DNA replication studies of N-nitroso compound–induced O6-alkyl-2′-deoxyguanosine lesions in Escherichia coli

N-Nitroso compounds (NOCs) are common DNA-alkylating agents, are abundantly present in food and tobacco, and can also be generated endogenously. Metabolic activation of some NOCs can give rise to carboxymethylation and pyridyloxobutylation/pyridylhydroxybutylation of DNA, which are known to be carcinogenic and can lead to gastrointestinal and lung cancer, respectively. Herein, using the competitive replication and adduct bypass (CRAB) assay, along with MS- and NMR-based approaches, we assessed the cytotoxic and mutagenic properties of three O6-alkyl-2′-deoxyguanosine (O6-alkyl-dG) adducts, i.e. O6-pyridyloxobutyl-dG (O6-POB-dG) and O6-pyridylhydroxybutyl-dG (O6-PHB-dG), derived from tobacco-specific nitrosamines, and O6-carboxymethyl-dG (O6-CM-dG), induced by endogenous N-nitroso compounds. We also investigated two neutral analogs of O6-CM-dG, i.e. O6-aminocarbonylmethyl-dG (O6-ACM-dG) and O6-hydroxyethyl-dG (O6-HOEt-dG). We found that, in Escherichia coli cells, these lesions mildly (O6-POB-dG), moderately (O6-PHB-dG), or strongly (O6-CM-dG, O6-ACM-dG, and O6-HOEt-dG) impede DNA replication. The strong blockage effects of the last three lesions were attributable to the presence of hydrogen-bonding donor(s) located on the alkyl functionality of these lesions. Except for O6-POB-dG, which also induced a low frequency of G → T transversions, all other lesions exclusively stimulated G → A transitions. SOS-induced DNA polymerases played redundant roles in bypassing all the O6-alkyl-dG lesions investigated. DNA polymerase IV (Pol IV) and Pol V, however, were uniquely required for inducing the G → A transition for O6-CM-dG exposure. Together, our study expands our knowledge about the recognition of important NOC-derived O6-alkyl-dG lesions by the E. coli DNA replication machinery.

capable of inducing cancer in ϳ40 different animal species, including higher primates, and are carcinogenic in multiple organs in animals (2). Humans are exposed to NOCs through dietary consumption, tobacco smoking, and endogenous metabolism (3,4). In this vein, two potential types of carcinogenic adducts formed from NOC exposure are of emerging concern, i.e. the pyridyloxobutyl (POB) and pyridylhydroxybutyl (PHB) adducts derived from tobacco-specific N-nitrosamines and carboxymethyl adducts arising from endogenous NOCs (5,6).
Tobacco-specific nitrosamines 4-(methylnitrosamino)-1-(3pyridyl)-1-butanone (NNK) and NЈ-nitrosonornicotine (NNN) are well-documented rodent carcinogens and classified as group I carcinogens by the International Agency for Research on Cancer (7). Metabolic activation of NNK and NNN via cytochrome P450 enzymes can produce reactive intermediates that can react with DNA to yield POB and/or PHB adducts along with other DNA lesions, which may induce mutations during DNA replication (5, 8 -15). Another type of extensively investigated promutagenic lesions, carboxymethyl (CM) adducts, can be induced in DNA through metabolic activation of endogenous NOCs, e.g. N-nitroso bile acid conjugate or N-nitrosoglycine (16 -18). It was suggested that the mutations in the TP53 gene found in gastrointestinal cancer emanate from carboxymethylation rather than methylation of DNA elicited by NOCs (19).
Multiple lines of evidence suggested that the O 6 position of 2Ј-deoxyguanosine is the most facile exocyclic atom to be alkylated on nucleobases (26), and the resulting lesions alter the hydrogen bonding property of the guanine base (6,(27)(28)(29)(30). In this vein, O 6 -methyl-dG (O 6 -Me-dG) and O 6 -POB-dG have been implicated in lung tumorigenesis mediated by NNK (12,(31)(32)(33), and O 6 -CM-dG may contribute to the mutations found in TP53 tumor suppressor gene of human gastrointestinal tumors (19) and is correlated with colorectal cancer (6). Thus, it is important to understand how the O 6 -alkyl-2Ј-deoxyguanosine (O 6 -alkyl-dG) lesions derived from the above-mentioned NOCs affect the efficiency and fidelity of DNA replication.
Several replication studies have been performed to assess the impact of O 6 -POB-dG and O 6 -CM-dG lesions on the flow of genetic information. O 6 -POB-dG was found to be highly mutagenic, and it induces exclusively G 3 A transition in Escherichia coli, whereas it elicits both G 3 A transition and G 3 T transversion in mammalian cells (15,33,34). In addition, multiple translesion synthesis DNA polymerases, particularly Pol , were found to be involved in the efficient bypass of O 6 -POB-dG (15), which is consistent with in vitro results showing that Pol exhibits much higher efficiency in bypassing O 6 -POB-dG than Pol , Pol , and Rev1 (35,36). O 6 -CM-dG is able to form Watson-Crick-type bp with thymidine, thereby inducing G 3 A transition mutation (37). In mammalian cells, Pol and Pol are required for the efficient bypass of this lesion, and only G 3 A transition was observed (38). However, it remains unknown how O 6 -PHB-dG compromises the efficiency and accuracy of DNA replication.

Results
The primary objective of this project was to elucidate how the O 6 -substituted guanine lesions arising from tobacco-derived and endogenous NOCs (Fig. 1) compromise the efficiency and fidelity of DNA replication in E. coli cells. To this end, ODNs harboring a series of O 6 -alkyl-dG lesions at a specific site were synthesized and confirmed by ESI-MS and MS/MS analyses ( Fig. 2 and Figs. S1-S7). A previously described CRAB assay ( Fig. 3 and Fig. S8) was employed to determine the impacts of these lesions on DNA replication in E. coli cells and to determine how replication across these lesions is influenced by SOS-induced DNA polymerases (39,40). After cellular replication, the region of interest in the progeny genomes was amplified by PCR, the ensuing products were digested with two restriction endonucleases, i.e. BbSI and MluCI, and the restriction fragments were analyzed by LC-MS/MS and native PAGE ( Fig. 4 and Figs. S9 -S14).

The impacts of O 6 -POB-dG and O 6 -PHB-dG on the efficiency and fidelity of DNA replication
Our results from PAGE and LC-MS/MS analyses revealed the lack of insertion or deletion mutations ( Fig. 4 and Fig.  S9 -S14). Hence, we calculated the bypass efficiencies of O 6 -POB-dG and O 6 -PHB-dG from the ratio of the 10-mer products emanating from lesion and control genomes, over the 13-mer product, generated from the competitor plasmid ( Fig. 4 and Figs. S11-S14), by taking into account the molar ratios of the lesion and control over the competitor genomes employed at the initial transfection. It turned out that these two lesions were moderate impediments to DNA replication in WT AB1157 cells, with O 6 -PHB-dG exerting a stronger blockage effect (Fig. 5A).
The roles of SOS-induced DNA polymerases in bypassing the O 6 -POB-dG and O 6 -PHB-dG were also investigated by performing the assay in E. coli strains deficient in Pol II, Pol IV, Pol V, or all three in combination. It turned out that a substantial diminution in bypass efficiency was observed only when all three SOS-induced DNA polymerases were depleted, which is consistent with the observations made previously for other O 6 -alkyl-dG lesions (30). No significant differences were observed for the bypass efficiencies of O 6 -POB-dG and O 6 -PHB-dG lesions in AB1157 cells and the isogenic cells with a single depletion of any of the three polymerases (Fig. 5A). This result supports that the three SOS-induced DNA polymerases assume somewhat redundant roles in bypassing O 6 -POB-dG and O 6 -PHB-dG.
The native PAGE analysis also facilitated the quantification of the mutation frequencies of O 6 -POB-dG and O 6 -PHB-dG. In agreement with previous findings (30), the major type of mutation was G 3 A transition for both O 6 -POB-dG and O 6 -PHB-dG in all E. coli strains (Fig. 5B). A low frequency of G 3 T transversion, however, was also detected for  (Fig. 5C), which differs from the previously reported results (34). We also found that the frequencies of the mutations conferred by O 6 -POB-dG and O 6 -PHB-dG were not affected by depletion of any of the three SOS-induced DNA polymerases, individually or all three together (Fig. 5, B and C). This result underscores that the observed mutations can be attributed to the intrinsic hydrogen-bonding properties of the O 6 -subsituted guanine, which preferentially forms bp with thymine.

The impacts of O 6 -CM-dG, O 6 -ACM-dG, and O 6 -HOEt-dG on the efficiency and fidelity of DNA replication
Different from O 6 -POB-dG, O 6 -PHB-dG, and the previously reported O 6 -alkyl-dG lesions (30), O 6 -CM-dG displayed strong blockage to DNA replication machinery in WT AB1157 cells, as manifested by a ϳ10% bypass efficiency (Fig. 6A). Interestingly, this bypass efficiency was augmented by almost 2-fold upon SOS induction.
The unique strong blockage effect of O 6 -CM-dG on cellular DNA replication prompted us to ask whether the negative charge located on the carboxymethyl functionality modulates the bypass efficiency of the lesion. To test this, we synthesized two other O 6 -modified derivatives with the carboxymethyl moiety being replaced with an aminocarbonylmethyl or hydroxyethyl group (Fig. 1), which are neutral under cellular pH. It turned out that O 6 -ACM-dG and O 6 -HOEt-dG also exhibited very low bypass efficiencies (Fig. 6A). Hence, we attribute the low bypass efficiencies of O 6 -CM-dG and the other two derivatives to the terminal hydrogen bond donor (i.e. hydroxyl or amino group) conjugated with the alkyl functionality adducted to the O 6 position of guanine. It is worth noting that, similar to what we observed for O 6 -POB-dG and O 6 -PHB-dG, the bypass efficiencies of these three lesions were not affected by individual depletion of Pol II, Pol IV, or Pol V (Fig. 6A).

Replication studies of NOC-induced O 6 -alkylguanine lesions
The mutation frequencies were also quantified, where only G 3 A transition was observed for all three lesions in all E. coli strains studied (Fig. 6B). In addition, the frequencies of G 3 A transition for O 6 -ACM-dG and O 6 -HOEt-dG were not influenced by depletion of any of the SOS-induced DNA polymerases, alone or all three in combination. However, O 6 -CM-dG distinguishes from other lesions, where Pol V and, to a lesser extent, Pol IV were required for the G 3 A transition mutation. Thus, the negative charge possessed by the carboxymethyl functionality affects the involvement of SOS-induced DNA polymerases in dTMP misincorporation opposite the lesion (Fig. 6B).

Discussion
There is established evidence for the association of tobacco use with lung cancer, and lung cancer is the second most common type of cancer diagnosed in both men and women (41). In addition, red meat-containing diet, especially processed meat, are risk factors for gastrointestinal cancer, which constitutes one of the leading causes of cancer-related deaths worldwide (42,43). Previous studies revealed that the POB, PHB, and CM adducts of DNA derived from the NOCs in tobacco and red meat may be important factors contributing to the development of lung and colorectal cancer.
We investigated the replicative bypass of O 6 -POB-dG, was studied here for the first time. Our results revealed several novel findings. First, we observed that O 6 -POB-dG and O 6 -PHB-dG were slight and moderate impediments to DNA replication, respectively, whereas replication past O 6 -CM-dG was strongly inhibited (Figs. 5A and 6A). Second, we found that the three SOS-induced DNA polymerases, i.e. Pol II, Pol IV, and Pol V, assume redundant roles in bypassing the three O 6alkyl-dG lesions ( Fig. 5A and 4A). Third, our results demonstrated that these lesions induce G 3 A transition exclusively, except that G 3 T transversion was also observed for O 6 -POB-dG (Fig. 5, B and C, and 6B). Fourth, we revealed that the mutation frequencies of O 6 -POB-dG and O 6 -PHB-dG were not affected by depletion of Pol II, Pol IV, Pol V, or all three polymerases; however, Pol V and, to a lesser extent, Pol IV play important roles in eliciting G 3 A transition for O 6 -CM-dG (Fig. 5, B and C, and 6B).
Previously, it was found that O 6 -alkyl-dG lesions, with the alkyl group ranging from methyl to sec-butyl, do not inhibit DNA replication in E. coli cells and the three SOS-induced DNA polymerases play somewhat redundant roles in bypassing these lesions (30). Here, we found that O 6 -POB-dG, O 6 -PHB-dG, and O 6 -CM-dG exhibited slight, moderate, and strong blockage to DNA replication, respectively.

Replication studies of NOC-induced O 6 -alkylguanine lesions
We further explored the mechanism underlying the very low bypass efficiency of O 6 -CM-dG. On the grounds that the carboxylic acid functionality in O 6 -CM-dG would be deproto-nated and carry a negative charge in cellular pH, we initially hypothesized that the negative charge may contribute to the diminished replicative bypass of O 6 -CM-dG. To test that hypothesis, we assessed how two neutral analogs of O 6 -CM-dG, i.e. O 6 -ACM-dG and O 6 -HOEt-dG, impede DNA replication. It turned out that the latter two lesions strongly block DNA replication, with the bypass efficiency of O 6 -ACM-dG being merely ϳ2% (Fig. 6A). These results unveiled that the poor replicative bypass of O 6 -CM-dG is not due to the presence of a negative charge situated on the carboxymethyl group of the lesion. Instead a hydrogen bond-donating hydroxyl or amino group at the terminus of the alkyl chain significantly impedes the replicative bypass (Fig. 6A), which is perhaps due to the disruption of the interaction between the lesions and the active site of polymerases (44,45). Similar mechanism is likely at play for the substantially lower bypass efficiency of O 6 -PHB-dG than O 6 -POB-dG (Fig. 5A). However, future structural studies are required for understanding the detailed molecular mechanism underlying the strong blockage effects of these lesions. In addition, it should be noted that, consistent with the previously reported results for simple O 6 -alkyl-dG lesions (30), genetic ablation of Pol II, Pol IV, or Pol V did not confer any significant difference in replicative bypass of any of the five O 6 -alkyl-dG lesions examined in the present study (Fig. 6A).
It was previously shown that in E. coli cells, simple aliphatic O 6 -alkyl-dG lesions only induce G 3 A transition, which also holds true for the complex O 6 -POB-dG lesion (30,34). In mammalian cells, the primary type of mutation elicited by O 6 -alkyl-dG lesions is the G 3 A transition, although a low frequency of G 3 T transversion was also detected for O 6 -POB-dG (15,34). In line with these previous observations, the five lesions examined in the present study directed primarily G 3 A transition (Figs. 5, B and C, and 6B). However, different from the results obtained from replication experiments employing colony picking and Sanger sequencing analysis (34), a low frequency of G 3 T transversion (ϳ2%) was also observed for O 6 -POB-dG in E. coli cells (Fig. 5C). This result underscores that the modified CRAB assay is more sensitive, thereby allowing for the detection of low frequencies of mutations better than the previously reported method (34).
The mutation frequencies for O 6 -POB-dG, O 6 -PHB-dG, O 6 -ACM-dG, and O 6 -HOEt-dG were not modulated by SOSinduced DNA polymerases, where genetic ablation of Pol II, Pol IV, and Pol V, alone or all three in conjunction, did not confer any appreciable changes in frequencies of G 3 A transition, or, for O 6 -POB-dG, G 3 T transversion mutation (Figs. 5, B and C, and 6B), suggesting that the mutagenic properties of the O 6 -alkyl-dG lesions are independent of the size or shape of the alkyl groups and arise from the intrinsic chemical property of the lesions, i.e. their capabilities in hydrogen bonding and in mispairing with thymidine.
Intriguingly, different from the above four O 6 -alkyl-dG lesions, two SOS-induced DNA polymerases, i.e. Pol IV and Pol V, play important roles in inducing G 3 A transition for O 6 -CM-dG (Fig. 6B), and the frequency of this mutation was increased by 2-fold upon SOS induction, substantiating that the mutagenic properties of O 6 -alkyl-dG lesions can be modulated by the negative charge situated on the O 6 position of guanine.

Replication studies of NOC-induced O 6 -alkylguanine lesions
For all the neutral substituents, from simple methyl to complex POB or PHB, the nucleotide insertion opposite the lesion was not affected by SOS-induced DNA polymerases. The mutagenic nucleotide insertion opposite the negatively charged O 6 -CM-dG is, however, influenced by Pol IV and Pol V (Fig.  6B). It should be noted that, in non-SOS-induced E. coli cells, Pol V could not be detected biochemically, and its expression level would increase by ϳ100-fold upon SOS induction (46 -48). In addition, Pol V plays crucial roles in replicative bypass of multiple lesions, such as oxidized guanine lesions and O 2 -alky-lated thymidine lesions, where it promotes error-prone replication under many circumstances (40,48,49).
It is worth comparing the findings made in E. coli cells with previously reported results in mammalian cells. It was found that O 6 -POB-dG directed primarily the G 3 A transition with a low frequency of G 3 T transversion in HEK293T cells (15), which is in agreement with what we observed in E. coli cells. However, translesion synthesis polymerases, i.e. Pol , Pol , Pol , and Pol , are all required for the efficient bypass of this lesion in human cells (15). Likewise, O 6 -CM-dG confers exclusively

Replication studies of NOC-induced O 6 -alkylguanine lesions
G 3 A transition in HEK293T cells, and it moderately impedes DNA replication (with a 38% bypass efficiency), where Pol and Pol are involved in bypassing this lesion (38). On the other hand, O 6 -CM-dG displays only a ϳ10% bypass efficiency in WT E. coli cells, and the three SOS-induced polymerases play redundant roles in bypassing this lesion.
It is also important to compare the results obtained for the major-groove O 6 -alkyl-dG lesions with previous findings made for minor-groove N 2 -alkyl-dG lesions. In this vein, the minorgroove N 2 -alkyl-dG lesions exert moderate blocking effects on DNA replication (50). For instance, a site-specifically incorporated N 2 -(1-carboxyethyl)-2Ј-deoxyguanosine (N 2 -CE-dG) was not a strong impediment to DNA replication in WT AB1157 cells, with the bypass efficiencies of S-and R-N 2 -CE-dG being ϳ80 and ϳ40%, respectively (50). However, the efficient and accurate replication across N 2 -alkyl-dG lesions in E. coli requires Pol IV, and their accurate bypass in human cells necessitates both Pol (an ortholog of Pol IV) and Pol (50 -54). These differences perhaps can be attributed to the differences in replication machineries of E. coli and human cells.
In summary, our shuttle vector-based study on a group of NOC-derived O 6 -alkyl-dG lesions provided important and novel insights into the impact of carboxymethyl and pyridyloxobutyl/pyridylhydroxybutyl DNA lesions on the efficiency and fidelity of DNA replication in cells. Our results revealed strong miscoding potentials of the O 6 -POB-dG and O 6 -PHB-dG derived from tobacco-specific nitrosamines. In addition, our results revealed the hydrogen-bonding donor and negative charge situated on the O 6 -alkyl group as important determinants of DNA replication in E. coli cells. It will be important to assess how these O 6 -alkyl-dG lesions compromise DNA replication and transcription in mammalian cells in the future.

MS and NMR
ESI-MS and MS/MS experiments were carried out on an LCQ Deca XP ion-trap mass spectrometer (Thermo Fisher Scientific). A mixture of acetonitrile and water (50:50, v/v) was used as the solvent for electrospray. The spray voltage was 3.0 kV, and the temperature of the ion transport tube was maintained at 275°C. High-resolution mass spectra were acquired on an Agilent 6510 Q-TOF LC/MS mass spectrometer (Agilent Technologies, Palo Alto, CA) equipped with an ESI source. 1 H and 31 P NMR spectra were recorded at 400 and 80 MHz on an Avance NEO 400 NMR spectrometer (Bruker, Billerica, MA) at 25°C, respectively.
O 6 -Aminocarbonylmethyl-dG (O 6 -AMC-dG)-containing ODN was deprotected by treating the solid support of O 6 -CM-dG containing ODN with concentrated ammonium hydroxide (57). All other 12-mer lesion-containing ODNs were cleaved from the solid support and deprotected with 0.25 M NaOH at room temperature overnight, and the resulting solution was subsequently neutralized with 0.5 M HCl. The O 6 -POB-dG-containing ODN was synthesized following previously published procedures (58), and the O 6 -PHB-dGcontaining ODN was obtained by incubation of the O 6 -POB-dG-containing ODN with sodium borohydride (5 eq) in H 2 O for 5 min in an ice-water bath.
The solvents were removed under vacuum, and the solid residues were redissolved in water and purified via reversed-phase HPLC, with the use of a Kinetex XB-C18 column (4.6 ϫ 150 mm, 5 m in particle size, and 100 Å in pore size; Phenomenex Inc., Torrance, CA), as described elsewhere (39,40). The HPLC traces for the purification of the 12-mer lesion-carrying ODNs are displayed in Fig. S5, and the mass spectrometric characterizations of the purified lesion-containing ODNs are shown in Fig. 2 and supporting Figs. S6 and S7. The purified 12-mer O 6 -alkyl-dG-containing ODNs were then ligated individually with a 10-mer ODN (5Ј-AGTGGAAGAC-3Ј) with T4 DNA ligase in the presence of a template ODN, and the desired 22-mer lesion-bearing ODNs were purified using denaturing PAGE, as described previously (39,40).

Replication studies of NOC-induced O 6 -alkylguanine lesions
16°C for 8 h. After the ligation, excess scaffolds and the unligated vector were degraded by incubating with T4 DNA polymerase (22.5 U) at 37°C for 4 h. The desired lesion-containing and the lesion-free M13 genomes were purified from the resultant mixture using a Cycle Pure kit (Omega). The constructed lesion-containing or control genomes were normalized against the lesion-free competitor genome following published procedures to determine accurately the relative concentrations of the constructed genomes (59).

Transfection of lesion-containing and competitor M13 genomes into E. coli cells
The control lesion-free M13 genome was mixed with the competitor genome at a molar ratio of 1:1. The O 6 -POB-dGand O 6 -PHB-dG-containing M13 genomes were mixed individually with the competitor genome at a molar ratio of 2:1. The O 6 -CM-dG-, O 6 -ACM-dG-, and O 6 -HOEt-dG-carrying M13 genomes were mixed with the competitor genome at a molar ratio of 5:1. The mixtures were transfected into SOS-induced, electrocompetent WT AB1157 E. coli cells and the isogenic E. coli cells that are deficient in pol II, pol IV, pol V, or all three polymerases (59). The SOS response was induced by irradiating E. coli cells with 254 nm light at a dose of 45 J/m 2 , as previously described (49).
The E. coli cells were subsequently grown in lysogeny broth culture medium at 37°C for 6 h. The phage was recovered from the supernatant by centrifugation at 13,000 r.p.m. for 5 min and further amplified in SCS110 E. coli cells to increase the progeny/lesion-genome ratio. The amplified phage was finally purified using the QIAprep Spin M13 kit (Qiagen) to obtain the ssM13 DNA template for PCR amplification.

Quantification of bypass efficiencies and mutation frequencies
A modified version of the CRAB assay was employed to assess the bypass efficiencies of the O 6 -alkyl-dG lesions in vivo (Fig. 3) (50,59). The region of interest in the purified ssM13 DNA template was amplified by PCR with Phusion high-fidelity DNA polymerase (NEB) using two primers, 5Ј-YCAGCTAT-GACCATGATTCAGTGAGTGGA-3Ј and 5Ј-YTCGGTGC-GGGCCTCTTCGCTATTAC-3Ј (where Y is an amino group). PCR amplification was conducted for 30 cycles, each of which consisted of 10 s at 98°C, 30 s at 65°C, and 15 s at 72°C, with a final extension at 72°C for 5 min.
The PCR products were then subjected to sequential restriction endonuclease digestion and native PAGE analysis for determination of bypass efficiency and mutation frequency ( Fig. 4 and Figs. S11-S14). Briefly, 80 ng of the PCR products were treated with BbsI (10 units) restriction endonuclease and shrimp alkaline phosphatase (SAP, 1 unit) in 10 l of CutSmart buffer (NEB) at 37°C for 30 min. The SAP was deactivated by heating at 80°C for 10 min, and the mixture was incubated with T4 polynucleotide kinase (PNK, 10 units), DTT (5 mM), and [␥-32 P]ATP at 37°C for 30 min to radiolabel the newly released 5Ј-termini of the restriction fragments. After deactivation of T4 PNK via heating at 80°C for 10 min, the resultant mixture was digested with MluCI (10 l) at 37°C for 30 min. Under these conditions, the 10-mer ODNs emanating from initial lesion-situated strand were radiolabeled as d(p*GGCGMGCTAT), where M indicates A, T, G, and C, and p* designates the 5Ј-32 Plabeled phosphate (Fig. 4A).
The digestion was subsequently quenched with 15 l of formamide gel-loading buffer (2ϫ), and the labeled fragments were separated using 30% native PAGE, as previously described ( Fig. 4 and Figs. S11-S14) (39,40). The intensities of the radiolabeled DNA bands were measured using a Typhoon 9410 variable mode imager, and the bypass efficiency was derived using the following equation: bypass efficiency (%) ϭ (lesion signal/ competitor signal)/(control signal/competitor signal) ϫ 100%, where the competitor signal was employed as the internal standard.
The mutation frequencies were determined according to previously published procedures (39,40). As shown in Fig. 4B, the two products with M being an A or G could be well-resolved by native PAGE analysis; hence, native PAGE analysis of the initial lesion-bearing strand was sufficient for quantifying the G 3 A transition frequency. In addition, the G 3 T transversion elicited by O 6 -POB-dG was further confirmed via PAGE analysis of the strand that is complementary to the original lesion-containing strand (Fig. S14).

Identification of mutagenic products by LC-MS/MS
The PCR products were digested with 50 units of BbsI restriction endonuclease and 20 units of shrimp alkaline phosphatase in 250 l of NEB CutSmart buffer at 37°C for 2 h, followed by deactivation of the enzymes through heating at 80°C for 20 min. To the mixture was added 50 units of MluCI, and the solution was incubated at 37°C for 1 h. The resulting solution was extracted once with phenol/chloroform/isoamyl alcohol (25:24:1, v/v). The aqueous layer was subsequently dried in a SpeedVac, desalted with HPLC, and dried again. The resultant solid residues were reconstituted in 20 l of water, and a 10-l aliquot was injected for LC-MS/MS analysis using an Agilent Zorbax SB-C18 column (0.5 ϫ 250 mm, 5 m in particle size). The gradient for LC-MS/MS analysis was 5 min of 5-20% methanol followed by 35 min of 20 -50% methanol in 400 mM 1,1,1,3,3,3-hexafluoro-2-propanol. The temperature for the ion-transport tube was maintained at 300°C. The LTQ linear ion trap mass spectrometer (Thermo Electron, San Jose, CA) was set up for monitoring the fragmentation of the [M-3H] 3Ϫ ions of the 10-mer d(GGCGMGCTAT), where M represents A, T, C, or G. The fragment ions detected in the MS/MS were manually assigned (Figs. S9 and S10).