Impact of Conformational Heterogeneity of OxoG Lesions and Their Pairing Partners on Bypass Fidelity by Y Family Polymerases

Summary

7,8-Dihydro-8-oxoguanine (oxoG), the predominant oxidative DNA damage lesion, is processed differently by high-fidelity and Y-family lesion bypass polymerases. Although high-fidelity polymerases extend predominantly from an A base opposite an oxoG, the Y-family polymerases Dpo4 and human Pol η preferentially extend from the oxoG•C base pair. We have determined crystal structures of extension Dpo4 ternary complexes with oxoG opposite C, A, G, or T and the next nascent base pair. We demonstrate that neither template backbone nor the architecture of the active site is perturbed by the oxoG(anti)•C and oxoG•A pairs. However, the latter manifest conformational heterogeneity, adopting both oxoG(syn)•A(anti) and oxoG(anti)•A(syn) alignment. Hence, the observed reduced primer extension from the dynamically flexible 3′-terminal primer base A is explained. Because of homology between Dpo4 and Pol η, such a dynamic screening mechanism might be utilized by Dpo4 and Pol η to regulate error-free versus error-prone bypass of oxoG and other lesions.

Introduction

7,8-Dihydro-8-oxoguanine (oxoG) is the major known product of oxidation of DNA by reactive oxygen species. It is induced in an aerobic environment by ionizing radiation, photochemical mechanisms, or normal cellular metabolic activity (Cadet et al., 2003). An increased risk of developing cancer has been linked to oxidative stress due to the overproduction of reactive oxygen species resulting from the response of cells to inflammation and infection (Grisham et al., 2000;Meira et al., 2008). Elevated levels of oxoG in genomic DNA have also been associated with neuro-degenerative diseases, aging, and cardiovascular disorders (Cooke et al., 2003; Lu et al., 2004).

The oxoG lesion is a potent mutagen inducing primarily G to T transversions in human cells (Tan et al., 1999; Tolentino et al., 2008). High-fidelity polymerases that replicate genomic DNA in vitro readily insert C or A opposite the oxoG lesion in varying proportions that depend on the polymerase, with extension occurring preferentially from oxoG•A noncanonical pairs (Einolf and Guengerich, 2001; Furge and Guengerich, 1997; Lowe and Guengerich, 1996; Shibutani et al., 1991). In contrast, the Y-family yeast and human Pol η (Carlson and Washington, 2005; Haracska et al., 2000) and Dpo4 (Eoff et al., 2007b; Rechkoblit et al., 2006; Zang et al., 2006) lesion bypass polymerases preferentially insert C opposite oxoG, and also favor extension from the oxoG•C base pair, thus achieving near error-free bypass of this lesion. The Y-family Pol η is proposed to be involved in oxoG lesion bypass in vivo based on increase of spontaneous mutations in the absence of Pol η in the glycosy-lase Ogg1Δ mutant yeast strain (Haracska et al., 2000), and is demonstrated by an siRNA knockdown approach to promote error-free bypass of oxoG lesions in mammalian cells (Lee and Pfeifer, 2008).

Dpo4 is a member of the DinB family of bypass DNA polymerases found in all three kingdoms of life. It has high homology to humanPol η and similar lesion by pass properties, there by making it a useful model system for studying Y-family polymerases (Boudsocq et al., 2001). Complexes of Dpo4 polymerase have been extensively studied by X-ray crystallography with unmodified DNA (Ling et al., 2001) and several types of lesions (Bauer et al., 2007; Eoff et al., 2007a; Irimia et al., 2007; Ling et al., 2003, 2004a, 2004b; Rechkoblit et al., 2006; Zang et al., 2005, 2006), or with mismatches (Trincao et al., 2004; Vaisman et al., 2005). The structural basis for the unique ability of the Y-family Dpo4 polymerase to preferentially extend the correct oxoG•C base pair remains unexplained and is a focus of this study.

High-fidelity polymerases avoid mismatches by producing tight-fitting, solvent-excluding reaction-ready active sites, resulting from closing the “O” helix of the finger domain on the flat surface of a complementary, Watson-Crick nascent base pair. Furthermore, multiple residues are employed to proofread the minor groove of five base pairs of growing DNA (Johnson and Beese, 2004; Steitz and Yin, 2004) and terminal mismatches are displaced to an exonuclease site. Y-family polymerases have more spacious and solvent-accessible active sites and there are no amino acid contacts with the minor groove edge of template/primer DNA, as observed in ternary complexes for archaeal Dpo4 (Ling et al., 2001), yeast Pol η (Alt et al., 2007), human Pol ι (Nair et al., 2004), and Pol κ (Lone et al., 2007). Importantly, Dpo4, unlike high-fidelity polymerases, does not contain an O helix to close upon dNTP binding and check the correctness of a nascent base pair. Instead, Dpo4 relies largely on Watson-Crick base pairing for fidelity check (Mizukami et al., 2006), and on stepwise translocation throughout the catalytic cycle, associated with a combination of rotational and translational motions (Rechkoblit et al., 2006). These structural and functional features enable Y-family polymerases to bypass a variety of DNA lesions that impede high-fidelity DNA polymerases, and concurrently cause a higher error rate and lower processivity on undamaged DNA templates (reviewed in Broyde et al., 2008; Yang and Woodgate, 2007). Recent studies suggest that access of translesion Y-family DNA polymerases to replication forks is tightly regulated, recruiting them only temporarily to overcome blocks to replicative polymerases (reviewed in Moldovan et al., 2007).

The available crystal structures for ternary complexes (template/primer-DNA polymerase-dNTP complex) of high-fidelity polymerases with template oxoG at the insertion site (Figure 1A) forming a nascent base pair with dCTP, reveal that template distortion associated with an oxoG in the anti conformation does not disrupt the tight reaction-ready active sites of repair gap-filling Pol β (Krahn et al., 2003) and high-fidelity Rb69 (Freisinger et al., 2004) and T7 (Brieba et al., 2004) polymerases. Previously, our group (Rechkoblit et al., 2006) has demonstrated that in the active site of the Dpo4 insertion ternary complex OxoG(anti) forms a base pair with dCTP, and the phosphate group of the oxoG residue is flipped about 180°, resulting in a 3.5 Å shift compared to the structure with unmodified G. The anti conformation of oxoG is stabilized by multiple and favorable contacts of the relocated phosphate group with Dpo4 amino acid residues and interaction of the O8 atom with Arg332 of the Dpo4 little-finger domain. These favorable interactions of oxoG(anti) with Dpo4 likewise persist in the postinsertion binary complex (template/primer-DNA polymerase complex) after covalent incorporation of C and translocation of the Dpo4 thumb domain.

Figure 1. Structure of the OxoG•C Dpo4 Extension Ternary Complex.

(A) Schematic of the pairing of the oxoG-containing 18-mer and 19-mer template strands with the 13-mer primer strand ending in a2′,3′-dideoxynucleotide in the extension ternary complex with Dpo4. The insertion position at the Dpo4 active site is denoted by (0), and the postinsertion position is denoted by (−1).

(B) Overall structure of the oxoG•C complex. Oxo-G(anti) at the (−1) position forms a base pair with the 3′-terminal C14 base of the primer strand. The next template base C5 is paired with an incoming dGTP at the active site. The first Ca²⁺, cation A, is coordinated by invariant D7, D105, and E106 residues, the second Ca²⁺, cation B, is chelated by the phosphate groups of the incoming dGTP.

(C) Structure of the active site of the oxoG•C complex. Arg332 of the Dpo4 little-finger domain forms hydrogen bonds with the O8 and the phosphate group of the OxoG(anti).

(D) OxoG(anti) paired with C(anti) of the primer strand and hydrogen-bonded with Arg332. Simulated annealing Fo-Fc omit electron density map contoured at the 3σ level is colored in blue (2.90 Å resolution).

In the crystal structures of high-fidelity polymerase complexes that represent extension past an oxoG lesion with the templating oxoG base at the postinsertion position of the active sites (Figure 1A), a template distortion associated with oxoG(anti) opposite covalently incorporated C was observed in T7 (Brieba et al., 2004) and in the Bacillus Pol I fragment BF (Hsu et al., 2004). In the case of BF, the template distortion affected the polymerase itself by disrupting the interactions with a minor groove proofreading residue and causing a distorted conformation of the O helix. However, neither the template nor the polymerase were distorted by oxoG in a syn conformation opposite A, thereby enabling the oxoG(syn)•A(anti) noncanonical pair to evade proofreading in both T7 (Brieba et al., 2004) and BF (Hsu et al., 2004) complexes. Thus, the disturbances caused by an oxoG(anti), but not by an oxoG(syn) conformation, provide a structural explanation for the predominant extension of the mutation-generating oxoG(syn)•A(anti) noncanonical pair by high-fidelity DNA polymerases.

The goal of our work was to understand the factors that allow the low-fidelity Dpo4 polymerase to efficiently and preferentially elongate the oxoG•C over the oxoG•A noncanonical pair, in contrast to high-fidelity polymerases that manifest the opposite behavior. For this purpose, we have determined crystal structures of extension Dpo4 ternary complexes with oxoG opposite C, A, G, or T and the next correct nascent base pair. Our results suggest that Dpo4 discriminates against extending from the oxoG•A base pairs by a dynamic screening process involving a novel synanti equilibrium of the oxoG residue and its partner base A. In the case of the correct oxoG•C base pair, the Dpo4 active site remains unperturbed, which allows for efficient extension. The inefficient extension from oxoG(anti)•G(syn), and wobble oxoG(anti)•T(anti) noncanonical pairs that are observed for the first time, is primarily caused by a misalignment of the catalytic metal ion. Our findings are compared with published crystal structures of oxoG•C and oxoG•A noncanonical pairs at the postinsertion position of polymerase active sites reported forT7 (Brieba et al., 2004) and BF (Hsu et al., 2004) high-fidelity DNA polymerases, and for Glu332 and Ala332 mutants of Dpo4 bypass polymerase (Eoff et al., 2007b).

Results

DNA Template/Primer Design and Crystal Structure Determination

In order to crystallize the Dpo4 extension ternary complexes with correct C and noncanonical A, G, or T at the 3′-end of the primer strand opposite oxoG at the (−1) position of the template strand (Figure 1A), we used an oxoG-modified 18-mer or 19-mer (containing a single unpaired residue at the 3′-end) template and 2′,3′-dideoxy terminated 13-mer primer strands. This resulted in a template/primer duplex containing a junctional nascent base pair between a C (5′ to the oxoG lesion on the template strand) and an incoming dGTP, positioned adjacent to a four base ingle-stranded 5′-template overhang (Figure 1A). In the case of the A base opposite oxoG, the use of the 18-mer template resulted in crystals in the P2₁ space group, with two molecules per asymmetric unit (AU) (designated oxoG•A-2 complexes). In the case of the 19-mer template (Figure 1A), the crystals grew in the P2₁2₁2 space group with one molecule per AU (designated oxoG•A-1 complex). Crystals with C, T, or G opposite oxoG were grown with either 18- or 19-mer templates; although the template length did not affect the space group or unit cell parameters, the crystals with 18-mer template had better diffraction and were used for data collection. The structures of the Dpo4 oxoG-modified extension ternary complexes were solved by the molecular replacement method employing the published Dpo4 oxoG-modified insertion ternary complex with incoming dCTP (Rechkoblit et al., 2006) as a search model, and refined. Necessary replacements of DNA bases and some conformational changes were then introduced with the help of initial electron density maps, and the models were refined. The crystal data, together with the data collection and refinement statistics for all structures, are summarized in Table 1.

Table 1. Data Collection and Refinement Statistics.

Extension Ternary Complex Type	OxoG•C	OxoG•A-2	OxoG•A-1	OxoG•G	OxoG•T
Data collection
Space group	P21212	P21	P21212	P21212	P21
Cell dimensions
a (Å)	99.86	51.26	95.76	99.67	53.29
b (Å)	112.40	107.41	111.65	111.92	110.68
c (Å)	52.83	97.87	52.71	53.16	102.19
β (°)	90.00	100.54	90.00	90.00	99.22
Complexes per AUa	1	2	1	1	2
X-ray source	APS; 24-ID	APS; 24-ID	APS; 24-ID	APS; 24-ID	APS; 24-ID
Wavelength	0.97949	0.97949	0.97942	0.97911	0.97922
Resolution range (Å)b	20-2.87 (2.96-2.87)	20-2.35 (2.42-2.35)	20-2.60 (2.67-2.60)	20-2.70 (2.80-2.70)	20-2.70 (2.78-2.70)
Rmerge(%)c	8.9 (65.6)	6.4 (42.4)	9.6 (52.8)	9.3 (49.3)	11.3 (46.5)
I / σI	15.9 (1.8)	20.1 (1.8)	16.2 (2.6)	18.6 (3.0)	12.6 (2.5)
Completeness (%)	94.9 (84.1)	92.7 (62.4)	94.2 (71.7)	93.8 (72.7)	99.2 (96.3)
Redundancy	4.7 (4.2)	3.8 (2.4)	6.2 (4.2)	6.1 (5.1)	4.1 (3.2)
Refinement
Resolution range (Å)	20-2.90	20-2.40	20-2.60	20-2.70	20-2.70
Number of reflections	12,368	36,655	16,068	15,082	29,786
Rfactor/Rfreed	23.2/ 27.7	20.9/ 25.3	20.2/ 26.5	21.4/ 27.7	20.0/ 26.0
Model composition (AU)
Protein	341	682	341	341	682
Templatee	18	32	19	18	34
Primer	13	26	13	13	26
Ligand (dGTP)	1	2	1	1	2
Ion (Ca2+)	3	6	3 (+1 Na+)	3	6
Water	17	239	65	40	118
B-factors
Protein	42.1	61.1	48.5	55.5	51.1
Template	43.6	64.1	52.3	56.7	54.1
Primer	47.2	63,2	51.5	65.0	55.9
Ligand	35.0	46.4	35.8	42.8	33.3
Ion	47.6	59.3	48.9	53.1	48.2
Water	15.3	57.6	48.0	56.7	40.5
Mean B-factor	42.4	61.1	41.4	56.2	43.5
Rmsd bond length (Å)	0.009	0.008	0.010	0.008	0.010
Rmsd bond angles (°)	1.61	1.60	1.65	1.52	1.62
PDB ID	3GIK	3GIJ	3GII	3GIM	3GIL

^aAU, asymmetric unit; APS, Advanced Photon Source, Argonne National Laboratory.

^bValues in parentheses correspond to the last resolution shell.

^cR_merge = Σ_hΣ_i|l_hi - < l_h> |/Σ < l_h >, where I_hi is the intensity of the i^th observation of reflection h, and < I_h > is the average intensity of redundant measurements of the h reflections.

^dR_factor = Σ‖ F_o| - |F_c‖/Σ‖F_o|, where F_o and F_c are the observed and calculated structure-factor amplitudes; R_free is monitored with the 5% reflections excluded from refinement.

^eResidues 1 to 4 of single-stranded 18-mer template overhang (chain J) in one molecule of the oxoG•A-2 complex and residues 1 and 2 (chain J) of the oxoG•T complex are disordered in the electron density maps.

Structure of the OxoG•C Extension Ternary Complex

The overall structure of the oxoG•C extension ternary complex (Figure 1B) is similar with a root-mean-square deviation (rmsd) of 0.46 Å to the insertion ternary complex with dCTP opposite oxoG (Rechkoblit et al., 2006) and the type I unmodified complex (Ling et al., 2001). Briefly, the Dpo4 polymerase embraces the 18-mer template/13-mer primer DNA by its four domains: palm (residues 1–10 and 78–166), finger (residues 11–77), thumb (residues 167–233), and little finger (residues 244–341). The thumb is joined to the little-finger domain by a 10 amino acid long tether (residues 234–243) that allows positioning of the little finger on the other side of the DNA duplex. There are three divalent cations (Figure 1B), identified previously as Ca²⁺ by anomalous scattering under identical crystallization conditions (Rechkoblit et al., 2006). The first Ca²⁺, cation A within the polymerase active site, is coordinated by invariant D7, D105, and E106 residues; the second Ca²⁺, cation B, is chelated by the phosphate groups of the incoming dGTP; the third cation is coordinated by the loop of the thumb domain (residues 181 and 186), adjacent to the tip of helix H (Figures 1B and 1C).

The residues in the Dpo4 polymerase active site pocket are outlined in Figure 1C. The “roof” of the active site, formed by the finger domain (in blue), is positioned directly over the C5 template base (5′ to adjacent oxoG) that forms a base pair with incoming dGTP. All protein and DNA residues including the 5′-C-T-A-A single-stranded template overhang are well ordered and defined in the 2F_o-F_c electron density map. The 2F_o-F_c map for the template/primer DNA, dGTP and Ca²⁺ ions spanning the Dpo4 active site is shown in Figure S1A (available online) and for the oxoG•C pair in Figure S1B. An oxoG in the anti conformation forms a Watson-Crick base pair with 2′,3′-dideoxycytosine at the 3′-end of the primer strand with a C1′-C1′ distance of 10.5 Å, a value typical for B-DNA. This pairing alignment is clearly evident from the simulated annealing F_o-F_c omit map contoured at the 3σ level shown in Figure 1D. The guanidinium group of Arg332 from the Dpo4 little-finger domain forms hydrogen bonds with the O8 atom and the phosphate group of an oxoG(anti) lesion (Figure 1D).

Comparison with G•C Extension Ternary Complex

The details of accommodation of the oxoG(anti)•C and G(anti)•C (from Protein Data Bank [PDB] ID 2AGQ, Vaisman et al., 2005) base pairs at the postinsertion (−1) position of Dpo4 extension ternary complexes are depicted in Figures 2A and 2B. The complexes are superimposed by Cα atoms of the little-finger domain, that are unique for Y-family polymerases. This domain creates an extensive interface for binding the template/primer DNA, and in the Dpo4 extension ternary complexes, makes electrostatic interactions with the phosphate groups of residues C5 to A9 of the template strand and residues A7 to G5 of the primer strand. Figure 2A depicts the interactions of the little-finger domain with template residues at the (−1) and (−2) positions. The side chains of the little-finger domain residues take similar positions, with Thr250 and Arg332 contacting the phosphate groups of oxoG or G, and Arg247 forming a hydrogen bond with the phosphate of the residue at the (−2) position. Surprisingly, at the (−1) position of the Dpo4 active site, the phosphate group of oxoG is conformationally similar to the phosphate of unmodified G (Figures 2A and 2B). The intrinsic steric clash of the O8 atom with the sugar-phosphate backbone that arises in oxoG(anti) is relieved by a change in the orientation of the oxoG base, via rotation around the glycosidic torsion angle χ to −67° from −111° for unmodified G and a slight rotation of the sugar ring due to change in the ζ, backbone torsion angle (Figures 2A and 2B; see Table S1 for backbone torsion angles for unmodified and oxoG template bases at the (0) and (−1) positions). The oxoG(anti)•C base pair is asymmetrically buckled by ∼12° with the oxoG base O8 atom inclined toward the 3′ side of the template strand and the C of the primer strand taking the normal orientation for the 3′-end base.

Figure 2. Comparison of OxoG(anti) with Unmodified-G(anti) Alignments Opposite C at the (−1) Position in the Extension Ternary Complex.

(A) Similar interactions of the little-finger domain with oxoG residue and unmodified-G at the (−1) position of the Dpo4 active site of the extension ternary complexes. Side chains of the little-finger domain of the oxoG-containing extension complex are shown in pink, C5-oxoG6-C7 residues are in orange, and Dpo4 side chains of unmodified complex and T-G-A residues are in white (PDB ID 2AGQ, (Vaisman et al., 2005); the complexes are superimposed via the Cα atoms of the little-finger domain of Dpo4.

(B) Similar phosphate backbone conformation of oxoG (orange) and unmodified G (white) (PDB ID 2AGQ) at the (−1) position of extension ternary complexes.

Thus, oxoG(anti) is easily accommodated at the (−1) position of the active site of the Dpo4 bypass polymerase, with minimal adjustment in orientation of the oxoG base and no disruptions or changes within the polymerase active site or position of the oxoG phosphate group.

OxoG(syn)•A(anti) Alignment in Molecule #1 of the OxoG•A-2 Extension Ternary Complex

The oxoG•A extension complex with the 18-mer oxoG-modified template crystallized in space group P2₁ (designated oxoG•A-2) with two molecules (labeled #1 and #2) in the AU unit. The overall arrangement of the active site of olecule #1 of the oxoG•A-2 complex (see 2F_o-F_c electron density map, Figure S2A) is similar to that observed for the oxoG•C complex (Figure 1C). The simulated annealing F_o-F_c omit map contoured at the 3σ level (Figure 3A) unambiguously indicates the syn conformation of the oxoG in a Hoogsteen alignment with A in the anti conformation. In this noncanonical pair, O6 (acceptor) and N7 (donor) of oxoG(syn) form two hydrogen bonds with the A(anti) N6 and N1 groups, respectively; the plane of the oxoG base is parallel to the adjacent C5 and C7 bases, the C1′ to C1′ distance is 10.4 Å, only 0.1 Å shorter than an ideal Watson-Crick base pair, and the phosphate group takes the normal position for the (−1) base in the Dpo4 ternary complex. The guanidinium group of Arg332 is relocated away from the base and phosphate group of the oxoG(syn) complex to avoid collision with N2 of the oxoG(syn) (Figure 3A), in contrast to being in hydrogen bonding distance to the O8 atom and phosphate group of oxoG in the oxoG(anti)•C base pair (Figure 1D).

Figure 3. (syn)-(anti) and (anti)-(syn) Arrangements of OxoG•A Non-canonical Pair and Different Orientations of Arg332 in Molecules #1 and #2 of the OxoG•A-2 and in the OxoG•A-1 Extension Ternary Complexes.

The oxoG•A-2 complex has two distinct molecules per asymmetric unit (AU), and the oxoG•A-1 has one molecule per AU.

(A) OxoG(syn)•A(anti) noncanonical pair in molecule #1 of the oxoG•A-2 complex. The Arg332 relocates to avoid collision of its guanidinium group with N2 of the oxoG(syn). Simulated annealing Fo-Fc omit electron density map contoured at the 3σ level is colored in blue (2.40 Å resolution).

(B) Predominant oxoG(anti)•A(syn) (80% occupancy) and minor oxoG(syn)•A(anti) (20% occupancy) noncanonical paring arrangements in molecule #2 of the oxoG•A-2 complex. Arg332 forms hydrogen bonds with the O8 and the phosphate group of the oxoG(anti). The hydrogen bond and C1′-C1′ distances are shown for the oxoG(anti)•A(syn) pair. Simulated annealing Fo-Fc omit electron density map contoured at the 3σ level is colored in blue (2.40 Å resolution).

(C) The O8 and the phosphate group of the oxoG(anti) of the oxoG•A-1 complex forms hydrogen bonds with the guanidinium group of Arg332. Simulated annealing Fo-Fc omit electron density map contoured at the 3σ level is colored in blue (2.60 Å resolution); the electron density for A opposite the oxoG is disordered.

Predominant OxoG(anti)•A(syn) Alignment in Molecule #2 of the OxoG•A-2 Extension Ternary Complex

The overall arrangement of the active site in molecule #2 of the oxoG•A-2 complex is similar to the oxoG•C complex (see 2F_o-F_c electron density map, Figure S2B). The simulated annealing F_o-F_c omit map for xoG, A14, Arg 332, and coordinated water molecule contoured at the 3σ level is shown in Figure 3B. Although the F_o-F_c electron density maps for molecules #1 and #2 are of equally high quality, the density between the oxoG and its base-pairing partner A in molecule #2 is continuous and is not sharply interrupted between the bases, as observed in molecule #1 (Figure 3A) or in the oxoG•C complex (Figure 1D). Two types of base-pair conformations are consistent with this map: oxoG(anti)•A(syn) with a C1′-C1′ distance of 10.5 A and oxoG(syn) •A(anti) with a C1′-C1′ distance of 10.3 Å (Figure 3B). The continuous electron density for the base-pair region is suggestive of an equilibrium between these two pairing modes. The possibility of an OxoG(anti)•A(anti) arrangement is excluded because such alignment produces a long, 12.6 Å, C1′-C1′ distance (Prive et al., 1987); when such a G(anti)•A(anti) noncanonical base pair is placed in the experimental electron density map a very short 1.9 Å distance between N1 of oxoG and N1 of A results after refinement of the model.

The oxoG in the oxoG(anti)•A(syn) noncanonical pair (Figure 3B) is accommodated in the same way as the oxoG (anti)•C(anti) base pair (Figure 1D) in the oxoG•C complex, with a glycosidic torsion angle χ = −68° (Table S1). Arg332 is well defined in the electron density map and forms hydrogen bonds with the O8 atom and the phosphate group of the OxoG(anti). The geometry of this base pair suggests a hydrogen bond between the O6 group of the oxoG and N6 of A. There is a water molecule that is well defined in the simulated annealing F_o-F_c omit map (Figure 3B) that bridges these two groups, and a bifurcated hydrogen bond between the donor N1 and N2 groups of oxoG and N7 (acceptor) of the A base is also noted. A water molecule about 2.8 Å away from the O6 group of G(anti) or oxoG (anti) is often found in high-resolution structures of free DNA duplexes or DNA-protein complexes, but does not participate in forming a base pair with cytosine (Auffinger and Westhof, 2000) (PDB ID's of oxoG-containing structures: 1U48, 1TKD).

The oxoG(syn)•A(anti) base pair in molecule #2 (shown in white in Figure 3B) is analogous to the one described previously in molecule #1 (Figure 3A), and can also be accommodated into this simulated annealing F_o-F_c omit electron density map. The water molecule that now can bridge the O6 of oxoG(syn) and the N6 of A(anti) is absent in molecule #1. Only one position of Arg332 is observed in molecule #2, with the NH₂ nitrogen of the guanidinium group just 3.1 Å away from the N2 nitrogen of oxoG (syn). The positions of the sugar and phosphate groups of the oxoG are identical in syn and anti conformations; in the A(anti) and A(syn) residues the phosphate groups take slightly different positions (Figure 3B, Table S1). Based on the observed position of Arg332 in molecule #2, which results in a close distance of the guanidinium group with the N2 of oxoG(syn), as well as the absence of bridging water in molecule #1, we conclude that the oxoG(anti) •A(syn) base pairisthe predominant alignment in molecule #2. The refinement of the oxoG•A-2 complex with 80% occupancy for oxoG(anti)•A(syn) and 20% occupancy for oxoG(syn) d A(anti) avoids any difference density on the F_o-F_c electron density map and at the same time fulfills the density on the simulated annealing F_o-F_c omit map and the regular 2F_o-F_c electron density map.

To our knowledge, only the oxoG(syn) •A(anti) alignment has been observed in free DNA duplexes in solution (Kouchakdjian et al., 1991) and in a crystal (McAuley-Hecht et al., 1994), as well as within the active sites of T7 (Brieba et al., 2004) and BF (Hsu et al., 2004) high-fidelity polymerases, and in Arg332 mutants of Dpo4 bypass polymerase (Eoff et al., 2007b).

OxoG(anti) Is Opposite Disordered A in the OxoG•A-1 Extension Ternary Complex

The oxoG•A extension complex with the 19-mer oxoG-modified template crystallized in space group P2₁2₁21 (designated oxoG•A-1) with one molecule in the AU unit. In the active site of the oxoG•A-1 complex, an oxoG lesion is found in the anti conformation with Arg332 contacting its O8 atom and phosphate group, as observed in the oxoG•C (Figure 1D) and molecule #2 of the oxoG•A-2 (Figure 3B) complexes. Despite the clear overall 2Fo-Fc map for the segment spanning the active site in this complex (Figure S2C), the map region for residue A14 is pretty poor and suggests its partial disorder (see the simulated annealing Fo-Fc omit map in Figure 3C). Hence, in this complex, A14 most likely adopts a range of conformations in different molecules within the crystal, which results in poor electron density at this site. The electron density map for the lesion site clearly indicates only the anti conformation of oxoG (Figure 3C). Thus, the oxoG•A-1 complex structure provides additional evidence for the ability of Arg332 to shift the oxoG equilibrium toward the anti conformation with adenine as a base-pairing partner.

OxoG(anti)•G(syn) Alignment in the OxoG•G Extension Ternary Complex

The 2F_o-F_c electron density map of the active site of the oxoG•G complex with 18-mer oxoG-containing template strand is well-ordered throughout the structure, including the lesion site and at G14 opposite oxoG (Figure S3A). The simulated annealing F_o-F_c omit map shown at the 3σ level (Figure 4A) nambiguously shows the anti conformation of oxoG, with Arg332 within hydrogen bonding distance to O8 and the phosphate group, as was observed in the oxoG•C complex (Figure 1D), and G14 in the syn conformation. The donor N1 and N2 atoms of the oxoG(anti) Watson-Crick edge form two hydrogen bonds with the Hoogsteen edge O6 and N7 of G(syn), respectively. The C1′ to C1′ distance in the oxoG(anti)•G(syn) noncanonical pair is 11.0 Å, 0.5 Å longer than an ideal Watson-Crick base pair.

Figure 4. OxoG(anti) Paired with G(syn) and T(anti) at the (−1) Position of the OxoG•G and the OxoG•T Extension Ternary Complexes, Respectively.

Simulated annealing Fo-Fc omit electron density maps contoured at the 3σ level are colored in blue (2.70 Å resolution).

(A) OxoG(anti)•G(syn) noncanonical pair of the oxoG•G complex. Arg332 forms hydrogen bonds with the O8 and the phosphate group of the OxoG(anti).

(B) OxoG(anti)•T(anti) noncanonical pair of the oxoG•T complex. Arg332 forms hydrogen bonds with the O8 and the phosphate group of the oxoG(anti).

The G•G mismatch in free DNA duplexes has been shown to adopt the G(anti)•G(syn) alignment both in solution (Cognet et al., 1991) and in a crystal (Skelly et al., 1993); the alternate G(anti)•G(anti) alignment is found less frequently in solution (Borden et al., 1992). Interestingly, oxoG opposite G in a DNA duplex in solution adopts the oxoG(syn)•G(anti) alignment (Thiviyanathan et al., 2003), in contrast to the oxoG(anti)•G(syn) alignment found in our Dpo4 extension complex.

OxoG(anti)•T(anti) Wobble Pair in the OxoG•T Extension Ternary Complex

The structures of the two molecules in the asymmetric unit of the oxoG•T extension complex with 18-mer oxoG-containing template strand are similar to each other, including the alignment of the oxoG•T noncanonical pair. As clearly evident from the well-ordered simulated annealing F_o-F_c omit map shown at the 3σ level in Figure 4B, the oxoG lesion takes the anti conformation with Arg332 forming hydrogen bonds withO8 and the phosphate group of oxoG. The T14(anti) is shifted into the major groove, where it adopts a wobble alignment in order to form hydrogen bonds betweenN3 and O2 withO6andN1 of oxoG(anti), respectively; the O4 atom of T14 is coordinated by a water molecule that is well-defined in the simulated annealing F_o-F_c omit map (Figure 4B). The C1′ to C1′ distance in the oxoG(anti)•T(anti) pair is 10.3 Å, very close to that of a Watson-Crick base pair; however, due to the wobble alignment, the sugar ring of T14 is shifted away from the Dpo4 catalytic triad residues and the active site Ca²⁺ ion by 2.1 Å.

The oxoG•T noncanonical pair has not been structurally characterized previously; the wobble G(anti)•T(anti) alignment is commonly found for G•T noncanonical pairs within free DNA duplexes in solution (Patel et al., 1982) and in the crystalline state (Hunter et al., 1987). In the BF high-fidelity polymerase postinsertion site, both G•T and T•G noncanonical pairs adopt the wobble alignment, which results in a displacement of the template G or T base, while the primer T or G bases remain in their normal position(Johnson and Beese, 2004). In the Dpo4 insertion ternary complex, the template T base formed a wobble base pair with incoming dGTP (Vaisman et al., 2005). However, at the postinsertion site of Dpo4 (−1 position), the template T and primer G non-canonical pair were observed in the reverse wobble alignment with the primer G base shifted into the major groove, thus deflecting the 3′-OH group away from the incoming dNTP, while the template T is accommodated normally (Trincao et al., 2004).

Discussion

Structural Basis for Efficient Primer Extension from the OxoG•C Base Pair by Dpo4

As we demonstrated in the Results section, the accommodation of the oxoG(anti)•C base pair at the postinsertion (−1) position of the Dpo4 ternary complex (Figure 1C) does not entail any changes within the Dpo4 active site. Analysis of the structures of ternary Dpo4 complexes with normal DNA bases suggests that there would be sufficient space to accommodate the O8 atom of an oxoG residue without changing the conformation of the sugar-phosphate backbone for an unmodified base at the (−1) position. The Arg332 residue of the Dpo4 little-finger domain further stabilizes the anti conformation of oxoG by forming direct hydrogen bonds with its O8 atom and phosphate group. The kinetic observations that primer extension catalyzed by Dpo4 occurs readily beyond the oxoG•C base pair (Eoff et al., 2007b; Rechkoblit et al., 2006) is consistent with these structural properties. Moreover, the ∼7.7-fold more efficient extension from an oxoG•C than from a G•C pair (Rechkoblit et al., 2006) is explained by an additional hydrogen bond between Arg332 and the O8 atom of oxoG, which is absent in the case of unmodified G.

Interestingly, pre-steady-state kinetic data indicated that wild-type Dpo4 is more efficient in extending primer strands from oxoG•C than from G•C pairs, compared with the opposite behavior of Arg332Glu and Arg332Ala mutants (Eoff et al., 2007b). Glu332 interacts through a water-mediated hydrogen bond with the O8 atom of oxoG, but fails to contact the phosphate group of oxoG (Eoff et al., 2007b). Furthermore, in the case of the Arg332Ala mutant, the Ala332 fails to contact both the O8 and the phosphate group of oxoG (Eoff et al., 2007b). The position of the phosphate group of oxoG was not affected by either of these Arg332 mutations. Thus, the contacts of Arg332 with the O8 atom and the phosphate group of the oxoG play a significant role in governing efficiency of primer extension from the oxoG•C pair.

Structural Basis for Impeded Primer Extension from the OxoG•C Pair by High-Fidelity Polymerases

In contrast, the anti conformation of oxoG at the postinsertion position of the active sites of high-fidelity T7 and BF polymerases induces local changes in the conformation of the template strand that reduces the ability to extend from the oxoG•C base pair. The replicative T7 polymerase is ∼300-fold less efficient in extending from the oxoG•C than from an unmodified G•C base pair (Furge and Guengerich, 1997). In the extension ternary complex of the T7 polymerase with the oxoG(anti)•C base pair at the (−1) post-insertion position, the phosphate of oxoG is rotated by ∼90° around the bond connecting C4′ and C5′ of the DNA backbone relative to the backbone conformation of an unmodified purine in order to accommodate the O8 atom (Brieba et al., 2004). As a consequence, the orientation of the side chain of His607 had to adjust to form a hydrogen bond with the relocated phosphate group of oxoG. Similarly, the BF polymerase required higher dNTP concentrations to extend from the oxoG•C base pair than from the G•C base pair. In the postinsertion binary complex of the BF polymerase, oxoG(anti) opposite C was accommodated by altering the orientation of the base and the position of the sugar of the oxoG residue, thereby disrupting the interaction with the minor groove proofreading residue Gln797 (Hsu et al., 2004). Such oxoG(anti)-induced template distortions resulted in altered conformations of the O and “O1” helices, preventing the next template base from entering the preinsertion site.

Conformational Flexibility of the OxoG•A Pair at the Dpo4 Active Site Inhibits Primer Extension

We have previously demonstrated that Dpo4 elongates primer strands from a terminal oxoG•C base pair ∼30 times more efficiently than from an oxoG•A noncanonical pair (Rechkoblit et al., 2006). In order to account for this observation, we now compare the alignment of the oxoG(anti)•C(anti) base pair with that of the oxoG(syn)•A(anti) (molecule #1 of the oxoG•A-2 complex) and oxoG(anti)•A(syn) (molecule #2 of the oxoG•A-2 complex) noncanonical pairs at the (−1) postinsertion position of the Dpo4 active site. The correct positioning of metal ion A and the 3′-primer terminus are the major factors determining Dpo4 primer elongation efficiency (Vaisman et al., 2005; Wang et al., 2007). In order to consider these factors, the structures of the oxoG•A containing extension-inhibited complex are superimposed on the oxoG•C extension-proficient complex by the palm and finger domains that form the Dpo4 active site (Figures 5A and 5B). As evident from this superposition, neither the oxoG(syn)•A(anti) (Figure 5A) nor the oxoG(anti)•A(syn) (Figure 5B) noncanonical pairs induce significant structural perturbations within the active site of Dpo4. The positions of the catalytic metal ion A are essentially overlapped with one another in all three complexes, despite the shift toward the minor groove of the sugar rings of 3′-terminal A(anti) and A(syn) by ∼0.5 Å and ∼0.8 Å relative to C(anti). The C3′ to αP distance is 4.31 Å in the oxoG(anti)•C(anti) complex, 4.11 Å for oxoG(syn) •A(anti), and 3.99 Å for oxoG(anti) •A(syn), thereby allowing for the positioning of a modeled 3′-OH (the 3′-OH isabsentduetothe2′, 3′-dideoxy 3′-terminal bases) at a distance not far from the ∼3.5 Å value necessary for reaction with αP (Wang et al., 2007). Although the absence of the 3′-OH should affect subangstrom alignment of a polymerase active site, it is reasonable to expect similar consequences of the missing 3-OH on the alignments of the Dpo4 active site in the oxoG•C and oxoG•A complex structures. Despite the missing 3′-OH, metal ion A maintains an octahedral coordination due to an additional water molecule. The reaction mechanism of Dpo4 includes a rate-limiting initial proton transfer from 3′-OH to αP via a water molecule (Wang et al., 2007). We hypothesize that continuing dynamic interchange of A conformations due to back-and-forth oxoG(anti)-(syn) transitions within the Dpo4 extension complex could raise the reaction energy barrier, thereby slowing the primer extension rate.

Figure 5. Comparison of the Alignments of OxoG Pairs with C, A, G, or T and the Catalytic Divalent Ions at the Active Site of the Extension Ternary Complexes.

The structures are superimposed by the palm and finger domains that form the Dpo4 active site. The oxoG•C complex is shown in color, the oxoG•A-2, oxoG•G, and oxoG•T complexes are in beige.

(A) Comparison of the oxoG(syn)•A(anti) noncanonical pair and the active site ions of molecule #1 of the oxoG•A-2 complex with the oxoG(anti)•C(anti) pair and ions of the oxoG•C complex.

(B) Comparison of the oxoG(anti)•A(syn) noncanonical pair and the active site ions of molecule #2 of the oxoG•A-2 complex with the oxoG(anti)•C(anti) pair and ions of the oxoG•C complex.

(C) Comparison of the oxoG(anti)•G(syn) noncanonical pair and the active site ions of oxoG•G complex with the oxoG(anti)•C(anti) pair and ions of the oxoG•C complex.

(D) Comparison of the oxoG(anti)•T(anti) noncanonical pair and the active site ions of oxoG•T complex with the oxoG(anti)•C(anti) pair and ions of the oxoG•C complex.

By contrast, the active sites of the oxoG•G and oxoG•T extension complexes appear to be disrupted. Superposition of the oxoG•G (Figure 5C) and oxoG•T (Figure 5D) complexes with the oxoG•C extension complex reveal that the metal ion A is relocated away from the 3′-end primer terminus by 1.3 Å and 1.7 Å, respectively, toward metal ion B. The presence of Ca²⁺ ions (coordination distance 2.32 Å) instead of physiological Mg²⁺(coordination distance 2.1 Å) in the crystals of Dpo4 ternary complexes does not seem to be a cause for the shift of ion A. It has been previously shown that a mismatched nascent base pair, rather than the use of Ca²⁺ instead of Mg²⁺ ions for crystal growth, caused the metal ion displacement in the spacious Dpo4 active site (Vaisman et al., 2005). Moreover, Dpo4 is able to catalyze primer extension in the presence of Ca²⁺ on both unmodified and oxoG-containing DNA templates (Irimia et al., 2006). In addition to the displaced A ions in the oxoG•G and oxoG•T complexes, the sugar ring of 3′-terminal G14 is shifted by ∼0.5 Å toward the active site, and the sugar ring of T14 is shifted by ∼2.2 Å. Interestingly, the C3′ to αP distances of dGTP are maintained at 4.3 Å, not far from the near-catalytically competent range.

Factors Impacting on OxoG (syn)/(anti) Distribution in OxoG•A Dpo4 Extension Complexes

We find that Dpo4 affects the (anti)-(syn) equilibrium of the template oxoG at the (−1) position opposite primer A as a result of the following factors: (1) Arg332 forms favorable interactions with oxoG(anti), but not with oxoG(syn); (2) the conformation of the phosphate group of the template base at the (−1) position allows for the accommodation of both conformations of oxoG, but the phosphate group is too distant to permit an internal hydrogen bond with N2 of oxoG(syn), which would have further stabilized this alignment.

To the best of our knowledge, an oxoG in the anti conformation with adenine as a base-pairing partner has never been reported in a free DNA or within a protein-binding pocket. However, unmodified G(anti) can form a mismatch with A in the G(anti)•A (syn) alignment, as shown in a free DNA duplex crystal (Brown et al., 1986). However, a G(anti)•A(anti) arrangement at neutral pH in solution (Patel et al., 1984) or in a crystal (Prive et al., 1987) has also been observed. In silico data demonstrate that during unrestrained molecular dynamic simulations of the oxoG (anti)•A(anti) noncanonical pair within a free DNA duplex, the OxoG(anti) undergoes a spontaneous transition to the thermodynamically more preferred oxoG(syn) (Cheng et al., 2005). Thus, it is therefore unexpected and particularly noteworthy that the favorable interactions of Arg332 with the O8 and phosphate groups of the oxoG residue at the (−1) position within the spacious Dpo4 active site (Figure 3B) shift the conformation of oxoG opposite partner A toward the (anti) form, which in turn induces changes in the conformations of the 3′-terminal primer base A.

Furthermore, the availability of kinetic and structural information for primer extension from the oxoG•A noncanonical base pair in wild-type Dpo4, as well as in the Glu332 and Ala332 mutants, provides additional insights into the key role of Arg332 on the fidelity of bypass of oxoG catalyzed by wild-type Dpo4. The Glu332 and Ala332 Dpo4 mutants (Eoff et al., 2007b) demonstrate faster rates in extending primers from the oxoG•A noncanonical pair than wild-type Dpo4 (Zang et al., 2006). In the mutant Glu332 Dpo4 extension ternary complexes, Glu332 either forms a favorable water-mediated contact with N2 of oxoG(syn) opposite A(anti) or with O8 of oxoG(anti) opposite C(anti). Thus, Glu332 interacts equally favorably with oxoG(syn) or (anti), which allows partnering base A to direct the oxoG•A noncanonical pair toward the oxoG(syn)-A(anti) alignment. Consequently, conformational flexibility of oxoG, and, in turn, its partner base A, is reduced resulting in a faster extension from the dynamically less flexible 3′-terminal primer base A. Similarly, in the mutant Ala332 Dpo4 extension complex, the conformation of the oxoG is governed predominantly by its partnering base A because of the lack of interaction between the Ala332 residue with oxoG(syn) in the oxoG•A pair, or oxoG(anti) in the oxoG•C base pair.

Structural Origins Favoring Extension from the OxoG(syn)•A(anti) Pair by High-Fidelity Polymerases

High-fidelity polymerases strongly prefer to extend from the oxoG•A noncanonical pair over the oxoG•C base pair (Einolf and Guengerich, 2001; Furge and Guengerich, 1997; Lowe and Guengerich, 1996; Shibutani et al., 1991). In the BF (Hsu et al., 2004) and T7 (Brieba et al., 2004) high-fidelity polymerase structures with oxoG•A at the postinsertion site, oxoG adopts a syn conformation and engages in a Hoogsteen base pair with adenine. Remarkably, neither the polymerase active site residues nor the DNA itself are disrupted, which explains why the oxoG(syn)•A(anti) base pair evades proofreading of its minor groove edge and allows for efficient DNA synthesis past an oxoG lesion.

Moreover, in the BF and T7 structures, the syn conformation of theoxoG lesion is further stabilized by a direct hydrogen bonding contact of the N2 group of oxoG with an oxygen atom of its own phosphate group. Although the phosphate group of oxoG(syn) at the (−1) position of the BF extension binary complex is disordered, at the (−3) position in the BF complex (PDB ID 1U4B), determined at 1.6 Å resolution, an oxygen atom of the phosphate group of oxoG(syn) is 2.9 Å away from the N2 (Hsu et al., 2004). In the T7 polymerase extension ternary complex determined at lower 2.5 Å resolution (1TKB) (Brieba et al., 2004), the N2 of oxoG(syn) is 3.4 Å away from an oxygen atom of its phosphate group, suggesting a weak hydrogen bond.

Feasibility of (anti)-(syn) Interchange in OxoG•A Polymerase Extension Complexes

The conformational transitions between the oxoG(anti)•A(syn) and oxoG(syn) •A(anti) alignments in a polymerase active site likely require a looping out of the bases of oxoG and A residues from the template/primer DNA, either sequentially or simultaneously, followed by their reinsertion into the double helix. Dpo4 and other structurally studied Y-family polymerases, including yeast Pol η (Alt et al., 2007), human Pol ι (Nair et al., 2004), and Pol κ (Lone et al., 2007), have contacts with the sugar-phosphate backbone of the template/primer DNA at and adjacent to the template/primer active site segment, but not with the minor groove of the DNA duplex. By contrast, the minor groove edges of five base pairs of the growing template/primer DNA spanning the active site of high-fidelity polymerases are contacted by the protein, thereby incorporating a proofreading element that checks the correctness of the newly formed base-pairing alignments (Johnson and Beese, 2004; Steitz and Yin, 2004). In addition to the spatial restriction of the minor groove at the active sites of high-fidelity polymerases, the hydrogen bonds between the minor groove edge of a base pair at the postinsertion site and the “proofreading” residues would need to be disrupted so as to allow a base to loop out. Thus, (anti)-(syn) and (syn)-(anti) conformational transitions could occur through a looping out of the base into the accessible minor groove in the open active sites of Y-family polymerases, but are expected to be very difficult in the tight active sites of high-fidelity polymerases.

Contrasts in OxoG Lesion Bypass Efficiency and Accuracy by High-Fidelity versus Y-Family Polymerases

In the present work we provide structural insights into the ability of the low-fidelity Dpo4 bypass polymerase to preferentially extend from a C base and discriminate against extension beyond an A base opposite an oxoG lesion. In contrast, high-fidelity polymerases exhibit the opposite behavior. We demonstrate that Dpo4 accommodates the oxoG(anti)•C base pair at the (−1) postinsertion position without any perturbations in the template backbone or the architecture of the active site, thereby allowing for efficient extension from the C base at the 3′ primer terminus. Unexpectedly, although minimal alterations in the alignment of the Dpo4 active site occur during accommodation of an oxoG•A noncanonical pair, Dpo4 engages the oxoG residue in an (anti)-(syn) equilibrium that, in turn, triggers (syn)-(anti) conformational interchange of the partner A base. Thus, we hypothesize that during the extension step, Dpo4 appears to discriminate against an A opposite an oxoG lesion based primarily on its dynamic conformational flexibility at the to-be-extended 3′ primer terminus. Surprisingly, screening of such dynamic conformational flexibility, rather than checking of minor groove integrity as employed by high-fidelity polymerases, appears to provide discrimination against extension from an oxoG•A noncanonical pair by the Dpo4 bypass polymerase.

Y-Family Polymerases Might Employ Conformational Heterogeneity of Lesions and Their Pairing Partners to Select More Stable Conformers for Bypass

The Y-family Pol η is the only known polymerase besides Dpo4 capable of predominant extension from a C base opposite an oxoG lesion, as demonstrated for yeast and human Pol η (Carlson and Washington, 2005; Haracska et al., 2000; Zhang et al., 2000). Given that Pol η generally exhibits bypass properties of other lesions similar to those of Dpo4 (Boudsocq et al., 2001), it is tempting to speculate that Pol η could also employ a dynamic screening mechanism to discriminate against extension of an A base opposite an oxoG lesion. The superposition of the structurally similar little-finger domains of Pol η (Alt et al., 2007; Trincao et al., 2001) and Dpo4 indicates that Lys498 of Pol η is positioned equivalently to Arg332 of Dpo4 (Eoff et al., 2007b); thus it might be used similarly by Pol η to stabilize oxoG in the anti conformation to promote error-free incorporation of C, while triggering conformational heterogeneity of a partnering A base.

More generally, such dynamic screening may be employed by Y-family polymerases Dpo4 and Pol η as a discrimination tool for allowing selection of a partnering base that promotes the stable “correct” alignment of a lesion or an unmodified base. This would support primer elongation and discourage incorporation of a partnering base that would generate heterogeneity and consequently inhibit DNA synthesis. For example, it has been previously noted that there was inefficient misinsertion of dGTP opposite template T by Dpo4, and that the bases of either the incoming nucleotide or the 3′ nucleotide of the primer strand were disordered in the electron density maps, indicating conformational heterogeneity (Vaisman et al., 2005). The array of possibilities in the accommodation of an abasic site within the Dpo4 active site includes base deletion and addition frameshift structures (Ling et al., 2004a); this correlates with a ∼200-fold slower insertion and extension opposite an abasic site compared with an unmodified template (Fiala et al., 2007) and various mutagenic events observed in the fully extended products (Fiala and Suo, 2007). Conformational heterogeneity of lesions and their pairing partners, involving either a 3′-primer terminus or an incoming dNTP (Yang and Woodgate, 2007), creates a challenge to achieving a reaction-ready alignment of the spacious active site, thereby suggestive of a dynamic fidelity-check screening mechanism in Y-family polymerases.

Experimental Procedures

Crystallization

The crystals of the Dpo4 extension ternary complexes containing oxoG-modified 18-mer or 19-mer templates and the 13-mer primer terminated by either 2′,3′-dideoxycytosine, 2′,3′-dideoxyadenosine, 2′,3′-dideoxyguanosine, or 2′,3′-dideoxythymidine were grown in the presence of dGTP and flash-frozen in liquid nitrogen forX-ray data collection under conditions described previously (Rechkoblit et al., 2006). Several rounds of microseeding were employed to produce the diffraction quality crystals of the oxoG•C, oxoG•A-1, oxoG•A-2, oxoG•G, and oxoG•T extension ternary complexes.

Structure Determination and Refinement

X-ray diffraction data were collected at the NE-CAT 24-ID beam line at the Advanced Photon Source (Argonne National Laboratory, Chicago, IL). The data were processed and scaled using the HKL2000 suite. The structure of the oxoG•C extension complex was solved by molecular replacement in the P2₁2₁2 space group (AMoRe; Navaza, 1994) using our published oxoG-modified insertion ternary Dpo4-DNA-dCTP structure (Rechkoblit et al., 2006) as a search model. The model building, including substitution of the DNA sequence, was manually finished in TURBO-FRODO (http://www.afmb.univ-mrs.fr/-TURBO-) based on the electron density map calculated in REFMAC, and the resulting model was refined in REFMAC at 2.90 Å resolution to a final R-factor/R-free of 23.2/27.7 (Table 1). The structure of the oxoG•C complex was used as a model to calculate initial electron density maps in REFMAC for the isomorphous crystals of the oxoG•A-1 and oxoG•G complexes. Molecular replacement with the oxoG•C structure as a model was used to solve the oxoG•A-2 and oxoG•T complexes in P2₁ space group. The crystal data, together with the data collection and refinement statistics for all structures, are summarized in Table 1. The simulated annealing omit maps were calculated in CNS with the oxoG, partner base, and Arg332 omitted from the models after they were heated to 2000 K and then slowly cooled.