Bypass of the Major Alkylative DNA Lesion by Human DNA Polymerase η

A wide range of endogenous and exogenous alkylating agents attack DNA to generate various alkylation adducts. N7-methyl-2-deoxyguanosine (Fm7dG) is the most abundant alkylative DNA lesion. If not repaired, Fm7dG can undergo spontaneous depurination, imidazole ring-opening, or bypass by translesion synthesis DNA polymerases. Human DNA polymerase η (polη) efficiently catalyzes across Fm7dG in vitro, but its structural basis is unknown. Herein, we report a crystal structure of polη in complex with templating Fm7dG and an incoming nonhydrolyzable dCTP analog, where a 2′-fluorine-mediated transition destabilization approach was used to prevent the spontaneous depurination of Fm7dG. The structure showed that polη readily accommodated the Fm7dG:dCTP base pair with little conformational change of protein and DNA. In the catalytic site, Fm7dG and dCTP formed three hydrogen bonds with a Watson–Crick geometry, indicating that the major keto tautomer of Fm7dG is involved in base pairing. The polη-Fm7dG:dCTP structure was essentially identical to the corresponding undamaged structure, which explained the efficient bypass of the major methylated lesion. Overall, the first structure of translesion synthesis DNA polymerase bypassing Fm7dG suggests that in the catalytic site of Y-family DNA polymerases, small N7-alkylguanine adducts may be well tolerated and form the canonical Watson–Crick base pair with dCTP through their keto tautomers.


Structure of Polη Incorporating dCTP opposite Templating Fm7dG
To gain structural insights into how TLS DNA polymerases bypass the major alkylative DNA lesion, we determined a ternary structure of polη in complex with templating Fm7dG paired with incoming nonhydrolyzable dCMPNPP (hereinafter dCTP*) ( Figure 3A). The nonhydrolyzable dCTP analog was used for this crystallographic study because it is isosteric to the natural nucleotide and its coordination with Mg 2+ ions is virtually identical to that of dCTP. The NH moiety of dCMPNPP replaces the bridging oxygen between Pα and Pβ, making the analog resistant to dCMP transfer as well as hydrolysis, which enables the capture of a ternary complex of a catalytically active polη bound to Fm7dG:dCTP* in the presence of Mg 2+ . This also allows the coordination of the primer terminus 3′-OH to the catalytic metal ion. The nonhydrolyzable dNTP* analog has been used in crystallographic studies of various DNA polymerases [38][39][40]. The polη-Fm7dG:dCTP* ternary complex was crystallized in the P61 space group with the cell dimension of a = b = 98.7 Å, c = 81.8 Å, α = β = 90.0°, and γ = 120° and one protein in the asymmetric unit. The polη ternary complex structure was refined to 2.3 Å with Rwork = 17.1% and Rfree = 23.3%.
The overall conformation of the polη-Fm7dG:dCTP* ternary structure is essentially identical to that of the published ternary structure with dG:dCTP insertion (PDB ID: 4O3N) [41]: The root-meansquare deviation for these structures was 0.211 Å. ( Figure 3B). The polη-Fm7dG:dCTP* structure displayed the thumb, palm, finger, and little finger domains of Y-family DNA polymerases ( Figure  3A). The incoming nucleotide resided between the palm and finger domains, whereas the templating Fm7dG was positioned between the finger and little finger domains. The statistics for data collection and the refinement are summarized in Table 2.

Structure of Polη Incorporating dCTP Opposite Templating Fm7dG
To gain structural insights into how TLS DNA polymerases bypass the major alkylative DNA lesion, we determined a ternary structure of polη in complex with templating Fm7dG paired with incoming nonhydrolyzable dCMPNPP (hereinafter dCTP*) ( Figure 3A). The nonhydrolyzable dCTP analog was used for this crystallographic study because it is isosteric to the natural nucleotide and its coordination with Mg 2+ ions is virtually identical to that of dCTP. The NH moiety of dCMPNPP replaces the bridging oxygen between P α and P β , making the analog resistant to dCMP transfer as well as hydrolysis, which enables the capture of a ternary complex of a catalytically active polη bound to Fm7dG:dCTP* in the presence of Mg 2+ . This also allows the coordination of the primer terminus 3 -OH to the catalytic metal ion. The nonhydrolyzable dNTP* analog has been used in crystallographic studies of various DNA polymerases [38][39][40]. The polη-Fm7dG:dCTP* ternary complex was crystallized in the P61 space group with the cell dimension of a = b = 98.7 Å, c = 81.8 Å, α = β = 90.0 • , and γ = 120 • and one protein in the asymmetric unit. The polη ternary complex structure was refined to 2.3 Å with R work = 17.1% and R free = 23.3%.
The overall conformation of the polη-Fm7dG:dCTP* ternary structure is essentially identical to that of the published ternary structure with dG:dCTP insertion (PDB ID: 4O3N) [41]: The root-mean-square deviation for these structures was 0.211 Å. ( Figure 3B). The polη-Fm7dG:dCTP* structure displayed the thumb, palm, finger, and little finger domains of Y-family DNA polymerases ( Figure 3A). The incoming nucleotide resided between the palm and finger domains, whereas the templating Fm7dG was positioned between the finger and little finger domains. The statistics for data collection and the refinement are summarized in Table 2.     The conformational difference between the polη-Fm7dG:dCTP* and published polη-dG:dCTP* structures (PDB ID: 4O3N) was confined to Fm7dG and the upstream of Fm7dG ( Figure 4A); their protein conformations were essentially the same ( Figure 3B). The nonbridging phosphate oxygen of the 5 -phosphodiester of Fm7dG shifted 1.3 Å relative to the dG:dCTP* structure ( Figure 4B). Slight conformational difference was also observed at the 5 side of the templating bases. It appears that the spacious catalytic site of polη well accommodates small alkyl-dG lesions with little distortion in protein and DNA conformations. most favored (%) 96.2 add. allowed (%) 3.3 RMSD bond lengths (Å) bond angles (degree) 0.009 1.55 a Values in parentheses are for the highest resolution shell; b Rmerge = Σ|I − <I>|/ΣI, where I is the integrated intensity of a given reflection; c Rwork = Σ|F(obs) − F(calc)|/ΣF(obs); d Rfree = Σ|F(obs) − F(calc)|/ΣF(obs), calculated using 5% of the data.
The conformational difference between the polη-Fm7dG:dCTP* and published polη-dG:dCTP* structures (PDB ID: 4O3N) was confined to Fm7dG and the upstream of Fm7dG ( Figure 4A); their protein conformations were essentially the same ( Figure 3B). The nonbridging phosphate oxygen of the 5′-phosphodiester of Fm7dG shifted 1.3 Å relative to the dG:dCTP* structure ( Figure 4B). Slight conformational difference was also observed at the 5′ side of the templating bases. It appears that the spacious catalytic site of polη well accommodates small alkyl-dG lesions with little distortion in protein and DNA conformations. The ternary complex structure revealed the base-pairing characteristics of Fm7dG and dCTP* in the replicating base pair site ( Figure 5). A 2Fo − Fc electron density map contoured at 1σ around Fm7dG:dCTP* clearly showed the presence of the N7 methyl moiety and 2′-β-fluorine of Fm7dG and incoming dCTP*. Polη tolerated the Fm7dG:dCTP* base pair with little distortion of the catalytic site. In particular, the flexible Arg61-Trp64 loop [32,36,38,42,43], which undergoes a significant conformational change upon binding of bulky DNA lesions (e.g., cyclobutene pyrimidine dimers, cisplatin-GpG, oxaliplatin-GpG, and phenanthriplatin-G), adopted essentially the same conformation as observed in the undamaged structure. In the nascent base pair site, templating Fm7dG formed a coplanar base pair with dCTP* and engaged in stacking interactions with the adjacent 5′ and 3′ bases ( Figure 5). The guanidinium moiety of Arg61 was oriented toward and stacked with the incoming dCTP*. The ternary complex structure revealed the base-pairing characteristics of Fm7dG and dCTP* in the replicating base pair site ( Figure 5). A 2F o − F c electron density map contoured at 1σ around Fm7dG:dCTP* clearly showed the presence of the N7 methyl moiety and 2 -β-fluorine of Fm7dG and incoming dCTP*. Polη tolerated the Fm7dG:dCTP* base pair with little distortion of the catalytic site. In particular, the flexible Arg61-Trp64 loop [32,36,38,42,43], which undergoes a significant conformational change upon binding of bulky DNA lesions (e.g., cyclobutene pyrimidine dimers, cisplatin-GpG, oxaliplatin-GpG, and phenanthriplatin-G), adopted essentially the same conformation as observed in the undamaged structure. In the nascent base pair site, templating Fm7dG formed a coplanar base pair with dCTP* and engaged in stacking interactions with the adjacent 5 and 3 bases ( Figure 5). The guanidinium moiety of Arg61 was oriented toward and stacked with the incoming dCTP*.  Figure  3A. The Arg61-Trp64 loop is shown in blue.
A close-up view of the metal-binding site gives insights into the facile bypass of Fm7dG by polη ( Figure 6A, Table 1). Both the catalytic "metal A" and nucleotide-binding "metal B" ions were observed in the active site. The catalytic Mg 2+ ion was coordinated with the 3′-OH of the primer terminus, a nonbridging oxygen of Pα, and catalytic carboxylates (Asp13, Asp115, and Glu116). The nucleotide-binding metal ion was complexed with Asp13, Met14, Asp115, and nonbridging oxygens of Pβ and Pγ. The nucleophilic 3′-hydroxyl group of the primer terminus was 3.6 Å from the Pα and poised for the in-line nucleophilic attack on the Pα, which would facilitate nucleotidyl transfer. Overall, the observation of the Fm7dG:dCTP* base pair with an ideal Watson-Crick geometry, together with the favorable metal ion coordination for catalysis, is consistent with the efficient incorporation of dCTP opposite Fm7dG by the enzyme (Table 1) [29].   Figure 3A. The Arg61-Trp64 loop is shown in blue.
A close-up view of the metal-binding site gives insights into the facile bypass of Fm7dG by polη ( Figure 6A, Table 1). Both the catalytic "metal A" and nucleotide-binding "metal B" ions were observed in the active site. The catalytic Mg 2+ ion was coordinated with the 3 -OH of the primer terminus, a nonbridging oxygen of P α , and catalytic carboxylates (Asp13, Asp115, and Glu116). The nucleotide-binding metal ion was complexed with Asp13, Met14, Asp115, and nonbridging oxygens of P β and P γ . The nucleophilic 3 -hydroxyl group of the primer terminus was 3.6 Å from the P α and poised for the in-line nucleophilic attack on the P α , which would facilitate nucleotidyl transfer. Overall, the observation of the Fm7dG:dCTP* base pair with an ideal Watson-Crick geometry, together with the favorable metal ion coordination for catalysis, is consistent with the efficient incorporation of dCTP opposite Fm7dG by the enzyme (Table 1) [29].  Figure  3A. The Arg61-Trp64 loop is shown in blue.
A close-up view of the metal-binding site gives insights into the facile bypass of Fm7dG by polη ( Figure 6A, Table 1). Both the catalytic "metal A" and nucleotide-binding "metal B" ions were observed in the active site. The catalytic Mg 2+ ion was coordinated with the 3′-OH of the primer terminus, a nonbridging oxygen of Pα, and catalytic carboxylates (Asp13, Asp115, and Glu116). The nucleotide-binding metal ion was complexed with Asp13, Met14, Asp115, and nonbridging oxygens of Pβ and Pγ. The nucleophilic 3′-hydroxyl group of the primer terminus was 3.6 Å from the Pα and poised for the in-line nucleophilic attack on the Pα, which would facilitate nucleotidyl transfer. Overall, the observation of the Fm7dG:dCTP* base pair with an ideal Watson-Crick geometry, together with the favorable metal ion coordination for catalysis, is consistent with the efficient incorporation of dCTP opposite Fm7dG by the enzyme (Table 1) [29].  The methyl modification on guanine N7 decreased the catalytic efficiency of polη and polβ by~3-and~300-fold, respectively, which suggests a differential impact of Fm7dG on polymerase activity [28]. The dramatic difference in the catalytic efficiency may have resulted from the variation in the sensitivity of polymerases toward the lesion. In the case of polβ, while Fm7dG:dCTP* formed the canonical Watson-Crick base pairing ( Figure 6B), coordination [44,45] between the 3 -OH of the primer terminus and the catalytic metal ion was lacking. In addition, the distance between the primer terminus 3 -OH and the P α of dCTP* was longer than that for correct insertion (4.8 Å vs.~3.4 Å). Furthermore, the nucleophilic 3 -OH was pointed away from the P α of dCTP*, thereby assuming a catalytically unfavorable conformation. This nonoptimal conformation for nucleotidyl transfer would slow the catalysis by polβ [28]. The polβ-Fm7dG:dCTP* complex would require conformational adjustments to reach the catalytically competent state, which may be a slow or rate-limiting step in catalysis. In the case of polη, the metal ion coordination for the polη-Fm7dG:dCTP* complex was essentially identical to that for the polη-dG:dCTP* complex and optimally positioned for catalysis.
The base-pairing characteristics of Fm7dG:dCTP* were very similar to those of dG:dCTP* (Figure 7). In the polη active site, Fm7dG:dCTP* adopted the canonical Watson-Crick base pair with an average hydrogen bond distance of 2.8 Å ( Figure 7A). Specifically, the O6, N1, and N2 of Fm7dG were hydrogen bonded to the N4, N3, and O2 of dCTP*, respectively, with distances of 3.0, 2.8, and 2.7 Å. The C1 -C1 distance for Fm7dG:dCTP* was 10.6 Å and the λ angles for Fm7dG and dCTP* were 57.8 • and 56.0 • , respectively, which were almost the same as observed in normal Watson-Crick base pairing. The b-factor value of the nucleobases of Fm7dG and dCTP* was in the 9-15 range (not shown), illustrating that the Fm7dG:dCTP* base pair was well ordered in the nascent base pair site. The propeller-twist angle for Fm7dG:dCTP* was~10 • , signifying that Fm7dG did not significantly distort base pair conformation ( Figure 7B). Altogether, the Fm7dG:dCTP* base pair was virtually indistinguishable from the dG:dCTP base pair, indicating that the positively charged N7 methyl moiety negligibly alters the base-paring property during correct nucleotide incorporation. The methyl modification on guanine N7 decreased the catalytic efficiency of polη and polβ by ~3-and ~300-fold, respectively, which suggests a differential impact of Fm7dG on polymerase activity [28]. The dramatic difference in the catalytic efficiency may have resulted from the variation in the sensitivity of polymerases toward the lesion. In the case of polβ, while Fm7dG:dCTP* formed the canonical Watson-Crick base pairing ( Figure 6B), coordination [44,45] between the 3′-OH of the primer terminus and the catalytic metal ion was lacking. In addition, the distance between the primer terminus 3′-OH and the Pα of dCTP* was longer than that for correct insertion (4.8 Å vs. ~3.4 Å). Furthermore, the nucleophilic 3′-OH was pointed away from the Pα of dCTP*, thereby assuming a catalytically unfavorable conformation. This nonoptimal conformation for nucleotidyl transfer would slow the catalysis by polβ [28]. The polβ-Fm7dG:dCTP* complex would require conformational adjustments to reach the catalytically competent state, which may be a slow or rate-limiting step in catalysis. In the case of polη, the metal ion coordination for the polη-Fm7dG:dCTP* complex was essentially identical to that for the polη-dG:dCTP* complex and optimally positioned for catalysis.
The base-pairing characteristics of Fm7dG:dCTP* were very similar to those of dG:dCTP* (Figure 7). In the polη active site, Fm7dG:dCTP* adopted the canonical Watson-Crick base pair with an average hydrogen bond distance of 2.8 Å ( Figure 7A). Specifically, the O6, N1, and N2 of Fm7dG were hydrogen bonded to the N4, N3, and O2 of dCTP*, respectively, with distances of 3.0, 2.8, and 2.7 Å. The C1′-C1′ distance for Fm7dG:dCTP* was 10.6 Å and the λ angles for Fm7dG and dCTP* were 57.8° and 56.0°, respectively, which were almost the same as observed in normal Watson-Crick base pairing. The b-factor value of the nucleobases of Fm7dG and dCTP* was in the 9-15 range (not shown), illustrating that the Fm7dG:dCTP* base pair was well ordered in the nascent base pair site. The propeller-twist angle for Fm7dG:dCTP* was ~10°, signifying that Fm7dG did not significantly distort base pair conformation ( Figure 7B). Altogether, the Fm7dG:dCTP* base pair was virtually indistinguishable from the dG:dCTP base pair, indicating that the positively charged N7 methyl moiety negligibly alters the base-paring property during correct nucleotide incorporation. The formation of Fm7dG:dCTP* with Watson-Crick geometry strongly suggests that, in the polη active site, the keto tautomer of Fm7dG, rather than the zwitterionic or enol tautomeric form of Fm7dG, exists as the major tautomer ( Figure 1A). The enolate or enol tautomer of Fm7dG has been observed when a lesion is paired with dT in the absence of polymerase contact [27]. The suppression of zwitterionic species could arise from the electron-rich microenvironment (e.g., bases and phosphate anions) around Fm7dG, which can diminish the impact of the positively charged N7methyl group on the ionization of N1 of Fm7dG.
In summary, the first structure of TLS polymerase catalyzing across a major methylated DNA lesion suggests that small N7-alkylguanine adducts can be readily accommodated in the catalytic site The formation of Fm7dG:dCTP* with Watson-Crick geometry strongly suggests that, in the polη active site, the keto tautomer of Fm7dG, rather than the zwitterionic or enol tautomeric form of Fm7dG, exists as the major tautomer ( Figure 1A). The enolate or enol tautomer of Fm7dG has been observed when a lesion is paired with dT in the absence of polymerase contact [27]. The suppression of zwitterionic species could arise from the electron-rich microenvironment (e.g., bases and phosphate anions) around Fm7dG, which can diminish the impact of the positively charged N7-methyl group on the ionization of N1 of Fm7dG.
In summary, the first structure of TLS polymerase catalyzing across a major methylated DNA lesion suggests that small N7-alkylguanine adducts can be readily accommodated in the catalytic site of TLS polymerases. In the polη active site, Fm7dG and dCTP* form three hydrogen bonds with an ideal Watson-Crick geometry, which explains the highly efficient nucleotidyl transfer opposite Fm7dG. It is highly likely that small N7-alkylguanine lesions engage in the canonical Watson-Crick base pairing with dCTP through their keto tautomers. It would be of interest to evaluate the impact of the steric bulkiness of the alkyl moiety (e.g., ethyl, benzyl, and nitrogen half mustard) on the base pairing characteristics of N7-alkylguanine lesions.

Synthesis of Fm7dG-Containing Oligonucleotide
The Fm7dG phosphoramidite was synthesized starting from a ribose derivative according to the synthetic procedures described previously [26]. The site-specific incorporated Fm7dG-containing DNA was custom synthesized by Midland Certified Reagent company (Midland, TX).

Cloning and Protein Expression and Purification
The catalytic core of human polη (amino acids 1-432) was cloned into pET28a plasmid with NcoI and BamHI restriction enzyme sites. E. coli BL21 (DE3) cells transformed with this plasmid were grown at 37 • C in LB medium supplemented with 50 µg/mL until the OD600 of 0.6. Protein expression was induced for 18 h at 20 • C by adding 0.3 mM isopropyl-β-thiogalactoside. Cells were collected by centrifugation at 8000× g for 20 min at 4 • C. Proteins were purified by Ni 2+ -NTA affinity, Heparin column, and Superdex-75 gel filtration chromatography. Purified human polη was concentrated to 15 mg/mL in a gel filtration buffer (25 mM Tris, pH 7.5, 300 mM KCl, 10% glycerol, and 2 mM dithiothreitol), aliquoted, and flash-frozen in liquid nitrogen to store at −80 • C.

Steady-State Kinetics of Single Nucleotide Incorporation Opposite Templating Fm7dG by Polη
Steady-state kinetic parameters for insertion opposite Fm7dG by polη were measured as described previously, with slight modifications [42]. The oligonucleotides for kinetic assays (primer, 5'-FAM/GGGGGCTCGTAAGGATTC-3 and template, 5´-CCGACT(Fm7dG)GAATCCTTACGAGC CCCC-3´) were synthesized by Midland Certified Reagent company (Midland, TX, USA) and Integrated DNA Technologies (Coralville, IA, USA). To prepare the duplex DNA substrate for polη, both oligonucleotides were annealed in hybridization buffer (10 mM Tris-HCl, pH 7.5; 1 mM EDTA) at 90 • C for 5 min. Enzyme activities were determined using the reaction mixture containing 40 mM Tris-HCl (pH 8.0), 60 mM KCl, 10 mM dithiothreitol (DTT), 250 µg/mL bovine serum albumin (BSA), 2.5% glycerol, 5 mM MgCl 2 , 50-100 nM recessed DNA (both control dG and Fm7dG), and varying concentrations of incoming dNTP. To prevent end-product inhibition and substrate depletion from interfering with accurate velocity measurement, the enzyme concentrations and reaction times were adjusted for each experiment (less than 20% insertion product formed). The reactions were initiated by the addition of the enzyme and stopped with a gel-loading buffer (95% formamide with 20 mM EDTA, 45 mM Tris-borate, 0.1% bromophenol blue, and 0.1% xylene cyanol). The quenched samples were separated on 20% denaturing polyacrylamide gels. The gels were visualized and analyzed using Typhoon Imager (GE Healthcare Life Sciences) to quantify product formation. The k cat and K m were determined by fitting the reaction rate over dNTP concentrations to the Michaelis-Menten equation. Each experiment was repeated three times to measure the average and the standard deviation of the kinetic results. The efficiency of nucleotide insertion was calculated as k cat /K m . The relative frequency of dNTP incorporation opposite Fm7dG was determined as f = (k cat /K m ) [dN:Fm7dG] /(k cat /K m ) [dC:dG] .

Crystallization, Data Collection, and Refinement
The template oligonucleotides for X-ray crystallographic studies (5 -CAT(X)ATGACGCT-3 , X = dG or Fm7dG) were synthesized by Midland Certified Reagent Co. (Midland, TX, USA). The primer, 5 -AGCGTCAT-3 , was synthesized by Integrated DNA Technologies (Coralville, IA, USA). The oligonucleotides were mixed at a 1:1 molar ratio and annealed in HEN buffer (10 mM HEPES, pH 8.0; 0.1 mM EDTA; and 50 mM NaCl) by heating for 10 min at 90 • C and slowly cooling to room temperature. The cocrystals of the ternary complex of polη-Fm7dG:dCTP* were grown at conditions similar to those described previously [46]. Briefly, the crystal was obtained by using the hanging drop vapor diffusion method in a reservoir buffer containing 0.1 M MES (pH 5.5), 5 mM MgCl 2 , and 15%-20% PEG 2K-MME. Diffraction data were collected on 3-5-day-old crystals that were cryoprotected with 20% glycerol at 100 K using the beamline 5.0.3 at the Advanced Light Source, Lawrence Berkeley National Laboratory. All diffraction data were processed using HKL 2000 (HKL research, Charlottesville, VA, USA) [47]. Structures were solved by molecular replacement using the polη-dG:dCTP* structure (PDB code 4O3N) as the search model [41]. The model was built using COOT [48] and refined using Phenix software [49]. MolProbity was used to make Ramachandran plots [50].