Structural insights into the bypass of the major deaminated purines by translesion synthesis DNA polymerase.

The exocyclic amines of nucleobases can undergo deamination by various DNA damaging agents such as reactive oxygen species, nitric oxide, and water. The deamination of guanine and adenine generates the promutagenic xanthine and hypoxanthine, respectively. The exocyclic amines of bases in DNA are hydrogen bond donors, while the carbonyl moiety generated by the base deamination acts as hydrogen bond acceptors, which can alter base pairing properties of the purines. Xanthine is known to base pair with both cytosine and thymine, while hypoxanthine predominantly pairs with cytosine to promote A to G mutations. Despite the known promutagenicity of the major deaminated purines, structures of DNA polymerase bypassing these lesions have not been reported. To gain insights into the deaminated-induced mutagenesis, we solved crystal structures of human DNA polymerase η (polη) catalyzing across xanthine and hypoxanthine. In the catalytic site of polη, the deaminated guanine (i.e., xanthine) forms three Watson-Crick-like hydrogen bonds with an incoming dCTP, indicating the O2-enol tautomer of xanthine involves in the base pairing. The formation of the enol tautomer appears to be promoted by the minor groove contact by Gln38 of polη. When hypoxanthine is at the templating position, the deaminated adenine uses its O6-keto tautomer to form two Watson-Crick hydrogen bonds with an incoming dCTP, providing the structural basis for the high promutagenicity of hypoxanthine.


INTRODUCTION
The genomic DNA is under persistent threats by DNA damaging events such as oxidation, UV irradiation, alkylation, and deamination. The exocyclic amines of nucleobases are susceptible to deamination by reactive oxygen species, reactive halogen species, nitric oxide, and water among others. These deamination events give rise to a wide variety of DNA lesions, including uracil (from cytosine), thymine (from 5-methylcytosine), hypoxanthine (from adenine), and xanthine (from guanine) [1][2][3] (Figure 1). Deaminated lesions are promutagenic as they change the donor-acceptor property of base pairing [4][5][6]. The deaminated purines in DNA are mainly recognized and removed by DNA glycosylases such as E. coli AlkA or Endo V in prokaryotes, and alkyladenine DNA glycosylase (AAG) or the homologs of Endo V in eukaryotes [1,[7][8][9][10][11] and the resulting abasic site is further processed by the downstream base excision repair (BER) enzymes.
Xanthine, as its monophosphate form (dXMP), is a key intermediate for de novo synthesis of dGMP, and its concentration (both dXMP and free xanthine) is closely related to the concentration of the entire guanine nucleotide pool [12,13]. Xanthine, which can arise by deamination of guanine in DNA, is a highly promutagenic damage. DNA polymerase α in eukaryotes frequently misincorporates dTMP opposite xanthine. While mammalian polβ accurately bypasses xanthine [14], translesion synthesis DNA polymerases η (polη) and κ (polη) preferentially incorporate dTTP opposite xanthine [15]. In mammalian cells, xanthine is a miscoding lesion that significantly promotes G to A transitions [16,17].
Hypoxanthine, the deaminated adenine, preferentially base pairs with dCTP and causes A to G mutations in many organisms [18,19]. The triphosphate form of hypoxanthine, dITP, is taken up readily by DNA polymerases [20]. Hypoxanthine is a significant block to replication by B-family Pfu DNA polymerase and Sulfolobus solfataricus Dpo1, while it is readily bypassed by Taq DNA polymerase and Y-family DNA polymerase PolY1, and Sulfolobus solfataricus Dpo4 [21,22]. Human DNA polymerases α, η, and truncated κ (κΔC) insert dCTP quite exclusively over dTTP [23]. In HEK293 and HCT116 human cell lines, hypoxanthine induces A to G mutations and deletion [24].
While there are many published studies on the promutagenicity of the deaminated purines, a structure of DNA polymerase bypassing those lesions has not been reported, significantly limiting our understanding of hypoxanthine/xanthine-induced mutagenesis. Herein, we present steady-state kinetic data of human polη incorporating nucleotide opposite deaminated purine lesions, xanthine and hypoxanthine, along with two crystal structures of polη complexed with a templating xanthine/hypoxanthine and incoming non-hydrolyzable dCTP analog. These crystal structures represent the first structure of DNA polymerase bypassing xanthine and hypoxanthine. Our studies revealed the base pairing properties of the major deaminated purines in the active site of polη, which provides valuable insights into the bypass of the deaminated lesions by DNA polymerases, especially translesion synthesis polymerases.

Protein expression and purification.
Polη was expressed and purified from E. coli with minor modifications of the method described previously [32]. Briefly, Polη was overexpressed in E. coli BL21(DE3) cells, and cultures were grown in Luria-Bertani medium at 37 °C until reaching the OD 600 of 0.7, and the cells were induced by adding 0.2 mM isopropyl β-D-α-thiogalactopyranoside. After incubating for 18 hours at 20 °C, the pelleted cells (6,000 RPM for 30 min) were resuspended in Ni-NTA column binding buffer A (50 mM sodium phosphate, pH 7.8, 500 mM NaCl and 10% glycerol) supplemented with 1 mg/ml lysozyme, 0.25% NP-40, 0.25% Triton X-100, and 0.25 mM phenylmethylsulfonyl fluoride (PMSF). After sonication for 90 seconds, the lysate was centrifuged at 15,000 g at 4 °C for 20 min. The supernatant was then filtered through 0.22 μm filter and further purified through Ni-NTA column (GE Healthcare). The elution fractions were pooled and further purified using the Heparin HiTrap column (GE Healthcare) followed by Superdex-75 size exclusion chromatography (GE Healthcare). The purity of the final product was confirmed by SDS-PAGE gel. The purified protein was concentrated, flash-frozen in liquid nitrogen, and stored at −80 °C for the future use.

Protein-DNA crystallization and structure determination.
Xanthine (XT)-and hypoxanthine (HX)-containing DNA were custom synthesized by Midland Certified Reagent Co. (Midland, TX). The primer (5′-AGCGTCAT-3′) was purchased from Integrated DNA Technologies (Coralville, IA). XT/HX-containing 12-mer template (5′-CAT[XT/HX]CTCACACT-3′) was annealed with complementary 8-mer primer (5′-AGTGTGAG-3′) in hybridization buffer (10 mM Tris-HCl pH 7.5, 30 mM NaCl, 1 mM EDTA) by heating for 5 min at 90°C followed by slow cooling to room temperature. The annealed lesion-containing DNA was incubated with ~9 mg/ml polη to form protein-DNA binary complexes with 1.2:1 molar ratio. Subsequently, a 10-fold molar excess of nonhydrolyzable dCMPNPP (Jena Bioscience) was added to the binary complex. The ternary polη-DNA complex co-crystals with nonhydrolyzable dCMPNPP (dCTP* hereafter) paired with templating XT/HX were grown in a buffer solution containing 100 mM MES pH 6.5, 14-23% PEG2000 MME, and 5 mM magnesium chloride. Crystals were cryoprotected in mother liquor supplemented with 20% glycerol and were flash-frozen in liquid nitrogen. Diffraction data were collected at 100 K at the beamline 23-ID-D at the Advanced Photon Source, Argonne National Laboratory. All diffraction data were processed using HKL 2000 [37], and the structures were solved by molecular replacement using Molrep [38]. Polη structure with an undamaged DNA (PDB ID 4O3N) was used as a search model. The model was built using COOT [39] and refined using PHENIX [40]. MolProbity was used to make Ramachandran plots [41]. All the crystallographic figures were generated by using Chimera [42].

Steady-state kinetics of single nucleotide incorporation opposite templating xanthine/ hypoxanthine by polη.
Steady-state kinetic parameters for nucleotide insertion opposite XT/HX by polη were measured as described previously [29,32]. Briefly, The oligonucleotides for kinetic assays (primer, 5´-FAM/GGGGG CTCGTAAGGATTC-3' and template, 5´-CCGACT[XT/ HX]GAATCCTTACGAGCCCCC-3´) were synthesized by Midland Certified Reagent company (Midland, TX) and Integrated DNA Technologies (Coralville, IA).To prepare DNA substrate containing XT/HX, each oligonucleotide was annealed in hybridization buffer (10 mM Tris-HCl pH 7.5, 30 mM NaCl, 1 mM EDTA) by heating for 5 min at 90°C followed by slow cooling to room temperature. Enzyme activities were determined using the reaction mixture containing 40 mM Tris-HCl pH 7.5, 60 mM KCl, 10 mM dithiothreitol, 250 μg/ml bovine serum albumin, 2.5 % glycerol, 5 mM MgCl 2 , 80 nM primer/template DNA, and the different concentration of incoming dNTP. To prevent end-product inhibition and substrate depletion from interfere with accurate velocity measurement, the enzyme concentrations and reaction-time intervals were adjusted for every experiment (less than 20% insertion product formed). The reactions were initiated by the addition of the enzyme and stopped with a gelloading buffer (95% formamide with 20 mM EDTA, 45 mM Tris-borate, 0.1% bromophenol blue, 0.1% xylene cyanol). The quenched samples were separated on 20% denaturing polyacrylamide gels. The gels were analyzed using ImageQuant (GE Healthcare) to quantify product formation. The k cat and K m were determined by fitting reaction rate over dNTP concentrations to Michaelis-Menten equation. Each experiment was repeated three times to measure the average of the kinetic results. The catalytic efficiency of nucleotide insertion was calculated as k cat /K m .

Xanthine uses its enol tautomer to form a Watson-Crick-like base pair with dCTP in polη active site.
Our kinetic studies show that polη efficiently incorporates dCTP opposite templating XT, which promotes accurate replication. To gain structural insight into the correct nucleotide insertion, we solved a ternary complex structure of polη incorporating a nonhydrolyzable dCMPNPP (dCTP* hereafter) opposite templating XT in the presence of Mg 2+ cofactors. The nonhydrolyzable nucleotide dCTP* was used because it is isosteric to dCTP, and its coordination to the active-site metal ions is essentially identical to that of dCTP [43]. The polη-XT:dCTP* ternary complex was crystallized in P6 1 space group with the cell dimension of a = 98.93 Å, b = 98.93 Å, c = 81.55 Å, α = 90.00°, β = 90.00°, and γ = 120.00°. The polη-XT:dCTP* ternary structure was refined to a resolution of 2.35 Å with R work = 18.9 % and R free = 23.8 % ( Table 2).
The polη-XT:dCTP* ternary complex structure provides the structural basis for correct insertion opposite XT by the enzyme (Figure 3). This 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 XT:dCTP* base pair is well accommodated within the enzyme's catalytic site ( Figure 3B) with strong electron density around XT and the incoming dCTP*, indicating the base pair is well accommodated in the catalytic site of polη. The guanidinium moiety of Arg61 engages in stacking interaction with the base of incoming dCTP* and Arg55 forms hydrogen bonds with phosphate oxygens of dCTP*, which stabilizes the incoming nucleotide in the catalytic site ( Figure 3B). The primer terminus 3´-OH is coordinated to the A-site magnesium ion and is about 3.4 Å away from the P α of dCTP* ( Figure 3C), thereby being optimally positioned for in-line nucleophilic attack on the P α of the dCTP*. The incoming dCTP* and primer terminus dT at the N-1 position are favorably positioned for stacking interaction ( Figure 3C). The templating XT adopts an anti-conformation and forms a Watson-Crick-like base pair with dCTP* with the inter-base hydrogen bonding distances of 3.2 Å, 2.8 Å, and 2.7 Å ( Figure  3D). The geometry of XT:dCTP* base pair displays the λ angles of 61.9° (XT) and 55.0° (dCTP*) and the C1´-C1´ distance of 10.4 Å ( Figure 3D), which are very similar to those of correct undamaged base pairs. The O2 of XT is 2.7 Å away from the O2 of dCTP*, indicating the formation of a strong hydrogen bond between the O2 atoms ( Figure 3E). This, in turn, indicates that the O2-enol tautomer of XT engages in hydrogen bonding interaction with the O2 of the incoming dCTP. The enol tautomerization of the O2 of XT nucleotide may be promoted the minor groove contact by Gln38, which is hydrogen bonded to the N3 of the templating XT. Overall, the polη-XT:dCTP* ternary complex structure with three Watson-Crick-like hydrogen bonds and two metal ions is consistent with the efficient dCTP insertion opposite XT by the enzyme.
Hypoxanthine forms a Watson-Crick base pair with incoming dCTP in the active site of polη.
Our kinetic studies show that polη efficiently catalyzes the insertion of dCTP opposite HX, which can promote A to G transition mutations. To gain structural insight into this incorrect bypass, we solved a ternary complex structure of polη incorporating a nonhydrolyzable dCMPNPP (dCTP* hereafter) opposite templating HX in the presence Mg 2+ ions. The polη-HX:dCTP* ternary complex was crystallized in P6 1 space group with the cell dimension of a = 98.54 Å, b = 98.54 Å, c = 81.66 Å, α = 90.00°, β = 90.00°, and γ = 120.00 °. The polη-HX:dCTP* ternary structure was refined to a resolution of 1.97 Å with R work = 17.1 % and R free = 20.8 % ( Table 2).
The polη-HX:dCTP* ternary complex structure provides the structural basis for the promutagenic replication past HX by the enzyme (Figure 4). As observed in the polη-XT:dCTP* structure, the polη-HX:dCTP* structure displays the conserved four characteristic domains of Y-family DNA polymerases ( Figure 4A). The incorrect HX:dCTP* base pair is well accommodated within the enzyme's catalytic site ( Figure 4B) with strong electron density around HX and dCTP*. Both catalytic (A-site) and nucleotide-binding (Bsite) metal ions are present in the catalytic site. Arg55 is hydrogen bonded to the phosphate oxygens of dCTP* and Arg61 forms π-stacking interaction with the nucleobase of dCTP*, thereby stabilizing the incoming nucleotide in the catalytic site ( Figure 4B). The primer terminus 3´-OH is coordinated to the A-site magnesium ion and is about 3.1 Å away from the P α of the incoming dCTP* ( Figure 4C), thereby being poised for in-line nucleophilic attack on dCTP*. The templating HX is in an anti-conformation and forms a Watson-Crick base pair with dCTP* with the inter-base hydrogen bonding distances of 2.8 Å (between N1 of HX and N3 of dCTP*) and 3.3 Å (between O6 of HX and N4 of dCTP*). The base-pair geometry of HX:dCTP* base pair displays the λ angles of 59.3° (HX) and 60.2° (dCTP*) and the C1´-C1´ distance of 10.5 Å ( Figure 4D), which are essentially identical to those of an undamaged correct base pair. The Watson-Crick HX:dCTP* base pair indicates the O6keto tautomer of HX involves in base pairing with dCTP*. Overall, the polη-HX:dCTP* ternary complex structure with Watson-Crick base pair and two magnesium ions is consistent with the efficient insertion of dCTP opposite HX by the polη.
The comparison of the polη-HX:dCTP* and polη-XT:dCTP* structures reveals conformational differences are mainly confined to the 5´ side of the templating purines and the primer terminus base ( Figure 5). Both HX:dCTP* and XT:dCTP* base pairs are well accommodated in the catalytic site without significant change of protein conformation. The conformations of the incoming dCTP in both structures are essentially identical. On the other hand, the conformation of the 5´ phosphate of the templating deaminated purines differ greatly ( Figure 5C and 5D). While the dT at the N+1 position engages in stacking interaction with templating XT, it does not stack with the templating HX ( Figure 5A). These conformational differences could contribute to the difference in the catalytic efficiency of dCTP insertion opposite HX and XT.
The deaminated purines and dG adopt similar conformation in the active site of polη.
The polη-xanthine:dCTP* and polη-hypoxanthine:dCTP* structures are very similar to the published polη-dG:dCTP* structure (PDB ID: 4O3N, Figure 6) [44]. In the active site of polη, dG and dCTP form Watson-Crick base pair ( Figure 6A). The templating dT at the N+1 position engages in π-stacking interaction with the templating dG at the N position. The superposition of the polη-XT:dCTP* structure with the polη-dG:dCTP* structure reveals that the conformations of templating XT/dG, incoming dCTP*, primer strand, and downstream template bases are essentially identical ( Figure 6B). Also, the guanidine moieties of Arg61 in the two structures stabilize the incoming dCTP* via π-stacking interaction ( Figure 6C). In addition, Arg55, which interacts with the γ-phosphate of incoming nucleotide, adopts the same conformation in the two structures. The only significant deviation is found at the templating bases at the N+1 and N+2 positions. The conformation of dT (N+1) in the polη-XT:dCTP* complex is not in the optimal position compared with that in polη-dG:dCTP* structure. In addition, dA at the N+2 position in the polη-XT:dCTP* complex shifts ~4 Å relative to the position in the polη-dG:dCTP* complex. These large conformational differences at the N+1 and N+2 positions may contribute to a 4-fold reduction in the catalytic efficiency ( Table 1).
The superposition of the polη-HX:dCTP* structure with the polη-dG:dCTP* structure (RMSD: 0.220 Å) shows that a significant conformational change is confined to the templating dT at the N+1 position. The templating dT(N+1) in the polη-HX:dCTP* structure rotates ~90° toward the major groove relative to the position in the polη-dG:dCTP structure, thereby not engaging in stacking interaction with the templating HX at the N position. On the other hand, the templating dA at the N+2 position of the polη-HX:dCTP* overlays very well with the corresponding dA of the polη-dG:dCTP* structure. In addition, the templating base at the N position, incoming dCTP*, primer strand, and downstream template bases in the two structures do not exhibit a significant conformational deviation ( Figure 6D). The guanidine moieties of Arg61 in the two structures stabilize incoming dCTP via stacking interaction ( Figure 6E). In addition, the orientation and conformation of Arg55 in the two structures are essentially the same (not shown). Interestingly, despite the disruption of the stacking interaction between dT(N+1) and HX(N), the catalytic efficiency of the HX:dCTP insertion is only slightly lower than that of the dG:dCTP insertion (37.4 vs. 45.6), indicating the stacking interaction between the templating bases at the N and N+1 positions may not significantly contribute to the catalytic efficiency of polη.

Promutagenic insertion opposite XT and HX by DNA polymerases.
Several kinetic reports on xanthine [15,45] and hypoxanthine [21,23,45] (Table 1) have highlithged the mutagenic potential of the major deaminated purines. In particular, bypass properties of XT and HX by human DNA polymerases α (polα), β (polβ), and κ (polκ) have been characterized [15,23]. For example, X-family polβ preferentially incorporates the correct nucleotide opposite XT with the fidelity (dCTP/dTTP) of ~70 (Table 1). On the other hand, Y-family polκ frequently incorporates dTTP with the fidelity (dCTP/dTTP) of ~0.5 (Table 1). In the case of Y-family polη (our study), dCTP incorporation opposite XT is only ~3-fold more efficient than dTTP insertion, highlighting the bypass fidelity of XT is greatly influenced by the microenvironment of DNA polymerase active site. At physiological pH, xanthine exists as an almost equal mixture of neutral and monoanionic (enolate) species ( Figure 7A) [46], the latter of which can take on an enol tautomer upon protonation. Our polη-XT:dCTP* crystal structure displays hydrogen bonding interactions between O2 of XT and O2 of dCTP* with the distance of 2.7 Å, which indicates that O2 of XT is in the enol tautomeric state ( Figure 7B). This observation is consistent with the higher insertion efficiency of dCTP over dTTP opposite XT in the active site of polη.
Varying degrees of minor groove contacts to the N3 of XT by DNA polymerases may contribute to the fidelity of XT bypass. Polκ does not engage in minor groove interaction at the replicating base pair site. In the case polη, the enzyme typically uses Gln38 to form hydrogen bonds with the minor groove edge and O4´ of a templating base during correct insertion, while it does not interact with the minor groove edge of incoming nucleotide. The minor groove contact to a templating XT could promote the formation of the enol tautomer of the base, which in turn can modulate the efficiency and fidelity of XT bypass.
While the bypass fidelity of XT varies significantly among DNA polymerases, that of HX varies to a much lesser extent. This would be because HX exists predominantly as the O6keto tautomeric species in base pairing ( Figure 7C), whereas XT exists as a ~1:1 mixture of the O2-keto and O2-enol tautomers and its tautomeric ratio can be significantly influenced by the microenvironment of DNA polymerases. In all three human DNA polymerases, polα, polη, and polκΔC, dCTP is preferentially inserted opposite HX with the fidelity (dTTP/ dCTP) ranging from 0.008 (hpolκΔC) to 0.038 (hpolη), which can facilitate A to G mutations. The crystal structure of the polη-HX:dCTP* complex shows that hypoxanthine behaves much like guanine and uses its O6-keto tautomer when paired with dCTP.

CONCLUSIONS
The crystal structure of the polη-HX:dCTP* complex shows the formation of two hydrogen bonds with a Watson-Crick geometry, which is consistent with the preferential insertion of dCTP opposite HX. The crystal structure of the polη-XT:dCTP* complex reveals that XT forms three Watson-Crick-like hydrogen bonds with incoming dCTP*, indicating the enol tautomeric species of XT involves in the correct nucleotide insertion. The enol tautomerization of the O2 of XT appears to be promoted by the minor groove contact by Gln38 of polη. These structures, which represent the first structures of TLS polymerase bypassing deaminated purines, provide structural insights into the mutagenic potential of the major deaminated purines.

Data Availability Statement
The atomic coordinates of polη-DNA complexes have been deposited in the Protein Data Bank with the following accession codes: polη-XT:dCTP (PDB Code: 6WK6) and polη-HX:dCTP (PDB Code: 6MQ8).  Deamination of purines alters hydrogen bonding properties of the bases. "A" and "B" denote hydrogen bond acceptor and donor, respectively. Incorporation of dCTP opposite xanthine (A) and hypoxanthine (B) by polη. An annealed DNA of 5´-FAM-labeled primer and XT-or HX-containing template was mixed with varying concentrations of dCTP, and the reactions were initiated by the addition of polη. 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.