, O6-[4-(3-pyridyl)-4-oxobut-1-yl]-guanine (O6-pobG), were determined in 0.1 N HCl hydrolysates and expressed relative to the guanine concentration in the samples (35).
A 144-bp duplex with a centrally located O6-pobdG adduct was prepared by annealing and ligation of six complementary and partially overlapping oligomers as described (36). Control substrates, a 144-bp undamaged (UM) duplex and a 136-bp duplex with a T[6-4]T, were prepared in a similar manner. Oligonucleotides employed to build the 144-bp duplex were obtained from Operon Technologies (Alameda, CA) or were prepared in house (37,38). The sequence of the O6-pobdG containing oligonucleotide was AATAGTAGCTG*GAGGC, where *G was O6-pobdG. This oliognucleotide was 5′ end-labeled with γ-32P-ATP and used to generate a 144-bp duplex with a radiolabel positioned 11 phosphodiester bonds 5′ to the damage. The dual incision event (excision) bracketing the lesion on the radiolabeled strand results in release of 24–32-mers containing both the damage and the radiolabel; this repair signal is detected and quantified as described below. The 136-bp duplex with T[6-4]T was prepared in a similar manner using an 8-mer containing the T[6-4]T substrate (36).
The six factors of the human excision nuclease, RPA, XPA, XPC•hR23B, TFIIH, XPF•ERCC1, and XPG, were purified and stored as described (36). Excision assays were performed as described (36) with 10 fmol substrate and the indicated amounts of repair factors in a 12.5 μL reaction. For kinetic assays, 20 fmol substrate in a 25 μL reaction volume was used and aliquots were removed at the indicated time points. Reactions were terminated by proteinase K treatment for 15 min at 55–60° C followed by phenol and phenol:chloroform extractions. DNA was recovered after ethanol precipitation, resuspended in formamide/dye mixture, and resolved in 10% sequencing gels. Excision was visualized by autoradiography and percentage excision was quantified using a PhosphorImager (Storm 860) and ImageQuant program (GE Healthcare, Piscataway, NJ). Excision levels for each reaction were determined as a percentage of radiolabel in the excision product (24–32 nt) region of the gel relative to the total radiolabel in the substrate migrating at 136 or 144 nt plus excision products in that lane.
The cytotoxicity and mutagenic activity of NNKOAc were determined in five CHO cell lines - AA8, UV5, EM9, CHOpcDNA3 and CHOAGT. All five cell lines were treated with increasing concentrations of NNKOAc for 1 h and then plated for assessment of cytotoxicity or mutagenesis. Cell survival was determined via a colony forming assay. As shown in Figure 2A, UV5 and EM9 cells were more sensitive to the cytotoxic effects of NNKOAc than the parental cell line, AA8. In addition, NNKOAc was more toxic in CHOpcDNA3 cells relative to CHOAGT cells (Figure 2C).
To determine the ability of the various repair pathways to influence the frequency of point mutations, the mutation frequency in the hprt gene was measured by selecting mutants with 6-thio-guanine. NNKOAc induced hprt mutations in the AA8 cells in a dose dependent fashion (Figure 2B). The frequency of mutations induced by NNKOAc was elevated in the UV5 cells at the two lowest concentrations and in EM9 at the two highest concentration (Figure 2B). The expression of hAGT did not influence the mutagenic activity of NNKOAc (Figure 2D).
The mutation spectra were characterized in NNKOAc-treated cells by sequencing the hprt cDNA isolated from 6-thioguanine-resistant mutants. In samples where the PCR product was absent, truncated or produced multiple amplicons, PCR analysis of the nine hprt exons was performed with genomic DNA (25,26). These latter studies allowed us to determine if the irregularities in the initial reverse transcript PCR reactions resulted from large deletions or mutations resulting in splicing defects. Results of the sequencing studies are displayed in Supplemental Tables 1–5 and summarized in Supplemental Table 6. The vast majority of hprt mutations induced by NNKOAc were point mutations (Supplemental Table 6). The spectrum of NNKOAc-induced mutations is significantly different than that reported for spontaneous mutations in these cell lines (p-value < 0.001, Table 2) (26,39). Most of these NNKOAc-induced point mutations (≥ 70%) occurred at AT base pairs, with tranversions as the dominant change (Table 2). The majority of these transversions were AT to CG mutations.
Alterations in the DNA repair status of the cell line had modest impact on the distribution of mutations (Table 2). For example, there were significant differences in the extent of AT to CG and AT to TA mutations between the cell lines (p < 0.05). The distribution of NNKOAc-derived mutations was much broader in the NER deficient cells (UV5) with fewer AT to CG mutations and more AT to TA mutations than the cell lines with functional NER (AA8). There were also more AT to TA mutations in the BER deficient EM9 cell line. While there is a trend towards fewer GC to AT transition mutations in CHOAGT relative to the other cell lines, this trend was not statistically significant.
To determine which adducts were differentially repaired in each cell line, levels of O2-pobC, 7-pobG, O2-pobdT and O6-pobdG were measured using established LC-MS/MS methods (31,32). Preliminary studies to investigate cell line differences in initial adduct levels were performed with a 1 h exposure of UV5, AA8, CHOAGT and CHOpcDNA3 cells to 32 μM NNKOAc. All cell lines had similar initial levels of each pyridyloxobutyl DNA adducts, except O6-pobdG (Figure 3). The levels of this adduct were significantly lower in CHOAGT cells, consistent with previous reports that O6-pobdG is readily repaired by AGT (7,18,40).
The time course of adduct removal was determined by exposing the CHO cell lines to 8 μM NNKOAc and then monitoring the disappearance of the four pyridyloxobutyl DNA adducts at 0, 4, 24 and 48 h after exposure. This concentration was chosen since it was only slightly toxic, killing less than 20–25% of cells in each cell line (Figures 2A and 2C). Also, we wanted to ensure that there were functioning repair systems at non-saturating adduct levels. As expected, there was no significant difference in adduct levels between any of the five cell lines immediately after the 1 h exposure. The only exception was O6-pobdG; it was significantly lower in CHOAGT cells. When CHOpcDNA3 and CHOAGT cells were compared, there was no significant difference in the rate of repair of O2-pobC, 7-pobG or O2-pobdT (data not shown). O6-pobdG was completely repaired in CHOAGT cells by 24 h whereas it persisted in CHOpcDNA3 cells (data not shown). The rates of removal of O2-pobC and 7-pobG were similar in AA8, UV5 and EM9 cells (Figure 4). In contrast, O2-pobdT was more persistent in UV5 cells which lack functional NER (Figure 4). There was no apparent difference in the removal of O6-pobdG in AA8, UV5 or EM9 cells. Since there was significant noise in the measurement of O6-pobdG adduct levels in these cell lines, DNA samples from AA8 and UV5 cells were also subjected to acid hydrolysis to measure the levels of the nucleobase, O6-pobG (35). When this method of analysis was employed, the data was cleaner and confirmed there was no significant difference in the rate of disappearance of O6-pobG between cell lines (Figure 4).
Since other large O6-alkylguanine adducts are substrates for NER (41), it was somewhat surprising that the repair of O6-pobdG was unaffected by the loss of ERCC2 activity in UV5 cells. To further explore this possibility, the ability of NER to repair O6-pobdG was determined in vitro. A reconstituted human excision nuclease RFI-VI (repair factors RPA, XPA, XPC, TFIIH, XPG and XPF-ERCC1) was incubated with a 32P-labeled 144-bp duplex containing a centrally located O6-pobdG adduct and the resulting excision products were separated by gel electrophoresis. Analysis of these gels indicated that a small amount of the O6-pobdG had been excised (Figure 5). Levels of excision for O6-pobdG (1.1%) were slightly greater than that observed for unmodified DNA (0.8%) but significantly less than the levels observed for a T[6-4]T photoproduct (32.7%). This excision proceeded in a time-dependent fashion (Supplemental Figure 1). These data confirm the cell-based results by demonstrating that O6-pobdG is a poor substrate for repair by the NER machinery.
The involvement of specific pyridyloxobutyl DNA adducts in the overall cytotoxic and mutagenic properties of a pyridyloxobutylating agent has not been previously investigated. Therefore, we determined the cytotoxic and mutagenic properties of the model pyridyloxobutylating agent, NNKOAc, in CHO cell lines that vary in DNA repair capacity. These effects were then compared to the lifetime of specific pyridyloxobutyl DNA adducts in the same cell lines to determine which of these adducts is responsible for the toxic and mutagenic activities of NNKOAc.
NNKOAc is toxic and mutagenic in CHO cells (Figure 2). The spectrum of mutations induced by NNKOAc in the hprt gene is significantly different from that observed in untreated cells (Table 2). NNKOAc caused mutations primarily at AT base pairs, regardless of repair background whereas the spontaneous mutations occurred primarily at GC base pairs (Table 2). In AA8 cells, AT to CG transversion mutations were the major type of base substitution in contrast to the GC to TA transversions that dominate the spontaneous mutation spectrum (Table 2). In vivo mutagenesis studies with NNK in the Muta™ Mouse and Big Bl™ mouse model systems indicated that AT to TA and/or AT to CG mutations are major mutations observed in the transgene in these animals (35,42). Our data suggest that these mutations may be caused by pyridyloxobutyl DNA adducts.
It is possible that O2-pobdT is responsible for the mutations observed at AT base pairs. In vitro polymerase reactions with O2-pobdT indicate that all four nucleotides can be incorporated opposite this modified base (17). However, we cannot rule out the possible involvement of less stable adenine or thymine adducts for the induction of mutations at AT base pairs since HPB-releasing adducts are a significant contributor to the overall DNA damage formed by pyridyloxobutylating agents (43) and these unstable adducts could form at AT base pairs. Mutations at GC base pairs represented approximately 30% of the NNKOAc-induced hprt mutations. The majority of these mutations are GC to AT transition mutations. Therefore, only 17–25% of the mutations were likely to be caused by O6-pobdG which has been shown to induce primarily GC to AT mutations in human cell lines (16).
The cytotoxic and mutagenic effects of NNKOAc were roughly doubled in the NER-compromised cell line, UV5, relative to AA8 cells. These results are consistent with the report that NNKOAc was modestly more cytotoxic in XPA (2.3 fold) or XPC (1.5 fold) deficient human lymphoid cell lines relative to NER competent cell lines (19). The impact of XPD deficiency on NNKOAc’s mutagenic effects in UV5 cells is comparable to that observed for structurally related alkylating agents, n-butylnitrosourea and ethylnitrosourea (24,29). n-Butylnitrosourea was 1.4-fold more cytotoxic and 40% more mutagenic in UV5 cells as compared with AA8 cells (24). The mutagenic activity of the ethylating agent, ethylnitrosourea increased by ~50% in UV5 cells relative to AA8 while the cytotoxicity was unaffected (29).
It has been reported that pyridyloxobutyl DNA adducts are repaired by NER as judged by the ATP-dependent incorporation of [α32P]dTTP into plasmid DNA upon incubation with extracts from NER competent and deficient human lymphoid cell lines; cell extracts from XPA and XPC-deficient cell lines were less active in their ability to repair pyridyloxobutylated DNA (19). Consistent with that report, our observations using the UV5 cell line suggest that NER may be involved in the repair of cytotoxic and mutagenic pyridyloxobutyl DNA adducts.
The only pyridyloxobutyl adduct that was impacted by the loss of XPD activity in UV5 cells was O2-pobdT (Figure 4). Since the reduced repair of this adduct is accompanied by increased cytotoxicity and mutagenicity, we propose that O2-pobdT is both a cytotoxic and mutagenic pyridyloxobutyl DNA adduct. This adduct is the dominant pyridyloxobutyl adduct detected in lung or liver DNA isolated from NNK or NNN-treated rats (9,10). Therefore, it is likely that this adduct contributes to the carcinogenic properties of pyridyloxobutylating nitrosamines.
There was an increase in the frequency of AT to TA mutations in UV5 relative to AA8 (Table 3), suggesting that NER is involved in the repair of pyridyloxobutyl DNA adducts that trigger this transversion mutation. Since O2-pobdT is poorly repaired in UV5 cells, we hypothesize that this adduct may trigger AT to TA mutations. Consistent with this hypothesis, structurally related O2-ethylthymine triggers AT to TA mutations in site-specific mutagenesis studies (44). Site-specific mutagenesis studies with O2-pobdT are required to directly demonstrate the genotoxicity of this adduct.
Surprisingly, there was no significant difference in the rate of repair of O6-pobdG between parental AA8 and NER-deficient UV5 cells. Consistent with the results in CHO cells, O6-pobdG is a poor substrate for human NER in vitro, although it is a better substrate for this repair pathway than O6-methylguanine (data not shown). These findings differ with those reported for other O6-alkylguanine adducts. For example, the bulky O6-butylguanine is thought to be repaired by nucleotide excision repair (41). It is possible that O6-pobdG is slowly repaired via NER since it does not accumulate in lung DNA in NNK-treated mice lacking AGT (35). This observation is in contrast to the accumulation of O6-mG in the lung DNA of the same animals.
The cytotoxic and mutagenic effects of NNKOAc were roughly doubled in EM9 cells relative to AA8 cells (Figure 2). EM9 cells lack functional XRCC1 which is an important scaffold protein that facilitates numerous steps in BER repair (45). Knockout of this protein sensitizes cells to the DNA damaging effects of a variety of alkylating agents (23). While our studies provide no information as to which pyridyloxobutyl DNA adducts are being repaired by BER, they suggest that DNA pyridyloxobutylation results in the formation of DNA damage that requires XRCC1 activity in order for the cell to escape the cytotoxic and mutagenic effects of this damage.
There are two potential pathways for pyridyloxobutyl DNA adducts to generate DNA damage that could enter into BER. First, there may be specific pyridyloxobutyl DNA adducts that are substrates for BER glycosylases. There is evidence that structurally related O6-butylguanine may be repaired by base excision repair in vivo (46). Therefore, it is possible that one or more of the pyridyloxobutyl DNA adducts are substrates for this group of repair proteins. Second, pyridyloxobutylation of DNA generates abasic sites since both 7-pobG and O2-pobC spontaneously depurinate/depyrimidate from DNA. These resultant abasic sites are expected to be repaired by the BER machinery.
There was an increase in the mutation frequency of AT to TA mutations in EM9 cells relative to AA8 cells (Table 3). Studies performed with site-specific abasic sites indicate that this type of lesion triggers predominantly transversions with AT to TA mutations more prevalent than AT to GC mutations (47). Therefore, we postulate that abasic sites generated either by glycosylase catalyzed removal of a pyridyloxobutylated base or by spontaneous depurination/depyrimidation can contribute to the mutagenic activity of pyridyloxobuylating agents.
The expression of human AGT in the CHO cell lines had a slight impact on the cytotoxicity of NNKOAc. These data suggest that O6-pobdG may be toxic to mammalian cells. This adduct was not toxic in bacteria (18). Surprisingly, there was no effect of AGT expression on the genotoxicity of NNKOAc. These observations contrast dramatically with those reported for methylating agents (48,49) and with results obtained for NNKOAc in bacteria expressing human AGT (18). Analysis for DNA adducts demonstrated that O6-pobdG was almost completely repaired in CHOAGT cells (Figure 3), confirming that this adduct is rapidly repaired by human AGT (18,40). Therefore, it appears that O6-pobdG is not a major contributor to the mutagenic activity of NNKOAc in the hprt gene of CHO cells.
The spectrum of NNKOAc-induced hprt mutations provides some clues to why O6-pobdG does not contribute significantly to the mutational activity of NNKOAc in the CHO model system. The majority of the mutations (83%) in the wild-type cell line, CHOpcDNA3, occur at AT base pairs, not GC base pairs. Therefore, it is likely that the repair of mutagenic O6-pobdG adducts does not affect the overall mutation frequency since they contribute only to a small percentage of the overall mutations. However, there is a decrease in the frequency of transition mutations at GC base pairs in CHOAGT cells (Table 3), consistent with AGT mediated repair of O6-pobdG. Our previous studies in Salmonella typhimurium demonstrated that human AGT protects against the mutagenic activity of NNKOAc (18). This particular model system requires mutagenesis at GC base pairs. Therefore, the ability of AGT to protect against the mutagenic effects of a pyridyloxobutylating agent will be dependent on the mutations required to alter the phenotype of the protein. Consistently, NNKOAc induced GC to AT transitions as well as GC to TA transversions in the twelfth codon of the K-ras oncogene of tumors in A/J mice (13). O6-pobdG is likely responsible for the GC to AT mutations observed in this system.
Overall, BER and NER are important pathways in the protection of cells against the cytotoxic and genotoxic effects of the pyridyloxobutylating pathways of tobacco-specific nitrosamines. It is likely that genetic variations in these repair pathways will influence the effectiveness of this protection and contribute to an individual’s sensitivity to the carcinogenic properties of the tobacco-specific carcinogens NNK and NNN. Our data are consistent with the hypothesis that multiple adducts contribute to the overall mutagenic properties of this pathway. This study suggests that O2-pobdT is a cytotoxic and mutagenic adduct and previous studies demonstrated that O6-pobdG is also a mutagenic adduct (16,18). Which adducts are crucial to the carcinogenic properties of this pathway are likely to vary depending on the system. If mutations at AT base pairs selectively produce proteins with oncogenic function, the formation of O2-pobdT and its repair by NER is probably important for tumor initiation by this pathway. On the other hand, if mutations at GC base pairs are important for initiating tumors, the formation O6-pobdG and its repair by AGT will be important. Future studies will investigate the importance of these repair pathways to protect against the genotoxic effects of pyridyloxobutyl DNA adducts in humans.
The authors would like to thank Dr. Peter Villalta for his assistance with the mass spectral analyses, Choua Vu for the synthesis of NNKOAc and Bob Carlson for his help with the figures. The Analytical Biochemistry Shared Resource is supported in part by Cancer Center Support Grant CA-77598. The research was supported by the following National Institutes of Health grants: CA-115309 (LAP), GM-32833 (AS), CA-018137 (AEP) and CA-081301 (SSH).
Supporting Information Available: Tables describing the locations and types of mutation in the coding region of the hprt gene in mutants derived from NNKOAc-treated CHO cells as well as figures showing the time course of DNA repair in CHO cells and by human excision nuclease.