Amino Acid Templating Mechanisms in Selection of Nucleotides Opposite Abasic Sites by a Family A DNA Polymerase

Background: Abasic sites are the most frequent DNA lesions and are often bypassed by incorporating an adenosine opposite that lesion.

Results: We determined structures of DNA polymerase in complex with different nucleotides opposite an abasic site.

Conclusion: Interaction of the incoming nucleotide with a single amino acid governs nucleotide selection opposite abasic sites.

Significance: This work furthers the understanding of the bypass of a mutagenic lesion by DNA polymerases.

Abstract

Cleavage of the N-glycosidic bond that connects the nucleobase to the backbone in DNA leads to abasic sites, the most frequent lesion under physiological conditions. Several DNA polymerases preferentially incorporate an A opposite this lesion, a phenomenon termed “A-rule.” Accordingly, KlenTaq, the large fragment of Thermus aquaticus DNA polymerase I, incorporates a nucleotide opposite an abasic site with efficiencies of A > G > T > C. Here we provide structural insights into constraints of the active site during nucleotide selection opposite an abasic site. It appears that these confines govern the nucleotide selection mainly by interaction of the incoming nucleotide with Tyr-671. Depending on the nucleobase, the nucleotides are differently positioned opposite Tyr-671 resulting in different alignments of the functional groups that are required for bond formation. The distances between the α-phosphate and the 3′-primer terminus increases in the order A < G < T, which follows the order of incorporation efficiency. Additionally, a binary KlenTaq structure bound to DNA containing an abasic site indicates that binding of the nucleotide triggers a remarkable rearrangement of enzyme and DNA template. The ability to resolve the stacking arrangement might be dependent on the intrinsic properties of the respective nucleotide contributing to nucleotide selection. Furthermore, we studied the incorporation of a non-natural nucleotide opposite an abasic site. The nucleotide was often used in studying stacking effects in DNA polymerization. Here, no interaction with Tyr-761 as found for the natural nucleotides is observed, indicating a different reaction path for this non-natural nucleotide.

Introduction

The most common DNA damage under physiological conditions consists of abasic sites resulting from spontaneous hydrolysis of the N-glycosidic bond between the sugar moiety and the nucleobase in DNA (1). Because the genetic information is lost by the cleavage of the nucleobase, abasic sites bear a high mutagenic potential (2–4). In most cases, the lesion is removed by DNA repair systems using the sister strand to guide for incorporation of the right nucleotide. However, undetected lesions, or those formed during S phase, pose a challenge to DNA polymerases and block replication (5, 6). Several studies indicated the mutagenic potential of these lesions in translesion synthesis, which is more pronounced in animal as compared with bacterial cells presumably due to higher translesion synthesis in eukaryotes (4, 7, 8). Interestingly, it has been shown that in human cells, an adenosine (A) is preferentially incorporated opposite abasic sites (4). Further in vitro and in vivo studies of DNA polymerases from family A (including human DNA polymerases γ and θ) and B (including human DNA polymerases α, ϵ, and δ) in the presence of the stabilized tetrahydrofuran abasic site analog F (supplemental Fig. S1) have shown that purines, in particular adenosine, and to a lesser extent guanosine, are most frequently incorporated opposite the lesion. The strong preference for adenosine has been termed “A-rule” (7–20).

A set of studies concerning the behavior of DNA polymerases from different sequence families showed that there are multiple mechanisms to overcome an abasic site. Most translesion DNA polymerases follow various loop-out mechanisms (11, 21–25). Thereby, the nucleotide selection is influenced by the following upstream templating bases resulting in deletions and complex mutation spectra. Recently, an amino acid templating mechanism was found for the “error-free” bypass of an abasic site by the yeast Rev1 DNA polymerase member of family Y (26). Because guanine is cleaved most frequently (2), the preference of Rev1 for dCMP incorporation opposite an abasic site represents the “best guess.”

However, the determinants of the A-rule are still controversially discussed. Structural and functional studies have added significantly to our understanding of the basic mechanisms of translesion synthesis by DNA polymerases (27, 28). Superior stacking as well as solvation properties of adenine have been discussed as the driving force behind adenine selection (17, 29–31). However, previously reported structures of KlenTaq, the large fragment of Thermus aquaticus (Taq) DNA polymerase I, which is a member of the sequence family A, suggests that this enzyme follows the A-rule by applying an amino acid templating mechanism (19, 20). Thereby, interaction with Tyr-671 seems to be the basis of the preference of purines. It facilitates nucleotide incorporation by mimicking a pyrimidine nucleobase directing for purine incorporation opposite abasic sites due to the enhanced geometric fit to the active site (32, 33). The crucial role of this tyrosine in translesion synthesis, which is highly conserved throughout evolution in DNA polymerase family A from bacteria to humans (34), was further probed by analysis of site-directed mutations (19). Interestingly, when the six-membered ring of tyrosine was mutated to the bicyclic indole of tryptophan, the A-rule was shifted to a “C/T-rule,” further supporting geometric factors as determinants of nucleotide selection opposite an abasic site.

Still, several aspects remain to be clarified to fully understand the mechanism of nucleotide selection opposite an abasic site by DNA polymerases from the family A. These aspects include the overall conformation of the enzyme-DNA complex in the binary state prior to nucleotide binding. Furthermore, the order of nucleotide incorporation efficiency opposite abasic sites by DNA polymerases such as KlenTaq is A, G, T, and C. The mechanistic understanding for this observation is sparse due to the lack of structural data. Here, we report a structure of a binary complex of KlenTaq DNA polymerase bound to the primer/template bearing an abasic site lesion. Additionally, we present several ternary structures of KlenTaq caught incorporating different nucleotides opposite an abasic site, providing insights into the preference order of nucleotide incorporation efficiency opposite this lesion.

EXPERIMENTAL PROCEDURES

Protein and Oligonucleotides

Protein expression and purification were conducted as described (35). In brief, an Escherichia coli codon-optimized KlenTaq gene (amino acids 293–832 of Taq gene; purchased from GeneArt) was cloned into a pET-21b vector without any purification tags and expressed in E. coli strain BL21 (DE3). Of note, codon optimization resulted in changes within the gene sequence without affecting the amino acid sequence. After heat denaturation and ultracentrifugation, a PEI precipitation was performed. The resulting material was purified by anion exchange (Q Sepharose) chromatography followed by size-exclusion chromatography (Superdex 75) in 20 mm Tris HCl, pH 7.5, 150 mm NaCl, 1 mm EDTA, 1 mm β-mercaptoethanol.

Oligonucleotides were purchased from Metabion or Thermo Scientific. The dideoxy cytosine modified primer was synthesized on an Applied Biosystems 392 DNA/RNA synthesizer using 5′-dimethoxytrityl-2′,3′-dideoxy cytosine modified N-succinoyl-long chain alkylamino controlled pore glass, which was purchased from Glen Research. The synthesized oligonucleotide was purified by preparative PAGE on a 12% polyacrylamide gel containing 8 m urea (DMT-OFF). The nucleotide dNITP³ was synthesized starting from dNI purchased from Berry & Associates Inc., according to published procedures (36).

Crystallization and Structure Determination

KlenTaq (buffer: 20 mm Tris HCl, pH 7.5, 150 mm NaCl, 1 mm EDTA, 1 mm β-mercaptoethanol) was incubated in the presence of DNA primer (5′-d(GAC CAC GGC GC)-3′), an abasic site F containing template (5′-d(AAA FNG CGC CGT GGT C)-3′); N represents the templating base directing the processing of ddGTP, ddTTP, or ddCTP.

KlenTaq_F-G-I

The crystallization was set up using purified KlenTaq (11 mg/ml), DNA template/primer duplex, and ddGTP in a molar ratio of 1:3:50 and in the presence of 20 mm MgCl₂. The crystallization solution was mixed in a 1:1 ratio with the reservoir solution containing 0.05 m sodium cacodylate (pH 6.5), 0.2 m NH₄OAc, 0.01 m Mg(OAc)₂, and 25% PEG 8000.

KlenTaq_F-G-II

The crystallization was set up as for KlenTaq_F-G-I except using KlenTaq (11 mg/ml):DNA template/primer duplex:ddGTP = 1:3:50 in the presence of 20 mm MgCl₂. The reservoir solution contained 0.05 m sodium cacodylate (pH 6.5), 0.2 m NH₄OAc, 0.01 m Mg(OAc)₂, and 28% PEG 8000.

KlenTaq_F-T

The crystallization was set up as for KlenTaq_F-G-I except using KlenTaq (11 mg/ml):DNA template/primer duplex:ddTTP = 1:3:50 in the presence of 20 mm MgCl₂. The reservoir solution contained 0.05 m sodium cacodylate (pH 6.5), 0.2 m NH₄OAc, 0.01 m Mg(OAc)₂, and 28% PEG 8000.

KlenTaq_F-binary-II

The crystallization was set up as for KlenTaq_F-G-I except using KlenTaq (11 mg/ml):DNA template/primer duplex:ddCTP = 1:1.5:60 in the presence of 20 mm MgCl₂. The reservoir solution contained 0.1 m Tris HCl (pH 8.0), 0.2 m Mg(formiate)₂, and 15% PEG 8000.

KlenTaq_F-binary

The crystallization was set up using purified KlenTaq (8 mg/ml), a template (5′-d(AAA FGG CGC CGT GGT C)-3′), and a previously synthesized primer with a dideoxy cytosine at its 3′-end, in a molar ratio of 1:1.4 and in the presence of 20 mm MgCl₂. The crystallization solution was mixed in 1:1 ratio with the reservoir solution containing 0.1 m Tris HCl (pH 7.5), 0.2 m Mg(formiate)₂, and 12% PEG 8000.

KlenTaq_F-NI

The crystallization was set up as for KlenTaq_F-binary except using KlenTaq (8 mg/ml):DNA template/primer duplex = 1:1.4 and in the presence of 20 mm MgCl₂. The reservoir solution contained 0.1 m Tris HCl (pH 7.0), 0.2 m Mg(formiate)₂, and 18% PEG 8000. The crystals were soaked overnight by adding 0.5 μl of dNITP (20 mm) solution. The occupancy of the bound dNITP was refined to 0.77.

Crystals were produced by the hanging drop vapor diffusion method by equilibrating against 1 ml of the reservoir solution for 5 days at 18 °C. The crystals were frozen in liquid nitrogen. Datasets were collected at beamline PXI (X06SA) of the Swiss Light Source (Paul Scherrer Institute, Villigen, Switzerland), at a wavelength of 1.000 or 0.900 Å using a PILATUS 6M detector. Data reduction was performed with the XDS package (37, 38). The structures were solved by difference Fourier techniques using KlenTaq wild type (Protein Data Bank (PDB) 3LWL) as model. Refinement was performed with PHENIX (39), and model rebuilding was done with COOT (40). Figures were made with PyMOL (41).

Primer Extension Assays

Primer extension was performed as described (19). In brief, for incorporation opposite F, 20 μl of the KlenTaq reactions contained 100 nm primer (5′-d(CGT TGG TCC TGA AGG AGG ATA GG)-3′), 130 nm template (5′-d(AAA TCA FCC TAT CCT CCT TCA GGA CCA ACG TAC)-3′), 100 μm dNTPs in buffer (20 mm Tris HCl, pH 7.5, 50 mm NaCl, and 2 mm MgCl₂), and 500 nm of the respective KlenTaq polymerase. Reaction mixtures were incubated at 37 °C. Incubation times are provided in the respective figure legends. Primer was labeled using [γ-³²P]ATP according to standard techniques. Reactions were stopped by the addition of 45 μl of stop solution (80% (v/v) formamide, 20 mm EDTA, 0.25% (w/v) bromphenol blue, 0.25% (w/v) xylene cyanol) and analyzed by 12% denaturing PAGE. Visualization was performed by phosphorimaging.

Enzyme Kinetics

The rates of single turnovers in pre-steady-state kinetics were determined as described (19). In brief, 15 μl of radiolabeled primer/template complex (200 nm) and DNA polymerase (2 mm) in reaction buffer (see primer extension assay) were rapidly mixed with 15 μl of a dNTP solution in reaction buffer at 37 °C. Quenching was achieved by adding previously described stop solution. For reaction times longer than 5 s, a manual quench was performed. The analysis of dNTP incorporation opposite to the abasic site primer (for sequences, see “Primer Extension Assays”) and templates (for sequences, see “Primer Extension Assays”) were applied. Quenched samples were analyzed on a 12% denaturing PAGE followed by phosphorimaging. For kinetic analysis, experimental data were fit by nonlinear regression using the program GraphPad Prism 4. The data were fit to a single exponential equation: (conversion) = A × (1 − exp(−k_obs t)). The observed catalytic rates (k_obs) were then plotted against the dNTP concentrations used, and the data were fit to a hyperbolic equation to determine the K_d of the incoming nucleotide. The incorporation efficiency is given by k_pol/K_d.

RESULTS

Overall Structure of KlenTaq Complexes in Presence of an Abasic Site

All KlenTaq crystals grew in the same space group and very similar cell parameters (supplemental Table S1) as those reported earlier (19, 20, 35, 42–45). Thus, significant differences in crystal packing forces acting on the active site should be negligible. Nevertheless, parts of the finger domain differ in their mobility (as measured by temperature factors) between the different structures. This is a consequence of the structural differences, which were induced in the active site assembly by the primer/template complexes and respective nucleotides that were investigated. In the following, the active sites for the various complexes are shown and described in detail.

Structure of KlenTaq in Binary Complex with DNA Duplex Containing an Abasic Site (KlenTaq_F-binary)

To address the issue, if the overall conformation of the enzyme-DNA complex is affected by an abasic site prior to nucleotide binding, we crystallized KlenTaq bound to primer/template duplex containing an abasic site following a similar strategy as reported recently (19, 20, 35, 43, 46). The structure was solved by difference Fourier techniques at a resolution of 1.9 Å (supplemental Table S1). KlenTaq_F-binary is very similar to one reported for KlenTaq in a binary complex bound to undamaged DNA duplex (PDB 4KTQ; KlenTaq_binary) (35) as reflected by an r.m.s.d. for C_α atoms of 0.60 Å (Fig. 1). However, remarkable structural changes were observed for the single-stranded 5′-overhang of the DNA template, which is rotated around the helical axis. Thereby, the lesion is flipped out of a developing DNA duplex (Fig. 1A). This conformation is stabilized by stacking of the 5′-upstream nucleobases of the template strand on the top of the primer/template duplex (Fig. 1, B and C) and a distinct hydrogen-bonding network of amino acid residues Glu-615 and Tyr-671 with the DNA template (Fig. 1D). In contrast, in KlenTaq_binary, the backbone rotates 5′ to the DNA duplex, redirecting the remainder of the single-stranded template out of the DNA polymerase active site (Fig. 1E).

FIGURE 1.

Structure of KlenTaq_F-binary (blue). A, template stacking assembly of A_n+2 and A_n+3 in KlenTaq_F-binary. The abasic site analog F_n+1 is located extrahelically. B, the stick and surface depiction highlights the template stacking arrangement. C, top view of the primer/template stacking arrangement. D, hydrogen-bonding network of the amino acid side chains Tyr-671 and Glu-615 with the template strand. E, same arrangement as in A for the superimposed structures of KlenTaq_F-binary (blue) and KlenTaq_binary (brown).

KlenTaq with ddGTP Opposite an Abasic Site (KlenTaq_F-G-I, KlenTaq_F-G-II)

To gain insights into the structural basis for the preference of adenosine over guanosine of KlenTaq in bypassing an abasic site, we crystallized KlenTaq as a ternary complex bound to a primer/template duplex and ddGTP opposite the abasic site. Two structures were reproducibly found in several crystallization trials and solved by difference Fourier techniques at resolutions of 2.3–2.4 Å (KlenTaq_F-G-I and KlenTaq_F-G-II) (supplemental Table S1). The obtained structures show conformational heterogeneity in the active site region. In particular, we found two different orientations of the incoming ddGTP (Fig. 2, A and B, and supplemental Fig. S2, A and B). However, the overall conformation of the two ddGTP-trapped structures is very similar to the ternary complex of KlenTaq harboring ddATP opposite an abasic site (PDB 3LWL; KlenTaq_F-A; r.m.s.d. for Cα of 0.42 Å) (19), showing the same remarkable structural changes as compared with the undamaged case (KlenTaq bound to an undamaged primer/template duplex processing a ddGTP; PDB 1QSS; KlenTaq_C-G) (42). Thus, the conformation of the O helix leaves the active site more open, similar to the one in KlenTaq_F-A, and somehow between the open (PDB 2KTQ; KlenTaq_open) (35) and closed (KlenTaq_C-G) conformations in the reported ternary complexes (Fig. 2E). Like in KlenTaq_F-A, we found that the abasic site is intrahelically located and that Tyr-671 is positioned opposite the incoming ddGTP at the place that is usually occupied by the templating nucleobase (Fig. 2A). However, remarkable differences between KlenTaq_F-A and KlenTaq_F-G-I appear in the positioning and interaction pattern of the incoming triphosphates (Fig. 2A and supplemental Fig. S3, A and B). In KlenTaq_F-G-I, Tyr-671 stabilizes ddGTP via hydrogen bonds between the Tyr-671 hydroxyl group and the N1 of guanine, whereas in KlenTaq_F-A, the N3 of adenine interacts with Tyr-671 (Fig. 2A and supplemental Fig. S3A). Furthermore, an important stabilizing factor of the incoming nucleotide is Arg-587. In KlenTaq_F-A, Arg-587 forms a hydrogen bond to the N7 of adenine (supplemental Fig. S3A). However, in KlenTaq_F-G-I, ddGTP is stabilized by cation-π interaction with Arg-587. Because this type of interaction strongly depends on the distribution of the electron density in the aromatic ring system, there is in the case of purines a clear preference for arginine to position itself below or above the imidazole five-atom ring system (47). Indeed, this is observed in KlenTaq_F-G-I for Arg-587 (Fig. 2, A and C). The binding arrangement of ddGTP causes a misalignment of the α-phosphate, resulting in a large distance of the α-phosphate to the 3′-primer terminus (8.7 Å). Hence, KlenTaq_F-G-I might represent an initial binding event of the incoming triphosphate (Fig. 2A). In comparison, the guanine base in KlenTaq_F-G-II is rotated around the N-glycosidic bond, and the N7 of guanine points to the hydroxyl group of Tyr-671 (Fig. 2, B and D, and supplemental Fig. S2B). Thereby, the sugar moiety is realigned, resulting in a shortened distance between the α-phosphate and 3′-primer terminus of about 6.9 Å. However, the reorientation of the nucleobase positions the six-membered heterocycle above Arg-587 (Fig. 2, B and D), a less favored arrangement observed only in rare cases (47). In summary, the distance between 3′-OH and the α-phosphate increases in the following direction KlenTaq_F-A < KlenTaq_F-G-II < KlenTaq_F-G-I that are all longer than the one observed in the canonical case (Fig. 2G).

FIGURE 2.

Structure of KlenTaq_F-G-I (black) and KlenTaq_F-G-II (gray). A, stabilization network of ddGTP in KlenTaq_F-G-I. The amino acid side chains Arg-587 and Tyr-671 are labeled. Gray, red, and black dashed lines indicate hydrogen-bonding interactions, cation-π interaction, and distance (Å), respectively. B, the same as in A for KlenTaq_F-G-II. C, top view of the nascent base pair opposite F (KlenTaq_F-G-I). In the front, the incoming ddGTP opposite Tyr-671 is depicted. The first nucleobase pair of the primer/template terminus is shown as transparent. D, same view as in C for KlenTaq_F-G-II. E, the O helices from the superimposition of KlenTaq_F-G-II (gray), KlenTaq_F-A (purple), KlenTaq_C-G (green), and KlenTaq_open (gold) are highlighted. F, comparison of the nascent base pairs of KlenTaq_F-G-II (gray), KlenTaq_F-A (purple), and KlenTaq_C-G (green). G, the incoming ddNTPs of KlenTaq_F-G-II (gray), KlenTaq_F-A (purple), and KlenTaq_C-G (green) and the respective 3′-primer terminus are shown. The arrow indicates the displacement of the α-phosphate regarding to the 3′-primer terminus.

Pyrimidines Opposite an Abasic Site

Pyrimidine nucleotides are incorporated opposite an abasic site by KlenTaq with poor efficiencies (19). Furthermore, we demonstrated that the purine preference could be switched to pyrimidine preference by a single site-directed mutagenesis of Tyr-671 to Trp (19). This suggests that the low incorporation efficiencies of pyrimidines relies on a specific interaction with this amino acid residue. To get structural insights, we crystallized KlenTaq in the presence of a primer/template duplex and ddTTP. The structure was solved by difference Fourier techniques at a resolution of 1.8 Å (KlenTaq_F-T) (supplemental Table S1). Although kinetic studies elucidated that thymidine incorporation opposite an abasic site is unfavorable, we observed a bound thymidine triphosphate in the active site (Fig. 3A and supplemental Fig. S2C). Once more we obtained a semiclosed enzyme conformation as can be seen from the superimposition of this structure with KlenTaq_A-T (PDB 1QTM; processing a canonical base pair with an incoming ddTTP) (42), KlenTaq_F-A, and KlenTaq_open (Fig. 3B). The O helix does not pack against the nascent base pair, allowing Tyr-671 to position itself between the incoming ddTTP and the abasic site as is observed for Tyr-671 in KlenTaq_F-A (Fig. 3A and supplemental Fig. S3A). The ddTTP is stabilized via H-bonding between the N1 of thymidine and the hydroxyl group of Tyr-671 as well as π-stacking interactions to Phe-667. In contrast to KlenTaq_F-A in KlenTaq_F-T, Arg-587 is released from the stabilization pattern of the incoming triphosphate, as observed for the purine-trapped structures, while still interacting with the phosphate backbone of the 3′-primer terminus. The resulting thymidine tyrosine pair is accommodated in the active site in a way that results in a misalignment of the α-phosphate and thereby positions the sugar moiety on top of the 3′-primer terminus (Fig. 3C).

FIGURE 3.

Structure of KlenTaq_F-T (yellow). A, stabilization network of ddTTP. The amino acid side chains Arg-587, Phe-667, and Tyr-671 are labeled. Gray and black dashed lines indicate hydrogen-bonding interactions and distance (Å), respectively. B, the O helices from the superimposition of KlenTaq_F-T (yellow), KlenTaq_F-A (purple), KlenTaq_A-T (magenta), and KlenTaq_binary (gold) are highlighted. C, the incoming ddNTPs of KlenTaq_F-T (yellow), KlenTaq_F-A (purple), and KlenTaq_A-T (berry) and the respective 3′-primer terminus are shown. The arrow indicates the displacement of the α-phosphate regarding the 3′-primer terminus.

The same approach as described before was used to co-crystallize KlenTaq in the presence of a primer/template duplex and ddCTP (KlenTaq_F-binary-II). However, in all conducted approaches, only the formations of binary complexes of KlenTaq and the primer/template were obtained. These structures are very similar to the binary structure KlenTaq_F-binary (supplemental Fig. S4).

Nucleotide Analog dNITP Incorporation Opposite an Abasic Site

Base stacking capability was suggested to play a decisive role in influencing the incorporation efficiency opposite an abasic site (17, 31, 48–50). Intriguingly, nucleotides that bear nucleobase surrogates with strong stacking but lacking hydrogen-bonding capability such as dNITP (Fig. 5A) are incorporated opposite abasic sites with higher efficiency as their natural counterparts. Therefore, non-natural nucleotides such as dNITP were used to investigate abasic site bypass as a probe for stacking. To get insights into how dNITP is processed opposite an abasic site by KlenTaq, we performed soaking experiments with crystals of a binary complex of KlenTaq in the presence of an abasic site and dNITP. The structure was solved by difference Fourier techniques at a resolution of 1.9 Å (supplemental Table S1 and supplemental Fig. S2D). In the presence of dNITP, the enzyme is accumulated in a closed and productive complex, very similar to reported canonical cases (e.g. PDB 1QTM; KlenTaq_A-T; r.m.s.d. for Cα of 0.53 Å) and in contrast to KlenTaq_F-A (Fig. 4, A and B). In disparity with the structures of KlenTaq complexed with natural nucleotides opposite an abasic site, Tyr-671 is released from its stacking interaction to the template strand and provides space for the incoming dNITP (Fig. 4A). The hydrophobic nucleotide analog perfectly stacks on the developing DNA duplex, resulting in a proper alignment of the α-phosphate and recruitment of two catalytically essential magnesium ions (Fig. 4C and supplemental Fig. S5).

FIGURE 5.

Nucleotide incorporation opposite an abasic site analog F. A, structure of nucleotide analog dNITP. B, partial primer/template sequence used in primer extension experiments. C, single nucleotide (nt) incorporation of KlenTaq wild type opposite F for 1, 10, or 60 min, respectively. The respective dNTP is indicated. Transient kinetic analysis of nucleotide incorporation opposite T/F by KlenTaq wild type was performed. *, kinetic data were first reported in Ref. 19. D, single nucleotide incorporation of KlenTaq Y671W mutant opposite F for 10, 60, or 120 min, respectively. The respective dNTP is indicated. Transient kinetic analysis was performed as in C for KlenTaq Y671W.

FIGURE 4.

Structure of KlenTaq_F-NI (orange). A, stabilization network of dNITP. The amino acid side chains Arg-587, Phe-667, and Tyr-671 are labeled. Black dashed lines indicate distance (Å). B, the O helices from the superimposition of KlenTaq_F-NI (cyan), KlenTaq_F-A (purple), KlenTaq_C-G (green), and KlenTaq_binary (gold) are highlighted. C, the incoming ddNTPs of KlenTaq_F-NI (cyan) and KlenTaq_C-G (green), the respective Mg²⁺ ions, and the respective catalytically relevant residues Asp-610, Tyr-611, and Asp-785 are shown.

Furthermore, kinetics show that dNIMP is incorporated more efficiently than any other natural nucleotide by KlenTaq opposite an abasic site resulting in the following order of incorporation efficiencies (k_pol/K_d): NI ≫ A > G ≫ T > C (Fig. 5C and supplemental Fig. S6A). Due to the decrease in K_d and simultaneous increase in k_pol, dNIMP shows a 22-fold increase in incorporation efficiency as compared with dAMP. Interestingly, if KlenTaq mutant Y671W is used, we obtained the following incorporation efficiency (k_pol/K_d) order: NI > C > T ≫ A > G (Fig. 5D and supplemental Fig. S6B). Therefore, one can assume that Tyr-671 does not influence the incorporation of dNIMP, whereas it clearly directs the incorporation of the natural nucleotides.

DISCUSSION

General Aspects

Although the bypass of a non-instructive lesion by DNA polymerases is extensively studied, the mechanisms are not fully understood yet. Besides initial binding of the nucleotide and the chemical reaction, noncovalent intermediates with energetic minima are crucial for the reaction pathway. These intermediates are believed to be responsible for preselection of nucleotides and represent kinetic checkpoints explaining the overall accuracy and fidelity of these enzymes (51–53). The present structural study provides further insights into how KlenTaq DNA polymerase, a member of sequence family A DNA polymerases, is able to bypass abasic sites (Table 1). Independent of the incoming natural nucleotide opposite the abasic site, the obtained ternary complexes show two noticeable alterations as compared with the canonical cases. Firstly, it is always observed that Tyr-671 is placed opposite the incoming nucleotides, and secondly, the enzyme adopts a semiclosed conformation. Such a conformation was also associated with mismatch incorporation (54). However, FRET studies of Klenow fragment DNA polymerase in the presence of mismatches indicate that along with the open form, the closed state is also partially occupied (55). Because crystallographic data only provide single snapshots, it is likely that in solution, there is also a distribution of the populations between the open and closed state even in the presence of an abasic site. Similarly to nucleotide discrimination between matched and mismatched nucleotides, DNA polymerases might also undergo several conformational changes as preselection steps before preceding the phosphoryl transfer in the presence of an abasic site. Therefore, the observed conformations might represent various local energy minima along the reaction coordinate. Along these lines, computer simulations of the fidelity of T7 DNA polymerase support the hypothesis that the formation of mismatches may also occur with high energy barriers from a partially open protein conformation (56). In general, there is an ongoing discussion whether the chemical step demands similar conformational states processing a correct or incorrect nucleotide (52–54, 57). However, the incoming nucleotides opposite an abasic site in the herein depicted structures are not properly aligned for the phosphoryl transfer. This suggests either that a rearrangement to the closed conformation has to occur or that the incorporation step proceeds from the semiclosed conformation after appropriate conformational changes of the nucleotide. Both scenarios support the observed overall poor incorporation efficiencies opposite abasic sites (19).

TABLE 1.

Summary of KlenTaq structures bound to an abasic site analog F

Structure of Binary Complex Containing Abasic Site and KlenTaq

In the binary complex of KlenTaq bound to an abasic site, we observed the 5′-upstream template strand stacking on the primer/template duplex. Nucleotide binding forces the release of the template strand from its stacking arrangement and triggers a remarkable rearrangement of the bound DNA template rather than a reorganization of the enzyme (supplemental Fig. S7). Furthermore, the ability to resolve the stacking arrangement might be dependent on the intrinsic properties of the respective nucleotide, such as the stacking and solvation ability of the nucleobase. It remains to be elucidated whether this arrangement is sequence-specific.

Nucleotide Selection Opposite Abasic Sites of KlenTaq, Adenosines versus Guanosines

The structural data offer valuable clues how nucleotide selection is performed opposite a non-instructive lesion and moreover support the previous kinetic analysis of abasic site bypass by KlenTaq (19). In the case of an incoming guanosine, we obtained several crystal structures showing the incoming triphosphate in two different orientations. This heterogeneity was not observed in the presence of the other nucleotides. In KlenTaq_F-G-I and KlenTaq_F-G-II, Tyr-671 is filling the space of the vacant nucleobase of the abasic site and is located opposite the incoming ddGTP and thereby roughly mimics the geometry of a nascent nucleobase pair. Of note, such a selection of purines mediated by Tyr-671 is also observed for KlenTaq_F-A. However, in KlenTaq_F-G-I and KlenTaq_F-G-II, the interaction with Tyr-671 causes misalignment of the α-phosphate resulting in enlarged distances to the 3′-primer terminus (Figs. 2F and 6, A and B). The different steric constraints of the purine structures account for the preference of adenosine over guanosine. In detail, the exocyclic C₂-NH₂ group of guanine prevents the same arrangement of the nucleotide, as was found for adenosine (Table 1). Due to the steric restriction of the active site, the assembly of the sterically more demanding guanosine results in an alignment with an enlarged distance between the α-phosphate and the 3′-primer terminus in comparison with adenosine. Interestingly, the complex showing the stronger cation-π interaction between the Arg-587 and the five-membered ring (KlenTaq_F-G-I) results in an arrangement of the ddGTP obviating a possible attack at the α-phosphate. The closer orientation of the nucleotide toward the 3′-primer terminus suggests that the structure of KlenTaq_F-G-II forming a less stable cation-π interaction (47) may represent one step ahead on the reaction coordinate in comparison with KlenTaq_F-G-I. In conclusion, we find that the distance between the α-phosphates and the 3′-primer terminus (see Table 1, 8.7–6.9 Å for guanosines versus 5.9 Å for adenosine) correlates well with the measured incorporation efficiencies. This indicates that active misalignment of the incoming guanosines governs their unfavorable processing by KlenTaq in comparison with adenosines.

FIGURE 6.

Geometry fit to active site. A, active site assembly in the presence of an abasic site analog F. The nascent base pair of Tyr-671 and ddGTP (KlenTaq_F-G-I) is depicted. The surface of the surrounding active site residues is shown in gray. B, same as in A for KlenTaq_F-G-II. C, active site assembly in the case of an undamaged template showing the nascent base pair of dC and ddGTP (KlenTaq_C-G). D, same as in A for KlenTaq_F-T. E, same as in A for KlenTaq_F-A. F, same as in C for KlenTaq_F-NI showing the nascent base pair of F and dNITP. A–F, the arrows indicate the positioning of the α-phosphate and the 3′-primer terminus, respectively.

Nucleotide Selection Opposite Abasic Sites by KlenTaq, Purines versus Pyrimidines

To shine light on the incorporation mechanisms of the less favored pyrimidines, we investigated a structure bearing an incoming ddTTP opposite an abasic site. The DNA polymerase interacts with the incoming nucleotide by hydrogen bonding of Tyr-671 with the nucleobase. This results in a nucleoside triphosphate conformation where the sugar moiety is positioned above the 3′-primer terminus instead of the α-phosphate (Figs. 3C and 6D). In this scenario, all components of the active site are assembled and organized in a topological and geometrical arrangement that does not allow the enzyme to proceed with the chemical step, explaining the very low incorporation efficiency. In contrast to purines, ddTTP shows low B-factors (supplemental Table S1), which indicates that it is well stabilized opposite Tyr-671. Thus, the tightly bound thymidine probably stalls the DNA polymerase in a “nonproductive” complex by active misalignment, whereas the purine nucleobases enhance the geometric fit to the active site, resulting in an arrangement with shortened distances between the α-phosphate and the 3′-primer terminus as compared with the pyrimidine base (Figs. 2G and 6, B and E).

Nucleobase Stacking and Abasic Site Bypass

Superior stacking and solvation properties of adenine have been discussed as a driving force behind adenine selection opposite abasic sites (17, 29–31). The nucleotide dNITP contains a nucleobase surrogate lacking hydrogen-bonding capability but increased stacking ability in comparison with their structural congener purines. These properties build the basis for the motivation to employ the nucleotide in abasic site bypass (17, 31, 48–50). We find, in contrast to the results obtained with nucleotides containing the natural nucleobases, that binding of the nucleotide analog dNITP readily allows the enzyme to change its active site conformation in a closed, productive complex that is similar to those found for undamaged DNA substrates. The accumulation of the enzyme in a productive conformation is clearly supported by the kinetic data, which show a significant increase in incorporation efficiency of dNIMP as compared with dAMP opposite the abasic site. Furthermore, the circumstance that dNITP is processed with higher efficiency than the favored nucleotides of either KlenTaq wild type or KlenTaq mutant Y671W shows that Tyr-671 apparently lost its selection criteria because both the dNITP and the Tyr-671 residues are not compatible with the geometric constraints of the active site. Our results suggest that due to its increased stacking ability, the nucleobase surrogate imposes active site conformations that differ significantly from those induced by the natural nucleobases. Consequently, results that are obtained by usage of dNITP instead of the natural nucleotides should be interpreted with caution toward their significance in abasic site bypass by natural nucleotides. It is noteworthy that dNITP is known as a chain terminator (59, 60) as well as universal base (61) and has shown no preference for incorporation opposite one of the four natural nucleobases. Again, enforcement of aberrant enzyme and DNA conformations due to the strong stacking ability of dNITP might be the origin of the observed properties.

Taken together, the present structural study of KlenTaq in complex with different nucleoside triphosphates opposite an abasic site reveals that natural nucleotides are bound, due to interaction with Tyr-671 and the geometric confines of the active site, in positions that require additional conformational changes for proper alignment to facilitate the chemical step. The degree of misalignment, measured as distance of the α-phosphate to the 3′-primer terminus, parallels the order of incorporation efficiency of KlenTaq opposite an abasic site. These conformational changes might be accompanied by barriers of varied energy explaining the observed decline in nucleotide incorporation efficiency from A and G to T and C as well as the strong block of the abasic site lesion. The resulting pausing of DNA synthesis might allow the DNA polymerase to be replaced by repair systems that use the sister strand for error-free repair.

The human DNA polymerase θ and γ are members of the sequence family A as KlenTaq. Interestingly, the human DNA polymerase θ also obeys the A-rule (18), and its intrinsically error-prone abasic site bypass might contribute to somatic hypermutation of Ig genes (62). Furthermore, the role of Tyr-671 in nucleotide selection opposite an abasic site is highlighted. Tyr-671 is highly conserved in family A DNA polymerases (19, 34) and is known to be involved in the discrimination process between canonical and noncanonical base pairs (63–65). Along these lines, Leob and co-workers (63) demonstrated that Tyr-671 is essential for maintaining Taq DNA polymerase I activity. Mutations at that position are hardly tolerated and compromise activity with the exception of phenylalanine. Nevertheless, mutations of the homolog residue Tyr-766 of the Klenow fragment from E. coli DNA polymerase I to serine or alanine result in significant decrease in fidelity (64). Interestingly, mutation of the corresponding Tyr-955 in human DNA polymerase γ has been attributed to progressive external ophthalmoplegia, stressing its importance in accurate function of DNA polymerases (58, 66).