Promutagenic bypass of 7,8-dihydro-8-oxoadenine by translesion synthesis DNA polymerase Dpo4.

Reactive oxygen species induced by ionizing radiation and metabolic pathways generate 7,8-dihydro-8-oxoguanine (oxoG) and 7,8-dihydro-8-oxoadenine (oxoA) as two major forms of oxidative damage. The mutagenicity of oxoG, which promotes G to T transversions, is attributed to the lesion's conformational flexibility that enables Hoogsteen base pairing with dATP in the confines of DNA polymerases. The mutagenesis mechanism of oxoA, which preferentially causes A to C transversions, remains poorly characterized. While structures for oxoA bypass by human DNA polymerases are available, that of prokaryotic DNA polymerases have not been reported. Herein, we report kinetic and structural characterizations of Sulfolobus solfataricus Dpo4 incorporating a nucleotide opposite oxoA. Our kinetic studies show oxoA at the templating position reduces the replication fidelity by ~560-fold. The catalytic efficiency of the oxoA:dGTP insertion is ~300-fold greater than that of the dA:dGTP insertion, highlighting the promutagenic nature of oxoA. The relative efficiency of the oxoA:dGTP misincorporation is ~5-fold greater than that of the oxoG:dATP misincorporation, suggesting the mutagenicity of oxoA is comparable to that of oxoG. In the Dpo4 replicating base pair site, oxoA in the anti-conformation forms a Watson-Crick base pair with an incoming dTTP, while oxoA in the syn-conformation assumes Hoogsteen base pairing with an incoming dGTP, displaying the dual coding potential of the lesion. Within the Dpo4 active site, the oxoA:dGTP base pair adopts a Watson-Crick-like geometry, indicating Dpo4 influences the oxoA:dGTP base pair conformation. Overall, the results reported here provide insights into the miscoding properties of the major oxidative adenine lesion during translesion synthesis.

The base pair geometry of syn-oxoG:anti-dA is essentially identical to that of the correct base pair; this facilitates the formation of the promutagenic oxoG:A base pair in the active site of DNA polymerases ( Figure 1B) [6][7][8], thus promoting G to T transversions [4]. OxoA, the cellular levels of which are up to one half of those of oxoG [9,10], is a major oxidative DNA modification found in γ-irradiated DNA and the genomes of human cancerous tissues [9,11,12]. Unlike oxoG, which favors a Watson-Crick-like base pair with dA, oxoA engages in a wobble base pair with dG ( Figure 1C), where N7-H and O8 of syn-oxoA are hydrogen bonded to O6 and N1-H of anti-dG, respectively [13]. While oxoA is significantly less mutagenic than oxoG in E. coli cells [14], the same lesion induces a significant number of A to C mutations in mammalian cells [3,15,16], suggesting differential mutagenicity of oxoA in prokaryotic and mammalian cells. OxoA opposite thymine is cleaved by E. coli mismatch-specific uracil DNA glycosylase (MUG) and human thymine DNA glycosylase (TDG) [3,[17][18][19].
The oxoA lesion bypass by high fidelity DNA polymerases in prokaryotes (e.g., Taq DNA polymerase and Escherichia coli DNA polymerase I) takes place in an error-free manner [20,21]. In contrast, mammalian DNA polymerases α and β frequently misincorporate dGTP and dCTP opposite the lesion in vitro [15]. The presence of oxoA at the templating position decreases the replication fidelity of human Polη by ~100-fold [22,23], highlighting the promutagenic properties of oxoA. Structural studies show that the minor groove contact by Gln38 of Polη stabilizes the base pairing between syn-oxoA and anti-dGTP, thereby promoting dGTP insertion opposite oxoA. Currently, the oxoA lesion bypass by prokaryotic Y-family DNA polymerases has not been reported and is the focus of the present study.
Here, we report steady-state kinetic studies of Dpo4 incorporating dTTP and dGTP opposite the major oxidative adenine lesion oxoA. We also present two crystal structures of ternary complexes of Dpo4 in complex with incoming nonhydrolyzable dTTP and dGTP analogs opposite templating oxoA. These structures highlight the conformational flexibility and miscoding properties of oxoA in the catalytic site of Dpo4. These results provide new insights into the dual coding potential of the major oxidative adenine lesion and its mechanisms of mutagenicity.

Protein expression, crystallization and structure determination.
Dpo4 was overexpressed and purified from E. coli using protocols described previously [28].
Ternary Dpo4 complexes with a templating oxoA and an incoming nonhydrolyzable dNTP analog were prepared and crystallized using minor modifications of published crystallization conditions [30]. To obtain ternary Dpo4-DNA complexes, Dpo4 was first incubated with a recessed DNA duplex containing a 18-mer template (5′-TCAT[oxoA]GAATCCTTCCCCC-3′) and a complementary 13-mer primer (5′-GGGGGAAGGATTC-3′), producing a Dpo4-DNA binary complex. Subsequently, a 10-fold molar excess of nonhydrolyzable dGMPNPP or dTMPNPP (Jena Bioscience) was added to the binary complex. The resulting ternary Dpo4-DNA complexes were crystalized in a buffer solution containing 50 mM Tris pH 7.5, 16-25% PEG3350, and 100 mM magnesium acetate. The ternary complex crystals were cryoprotected in the mother liquor supplemented with 15% ethylene glycol and were flash-frozen in liquid nitrogen. Diffraction data were collected at 100 K at the beamline 19-ID at the Advanced Photon Source and at the beamline BL 5.0.1. at the Lawrence Berkeley National Laboratory. All diffraction data were processed using HKL2000. Structures were solved by molecular replacement with a ternary Dpo4 complex structure (PDB ID: 1JX4) as a search model using Molrep [35]. The molecular model was built using COOT [36], and was refined using PHENIX [37]. MolProbity was used to make Ramachandran plots [38]. All the crystallographic figures were generated using Chimera [39].

Steady-state kinetics of single nucleotide incorporation opposite templating oxoA by Dpo4.
Steady-state kinetic parameters for Dpo4-catalyzed nucleotide insertion opposite templating oxoA were measured as similarly described previously [23,40]. Briefly, 5´-FAM-labeled primer (5´-FAM/GGGGGAAGG ATTC-3´) and oxoA-modified template (5´-TCAT(oxoA)GAATCCTTCCCCC-3´) were annealed in a hybridization buffer (10 mM Tris-HCl pH 7.5, 1 mM EDTA) by heating at 90°C for 5 min and slowly cooling to room temperature. Enzyme activities were determined using the reaction mixture containing 40 mM Tris-HCl pH 8.0, 60 mM KCl, 10 mM dithiothreitol, 250 μg/ml bovine serum albumin, 2.5% glycerol, 5 mM MgCl 2 , 100 nM annealed DNA, and varying concentration of an individual dNTP. To prevent the end-product inhibition and substrate depletion, the enzyme concentrations and reaction-time intervals were adjusted for each experiment (less than 20% insertion product formed). Nucleotide insertion reactions were initiated by adding the enzyme and stopped by adding a gel-loading buffer (95% formamide with 20 mM EDTA, 45 mM Tris-borate, 0.1% bromophenol blue, 0.1% xylene cyanol). The quenched reaction mixtures were separated on 20% denaturing polyacrylamide gels. The gels were analyzed using ImageQuant (GE Healthcare) to quantify the nucleotide insertion products. The k cat and K m were determined by fitting reaction rate over dNTP concentrations to the Michaelis-Menten equation. Each experiment was repeated three times to determine the average and the standard deviation of the kinetic results. The efficiency of nucleotide insertion was calculated as k cat /K m . The relative efficiency of dGTP misincorporation opposite oxoA was determined as f = (k cat /K m ) [dGTP:oxoA] /(k cat /K m ) [dTTP:oxoA] .

OxoA forms a Watson-Crick base pair with an incoming dTTP* in the active site of Dpo4.
Our steady-state kinetic studies show Dpo4 efficiently incorporates dTTP opposite a templating oxoA. To gain structural insights into how this occurs, we solved a ternary complex structure of Dpo4 incorporating a nonhydrolyzable dTMPNPP (dTTP* hereafter) opposite a templating oxoA in the presence Mg 2+ ions. The nonhydrolyzable nucleotide dTTP* was used because it is isosteric to dTTP, and its coordination to the active-site metal ions is virtually indistinguishable from that of dTTP. The nonhydrolyzable dNTP analogs have been frequently used in the determination of ternary structures of various DNA polymerases [41,42]. The Dpo4-oxoA:dTTP* ternary complex was crystallized in the P2 1 space group with cell dimensions of a = 52.84 Å, b = 97.46 Å, c = 101.24 Å, α = 90.00°, β = 90.05°, and γ = 90.00°. The Dpo4-oxoA:dTTP* ternary structure was refined to a resolution of 2.84 Å with R work = 23.0% and R free = 28.0% ( Table 2).
The Dpo4-oxoA:dTTP* ternary complex structure provides the structural basis for the correct insertion past oxoA by the enzyme (Figure 3). The ternary structure displays the conserved secondary structures and the four characteristic domains (thumb, palm, finger, and little finger) of Y-family DNA polymerases ( Figure 3A). The oxoA:dTTP* base pair is well accommodated within the enzyme's catalytic site ( Figure 3B). The strong electron density around anti-oxoA and the incoming dTTP* indicates oxoA:dTTP* is ordered in the replicating base pair site of Dpo4. The guanidinium moieties of Arg331 and Arg332 engage in hydrogen bond interactions with the 5'-phosphate oxygens and the 8-oxo group of oxoA, respectively, thereby stabilizing oxoA at the templating position ( Figure 3C). On the other hand, the side chains of Tyr48 and Arg51 are hydrogen bonded to the phosphate oxygens of the incoming dTTP*. The templating oxoA adopts an anti-conformation and forms a Watson-Crick base pair with dTTP* at inter-base hydrogen bonding distances of 2.7 Å and 3.0 Å ( Figure 3D). The base pair geometry of oxoA:dTTP* is essentially identical to that of the canonical base pair. Specifically, oxoA:dTTP* base pair displays the λ angles of 49° (oxoA) and 59° (dTTP*) and a C1´-C1´ distance of 10.5 Å ( Figure 3D). The similar oxoA:dTTP* base pair conformations have been observed in published polβ and polη structures [22,23]. The primer terminus 3´-OH is coordinated to the A-site magnesium ion and is about 4.4 Å away from the P α of the incoming dTTP* ( Figure 3E), poised for in-line nucleophilic attack on the P α of the incoming nucleotide. The A-and B-site magnesium ions are coordinated with ligands including Asp7, Phe8, Asp105, Glu106, phosphate oxygens, and primer terminus 3´-OH ( Figure 3E). In addition, dTTP* stacks with the primer terminus dT ( Figure 3E). Overall, the Dpo4-oxoA:dTTP* ternary complex structure containing two metal ions and exhibiting Watson-Crick base pairing is consistent with the efficient dTTP incorporation opposite oxoA by the enzyme.
Comparison of published Dpo4 ternary structures with our oxoA:dTTP* structures reveals the impact of the templating oxoA on the DNA conformation ( Figure 4). While the nascent base pairs of the Dpo4-oxoA:dTTP* and published Dpo4-dG:dCTP structures adopt a canonical Watson-Crick conformation, DNA conformations at the N+1 and N+2 positions of these structures significantly differ ( Figure 4A and 4B). Unlike the Dpo4-oxoA:dTTP* structure, in the Dpo4-dG:dCTP structure, Arg332 does not form a hydrogen bond with the 5 -phosphate oxygen of the templating base. Similar conformations for Arg331 and Arg332 are observed in the published Dpo4-oxoG:dCTP* structure [43], where Arg331 and Arg332 are hydrogen bonded to the 5´-phosphate and the 8-oxo moiety of oxoG, respectively. These results indicate Arg332 may play a role in accommodating 8-oxopurine at the templating position [43]. Despite the similar conformations for Arg331 and Arg332 in the oxoA:dTTP* and oxoG:dCTP* structures, their DNA conformations at the N+1 and N+2 positions show a large deviation ( Figure 4C and 4D).

OxoA forms a Watson-Crick-like base pair with dGTP* in the active site of Dpo4.
Our steady-state kinetic analysis shows substituting dA for oxoA at the templating position greatly (~560-fold) promotes mutagenic replication (Table 1). To gain structural insights into the promutagenic nature of oxoA in the active site of Dpo4, we determined the ternary structure of Dpo4 in complex with the templating oxoA and incoming dGTP*. The ternary complex of Dpo4-oxoA:dGTP* was crystallized in P2 1 space group with cell dimensions of a = 52.98 Å, b = 97.76 Å, c = 101.82 Å, α = 90.00° , β = 89.99, and γ = 90.00°. The Dpo4-oxoA:dGTP* structure was refined to a resolution of 3.08 Å with R work = 25.6% and R free = 30.3% ( Table 2).
The Dpo4-oxoA:dGTP* ternary complex structure reveals the promutagenic properties of oxoA in the Dpo4 active site ( Figure 5). The overall structure of the oxoA:dGTP* structure is very similar to that of the oxoA:dTTP* structure (RMSD = 0.29 Å). In the replicating base pair site of Dpo4, oxoA is in a syn-conformation and the incoming dGTP* is in an anticonformation ( Figure 5B). Both A-and B-site metal ions are present in the catalytic site. Arg332 no longer engages in hydrogen bond interactions with oxoA, while Arg331 remains hydrogen bonded to the 5´-phosphate oxygen of the templating oxoA ( Figure 5C). Unlike the Dpo4-oxoA:dTTP* structure, the primer terminus G:C base pair in the oxoA:dGTP* structure does not form Watson-Crick hydrogen bonds despite its correct base pairing, suggesting oxoA:dGTP affects the conformation of the primer terminus base pair. Interestingly, oxoA and dGTP* form three Hoogsteen hydrogen bonds with a Watson-Cricklike geometry ( Figure 5D). The inter-base hydrogen bonding distances for oxoA:dGTP* are 3.0 Å, 2.9 Å, and 3.1 Å. The λ angles for oxoA:dGTP* are 53° (oxoA) and 54° (dGTP*). The C1´ (oxoA)-C1´ (dGTP*) distance is 10.4 Å. Therefore, the base pair geometry for oxoA:dGTP* is virtually identical to that observed in correct base pairs.
The coordination of the A-site magnesium ion is significantly different from those observed in Dpo4 structures catalyzing the correct insertion ( Figure 5E). Specifically, the primer terminus 3´-OH is not coordinated to the A-site Mg 2+ ion and is about 5.4 Å away from the P α of the incoming dGTP*. In addition, the primer terminus 3´-OH points away from the P α and thus is not optimally positioned for in-line nucleophilic attack on the P α . The lack of coordination of the primer terminus 3´-OH appears to be caused by a relatively large propeller-twist angle (20°) for oxoA:dGTP*. The presence of two purine rings (syn-oxoA and dGTP*) at the N position could influence the conformation of the primer terminus base pair, possibly inducing an unfavorable alignment of 3´-OH for nucleotidyl transfer reaction. Our kinetic studies show that the catalytic efficiency of the oxoA:dGTP insertion is ~320fold greater than that of the dA:dGTP insertion, while it is ~14-fold less than that of the oxoA:dTTP insertion (Table 2). Overall, the Dpo4-oxoA:dGTP* ternary complex structure with a Watson-Crick-like oxoA:dGTP* base pair conformation but non-optimal orientation of the 3´-OH primer terminus is consistent with our kinetic data.
The comparison of the Dpo4-oxoA:dGTP* and Dpo4-oxoA:dTTP* structures provide insights into the difference in the catalytic efficiency of the oxoA:dGTP and oxoA:dTTP insertions. In Dpo4 catalytic site, dTTP insertion opposite oxoA is 14-fold more efficient than dGTP insertion opposite oxoA. The superposition of the oxoA:dGTP* and oxoA:dTTP* structures reveals large conformational changes of DNA at the N+1 and N+2 positions ( Figure 6A). In particular, the templating dT at the N+1 position stacks with anti-oxoA in the oxoA:dTTP* structure, while it is flipped out of the helix in the oxoA:dGTP* structure. The lack of the stacking interaction between the templating oxoA and dT (N+1) in during the oxoA:dGTP insertion could result in an increased K m . The geometry of the oxoA:dTTP* and oxoA:dGTP* base pairs in the Dpo4 active site are much alike ( Figure  6B), indicating both base pairs are well tolerated in the nascent base pair site of Dpo4. OxoA:dTTP with Watson-Crick geometry can occur through their major tautomers. On the contrary, oxoA:dGTP with Watson-Crick-like geometry requires either of the nucleobases to undergo tautomerization in the replicating base pair site. Such requirement would make the oxoA:dGTP insertion less favorable than the oxoA:dTTP. One notable conformational difference is found in Arg332 of Dpo4 ( Figure 6C), which is positioned near the template DNA. In the oxoA:dTTP* structure, Arg332 is hydrogen bonded to the 5´-phophate oxygen and O8 of the templating oxoA at the N position. In the presence of the incoming dGTP*, Arg332 moves toward the templating base at the N-1 position and thus no longer engages in hydrogen bonding interactions with the oxoA nucleotide. Overall, the comparison of the Dpo4-oxoA:dTTP* and Dpo4-oxoA:dGTP* structures highlights that both base pairs are readily accommodated in the Dpo4 active site without triggering large-scale conformational changes of protein.

DISCUSSION
While the base pairing properties of oxoA:dTTP* in the catalytic sites of DNA polymerases are essentially the same ( Figure 7A and 7B), those of oxoA:dGTP* vary significantly among DNA polymerases. In the absence of protein contact, oxoA:dG forms two Hoogsteen hydrogen bonds with a wobble geometry. In the confines of the X-family Polβ, the oxoA:dGTP* forms three Hoogsteen hydrogen bonds with a Watson-Crick-like geometry. The formation of Watson-Crick-like oxoA:dGTP* base pair in Polβ active site may be induced by the minor groove contact by Asn279, which is hydrogen bonded to the N3 of dGTP. The minor groove interaction would lower the pK a of oxoA and dGTP, thereby increasing the population of the rare enol tautomer of oxoA or dGTP. The involvement of the rare enol tautomer of oxoA or dGTP will allow to form Watson-Crick-like oxoA:dGTP conformation. Abrogation of the minor groove interaction via an Asn279Ala mutation results in the adoption of syn-oxoA:syn-dGTP base pair conformation. In the active site of human Y-family Polη, oxoA:dGTP* takes on a wobble geometry ( Figure 7C), which is greatly stabilized by Gln38-mediated minor groove contacts to both oxoA and dGTP. The conformational differences of oxoA:dGTP in the active sites of Polβ and Polη suggest that the minor groove interaction is not the sole determining factor for Watson-Crick-like oxoA:dGTP conformation. In the catalytic site of prokaryotic Y-family DNA polymerase Dpo4, oxoA:dGTP does not engage in minor groove interaction with protein, yet it takes on a Watson-Crick-like conformation, indicating that the microenvironment of Dpo4 promotes the formation of Watson-Crick-like oxoA:dGTP conformation. The active site of Dpo4, which is more spacious than that of Polβ and Polη, might be more suitable for accepting Watson-Crick-like oxoA:dGTP conformation in the absence of the minor groove contacts to the nascent base pair.
The bypass of the major oxidative adenine lesion by Dpo4 and Polη is more error prone than that of the major oxidative guanine lesion. Dpo4 incorporates dCTP (correct nucleotide) opposite oxoG 70-fold more efficiently than dGTP [44]. On the other hand, the same enzyme inserts dTTP (correct nucleotide) opposite oxoA only 14-fold more efficiently than dGTP, indicating that oxoA is more promutagenic than oxoG. The catalytic efficiency of the correct insertion opposite oxoA is comparable to that of the correct insertion opposite oxoG [44]. The catalytic efficiency for the incorrect oxoA:dGTP insertion, however, is ~4-fold greater than that of the incorrect oxoG:dATP insertion. The lower catalytic efficiency for the oxoG:dATP insertion is mainly caused by a higher K m . In addition to Dpo4, human Polη replicates past oxoA ~2-fold less accurately than oxoG, highlighting the promutagenic nature of oxoA.
The structural comparison of the Dpo4-oxoA:dGTP and Polη-oxoA:dGTP complexes provide insights into the differences in the oxoA bypass fidelity of these polymerases. The replication fidelity of oxoA bypass by Solfolobus solfataricus Dpo4 is ~7-fold greater than that by human Polη [23]. This difference in fidelity is caused by the decrease in the catalytic efficiency of the oxoA:dGTP insertion by Dpo4 relative to Polη. The oxoA:dGTP insertion efficiency of Polη is ~8-fold greater than that of Dpo4. The high misincorporation rate for Polη is mainly due to a lower K m value compared to Dpo4. The crystal structure of Polη-oxoA:dGTP* reveals that Gln38 of Polη stabilizes the binding of dGTP at the replicating base pair site by making minor groove contacts to both the templating oxoA and incoming dGTP ( Figure 7D). The removal of the minor groove interactions by Gln38Ala Polη increases the K m by ~20-fold [22], illustrating the role of the minor groove interaction in the oxoA:dGTP insertion. In contrast, the corresponding position where Gln38 of polη is located is occupied by a hydrophobic residue (Val32) in Dpo4, precluding minor groove hydrogen bonding interactions with oxoA:dGTP. In addition, while the 3´-OH of primer terminus dT in the Polη-oxoA:dGTP* structure is optimally positioned for in-line nucleophilic attack, the corresponding 3´-OH in the Dpo4-oxoA:dGTP* structure is in a non-optimal position for the nucleotidyl transfer reaction, which would slow incorporation of dGTP opposite oxoA. Overall, these structural differences may contribute to the observed differences in the replication fidelities of oxoA bypass by Dpo4 and Polη.

CONCLUSION
Here, our kinetic and structural characterizations of the Dpo4-mediated bypass of oxoA provides insights into the miscoding potential of the major oxidative adenine lesion. The dGTP incorporation opposite oxoA by Dpo4 is ~320-fold more efficient than that opposite dA. The substitution of a templating dA for oxoA reduces the fidelity by ~560-fold, suggesting that oxoA promotes mutagenic replication by translesion synthesis DNA polymerases. In the catalytic site of Dpo4, oxoA adopts an anti conformation when paired with an incoming dTTP* and a syn conformation when paired with an incoming dGTP*, highlighting the conformational flexibility of oxoA in the templating position. Interestingly, the oxoA:dGTP base pair adopts a Watson-Crick-like geometry, indicating the promutagenic oxoA:dGTP base pair is well tolerated within the catalytic site of Dpo4. The relative efficiency of the oxoA:dGTP insertion by Dpo4 is ~5-fold greater than that of the oxoG:dATP insertion, indicating the oxoA lesion bypass by some TLS polymerases is more error-prone than the oxoG lesion bypass. In sum, these findings presented here further our understanding of the promutagenic nature of oxoA and the bypass of oxoA in prokaryotes by a translesion synthesis polymerase.  Insertion of dTTP (A) or dGTP (B) opposite oxoA by Dpo4. An annealed DNA of 5´-FAMlabeled primer and an oxoA-modified template was mixed with varying concentrations of dTTP or dGTP, and the reactions were initiated by the addition of Dpo4. All the reactions were conducted at 37 °C, and the quenched reaction samples were separated on 20% denaturing polyacrylamide gels.      d R free = Σ|F(obs)-F(calc)|/ΣF(obs), calculated using 5% of the data.