Error-prone Replication Bypass of the Primary Aflatoxin B1 DNA Adduct, AFB1-N7-Gua*

Background: Aflatoxin B1 exposure causes mutations that are associated with liver cancer. Results: The AFB1-N7-Gua adduct is highly mutagenic in primate cells, with contributions by both replicative and translesion synthesis DNA polymerases. Conclusion: AFB1-N7-Gua adduct is a biologically relevant DNA adduct. Significance: This is the first study demonstrating the mutagenicity of AFB1-N7-Gua in mammalian cells and the identification of candidate DNA polymerases involved in these processes. Hepatocellular carcinomas (HCCs) are the third leading cause of cancer deaths worldwide. The highest rates of early onset HCCs occur in geographical regions with high aflatoxin B1 (AFB1) exposure, concomitant with hepatitis B infection. Although the carcinogenic basis of AFB1 has been ascribed to its mutagenic effects, the mutagenic property of the primary AFB1-DNA adduct, AFB1-N7-Gua, in mammalian cells has not been studied extensively. Taking advantage of the ability to create vectors containing a site-specific DNA adduct, the mutagenic potential was determined in primate cells. This adduct was highly mutagenic following replication in COS-7 cells, with a mutation frequency of 45%. The spectrum of mutations was predominantly G to T base substitutions, a result that is consistent with previous mutation data derived from aflatoxin-associated HCCs. To assess which DNA polymerases (pol) might contribute to the mutational outcome, in vitro replication studies were performed. Unexpectedly, replicative pol δ and the error-prone translesion synthesis pol ζ were able to accurately bypass AFB1-N7-Gua. In contrast, replication bypass using pol κ was shown to occur with low fidelity and could account for the commonly detected G to T transversions.


Hepatocellular carcinomas (HCCs) are the third leading cause of cancer deaths worldwide. The highest rates of early onset HCCs occur in geographical regions with high aflatoxin B 1 (AFB 1 ) exposure, concomitant with hepatitis B infection.
Although the carcinogenic basis of AFB 1 has been ascribed to its mutagenic effects, the mutagenic property of the primary AFB 1 -DNA adduct, AFB 1 -N7-Gua, in mammalian cells has not been studied extensively. Taking advantage of the ability to create vectors containing a site-specific DNA adduct, the mutagenic potential was determined in primate cells. This adduct was highly mutagenic following replication in COS-7 cells, with a mutation frequency of 45%. The spectrum of mutations was predominantly G to T base substitutions, a result that is consistent with previous mutation data derived from aflatoxin-associated HCCs. To assess which DNA polymerases (pol) might contribute to the mutational outcome, in vitro replication studies were performed. Unexpectedly, replicative pol ␦ and the error-prone translesion synthesis pol were able to accurately bypass AFB 1 -N7-Gua. In contrast, replication bypass using pol was shown to occur with low fidelity and could account for the commonly detected G to T transversions.
Hepatocellular carcinomas (HCCs) 2 are the most common form of liver cancer and the third leading cause of cancer death worldwide, with about half a million new cases diagnosed each year. Most cases occur in sub-Saharan Africa, Southeast Asia, and China because of the high prevalence of hepatitis B virus (HBV) infection and dietary exposure to aflatoxin B 1 (AFB 1 ) (1,2). Chronic exposure to AFB 1 in HBV-infected patients has been shown to correlate with onset of HCC 20 years earlier than in individuals who suffer from HBV infection alone (3). Thus, it is essential to understand the mechanism of action of AFB 1 in hepatocellular carcinogenesis for the development of future effective intervention and therapeutic strategies. AFB 1 is the most potent hepatocarcinogen of the known natural aflatoxins and is produced as a secondary metabolite by Aspergillus flavus and Aspergillus parasiticus. These fungi commonly contaminate agricultural crops such as corn and peanuts during their growth in the field and during post-harvest storage (2,4). Upon ingestion, liver metabolic cytochrome P450 enzymes convert AFB 1 to the reactive intermediate, AFB 1 -8,9epoxide (5, 6), which can conjugate on the N7 atom of deoxyguanosine in DNA to form a quantitatively abundant AFB 1 -DNA adduct, 8,9-dihydro-8-(N7-guanyl)-9-hydroxyaflatoxin B 1 (AFB 1 -N7-Gua) (see Fig. 1) (7)(8)(9). This primary DNA adduct is chemically unstable due to the positive charge on the imidazole ring, which further promotes rearrangement to the open ring form of AFB 1 -formamidopyrimidine (AFB 1 -FAPY) or depurination to generate an abasic site (AP) (10,11). Kinetic analyses of the formation and fate of these aflatoxin DNA adducts were determined from rat liver DNA after a single dose exposure. These data revealed that the maximum level of AFB 1 -N7-Gua adducts was attained at or before 2 h, with a half-life of 7.5 h. In 24 h, about 20% of the initial AFB 1 -N7-Gua was converted to the persistent AFB 1 -FAPY adduct, which was continuously detected during the 72-h period studied (12). Thus, the timing of DNA replication in cells relative to the time of the initial adduct formation would significantly determine whether the DNA polymerase (pol) would encounter the AFB 1 -N7-Gua, or AFB 1 -FAPY adducts, AP sites, or a mixture of all types of DNA lesions.
Mutagenesis that arises from replication of a damaged DNA template in tumor suppressor genes or oncogenes is considered to be the initial step of cancer development. Analyses of two surveys conducted in regions with high prevalence of HBV infection and aflatoxin exposure revealed that more than half of the HCC samples contained a G to T mutation at the third position in codon 249 (AGG) of the p53 tumor suppressor gene (13,14). This mutation represents a biomarker of aflatoxin exposure as it is not detected in solely HBV-associated HCCs. Multiple experimental mutagenesis studies have shown that the predominant mutation caused by AFB 1 exposure is G to T transversion (15)(16)(17)(18). Among the three AFB 1 -induced DNA lesions mentioned above, AP sites appear to make only a minor contribution to the observed G to T mutation in Escherichia coli (19). Recently, our group and others have demonstrated that AFB 1 -FAPY adducts are highly mutagenic precursors in both primate cells and E. coli (mutation frequencies of 97 and 32%, respectively), with the G to T transversion being the predominant type of mutation (20,21). However, the primary AFB 1 -N7-Gua adducts appear to be only slightly mutagenic in E. coli, with an error rate of 4% (19). There are no studies regarding the mutagenic potential of AFB 1 -N7-Gua in mammalian cells, thus prompting this investigation.
DNA lesions that stall the progression of replication fork can be bypassed using specialized translesion synthesis (TLS) polymerases. However, this bypass may occur at the detriment of genome stability, with the mutation outcome depending on the type of DNA damage and the TLS polymerase utilized (22)(23)(24). Recently, pol has been suggested to play a role in replication bypass of AFB 1 -FAPY adducts, contributing to the predominant G to T transversions (20). There is only limited information regarding how DNA polymerases process the AFB 1 -N7-Gua adducts. The Klenow fragment (exo Ϫ ) of bacterial DNA pol I could insert either a correct C or a mismatched A opposite AFB 1 -N7-Gua in vitro, but only the mispaired A allowed fulllength extension, a result consistent with the observation of G to T mutations (25). In contrast, the archaea Sulfolobus solfataricus DNA polymerase IV (Dpo4), a homolog of human pol , exhibits error-free bypass of AFB 1 -N7-Gua in vitro (26). The present study was designed to examine the mutagenic potential of AFB 1 -N7-Gua adducts in primate cells using a site-specific mutagenesis approach and subsequently test the potential role of eukaryotic DNA polymerases in replication bypass of this lesion by in vitro replication assays.
Oligodeoxynucleotides-Oligodeoxynucleotides (5Ј-ATAATT-XAATCC-3Ј) adducted with AFB 1 -N7-Gua (see Fig. 1) at the position of X were prepared as described previously (6,30). Nondamaged (ND) oligodeoxynucleotides (12-and 46-mers) with a dG in place of AFB 1 -N7-Gua and primer DNAs were obtained from Integrated DNA Technologies, Inc. The stability of the AFB 1 -N7-Gua adduct in the 12-mer oligodeoxynucleotides was tested by assaying for spontaneous depurination. Single-stranded (ss) DNAs were dissolved in a buffer consisting of 10 mM Tris-HCl (pH 7.5) and 1 mM EDTA, and after 3 months, they were digested with human AP endonuclease in 50 mM HEPES, 150 mM KCl, and 2 mM MgCl 2 . Following electrophoretic separation of the DNAs, there was no evidence of any site-specific depurination of the AFB 1 -N7-Gua site, suggesting that spontaneous depurination rates are negligible under the storage and manipulation conditions used throughout this study.
Site-specific Mutagenesis in COS-7 Cells-The insertion of modified oligodeoxynucleotide (12-mer) containing an AFB 1 -N7-Gua adduct into ss pMS2 shuttle vector, cellular transfection, extraction of replicated progeny, and differential hybridization analyses were performed as reported previously (20). Briefly, the 12-mer oligodeoxynucleotide was ligated into an EcoRV-linearized ss pMS2 vector through the complementarity to the center of a 44-mer scaffold DNA (indicated in bold) (5Ј-CUCGAGGGCCCCUGCAAGCGAUGGAUUCAAUUA-UAUCGCUGGUACCGAGCUCGAAUUC-3Ј), whereas the flanking sequences were annealed to the linearized vector sequences upon EcoRV cleavage of the spontaneous duplex hairpin. Subsequently, the scaffold DNA was removed by uracil DNA glycosylase and T4 DNA polymerase treatments in the absence of dNTPs. Following transfection of site-specifically AFB 1 -N7-Gua-modified ss pMS2 vector into COS-7 cells, replicated DNAs were extracted and used to transform E. coli DH5␣ cells. Individual colonies were grown in liquid culture and lysed, and then the DNA was cross-linked to a Hybond membrane using a UV Stratalinker.
Differential DNA hybridizations were carried out using a total of four 5Ј-radiolabeled probes (5Ј-GATATAATT-NAATCCATCGCTT-3Ј, where N refers to G, T, A, or deletion) at 44°C overnight. To confirm the successful insertion of the 12-mer oligodeoxynucleotides, a bridge probe (5Ј-ATCCATC-GCTTGCAGGGG-3Ј) complementary to the sequence at the joint of the vector and insert was used. DNAs that contained the 12-mer insert, but did not hybridize with any of the mutation or deletion probes, were analyzed by DNA sequencing.
In Vitro DNA Replication Assay-A series of primers whose 3Ј end was designed to hybridize upstream (Ϫ), opposite (0), or downstream (ϩ) of the lesion site was used to anneal to different regions of the 46-mer template as described previously (20) and shown in the figure insets. Three additional variations of the 0 primer were designed to contain a mismatched nucleotide at the 3Ј end (0-A, 0-T, and 0-G primers, respectively). The primers were 32 P-labeled and annealed to the 46-mer template at a 1:2 molar ratio in the presence of 40 mM NaCl, heated at 90°C for 2 min, and cooled to room temperature. The components of the in vitro primer extension reactions for pol ␦, , , and were generally designed from previous studies and included 5 nM primer-template duplex in 25 mM Tris-HCl (pH 7.5), 8 mM MgCl 2 , 10% glycerol, 100 g/ml bovine serum albumin, 8 mM NaCl, and 5 mM dithiothreitol (31)(32)(33). For pol 4 , a different buffer was used, containing 40 mM Tris-HCl, pH 7.6, 0.2 mg/ml bovine serum albumin, 8 mM magnesium acetate, and 120 mM NaCl (29). All the reactions were conducted at 37°C for 30 min unless specified in the figure legends. Concentrations of dNTP(s) and polymerases are given in the figure legends. An equal volume of a solution containing 95% (v/v) formamide, 20 mM EDTA, 0.02% (w/v) xylene cyanol, and 0.02% (w/v) bromphenol blue was added to terminate the reaction. The reaction products were resolved by electrophoresis through 15% acrylamide denaturing gels containing 8 M urea and visualized by a PhosphorImager screen (GE Healthcare).

Mutagenic Potential of AFB 1 -N7-Gua Adducts in Primate
Cells-To assess whether the AFB 1 -N7-Gua adducts (Fig. 1) are mutagenic precursors in mammalian cells, we generated an ss shuttle vector pMS2 containing site-specific AFB 1 -N7-Gua adduct or ND dG control; the vector was subsequently transfected into African green monkey kidney COS-7 cells. The utilization of an ss vector, which was not a suitable substrate for DNA repair, allowed for a reliable determination of the mutagenic frequency and spectra derived from the replication bypass of the lesion. After a 48-h incubation period, the replicated plasmid progenies were extracted and transformed into E. coli to form individual colonies. DNAs extracted from bacteria were further hybridized with various DNA probes that were designed to recognize the accurate or mutated sequence at the AFB 1 -N7-Gua adduct or ND dG insertion region. A bridge probe, spanning both the 12-mer insert and the vector, was used to ensure the successful construction of an ss pMS2 vector containing either AFB 1 -N7-Gua or ND dG. Thus, a mutagenic spectrum could be inferred through the results of differential hybridization. Representative colonies were picked for DNA sequencing to verify the correct identification of the mutation by each sequence-specific probe. If the bacterial DNA only hybridized with the bridge probe, indicative of the presence of the adducted 12-mer, but not with any sequence-specific mutation probes, plasmid DNAs were extracted for DNA sequencing across the region of the insert to determine the sequence of the progeny DNA. Analyses of these data revealed that replication past the AFB 1 -N7-Gua adduct resulted in a high error frequency (ϳ45%) ( Table 1). No mutations were observed with the control sequence ( Table 1). The mutation spectrum included single base substitutions and deletions, with the predominant mutations being G to T transversions, accounting for 82% of all mutations scored. The second most common mutation type was G to A transitions (ϳ10%) followed by G to C transversions and deletions (Ͻ3%). Taken together, these data indicate that AFB 1 -N7-Gua is a highly mutagenic precursor in primate cells, predominantly inducing a G to T transversion, which is consistent with observations of HCC patient samples from areas experiencing high aflatoxin exposure (13,14).
Accurate Replication Bypass of AFB 1 -N7-Gua by pol ␦-In light of the in vivo mutagenesis results, studies were initiated to determine which replicative or TLS DNA polymerases may be involved in synthesis past the AFB 1 -N7-Gua adducts in COS-7 cells using in vitro replication assays. Although the AFB 1 -FAPY adduct was highly effective in blocking replication    templates were efficiently extended, synthesis by pol ␦ was predominantly inhibited one nucleotide prior to AFB 1 -N7-Gua ( Fig. 2A). Surprisingly, unlike comparable replication studies using DNAs containing the AFB 1 -FAPY adduct, pol ␦ generated small amounts of full-length bypass product of AFB 1 -N7-Gua ( Fig. 2A, lane 6). In addition, the presence of the adduct also prevented efficient loading of pol ␦ at low concentrations, as evidenced by significant reductions in primer utilization ( Fig.  2A, comparison of lanes 2 and 5).
Single-nucleotide incorporation reactions were conducted to determine the specificity of the insertion by pol ␦ opposite the lesion. As shown in Fig. 2B, using the Ϫ1 primer, the overall reaction was dominated by the exonucleolytic activity of pol ␦ (as evidenced by significant degradation of the Ϫ1 primers). However, when primer extension was observed, pol ␦ exclusively inserted the correct C opposite AFB 1 -N7-Gua, and fulllength bypass products were observed in the presence of all dNTPs. In reactions containing only dATP or dGTP, the Ϫ1 primer was degraded by the exonuclease proofreading activity of pol ␦ to Ϫ3 or Ϫ4 sites, respectively. In the presence of dTTP, primer degradation paused due to polymerase idling at the Ϫ1 site. These data suggested that under conditions restricting dNTP availability, the exonuclease reaction dominated the polymerization reaction when pol ␦ encountered AFB 1 -N7-Gua adducts. Although these data suggested that pol ␦ insertion was primarily error-free, an additional experimental strategy was used to corroborate this conclusion. Because the two downstream template nucleotides were both T, it was hypothesized that the addition of dATP with the other individual dNTPs in the reaction could facilitate extension beyond the nucleotide that had been inserted opposite the lesion. As shown in Fig. 2C, correct C insertion opposite the lesion followed by incorporation of two additional nucleotides by pol ␦ was observed, yielding a product band at the ϩ2 site (Fig. 2C, lane 4). All other permutations of two nucleotides resulted in exonuclease resection back to the nearest complementary nucleotide (Fig. 2C, lanes 3 and 5). However, when pol ␦-Exo was assayed, it inserted all four dNTPs opposite AFB 1 -N7-Gua and readily extended the misinserted A (Fig. 2E), indicating that the presence of the pol ␦ proofreading function plays a critical role in preventing mutagenesis of AFB 1 -N7-Gua. Although these data suggested that pol ␦ catalyzed high fidelity replication, the results did not eliminate the possibility that if another polymerase inserted an A opposite the adduct, then pol ␦ could extend it, contributing to the predominant G to T mutations induced by AFB 1 -N7-Gua in COS-7 cells (Table 1). To determine whether pol ␦ was capable of extending a primer constructed with a mispaired A opposite the lesion, as compared with a primer with a matched C, primer extension reactions were carried out. Consistent with the data in Fig. 2, B and C, for both primers, the exonucleolytic reaction was more efficient than the extension when replicating through AFB 1 -N7-Gua. However, the mispaired primer was totally resected to the Ϫ1 site (Fig. 2D, lane 6), whereas the matched C primer was more resistant to the exonuclease processing (Fig. 2D, lane 4). Although both reactions resulted in full-length products, the abundance was greater for the matched primer. Given the efficiency of the exonucleolytic reaction on the mispaired primer, it is inferred that the full-length products were generated from the resected Ϫ1 position.
Thus, these data reveal for the first time that pol ␦ can replicate past AFB 1 -N7-Gua by accurate incorporation opposite, and extension beyond, the lesion. These data suggested that pol ␦ could be involved in limiting mutagenesis of AFB 1 -N7-Gua in vivo and contributing to the error-free portion of the replication (Table 1).

Replication Bypass of AFB 1 -N7-Gua by TLS Polymerases-
Given that nearly 50% of the DNA synthesis past AFB 1 -N7-Gua in COS-7 cells was mutagenic (Table 1) and that pol ␦ catalyzed high fidelity bypass of the lesion (Fig. 2), it was hypothesized that one or more of the specialized TLS polymerase(s) may play a role in the AFB 1 -N7-Gua-induced mutagenesis. To determine the identity of the nucleotide inserted opposite the lesion, single-nucleotide incorporation reactions were carried out with yeast pol 4 , human pol , , and .
As shown in Fig. 3A lane 9, pol 4 preferentially inserted the correct C opposite the lesion, whereas A, G, and T were utilized, albeit less efficiently (Fig. 3A, lanes 8, 10, and 11, respectively). With the same experimental approach, pol displayed no nucleotide preference (Fig. 3B, lanes 8-11), whereas pol preferentially incorporated A and G (Fig. 3C). In contrast, pol preferred insertion of C and T (Fig. 3D).
Given that none of the TLS polymerases tested above exhibited selectivity of insertion of dA opposite the AFB 1 -N7-Gua adduct, which could play a major role in contributing to the predominant AFB 1 -N7-Gua-induced G to T mutation, it was hypothesized that TLS polymerases (pols 4 , , , and ) could preferentially extend a mispaired primer opposite the lesion. To experimentally test this, four primers were synthesized such that the 3Ј-nucleotide represented the matched C and the three mismatches (Fig. 4). Comparative analyses revealed that pol 4 extended the accurate base-paired C primer somewhat more efficiently than any of the other three mispaired primers annealed to the damaged template (Fig. 4B).
In the case of pol , although full-length bypass products were detected from all four primers hybridized with the damaged template, there was an increased preference for extension from the mispaired A primer (Fig. 4C, lanes 5-8). However, using conditions to observe these extensions on damaged DNA, all primers were extended on ND templates (Fig. 4C, lanes 1-4). Similar to that observed for pol , pol preferentially catalyzed primer extension from the mismatched A and T primers for the adducted template (Fig. 4D,  lanes 5 and 8). Moreover, a pol -deficient COS-7 cell line was made available 3 to test for evidence for a role of pol in the bypass of this lesion. Using the same site-specific mutagenesis strategy as described above, no differences were observed in the mutation frequency or spectrum in the pol -defective cells as compared with WT cells. The mutation frequency of AFB 1 -N7-Gua was 50% in pol -deficient cells with the predominant mutation type being G to T transversions as 89% of all mutants followed by 11% of G to A transitions.
Consistent with the results obtained for pol 4 , pol preferentially extended the matched C primers using either the ND or the AFB 1 -N7-Gua-modified template, albeit inefficiently on the damaged substrate (Fig. 4E). In summary, these data provide the first biochemical evidence supporting the possible role of pol 4 , in addition to pol ␦, in accurate TLS past AFB 1 -N7-Gua. Pol may contribute to misinsertion opposite the adduct and extension from the mispaired A, thus accounting for the predominant G to T mutation in mammalian cells.
Resumption of Efficient Replication by Replicative Polymerase-To address how close to the adduct normal DNA synthesis by replicative polymerase can resume, primer extension reactions were carried out with pol ␦ using ND or AFB 1 -N7-Gua modified templates annealed to ϩ2, ϩ3, and ϩ5 primers. These primers contained either matched C or mismatched A opposite the lesion, representing the accurate and mutagenic bypass observed in vivo, respectively (Fig. 5). Pol ␦ extended all primers efficiently when the correct C was paired across control dG or the adduct (Fig. 5A, lanes 4 -9). However, a small percentage of the primer annealed to the templates containing the lesion was resected, primarily back to the ϩ1 site, with slightly higher exonucleolytic activity on the ϩ2 primer. Thus, the extension efficiencies increased as the number of correct base pairs beyond the lesion increased. All primers with the ND template were completely utilized. Conversely, minimal amounts of fulllength products were observed when a mispaired A was placed opposite the lesion in the ϩ2 or ϩ3 primer; the exonuclease digested back to the Ϫ1 position (Fig. 5B, lanes 7-8). Interestingly, when the primer terminus was five nucleotides downstream from the mismatched A across the lesion, a substantial amount of full-length extended products was readily detected, with some portion of the ϩ5 primers still excised to the Ϫ1 site (Fig. 5B, lane 9). Moreover, pol ␦ efficiently extended all the mismatched primers annealed to the ND template (Fig. 5B,  lanes 4 -6).
In summary, these data revealed that the position of the mispair across the lesion upstream of the primer terminus could affect the DNA synthesis ability of pol ␦. The balance between the exonuclease and polymerase activities shifts toward extension when the mismatch was located deeper in the duplex, suggesting the requirement of continuous synthesis by TLS polymerase(s) beyond the lesion to reserve the mutagenic bypass before switching to replicative polymerase.

DISCUSSION
Exposure to aflatoxin, the most potent environmental chemical carcinogen, poses a great risk to public health for the devel-opment of HCCs. About 4.6 -28.2% of more than half a million new HCC cases worldwide each year can be attributed to aflatoxin exposure alone (34). Despite intense studies to identify the DNA adducts, mutagenesis, and carcinogenesis induced by aflatoxin, the underlying molecular mechanisms driving these processes are far from completely understood. Recently, we reported that the AFB 1 -FAPY adducts are highly mutagenic in primate cells (20). The current study utilized exactly the same sequence context-and site-specific mutagenesis approach to demonstrate that the primary AFB 1 -DNA adduct, AFB 1 -N7-Gua, was also highly mutagenic (Table 1), although it was about half of that measured for AFB 1 -FAPY adducts (20). Consistent with the observation made for AFB 1 -FAPY, the mutagenicity of AFB 1 -N7-Gua in primate cells was much higher (Table 1) than that in E. coli cells (4%) (19). This underscores the advantage of our biologically relevant system to address the mutagenic effects of aflatoxin in animals and humans.
Furthermore, the predominant G to T transversion at the lesion site measured in primate cells was in agreement with that reported in E. coli (19). The mutation spectrum also correlated well with other mutagenesis studies in various experimental model systems (15,17,18,35,36) and aflatoxin-related HCC samples (13,14). However, no base substitutions located at the 5Ј base of AFB 1 -N7-Gua were detected in our study, suggesting that the mutation at the 5Ј base of the lesion identified in E. coli (19) could be sequence context-dependent; we can rule out the possible effect of DNA adduct formation and DNA repair on the mutagenesis of this lesion because an ss vector that had been modified by the adduct at a specific site was used. Alternatively, it could also be cell type-specific, considering the differences between primate and SOS-induced bacterial DNA polymerases that may be involved in TLS across AFB 1 -N7-Gua adducts. Overall, our results support the following conclusions: 1) AFB 1 -N7-Gua could serve as a mutagenic precursor contributing to the commonly observed G to T transversion as the initiating event of hepatocellular carcinogenesis, and 2) primate cells, and by inference human cells, are more susceptible to the mutagenic effect of AFB 1 than E. coli, thus providing a basis for AFB 1 -associated HCC carcinogenesis. We speculate that the data derived from replication of these single-stranded vectors For this lesion, the model proposes that pol ␦ can replicate past the lesion by correct insertion and extension from the damage site. Alternatively, if pol ␦ is blocked and dissociates one nucleotide prior to the adduct, pol 4 can preferentially insert a correct base opposite and extend from the lesion (green lines). The mutagenic pathway involves the blocking of pol ␦ one nucleotide prior to the lesion followed by recruitment of pol to catalyze the insertion step and the more efficient extension from a mispaired A:G terminus (blue lines). Efficient resumption of normal replication by pol ␦ after the second polymerase switch is proposed to occur at least five nucleotides downstream of the mismatched site.
in African green monkey cells are germane to mutagenic processes in human cells based on a very high degree of sequence identity between human and African green monkey. This assumption is based on unpublished data 4 in which we have cloned and sequenced 16 DNA repair and replication genes from COS-7 cells, and overall, the conservation was 97% for DNA sequence and 98% for amino acid identity.
A model of TLS past AFB 1 -N7-Gua adducts was inferred based on the biochemical studies (Fig. 6). High fidelity replication bypass of the AFB 1 -N7-Gua adduct occurred in about half of the progeny, and these data are consistent with a bypass mechanism mediated by pol ␦ or pol 4 (Fig. 6, left). If, however, the AFB 1 -N7-Gua adduct causes blockage of the replicative polymerase, this could lead to the recruitment of pol that would subsequently catalyze incorporation of all dNTPs opposite the lesion and preferentially extend the mismatched dA (Fig. 6, right). We favor this explanation to account for the G to T transversion over a slippage realignment model, in which the 5Ј-dT would be used as the template for the exclusive insertion of dA. The insertion of all dNTPs equally by pol without preference for dA (Fig. 3B) argues against the slippage realignment model; instead, the preferential extension off of dA by pol could potentially account for the observed mutagenesis. These results will aid us in focusing future studies into the biological role of these polymerases in the bypass of AFB 1 -DNA adducts.
Although both AFB 1 -N7-Gua and AFB 1 -FAPY adducts intercalate at the 5Ј face of the modified base in duplex DNA (37,38), potentially explaining the similar mutation spectrum, with predominant mutation being G to T transversion, it does not provide insight for the differential mutagenic potential between these two lesions (Table 1) (20). Recently, crystallographic studies revealed that the conformation of the AFB 1 moiety within the active site of the archaea polymerase Dpo4 differed between AFB 1 -N7-Gua and AFB 1 -FAPY due to the distinct orientations of the N7-C8 bond of AFB 1 -N7-Gua as compared with the N 5 -C8 bond of AFB 1 -FAPY (26). To keep the N7-C8 bond of AFB 1 -N7-Gua in plane with the adducted dG, the AFB 1 moiety of AFB 1 -N7-Gua is 16°out of plane relative to the modified dG, forming a wedge in the DNA; this is believed to allow better access for the incoming dCTP (26). This could be a plausible mechanism underlying the more accurate bypass of AFB 1 -N7-Gua that was observed in this study. However, bacterial DNA pol I Klenow exo Ϫ has been shown to be capable of inserting both dA and dC opposite AFB 1 -N7-Gua, with the exclusive extension from the mispaired primer template (25). Thus, it warrants further investigation of crystal structures of these eukaryotic polymerases in complex with AFB 1 -N7-Gua adducts to elucidate the mechanisms that may contribute to the mutagenic outcome in mammalian cells.
In the current study, we established that AFB 1 -N7-Gua lesions are biologically relevant DNA adducts causing primarily G to T transversions, the predominant mutation induced by AFB 1 exposure, in primate cells. Furthermore, biochemical evidence of DNA replication bypass of these adducts was pro-vided, and from this, attention could be focused particularly on pol ␦, , and for in vivo assays to advance our understanding of cellular pathways that involve in AFB 1 -induced mutagenesis as the initiation of AFB 1 -related HCC.