Significant impact of divalent metal ions on the fidelity, sugar selectivity, and drug incorporation efficiency of human PrimPol

Abstract

Human PrimPol is a recently discovered bifunctional enzyme that displays DNA template-directed primase and polymerase activities. PrimPol has been implicated in nuclear and mitochondrial DNA replication fork progression and restart as well as DNA lesion bypass. Published evidence suggests that PrimPol is a Mn²⁺-dependent enzyme as it shows significantly improved primase and polymerase activities when binding Mn²⁺, rather than Mg²⁺, as a divalent metal ion cofactor. Consistently, our fluorescence anisotropy assays determined that PrimPol binds to a primer/template DNA substrate with affinities of 29 and 979 nM in the presence of Mn²⁺ and Mg²⁺, respectively. Our pre-steady-state kinetic analysis revealed that PrimPol incorporates correct dNTPs with 100-fold higher efficiency with Mn²⁺ than with Mg²⁺. Notably, the substitution fidelity of PrimPol in the presence of Mn²⁺ was determined to be in the range of 3.4×10⁻² to 3.8×10⁻¹, indicating that PrimPol is an error-prone polymerase. Furthermore, we kinetically determined the sugar selectivity of PrimPol to be 57–1,800 with Mn²⁺ and 150–4,500 with Mg²⁺, and found that PrimPol was able to incorporate the triphosphates of two anticancer drugs (cytarabine and gemcitabine), but not two antiviral drugs (emtricitabine and lamivudine).

1. INTRODUCTION

Since the discovery of human PrimPol in 2013 ^{1; 2; 3}, there has been extensive research to determine its precise role during genome replication in vivo. PrimPol is a bifunctional enzyme that is able to prime single-stranded DNA and subsequently catalyze primer extension via a polymerase-like, nucleotidyl transfer activity. PrimPol is only the second primase to be identified in humans and contains two distinct domains: an evolutionarily conserved Archaeo-Eukaryotic Primase (AEP) domain, and a UL52-like zinc finger domain that is required for its primase activity ⁴. Interestingly, PrimPol is found in both the nucleus and the mitochondria and has been implicated in replication fork progression and restart, as well as DNA lesion bypass ^{1; 3; 5; 6}. Gene silencing of PrimPol in human cells causes profound arrest in mitochondrial DNA (mtDNA) synthesis, decreased replication fork progression rates, and increased replication protein A (RPA) foci, which is indicative of replicative stress ^{1; 3; 5; 7}. In PrimPol^−/− derived mouse embryonic fibroblasts, the presence of chromosomal aberrations such as chromatid breaks ^{2; 3} and micronuclei ⁵ indicate that PrimPol is essential in the maintenance of genomic integrity. PrimPol is only the second enzyme with DNA polymerase activity to be identified in mitochondria, the other being DNA polymerase γ which is responsible for rapid and faithful mtDNA replication ^{8; 9}. Furthermore, PrimPol is found to play a key role in UV damage resistance, as cells lacking PrimPol have increased sensitivity to UV-C radiation ³ and GFP-tagged PrimPol was shown to be rapidly recruited to chromatin in cells treated with UV-A radiation ⁵. Consistently, PrimPol can bypass common UV-induced lesions such as cis-syn thymine-thymine cyclobutane pyrimidine dimers and (6–4) photoproducts in vitro ^{3; 5}. PrimPol has also been shown to bypass common single-base lesions including 8-oxoguanine ⁶ and abasic sites ^{1; 5; 6}. During DNA synthesis with undamaged DNA templates, PrimPol makes one error per 10⁴-10⁵ nucleotide incorporations in the presence of Mg²⁺ as measured by lacZ, HSV-tk, and M13mp2 forward mutation assays and the errors are primarily base insertions and deletions, not substitutions ^{4; 10; 11}. In vitro, the primase and polymerase activities of PrimPol ^{4; 6; 11} are shown to be differentially influenced by Mg²⁺ and Mn²⁺, the known divalent metal ion cofactors utilized by other primases and polymerases for catalysis. However, such an effect of the divalent metal ion cofactor has not been quantitatively and rigorously analyzed. To fill the void, we employed fluorescence anisotropy and pre-steady-state kinetic assays to determine the effect of the divalent metal ion cofactor on the DNA binding affinity and nucleotide incorporation efficiency of human PrimPol. In addition, we used pre-steady-state kinetics to determine the ability of human PrimPol to discriminate against ribonucleotides (rNTPs) and to incorporate the triphosphates of four nucleoside analog drugs in the presence of Mn²⁺ or Mg²⁺.

2. Materials and Methods

2.1. Materials.

Reagents were purchased from the following companies: OptiKinase from USB corp., [γ−³²P]-ATP from PerkinElmer, deoxyribonucleotides (dNTPs) and rNTPs from Bioline, 2′-deoxy-2′,2′-difluorodeoxycytidine 5′-triphosphate (GemCTP) and 2′-aracytidine 5′-triphosphate (AraCTP) from TriLink BioTechnologies, 5′-triphosphate of lamivudine ((−)3TC-TP) and emtricitabine ((−)FTC-TP) from Gilead Sciences, and oligonucleotides from Integrated DNA Technologies.

2.2. Expression and Purification of Human PrimPol.

Human PrimPol containing an N-terminal 6x-histidine tag was subcloned as previously described³, and transformed into Escherichia coli Rosetta (DE3) competent cells. A single colony was used to inoculate 100 mL of LB media (30 μg/mL kanamycin and 34 μg/mL chloramphenicol) and the culture was grown at 37 °C overnight to an OD₆₀₀ of 1.5. The starter culture was then used to inoculate 6 × 1 L of fresh LB media (30 μg/mL kanamycin and 34 μg/mL chloramphenicol) and the overexpression cultures were grown at 37 °C to an OD₆₀₀ of 0.8 followed by rapid cooling on ice. The cultures were induced at an OD₆₀₀ of 1.0 with 0.1 mM Isopropyl β-D-1-thiogalactopyranoside and allowed to grow at 16 °C for an additional 15 hours following induction before pelleting by centrifugation. The cell pellet was re-suspended in buffer A (50 mM Tris-HCl pH 7.5, 300 mM NaCl, 10% glycerol, 0.1% β-mercaptoethanol, 10 mM imidazole, 0.01 mM EDTA, and 0.1% IGEPAL) and supplemented with EDTA-free Protein Inhibitor Cocktail tablets (Roche) and 1 mM phenylmethanesulfonylfluoride. Cells were lysed by three passages through a French pressure cell press at 15,000 psi. The soluble fraction was isolated by ultracentrifugation at 40,000 rpm for 40 min. The cleared lysate was then incubated with charged nickel nitrilotriacetic acid (Ni-NTA) resin for 3 hr at 4 °C. The Ni-NTA beads were packed into a tricorn FPLC column and were washed with 20 column volumes (CV) of buffer A and further washed with 10 CV of 4% buffer B (buffer A containing 500 mM imidazole). Protein was eluted with a linear gradient of 4 to 100% buffer B over 15 CV and fractions were analyzed by SDS-PAGE. Protein-containing fractions were pooled and then loaded onto a HiTrap Heparin HP column (GE Healthcare). The column was washed with 5 CV buffer C (buffer A without imidazole) and then with 10 CV of 10% buffer D (buffer C containing 1 M NaCl). The protein was eluted with a gradient of 10 to 100% buffer D over 10 CV. Stepwise dialysis was performed overnight at 4°C to decrease the NaCl concentration of the eluted protein solution from 700 mM to 125 mM. The purest samples were pooled and concentrated to 500 μL using an Amicon Ultra-15 Centrifugal filter (Millipore). The protein sample was further purified in a Superdex 200 size exclusion chromatography column (GE Healthcare) to isolate full-length human PrimPol (66.5 kDa). Fractions were analyzed via SDS-PAGE and the most pure samples were pooled and dialyzed against storage buffer (50 mM Tris-HCl, pH 7.5, 400 mM NaCl, 50% glycerol, 1 mM DTT, and 0.1 mM EDTA). Using the predicted extinction coefficient (ε₂₈₀ = 77,655 M⁻¹cm⁻¹), the final concentration of purified PrimPol was determined by UV-Vis spectroscopy at 280 nm and the final yield of protein was 1 mg/L of culture.

2.3. Radiolabeling and annealing DNA substrates.

All oligonucleotides were purified via PAGE and reverse-phase chromatography (Sep-Pak classic C-18 cartridges). The 21-mer primer used for the single-turnover assays was 5′-[³²P]-labeled by incubating it with [γ−³²P]-ATP and OptiKinase (USB) for 3 hr at 37 °C. The reaction was terminated by heating at 95 °C for 2 min to denature Optikinase. The radiolabeled primer was purified from any unreacted [γ−³²P]-ATP using a Bio-spin 6 column (Bio-Rad). The primer was then annealed to a DNA template (Table 1) in a 1:1.35 molar ratio by first heating the reaction mixture to 95 °C for 5 min and then slowly cooling the mixture to room temperature overnight.

Table 1.

21/41-mer DNA substrates

D-1

5′-CGCAGCCGTCCAACCAACTCA-3′

3′-GCGTCGGCAGGTTGGTTGAGTAGCAGCTAGGTTACGGCAGG-5′

D-6

5′-CGCAGCCGTCCAACCAACTCA-3′

3′-GCGTCGGCAGGTTGGTTGAGTGGCAGCTAGGTTACGGCAGG-5′

D-7

5′-CGCAGCCGTCCAACCAACTCA-3′

3′-GCGTCGGCAGGTTGGTTGAGTTGCAGCTAGGTTACGGCAGG-5′

D-8

5′-CGCAGCCGTCCAACCAACTCA-3′

3′-GCGTCGGCAGGTTGGTTGAGTCGCAGCTAGGTTACGGCAGG-5′

2.4. Fluorescence anisotropy titration.

The Cy3-labeled DNA substrate 17/30-mer (Figure 1A, 30 nM) was titrated with increasing amounts of PrimPol and the anisotropy was monitored using a FluoroMax-4 (Horiba). Assays were carried out at 25 °C in buffer E [50 mM Tris–HCl (pH = 7.5 at 25 °C), 50 mM NaCl, 0.01 mM EDTA] without divalent metal ions or in the presence of 5 mM MnCl₂, or 5 mM MgCl₂ where indicated. Excitation and emission for the Cy3 fluorophore were set to 540 and 568 nm, respectively, with a 10 nm slit width and 2 s integration time. The data obtained from anisotropy measurements were fit to Eq. 1:

$$\mathrm{\Delta }A=\left(\mathrm{\Delta }{A}_T/2D_0\right)\times \left\{\left(K_{\text{d},\text{DNA}}+D_0+E_0\right)–{\left[{\left(K_{\text{d},\text{DNA}}+D_0+E_0\right)}^2-4E_0{D}_0\right]}^{\mathrm{½}}\right\}$$ Eq. 1

FIGURE 1.

Determination of the binding affinity of PrimPol to DNA in the presence or absence of a divalent metal ion. (A) Cy3-labeled DNA 17/30-mer. Increasing amounts of PrimPol were titrated into a fixed concentration of Cy3-labeled 17/30-mer (30 nM). The plot of the concentration of the binary complex PrimPol●DNA versus the concentration of PrimPol was fit to Eq. 1 (Materials and Methods) to obtain K_d,DNA. The K_d,DNA value was 41 ± 5 nM in the absence of any divalent metal ions (B), 29 ± 5 nM in the presence of 5 mM MnCl₂ (B), 979 ± 119 nM in the presence of 5 mM MgCl₂ (C).

Where ΔA is the change in anisotropy, ΔA_T is the maximum anisotropy change, D₀ and E₀ are the initial concentrations of DNA and PrimPol, respectively, and K_d,DNA is the equilibrium dissociation constant of the PrimPol●DNA binary complex.

2.5. Single-turnover kinetic assays.

PrimPol (300 nM) and 5′-[³²P]-labeled DNA (30 nM) were pre-incubated at 37 °C for 5 min in buffer E (pH = 7.5 at 37 °C) containing 5 mM DTT and 0.1 μg/ml BSA, before mixing with increasing concentrations of a single dNTP. Reaction mixtures (10 μL) were quenched with EDTA to a final concentration of 0.37 M at increasing time points. Reaction products were separated via PAGE (17% polyacrylamide, 8 M urea) and visualized using a PhosphorImager plate (Amersham Biosciences) and TyphoonTrio scanner (GE Healthcare). The product was quantified using ImageQuant software (Molecular Dynamics). The plot of product concentration versus time from each time course was fit to a single-exponential equation (Eq. 2) using non-linear regression software KaleidaGraph (Synergy Software) to determine the observed nucleotide incorporation rate constant (k_obs).

$$\left[\text{Product}\right]=\text{A}\left[1-\text{exp}\left(-k_{\text{obs}}t\right)\right]$$ Eq. 2

where A is the reaction amplitude, which is equal to the initial concentration of the PrimPol●DNA binary complex. The k_obs values obtained from Eq. 2 were plotted against the respective concentrations of dNTP and the data were fit to a hyperbolic equation (Eq. 3):

$$k_{\text{obs}}=k_{\text{p}}\left[\text{dNTP}\right]/\left(\left[\text{dNTP}\right]+K_{\text{d}}\right)$$ Eq. 3

Where K_d is the apparent dissociation constant of dNTP from the PrimPol●DNA●dNTP ternary complex and k_p is the maximum rate constant of dNTP incorporation.

3. Results

3.1. Binding affinity of human PrimPol to DNA in the presence of Mn²⁺ or Mg²⁺.

Human PrimPol was expressed in Escherichia coli and purified to >98% purity through column chromatography (Materials and Methods). To investigate the effect of divalent metal ions on the binding affinity of human PrimPol to DNA, we employed a fluorescence anisotropy assay (Materials and Methods) and determined the dissociation equilibrium constant (K_d,DNA) for the PrimPol●DNA binary complex in the presence, or absence, of 5 mM MnCl₂ or MgCl₂. Similar concentrations of the divalent metal ions have been used in published studies of human PrimPol ^{4; 6; 11}. In the absence of any supplemental divalent metal ions, the K_d,DNA for the binding of PrimPol to the Cy3-labeled DNA substrate 17/30-mer (Figure 1A) was measured to be 41 ± 5 nM (Figure 1B). In the presence of 5 mM Mn²⁺ or Mg²⁺, the K_d,DNA value was determined to be 29 ± 5 or 979 ± 119 nM, respectively (Figures 1B and 1C). Thus, the affinity (K_d,DNA = 979 nM) of our prepared PrimPol from E. coli to DNA in the presence of 5 mM Mg²⁺ is ~24-fold lower than in the absence (K_d,DNA = 41 nM) while the presence of 5 mM Mn²⁺ slightly enhanced the affinity (K_d,DNA = 29 nM). This suggests that the presence of a molar excess of Mg²⁺ likely disrupted the interaction between the Zn²⁺ and its coordinating amino acid residues (Cys-His-Cys-Cys), which subsequently altered the binding of the zinc finger domain to the single-stranded template region in the DNA substrate ⁴ (Figure 1A) and weakened the overall DNA binding affinity of PrimPol (Figure 1C). Previously, similarly prepared human PrimPol to ours was found to contain Zn²⁺ in its C-terminal zinc finger domain with an occupancy of ~80% measured by using inductively coupled plasma mass spectrometry (ICP-MS) ⁴. This suggests that the presence of 5 mM Mn²⁺, a transition divalent metal ion known to bind to zinc finger motifs ^{12; 13}, likely bound to 20% of our prepared PrimPol, which lacked Zn²⁺ based on the published ICP-MS assay results ⁴, and slightly improved the DNA binding affinity of our PrimPol (Figure 1B).

3.2. Correct nucleotide incorporation efficiency in the presence of Mn²⁺ or Mg²⁺.

Besides DNA binding, Mn²⁺ and Mg²⁺ may impact the nucleotide incorporation efficiency of PrimPol differently since its AEP domain binds to divalent metal ions. To examine this possibility, single-turnover kinetic assays were performed by mixing a pre-incubated solution of PrimPol (300 nM) and 5′-[³²P]-labeled 21/41-mer DNA substrate D-7 (30 nM, Table 1) with increasing concentrations of correct dATP in the presence of 5 mM Mn²⁺ for various times before being quenched with 0.37 M EDTA (Materials and Methods). After the reaction products were separated and quantitated and the time courses were analyzed, we determined the maximum dATP incorporation rate constant (k_p) of 0.066 ± 0.003 s⁻¹ and the apparent equilibrium dissociation constant (K_d) of 11 ± 1 μM for dATP binding (Figure 2). The substrate specificity (k_p/K_d) of dATP was then calculated to be 6.0×10⁻³ μM⁻¹ s⁻¹ (Table 2).

FIGURE 2.

Determination of kinetic parameters for dATP incorporation opposite dT in the presence of Mn²⁺ at 37 °C. (A) A pre-incubated solution of PrimPol (300 nM) and D-7 DNA substrate (30 nM, Table 1) was mixed with increasing concentrations of dATP in the presence of 5 mM Mn²⁺ at 37 °C. The product concentrations were plotted against reaction times and each time course was fit to Eq. 2 (Materials and Methods) to yield k_obs. (B) The k_obs values were plotted against respective concentrations of dATP and the plot was fit to Eq. 3 (Materials and Methods) to yield a k_p of 0.066 ± 0.003 s⁻¹ and a K_d of 11 ± 1 μM.

TABLE 2.

Pre-steady-state kinetic parameters for nucleotide incorporation onto 21/41-mer DNA substrates catalyzed by PrimPol in the presence of 5 mM Mn²⁺ at 37 °C.

dNTP	kp (s−1)	Kd (μM)	kp/Kd (μM−1 s−1)	Fidelitya	Sugar Selectivityb
D-1 (Template dA)
dTTP	0.096 ± 0.004	17 ± 2	5.8 × 10−3	-
dATP	0.0083 ± 0.0001	4.1 ± 0.3	2.0 × 10−3	2.6 × 10−1
dCTP	0.014 ± 0.0009	13 ± 4	1.1 × 10−3	1.7 × 10−1
dGTP	0.0047 ± 0.0007	13 ± 4	3.6 × 10−4	6.0 × 10−2
rUTP	0.0018 ± 0.0002	561 ± 160	3.2 × 10−6	5.5 × 10−4	1800
D-6 (Template dG)
dCTP	0.060 ± 0.005	16 ± 4	3.8 × 10−3	-
dATP	0.011 ± 0.00005	20 ± 3	5.5 × 10−4	1.3 × 10−1
dGTP	0.035 ± 0.003	27 ± 8	1.3 × 10−3	2.6 × 10−1
dTTP	0.0058 ± 0.0005	14 ± 6	4.2 × 10−4	9.9 × 10−2
rCTP	0.0092 ± 0.0002	136 ± 7	6.7 × 10−5	1.7 × 10−2	57
D-7 (Template dT)
dATP	0.066 ± 0.003	11 ± 1	6.0 × 10−3	-
dCTP	0.013 ± 0.002	17 ± 9	7.6 × 10−4	1.1 × 10−1
dGTP	0.017 ± 0.001	22 ± 4	7.7 × 10−4	1.1 × 10−1
dTTP	0.0029 ± 0.0002	3 ± 1	1.0 × 10−3	1.4 × 10−1
rATP	0.0047 ± 0.0006	744 ± 234	6.3 × 10−6	1.0 × 10−3	950
D-8 (Template dC)
dGTP	0.036 ± 0.002	16 ± 3	2.3 × 10−3	-
dATP	0.0049 ± 0.0003	16 ± 3	3.0 × 10−4	1.2 × 10−1
dCTP	0.014 ± 0.001	10 ± 3	1.4 × 10−3	3.8 × 10−1
dTTP	0.0024 ± 0.0004	30 ± 7	8.0 × 10−5	3.4 × 10−2
rGTP	0.0015 ± 0.0001	110 ± 29	1.4 × 10−5	6.1 × 10−3	160

^aCalculated as (k_p/K_d)_incorrect/[(k_p/K_d)_correct + (k_p/K_d)_incorrect]

^bCalculated as (k_p/K_d)_dNTP/(k_p/K_d)_rNTP

Similar single-turnover kinetic assays were performed for correct dTTP, dCTP, dGTP incorporation onto D-1, D-6, and D-8 (Table 1), respectively, and the kinetic parameters are listed in Table 2. In the presence of 5 mM Mn²⁺, correct dNTP incorporation occurs with a k_p in the range of 0.036–0.096 s⁻¹, a K_d of 11–17 μM, and a k_p/K_d of (2.3–6.0)x10⁻³ μM⁻¹s⁻¹ (Table 2). For comparison, we performed similar single-turnover kinetic assays for a single correct dNTP incorporation onto a corresponding DNA substrate (Table 1) by PrimPol in the presence of 5 mM Mg²⁺, e.g. dGTP onto D-8 (Figure S1). The kinetic data (Table 3) show that PrimPol incorporates correct dNTPs in the presence of 5 mM Mg²⁺ with a k_p of 0.011–0.020 s⁻¹, a K_d of 262–895 μM, and a k_p/K_d of (2.2–5.0)x10⁻⁵ μM⁻¹s^-1. Thus, on average, the presence of 5 mM Mg²⁺ makes PrimPol about 100-fold less efficient during correct dNTP incorporation than in the presence of 5 mM Mn²⁺ (Tables 2 and 3).

TABLE 3.

Pre-steady-state kinetic parameters for dNTP incorporation onto 21/41-mer DNA substrates catalyzed by PrimPol in the presence of 5 mM Mg²⁺ at 37 °C.

dNTP	kp (s−1)	Kd (μM)	kp/Kd (μM−1 s−1)	Fidelitya
D-1 (Template dA)
dTTP	0.020 ± 0.002	526 ± 99	3.8 × 10−5
D-6 (Template dG)
dCTP	0.013 ± 0.0008	262 ± 55	5.0 × 10−5
D-7 (Template dT)
dATP	0.011 ± 0.0007	388 ± 65	2.8 × 10−5	-
dCTP	Not determined	Not determined	3.1 × 10−7	1.1 × 10−2
dGTP	Not determined	Not determined	1.3 × 10−8	4.6 × 10−4
dTTP	Not determined	Not determined	1.7 × 10−8	6.1 × 10−4
D-8 (Template dC)
dGTP	0.020 ± 0.0006	895 ± 60	2.2 × 10−5

^aCalculated as (k_p/K_d)_incorrect/[(k_p/K_d)_correct + (k_p/K_d)_incorrect]

3.3. Substitution fidelity of human PrimPol in the presence of Mn²⁺ or Mg²⁺.

To kinetically estimate the incorporation fidelity of PrimPol, we employed similar single-turnover assays to determine the pre-steady-state kinetic parameters (Table 2) for the 12 possible incorrect incorporations in the presence of 5 mM Mn²⁺, e.g. dGTP incorporation opposite dT (Figure S2). Similar approaches have been used by us to determine the fidelities of other DNA polymerases ^{14; 15; 16; 17; 18}. The kinetic data in Table 2 demonstrate that PrimPol incorporates incorrect dNTPs in the presence of 5 mM Mn²⁺ with a k_p of 0.0024–0.035 s⁻¹, a K_d of 3–30 μM, and a k_p/K_d of (8.0×10⁻⁵-2.0×10⁻³) μM⁻¹s^-1. Relative to the kinetic parameters with correct dNTP incorporations (see above), incorrect dNTP incorporations occur with lower k_p and k_p/K_d values but with comparable K_d values. The substitution fidelity of PrimPol, defined as (k_p/K_d)_incorrect/[(k_p /K_d)_incorrect + (k_p/K_d)_correct], in the presence of 5 mM Mn²⁺ is calculated to be in the range of 3.4×10⁻² to 3.8×10⁻¹ (Table 2). This fidelity indicates that PrimPol tends to make one substitution error out of every 4 to 29 nucleotide incorporations in the presence of Mn²⁺. This substitution fidelity is about 100-fold lower than the fidelity range of 10⁻²-10⁻⁴ previously estimated by Zafar et. al. through steady-state kinetic methods in the presence of 10 mM Mn^{2+ 6}. The large discrepancy is likely because the steady-state kinetic parameters are complicated by fast DNA product dissociation ¹⁹ and are inaccurately determined when nucleotide incorporation, e.g. misincorporation, is inefficient and yields little products. The low fidelity of PrimPol can be illustrated in the gel images of the time courses of incorrect dNTP incorporations, especially when multiple dCTP misincorporations occurred and even made the products longer than the template 41-mer in the presence of Mn²⁺ (Figure S3). Similar template-independent dCTP misincorporations by PrimPol have been observed previously ^{4; 11}.

To examine if the switch from Mn²⁺ to Mg²⁺ increases the fidelity of PrimPol, we determined the kinetic parameters for individual misincorporations onto D-7 (Table 1) in the presence of 5 mM Mg²⁺. Opposite dT, each incorrect dNTP was inefficiently incorporated and only a small amount of products were formed even after 3 hours of incubation (Figure S4C). As a result, we instead determined the k_p/K_d value, not individual k_p and K_d values, for each dNTP misincorporation and calculated the fidelity (10⁻²-10⁻⁴) of PrimPol (Table 3). Notably, the dCTP:dT misincorporation occurred with a 100-fold lower fidelity value than the other two misincorporations (Table 3). This is likely due to two consecutive dCTP incorporations onto D-7 (Table 1) with the second being a correct incorporation, while dGTP or dTTP was only misincorporated once onto D-7 (Figure S4), leading to a 100-fold difference in their k_p/K_d values (Table 3). Thus, the substitution fidelity of PrimPol in the presence of Mg²⁺ is estimated to be ~10⁻⁴, which is 100–1,000 fold higher than its fidelity in the presence of Mn²⁺ (Table 2). Our fidelity value of ~10⁻⁴ is also comparable with the fidelity values estimated from lacZ, HSV-tk, and M13mp2 forward mutation assays in the presence of Mg^{2+ 4; 10; 11}.

3.4. Sugar selectivity of human PrimPol in the presence of Mn²⁺ and Mg²⁺.

Eukaryotic primases are responsible for catalyzing de novo primer synthesis on single-stranded DNA for the initiation of leading and lagging strand synthesis²⁰. Specifically, primases utilize rNTPs to catalyze de novo RNA primer synthesis (7–10 bases), which can then be extended by a specialized DNA polymerase in the presence of dNTPs to generate a hybrid 5′-RNA-DNA-3′ primer for DNA replication²¹. Contrastingly, human PrimPol prefers to utilize dNTPs for both de novo primer synthesis^{1; 3} and subsequent primer extension¹¹. Replicative polymerases can extend PrimPol-generated DNA primers during lesion bypass and replication fork restart, as there is no need for further processing of the RNA strands²². To quantitatively determine how much human PrimPol discriminates against rNTPs, we performed single-turnover kinetic assays (Materials and Methods) to investigate matched rNTP incorporation onto DNA/DNA substrates (Table 1) in the presence of either 5 mM MnCl₂ or 5 mM MgCl₂ (data not shown). PrimPol incorporated matched rNTPs with a range of k_p values from 0.0015–0.0092 s⁻¹ and a range of K_d values of 110–744 μM, yielding calculated k_p/K_d values of 3.2×10⁻⁶ to 6.8×10⁻⁵ μM⁻¹s⁻¹ in the presence of 5 mM Mn²⁺ (Table 2). The ratio of catalytic efficiencies of correct dNTPs over matched rNTPs [(k_p/K_d)_dNTP/(k_p/K_d)_rNTP] (Table 2), was calculated to give the sugar selectivity of PrimPol, which has been done similarly for other DNA polymerases ^{23; 24; 25; 26; 27}. The sugar selectivity of PrimPol, was found to be 57–1,800 in the presence of 5 mM Mn²⁺ (Table 2). Interestingly, both a decrease in the k_p (24-fold on average) and an increase in the K_d (30-fold on average) are responsible for rNTP discrimination by human PrimPol.

In the presence of Mg²⁺, the individual kinetic parameters (k_p and K_d) could not be determined due to extremely slow incorporation of matched rNTPs (see above). Instead, we obtained k_obs values for each matched rNTP at 1 mM concentrations, and found that PrimPol incorporated matched rNTPs in the range of (6.8–9.5)x10⁻⁵ s⁻¹ (Table S2). Since the k_obs for correct incorporation of each dNTP at 1 mM was determined above, the ratios of observed nucleotide incorporation rate constants (k_{obs, dNTP}/k_{obs, rNTP}) were calculated to be in the range of 105–260 (Table S2). Thus, PrimPol incorporated correct dNTPs 150-fold (on average) more rapidly than matched rNTPs in the presence of Mg²⁺. Although we could not determine the K_{d, rNTP}/K_{d, dNTP} ratios in the presence of Mg²⁺, we assume they are similar to the average 30-fold binding affinity ratio determined in the presence of Mn²⁺ (see above). Together, our kinetic data indicate that the sugar selectivity of human PrimPol in the presence of Mg²⁺ will be greater than 150 and could be as high as 4,500.

3.5. Incorporation efficiencies of the triphosphates of four cytidine analog drugs gemcitabine, cytarabine, emtricitabine, and lamivudine by human PrimPol.

Chain-terminating nucleoside analogs have been developed as chemotherapeutic drugs to treat cancers and antiviral drugs to combat viral infections. For example, gemcitabine (GemC) and cytarabine (AraC) have been used to treat pancreatic adenocarcinoma²⁸, non-muscle invasive bladder cancer²⁹, non-small cell lung cancer^{30; 31; 32}, and acute myeloid leukemia³³ while lamivudine ((−)3TC) and emtricitabine ((−)FTC) are widely prescribed drugs against human immunodeficiency virus (HIV) and Hepatitis B (HBV) infections ^{34; 35; 36}. The chain-terminating nucleoside analog drugs are cellularly activated into their triphosphate forms which compete against natural dNTPs for incorporation and terminate cellular and viral genomic replication once incorporated. Besides clinical efficacy, the drugs also cause various clinical toxicities and some of them have not been explained well at a molecular level. Considering human PrimPol is a recently discovered enzyme, it could incorporate the triphosphates of various nucleoside analog drugs and contribute to their observed clinical toxicities. Here, we investigated the abilities of human PrimPol to incorporate the triphosphates of four cytidine analogs (Figure S5) by performing similar single-turnover kinetic assays (Materials and Methods) in the presence of 5 mM MnCl₂ or MgCl₂ (data not shown). The measured kinetic data are listed in Table 4. In the presence of 5 mM Mn²⁺, AraCTP (K_d = 21 ± 4 μM) was bound by PrimPol with a similar affinity as dCTP (K_d = 16 ± 4 μM). However, the incorporation of AraCTP (k_p = 0.0057 ± 0.0004 s⁻¹) was 10-fold slower than dCTP (k_p = 0.060 ± 0.005 s⁻¹), leading to a discrimination value, calculated as [(k_p/K_d)_dCTP/(k_p/K_d)_analog], of 14 (Table 4). In comparison, GemCTP was incorporated with similar k_p and K_d values as dCTP, leading to a discrimination value of 2.9 (Table 4).

TABLE 4.

Pre-steady-state kinetic parameters for incorporation of dCTP analogs onto D-6 (Table 1) catalyzed by human PrimPol at 37 °C.

NTP	kp (s−1)	Kd (μM)	kp/Kd (μM−1 s−1)	Discriminationa
In the presence of 5 mM Mn2+
AraCTP	0.0057 ± 0.0004	21 ± 4	2.7 × 10−4	14
GemCTP	0.058 ± 0.0006	45 ± 2	1.3 × 10−3	2.9
(−)3TC-TP	Not determined	Not determined	1.2 × 10−6	3,200
(−)FTC-TP	Not determined	Not determined	9.0 × 10−7	4,200
In the presence of 5 mM Mg2+
AraCTP	0.0076 ± 0.0003	316 ± 44	2.4 × 10−5	2.1
GemCTP	0.0071 ± 0.0006	1,380 ± 255	5.1 × 10−6	9.8
(−)3TC-TP	No observed incorporation
(−)FTC-TP	No observed incorporation

^aCalculated as (k_p/K_d)_dCTP/(k_p/K_d)_analog with the (k_p/K_d)_dCTP values from Tables 2 and 3.

In the presence of 5 mM Mg²⁺ (Tables 2–4), both AraCTP and GemCTP were incorporated by PrimPol with 2-fold lower k_p values than dCTP (Tables 3 and 4). However, the K_d values of AraCTP and dCTP are comparable while GemCTP was bound with a 5-fold higher K_d (Tables 3 and 4). The discrimination factors were calculated to be 2.1 and 9.8 for AraCTP and GemCTP, respectively (Table 4).

For (−)3TC-TP and (−)FTC-TP, the two antiviral drugs with L-stereochemistry, we also kinetically investigated their incorporation by human PrimPol. In the presence of 5 mM Mn²⁺, both (−)3TC-TP and (−)FTC-TP were incorporated inefficiently and only their k_p/K_d values were determined to be 1.2×10⁻⁶ and 9.0×10⁻⁷ μM⁻¹s⁻¹, respectively (Table 4). In the presence of 5 mM Mg²⁺, these L-analogs were not incorporated by PrimPol even after 3 hours at 37 °C (data not shown).

4. Discussion

4.1. Mn²⁺, not Mg²⁺ is likely the divalent metal ion cofactor for the AEP domain of human PrimPol.

In this paper, we employed a fluorescence anisotropy assay and determined the K_d,DNA values for the PrimPol●DNA binary complex in the presence or absence of 5 mM Mn²⁺ or Mg²⁺ (Figure 1). Strikingly, in the presence of Mg²⁺, the affinity of PrimPol to DNA (K_d,DNA = 979 nM) is significantly lower than other human DNA polymerases (Table S1) and the switch from Mg²⁺ to Mn²⁺ increased the DNA binding affinity (K_d,DNA = 29 nM) of PrimPol by 34-fold. The 29 nM DNA binding affinity is likely caused by the C-terminal zinc finger domain of PrimPol which binds to the single-stranded template region in a primer/template substrate and increases the DNA binding affinity of PrimPol ⁴. Since Mn²⁺ can substitute Zn²⁺ and bind to zinc finger motifs ^{12; 13} while Mg²⁺ cannot, the presence of a large molar excess of Mg²⁺ (5 mM) in our fluorescence anisotropy assay (Figure 1) likely disrupted the zinc finger domain of PrimPol and significantly decreased its DNA binding affinity. Furthermore, the substrate specificity (k_p/K_d) values for correct dNTP incorporations determined through pre-steady-state kinetic assays (Figure 2) are 100-fold higher with 5 mM Mn²⁺ ((2.3–6.0)x10⁻³ μM⁻¹s⁻¹, Table 2) than with 5 mM Mg²⁺ ((2.2–3.8)x10⁻⁵ μM⁻¹s⁻¹, Table 3). The 100-fold efficiency difference is mainly contributed by the ~48-fold dNTP binding affinity ratio since the k_p was only ~2-fold lower with Mg²⁺ versus Mn²⁺ (Tables 2 and 3). After multiplying the k_p/K_d difference by the 34-fold DNA binding affinity ratio, PrimPol is 3,400-fold less efficient as a DNA polymerase to bind and elongate a DNA/DNA substrate in the presence of 5 mM Mg²⁺ over an equal concentration of Mn²⁺. Considering the very low DNA and dNTP binding affinities and correct dNTP incorporation rates and efficiencies relative to other kinetically characterized human DNA polymerases in the presence of Mg²⁺ (Table S1), PrimPol is incapable of competing for access to DNA and is not efficient enough to function as a meaningful DNA polymerase during DNA replication and lesion bypass in vivo. For example, PrimPol is 2×10⁶ fold less efficient than human DNA polymerase γ with Mg²⁺ and may not be able to play a role in mitochondrial DNA replication. However, in the presence of Mn²⁺, PrimPol bound both DNA and dNTPs with moderately high affinities (Table 2) and incorporated correct dNTPs with comparable efficiencies as human DNA polymerases μ, κ, and ι with Mg²⁺ (Table S1). In parallel, the primase activity of PrimPol has been barely observed in the presence of 10 mM Mg²⁺ but is activated by Mn²⁺ with a concentration as low as 50 μM ¹. Together, these results suggest that the AEP domain of PrimPol likely utilizes Mn²⁺, not Mg²⁺, as divalent metal ion cofactor in order to carry out both polymerase and primase activities in vivo. Recently, Polymerase Delta Interacting Protein 2 (PolDIP2) has been shown to interact with the AEP domain and moderately enhance the DNA binding affinity and polymerase processivity of human PrimPol in the presence of Mg^{2+ 37}. If the enhancement is also true in the presence of Mn²⁺, the combined favorable effect of PolDIP2 and Mn²⁺ will further solidify the role of PrimPol as a legitimate bifunctional enzyme in replication fork progression and restart, as well as DNA lesion bypass ^{1; 3; 5; 6}. More studies, especially structural investigation, are needed to verify the modulation effect of both PolDIP2 and Mn²⁺ on the structure and function of PrimPol.

Interestingly, some of the thermodynamic and kinetic parameters reported here are different than those reported by Mislak and Anderson (Antimicrob. Agents Chemother. 60, 561–569) in 2016 ³⁸. Specifically, their reported K_d,DNA values ³⁸, measured by electrophoretic mobility shift assays (EMSAs), are larger than ours by 11- to 25-fold in the presence of Mn²⁺ and 8-fold in the presence of Mg²⁺. The relative enhancement in the DNA binding affinities by switching from Mg²⁺ (K_d,DNA = 340–720 nM) to Mn²⁺ (K_d,DNA = 8 μM) in their studies ³⁸ is comparable to the 34-fold DNA binding affinity enhancement reported here. The large discrepancy between our and their K_d,DNA values is likely due to the different methods used the two studies. The EMSAs employed by Mislak and Anderson ³⁸ are not performed under true equilibrium conditions and can only be used to estimate the K_d,DNA values. In comparison, our fluorescence anisotropy assays are a superior method because the protein and DNA binding and dissociation were allowed to come to equilibrium in solution, and polarization was measured promptly after equilibrium was achieved. Furthermore, the sole variable to affect equilibrium in an fluorescence anisotropy assay is the concentration of DNA or PrimPol in their binding interactions while the gel matrix and electrical current can perturb the protein-DNA complex formation in an EMSA. Likewise, the slightly lower k_p and k_p/K_d values measured here may have been caused by differences in the reaction buffers used in our single-turnover assays (Materials and Methods) and those used by Mislak and Anderson (10 mM Bis-Tris propane, pH = 7.0, 1 mM DTT, 10 mM MnCl₂) ³⁸. Furthermore, the slow single-turnover rates (Tables 2 and 3) measured here are comparable to the steady-state rates reported by Zafar et al.⁶ Therefore, we expect not to observe an initial burst of product formation under pre-steady-state kinetic conditions in which the DNA substrate is only a few fold molar excess over the enzyme. Additionally, it should be noted that PrimPol is an atypical DNA polymerase and its kinetic mechanism is likely to be different from those of canonical polymerases. Thus, the assignments of the steady-state and pre-steady state rates within the mechanism of polymerization catalyzed by PrimPol remain to be established.

4.2. The polymerase activity of human PrimPol is error-prone in the presence of Mn²⁺.

The substitution fidelity of PrimPol was kinetically determined to be in the range of 3.4×10⁻² to 3.8×10⁻¹ in the presence of Mn²⁺ (Table 2). Thus, PrimPol poorly discriminated against incorrect dNTPs during primer elongation. The fidelity of PrimPol is actually comparable to the low fidelities of human Y-family DNA polymerases which bypass DNA lesions in vivo ³⁹. PrimPol incorporated an incorrect dNTP with only 2- to 23-fold slower k_p than a correct dNTP and bound to all dNTPs with comparable affinities (K_d = 3–30 μM), demonstrating a kinetic basis for the low fidelity of PrimPol. These data also indicate that incorrect dNTPs are strong competitive inhibitors against correct dNTPs during polymerization in the presence of Mn²⁺. Although significantly less efficient, PrimPol has a fidelity of 10⁻⁴ in the presence of Mg²⁺ (Table 3) which is 100–1,000 fold more faithful than in the presence of Mn²⁺ (Table 2). Consistently, the switch from Mg²⁺ to Mn²⁺ as the divalent metal ion cofactor is known to increase misincorporation frequency for many DNA polymerases and can lead to increased genomic instability ^{40; 41; 42; 43; 44}.

4.3. Human PrimPol displays moderate sugar selectivity in the presence of Mn²⁺ and Mg²⁺.

Human PrimPol has been shown to select dNTPs over rNTPs during primer extension with both DNA and RNA primers ¹¹. Consistently, our kinetic studies quantitatively estimated the sugar selectivity of PrimPol to be 57–1,800 in the presence of Mn²⁺ (Table 2) and 150–4,500 in the presence of Mg²⁺ (see 3.4 in Results). Thus, human PrimPol displays a modest sugar selectivity (740 on average with Mn²⁺) ⁴⁵, which is comparable to the sugar selectivities of human DNA polymerases α (500), γ (1,000) and Rev1 (280), measured using similar kinetic methods^{31; 45; 46}. Interestingly, a very recently published crystal structure of the AEP domain of human PrimPol in complex with DNA and an incoming dATP ⁴⁷ was modeled with an rATP to replace the dATP in the enzyme active site. Notably, the 2′ hydroxyl of the rATP ribose sterically clashed with the backbone carbonyl of Asn289. Thus, the 67-fold difference between the K_d values of correct dATP and matched rATP in the presence of Mn²⁺ (Table 1) likely arises from the clash between the rATP and the steric gate residue Asn289 ⁴⁵, similar to what has been observed with other DNA polymerases ^{30; 31; 32; 44; 45}, e.g. rNTP with Tyr12 in Sulfolobus solfataricus DNA polymerase IV (Dpo4) ^{48; 49}. Furthermore, since the AEP domain of PrimPol contains a single active site that catalyzes both primase and polymerase activities^{1; 2; 3}, the same mode of sugar selection is likely utilized by both the primase and polymerase activities of PrimPol. Our ongoing structural and mutagenic studies of PrimPol will further characterize the steric gate residue in this bifunctional enzyme.

4.4. Human PrimPol was able to incorporate the triphosphates of gemcitabine and cytarabine, but not emtricitabine and lamivudine.

Nucleoside Reverse Transcriptase Inhibitors (NRTIs) specifically target the active site of RTs and halt viral genomic replication. Mitochondrial toxicity of NRTIs has been associated with the incorporation of the triphosphates of NRTIs by human DNA polymerase γ, a main off-target issue. Based on the existence of PrimPol in the mitochondria, it is possible that PrimPol could potentially incorporate NRTIs and contribute to their mitochondrial toxicology. Recently, human PrimPol has been shown to incorporate the triphosphates of anti-HIV NRTI drugs zidovudine and abacavir ³⁸ and re-prime downstream from NRTI-terminated DNA primers ⁵⁰. In this paper, we found that human PrimPol incorporated the triphosphates of antiviral drugs emtricitabine ((−)3TC-TP) and lamivudine ((−)FTC-TP) with extremely low efficiencies (10⁻⁶ to 10⁻⁷ μM⁻¹ s⁻¹) in the presence of Mn²⁺ (Table 4) but did not incorporate them even after 3 hours in the presence of Mg²⁺ at 37 °C. Since (−)3TC-TP and (−)FTC-TP, unlike natural dNTPs, possess L-stereochemistry (Figure S5), our results suggest that PrimPol, like Dpo4⁵¹ and human DNA polymerase λ³⁶, possesses strong D-stereoselectivity. Our results further suggest that human PrimPol is an unlikely off-target of these two drugs in vivo.

In comparison, human PrimPol incorporated AraCTP, GemCTP, and dCTP, which possess D-stereochemistry (Figure S5), with comparable k_p/K_d values in the presence of Mn²⁺ or Mg²⁺, leading to the discrimination factors in the range of 2–14 (Table 4). It suggests that the inhibition of PrimPol by these anticancer chain-terminating nucleotide analogs likely contributes to their clinical toxicities. Interestingly, there is an order for the K_d values: dCTP < AraCTP < GemCTP < rCTP (Tables 2–4). This order is closely correlated to the size of the 2′-group oriented below the sugar ring, likely due to its clash with an unidentified steric gate residue in PrimPol (see above). As the size of the 2′-group increases, the K_d value increases simultaneously. Specifically, the 2′-group is a small hydrogen atom in both dCTP and AraCTP, a medium sized fluorine atom in GemCTP, and a comparatively larger sized hydroxyl group in rCTP (Figure S5), which leads to the above K_d order. The slightly larger K_d value of AraCTP over dCTP may be contributed by the size difference between their 2′-groups oriented above the sugar ring.

In summary, our fluorescence anisotropy and pre-steady-state kinetic studies demonstrate that human PrimPol is a moderately efficient and highly unfaithful DNA polymerase in the presence of Mn²⁺ but is too inefficient to function as a DNA polymerase in the presence of Mg²⁺. Human PrimPol possesses modest sugar selectivity and can incorporate the triphosphates of anticancer gemcitabine and cytarabine, but not antiviral emtricitabine and lamivudine.

1. Abbreviations:

(−)3TC

(−)-β-L-2′−3′-dideoxy-3′-thiacytidine

(−)3TC-TP

5′-triphosphate of (−)3TC or lamivudine

(−)FTC

(−)-β -L-2′−3′-dideoxy-5-fluoro-3′-thiacytidine

(−)FTC-TP

5′-triphosphate of (−)FTC or emtricitabine

AEP

Archaeo-Eukaryotic Primase

dNTP

deoxyribonucleotide

AraCTP

2′-aracytidine-5′-triphosphate

Dpo4

DNA polymerase IV

GemCTP

2′-deoxy-2′,2′-difluorodeoxycytidine-5′-triphosphate

mtDNA

mitochondrial DNA

NRTI

Nucleoside Reverse Transcriptase Inhibitor

PolDIP2

Polymerase Delta Interacting Protein 2

ICP-MS

inductively coupled plasma mass spectrometry

RPA

replication protein A

rNTP

ribonucleotide