Insights into the mismatch discrimination mechanism of Y-family DNA polymerase Dpo4.

Nucleobases within DNA are attacked by reactive oxygen species to produce 7,8-dihydro-8-oxoguanine (oxoG) and 7,8-dihydro-8-oxoadenine (oxoA) as major oxidative lesions. The high mutagenicity of oxoG is attributed to the lesion's ability to adopt syn-oxoG:anti-dA with Watson-Crick-like geometry. Recent studies have revealed that Sulfolobus solfataricus P2 DNA polymerase IV (Dpo4) inserts nucleotide opposite oxoA in an error-prone manner and accommodates syn-oxoA:anti-dGTP with Watson-Crick-like geometry, highlighting a promutagenic nature of oxoA. To gain further insights into the bypass of oxoA by Dpo4, we have conducted kinetic and structural studies of Dpo4 extending oxoA:dT and oxoA:dG by incorporating dATP opposite templating dT. The extension past oxoA:dG was ~5-fold less efficient than that past oxoA:dT. Structural studies revealed that Dpo4 accommodated dT:dATP base pair past anti-oxoA:dT with little structural distortion. In the Dpo4-oxoA:dG extension structure, oxoA was in an anti conformation and did not form hydrogen bonds with the primer terminus base. Unexpectedely, the dG opposite oxoA exited the primer terminus site and resided in an extrahelical site, where it engaged in minor groove contacts to the two immediate upstream bases. The extrahelical dG conformation appears to be induced by the stabilization of anti-oxoA conformation via bifurcated hydrogen bonds with Arg332. This unprecedented structure suggests that Dpo4 may use Arg332 to sense 8-oxopurines at the primer terminus site and slow the extension from the mismatch by promoting anti conformation of 8-oxopurines.


INTRODUCTION
Human DNA is under constant threat from both endogenous and exogenous sources, and reactive oxygen species are one of the major DNA damaging agents [1]. These reactive oxygen species are usually formed from exogenous ionizing radiation and/or endogenous free radicals, and they are known to generate a wide range of DNA lesions including 7,8-dihydro-8-oxoguanine (oxoG) and 7,8-dihydro-8-oxoadenine (oxoA) [2] (Figure 1). The overproduction of these reactive oxygen species is reported to increase the oxidative stress, which results in many harmful effects in the cells including the increased risk of cancer development [3]. OxoG, one of the best studied lesions caused by reactive oxygen species, induces G to T transversions in human cells [4]. This usually happens when high-fidelity polymerases insert A opposite oxoG with favorable extension from oxoG:dA pair [5,6]. On the other hand, the Y-family DNA polymerases including yeast and human polη [3,7] and Dpo4 [8] preferentially insert C opposite oxoG, and the extension past oxoG:dC occurs more favorably than other pairs [8,9].
Unlike oxoG, whose bypass and mutagenicity have been intensively investigated, oxoA has not been in the center of scientific focus. The oxoA, whose cellular levels are about one third to one half of oxoG [10,11], is shown to be cleaved from oxoA:dT, oxoA:dG, and oxoA:dC base pairs by human thymine DNA glycosylase (TDG) and E. coli mismatch-specific uracil DNA glycosylase (MUG) [12,13].
The oxoA in oxoA:dC base pair can also be removed by 8-Oxoguanine DNA glycosylase (hOGG1) and endonuclease VIII-like protein 1 (NEIL1) in vitro [14,15]. While oxoA blocks the activity of RNA polymerase II significantly, oxoG is readily bypassed by the same enzyme [16]. The correct insertion (dTTP) opposite oxoA is almost dominantly observed from prokaryotic DNA polymerases such as Escherichia coli DNA polymerase I and Taq DNA polymerase [17,18], while the incorrect insertion of dGTP opposite oxoA is frequently observed from mammalian DNA polymerases α and β in vitro [19].
While the amount of the biochemical information on oxoA increases, the structural basis for the mutagenicity of oxoA is limited. In the absence of protein contact, oxoA adopts a syn conformation and forms a Hoogsteen base pair with a wobble geometry [20]. In the replicating base pair site of human polymerase β (polβ) Sulfolobus solfataricus P2 DNA polymerase IV (Dpo4), oxoA forms Hoogsteen pairing with dGTP with Watson-Crick-like geometry, promoting A to C transversions [21,22]. Insertion of dGTP opposite oxoA by Dpo4 is ~300-fold more efficient than the dA:dGTP insertion, revealing the promutagenic characteristics of oxoA. In the catalytic site of human polymerase η (polη), oxoA forms a Hoogsteen base pairing with dGTP with a wobble geometry [23]. These studies have revealed, unlike oxoG, the mutagenicity of oxoA is induced by minor groove contacts by DNA polymerases. Despite these advances, a structure of a DNA polymerase extending oxoA:dT or oxoA:dG has not been reported, limiting our understanding of oxoA-mediated mutagenesis.
Herein, we report kinetic data for extension past oxoA:dT and oxoA:dG by Dpo4 along with two crystal structures of Dpo4 extending past oxoA:dT and oxoA:dG. These kinetic and structural studies provide new insights into the mutagenicity of oxoA in prokaryotes and the mismatch discrimination mechanism of Dpo4.

MATERIALS AND METHODS
Protein expression, crystallization and structure determination. Dpo4 was expressed and purified from E. coli with minor modifications of the method described previously [24]. The ternary Dpo4 complex with the templating oxoA was prepared and crystallized as described previously with minor optimization [24]. To obtain the ternary complex of Dpo4-DNA extension complex past oxoA:dT and oxoA:dG crystals, a 19-mer template (5'-TTCAT[oxoA]GAATCCTTCCCCC-3' Midland Certified Reagent, Midland, TX) and a 14-mer primer (5'-GGGGGAAGGATTCY-3' Y = G or T, Integrated DNA technologies, Coralville, IA) were synthesized. The template and the primer oligonucleotides were annealed in hybridization buffer (10 mM Tris-HCl pH 7.5, 1 mM EDTA) by heating at 90 o C for 5 min followed by slow cooling to an ambient temperature. The annealed DNA (1.2 molar excess) was mixed with 8 mg/ml Dpo4 to form a binary complex. Subsequently, a 10-fold molar excess of nonhydrolyzable dAMPNPP (Jena Bioscience) was added to the binary complex. Ternary Dpo4-DNA complex co-crystals with nonhydrolyzable dATP analog (dATP*) paired with templating dT one base upstream of oxoA lesion were grown in a buffer solution containing 50 mM Tris pH 7.5, 16-25% PEG3350, and 100 mM magnesium acetate. Crystals were cryoprotected in mother liquor supplemented with 15% ethylene glycol and were flashfrozen in liquid nitrogen. Diffraction data were collected at 100 K at the beamline 19 ID at the Advanced Photon Source and at the beamline BL 5.0.1. at the Lawrence Berkeley National Laboratory. All diffraction data were processed using HKL 2000 [32]. Structures were solved by molecular replacement with a ternary Dpo4 complex structure (PDB ID: 1JX4) as a search model using Molrep [33]. The model was built using COOT [34] and was refined using PHENIX [35]. MolProbity was used to make Ramachandran plots [36]. All the crystallographic figures were generated using Chimera [37].

Steady-state kinetics of single nucleotide incorporation and extension opposite templating oxoA by
Dpo4. Steady-state kinetic parameters for insertion opposite oxoA and extension past oxoA:dN by Dpo4 were measured as described previously with minor modification [21]. Briefly, the oligonucleotides DNAs for kinetic assays (primer, 5´-FAM/GGGGGAAGGATTCN-3´ (N = G or T) and template, 5´-

OxoA at templating strand promotes error-prone bypass by Dpo4.
To evaluate the efficiency of extension past oxoA by Dpo4, we determined kinetic parameters for Dpo4 incorporating dATP opposite the templating dT past oxoA:dT or oxoA:dG along with the extension past dA:dT and dA:dG (Table 1 and Figure 2). Similar to the insertion opposite oxoA lesion, which increased the reaction efficiency (k cat /K m ) for the incorrect insertion (dGTP) greatly compared to the incorrect insertion opposite the control dA [22], the extension past oxoA:dG displayed the increased reaction efficiency from the extension past dA:dG. In the presence of the correct oxoA:dT base pair at the primer terminus position, Dpo4 incorporated dATP opposite the templating dT with a relative efficiency (k cat /K m ) of 0.21 (2.94x10 -3 s -1 μM -1 for the extension past oxoA:dT vs. 13.80x10 -3 s -1 μM -1 for the extension past dA:dT). Upon the introduction of oxoA lesion, K m was increased about 3-fold (15.29 vs. 5.44 μM), and k cat was reduced by about 60% (44.5x10 -3 s -1 vs. 74.8x10 -3 s -1 ) relative to the extension past undamaged dA:dT. In the presence of an incorrect oxoA:dG base pair at the primer terminus position, Dpo4 incorporated dATP opposite the templating dT with a relative efficiency of 0.64, which was ~2-fold more efficient than the extension past dA:dG (0.34 of k cat /K m for the extension past dA:dG vs. 0.64 for the Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20210162/908586/bcj-2021-0162.pdf by guest on 22 April 2021 extension past oxoA:dG), indicating that Dpo4's efficiency for bypassing oxoA lesion for the incorrect insertion (dGTP) is maintained comparable in the extension step as well. The catalytic efficiency (k cat /K m ) of dATP extension past oxoA:dT was about 20% of that for the same extension past dA:dT (2.94 10 -3 s -1 μM -1 vs. 13.80). The catalytic efficiency (k cat /K m ) of dATP extension past oxoA:dG was about twice as much as for the same extension past dA:dG (0.64x10 -3 s -1 μM -1 vs. 0.34x10 -3 s -1 μM -1 ). While Dpo4 extended dA:dG ~40-fold less efficiently than dA:dT, it extended oxoA:dG only ~5-fold less efficiently than oxoA:dT, indicating that the presence of oxoA at template strand promotes mutagenic replication of Dpo4.

Structure of Dpo4 extending oxoA:dT by incorporating dATP opposite templating dT
Our kinetic studies showed the extension of oxoA:dT by Dpo4 is only ~5-fold less efficient than the extension of dA:dT. To gain structural insights into the relatively efficient extension of oxoA lesion by Dpo4, we solved ternary complex structures of Dpo4 incorporating a nonhydrolyzable dAMPNPP (dATP* hereafter) opposite a templating dA past oxoA:dT in the presence of Mg 2+ . The ternary complex of the extension past oxoA:dT was crystallized in P 1 space group with the cell dimension of a = 53.3 Å, b = 99.0 Å, c = 102.9 Å, α = 90.03 º , β = 89.99 º , and γ = 89.92 º , and the structure was refined to 2.42 Å with R work = 19.9% and R free = 24.2%. The statistics for data collection and the refinement were summarized in Table. 2. Similar to the structure of Dpo4 incorporating dTTP opposite templating oxoA, the structure of Dpo4 extending past oxoA:dT exhibited the same conserved secondary structures and the four characteristic domains (thumb, palm, finger, and little finger) of Y-family DNA polymerases (Figure 3-A). The electron density around the incoming dATP* at the extension position, anti-oxoA, and dT at the primer terminus is well-ordered at the contour level of 1σ (Figure 3-B). The extension past the correct insertion (oxoA:dT) structure showed about 20 o distortion between the anti-oxoA and the primer terminus dT base pair, and their inter-base hydrogen bonding distances were 2.8 Å and 3.5 Å (Figure 3-C). The 8oxo moiety of anti-oxoA engages in hydrogen bonding interactions with Arg332. The extension base pair between the templating dT and the incoming dATP* displayed optimal Watson-Crick base pairing with oxoA retaining an anti-conformation with no noticeable distortion (Figure 3-C), and the inter-base hydrogen bonding distances between the templating dT and the incoming dATP* were 2.7 Å and 3.2 Å.
Compared to the published insertion structures (Dpo4-oxoA:dTTP (PDB Code: 6VGM) and Dpo4-oxoA:dGTP (PDB Code: 6VG6) structures [22]), the extension structure past oxoA:dT showed much more optimal metal coordination for the reaction. The primer terminus OH group was coordinated to the A-site metal, 3.2 Å away from the P α of incoming dATP*, and it is well aligned for an in-line nucleophilic attack on the P α (Figure 3-D). Both A-and B-site magnesium ions were fully coordinated with Asp7, Phe8, Asp105, Glu106, phosphate oxygens, and 3'-OH of primer terminus, showing that Dpo4 extension of dATP* past oxoA:dT ternary complex was poised for nucleotidyl transfer reaction at replication site (Figure 3-D).

Structure of Dpo4 extending oxoA:dG by incorporating dATP opposite templating dT
Our kinetic studies showed that Dpo4's efficiency of the oxoA:dG extension was only ~5-fold less efficient than that of the oxoA:dT extension. To gain insight into the promutagenic nature of oxoA, we solved structure of Dpo4 incorporating dATP opposite dT in the presence of oxoA:dG at the primer terminus position. The ternary complex was crystallized in P2 1 space group with the cell dimension of a = 53.1 Å, b = 103.7 Å, c = 99.1 Å, α = 90.00 º , β = 89.94 º , and γ = 90.00 º , and the structure was refined to 2.54 Å with R work = 21.0% and R free = 26.0%. The statistics for data collection and the refinement were summarized in Table. 2. Similar to the structure of Dpo4 extending past oxoA:dT, the structure of Dpo4 extending past oxoA:dG exhibited the same conserved secondary structures and the four characteristic domains (thumb, palm, finger, and little finger) of Y-family DNA polymerases (Figure 4-A). The overall structures of the extensions past oxoA:dT and oxoA:dG were almost identical with an RMSD value of 0.30 Å.
The Dpo4-oxoA:dG extension structure reveals a novel strategy by which the enzyme deters extension of a mismatched base pair. While the nascent dT:dATP base pair in the Dpo4-oxoA:dG extension structure adopts a Watson-Crick base pair geometry, the primer terminus dG did not form base pairing with oxoA. Unlike the Dpo4-oxoA:dGTP insertion structure that displayed syn-oxoA and anti-dGTP conformations, oxoA in the Dpo4-oxoA:dG extension structure adopted an anti conformation, thereby engaging in bifurcate hydrogen bonding interaction with Arg332. Interestingly, the primer terminus dG exited the primer terminus site and resided in an extrahelical site near the immediate two upstream bases. The primer terminus dG was in a syn conformation, which allowed the formation of minor groove contacts to the upstream bases via the N2 of the dG (Figure 4-B). The further stabilization of the flipped-out dG conformation was provided by the hydrogen bonds among the 3'-OH of the dG and Glu106 and Tyr108 of Dpo4 (Figure 4-D). It appears this unprecedented conformation was caused by the stabilization of anti-oxoA conformation by Arg332-mediated bifurcate hydrogen bonds, which would generate a steric clash between anti-oxoA and anti-dGTP. The strong electron densities for the incoming dATP*, anti-oxoA, and the primer terminus dG are observed at the contour level of 1σ (Figure 4-B).
Despite of the formation of the unusual conformation at the primer terminus site, the base pairing between the incoming dATP and templating dT was not significantly affected, forming two Watson-Crick hydrogen bonds with the distances of 2.9 Å and 3.1 Å (Figure 4-C).
Not surprisingly, the Dpo4-oxoA:dG extension structure showed much less optimal metal coordination for the reaction compared with the extension past oxoA:dT. The primer terminus dG was not coordinated to the A-site metal. Instead, it moved away from the A-site metal and was hydrogen bonded to Glu106 that was coordinated to the catalytic metal ion (metal A). In addition, 3'-OH of the extrahelical dG was 9.6 Å away from the P α of incoming dATP* (Figure 4-D), suggesting that the extension reaction would be greatly hindered with this conformation. Nevertheless, both A-and B-site magnesium ions were fully coordinated with Asp7, Phe8, Asp105, Glu106, phosphate oxygens, and water molecules (Figure 4-D), indicating that Dpo4's extension reaction past oxoA:dG would be feasible should the dG is brought back into the active site. Overall, the Dpo4-oxoA:dG extension structure presented here does not represent a catalytically competent conformation. The fact that the oxoA:dG extension is ~5-fold less efficient than the oxoA:dT extension being considered, the energy difference between the Dpo4-oxoA:dG extension complex and a catalytically competent complex would not be significant.

The role of Arg332 of Dpo4 in stabilizing 8-oxopurines in an anti conformation.
The bifurcated hydrogen bonds between Arg332 of Dpo4 and the 8-oxo moiety of oxopurines would promote the formation of the oxopurines in an anti conformation, which in turn would deter the incorporation of purines opposite the lesions. Our oxoA:dT and oxoA:dG extension structures show that oxoA does not undergo anti to syn conformational change. In both structures, Arg332 is proximal to the 8-oxo group of oxoA (Figure 3-C and 4-C) and keeps the oxoA from going through anti-to-syn conformational change. To evaluate whether Arg332-mediated interaction is Dpo4's general extension strategy to slow the extension of mispaired 8-oxopurines, we compared our Dpo4-oxoA:dG extension structure with the published Dpo4-oxoG:dA extension structures [9]. Interestingly, the authors reported two structures bearing either syn or anti conformation of oxoG at the primer terminus base pair site. The structure with syn-oxoG:dA (PDB Code: 3GIJ) shows electron density for both oxoG and dA, while the structure with anti-oxoG:dA (PDB Code: 3GII) shows electron density for oxoG only, indicating the primer terminus dA is disordered. A comparison of our oxoA:dG extension structure with the syn-oxoG:dA extension structure reveals noticeable conformational difference for Arg332. In the syn-oxoG:dA structure, Arg332 moves away from syn-oxoG (Figure 5-A), which would prevent steric clash between the guanidinium moiety of Arg332 and the N2 of oxoG. Instead of interacting with oxoG, Arg332 in the syn-oxoG:dA extension structure engages in hydrogen bonding interaction with a phosphate oxygen of template strand. In contrast, Arg332 in the oxoA:dG extension structure forms two hydrogen bonds with anti-oxoA, thereby stabilizing anti-oxoA conformation.
Another notable difference between the oxoA:dG and syn-oxoG:dA structures is found in the position of A-site metal. While the conformations of the nascent base pairs are well overlaid, the position of A-site metal shifted 2.4 Å relative to Ca 2+ in oxoG:dA extension structure probably due to the movement of primer terminus dG in the oxoA:dG extension structure (Figure 5-B).
When our anti-oxoA:dG extension structure is compared with the anti-oxoG:dA extension structure, the overall conformations of the two structures are almost identical with no significant deviation in both Dpo4 and DNA (Figure 5-C). However, while dG across anti-oxoA had a clear density with hydrogen-bonding interactions with Glu106 and Tyr108 of Dpo4 and the primer bases, dA across anti-oxoG is disordered (Figure 5-D). This conformational heterogeneity of the oxoG:dA structures could impact the fidelity and efficiency of the bypass across oxoG in Dpo4 [9], which would partially explain a reported ~30-fold reduction in the extension efficiency for oxoG:dA compared with that for oxoG:dC [8].
Similarly, Dpo4 extension past oxoA:dG, although its syn-oxoA:dG base pair has not been crystallized, would be slowed by this conformational heterogeneity, which could contribute to ~5-fold reduction in the extension efficiency for oxoA:dG compared with that for oxoA:dT.

Dpo4 and polη use different mechanisms for the bypass across oxidative lesions.
Dpo4 has been shown to bypass oxoG in a highly accurate manner (70:1 in the insertion and 30:1 in the extension) [8], while it bypasses oxoA relatively error-free manner (14:1 in the insertion and 5:1 in the extension) [22]. Unlike Dpo4, human polη bypasses oxoG in an error-prone manner (3:1 in the insertion and 1:1 in the extension) [38,39]. Also, the insertion bypass of oxoA by polη is promutagenic (3:2 between correct and incorrect insertion) [23]. To understand what contributes to these differences in the reaction fidelity in the Y-family DNA polymerases in prokaryotes and eukaryotes, we compared Dpo4 extension structures with polη extension past oxoG:dC (correct insertion) and oxoG:dA (incorrect insertion) structures (PDB Code:4O3S and 4O3R, respectively) [38]. The superposition of the Dpo4-oxoA:dT extension structure with polη-oxoG:dC extension structure reveals the oxo moiety of 8oxopurines is adjacent to Arg332 of Dpo4 and Asn324 of polη, respectively (Figure 6-A). However, while Arg332 of Dpo4 makes bifurcate hydrogen bonds with the 8-oxo moiety of anti-8-oxopurines, Asn324 is 3.6 Å away from the 8-oxo moiety of oxoG, thereby not being able to effectively stabilize 8oxopurines in an anti conformation. The ability of Dpo4 to stabilize 8-oxopurine in an anti conformation would discourage the formation of syn-oxopurine:anti-purine (e.g., syn-oxoG:anti-dA, syn-oxoA:anti-dG) base pair conformations at the primer terminus site. As anti-oxopurine:anti-purine base pair conformation would be strongly discouraged by the geometric restraint at the primer terminus site of Dpo4, the presence of Arg332-mediated hydrogen bond to anti-oxopurines will preferentially promote the extension of oxopurines:pyrimidine (e.g., oxoG:dC, oxoA:dT) but not oxopurines:purines (e.g., oxoG:dA, oxoA:dG). This is in consistent with the observed error-free extension of 8-oxopurines by Dpo4. Due to The superposition of the Dpo4-oxoA:dG extension structure with published polη-oxoG:dA extension structure reveals large conformational differences (Figure 6-B). First, oxoG and dA at the primer terminus site in polη forms Hoogsteen base pairing with a Watson-Crick-like geometry. In stark contrast, oxoA does not pair with dG at the primer terminus site due to the extrusion of primer dG from the active site. As described above, the primer dG opposite oxoA is positioned right outside the active site, yet the two catalytic magnesium ions and the incoming nucleotide (dATP*) are well positioned in the active site, indicating that the reaction can occur via the relocation of the extrahelical dG to the active site ( Figure 6B).
Dpo4 participates in both insertion and extension steps for the bypass of many lesions, while polη primarily involves in the insertion step and engages another polymerase for the extension step [40,41], with some exceptions of the lesion bypass including cyclobutane pyrimidine dimer lesion [42]. Dpo4 has been shown to utilize template translocation and misalignment to discriminate the correct bypass over the incorrect [43]. Our study presented here along with the previous Dpo4-oxoA/oxoG bypass studies display that the smaller active site could help Dpo4 with effectively discriminating the incorrect bypass at the insertion and extension steps. The heterogeneity of the oxoA/oxoG base pairs in the extension step and the interactions with the Dpo4 residues also contributed to this discrimination. Dpo4's unique strategy is an example of how consecutive promutagenic bypass on the insertion and extension steps is avoided via various mechanisms of geometric constraints.

CONCLUSION
In summary, our kinetic and structural studies provide further insights into the mutagenic potential of 8oxoadenine and a unique discrimination mechanism Dpo4 uses to increase the replication fidelity past 8oxopurines. Arg332 of Dpo4 facilitates 8-oxopurines to adopt an anti conformation at the template opposite primer terminus, thereby slowing the formation of incorrect base pairs such as syn-oxoG:anti-dA and syn-oxoA:anti-dG. In the presence of anti-oxoG or anti-oxoA at the primer terminus base pair site, a purine opposite the 8-oxopurine does not form a stable base pair with the lesion, thereby either being disordered (8-oxoG:dA) or resided on an extrahelical site (8-oxoA:dG). Our findings presented here give us more in-depth understanding regarding the prokaryotic cell's selection for a specific TLS polymerase for the bypass of various DNA lesions to achieve an error-free DNA lesion bypass.         a Values in parentheses are for the highest resolution shell. b R merge = Σ|I-<I>|/ ΣI where I is the integrated intensity of a given reflection. c R work = Σ|F(obs)-F(calc)|/ΣF(obs). d R free = Σ|F(obs)-F(calc)|/ΣF(obs), calculated using 5% of the data.