Translesion synthesis DNA polymerases promote error-free replication through the minor-groove DNA adduct 3-deaza-3-methyladenine

N3-Methyladenine (3-MeA) is formed in DNA by reaction with S-adenosylmethionine, the reactive methyl donor, and by reaction with alkylating agents. 3-MeA protrudes into the DNA minor groove and strongly blocks synthesis by replicative DNA polymerases (Pols). However, the mechanisms for replicating through this lesion in human cells remain unidentified. Here we analyzed the roles of translesion synthesis (TLS) Pols in the replication of 3-MeA-damaged DNA in human cells. Because 3-MeA has a short half-life in vitro, we used the stable 3-deaza analog, 3-deaza-3-methyladenine (3-dMeA), which blocks the DNA minor groove similarly to 3-MeA. We found that replication through the 3-dMeA adduct is mediated via three different pathways, dependent upon Polι/Polκ, Polθ, and Polζ. As inferred from biochemical studies, in the Polι/Polκ pathway, Polι inserts a nucleotide (nt) opposite 3-dMeA and Polκ extends synthesis from the inserted nt. In the Polθ pathway, Polθ carries out both the insertion and extension steps of TLS opposite 3-dMeA, and in the Polζ pathway, Polζ extends synthesis following nt insertion by an as yet unidentified Pol. Steady-state kinetic analyses indicated that Polι and Polθ insert the correct nt T opposite 3-dMeA with a much reduced catalytic efficiency and that both Pols exhibit a high propensity for inserting a wrong nt opposite this adduct. However, despite their low fidelity of synthesis opposite 3-dMeA, TLS opposite this lesion replicates DNA in a highly error-free manner in human cells. We discuss the implications of these observations for TLS mechanisms in human cells.

N3-Methyladenine (3-MeA) 2 is formed in DNA by reaction with alkylating agents and by reaction with S-adenosylmethionine, the reactive methyl donor in most cellular reactions (1,2). It has been estimated that the action of S-adenosylmethionine on DNA generates ϳ600 3-MeA adducts in a mammalian cell per day and the half-life of 3-MeA is between 4 and 24 h (3).
3-MeA is highly cytotoxic because it blocks DNA synthesis by the replicative DNA polymerases (4). Structural studies with the catalytic subunit of DNA polymerase (Pol) ␦, the major eukaryotic replicative Pol, have shown that a number of residues in its active site interact with bases in the DNA minor groove, such that a mispaired base can be sensed up to 4 bases away from the primer terminus (5). Thus, an infringement of the minor groove at the N3 position of a purine would be highly blocking to replication by Pol␦, necessitating the requirement of translesion synthesis (TLS) DNA Pols for inserting a nucleotide (nt) opposite 3-MeA and then extending synthesis from the inserted nt for a distance of at least 4 -5 nts before Pol␦ could take over.
Among the human TLS Pols, Pols , , , and Rev1, Pol is most suited to incorporate a nt opposite lesions that protrude into the minor groove. The active site of Pol is much narrower than that of other TLS Pols; consequently, the purine template A or G is pushed into the syn conformation by the incoming nt, and two hydrogen bonds are formed between the Hoogsteen edge of the purine template and the Watson-Crick edge of the incoming nt (6 -8). Because the adoption of a syn conformation by 3-MeA would move this minor groove lesion into the more spacious major groove, Pol could incorporate the correct nt T opposite 3-MeA by Hoogsteen base pairing. Structural studies with Pol have indicated that its active site can accommodate lesions that protrude into the minor groove at the templateprimer junction, whereby it could proficiently extend synthesis from the 3-MeA⅐T base pair at the template-primer junction (9). Thus, proficient and error-free replication through the 3-MeA lesion could be performed by the sequential action of Pols and .
To analyze the genetic control of TLS through 3-MeA in human cells, we used the stable 3-deaza analog of this adduct, 3-dMeA. We show that TLS through this adduct occurs via three different pathways. As expected from structural studies, we provide evidence that Pols and function together in mediating TLS through this lesion. In addition, we find that Pol and Pol are also required and they mediate replication through this lesion via two independent pathways. Even though Pols and incorporate nts opposite 3-dMeA with a low fidelity, replication through the lesion occurs in an error-free manner in human cells. We discuss the implications of these and other observations for the role of TLS in promoting proficient and predominantly error-free replication through DNA lesions in human cells.

Genetic control of replication through the 3-dMeA lesion in human cells
3-dMeA was incorporated in the lacZ target sequence in the leading strand (Fig. 1A) and the schematic of siRNA depletions and the various steps for the analyses of TLS frequency and for mutation analyses is shown in Fig. 1B. In the plasmid system, TLS through the lesion generates Kan ϩ blue colonies and the frequency of Kan ϩ blue colonies among the total Kan ϩ colonies gives a very reliable and highly repeatable estimate of TLS frequency.
To identify the TLS pols required for replicating through the 3-dMeA lesion, we examined the effects of siRNA depletion of various TLS Pols, including Pols , , , and . For all the TLS Pols analyzed, we ascertained that siRNA treatment led to a highly efficient depletion of the intended protein similar to that shown previously (10,11). The TLS data we obtained from independent TLS assays in human cells are highly reproducible as evidenced from the high repeatability of data for TLS opposite UV-induced lesions, cis-syn TT dimer, and the (6 -4)-TT photoproduct reported in different studies (10,12,13).
TLS opposite 3-dMeA in normal human fibroblasts treated with control (NC) siRNA occurs with a frequency of ϳ46% (Table 1), and the TLS frequency remains the same in Poldepleted cells indicating that Pol is not required. By contrast, TLS frequency is reduced to ϳ27% in Polor Pol-depleted cells. To determine whether Pols and function together or independently, we examined the effects of their simultaneous depletion on TLS frequency. Our observation that their simultaneous depletion causes no further reduction in TLS frequency than that conferred by their individual depletion indicates that they function together in mediating replication through the 3-dMeA lesion. Depletion of Pol or the Rev3 catalytic or Rev7 accessory subunit of Pol also led to a reduction in TLS frequency to ϳ25%. To determine whether Pol and Pol function together in one TLS pathway or whether they constitute independent pathways, we exam-

Genetic control of replication through dMeA
ined the effects of their co-depletion on TLS frequency. Our results that TLS frequency is reduced to ϳ13% upon the simultaneous depletion of Pols and indicate that these Pols act independently.
Our genetic results suggested that replication through 3-dMeA is mediated via three independent pathways in which Pols and function together in one pathway and Pol and Pol function in two additional pathways. To verify this inference, we determined whether simultaneous depletion of Pol or Pol together with Pol or with Pol leads to a further reduction in TLS frequency. In accord with this inference, co-depletion of Pol or Pol with Pol confers a reduction in TLS frequency to ϳ14%, and a similar reduction in TLS frequency occurs upon the simultaneous depletion of Pol or Pol with Pol (Table 1). Thus, TLS opposite 3-dMeA occurs via three independent Pol/, Pol, and Pol pathways (Fig. 2), and each of these pathways contributes about equally to the replication of 3-deMeAdamaged DNA.

Non-catalytic role of Rev1 in the Pol/Pol pathway
Rev1 specifically incorporates a C opposite template G (14). In the Rev1 active site, G does not form a base pair with the incoming C; instead, template G is pushed out into a solventfilled cavity and a conserved Arg residue in Rev1 then pairs with the incoming C (15,16). Thus, Rev1 DNA polymerase activity is specifically adapted to insert a C opposite N 2 -dG DNA adducts and would play no role in TLS opposite 3-dMeA. However, in mammalian cells, Rev1 also plays an indispensable role as a scaffolding component of Y-family Pols , , and (10). Hence, even though the Rev1 catalytic activity would play no role in TLS opposite 3-dMeA, its scaffolding role would be required for TLS mediated by the Pol/Pol pathway but not for TLS mediated by the Pol or Pol pathways. To confirm this, we examined the effects of co-depletion of Rev1 with Pol, Pol, Pol, or Pol. As shown in Table 2, Rev1 depletion reduces the TLS frequency to ϳ23%, and this frequency remained the same upon co-depletion of Rev1 with Pol or Pol. By contrast, simultaneous depletion of Rev1 with Pol or Rev3 reduced TLS frequency to ϳ12%. The epistasis of Rev1 with Pol or Pol and the reduction in TLS frequency upon co-depletion of Rev1 with Pol or Pol is consistent with the requirement of Rev1 for Pol/Pol-dependent TLS but not for TLS mediated by Pol or Pol.

Error-free replication through 3-dMeA in human cells
Sequence analyses of TLS products from cells treated with control siRNA revealed no mutational events, and we also found no evidence of mutational events among the TLS products obtained from Pol-, Rev1-, Pol-, or Pol-depleted cells (Table 3). Thus, TLS opposite 3-dMeA occurs in a highly errorfree manner in human cells.

Catalytic efficiency and fidelity of TLS Pols for DNA synthesis opposite 3-dMeA
Pol primarily inserts a T opposite 3-dMeA, and in the presence of 4 dNTPs, it inserts a T opposite the lesion but does not extend synthesis from the inserted nt (Fig. 3). Using steadystate kinetic analyses, we determined the catalytic efficiency and fidelity of Pol for nt insertion opposite undamaged A versus 3-dMeA. As shown in Table 4, Pol inserts a T nt opposite an undamaged A with over 1000-fold higher catalytic efficiency than an incorrect nt. However, compared with the incorporation of a T opposite undamaged A, T is incorporated opposite 3-dMeA with an ϳ8-fold lower efficiency. Moreover, compared with the efficiency for T incorporation opposite 3-dMeA, the efficiency for incorporating the incorrect nt A or G opposite 3-dMeA is reduced by only ϳ25-30-fold, and the efficiency for incorporating a C is reduced by ϳ90-fold. Thus, the catalytic efficiency of Pol for incorporating the correct nt opposite 3-dMeA is reduced and the efficiency for incorporating an incorrect nt is enhanced. As determined by steady-state kinetic analyses, Pol extends synthesis from the 3-dMeA:T base pair by incorporating dATP opposite the next 5Ј template base with

Genetic control of replication through dMeA
an ϳ10-fold reduced efficiency compared with that for the undamaged A:T base pair (Table 5). Pol primarily incorporates a T opposite 3-dMeA; however, it incorporates an A also, albeit not as well, and in the presence of 4 dNTPs, Pol replicates DNA through the 3-dMeA lesion (Fig. 4). Steady-state kinetic analyses indicate that opposite undamaged A, Pol incorporates a T with an over 10 3 higher catalytic efficiency than any of the incorrect nts; however, compared with that opposite undamaged A, the catalytic efficiency of Pol for T incorporation opposite 3-dMeA is reduced almost 100-fold (Table 6). Moreover, compared with the efficiency for T incorporation opposite 3-dMeA, Pol incorporates an A opposite this lesion with only an ϳ10-fold reduction in efficiency, a G with an ϳ30-fold reduction in efficiency, and a C with an ϳ120-fold reduction in efficiency (Table 6). Thus, the catalytic efficiency and fidelity of Pol for incorporating the correct nt opposite 3-dMeA is greatly reduced. The efficiency of Pol for incorporating dATP opposite the next 5Ј T template base is reduced ϳ20-fold compared with the efficiency for extending synthesis from the undamaged A:T base pair (Table 7).
Pol is a proficient extender of synthesis from nts incorporated opposite DNA lesions by another TLS Pol (17)(18)(19)(20)(21). We find that the proficiency of Pol for extension of synthesis from the 3-dMeA:T base pair is reduced ϳ30-fold compared with that for extending the undamaged A:T base pair (Table 8). Because Pol exhibits poor proficiency for inserting nts opposite DNA lesions, we presume that another as yet unidentified TLS Pol inserts nts opposite 3-dMeA from which Pol then extends synthesis.

A major role of TLS in promoting replication through the 3-dMeA lesion
Our results indicating that TLS in normal human fibroblasts accounts for almost 50% lesion bypass strongly suggest that TLS provides a primary means for replicating through this DNA lesion. Because the 3-dMeA lesion could have been removed from the plasmid by the 3-alkyladenine DNA glycosylase (22), or by nucleotide excision repair, and because removal of the lesion by these repair processes will generate Kan ϩ white colonies, the proportion of Kan ϩ blue colonies, indicative of TLS, will be correspondingly reduced. Thus, the TLS frequency in excision repair proficient cells underestimates the frequency with which TLS actually occurs. The high frequency of TLS in repair proficient cells supports the premise that TLS provides the primary mechanism for replicating through the 3-dMeA lesion and that alternative lesion bypass mechanisms such as filling-in of the gap opposite the lesion site by template switching play a much less significant and subsidiary role. TLS also plays a prominent role in promoting replication through other DNA lesions such as UV induced cis-syn TT dimer (12) and (6 -4)-TT photoproduct (13), thymine glycol (11, 23) generated by the reaction of oxygen-free radicals with thymine; and N1-methyladenine (24) generated by reaction with naturally occurring methyl halides and by reaction with environmental methylating agents.

Action mechanisms of TLS Pols in DNA synthesis opposite 3-dMeA
The ability of Pol to push the purine template into a syn conformation and to form a Hoogsteen base pair with the incoming pyrimidine nt provides a mechanism by which Pol could insert a T opposite 3-dMeA (7,8), and the ability of Pol to accommodate the 3-dMeA:T base pair at the primer terminus (9) would allow it to extend synthesis from this base pair. In the absence of structural information, it is unclear whether Pol pushes the 3-dMeA lesion into a syn conformation for forming a Hoogsteen base pair with the incoming dTTP and then it extends synthesis from the 3-dMeA:T Hoogsteen base pair; or whether the minor groove disruption by 3-dMeA presents no steric hindrance to the Pol active site; consequently, it forms a 3-dMeA:T Watson-Crick (W-C) base pair and then extends synthesis from the W-C base pair. For the Pol-dependent pathway, we expect Pol to extend synthesis from the T nt inserted opposite 3-dMeA by a DNA polymerase whose identity remains to be determined. However, the possibility that human Pol can insert a nt opposite 3-dMeA as well as extend synthesis from the inserted nt cannot be entirely excluded at this point.

Highly error-free replication through the 3-dMeA lesion
Sequence analysis of TLS products has indicated that all three pathways dependent upon Pol/Pol, Pol, and Pol, respectively, conduct TLS opposite 3-dMeA in an error-free manner. In striking contrast to the lack of mutational TLS products in human cells, in vitro biochemical studies have indicated that TLS Pols synthesize DNA opposite 3-dMeA with a  Genetic control of replication through dMeA low fidelity. Thus, whereas Pol incorporates a T opposite undamaged A with a 1000-fold or even higher catalytic efficiency than an incorrect nt, it incorporates a T opposite 3-dMeA with a reduced efficiency than opposite undamaged A and the catalytic efficiency for inserting the incorrect nt opposite 3-dMeA is enhanced. For Pol, the proficiency for incorporating a T opposite 3-dMeA is greatly reduced compared with that opposite undamaged A, and the catalytic efficiency of Pol for inserting an incorrect nt opposite 3-dMeA is increased. The striking discrepancy between the low fidelity of DNA synthesis by Pol or Pol opposite 3-dMeA observed in in vitro biochemical studies versus the high fidelity of TLS opposite this lesion observed in human cells strongly suggests that the fidelity of TLS Pols is tightly regulated during replication through this DNA lesion in human cells. We have previously noted that TLS opposite DNA lesions such as cis-syn TT dimer (12), (6 -4)-TT photoproduct (13), thymine glycol (11,23), or an N1-methyladenine (24) also occurs in a much more errorfree manner in human cells than biochemical studies would have predicted. This accumulating evidence reinforces the notion that the catalytic efficiency and fidelity of TLS Pols opposite DNA lesions are regulated such that replication through DNA lesions occurs in a predominantly error-free manner in human cells.

Construction of plasmid vectors containing 3-dMeA
Since the half-life of 3-MeA in vitro has been established to be between 12 and 24 h (3), that precludes the construction of oligos containing this DNA adduct. To circumvent this problem, a stable 3-deaza analog of the nucleoside 3Ј-methyl-2Јdeoxyadenosine was incorporated into oligos as 3-dMeA. 3-dMeA projects into the DNA minor groove and blocks synthesis by replicative Pols. Hence, for all the genetic and biochemical studies, we used 3-dMeA lesion-containing DNAs.
3-Deaza-3-methyldeoxyadenosine oligonucleotides were synthesized on a Model 8909 Expedite DNA synthesizer using standard DNA synthesis chemistry. The 3-deaza-3-methyl-deoxyadenosine was incorporated using an offline coupling mode for incorporation of the 3-deaza-3-Me-dA-cyanoethyl phosphoramidite (purchased from Berry and Associates, Inc., Dexter, MI 48130). The oligos were deprotected using standard concentrated ammonia deprotection, and purified and analyzed by reverse phase HPLC on a Beckman System Gold

Steady-state kinetic analyses of nucleotide incorporation opposite undamaged A or 3-dMeA by human Pol
Pol (0.02-0.2 nM) was incubated with primer:template DNA substrate (10 nM) and increasing concentrations of dNTPs for 10 min at 37°C. The nucleotide incorporation rate was plotted against dNTP concentration and the data were fit to the Michaelis-Menten equation. Apparent K m and k cat values were obtained from the fit and used to calculate the efficiency of deoxynucleotide incorporation (k cat /K m ).  Genetic control of replication through dMeA HPLC. Oligonucleotides were additionally analyzed and confirmed by MALDI-MS on a Bruker Autoflex MALDI mass spectrophotometer. The in-frame target sequence of the laczЈ gene containing 3-dMeA is shown in Fig. 1A. Because the laczЈ sequence in the 3-dMeA-containing DNA strand is in-frame, it encodes functional ␤-galactosidase (␤-gal); the opposite DNA strand harbors an SpeI restriction site containing a ϩ1 frameshift, which makes it non-functional for ␤-gal. The 3-dMeA-containing strand carries the kanamycin gene (Kan ϩ ), whereas the other DNA strand has the kan Ϫ gene (Fig. 1A). The detailed methods for the construction of lesion-containing SV40-based duplex plasmids have been published previously (12,13).

Assays for translesion synthesis and mutation analyses of TLS products in human cells
The schematic of TLS assay is shown in Fig. 1B and the detailed methods for TLS assays have been published previously (12,13). Briefly, human fibroblast GM637 cells were transfected with the particular siRNA and after 48 h of incubation, the target vector DNA and siRNA (second transfection) were co-transfected. After 30 h of incubation, plasmid DNA was transfected into Escherichia coli XL1-Blue super competent cells (Stratagene) and cells plated on LB/kan plates containing isopropyl 1-thio-␤-D-galactopyranoside (GenDEPOT) and 100 g/ml of X-gal (GenDEPOT). TLS frequency is determined from the number of blue colonies among total colonies growing on LB/Kan plates and mutation frequencies and mutational changes were analyzed by DNA sequencing.

DNA polymerase assays
DNA substrates consisted of a radiolabeled oligonucleotide primer annealed to a 75-nt oligonucleotide DNA template by heating a mixture of primer/template at a 1:1.5 molar ratio to 95°C and allowing it to cool to room temperature for several hours. The template 75-mer oligonucleotide contained the sequence 5Ј-AGC AAG TCA CCA ATG TCT AAG AGT TCG TAT AAT GCC TAC ACT GGA GTA CCG GAG CAT CGT CGT GAC TGG GAA AAC-3Ј and it contained an undamaged A or 3-dMeA at the underlined position. For examining the incorporation of dATP, dTTP, dCTP, or dGTP nucleotides individually, or of all 4 dNTPs, a 44-mer primer 5Ј-GTT TTC CCA GTC ACG ACG ATG CTC CGG TAC TCC AGT GTA GGC AT-3Ј was annealed to the above 75-mer template.
The standard DNA polymerase reaction (5 l) contained 25 mM Tris⅐HCl (pH 7.5), 5 mM MgCl 2, 1 mM dithiothreitol, 100 g/ml of BSA, 10% glycerol, 10 nM DNA substrate, and 0.2 nM Pol or 1 nM Pol. For nucleotide incorporation assays, 50 or 25 M dATP, dTTP, dCTP, or dGTP (Roche Biochemicals) were used for Pol and Pol, respectively, and for examining synthe-   Genetic control of replication through dMeA sis through the 3-dMeA lesion, 50 or 25 M each of all 4 dNTPs were used for Pol and Pol, respectively. Reactions were carried out for 10 min at 37°C. Reaction products were resolved on a 12% polyacrylamide gel containing 8 M urea and analyzed by a PhosphorImager.

Steady-state kinetic analysis
Steady-state kinetic analyses for deoxynucleotide incorporation were performed as described (25). Gel band intensities of the substrate and products of the deoxynucleotide incorporation reactions were quantified using a PhosphorImager and ImageQuant software (Molecular Dynamics). The observed rate of deoxynucleotide incorporation, v obs , was determined by dividing the amount of product formed by the reaction time and protein concentration. The v obs was graphed as a function of the deoxynucleotide concentration, and the data were fit to the Michaelis-Menten equation describing a hyperbola: v obs ϭ (k cat [E] ϫ [dNTP])/(K m ϩ [dNTP]). From the best fit curve, the apparent K m and k cat steady-state kinetics parameter were obtained for the incorporation of dATP, dGTP, dCTP, and dTTP and the efficiencies of the nucleotide incorporation (k cat / K m ) were determined.
Author contributions-J. H. Y. performed and analyzed the experiments on the genetic control of TLS and mutagenicity, J. R. C. performed the biochemical experiments and analyzed the data, J. P. contributed to the genetic experiments, S. P. and L. P. designed and coordinated the study and wrote the paper. All authors reviewed the results and approved the final version of the manuscript.