Klenow Fragment Discriminates Against the Incorporation of the Hyperoxidized dGTP Lesion Spiroiminodihydantoin into DNA

Abstract

Defining the biological consequences of oxidative DNA damage remains an important and ongoing area of investigation. At the foundation of understanding the repercussions of such damage is a molecular-level description of the action of DNA-processing enzymes, such as polymerases. In this work we focus on a secondary, or hyperoxidized, oxidative lesion of dG which is formed by oxidation of the primary oxidative lesion 2′-deoxy-8-oxo-7,8-dihydroguanosine (8-oxodG). In particular, we examine incorporation into DNA of the diastereomers of the hyperoxidized guanosine triphosphate lesion spiroiminodihydantoin-2′-deoxynucleoside-5′-triphosphate (dSpTP). Using kinetic parameters we describe the ability of the Klenow fragment of E. coli DNA polymerase I lacking 3'→5' exonuclease activity (KF⁻) to utilize (S)-dSpTP and (R)-dSpTP as building blocks during replication. We find that both diastereomers act as covert lesions, similar to a Trojan horse: KF⁻ incorporates the lesion dNTP opposite dC, which is a non-mutagenic event, however, during the subsequent replication it is known that dSp is nearly 100% mutagenic. Nevertheless, using k_pol/K_d to define the nucleotide incorporation specificity we find that the extent of oxidation of the dGTP-derived lesion correlates with its ability to be incorporated into DNA. KF⁻ has the highest specificity for incorporation of dGTP opposite dC. The selection factors for incorporating 8-oxodGTP, (S)-dSpTP, and (R)-dSpTP are 1,700-, 64,000- and 850,000-fold lower respectively. Thus, KF⁻ is rigorous in its discrimination against incorporation of the hyperoxidized lesion and these results suggest that the specificity of cellular polymerases provides an effective mechanism to avoid incorporating dSpTP lesions into DNA from the nucleotide pool.

Graphical Abstract

INTRODUCTION

Genomic DNA is exposed to a variety of endogenous and exogenous reactive oxygen species (ROS).^{1, 2} Due to its low redox potential,³ dG is the major target for ROS and a prototypic and well-studied oxidation product of dG is 2′-deoxy-8-oxo-7,8-dihydroguanosine (8-oxodG) (Figure 1).^4–6 Indeed, in vivo 8-oxodG is found at steady-state levels of 0.3–4.2 per 10⁶ dG.⁷ Studies performed in vitro have shown that 8-oxodG has a high miscoding frequency because it can base pair with both dC and dA, with the latter leading to G→T transversion mutations.^{8, 9} Despite its high miscoding frequency in vitro, 8-oxodG has a low mutation frequency in vivo yielding less than 10% G→T transversions.^9–11 This low mutagenicity in vivo derives from the presence of an extensive repair system that counters the genetic effects of 8-oxodG.¹²

Figure 1.

Structures of dG, 8-oxodG, (S)-dSp, and (R)-dSp. The A and B rings of dSp are indicated.

In addition to the mutagenic potential of 8-oxodG, the lesion itself is chemically labile towards further oxidation.¹³ Several hyperoxidized lesions have been identified, including 2′-deoxy-spiroiminodihydantoin (dSp), which exists as a pair of diastereomers (Figure 1).^14–20 Furthermore, the dSp lesion has been detected in genomic DNA from bacteria and mammalian cells. Hailer and co-workers detected the dSp lesion in E. coli treated with chromate,²¹ and more recently, Mangerich and co-workers identified dSp in Helicobacter hepaticus-infected mice, which develop inflammation-mediated carcinogenesis at levels of 1–5 per 10⁸ nucleotides in genomic DNA.²² Several DNA glycosylases have been found to excise the dSp lesion from DNA in vitro and initiate the base excision repair (BER) process. The bacterial glycosylases FPG, Nei, and Nth remove dSp with differing preferences for the opposing base pair partner.^23–25 The dSp lesion is not a substrate for the human enzyme OGG1 which is responsible for removing 8-oxodG, but the lesion is a substrate for yeast OGG1 and OGG2.²⁶ Finally, eukaryotic^27–30 and viral³¹ NEIL1 glycosylases can excise dSp from double-stranded DNA with some preference for the opposing base pair partner, and can also remove dSp from single-stranded DNA, bubble, and bulge structures. More recent results suggest that dSp can also be removed from DNA by nucleotide excision repair.³² If dSp is not removed prior to replication, in contrast to the mildly mutagenic 8-oxodG, the hyperoxidized dSp lesion is nearly 100% mutagenic in vitro^{33, 34} and in vivo,^35–37 causing both G→T and G→C transversions.

As the primary and secondary oxidation products of dG, the mutagenicity of both 8-oxodG and dSp have been well-studied when the lesion is present in DNA. However, in order to understand the full mutagenic potential of a lesion we must also understand the extent to which polymerases utilize the oxidized lesions when they are part of the nucleotide pool. Notably, it has been shown that a nucleobase is more susceptible to oxidation when it is part of a free nucleotide than when incorporated in duplex DNA. For instance, dATP in the nucleotide pool is 67-fold more easily oxidized than dA in duplex DNA.³⁸ Although the preference is less pronounced for dGTP, it is still 9-fold easier to oxidize in the nucleotide pool relative to dG in duplex DNA.³⁸ Therefore, it is important to consider the nucleotide pool as a source of oxidative damage. Indeed, it is known that 8-oxodGTP can be incorporated into DNA by several bacterial and mammalian polymerases.^39–43 However, it remains unknown if polymerases can utilize dSpTP as a building block during DNA replication.

Here we determine the extent to which each diastereomer of dSpTP is a substrate for the Klenow Fragment of E. coli DNA polymerase I lacking 3'→5' exonuclease activity (KF⁻). We establish that KF⁻ can incorporate each lesion into DNA, with a preference for (S)-dSpTP over (R)-dSpTP. We also define the kinetic parameters for incorporation of each lesion and discuss the incorporation efficiency relative to canonical dGTP and the primary oxidation product 8-oxodGTP.

EXPERIMENTAL PROCEDURE

Materials

Klenow Fragment 3'→5' exo⁻ (KF⁻), calf intestinal alkaline phosphatase, and T4 polynucleotide kinase were purchased from New England Biolabs (Ipswich, MA). The canonical dNTPs were from Thermo Scientific (Waltham, MA) and 8-oxodGTP was from TriLink BioTechnologies (San Diego, CA).

Oligonucleotide Synthesis and Purification

Oligonucleotides were synthesized and purified according to methods published previously by our laboratory.⁴⁴ Oligonucleotide concentrations were determined at 90 °C using molar extinction coefficients as determined by nearest-neighbor theory for single-stranded DNA^{45, 46} on a Beckman Coulter DU800 UV-visible spectrophotometer equipped with a peltier thermoelectric device. The identity of the oligonucleotides was confirmed by ESI-MS.

Synthesis, Purification, and Absolute Configuration of Spiroiminodihydantoin-2′-deoxynucleoside-5′-triphosphate

The hyperoxidized nucleotide triphosphate dSpTP was synthesized according to literature procedures from dGTP.⁴⁷ The resulting solution was dried in vacuo, resuspended in a small amount of deionized water, and purified by ion-pairing HPLC on a Varian Microsorb C18 column (250 × 4.6 mm) (Agilent Technologies, Inc., Santa Clara, CA) using acetonitrile (mobile phase A) and 10 mM tetrabutylammonium hydroxide, 10 mM monopotassium phosphate, 0.25% methanol, pH = 7.0 (mobile phase B) as the mobile phases. As the gradient mobile phase A increased from 5 to 65% over 40 min at a flow rate of 1 mL/min. The dSpTP diastereomers were collected as a mixture and dried in vacuo; Figure S1A demonstrates that the mixture of dSpTP diastereomers was successfully separated from the dGTP starting material. The sample was resuspended in deionized water and the diastereomers were separated on the same HPLC column as above using mobile phase B (same as above) and mobile phase C (2.8 mM tetrabutylammonium hydroxide, 100 mM monopotassium phosphate, 30% methanol, pH = 5.5); as the gradient mobile phase C was increased from 50 to 70% over 40 min at flow rate 0.9 mL/min (Figure S1B). Removal of the ion-pairing reagent from the separated diastereomers was achieved by multiple rounds of purification on a Dionex DNAPac PA-100 anion exchange HPLC column (250 × 4.6 mm) (Thermo Scientific, Waltham, MA) using mobile phase D (25 mM ammonium bicarbonate, pH=8.5) and mobile phase E (800 mM ammonium bicarbonate, pH=8.5); as the gradient mobile phase E was increased from 0 to 50% over 15 min at a flow rate of 1 mL/min. Multiple rounds of anion exchange HPLC were performed to the point that the experimentally determined rate of nucleotide incorporation did not change. Notably, the diastereomers of dSpTP elute in the same order from the C18 and anion exchange HPLC columns. The volatile ammonium bicarbonate was removed by repetitive resuspension in deionized water and drying in vacuo. For all the HPLC analyses described above, absorbance was monitored at both 225 nm and 260 nm. The purified dSpTP diastereomers were characterized along with dGTP by anion exchange HPLC using mobile phase D (25 mM ammonium bicarbonate, pH=8.5) and mobile phase E (800 mM ammonium bicarbonate, pH=8.5); as the gradient mobile phase E was increased from 0 to 50% over 60 min at a flow rate of 1 mL/min (Figure S1C). The concentration of each diastereomer of dSpTP was determined by UV absorbance based on the extinction coefficient ε_230nm = 4,900 M⁻¹ cm⁻¹.¹⁴ The molecular weights were verified by ESI-MS (Figure S2A, B).

The absolute configuration of the diastereomers of dSpTP was then assigned based on literature reports.^{48, 49} Each diastereomer of dSpTP (0.2 µmol) was resuspended in 1× Cutsmart buffer (20 mM Tris acetate, 50 mM potassium acetate, 10 mM magnesium acetate, 0.1 mg/mL BSA, pH 7.9) (New England Biolabs, Ipswich, MA) and treated with 20 U calf intestinal alkaline phosphatase at 37 °C for 40 h. The phosphatase was removed by passing the solution through a 3,000 MWCO centrifugal filter (Sartorius Corp., Bohemia, NY). The resulting dSp nucleosides were then analyzed on a Thermo Scientific HyperCarb column (150 × 4.6 mm) (Thermo Scientific, Waltham, MA) with an isocratic flow of deionized water containing 0.1 % acetic acid at 1 mL/min; the absorbance was monitored at 240 nm (Figure S1D). The dSp nucleosides were then analyzed by ESI-MS (Figure S2C, D). Consistent with literature reports,⁴⁹ removal of the triphosphate group switches the order of elution on the HyperCarb column relative to the C18 and anion exchange columns. Based on literature precedent,⁴⁹ we assigned the absolute configurations of (S)-dSpTP (triphosphate elutes first from C18 and anion exchange columns, nucleoside elutes second from HyperCarb column) and (R)-dSpTP (triphosphate elutes second from C18 and anion exchange columns, nucleoside elutes first from HyperCarb column).

Preparation of DNA Primer/Template Assemblies

The primer strand was radiolabeled at the 5′-end with ³²P using T4 polynucleotide kinase and annealed to the template in a 1:1.5 ratio of primer:template in a buffer containing 10 mM Tris-HCl, 50 mM NaCl, 10 mM MgCl₂, and 1 mM DTT, pH 7.9 by heating the solution at 90 °C for 5 min and allowing it to slowly return to room temperature over approximately 2.5 h.

Qualitative Single Nucleotide Triphosphate Incorporation Reactions

Primer/template DNA was incubated with KF⁻ for 5 min on ice where the final sample contained 5 nM primer/template DNA, 25 nM KF⁻, 10 mM Tris-HCl, 50 mM NaCl, 10 mM MgCl₂, 1 mM DTT and 0.04 mg/mL BSA, pH 7.9. The incorporation reaction was initiated by addition of nucleotide triphosphate to yield a final concentration of 20 nM dGTP, 5 µM 8-oxodGTP, 100 µM (S)-dSpTP, or 100 µM (R)-dSpTP and the samples were incubated for 15 min at room temperature. The reactions were quenched by addition of gel loading buffer (80% formamide, 1 mg/mL xylene cyanol, 1 mg/ml bromophenol blue, and 10 mM EDTA), electrophoresed through a 16% denaturing PAGE gel, and imaged by phosphorimagery.

Primer Extension after the Incorporation of Sp Nucleotide Triphosphate

To examine whether primer extension will continue after the incorporation of the dSpTP nucleotide triphosphates, DNA with template X = C was used. In addition to 100 µM of dSpTP, 20 nM of both dATP and dCTP were added to the reaction while the rest of conditions are the same as described above for the qualitative single nucleotide incorporation experiment.

Single-Turnover Kinetics Reactions

Reactions were performed at 25 °C using a Rapid Quench Flow apparatus (RQF-3) from KinTek Corporation (Austin, TX). A solution consisting of 10 nM primer/template DNA, 50 nM KF⁻, 10 mM Tris-HCl, 50 mM NaCl, 10 mM MgCl₂, 1 mM DTT, 0.08 mg/mL BSA, pH 7.9 was incubated for 5 min on ice. The incorporation reaction was initiated by adding various concentrations of each nucleotide triphosphate (listed in each figure legend) in buffer containing 10 mM Tris-HCl, 50 mM NaCl, 10 mM MgCl₂, and 1 mM DTT, pH 7.9. For each reaction in the RQF, 15 µL DNA·KF⁻ and 15 µL dNTP were loaded into the reaction loops, rapidly mixed and incubated for the designated time before being quenched by 85 µL of 100 mM EDTA. The final reaction concentrations were 5 nM primer/template DNA, 25 nM KF⁻, dNTP (various concentrations), 10 mM Tris-HCl, 50 mM NaCl, 10 mM MgCl₂, 1 mM DTT, 0.04 mg/mL BSA, pH 7.9. Gel loading buffer was added and the samples were analyzed by 16% denaturing PAGE gel and imaged by phosphorimagery.

Data Analysis

KinTek Explorer software (KinTek Corporation, Austin, TX) was used to globally fit the kinetic data.^{50, 51} The data were iteratively fit until a best fit was obtained according to a specific enzyme model. Because binding of a canonical nucleotide triphosphate to a polymerase is diffusion limited, k₁ was fixed at 150 µM⁻¹ s⁻¹ for dGTP.⁵² However, for the oxidized and hyperoxidized nucleotide triphosphates 8-oxodGTP and dSpTP, k₁ was fixed at 10 µM⁻¹ s⁻¹ due to their reduced binding.⁵¹ The lower and upper limits associated with each kinetic parameter were defined using the FitSpace function of KinTek Explorer. The best fit curves and experimental data were exported from KinTek Explorer and plotted in KaleidaGraph (Synergy software, Reading, PA).

RESULTS AND DISCUSSION

Qualitative Analysis of Incorporation of Spiroiminodihydantoin Nucleotide Triphosphate by Klenow Fragment

In order to study incorporation of the hyperoxidized nucleotide triphosphate dSpTP into DNA, templates were designed with either X = dA, dC, dG, or dT at the site of incorporation (Figure 2A). The diastereomers of dSpTP were resolved and examined separately. Primer extension reactions were catalyzed by KF⁻ and were initiated by the addition of either dGTP, 8-oxodGTP, (S)-dSpTP, or (R)-dSpTP. The results are shown in Figure 2B. As expected, dGTP is incorporated exclusively opposite dC, and 8-oxodGTP is incorporated opposite both dC and dA.⁸ Both diastereomers of dSpTP are incorporated opposite dC, with more incorporation observed for (S)-dSpTP relative to (R)-dSpTP. Notably, in order to observe sufficient amounts of the extended primer product, the concentration of 8-oxodGTP was 250-fold greater than dGTP, and the concentration of the dSpTP diastereomers was 5,000-fold greater than dGTP. The experiments were also performed with dSpTP that was prepared using a different synthetic strategy (using 8-oxodGTP as the starting material⁴⁷ rather than dGTP), and the same results were obtained (data not shown). Obtaining the same results with these different preparations of dSpTP indicates that these polymerase incorporation results are not due to trace contaminants of the dGTP starting material that may remain after synthesis and purification, but rather to dSpTP itself. That the product observed in Figure 2A corresponds to incorporation of (S)- and (R)-dSpTP rather than potential contaminating dGTP starting material is further supported by the difference in the migration between the single nucleotide incorporation products when dGTP or our preparation of dSpTP are added to the primer (Figure S3).

Figure 2.

(A) Sequence of primer and template strands. Primer was ³²P-radiolabeled at 5'-end as indicated by the asterisk. (B) Primer extension on different template DNA (X = A, C, G or T) with either dGTP, 8-oxodGTP, (S)-dSpTP, or (R)-dSpTP. The five lanes on the left in panel B are the primer and the positive controls with correct nucleotide for each template DNA. (C) Primer extension after the incorporation of dSpTP. See Experimental Procedures for reaction conditions.

Having observed that both (S)-dSpTP and (R)-dSpTP can be incorporated into DNA from the nucleotide pool, we next examined whether the polymerase can extend beyond dSpTP or if incorporation of the hyperoxidized nucleotide would terminate replication. To avoid competition between dSpTP and dGTP, the dGTP was excluded from the reactions. Both dATP and dCTP were included and under these experimental conditions the full extension product would be a 27-mer. As shown in Figure 2C, KF⁻ can extend past both diastereomers of dSpTP to yield the 27-mer product, suggesting that if incorporated from the nucleotide pool the lesions are not absolute blocks to replication. It is noteworthy that these qualitative experiments were performed using an excess of polymerase, and that under conditions limiting in polymerase stalling at the lesion site could occur.

In previous work, it was shown that when dSp is located in the template strand, KF⁻ incorporates dATP and dGTP opposite the lesion both in vitro and in vivo.^33–37 In fact, the correct base dCTP is almost never incorporated opposite dSp, meaning that the lesion is nearly 100% mutagenic if formed in genomic DNA. Based on these prior observations, we expected the dSpTP diastereomers to be incorporated into templates that contain X = dA and dG. Consistent with this logic, 8-oxodG is paired with dA and dC regardless of whether the lesion is in the template or is incorporated from the nucleotide pool.^{8, 33, 40, 42} Our observation that KF⁻ only incorporates the dSpTP diastereomers opposite of dC in the template provides an intriguing example where a polymerase exhibits a different preference based on whether the lesion is in the template or in the nucleotide pool.

It is of note that because the hyperoxidized lesion is a dG-derived lesion, the incorporation of dSpTP opposite dC does not immediately represent a mutagenic event. However, in a subsequent round of replication the hyperoxidized lesion will serve as the template and in this context is known to be nearly 100% mutagenic. Thus, one could liken the dNTP form of the hyperoxidized lesion to a Trojan horse; the dNTP is incorporated into DNA by polymerase but will not exhibit mutagenicity until a G → C or G → T mutation occurs in the next round of replication.

It is possible to rationalize how a polymerase could use different base pairing for a lesion, depending on whether the lesion serves as the template or is incorporated from the nucleotide pool. Prior to making a covalent linkage between the primer and incoming nucleotide, a polymerase probes the template nucleobase for favorable electrostatic interactions. It has been proposed that for KF⁻ the electrostatic profile is a significant determinant in nucleotide selectivity.⁵³ Interestingly, the B-ring of the lowest energy tautomer of the diastereomers of dSpTP⁴⁹ can form two of the same hydrogen bonds as the Watson-Crick face of dGTP (using a lone pair on the carbonyl and a hydrogen of the exocyclic amine) suggesting a means by which the hyperoxidized lesion pairs with dC. Other tautomers of dSpTP (i.e., the imino tautomer or the amino unconjugated tautomer⁴⁹), although energetically less favorable, can form three hydrogen bonds with dC. Furthermore, a nucleotide triphosphate in solution has more conformational freedom than a nucleobase in a template strand because it is not confined by base stacking, base pairing, and covalent attachment to the sugar-phosphate backbone. This conformational freedom could allow incoming dSpTP to adopt the anti conformation and present the B-ring to base pair with dC in the template. However, characterizations of dSp have shown that the A- and B-rings are perpendicular to one another.⁵⁴ Due to this structural constraint, when the lesion is located in the template, the A-ring of dSp could be presented for base pairing with an incoming nucleotide, which would explain the lack of pairing with dCTP when the lesion is in the template. Indeed, a recent crystal structure of (S)-dSp as the templating nucleobase in the active site of DNA polymerase β (pol β) reveals that the lesion is in the syn conformation and uses the A-ring as the base pairing face.⁵⁴

While the information is not yet available for Sp, crystal structures of primer/template-polymerase-dNTP ternary complexes where dSp is either the templating base or dSpTP is the incoming nucleotide would greatly inform our results. Previous ternary structures of pol β bound to DNA containing 8-oxodG as the templating base or as the incoming 8-oxodGTP revealed the molecular basis for different base pairing. When 8-oxodG served as the templating base, the lesion was observed while base pairing with an incoming dCTP. Both 8-oxodG and dCTP were in the anti conformation and Watson-Crick base pairing was observed.⁵⁵ In the ternary structure where 8-oxodGTP was the incoming nucleotide the templating base was dA.⁵⁶ While dA remained anti the 8-oxodGTP was in the syn conformation. The syn conformation was stabilized by Hoogsteen hydrogen-bonding with the templating dA. In other work with human DNA polymerase η, which replicates a templating 8-oxodG in a nearly error-free manner, ternary complexes revealed that an Arg residue from a finger domain prevents formation of the 8-oxodG:A mispair.⁵⁷ Similar structural studies would reveal the molecular basis for why a templating dSp base pairs with dATP and dGTP, whereas the nucleotide version dSpTP base pairs with a templating dC.

Kinetic Analyses of Nucleotide Triphosphate Incorporation

To evaluate the biological relevance of incorporation of dSpTP into DNA from the nucleotide pool, we performed experiments to compare the kinetic parameters k_pol and K_d of the hyperoxidized nucleotide to dGTP and 8-oxodGTP. To avoid the complications that arise from multiple turnovers of a processive polymerase, i.e., the rate being dominated by dissociation of the DNA·KF⁻ complex, the incorporation assays were performed under single-turnover conditions in which [KF⁻]>>[DNA]. Because single-turnover experiments using DNA polymerases occur rapidly and are too fast to be mixed and quenched manually, we used a rapid quench flow (RQF) apparatus for these experiments. Furthermore, because dGTP, 8-oxodGTP, (S)-dSpTP, and (R)-dSpTP share the ability to be incorporated opposite C, the kinetic analyses were performed using the template where X = dC.

Single nucleotide incorporation assays were performed using varying concentrations of dNTP in order to examine the concentration dependence of the rate of incorporation and amplitude of product formation. Under these conditions, the concentration dependence of the rate of incorporation reports k_pol and K_d. k_pol is the rate of the slowest step up to and including the phosphoryl transfer step in which the nucleotide is covalently attached to the primer, and K_d is the ground state nucleotide dissociation constant K_d. The DNA and KF⁻ were pre-incubated prior to initiating the reaction by addition of dNTP. In this manner, the measured K_d is for nucleotide binding to the KF⁻·DNA complex rather than binding of KF⁻ to the DNA. In this report, we use k_pol/K_d to define the nucleotide incorporation specificity.⁵⁸

Single-Turnover Kinetic Analysis of dGTP and 8-oxodGTP Incorporation

As the canonical nucleotide to incorporate opposite dC, dGTP was employed as a control for the incorporation experiments. A time course was performed at several dGTP concentrations and the data were fit globally to the model shown in Figure 3A. In this model K_d is defined by k₋₁/k₁, k₂ is k_pol, and k₃ is release of pyrophosphate. For KF⁻ incorporating a canonical dNTP, it is generally accepted that k_pol includes a rate limiting conformational change⁵⁹ that controls nucleotide specificity,⁶⁰ followed by phosphoryl transfer; however, other results suggest that the phosphoryl transfer step might limit the rate.⁶¹ It is also known that for incorporation of canonical nucleotides k₃ is faster than k₂.⁶² In our analysis, a K_d of 0.8 µM and k_pol of 41 s⁻¹ define a specificity constant k_pol/K_d (or k₂/K_d) of 51 µM⁻¹ s⁻¹ for the incorporation of dGTP opposite dC (Table 1). It is notable that the final amplitude of 21-mer product is the same at each concentration of dGTP and the slope of the rise converge with increasing concentration (Figure 3C). These observations show that k_pol is fast and there are no slower steps after incorporation that would cause incorporation to be reversible. Both of these conditions are accounted for in the model in Figure 3A.

Figure 3.

Kinetic analysis of dGTP and 8-oxodGTP incorporation by KF⁻. Enzyme model used for (A) dGTP and (B) 8-oxodGTP. (C) Global fitting of incorporation of dGTP opposite of template dC. The concentrations of dGTP are 0.5 µM (closed circles), 1 µM (open circles), 10 µM (closed squares), 50 µM (open squares), and 100 µM (closed triangles). (D) Global fitting of incorporation of 8-oxodGTP opposite of template dC. The concentrations of 8-oxodGTP are 1 µM (closed circles), 2 µM (open circles), 5 µM (closed squares), 10 µM (open squares), and 50 µM (closed triangles). The curves superimposed with the experimental data were generated by KinTek Explorer fitting.

Table 1.

Kinetic Parameters for the Incorporation of dNTPs Opposite Template dC.

Nucleotide	Kd (µM)	k2 (s−1)	k2/Kd (µM−1 s−1)	selection factora
dGTP	0.8	41	51	1
8-oxodGTP	15.2	0.45	0.030	1,700
(S)-dSpTP	1,700b	1.4b	8×10−4	64,000
(R)-dSpTP	7,000b	0.45b	6×10−5	850,000

^aSelection factor is the ratio of the specificity of incorporation (k₂/K_d) of dGTP relative to the dNTP of interest.

^bValues are the lower bound provided by KinTek Explorer global fitting.

In the case of 8-oxodGTP incorporation opposite of dC, the single-turnover plots at each nucleotide concentration are noticeably biphasic (Figure 3D). Additionally, the amplitude is dependent on nucleotide concentration suggesting that after incorporation there is a slow step that allows phosphoryl transfer to be reversible. Indeed, similar results were obtained for the incorporation of 8-oxodGTP opposite dC by the human mitochondrial DNA polymerase.⁴² In accordance with Hanes and co-workers,⁴² we accounted for the differences observed for incorporation of 8-oxodGTP relative to dGTP by fitting the data to the model shown in Figure 3B, which allows for reversibility of the phosphoryl transfer. A K_d of 15.2 µM describes nucleotide binding while the incorporation step has a forward rate (k₂) of 0.45 s⁻¹ and a reverse rate (k₋₂) of 0.08 s⁻¹. In contrast to canonical nucleotides, for modified nucleotides it is not known whether k₂ is limited by the conformational change, but it has been shown that when KF incorporates a mismatched nucleotide, the rate of phosphoryl transfer is not limited by the preceding conformational change.⁶³ Pyrophosphate release is indeed slower than incorporation and occurs at a rate of 0.09 s⁻¹, which allows k₂ to be reversible. Therefore, unlike for dGTP, k_pol ≠ k₂. When taking into account the slow pyrophosphate release, the k_pol/K_d value (0.016 µM⁻¹ s⁻¹, where k_pol = k₂k₃/(k₋₂+k₃))⁴² is slightly smaller than the k₂/K_d (0.03 µM⁻¹ s⁻¹) for 8-oxodGTP. Interestingly, it has been proposed that slow pyrophosphate release is an additional factor that limits incorporation of 8-oxodGTP by mitochondrial polymerase.⁴²

Single-Turnover Kinetic Analysis of dSpTP Incorporation

Compared to dGTP and 8-oxodGTP, experiments using the diastereomers of dSpTP required significantly higher nucleotide triphosphate concentrations and extended incubation times to observe a sufficient amount of product for analysis (Figure 4B, 4C). We also observed differences in incorporation between the two dSpTP diastereomers. Specifically, the amount of product accumulated after a particular reaction time for (S)-dSpTP is significantly greater than for (R)-dSpTP at the same nucleotide concentration, consistent with the results of the qualitative incorporation presented in Figure 2B. Global fitting of the data to the models in Figure 3A and 3B reveal that the data are described better by a model (Figure 4A) that allows reversibility of the incorporation step (data not shown), similar to the results observed for 8-oxodGTP. However, unlike what we observed for incorporation of dGTP or 8-oxodGTP, for the diastereomers of dSpTP the parameters in the kinetic scheme were not well constrained (Figure 4D, E and Figure S4). As a consequence, we could not obtain individual quantitative values for K_d and k_pol. Because k_pol is limited by the amount of available DNA·KF⁻·dNTP complex, poor binding of the hyperoxidized nucleotide triphosphate to the DNA·KF⁻ complex likely causes the lack of parameter constraint. Importantly, these data indicate that the K_d for both diastereomers of dSpTP is greater than the nucleotide concentrations used here and these values are at least on the millimolar scale. Even though k₋₁ and k₂ are not constrained and can vary across a wide range for both diastereomers of dSpTP, analysis by KinTek Explorer FitSpace reveals there is a linear correlation between k₂ and k₋₁ (Figure 4D, E). Since k₁ was fixed at 10 µM⁻¹ s⁻¹ during the fitting (see Experimental Procedures), there is also a linear correlation between k₂ and K_d (K_d = k₋₁/k₁ = k₋₁/10). Therefore we use k₂/K_d to compare the incorporation specificities of the dNTPs used in this study, rather than k_pol/K_d. From the slope of FitSpace analysis k₂/k₋₁ (or k₂/(10×K_d)), we defined the incorporation specificity k₂/K_d for (S)-dSpTP and (R)-dSpTP as 8×10⁻⁴ µM⁻¹ s⁻¹ and 6×10⁻⁵ µM⁻¹ s⁻¹, respectively.

Figure 4.

Kinetic analysis of (S)-dSpTP and (R)-dSpTP incorporation by KF⁻. (A) Enzyme model used for global fitting. (B) Global fitting of incorporation of (S)-dSpTP opposite of template dC. (S)-dSpTP concentrations are 10 µM (closed circles), 20 µM (open circles), 50 µM (closed squares), 100 µM (open squares), and 200 µM (closed triangles). (C) Global fitting of incorporation of (R)-dSpTP opposite of template dC. (R)-dSpTP concentrations are 50 µM (closed circles), 100 µM (open circles), 200 µM (closed squares), 400 µM (open squares), and 600 µM (closed triangles). The curves superimposed with the experimental data were generated by KinTek Explorer fitting. FitSpace analysis by KinTek Explorer showing the results of the initial excursions to map the boundaries of a good fit for (D) (S)-dSpTP and (E) (R)-dSpTP.

Our kinetic results demonstrate that KF⁻ exhibits a stereochemical preference and more readily incorporates (S)-dSpTP over (R)-dSpTP. A general stereochemical preference for the (S) diastereomer has been reported previously for several other DNA-processing enzymes. The Burrows laboratory showed that KF⁻ was more efficient at inserting dATP opposite dSp1 relative to dSp2, where the diastereomers were first described based on their order of elution from an anion exchange HPLC column.³⁴ Subsequent studies revealed that dSp1 and dSp2 are (S)-dSp and (R)-dSp, respectively.^{48, 49} The Essigmann laboratory similarly showed that pol II and pol IV more efficiently bypass dSp1 in vivo⁶⁴ and that the relative amount of G → C and G → T transversions can vary for dSp1 and dSp2 depending on the sequence context.^35–37 Furthermore, nuclease P1⁶⁵ and the BER glycosylase human NEIL1^{28, 30, 32} also have a preference for the dSp1 configuration. Molecular dynamics simulations showed that hNEIL1 makes better contacts with (S)-dSp in its binding pocket, justifying the stereochemical preference of the enzyme.⁶⁶

Comparing the Kinetic Parameters for dGTP, 8-oxodGTP, (S)-dSpTP, and (R)-dSpTP

Our results show that the extent of oxidation of the dGTP-derived lesion is correlated with its ability to be incorporated into DNA. The specificity of incorporation is highest for dGTP opposite dC by KF⁻. Compared to dGTP, the selection factors for incorporating 8-oxodGTP, (S)-dSpTP, and (R)-dSpTP are 1,700-, 64,000- and 850,000-fold lower respectively (Table 1). Notably, since pyrophosphate release is slow for 8-oxodGTP, (S)-dSpTP, and (R)-dSpTP, k_pol/K_d must be smaller than k₂/K_d, so these selection factors are lower limits for specificity.

In order to provide a graphical representation of the difference in specificity for incorporation of the canonical, oxidized, and hyperoxidized guanine lesions we also analyzed our kinetic data using traditional Michaelis-Menten methods, in which the observed rate (k_obs) is plotted against nucleotide triphosphate concentration (Figure 5). The slope of the rise can be used to define incorporation specificity k_pol/K_d. In fact, the k_pol/K_d from this Michaelis-Menten analysis (data not shown) is comparable to the k₂/K_d obtained by global fitting. Clearly, the incorporation kinetics for dGTP are significantly different from those of oxidized nucleotides (Figure 5 inset, open triangles). For the lesion dNTPs, a rise in k_obs as a function of dNTP concentrations is observed for 8-oxodGTP and (S)-dSpTP, with 8-oxodGTP more steep than (S)-dSpTP. However, for (R)-dSpTP, k_obs was extremely slow across the entire concentration range, and there was no obvious dependence of the rate on nucleotide concentration. These results are entirely consistent with the quantitative parameters obtained by global fitting.

Figure 5.

Analysis of dGTP (open triangles), 8-oxodGTP (closed circles), (S)-dSpTP (open squares), and (R)-dSpTP (closed triangles) kinetic results using Michaelis-Menten techniques where the observed rates are plotted against nucleotide concentrations. The insert is a zoomed out view of the main panel and allows for comparison of dGTP to the three oxidized nucleotides.

Biological Considerations

In addition to the kinetic parameters describing incorporation of lesion dNTPs, one must also consider the action of enzymes responsible for cleansing the nucleotide pool. The importance of removing 8-oxodGTP from the nucleotide pool is underscored by the fact that E. coli lacking MutT, a phosphatase that converts 8-oxodGTP to 8-oxodGMP,³⁹ have a 100- to 10,000-fold higher mutation rate compared to wild type E. coli.⁶⁷ This dramatic increase in mutation rate in the absence of MutT indicates that the nucleotide pool represents a biologically significant source of 8-oxodGTP.^{68, 69} However, it has been demonstrated that the diastereomers of dSpTP are not good substrates for MutT or other known MutT-type nucleotide pool sanitization enzymes in E. coli.⁷⁰ Therefore, preventing the incorporation of dSpTP into DNA may depend solely on the ability of a polymerase to discriminate the hyperoxidized nucleotide triphosphate from dGTP. Indeed, consistent with our kinetic results, in vivo studies demonstrate that introduction of dSpTP1 or dSpTP2 into E. coli cells does not significantly increase the mutation frequency.⁷⁰ This result is in contrast to the high mutagenicity observed in E. coli when the Sp lesions are in the DNA template.^{35–37, 64} Our results suggest that the lack of mutagenicity of dSpTP is because the E. coli cellular polymerases effectively discriminate the hyperoxidized nucleotides and do not use them as building blocks during replication. It is notable, however, that although it is difficult to force dSpTP into DNA from the nucleotide pool, if incorporation occurs, the lesion is a powerful source of mutations. Indeed, the “Trojan horse” quality of dSpTP makes it an intriguing player in the field of lethal mutagenesis^{71, 72} where a miscoding nucleotide is used as an antiviral agent to accelerate viral mutation rates and drive a viral population to extinction. In order to be effective in this manner, however, the incorporation efficiency of dSpTP would likely need to be higher than that observed for KF⁻.

In this work, we used KF⁻ as a model polymerase. Similar incorporation efficiencies have been observed for KF⁻ and some mammalian replicative polymerases incorporating 8-oxodGTP into DNA,⁷³ although mammalian polymerases have a higher preference to incorporate 8-oxodGTP opposite the correct dC instead of the mismatched dA template.⁷⁴ Therefore, we expect the selection factor for mammalian polymerases incorporating the hyperoxidized dNTP to be even more rigorous than KF⁻ and to prevent incorporation of dSpTP from the nucleotide pool.

As a final consideration the in vivo concentrations of (S)-dSpTP and (R)-dSpTP will also influence the extent to which the hyperoxidized lesion can be incorporated into DNA. These values remain to be determined but are most certainly well below the millimolar K_d values suggested by our experiments. Thus, we conclude that while dSpTP can be described as a Trojan horse, the specificity of cellular polymerase provides an effective defense against using the hyperoxidized nucleotide as a building block during DNA replication, and that the dSp lesion is a much more potent mutagen when it is formed in genomic DNA.

ABBREVIATIONS

8-oxodGTP

2'-deoxy-8-oxo-7,8-dihydroguanosine-5'-triphosphate

BER

base excision repair

dGTP

2'-deoxyguanosine-5'-triphosphate

dSpTP

spiroiminodihydantoin-2'-deoxynucleoside-5'-triphosphate

ESI-MS

electrospray ionization mass spectrometry

KF⁻

Klenow Fragment 3'→5' exo⁻

RQF

rapid quench flow

ROS

reactive oxygen species