Translesion Synthesis of 8,5′-Cyclopurine-2′-deoxynucleosides by DNA Polymerases η, ι, and ζ*

Background: The 8,5′-cyclopurine-2′-deoxynucleosides (cPus) are important types of oxidative DNA damage. Results: cPus exhibit both inhibitory and mutagenic effects on replication; polymerases (Pols) η, ι, and ζ are involved in translesion synthesis of these lesions. Conclusion: Pols η, ι, and ζ cooperatively promote translesion synthesis across cPu lesions. Significance: This work revealed the effects of cPus on DNA replication in mammalian cells. Reactive oxygen species can give rise to a battery of DNA damage products including the 8,5′-cyclo-2′-deoxyadenosine (cdA) and 8,5′-cyclo-2′-deoxyguanosine (cdG) tandem lesions. The 8,5′-cyclopurine-2′-deoxynucleosides are quite stable lesions and are valid and reliable markers of oxidative DNA damage. However, it remains unclear how these lesions compromise DNA replication in mammalian cells. Previous in vitro biochemical assays have suggested a role for human polymerase (Pol) η in the insertion step of translesion synthesis (TLS) across the (5′S) diastereomers of cdA and cdG. Using in vitro steady-state kinetic assay, herein we showed that human Pol ι and a two-subunit yeast Pol ζ complex (REV3/REV7) could function efficiently in the insertion and extension steps, respectively, of TLS across S-cdA and S-cdG; human Pol κ and Pol η could also extend past these lesions, albeit much less efficiently. Results from a quantitative TLS assay showed that, in human cells, S-cdA and S-cdG inhibited strongly DNA replication and induced substantial frequencies of mutations at the lesion sites. Additionally, Pol η, Pol ι, and Pol ζ, but not Pol κ, had important roles in promoting replication through S-cdA and S-cdG in human cells. Based on these results, we propose a model for TLS across S-cdA and S-cdG in human cells, where Pol η and/or Pol ι carries out nucleotide insertion opposite the lesion, whereas Pol ζ executes the extension step.

Living cells are constantly challenged by endogenous and exogenous genotoxins (1). The resultant DNA damage can lead to mutations and altered gene function, cell senescence, or apoptosis (2). To maintain normal cellular function, multiple DNA damage surveillance and repair systems have evolved (3). Among them, translesion synthesis (TLS) 3 is a damage toler-ance pathway by which mammalian cells overcome replication blockages imposed by various types of DNA damage (4,5). The most abundant class of TLS polymerases belong to the Y-family, including DNA polymerase (Pol ), Pol , Pol , and REV1 (5). In this regard, Pol , Pol , and Pol usually play major roles in the insertion step of TLS with occasional involvement in the subsequent extension step (6). In addition, REV1 is widely thought to serve as a scaffold protein to facilitate the recruitment of other TLS polymerases, including Pol , Pol , Pol , and, a B-family polymerase, Pol (7)(8)(9)(10). Recently, it was suggested that yeast (and presumably mammalian) Pol comprised four subunits, including REV3, REV7, Pol31, and Pol32 (11,12).
Reactive oxygen species, formed as byproducts of normal aerobic metabolism as well as from exposure to ionizing radiation, can induce a battery of DNA damage products including the 5ЈR and 5ЈS diastereomers of 8,5Ј-cyclo-2Ј-deoxyadenosine (cdA) and 8,5Ј-cyclo-2Ј-deoxyguanosine (cdG) (Fig. 1) (13,14). 8,5Ј-Cyclopurine-2Ј-deoxynucleosides (cPus) are stable lesions and are found to be reliable markers for oxidative DNA damage (13,15,16). These cPu lesions have been readily detected in vitro and in vivo under various conditions (13,(17)(18)(19). Moreover, it was recently shown that cPu lesions can accumulate in mammalian tissues in an age-dependent and tissue-specific manner, which may accelerate the natural processes of aging and contribute to the development of cancer, neurodegeneration, and other human diseases (13,16,18,20,21).
Unlike most other reactive oxygen species-induced DNA lesions that are removed by base excision repair, cPu lesions are unique due to the presence of a C8-C5Ј bond between the purine base and 2-deoxyribose of the same nucleoside (Fig. 1). This additional covalent bond induces helical distortion to DNA and substantially stabilizes the glycosidic bond against acid-induced hydrolysis, which may prevent initiation of base excision repair by a DNA glycosylase and render the cPu lesions attractive substrates for nucleotide excision repair (13,21). Indeed, multiple lines of evidence supports that the cPu lesions are repaired by nucleotide excision repair, but not by base excision repair or direct enzymatic reversal (22)(23)(24)(25)(26).
Unrepaired cPu lesions may lead to detrimental biological consequences. In this regard, S-cdA and S-cdG located in the template DNA inhibit strongly DNA transcription and induce the generation of mutant transcripts in mammalian cells (22,23,26). In addition, a single cPu lesion may alter gene expression by preventing the binding of transcription factors and regulatory proteins to their recognition sequences (27,28). S-cdA and S-cdG can also act as strong inhibitors to DNA replication and induce nucleobase substitutions at the lesion site in Escherichia coli cells (25,29). However, it is still unclear how these cPu lesions compromise DNA replication in mammalian cells. In addition, although previous biochemical studies have suggested a role for Pol in the TLS across the cPu lesions in vitro (30,31), it remains unknown whether Pol is involved in the replicative bypass of cPu lesions in mammalian cells, and much less is known about whether other TLS polymerases, including Pol , Pol , and Pol , participate in the TLS across cPu lesions in vitro and in mammalian cells.
To address these questions, herein we performed in vitro steady-state kinetic assays to understand the roles of Pol , Pol , and Pol in the replicative bypass of cPu lesions. We also developed a strand-specific PCR-based competitive replication and adduct bypass (SSPCR-CRAB) assay and quantitatively examined how S-cdA and S-cdG lesions compromise the efficiency and accuracy of DNA replication in human cells. Moreover, we investigated the relative roles of TLS polymerases, including Pol , Pol , Pol , and Pol , in the bypass of S-cdA and S-cdG using human cells that are completely or partially deficient in one or more of these TLS Pols.

EXPERIMENTAL PROCEDURES
Materials and Cell Culture Conditions-Unmodified oligodeoxyribonucleotides (ODNs), shrimp alkaline phosphatase, chemicals, and [␥-32 P]ATP were purchased from Integrated DNA Technologies, U. S. Biochemical Corp., Sigma, and PerkinElmer Life Sciences, respectively. The 12-mer lesionbearing ODNs 5Ј-ATGGCGXGCTAT-3Ј ("X" represents S-cdA or S-cdG) were previously synthesized following published procedures (29,32). Human DNA Pol , Pol , and a two-subunit Saccharomyces cerevisiae DNA Pol complex (Rev3/Rev7) were purchased from Enzymax, and human DNA Pol was provided by Professor Roger Woodgate. All other enzymes unless otherwise specified were purchased from New England BioLabs. The 293T human embryonic kidney epithelial cells were purchased from ATCC. The SV40-transformed Pol -deficient XP30RO fibroblasts and the corrected cells (XP30RO ϩ Pol ) were provided by Professor James E. Cleaver (33,34). Cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Invitrogen), 100 units/ml of penicillin, and 100 g/ml of streptomycin (ATCC), and incubated at 37°C in 5% CO 2 atmosphere.
Steady-state Kinetic Measurements-Steady-state kinetic assays were carried out following previously published procedures (35). The primer-template complex consisted of a 20-mer S-cdA-or S-cdG-bearing template and a 5Ј-32 P-labeled 13-mer, 14-mer (with the correct nucleosides being placed opposite the lesions), or 15-mer (carrying the correct nucleosides opposite the lesions and their adjacent 5Ј nucleoside, Figs. 2 and 3) primer (0.05 M).
Template Preparation for in Vivo Replication Studies-We first constructed a parent vector for S-cdA by modifying the sequence of the original pTGFP-Hha10 plasmid, which contains an SV40 origin and is able to replicate in the SV40-transformed mammalian cells (36,37). To this end, a 50-mer ODN with the sequence 5Ј-AATTCGCAGCGAGTCATCCATGG-CGAGGTATTGTGGAGTCGATGCATCCG-3Ј was annealed with its complementary strand and ligated to an NheI-EcoRI restriction fragment from the pTGFP-Hha10 plasmid. We next constructed the S-cdA-bearing double-stranded shuttle vector by using a previously described method (Fig. 4A) (36,37). Briefly, we nicked the parent vector for S-cdA with Nt.BstNBI to produce a gapped vector by removing a 25-mer singlestranded ODN, followed by filling the gap with a 12-mer S-cdAbearing ODN (5Ј-ATGGCGXGCTAT-3Ј, X ϭ S-cdA) and a 13-mer unmodified ODN (5Ј-TGTGGAGTCGATG-3Ј). The ligation mixture was incubated with ethidium bromide, and the resulting supercoiled lesion-bearing plasmid was purified using agarose gel electrophoresis. Using the same method, we prepared the lesion-free control and competitor vectors for S-cdA, where the 12-mer S-cdA-containing ODN was replaced with a 12-mer ODN (5Ј-ATGGCGAGCTAT-3Ј) and a 15-mer ODN (5Ј-ATGGCGAGCAGCTAT-3Ј), respectively. The S-cdG-bearing vector as well as its corresponding undamaged control and competitor vectors were constructed in a similar fashion.
In Vivo Transfection and Plasmid Isolation-The lesionbearing or the corresponding non-lesion control plasmids were premixed with the competitor genome for in vivo transfection, with the molar ratios of competitor vector to control and lesion-bearing genome being 1:1 and 1:19, respectively. The 293T human embryonic kidney epithelial cells, XP30RO, and XP30RO ϩ Pol cells (1 ϫ 10 5 ) were seeded in 24-well plates and cultured overnight, after which they were transfected with 300 ng of mixed genome by using Lipofectamine 2000 (Invitrogen) following the manufacturer's instructions. The cells were harvested 24 h after transfection, and the progenies of the plasmid were isolated using the Qiagen Spin Kit (Qiagen), with minor changes (38). The residual unreplicated plasmid was further removed by DpnI digestion, followed by digesting the resulting linear DNA with exonuclease III as described elsewhere (39 -41).
Highly efficient depletion of specific TLS DNA polymerases was achieved using siRNAs that were previously validated (42,43). All siRNAs were purchased from Dharmacon: POLK
Real-time Quantitative RT-PCR-Total RNA was extracted from the cells 48 h after transfection with siRNA using the Total RNA Kit I (Omega), and subjected to DNase I treatment with the DNA-free kit (Ambion) to eliminate the DNA contamination. cDNA was generated by using Moloney murine leukemia virus reverse transcriptase (Promega) and an oligo(dT) 16 primer. Real-time quantitative RT-PCR for evaluating the siRNA knockdown efficiency was performed by using the iQ SYBR Green Supermix kit (Bio-Rad) and gene-specific primers for POLK, POLI, POLH, REV3L, or the control gene GAPDH as described elsewhere (22), and the primers for real-time RT-PCR shown in supplemental Table S1.
Western Blot Analysis-Western analysis was performed with a total of 40 g of whole cell lysate. Antibodies that specifically recognize human Pol , Pol , Pol , or REV3L were purchased from Santa Cruz Biotechnology and used at a 1:10,000 dilution. Human ␤-actin antibody (Abcam) was used at a 1:5,000 dilution. Horseradish peroxidase-conjugated secondary goat anti-mouse antibody (Santa Cruz Biotechnology), donkey anti-goat antibody (Santa Cruz Biotechnology), and goat anti-rabbit antibody (Abcam) were used at a 1:10,000 dilution.
PCR, PAGE, and LC-MS/MS Analyses-The progeny genomes arising from in vivo replication were amplified in a PCR (GoTaq green master mix, Promega) containing a pair of primers whose products cover the initial lesion site and span 8 DpnI recognition sites. The primers were 5Ј-GCTAGCGGATGCATCGAC-TCCACAACAG-3Ј and 5Ј-GGCTCCCTTTAGGGTTCCGA-TTTAGTG-3Ј, and the PCR amplification started at 95°C for 2 min; then, 35 cycles at 95°C for 30 s, 65°C for 30 s, and 72°C for 1.5 min, and a final 5-min extension at 72°C. The PCR products were purified using a QIAquick PCR Purification Kit (Qiagen) and stored at Ϫ20°C until use.
For PAGE analysis, a portion of the PCR fragments was treated with 5 units of NcoI and 1 unit of shrimp alkaline phosphatase at 37°C in 10 l of New England Biolab buffer 3 for 1 h, followed by heating at 80°C for 20 min to deactivate the shrimp alkaline phosphatase. The above mixture was then treated in 15 l of New England Biolabs buffer 3 with 5 mM DTT, ATP (50 pmol of cold, premixed with 1.66 pmol of [␥-32 P]ATP), and 5 units of T4 polynucleotide kinase. The reaction was continued at 37°C for 1 h, followed by heating at 65°C for 20 min to deactivate the T4 polynucleotide kinase. To the reaction mixture 5 units of SfaNI was subsequently added, and the solution was incubated at 37°C for 1 h, followed by quenching with 15 l of formamide gel loading buffer. The mixture was loaded onto a 30% polyacrylamide gel (acrylamide:bis-acrylamide ϭ 19:1) and products were quantified by phosphorimager analysis. Similar to a previously described method for assessing the impact of DNA lesions on transcription (22,44), we determined the effects of DNA lesions on replication efficiency and fidelity by the "relative bypass efficiency" (RBE) and "relative mutation frequency" (RMF) values, respectively. Briefly, the RMF value was determined from the relative amounts of different products arising from replication of the lesion-containing genome. The RBE value was calculated using the following formula, %RBE ϭ (lesion signal/competitor signal)/(non-lesion control signal/ competitor signal) (45,46). LC-MS/MS was used to further identify the replication products arising from S-cdA-or S-cdG-bearing substrates similar to those described elsewhere (22,29,37,44). Briefly, PCR products were treated with 50 units of NcoI and 20 units of shrimp alkaline phosphatase in 250 l of New England Biolabs buffer 3 at 37°C for 4 h, followed by heating at 80°C for 20 min. To the resulting solution was added 50 units of SfaNI, and the reaction mixture was incubated at 37°C for 4 h followed by extraction with phenol/chloroform/isoamyl alcohol (25:24:1, v/v). The aqueous portion was dried with Speed-Vac, desalted with HPLC, and dissolved in 12 l of water. The ODN mixture was subjected to LC-MS/MS analysis, as described elsewhere (18,25,31,38). An LTQ linear ion trap mass spectrometer (Thermo Electron) was set up for monitoring fragmentation of the [M-3H] 3Ϫ ions of the 13-mer ODNs (d(CATGGCGMGCTGT), where "M" designates A, T, C, or G).

Roles of TLS Pols in the Bypass of S-cdA and S-cdG in Vitro-
We first performed steady-state kinetic assays to determine the ability of human Pol and Pol , as well as yeast Pol to insert nucleotides opposite the S-cdA and S-cdG, and their respective undamaged substrates ( Fig. 2A). In this regard, due to the lack of availability of recombinant mammalian Pol (6, 47), herein we used a two-subunit S. cerevisiae Pol complex (REV3/REV7) instead. In this regard, a heterodimeric complex, including the REV3 and REV7 subunits, was previously identified as the minimal assembly required for catalytic activity of Pol in vitro (4,11). Our results showed that, like Pol (31), Pol was highly error-free and efficient in inserting nucleotides opposite the two cPu lesions; relative to the corresponding insertion for the unmodified substrates, incorporation of the correct nucleotides opposite S-cdA and S-cdG occurred at frequencies of ϳ59 and ϳ61%, respectively (Figs. 2, B and C, and 3, and Table 1). On the other hand, Pol and Pol (REV3/REV7) incorporated nucleotides opposite the cPu lesions with extremely low efficiency (Fig. 2, B and C, and supplemental Tables S2 and S3).
We next assessed the accuracy and efficiency of nucleotide incorporation opposite the adjacent 5Ј nucleoside of the cPu lesions, where we employed a 14-mer primer with the correct dNTP being placed opposite the cPu lesions and their respective controls (supplemental Fig. S1). It turned out that the efficiency for nucleotide incorporation opposite the 5Ј neighboring nucleoside (at ϩ1 position) was very poor with Pol or Pol (Fig. 2, B and C, and supplemental Tables S4 and S5). However, Pol (REV3/REV7) was modestly efficient, albeit error-prone, in extending past the S-cdA:dT and S-cdG:dC pairs (Fig. 2, B and C, and supplemental Tables S6). In this regard, when the correct S-cdA:dT pair was placed at the primer-template junction, yeast Pol (REV3/REV7) preferentially incorporates the correct dCMP opposite dG at the ϩ1 position, with misincor-   Table S6). Similarly, Pol (REV3/REV7) was highly error-prone in extending past the S-cdG:dC pair, with misincorporation of dAMP, dTMP, and dGMP opposite the ϩ1 nucleoside (dG) occurring at frequencies of 10, 21, and 46%, respectively (supplemental Table S6). We further obtained the steady-state kinetic parameters for yeast Pol (REV3/REV7)-, as well as human Poland Pol -mediated nucleotide insertion at the ϩ2 position with respect to the two cPu lesions. The results showed that yeast Pol (REV3/REV7) extended past the S-cdA-and S-cdG-bearing substrates with high accuracy and efficiencies; relative to the corresponding insertion for the unmodified substrates, insertion of the correct nucleotides opposite the ϩ2 base in the S-cdA-and S-cdG-bearing templates occurred at frequencies of ϳ86 and ϳ68%, respectively (Fig. 2, B and C, and supplemental Table S7). In addition, Pol -mediated extension past the lesion-containing substrates was moderately efficient, as insertion of the correct dGMP occurred at efficiencies of ϳ40% relative to the corresponding extension past the dA-and dG-containing substrates (Fig. 2, B and C, and supplemental Fig. S2 and Table S8). On the other hand, human Pol was also able to extend past S-cdA and S-cdG when the correct nucleosides are placed opposite the lesions and their adjacent 5Ј nucleosides, albeit much less efficiently (by ϳ6 -10-fold) than the corresponding extension for the unmodified substrates (supplemental Table S8).
Effects of S-cdA and S-cdG on Replication in Human Cells-To investigate how S-cdA and S-cdG compromise the efficiency and fidelity of DNA replication in human cells, we developed a new SSPCR-CRAB assay. This method evolved from the principle of the traditional competitive replication and adduct bypass assay, which has been widely used for assessing the cytotoxic and mutagenic properties of DNA lesions placed in a single-stranded M13 genome in E. coli cells (45,46).
We constructed double-stranded shuttle vectors containing a single site-specific lesion (S-cdA or S-cdG), as well as the corresponding non-lesion control vectors housing an unmodified nucleotide (A or G) at the lesion site. Given that the lesionbearing strand may be replicated at a reduced efficiency from its opposing undamaged strand in mammalian cells (48,49), we employed a similar strategy as described previously (37,48,49) and incorporated a C/C mismatch two nucleotides away from the lesion site to distinguish the replication products formed from the two strands (Fig. 4A). The lesion-bearing or undamaged control vectors were mixed individually with an undamaged competitor vector at given molar ratios and co-transfected into human cells (Fig. 4B). In this regard, the competitor vector contains three more nucleotides than the control vector situated between the two restriction sites used for SSPCR-CRAB assay. The progenies of the plasmids were isolated from human cells 24 h after transfection, and residual unreplicated plasmids were removed by a combined treatment with DpnI and exonuclease III as described previously (39 -41). The progeny genomes were subsequently amplified using a pair of PCR primers spanning the lesion site. Notably, one of the primers (P1) carries a G as the terminal 3Ј-nucleotide corresponding to the C/C mismatch site (Fig. 4B), which is used to amplify selectively the progeny genomes arising from the replication of the bottom, but not the top strand of the plasmids under appropriate conditions (50). In addition, P1 was designed deliberately to carry a C/A mismatch three bases from its 3Ј-end (Fig. 4B) to increase the specificity of strand-specific PCR, as described previously (50). The resulting PCR products were digested with appropriate restriction enzymes, i.e. NcoI and SfaNI, and subjected to LC-MS/MS and PAGE analyses.
We first asked whether Pol is involved in TLS across cPu lesions in human cells. To this end, we performed the SSPCR-CRAB assay using the Pol -deficient XP30RO fibroblasts and the corrected cells (XP30RO ϩ Pol ) as hosts for in vivo replication. PAGE analysis of restriction fragments of PCR products showed that the RBE values for S-cdA and S-cdG in Pol -deficient XP30RO cells were ϳ3 and ϳ2%, respectively, and the Only the construction of S-cdAbearing vector is shown. SV40 ORI and X indicate SV40 replication origin and S-cdA, respectively. The C/C mismatch site is underlined. The cleavage sites of Nt.BstNBI, NcoI, and SfaNI are designated with arrows. P1 represents one of the primers for PCR, i.e. d(GCTAGCGGATGCATCGACTCCACAACAG). P1 contains a G as the terminal 3Ј-nucleotide corresponding to the C/C mismatch site of lesion-bearing genome, and it also contains a C/A mismatch three bases away from its 3Ј-end for improving the specificity of PCR. M represents the nucleobase formed at the lesion site after replication, and N designates the paired nucleobase of M in the complementary strand.
RBE values for the two lesions were substantially higher in Pol -rescued human cells at 24 h after transfection (Fig. 5, A and  B). PAGE analysis also showed that S-cdA was mutagenic during replication, with A 3 T mutation occurring at frequencies of ϳ5 and ϳ9% in Pol -deficient and Pol -complemented cells, respectively (Fig. 5, A and C). In addition, replicative bypass of S-cdG in Pol -deficient and Pol -rescued cells can give rise to G 3 A transition at frequencies of ϳ3 and ϳ11%, respectively, and G 3 T transversion at frequencies of ϳ27 and ϳ32%, respectively (Fig. 5, A and C). We also determined the identities of these mutated products by LC-MS/MS analysis (supplemental Fig. S3).
We further confirmed the involvement of Pol in TLS across S-cdA and S-cdG by using 293T cells in which the expression of Pol was knocked down by siRNA. Consistent with a previous report (43), our real-time PCR and Western blot results showed that siRNA knockdown was highly efficient for the POLH gene (Fig. 6, A and B). As expected, the RBE values for S-cdA and S-cdG were markedly lower in Pol knockdown cells than in control siRNA-treated cells (Fig.  7, A and C, and supplemental Fig. S4, A and B). In addition, we found that knocking down the expression of Pol caused a considerable decrease in the A 3 T mutation induced by S-cdA in 293T cells (Fig. 7B and supplemental Fig. S4A). On the other hand, partial depletion of Pol with siRNA did not confer a significant reduction in G 3 A or G 3 T mutations induced by S-cdG in human cells (Fig. 7, D and E, and supplemental Fig.  S4B).
We next used siRNAs to inhibit the expression of other TLS polymerases, including Pol , Pol , and REV3L (Fig. 6, A and B), and determined their possible roles in TLS across cPu lesions in human cells. Relative to 293T cells treated with control siRNA, the RBE values for S-cdA and S-cdG were considerably lower in the cells treated with Pol or REV3L siRNAs (Fig. 7, A and C, and supplemental Fig. S4, A and B). On the other hand, we did not find any change in RMF values for S-cdA or S-cdG caused by siRNA knockdown of Pol or REV3L, except that knockdown of Pol led to a decrease in the G 3 A mutation for S-cdG (Fig. 7, B, D, and E, and supplemental Fig. S4, A and B). In addition, the siRNA depletion of Pol had no effect on the RBE or RMF values for S-cdA and S-cdG in human cells (Fig. 7, A-E,  and supplemental Fig. S4, A and B).
The above results suggested that Pol , Pol , and Pol had important roles in promoting replication through cPu lesions in   OCTOBER 4, 2013 • VOLUME 288 • NUMBER 40 human cells. To further ascertain whether these three polymerases function in the same or different pathways for replicative bypass of cPu lesions, we assessed the effects of simultaneous knockdowns of two of the three TLS Pols (Fig. 6, A and B). Our results showed that knockdown of both Pol and REV3L had an effect similar to that of Pol alone (Fig. 7, A-E, and supplemental Fig. S4, C and D). Similarly, we found that the dual knockdown of Pol and Pol , or Pol and REV3L, did not cause further change in TLS frequencies relative to that observed upon the knockdown of one of these polymerases alone (Fig. 7, A-E, and supplemental Fig. S4, C and D). Therefore, Pol , Pol , and Pol may cooperate with each other to carry out TLS across S-cdA and S-cdG in human cells.

DISCUSSION
It has been previously shown that S-cdA is a strong blockage to mammalian DNA polymerase ␦ and T7 DNA polymerase in vitro (23), and the bypass efficiencies of S-cdA and S-cdG during replication are very low (less than ϳ5%) in wild-type E. coli cells (25,29). In agreement with these findings, our results demonstrated that both S-cdA and S-cdG constituted strong blockage to DNA replication machinery in human cells. We also found that replicative bypass of S-cdA and S-cdG generated mutations in human cells. Similar to the observation in E. coli cells (25,29), S-cdA was weakly mutagenic and induced A 3 T transversion at a frequency of ϳ10% in 293T cells. On the other hand, the major type of mutation induced by S-cdG is the G 3 A transition occurring at a frequency of ϳ20%, although a low frequency (less than ϳ5%) of G 3 T transversion was also observed in E. coli cells (25,29). Compared with E. coli cells, replicative bypass of S-cdG was more mutagenic in human 293T cells (ϳ50%), where we observed a higher frequency of G 3 T than the G 3 A mutation. The strong blocking and mutagenic effects of the cPu lesions on DNA replication in human cells, together with the abundant presence of these lesions in mammalian tissues (13,16,18,20,21), suggest that these lesions may constitute significant endogenous DNA lesions that may play important roles in the development of human diseases.
We further demonstrated a role for Pol in promoting replication through the cPu lesions in vivo, by using human cells that are completely or partially deficient in Pol . Consistent with our results, depletion of its ortholog, i.e. Pol V, also caused decreased bypass efficiencies of S-cdA and S-cdG in E. coli cells (25,29). In addition, depletion of Pol in human cells conferred a considerable reduction in mutation frequency of S-cdA, suggesting that Pol may be involved in error-prone replicative bypass of this lesion in mammalian system. This result is somewhat surprising, as our previous biochemical studies have sug- gested a role for Pol in the error-free nucleotide insertion opposite cPu lesions (31). The difference in the fidelity of lesion bypass by Pol between in vitro and in vivo studies may be attributed to two factors. First, it has been proposed that the TLS machinery requires the functional interaction between TLS Pols and other accessory proteins, which may modulate the fidelity and/or efficiency of TLS Pols in lesion bypass in cells (6,43,51,52). Second, the in vitro steady-state kinetic experiments were conducted in the presence of one nucleotide at a time, whereas the replication in cells occurs in the mutual presence of all four canonical nucleotides. Thus, the absence or presence of competition in nucleotide incorporation under the two replication conditions may also result in different efficiencies and fidelities of TLS (53).
We also demonstrated, for the first time, that Pol and Pol have important roles in translesion synthesis of cPu lesions. Our biochemical results demonstrated that human Pol and yeast Pol (REV3/REV7) could function efficiently in the insertion and extension steps, respectively, of TLS across S-cdA and S-cdG in vitro. Along this line, siRNA knockdown of Pol and REV3L led to a substantial decrease in RBE values for S-cdA and S-cdG in human cells. Like in the case of Pol , although in vitro biochemical results showed that Pol predominantly incorporated the correct nucleotides opposite the cPu lesions, our in vivo replication studies showed that, siRNA knockdown of Pol could cause a reduction in S-cdG-induced G 3 A mutation. This result suggested that Pol may play a role in errorprone insertion opposite this lesion during replication in human cells.
Moreover, we found that the simultaneous depletion of any two of Pol , Pol , and Pol had no additive effects on the replicative bypass of cPu lesions, indicating that these three polymerases might function in the same or partly overlapping genetic pathways for bypassing cPu lesions in human cells. On the other hand, although human Pol exhibited its ability to extend past these cPu lesions in vitro, knocking down the expression of Pol had no effect on TLS across S-cdA and S-cdG in human cells. This is in keeping with previous data indicating that Pol IV, an ortholog of Pol , may not be able to bypass these cPu lesions in E. coli cells (25,29).
It should be noted that we observed lower frequencies of the G 3 A mutation induced by S-cdG in Pol -deficient human fibroblast cells than the isogenic cells rescued with wild-type human Pol ; however, siRNA knockdown of Pol had no significant effect on the mutation frequencies of S-cdG in 293T cells. This result may reflect a role for the residual Pol in TLS across S-cdG in cells. In this vein, it was suggested that residual TLS Pols might be sufficient for keeping the miscoding properties of some DNA lesions unchanged, even though sometimes lesion bypass efficiencies could be reduced by the siRNA knockdown of some specific TLS Pols (37,42,43).
Based on our findings, we proposed a model for the TLS across cPu lesions in human cells. In this model, a replicative polymerase stalled by cPu lesions on the template DNA is displaced by Pol and/or Pol , which inserts a nucleotide opposite the lesion. A second polymerase switch can then be made, where another TLS polymerase (i.e. Pol ) accurately extends the primer terminus past the damage site. In keeping with our study, two-polymerase mechanisms, in which insertion by one (or more) specific TLS Pol(s) is followed by extension with another polymerase, has been suggested to play important roles in the replicative bypass of several other DNA lesions in mammalian cells. These included thymidine glycol, cis,syn-cyclobutane pyrimidine dimer, and pyrimidine(6 -4)pyrimidone photoproducts in mammalian cells (6,42,43,51,52). It would be interesting to unveil how different inserter and extender Pols as proposed here and in other studies (6,42,43,51,52) coordinate their actions during TLS across DNA damage in human cells.