Pre-Steady-State Kinetic Investigation of Bypass of a Bulky Guanine Lesion by Human Y-family DNA Polymerases

Abstract

3-Nitrobenzanthrone (3-NBA), a byproduct of diesel exhaust, is highly present in the environment and poses a significant health risk. Exposure to 3-NBA results in formation of N-(2′-deoxyguanosin-8-yl)-3-aminobenzanthrone (dG^C8-N-ABA), a bulky DNA lesion that is of particular importance due to its mutagenic and carcinogenic potential. If not repaired or bypassed during genomic replication, dG^C8-N-ABA can stall replication forks, leading to senescence and cell death. Here we used pre-steady-state kinetic methods to determine which of the four human Y-family DNA polymerases (hPolη, hPolκ, hPolι, or hRev1) are able to catalyze translesion synthesis of dG^C8-N-ABA in vitro. Our studies demonstrated that hPolη and hPolκ most efficiently bypassed a site-specifically placed dG^C8-N-ABA lesion, making them good candidates for catalyzing translesion synthesis (TLS) of this bulky lesion in vivo. Consistently, our publication (Biochemistry 53, 5323–31) in 2014 has shown that small interfering RNA-mediated knockdown of hPolη and hPolκ in HEK293T cells significantly reduces the efficiency of TLS of dG^C8-N-ABA. In contrast, hPolι and hRev1 were severely stalled by dG^C8-N-ABA and their potential role in vivo was discussed. Subsequently, we determined the kinetic parameters for correct and incorrect nucleotide incorporation catalyzed by hPolη at various positions upstream, opposite, and downstream from dG^C8-N-ABA. Notably, nucleotide incorporation efficiency and fidelity both decreased significantly during dG^C8-N-ABA bypass and the subsequent extension step, leading to polymerase pausing and error-prone DNA synthesis by hPolη. Furthermore, hPolη displayed nucleotide concentration-dependent biphasic kinetics at the two polymerase pause sites, suggesting that multiple enzyme•DNA complexes likely exist during nucleotide incorporation.

Graphical Abstract

1. INTRODUCTION

DNA damage caused by prevalent chemical agents in the environment poses significant threats to cell survival and proliferation and represents a major concern for occupational and public health. One molecule in particular, 3-nitrobenzanthrone (3-NBA)^a, a nitro-polycyclic aromatic hydrocarbon (nitroPAH), was discovered in diesel exhaust as bound to particulate air matter ¹. 3-NBA is ubiquitously present in the environment, especially in areas exposed to high levels of diesel emissions, such as cities or areas of heavy construction ². It has one of the most mutagenic profiles of known pollutants tested in the Ames Salmonella typhimurium (TA98) assay and is responsible for inducing mostly G→T transversions ^{1; 3}. Upon inhalation, 3-NBA is metabolized into various mutagenic and potentially carcinogenic metabolites that can react with purines in DNA ⁴. These reactions lead to the formation of bulky, aromatic DNA lesions which act as roadblocks to genomic replication. One such product, N-(2′-deoxyguanosin-8-yl)-3-aminobenzanthrone (dG^C8-N-ABA), is of particular interest because it is a major lesion formed in mice after intraperitoneal treatment with 3-NBA ⁵. The bulky lesion dG^C8-N-ABA, produced by the electrophilic attack of deoxyguanosine (dG) by a 3-NBA metabolite ⁴ (Figure S1), has been found to severely block replication fork progression ⁶.

During normal DNA replication, highly efficient and faithful DNA polymerases perform the bulk of the work. It is known that replicative DNA polymerases possess tight and selective active sites that allow them to stringently choose against incorrect nucleotides but also limit their abilities to catalyze translesion synthesis (TLS) through damaged template nucleotides ^{7; 8}. When a damaged site is encountered, replicative polymerases are stalled due to a significantly slowed nucleotide incorporation activity and a competing 3′→5′ exonuclease activity, resulting in futile cycles of DNA synthesis and excision⁹. In contrast, Y-family DNA polymerases, which lack proofreading activity and possess comparatively flexible and solvent accessible active sites, are more suited to accommodate DNA lesions and catalyze TLS ¹⁰. Notably, the human genome encodes four Y-family members: DNA polymerase eta (hPolη), kappa (hPolκ), iota (hPolι), and hRev1. These human polymerases vary in their lesion bypass capabilities and are differentially recruited to damaged DNA to catalyze TLS depending on the lesion that is present ¹⁰. In this study, we sought to determine which of the human Y-family DNA polymerase(s) are able to efficiently bypass dG^C8-N-ABA in vitro. Previously we utilized pre-steady-state kinetic methods to study the bypass of dG^C8-N-ABA catalyzed by a model Y-family polymerase, Sulfolobus solfataricus DNA Polymerase IV (Dpo4) ¹¹. We observed that Dpo4 is significantly less efficient and faithful during dG^C8-N-ABA bypass and the subsequent extension compared to undamaged DNA, implicating the mutagenic effect of this lesion in vivo. Here we applied similar methods to determine the most active human Y-family polymerase for dG^C8-N-ABA bypass and establish a kinetic mechanism for TLS of dG^C8-N-ABA. Initially, running start assays were performed to study the effect that a site-specifically placed, bulky dG^C8-N-ABA lesion had on each human Y-family polymerase. Our results indicate that hPolη is a likely candidate to catalyze TLS of dG^C8-N-ABA in vivo. Subsequently, we performed single-turnover kinetic assays to determine the DNA binding affinities (K_{d, DNA}), maximum nucleotide incorporation rate constants (k_p), apparent nucleotide binding affinities (K_{d, dNTP}), and the substrate specificities (k_p/K_{d, dNTP}) for nucleotide incorporations by hPolη at sites upstream, opposite, and downstream from the lesion. Finally, biphasic kinetic assays were performed to establish a minimal kinetic mechanism for nucleotide incorporation opposite dG^C8-N-ABA and the subsequent extension of the lesion bypass product.

2. MATERIALS AND METHODS

2.1 Materials

All reagents were purchased from the following companies: OptiKinase from United States Biochemical, [γ-³²P] ATP from Perkin-Elmer, Biospin-6 columns from Bio-Rad Laboratories, and dNTPs from Bioline. Full-length hPolη and truncated fragments of hPolκ (9–518), hPolι (1–420), and hRev1 (341–829) were overexpressed and purified as described previously ¹².

2.2 DNA Substrates

A DNA template (26-mer) containing a dG^C8-N-ABA lesion at the 21^st position was synthesized as described previously ¹¹. DNA containing the dG^C8-N-ABA lesion was kept in dark conditions to preserve the light-sensitive substrate. All other synthetic oligonucleotides were purchased from Integrated DNA Technologies (IDT) and gel purified by 17% denaturing polyacrylamide gel electrophoresis (PAGE). Primers were 5′-radiolabeled by incubating with [γ-³²P]-ATP and OptiKinase for 3 hr, followed by heat inactivation (95 °C for 2 min), and isolation of radiolabeled DNA using a Biospin-6 column to remove excess, unreacted [γ-³²P]-ATP. A primer was annealed to a template at a 1:1.35 ratio. Annealing solutions containing undamaged DNA and dG^C8-N-ABA-containing DNA were heated to 95 and 72 °C, respectively, for 5 minutes, followed by slow cooling to room temperature. All DNA substrates used in this study are listed in Table 1.

Table 1.

Sequences of DNA oligonucleotides

Primers (Position)a	Sequence
17-mer (−4)	5′-AACGACGGCCAGTGAAT-3′
18-mer (−3)	5′-AACGACGGCCAGTGAATT-3′
19-mer (−2)	5′-AACGACGGCCAGTGAATTC-3′
20-mer (−1)	5′-AACGACGGCCAGTGAATTCG-3′
21-mer (0)	5′-AACGACGGCCAGTGAATTCGC-3′
22-mer (+1)	5′-AACGACGGCCAGTGAATTCGCG-3′
Templates
26-mer	3′-TTGCTGCCGGTCACTTAAGCGCGCCC-5′
26-mer-dGC8-N-ABA	3′-TTGCTGCCGGTCACTTAAGCGCGCCC-5′

G indicates the dG^C8-N-ABA lesion at the 21^st position in the template.

^aPosition of primer 3′ terminus relative to the templating dG^C8-N-ABA.

2.3 Reaction Buffers and Assay Analysis

All chemical quench assays were performed with a KinTek Rapid Chemical Quench-Flow apparatus at 37 °C in buffer R (50 mM HEPES, pH 7.5 at 37 °C, 50 mM NaCl, 5 mM MgCl₂, 0.1 mM EDTA, 10% glycerol, 5 mM DTT, and 0.1 mg/mL Bovine Serum Albumin (BSA)). Electrophoretic mobility shift assays (EMSAs) were performed at 23 °C in buffer S (50 mM Tris-HCl, pH 7.5 at 23 °C, 50 mM NaCl, 5 mM MgCl₂, 0.1 mM EDTA, 10% glycerol, 5 mM DTT, and 0.1 mg/mL BSA). All concentrations listed were final upon mixing and all gels were scanned using a TyphoonTrio (GE Healthcare). Quantification was carried out using ImageQuant (GE Healthcare) and Kaleidagraph (Synergy) software.

2.4 Running Start Assays

A pre-incubated solution of hPolη (100 nM), hPolκ (100 nM), hPolι (1 μM), or hRev1 (1 μM) and 5′-[³²P]-17/26-mer (100 nM) was rapidly mixed with all four dNTPs (200 μM each). After various incubation periods, the samples were quenched with the addition of 0.37 M EDTA and separated by 17% PAGE (8 M Urea, 1X TBE). Analysis and quantification of gels were performed as described in Reaction Buffers and Assay Analysis. The relative lesion bypass efficiencies (bypass %) were calculated for each polymerase as a function of reaction time. For each time point t, the bypass % was calculated as a ratio of bypass to encounter events (Eq. 1) as described previously ^{12; 13; 14}.

$${\text{dG}}^{\mathrm{C}8-N-\text{ABA}}\text{bypass}\%=(\mathrm{B}/\mathrm{E})\times 100=[\mathrm{B}/([20-\text{mer}]+\mathrm{B})]\times 100$$ Eq. 1

Quantification of running start assays (Figure 1A–H) was achieved by treating the total encounter (E) events as equivalent to the concentration of 20-mer, and bypass (B) events as equivalent to the sum of the concentrations of all products with sizes greater than or equal to the 21-mer. The bypass % was calculated using Eq. 1 then plotted against reaction times, t. To compare the relative amount of time it took for each polymerase to bypass dG^C8-N-ABA, the t₅₀bypass was determined as the time to bypass of 50% of the total lesions encountered. In a similar fashion, the t₅₀ was determined as the time to bypass 50% of the undamaged dG base.

Figure 1. Running start assays at 37 °C.

A pre-incubated solution of 100 nM 5′-[³²P]-labeled 17/26-mer DNA (A, C, E, and G) or 17/26-mer-dG^C8-N-ABA (B, D, F, and H) and 100 nM hPolη (A and B), 100 nM hPolκ (C and D), 1 μM hPolι (E and F), or 1 μM hRev1 (G and H) was rapidly mixed with all four dNTPs (200 μM each) for various incubation periods before being quenched with the addition of 0.37 M EDTA. Sizes of important products are indicated. The dG^C8-N-ABA lesion is located at the 21^st position from the 3′ terminus of the DNA template.

2.5 Electrophoretic Mobility Shift Assays

hPolη (2–650 nM) was titrated into solutions containing 5′-[³²P]-labeled DNA (10 nM) and allowed to equilibrate in buffer S for 15 min at room temperature (23 °C). The samples were separated using 4.5% native PAGE which were analyzed and quantified as described in Reaction Buffers and Assay Analysis. The concentration of hPolη•DNA complex was plotted against the total concentration of hPolη and the data were fit to a quadratic equation (Eq. 2):

$$[\mathrm{E}•\text{DNA}]=0.5(K_{\mathrm{d},\text{DNA}}+E_{\mathit{0}}+D_{\mathit{0}})-0.5{[{(K_{\mathrm{d},\text{DNA}}+E_{\mathit{0}}+D_{\mathit{0}})}^2-4E_{\mathit{0}}{D}_{\mathit{0}}]}^{1/2}$$ Eq. 2

Where E₀ is the initial enzyme concentration, D₀ is the initial DNA concentration, and K_{d, DNA} is the apparent equilibrium dissociation constant for the hPolη•DNA complex.

2.6 Standing Start Assay for hRev1

A pre-incubated solution of hRev1 (200 nM) and 5′-[³²P]-labeled undamaged (20/26-mer) or damaged (20/26-mer-dG^C8-N-ABA) DNA (200 nM) in buffer R was mixed with dATP, dCTP, dGTP, dTTP, or all four nucleotides (200 μM each). After 10 min, the reaction was quenched with the addition of 0.37 M EDTA and the samples were separated using 17% denaturing PAGE.

2.7 Single-Turnover Assays

A pre-incubated solution of 5′-[³²P]-labeled DNA (20 nM) and hPolη (130 nM) in buffer R was rapidly mixed with varying concentrations of dNTP (10 – 1200 μM). After various incubation times, the reaction was quenched with the addition of 0.37 M EDTA. The samples were then analyzed using 17% denaturing PAGE. Product concentration was plotted against time and the data were fit to a single-exponential equation (Eq. 3).

$$[\text{Product}]=A[1-exp(-k_{\text{obs}}t)]$$ Eq. 3

Where k_obs is the observed rate constant of nucleotide incorporation and A is the reaction amplitude. The k_obs values obtained using Eq. 3 were plotted against the respective dNTP concentration and the data were fit to a hyperbolic equation (Eq. 4)

$$k_{\text{obs}}=k_{\mathrm{p}}[\text{dNTP}]/([\text{dNTP}]+K_{\mathrm{d},\text{dNTP}})$$ Eq. 4

Where k_p is the maximum nucleotide incorporation rate constant and K_{d, dNTP} is the apparent equilibrium dissociation constant for the binding of dNTP to hPol•DNA.

2.8 Biphasic Assays

Biphasic assays were performed by rapidly mixing a pre-incubated solution of hPolη (120 nM) and 5′-[³²P]-labeled DNA (30 nM) with various concentrations of dNTP (25–1,000 μM) and excess, unlabeled DNA trap (5 μM). After various incubation periods, reactions were quenched with the addition of 0.37 M EDTA and samples were separated using 17% denaturing PAGE. Product concentration was plotted against time and the data were fit to a double-exponential equation (Eq. 5)

$$[\text{Product}]=A_{\mathrm{f}}[1-exp(-k_{\mathrm{f}}t)]+A_{\mathrm{s}}[1-exp(-k_{\mathrm{s}}t)]$$ Eq. 5

Where A_f and A_s are the reaction amplitudes of the fast and slow phases, respectively, and k_f and k_s are the rate constants of the fast and slow phases, respectively.

3. RESULTS

3.1 Measuring Relative Polymerase Efficiency during dG^C8-N-ABA Bypass

To test if each individual human DNA polymerase could bypass a single dG^C8-N-ABA lesion, running start assays were performed by rapidly mixing pre-incubated hPolη, hPolκ, hPolι, or hRev1 and undamaged 17/26-mer or damaged 17/26-mer-dG^C8-N-ABA containing a site-specifically placed dG^C8-N-ABA (Table 1) with all four dNTPs. The lesion stalled DNA synthesis by each of the Y-family polymerases, but the degree of stalling as well as the positions at which stalling was observed varied between each polymerase. For hPolη, moderate pausing was observed at the 19-mer (−2), 20-mer (−1), 21-mer (0), and 22-mer (+1) positions. Significant accumulation of the 20-mer product indicated that incorporation of nucleotide at the bypass step (Position −1) was the most problematic for hPolη (Figure 1B). Despite this stalling, the full-length product appeared after 6 s, which was only 2-fold longer than the time required for hPolη to synthesize the full-length product with the undamaged 17/26-mer. hPolκ stalled significantly during the bypass and extension steps, indicated by the 20-mer and 21-mer accumulation at these positions (Figure 1D). hPolκ produced the full length products within 1 s with the undamaged 17/26-mer and took only 3-fold longer (3 s) with the 17/26-mer-dG^C8-N-ABA substrate, indicating that hPolκ was similarly efficient as hPolη when bypassing dG^C8-N-ABA. The observed accumulation of the 25-mer product is likely caused by polymerase slippage and template repositioning within the 5′-dC-rich region of the template (Table 1), resulting in the “−1” frameshift ^{15; 16}. Additionally, the full length product (25-mer) produced by this mechanism was likely further extended via a blunt-end addition ^{17; 18}. Interestingly, the bulky lesion caused significant stalling of hPolι and hRev1 with no appearance of the full-length product even after long reaction periods (3 and 10.5 hr, respectively). With 17/26-mer-dG^C8-N-ABA, hPolι synthesized the bypass and extension products after 300 and 3,900 s, respectively, which are 30- and 65-fold, respectively, longer than with the undamaged 17/26-mer (Figure 1E–F). In comparison, hRev1 synthesized the lesion bypass product within 1 hr but was unable to extend it even after 10.5 hr. The failure of hRev1, a dCTP polymerase, to efficiently catalyze TLS of dG^C8-N-ABA was likely caused by multiple template-independent misincorporations of dCTP by hRev1 ¹⁹, resulting in DNA distortion. This is supported by the fact that hRev1was not able to synthesize the full-length product even with the undamaged DNA substrate (Figure 1H). The results of the running start assay with hRev1 prompted us to perform standing start primer extension assays to qualitatively evaluate the efficiency and fidelity of hRev1 to incorporate a nucleotide directly opposite from dG^C8-N-ABA. A pre-incubated solution of hRev1 and 20/26-mer-dG^C8-N-ABA (or 20/26-mer) was mixed and reacted with a solution containing either dATP, dCTP, dGTP, dTTP, or all four dNTPs. Interestingly, hRev1 showed greater selectivity with damaged than with undamaged DNA by predominantly incorporating dCTP (Figure S1). With undamaged DNA, hRev1 incorporated all four nucleotides with a preference of dCTP > dATP = dGTP = dTTP.

To provide a comparison for the amount of time it took for each human Y-family polymerase to bypass dG^C8-N-ABA, gels from the running start assays (Figure 1 A–H) were quantified and Eq. 1 was used to calculate values for t₅₀ and t₅₀bypass which represent the times it took to elongate 50% of the 20-mer to the 21-mer or longer products with undamaged and damaged DNA, respectively. The t₅₀ and t₅₀bypass times measured for each of the Y-family polymerases is summarized in Table 2. Notably, hPolι and hRev1 did not bypass 50% of the dG^C8-N-ABA lesion in the allowed reaction time for the running start assays. In contrast, the relatively short t₅₀bypass times for hPolη (13 s) and hPolκ (7 s) suggest a potential role involving these two polymerases in the bypass and/or extension of dG^C8-N-ABA lesions in vivo. Consistently, a recent publication reports ~25 and ~30% efficiency reductions in TLS of dG^C8-N-ABA in human embryonic kidney (HEK293T) cells when hPolη and hPolκ, respectively, were subject to gene knockdown ²⁰. Based on the slightly smaller perturbation of DNA synthesis by hPolη in the presence of dG^C8-N-ABA, we decided to use pre-steady-state kinetic methods to initially characterize the efficiency and fidelity of TLS of dG^C8-N-ABA catalyzed by hPolη. However, it is apparent based on both the running start and HEK293T cell-based assays that hPolκ is also potentially involved in TLS of dG^C8-N-ABA in vivo. Accordingly, we are currently performing similar kinetic studies to determine the effect of dG^C8-N-ABA on nucleotide incorporation kinetics with hPolκ.

Table 2.

dG^C8-N-ABA bypass halftimes of human Y-family polymerases at 37 °C

Polymerase	t50(s)a	t50bypass(s)b	t50bypass/t50
hPolη	0.7	13	19
hPolκ	0.3	7	23
hPolι	200	>10,800	>55
hRev1	1,800	>37,000	>20

^aCalculated as the time required for traversing 50% of undamaged dG.

^bCalculated as the time required for bypassing of 50% of dG^C8-N-ABA.

3.2 DNA Binding Affinity Estimated through Electrophoretic Mobility Shift Assays

To determine the apparent equilibrium dissociation constant for the binding of hPolη to DNA containing dG^C8-_N^-ABA (K_{d, DNA}), EMSAs were performed. A representative gel image and a plot for an EMSA experiment are shown in Figure S2 for hPolη binding to 21/26-mer-dG^C8-N-ABA. For all DNA substrates tested, the K_{d, DNA} values were within a 2-fold range for the corresponding undamaged and damaged DNA substrates (Figure S3D and Table 3). Interestingly, the binding affinities of hPolη for 19/26-mer-dG^C8-N-ABA and 20/26-mer-dG^C8-N-ABA (K_{d, DNA} =12 ± 2 nM for both) were 2-fold tighter than the corresponding undamaged DNA substrates (K_{d, DNA} =25 ± 5 nM; 23 ± 2 nM). Based on our ternary crystal structure of hPolη in complex with dG^C8-N-ABA-DNA and dCTP²¹, the slight increase in binding affinity likely arises from positioning of the ABA moiety of dG^C8-N-ABA within a hydrophobic cleft in the hPolη active site thereby stabilizing the conformation of the damaged base. Similar effects are observed in the crystal structures of Dpo4 with dG^BPDE and dG^{AP 22; 23}. Overall, hPolη binds to dG^C8-N-ABA-containing DNA tightly enough such that it does not significantly contribute to the polymerase pausing pattern observed in the running start assay (Figure 1B). Similar results have been obtained for dG^AP bypass by hPolη ¹³ as well as dG^AP, dG^C8-N-ABA, and abasic site bypass by Dpo4 ^{11; 24; 25}.

Table 3.

DNA binding affinity of hPolη at 23 °C

DNA Substrate	Kd, undamaged DNA (nM)	Kd, damaged DNA (nM)a	Affinity Ratiob
18/26-mer	28 ± 3	32 ± 5	1.1
19/26-mer	25 ± 5	12 ± 2	0.5
20/26-mer	23 ± 2	12 ± 2	0.5
21/26-mer	26 ± 1	26 ± 7	1.0
22/26-mer	17 ± 6	24 ± 4	1.4

^aDamaged DNA refers to template 26-mer containing a dG^C8-N-ABA lesion at the 21^st position.

^bCalculated as K_{d, damaged DNA}/K_{d, undamaged DNA}.

3.3 Determination of Kinetic Parameters for Nucleotide Incorporations upstream, opposite, and downstream from dG^C8-N-ABA

Since DNA binding did not significantly affect TLS of dG^C8-N-ABA by hPolη, it is possible that the stalling of hPolη during TLS (Figure 1B) was caused by decreased nucleotide binding affinity and incorporation efficiency. To examine this hypothesis, we utilized single-turnover assays to measure the apparent equilibrium dissociation constant for nucleotide binding (K_d, _dNTP) and the maximum rate constant of nucleotide incorporation (k_p) at positions upstream, downstream, and opposite from dG^C8-N-ABA. By varying the primer length, we were able to measure the kinetic parameters at specific positions along the template strand relative to dG^C8-N-ABA, and thus determine the effect of dG^C8-N-ABA on the efficiency and fidelity of nucleotide incorporation during each step of TLS catalyzed by hPolη. Briefly, a pre-incubated solution of 5′-[³²P]-labeled DNA (20 nM, Table 1) and hPolη (130 nM) was rapidly mixed with a single dNTP at varying concentrations. Reactions were analyzed as described (Materials and Methods), and the data were fit to Eq. 3 to yield an observed rate constant of incorporation (k_obs). The k_obs values were then plotted against the respective dNTP concentrations and fit to a hyperbolic equation (Eq. 4) to yield the k_p and K_{d, dNTP} (Figure 2). hPolη showed a decrease in nucleotide incorporation efficiency (k_p/K_{d, dNTP}) at positions upstream, opposite, and downstream from the lesion (Position −3, −2, −1, 0, +1). With 18/26-mer-dG^C8-N-ABA at Position −3, hPolη correctly incorporated dCTP with a k_p of 66 ± 1 s⁻¹, although the dCTP binding affinity was weakened by 4-fold (K_{d, dCTP} =200 ± 11 μM) relative to undamaged 18/26-mer. As a result, the polymerase efficiency (k_p /K_{d, dCTP} = 3.3 × 10⁻¹ s⁻¹ μM⁻¹) was only slightly reduced compared to the value with 18/26-mer (k_p/K_{d, dCTP} = 1.5 s⁻¹ μM⁻¹). Similar experiments were carried out with both undamaged and damaged DNA substrates containing primers of 19-mer (Position −2), 20-mer (Position −1), 21-mer (Position 0), and 22-mer (Position +1) and the kinetic parameters are listed in Tables 3 and S1. The most significant stalling occurred during incorporation opposite dG^C8-N-ABA (Position −1), caused by a 25-fold decrease in k_p (1.9 ± 0.1 s⁻¹) and a 4-fold increase in K_{d, dCTP} (325 ± 44 μM) which resulted in an overall reduction in efficiency by approximately 100-fold (k_p/K_{d, dCTP} = 5.8 × 10⁻³ s⁻¹ μM⁻¹) with 20/26-mer-dG^C8-N-ABA compared to undamaged 20/26-mer (k_p/K_{d, dCTP} = 5.6 × 10⁻¹ s⁻¹ μM⁻¹). At the extension step (Position 0), hPolη displayed a k_p that was 49-fold slower (0.8 ± 0.03 s⁻¹) but surprisingly bound the incoming dGTP more tightly (K_{d, dGTP} = 28 ± 5 μM). With 21/26-mer-dG^C8-N-ABA, hPolη showed an approximate 5-fold increase in efficiency (k_p/K_{d, dGTP} = 2.8 × 10⁻² s⁻¹ μM⁻¹) compared to the lesion bypass step (Position −1). With 22/26-mer-dG^C8-N-ABA (Position +1), the efficiency of hPolη was similar (k_p/K_{d, dCTP} = 4.2 × 10⁻² s⁻¹ μM⁻¹) to that measured at position −2 (Table 4) which is indicated by similarly moderate accumulation at these positions (Figure 1B).

Figure 2. Determination of the kinetic parameters for dCTP incorporation into 20/26-mer-dG^C8-N-ABA at 37 °C.

(A) A pre-incubated solution of hPolη (120 nM) and 5′-[³²P]-labeled 20/26-mer-dG^C8-N-ABA (20 nM) was rapidly mixed with increasing concentrations of dCTP (25 μM, ●; 50 μM, ○; 100 μM, ■;200 μM, □; 400 μM, ◆; 600 μM, ◇; and 800 μM, ▲) before being quenched by the addition of 0.37 M EDTA at the indicated time points. Product formation was plotted against time and the data were fit to Eq. 3 to yield the observed rate constant of product formation (k_obs). (B) Each k_obs value was plotted against its respective dCTP concentration and the data were fit to Eq. 4 to yield the maximum dCTP incorporation rate constant (k_p) of 1.9 ± 0.1 s⁻¹ and dCTP equilibrium dissociation constant (K_{d, dCTP}) of 325 ± 44 μM.

Table 4.

Kinetic parameters for dNTP incorporation into dG^C8-N-ABA-containing DNA at 37 °C

dNTP	Kd, dNTP (μM)	kp(s−1)	kp/Kd, dNTP (μM−1s−1)	Efficiency Ratioa,b	Fidelityc	Fidelity ratiod	Probability e (%)
18/26-mer-dGC8-N-ABA - Template dG (Position −3)
dCTP	200 ± 11	66 ± 1	3.3 × 10−1	5	-	-	-
dATP	369 ± 48	2.1 ± 0.1	5.7 × 10−3	0.6	1.7 × 10−2	7.4	-
19/26-mer-dGC8-N-ABA - Template dC (Position −2)
dGTP	150 ± 34	6.6 ± 0.4	4.4 × 10−2	23	-	-	-
dCTP	90 ± 19	0.093 ± 0.0001	1.0 × 10−3	5	2.2 × 10−2	4.8	-
20/26-mer-dG C8-N-ABA - Template dGC8-N-ABA (Position −1)
dCTP	325 ± 44	1.9 ± 0.1	5.8 × 10−3	97	-	-	77
dATP	161 ± 38	0.18 ± 0.01	1.1 × 10−3	3	1.6 × 10−1	33	15
dGTP	39 ± 5	0.014 ± 0.0003	3.6 × 10−4	3	5.8 × 10−2	29	5
dTTP	475 ± 85	0.14 ± 0.01	2.9 × 10−4	23	4.8 × 10−2	4	4
21/26-mer-dGC8-N-ABA - Template dC (Position 0)
dGTP	28 ± 5	0.8 ± 0.03	2.9 × 10−2	29	-	-	74
dATP	473 ± 25	3.2 ± 0.10	6.8 × 10−3	2	1.9 × 10−1	11	17
dCTP	201 ± 20	0.4 ± 0.02	2.0 × 10−3	9	6.4 × 10−2	3	5
dTTP	392 ± 108	0.6 ± 0.01	1.5 × 10−3	3	4.9 × 10−2	9	4
22/26-mer-dGC8-N-ABA - Template dG (Position +1)
dCTP	215 ± 67	9.1 ± 1.0	4.2 × 10−2	4	-	-	-
dGTP	41 ± 8	0.028 ± 0.001	6.8 × 10−4	0.6	1.6 × 10−2	6.5	-

^aCalculated as [(k_p/K_{d, dNTP})_{undamaged DNA}/(k_p/K_{d, dNTP})_{damaged DNA}].

^bKinetic parameters for undamaged DNA from Table S1 of Sherrer et al.¹³

^cCalculated as (k_p/K_{d, dNTP})_incorrect/[(k_p/K_{d, dNTP})_incorrect) + (k_p/K_{d, dNTP})_correct].

^dCalculated as Fidelity_{damaged DNA}/Fidelity_{undamaged DNA}.

^eCalculated as [(k_p/K_{d, dNTP})_{damaged DNA}/Σ(k_p/K_{d, dNTP})_{damaged DNA}] × 100.

Previously, it has been shown that dG^C8-N-ABA bypass carried out by hPolη is error-prone in HEK293T cells ²⁰. Thus, we investigated the effect of dG^C8-N-ABA on the fidelity of hPolη during dG^C8-N-ABA bypass and the subsequent extension step. The fidelity of hPolη, defined as (k_p /K_{d, dNTP})_incorrect/[(k_p/K_{d, dNTP})_correct + (k_p/K_{d, dNTP})_incorrect], was calculated for the bypass (20/26-mer-dG^C8-N-ABA) and extension (21/26-mer-dG^C8-N-ABA) steps by using the single-turnover kinetic data of both correct and three incorrect nucleotide incorporations (Table 4). At these two steps, the fidelity of hPolη (Table 4) was determined to be in the ranges of 4.8 × 10⁻² to 1.6 × 10⁻¹ and 4.9 × 10⁻² to 1.9 × 10⁻¹, respectively. In comparison, the fidelity of hPolη with undamaged DNA was previously determined to be in the range of 10⁻² to 10⁻³ (Table S1) by pre-steady-state kinetics ¹³ and this range is consistent with the values determined using phage plaque assays ^{26; 27}. Thus, the presence of dG^C8-N-ABA significantly reduced the fidelity of hPolη at both the lesion bypass and subsequent extension steps, especially during misincorporations of dGTP (29-fold) and dATP (33-fold) opposite dG^C8-N-ABA (Table 4). For these two TLS steps, nucleotide incorporation probabilities, given by [(k_p/K_{d, dNTP})_{damaged DNA}/Σ(k_p/K_{d, dNTP})_{damaged DNA}] × 100, were also calculated. The probability of correct incorporation was reduced from 98% and 96% with undamaged DNA (Table S1) to 77% and 74% with damaged DNA (Table 4) for dG^C8-N-ABA bypass and the subsequent extension step, respectively. Notably, with damaged DNA, the probability of dATP misincorporation was 15 and 17% at the bypass and extension steps, respectively, and these probabilities are the highest among all misincorporations (Table 4). Consistently, the G→T transversion was found to occur at a high rate during TLS of dG^C8-N-ABA in vivo ^{5; 20; 28}.

3.4 Determination of Biphasic Kinetic Parameters at Polymerase Pause Sites

Nucleotide incorporation during TLS has often been shown to follow biphasic kinetics. Thus, we performed biphasic kinetic assays as described previously ^{13; 24; 25; 29} to test if hPolη followed the same trend during dG^C8-N-ABA bypass and the subsequent extension step. A pre-incubated solution of 5′-[³²P]-labeled DNA (30 nM) and hPolη (120 nM) was rapidly mixed with increasing concentrations of a correct nucleotide (25 to 1,000 μM) and an unlabeled trap DNA substrate 20/26-mer (5 μM). The trap DNA was added to a 150-fold molar excess over the labeled DNA so that any enzyme that had dissociated from the radiolabeled DNA substrate during the incubation time would be sequestered by the trap DNA. Thus, only labeled DNA product formed during a single enzyme binding event would be observed. The trap DNA concentration (5 μM) was determined to be sufficiently high enough to sequester any hPolη that had dissociated from 5′-[³²P]-labeled DNA ²⁴. During the incorporation of dCTP (1,000 μM) onto 20/26-mer-dG^C8-N-ABA, hPolη indeed followed biphasic kinetics (Figure 3A). The plot of reaction time versus product concentration was fit to Eq. 5 (Materials and Methods) to yield reaction amplitudes of A_f = 5.8 ± 0.8 nM (19%) and A_s = 8.6 ± 0.8 nM (28%) with rate constants of k_f = 7.2 ± 1.4 s⁻¹ and k_s = 0.42 ± 0.06 s⁻¹ for the fast and slow phases, respectively (Table 5). To test if the kinetic parameters were affected by dCTP concentration, similar biphasic kinetic experiments were performed individually with 300 and 600 μM dCTP. Indeed, as the dCTP concentration increased, the reaction rate constants (k_f and k_s) also increased while the reaction amplitudes remained relatively unchanged (Table 5). In addition, the biphasic kinetic experiments were also performed with 21/26-mer-dG^C8-N-ABA. As with 20/26-mer-dG^C8-N-ABA, a similar biphasic kinetic trend and a dNTP concentration-dependent increase in the rate constants were observed with 21/26-mer-dG^C8-N-ABA during the extension step (Figure 3B and Table 5). Interestingly, the dNTP concentration-dependent increase in the rate constants (k_f and k_s) had been observed previously with Dpo4 when it bypassed the dG^{C8-N-ABA 11} and dG^1,8 lesions ³⁰. However, our current study is the first one to report this concentration-dependence for lesion bypass catalyzed by a human Y-family DNA polymerase. Notably, kinetic parameters for the DNA trap experiments performed with undamaged 20/26-mer and 21/26-mer DNA are reported in Table S2, showing that even with undamaged DNA substrates, hPolη also displays biphasic kinetics of nucleotide incorporation.

Figure 3. Biphasic kinetics of dNTP incorporation in the presence of a DNA trap at 37 °C.

A pre-incubated solution of 120 nM hPolη and 30 nM 20/26-mer-dG^C8-N-ABA (A) or 21/26-mer-dG^C8-N-ABA (B) was rapidly mixed with 5 μM unlabeled 20/26-mer trap DNA and an indicated concentration of dCTP (A) or dGTP (B). After various incubation periods, the reaction was quenched with the addition of 0.37 M EDTA. Product formation was plotted against time and the data were fit to Eq. 5 to yield A_f and A_s, which are the reaction amplitudes of the fast and slow phases, respectively, as well as k_f and k_s, which are the rate constants of the fast and slow phases, respectively. The insets are magnifications of shorter time points to show the fast phase of product formation.

Table 5.

Kinetic parameters from biphasic kinetic assays with hPolη at 37 °C

[dNTP]	Af (nM)	kf (s−1)	As (nM)	ks (s−1)
20/26-mer-dGC8-N-ABA
300 μM dCTP	5.6 ± 0.8 (19%)	4.6 ± 1.3	8.3 ± 0.8 (28%)	0.32 ± 0.06
600 μM dCTP	5.5 ± 0.7 (18%)	6.3 ± 1.5	8.5 ± 0.7 (28%)	0.36 ± 0.05
1000 μM dCTP	5.8 ± 0.8 (19%)	7.2 ± 1.4	8.6 ± 0.8 (29%)	0.42 ± 0.06
21/26-mer-dGC8-N-ABA
25 μM dGTP	4.4 ± 0.7 (15%)	3.8 ± 1.5	9.2 ± 0.8 (31%)	0.082 ± 0.002
50 μM dGTP	4.4 ± 0.8 (15%)	5.2 ± 1.7	9.1 ± 0.6 (30%)	0.13 ± 0.02
90 μM dGTP	4.3 ± 0.6 (14%)	11 ± 4	9.3 ± 0.7 (31%)	0.16 ± 0.03

A_f and A_s are the reaction amplitudes of the fast and slow phases, respectively, while k_f and k_s are the rate constants of the fast and slow phases, respectively.

4. DISCUSSION

In this paper, we performed running start assays to determine which of the four human Y-family DNA polymerases were able to most efficiently catalyze TLS of dG^C8-N-ABA in vitro. Both hPolη and hPolκ were able to bypass dG^C8-N-ABA, while hPolι and hRev1 were completely stalled by the lesion (Figure 1). Consistently, small interfering RNA (siRNA)-mediated knockdown of hPolι had a minimal effect on overall TLS of dG^C8-N-ABA in HEK293T cells, suggesting that hPolι plays a very minor role during dG^C8-N-ABA bypass in vivo ²⁰. Surprisingly, siRNA-mediated knockdown of hRev1 proved to be detrimental to dG^C8-N-ABA bypass in HEK293T cells. Based on its inability to catalyze the bypass of dG^C8-N-ABA in vitro, hRev1 likely serves as a scaffold protein that synergistically interacts with proliferating cell nuclear antigen (PCNA) and other TLS polymerases to aid in TLS of dG^C8-N-ABA and likely other DNA lesions in vivo ^{13; 31; 32; 33}.

In contrast, hPolη and hPolκ were able to bypass dG^C8-N-ABA with similarly small t₅₀bypass values (Table 2), indicating potential roles for these polymerases in dG^C8-N-ABA bypass in vivo. Consistently, knockdown of these two polymerases in HEK293T cells significantly reduced the efficiency of TLS of dG^{C8-N-ABA 20}. Knockdown of hPolκ resulted in no change in the percentage of mutagenic dG^C8-N-ABA bypass events, suggesting that hPolκ carries out relatively error-free TLS of dG^{C8-N-ABA 20}. On the other hand, knockdown of hPolη resulted in a 39% decrease in mutation frequency, indicating that the bypass of dG^C8-N-ABA by hPolη in vivo is error-prone ²⁰. Taken together, the bypass of dG^C8-N-ABA by hPolη is likely a mechanism through which 3-NBA promotes lesion-induced mutagenesis and tumorigenesis. To determine a kinetic basis for mutagenic bypass of dG^C8-N-ABA by hPolη, we utilized pre-steady-state kinetic methods to investigate the effect of a site-specifically placed dG^C8-N-ABA on hPolη-catalyzed nucleotide incorporation at and near the lesion.

4.1 Kinetic basis for hPolη pausing caused by dG^C8-N-ABA

Overall, the binding affinity of hPolη to DNA (Table 3) is insignificantly affected by the presence of dG^C8-N-ABA and therefore does not significantly contribute to the polymerase stalling observed in Figure 1B. Instead, the decrease of nucleotide incorporation rate constants and the weakened nucleotide binding affinities are likely responsible for the observed accumulation of intermediate products. A similar kinetic trend is seen for the conversion of 18-mer→19-mer→20-mer as well as 19-mer→20-mer→21-mer. Based on the kinetic data in Table 4, the first conversion step has a higher rate constant (k_p) and also a higher efficiency ratio (k_p/K_{d, dNTP}) than the second, less efficient conversion step, leading to the observed accumulation of the 19-mer and 20-mer products (Figure 1B). The 19-mer and 20-mer accumulation was absent at later time points (30–1200 s) due to their eventual conversion to the 20-mer and 21-mer products, respectively. For the reaction series of the conversion of 20-mer→21-mer→22-mer and 21-mer→22-mer→23-mer, the first conversion step was less efficient than the second conversion step. This kinetic pattern led to accumulation of 20-mer and 21-mer as seen in Figure 1B. Taken together, it is clear that the conversion of 20-mer→21-mer is less efficient than both the preceding and subsequent reactions leading to the strong accumulation of the 20-mer product. For 23-mer, 24-mer, and 25-mer, their accumulation appeared the same in Figure 1B, indicating that the nucleotide incorporation in the consecutive conversion steps associated with these intermediate products were similarly efficient, and that hPolη likely resumed its normal polymerase efficiency after being several positions downstream from the lesion. Thus, the conversion rate and efficiency of 23-mer→24-mer, although not measured here, are likely as high as the conversion of 18-mer→19-mer (k_p = 66 s⁻¹, k_p/K_{d, dNTP} = 0.33 μM⁻¹s⁻¹) but higher than the conversion of 22-mer→23-mer (k_p = 9.1 s⁻¹, k_p/K_{d, dNTP} = 0.042 μM⁻¹s⁻¹), contributing to the observed 22-mer accumulation (Figure 1B).

4.2 Kinetic mechanism for dG^C8-N-ABA bypass and subsequent extension of the lesion bypass product

Our biphasic kinetic assays revealed that two distinct phases exist when hPolη incorporated correct dNTPs opposite dG^C8-N-ABA (Position −1) and one base pair downstream (Position 0) from the lesion. The fast phases were characterized by relatively small reaction amplitudes and fast rate constants while the slow phases had larger reaction amplitudes and slower rate constants (Table 5). Individual contributions of the amplitudes (A_f and A_s) and rate constants (k_f and k_s) were used to calculate the overall nucleotide incorporation rate constant in the presence of the DNA trap. For example, in the presence of 20/26-mer-dG^C8-N-ABA and 1,000 μM dGTP (Table 5), the sum of the individual contributions of the fast phase [(7.2 s⁻¹) × 19%] and slow phase [(0.42 s⁻¹) × 29%] gave an overall rate constant of 1.5 s⁻¹. As expected, 1.5 s⁻¹ is close to the k_obs of 1.4 s⁻¹, estimated using Eq. 4, 1.0 mM dCTP, and corresponding measured k_p and K_{d, dCTP} values (Table 4), for dCTP incorporation opposite dG^C8-N-ABA under single-turnover reaction conditions. Similarly, in the presence of 21/26-mer-dG^C8-N-ABA and 50 μM dGTP, the sum of the individual contributions of the fast phase [(5.2 s⁻¹) × 15%] and slow phase [(0.13 s⁻¹) × 30%] gave an overall nucleotide incorporation rate constant of 0.8 s⁻¹, which is similar to the k_obs of 0.5 s⁻¹, calculated using Eq. 4, 50 μM dGTP, and corresponding measured k_p and K_{d, dGTP} values (Table 4), for dGTP incorporation during the extension step. Thus, the biphasic kinetic feature of nucleotide incorporation at Positions −1 and 0 under single-turnover conditions was successfully deconvoluted into a fast and a slow phase through the DNA trap experiments (Figure 3). These two kinetic phases prompted us to propose a kinetic mechanism for dG^C8-N-ABA bypass and the subsequent extension by including the formation of productive (E•DNA_n^P) and non-productive (E•DNA_n^N) complexes (Scheme 1A). The portion of the hPolη population bound in the E•DNA_n^P will bind correct dNTP and rapidly elongate DNA with a fast rate constant of k_f. The fast phase k_f should be dNTP concentration-dependent and can reach its maximum value of k₁. The remainder of hPolη bound in the E•DNA_n^N complex will elongate DNA only after the slow conversion (k_e) of E•DNA_n^N → E•DNA_n^P. The slow phase rate constant of k_s is a function of both k_e and k₁ but should be limited by k_e. The bypass of a bulky dG^AP lesion by hPolη as well as the bypass of an abasic site, a cisplatin-d(GG) adduct, and a dG^AP lesion by Dpo4 have previously been proposed to follow this mechanism ^{13; 24; 25; 29}. If this mechanism is correct, the E•DNA_n^N → E•DNA_n^P conversion rate (k_e) and reaction amplitudes should not depend on dNTP concentration. However, both k_f and k_s increased with dNTP concentration at Positions −1 and 0 while both the fast and slow phase reaction amplitudes remained unchanged. To accommodate these observations, an alternative kinetic mechanism (Scheme 1B) was proposed for hPolη to bypass dG^C8-N-ABA and to subsequently extend the lesion bypass product. The same mechanism has been proposed by us for dG^C8-N-ABA bypass by Dpo4 ¹¹. This mechanism shows that k_s is actually a function of an exchange rate (k_e) and a slow nucleotide incorporation rate constant (k₂). If k_s had remained the same as dNTP concentration was increased, this would indicate that k_s were solely a function of k_e, and the mechanism could be simplified to reflect Scheme 1A. Because k_s increased with higher dNTP concentration, especially during the extension step, k_s was predominantly a function of k₂ during TLS of dG^C8-N-ABA by hPolη.

Scheme 1.

Proposed kinetic mechanism for dG^C8-N-ABA bypass and subsequent extension.

Notably, the total reaction amplitudes for the bypass and extension steps (48% and 46%, Table 5) are significantly smaller than the corresponding values with undamaged DNA (88% and 86%, Table S2). These differences indicate that hPolη may have bound to damaged DNA in a catalytically incompetent complex (E•DNA_n^D) where conformational change(s) of hPolη would not be able to produce a catalytically competent complex (Scheme 1B). In instances when any molecules of the E•DNA^D complex dissociated, the enzyme would bind to excess, unlabeled trap DNA resulting in no detectable product formation. Thus, the presence of such a dead-end binary complex can only be inferred by the sub-maximal product formation observed over the course of TLS of dG^C8-N-ABA by hPolη. Similar results have been observed in previous kinetic studies of lesion bypass under comparable conditions ^{13; 24; 25; 29}.

Although the hPolη•dG^C8-N-ABA-DNA binary structures are not available yet, some of their binding conformations could be inferred from an NMR structure of a 12-mer DNA duplex containing an embedded dG^C8-N-ABA:dC basepair ³⁴ and crystal structures of yeast Polη (yPolη) in complex with DNA containing a bulky 2-acetylaminofluorine lesion (dG^AAF) ³⁵. In the NMR structure, the ABA moiety of dG^C8-N-ABA is intercalated in the DNA duplex and takes the place of a base pair, with the damaged guanine in the syn conformation and extruded into the major groove, and with the opposing cytosine also pushed into the major groove ³⁴. Similarly, one yPolη•dG^AAF-DNA structure shows that the bulky adduct stacks on the primer/template junction base pair, occupies the polymerase active site, and precludes the incoming nucleotide from binding ³⁵. If such a binding conformation also exists in hPolη•dG^C8-N-ABA-DNA, it likely represents E•DNA_n^D in Scheme 1. Interestingly, in the other structure of yPolη•dG^AAF-DNA ³⁵, the AAF adduct is partially rotated out of the DNA helix to allow some excess for incoming dCTP and the structure likely represents the binding conformation of E•DNA_n^N (Scheme 1). Lastly, our ternary crystal structure of hPolη•dG^C8-N-ABA-DNA•dCTP shows that the adducted guanine forms a standard Watson-Crick pair with the incoming dCTP, with the ABA moiety is accommodated inside a hydrophobic cleft to the side of the active site of hPolη²¹. Such a binding conformation of hPolη•dG^C8-N-ABA-DNA is considered to be E•DNA_n^P (Scheme 1).

4.3 Error-prone bypass of dG^C8-N-ABA and subsequent extension of the lesion bypass product

During dG^C8-N-ABA bypass, hPolη showed a 10-fold reduced fidelity (4.8 × 10⁻²–1.6 × 10⁻¹) relative to undamaged DNA (2.0 × 10⁻³–1.2 × 10⁻², Table S1) with nucleotide incorporation preference following the order of dCTP > dATP > dGTP > dTTP (Table 4). Thus, the presence of dG^C8-N-ABA decreased the ability of hPolη to discriminate against incorrect nucleotides, especially dATP. Our kinetic studies show that hPolη will misincorporate dATP 15% of the time when bypassing dG^C8-N-ABA (Table 4). Interestingly, our previous studies have shown that in NER-proficient HEK293T cells, G→T transversions were the most common mutation resulting from dG^C8-N-ABA bypass, followed by G→A transitions and G→C transversions, albeit at much lower rates ²⁰. Furthermore, our previous studies have also demonstrated that siRNA-mediated knockdown of hPolη leads to a 39% decrease in mutation frequency in HEK293T cells, and strongly suggest that hPolη plays a major role in the error-prone bypass of dG^C8-N-ABA in vivo ²⁰. Therefore, these observations from our published cell-based assays ²⁰ are consistent with the results of our kinetic studies with hPolη (Table 4).

Highlights.

• hPolη and hPolκ efficiently bypassed a site-specifically placed dG^C8-N-ABA lesion

• The fidelity of hPolη decreased during dG^C8-N-ABA bypass and extension

• hPolη predominantly misincorporated dATP during dG^C8-N-ABA bypass

• Nucleotide incorporation at hPolη pause sites followed biphasic kinetics

• A kinetic basis was established for mutagenic bypass of dG^C8-N-ABA in vivo