Pre-Steady State Kinetic Investigation of the Incorporation of Anti-Hepatitis B Nucleotide Analogs Catalyzed by Non-Canonical Human DNA Polymerases

Abstract

Antiviral nucleoside analogs have been developed to inhibit the enzymatic activities of the hepatitis B virus (HBV) polymerase, thereby preventing the replication and production of HBV. However, the usage of these analogs can be limited by drug toxicity because the 5′-triphosphates of these nucleoside analogs (nucleotide analogs) are potential substrates for human DNA polymerases to incorporate into host DNA. Although they are poor substrates for human replicative DNA polymerases, it remains to be established whether these nucleotide analogs are substrates for the recently discovered human X- and Y-family DNA polymerases. Using pre-steady state kinetic techniques, we have measured the substrate specificity values for human DNA polymerases β, λ, η, ι, κ, and Rev1 incorporating the active forms of the following anti-HBV nucleoside analogs approved for clinical use: adefovir, tenofovir, lamivudine, telbivudine, and entecavir. Compared to the incorporation of a natural nucleotide, most of the nucleotide analogs were incorporated less efficiently (2 to >122,000) by the six human DNA polymerases. In addition, the potential for entecavir and telbivudine, two drugs which possess a 3′-hydroxyl, to become embedded into human DNA was examined by primer extension and DNA ligation assays. These results suggested that telbivudine functions as a chain terminator while entecavir was efficiently extended by the six enzymes and was a substrate for human DNA ligase I. Our findings suggested that incorporation of anti-HBV nucleotide analogs catalyzed by human X- and Y-family polymerases may contribute to clinical toxicity.

INTRODUCTION

With more than two billion people infected worldwide, hepatitis B virus (HBV) remains an important global health concern. Chronic HBV infection, which affects more than 350 million people, is a major cause of hepatocellular carcinoma and liver cirrhosis, two life-threatening disease states of the liver. Thus, HBV treatment is important to prevent or to slow the progression of these severe liver complications. Currently, seven antiviral agents are approved by the United States Food and Drug Administration (FDA) for treatment of HBV: two immune modulators (interferon-alpha and pegylated interferon-alpha) and five nucleoside/nucleotide analogs [adefovir (PMEA), tenofovir (PMPA), lamivudine (L-3TC), telbivudine (L-TBV), and entecavir (ETV) (Figure 1)]. Following cellular uptake, these analogs undergo either two (PMEA and PMPA) or three (L-3TC, L-TBV, and ETV) phosphorylation events to be activated to their di- (PMEA-DP and PMPA-DP) or triphosphate (L-3TC-TP, L-TBV-TP, and ETV-TP) forms, respectively. These activated nucleotide analogs target the HBV DNA polymerase (Pol) which has enzymatic activity for a unique protein-priming event, RNA-dependent and DNA-dependent DNA synthesis, and degradation of RNA in a RNA/DNA duplex (i.e. RNase H). Depending on the analog, these drugs may function as competitive inhibitors against natural dNTP substrates and/or as obligate or masked chain terminators that inhibit the priming and/or polymerization activities of the HBV Pol. Unfortunately, the usage of anti-HBV nucleoside analogs can be limited by drug resistance and adverse side effects.^{1, 2} It has been postulated that cellular DNA polymerases, such as human DNA polymerase γ (Pol γ), may be potential drug targets and the cause of observed clinical toxicity, since nucleoside analogs approved for human immunodeficiency virus type 1 (HIV-1) are associated with mitochondrial toxicity.^{3, 4} However, mitochondrial toxicity induced by nucleotide analog incorporation catalyzed by Pol γ does not account for all of the unwanted side effects.⁵ The human genome encodes at least 15 other DNA polymerases, which are members of the A-, B-, X- or Y-family, that may be potential candidates for generating cellular toxicity via analog incorporation into nuclear DNA.^5–8

Figure 1.

Chemical structures of anti-HBV nucleoside/nucleotide analogs and their natural counterpart.

Using pre-steady state kinetic techniques, we determined the incorporation efficiency of five anti-HBV nucleotide analogs (Figure 1) with six non-canonical human DNA polymerases: Pols β, λ, η, ι, κ, and Rev1. Both Pols β and λ are X-family DNA polymerases. Pol β functions in base excision repair while Pol λ is putatively involved in base excision repair, nonhomologous end-joining, and antibody generation.⁹ The human genome encodes four Y-family DNA polymerases (Pols η, ι, κ, and Rev1) that catalyze translesion DNA synthesis and may play a part in somatic hypermutation.¹⁰ Therefore, inhibition of these selected X- and Y-family pols could lead to unwanted toxicity including apoptosis, genetic instability, and immunodeficiency. Our kinetic data showed that most of the analogs were substrates for the non-canonical pols and that the kinetic basis of incorporation varied for each analog. These results suggested that human X- and Y-family enzymes are capable of inserting nucleotide analogs in vivo and established structure-function relationships that are important for future anti-HBV drug design.

EXPERIMENTAL PROCEDURES

Materials

These chemicals were purchased from the following companies: [γ- ³²P]ATP, MP Biomedicals; ATP, GE Healthcare; Bio-Spin 6 columns, Bio-Rad Laboratories; deoxyribonucleotide 5′-triphosphates, GE Healthcare; β-L-2′-deoxythymidine 5′-triphosphate (L-TBV-TP), TriLink Biotechnologies; OptiKinase™, USB Corporation; synthetic oligodeoxyribonucleotides 21-mer, 5′-phosphorylated 19-mer, and 41-mers, Integrated DNA Technologies. The diphosphate of 9-[2-(phosphonomethoxy)ethyl]adenine (PMEA-DP) and 9-[2-(phosphonomethoxy)propyl]adenine (PMPA-DP), β-L-2′,3′-dideoxy-3′-thiacytidine 5′-triphosphate (L-3TC-TP), and entecavir 5′-triphosphate (ETV-TP) were kind gifts from Gilead Sciences, Inc.

Preparation of human DNA polymerases and DNA substrates

The plasmids, expression, and purification of human DNA polymerases β,^11–13 λ,^14–16 η,¹⁷ truncated ι (1–420),^{17, 18} truncated κ (9–518),¹⁷ and truncated Rev1 (341–829)^19–21 were described previously and the purity of each enzyme is greater than 95% based on a Coomassie-stained polyacrylamide gel. Purified human Δ235 DNA ligase I was a kind gift of Dr. Tom Ellenberger.²² Commercially synthesized oligonucleotides in Table 1 were purified using polyacrylamide gel electrophoresis.^23–25 The 21-mer primer was 5′-radiolabeled with [γ-³²P]ATP and OptiKinase™ according to the manufacturer’s protocol, and the unreacted [γ-³²P]ATP was subsequently removed via a Bio-Spin 6 column. The 21-41mer primer-template DNA substrates²³ and 21-19-41mer single-nucleotide gapped DNA substrates^{4, 26, 27} were annealed as described previously.

Table 1.

Sequences of oligonucleotides^a

21mer
19merC
19merA
41merGT
41merTG
41merCG
41merAG

^aThe 21mer strand was 5′-radiolabeled. For single-nucleotide gap DNA substrates, the downstream 19mer strand was 5′-phosphorylated. The identity of important base changes is highlighted in bold.

Single-turnover kinetic assays to measure the k_p and K_d

Kinetic assays were completed using buffer B (50 mM Tris-HCl pH 7.8 at 37 °C, 5 mM MgCl₂, 50 mM NaCl, 0.1 mM EDTA, 5 mM DTT, 10% glycerol, and 0.1 mg/ml of bovine serum albumin (BSA)) for Pol β, buffer L (50 mM Tris-HCl pH 8.4 at 37 °C, 5 mM MgCl₂, 100 mM NaCl, 0.1 mM EDTA, 5 mM DTT, 10% glycerol, and 0.1 mg/ml of BSA) for Pol λ,¹⁴ and buffer Y (50 mM HEPES pH 7.5 at 37 °C, 5 mM MgCl₂, 50 mM NaCl, 0.1 mM EDTA, 5 mM DTT, 10% glycerol, and 0.1 mg/ml of BSA) for Pols η, ι, κ, and Rev1. The primer-template DNA substrates were used in assays with Pols η, ι, κ, and Rev1 while the gap-filling Pols β and λ were provided 21-19-41mer gap DNA. All kinetic experiments described herein were performed at 37 °C, and the reported concentrations were final after mixing all the components. A pre-incubated solution of the Pol (120 or 300 nM) and 5′-[³²P]-radiolabeled DNA substrate (30 nM) was mixed with increasing concentrations (0.01–900 µM) of a single nucleotide or nucleotide analog in the appropriate buffer at 37 °C. The enzyme was in molar excess over DNA, whereby the enzyme to DNA ratio was 4:1 for Pols λ, η, ι, and Rev1 and 10:1 for Pols β and κ. Aliquots of the reaction mixtures were quenched at various times using 0.37 M EDTA. A rapid chemical-quench flow apparatus (KinTek) was utilized for fast nucleotide incorporations. Reaction products were resolved using sequencing gel electrophoresis (17% acrylamide, 8 M urea) and quantitated with a Typhoon TRIO (GE Healthcare). The time course of product formation at each nucleotide concentration was fit to a single-exponential equation (Eq. 1) using a nonlinear regression program, KaleidaGraph (Synergy Software), to yield an observed rate constant of nucleotide incorporation (k_obs). The k_obs values were then plotted as a function of nucleotide concentration and fit using the hyperbolic equation (Eq. 2) which resolved the maximum rate of incorporation (k_p) and the equilibrium dissociation constant (K_d) of an incoming nucleotide for each Pol•DNA complex.

$$[\text{Product}]=\mathrm{A}[1-\text{exp}(-k_{\mathit{\text{obs}}}\mathrm{t})]$$ Equation 1

$$k_{\mathit{\text{obs}}}=k_p[\text{dNTP}]/\{[\text{dNTP}]+K_d\}$$ Equation 2

Extension assay for non-chain terminators

A pre-incubated solution of Pol (240 or 600 nM) and 5′-[³²P]-radiolabeled 21/41mer DNA (60 nM) in the appropriate buffer was mixed with either ETV-TP•Mg²⁺ (1–50 µM) or L-TBV-TP•Mg²⁺ (100–500 µM) to allow sufficient extension (at least 8 half-lives) of the primer before adding the four natural dNTPs (200 µM) for various reaction times. The enzyme to DNA ratio was 4:1 for Pols λ, η, ι, and Rev1 and 10:1 for Pols β and κ. Aliquots of the reaction mixtures were quenched using 0.37 M EDTA, and the reaction products were resolved using sequencing gel electrophoresis (20% acrylamide, 8 M urea).

DNA ligation assay for non-chain terminators

A pre-incubated solution of Pol β (120 nM) and 5′-[³²P]-radiolabeled 21-19/41mer DNA (30 nM) in buffer B was mixed in independent reactions with dGTP•Mg²⁺ (1 µM), dTTP•Mg²⁺ (1 µM), ETV-TP• Mg²⁺ (1 µM) or L-TBV-TP•Mg²⁺ (500 µM) to allow sufficient extension (at least 8 half-lives; 30 s for dGTP and dTTP, 5 min for ETV-TP, and 120 min for L-TBV-TP) of the primer before adding human Δ235 DNA ligase I (30 nM) and 1 mM ATP to the polymerization mixture. Aliquots of the reaction mixtures were quenched at various times by adding them to 0.37 M EDTA and immediate heat denaturation for 2 min at 95 °C. The DNA products were resolved using sequencing gel electrophoresis (20% acrylamide, 8 M urea).

RESULTS

Determination of selection factors

The mechanism of anti-HBV nucleotide analog incorporation catalyzed by Pols β, λ, η, ι, κ, and Rev1 was determined by utilizing single-turnover kinetic methodology. When the enzyme is in molar excess over DNA, the conversion of the DNA substrate into product is observed directly in a single pass through the enzymatic pathway so that the nucleotide concentration dependence on the observed rate constant (k_obs) can resolve the equilibrium dissociation constant (K_d) and maximum rate of incorporation (k_p) for an incoming nucleotide.^28–30 As a representative example, a pre-incubated solution of Pol λ and 5′-[³²P]-radiolabeled 21-19C-41merTG (Table 1) was mixed with increasing concentrations of PMEA-DP (see Experimental Procedures, Figure 2A). After plotting and fitting the data to equations 1 and 2 (Figure 2B and 2C), the single-turnover kinetic parameters were resolved: a k_p of 0.175 ± 0.004 s⁻¹ and a K_d of 4.5 ± 0.3 µM. Using similar single-turnover kinetic assays, the kinetic parameters were measured for each enzyme incorporating a natural nucleotide or one of the analogs, which were subsequently used to calculate the substrate specificity constants (k_p/K_d) and selection factors ((k_p/K_d)_dNTP/(k_p/K_d)_analog) in Tables 2–5.

Figure 2.

Concentration dependence on the pre-steady state rate constant of PMEA-DP incorporation catalyzed by Pol λ. A pre-incubated solution of Pol λ (120 nM) and 21-19C-41merTG DNA (30 nM) was rapidly mixed with increasing concentrations of PMEA-DP•Mg²⁺ (0.2 µM, ●; 0.5 µM, ○; 1 µM, ■; 2 µM, □; 5 µM, ▲; 10 µM, △; and 25 µM, ◆) for various time intervals. (A) A representative gel image for PMEA-DP incorporation at 25 µM is shown. The length of the DNA primer is indicated in the right margin. (B) The concentration of the DNA product is plotted as a function of time. The solid lines are the best fits to a single-exponential equation which determined the observed rate constant, k_obs. (C) The k_obs values were plotted as a function of PMEA-DP concentration. The data (●) were then fit to a hyperbolic equation, yielding a k_p of 0.175 ± 0.004 s⁻¹ and a K_d of 4.5 ± 0.3 µM.

Table 2.

Kinetic parameters for nucleotide incorporation opposite template dT at 37 °C.

DNA Polymerase	dNTP	kp (s−1)	Kd (µM)	kp/Kd (µM−1s−1)	Selection Factora
Pol β	dATPb	32 ± 1	9.2 ± 1.0	3.5
	PMPA-DPb	4.7 ± 0.5	50 ± 10	9.4 × 10−2	40
	PMEA-DP	1.31 ± 0.04	65 ± 6	2.0 × 10−2	170
Pol λ	dATPc	1.5 ± 0.1	0.9 ± 0.3	1.7
	PMPA-DPb	0.095 ± 0.008	3.7 ± 0.9	2.6 × 10−2	65
	PMEA-DP	0.175 ± 0.004	4.5 ± 0.3	3.9 × 10−2	44
Pol η	dATPb	35 ± 3	130 ± 30	2.7 × 10−1
	PMPA-DPb	0.0134 ± 0.0007	90 ± 10	1.5 × 10−4	1,800
	PMEA-DP	0.069 ± 0.008	55 ± 18	1.3 × 10−3	210
Pol κ	dATPb	2.49 ± 0.08	7.0 ± 1.0	3.6 × 10−1
	PMPA-DPb	0.010 ± 0.005	3000 ± 2000	3.3 × 10−6	110,000
	PMEA-DP	0.039 ± 0.001	23 ± 3	1.7 × 10−3	210
Pol ι	dATPb	0.015 ± 0.001	260 ± 40	5.8 × 10−5
	PMPA-DPb	No observed incorporation			high
	PMEA-DP	0.00047 ± 0.00004	180 ± 40	2.6 × 10−6	48
Rev1	dATPb	0.00152 ± 0.00005	2.0 ± 0.3	7.6 × 10−4
	PMPA-DPb	No observed incorporation			high
	PMEA-DP	0.00037 ± 0.00002	120 ± 20	3.1 × 10−6	250
Pol γ	dATPd	45 ± 1	0.8 ± 0.1	56
	PMPA-DPe	0.21 ± 0.01	40.3 ± 5.7	5.2 × 10−3	10,800
T7 exo	dATPf	156 ± 8	8 ± 2	19.5
	PMPA-DPf	0.096 ± 0.009	268 ± 39	3.6 × 10−4	54,400

^aCalculated as (k_p/K_d)_dATP/(k_p/K_d)_Analog.

^bKinetic parameters are from reference.³¹

^cKinetic parameters are from reference.¹⁴

^dKinetic parameters are from reference.⁴⁹

^eKinetic parameters are from reference.³⁸

^fKinetic parameters are from reference.⁵⁰

Table 5.

Kinetic parameters for nucleotide incorporation opposite template dC at 37 °C.

DNA Polymerase	dNTP	kp (s−1)	Kd (µM)	kp/Kd (µM−1s−1)	Selection Factora
Pol β	dGTP	18.8 ± 0.4	8.7 ± 0.4	2.2
	ETV-TP	0.054 ± 0.002	0.26 ± 0.04	2.1 × 10−1	10
Pol λ	dGTPb	2.5 ± 0.1	2.1 ± 0.3	1.2
	ETV-TP	0.034 ± 0.002	0.09 ± 0.02	3.8 × 10−1	3
Pol η	dGTP	38 ± 2	80 ± 10	4.8 × 10−1
	ETV-TP	0.24 ± 0.01	2.4 ± 0.4	1.0 × 10−1	5
Pol κ	dGTP	4.4 ± 0.1	10.6 ± 0.9	4.2 × 10−1
	ETV-TP	0.370 ± 0.004	4.2 ± 0.1	8.8 × 10−2	5
Pol ι	dGTP	0.095 ± 0.002	141 ± 8	6.7 × 10−4
	ETV-TP	0.0097 ± 0.0004	33 ± 5	2.9 × 10−4	2
Rev1	dGTP	0.0038 ± 0.0002	3.9 ± 0.8	9.7 × 10−4
	ETV-TP	0.0017 ± 0.0001	56 ± 10	3.0 × 10−5	32
HIV-1 RT	dGTPc	18.3 ± 1.30	1.76 ± 0.51	10.4
	ETV-TPc	0.107 ± 0.007	2.23 ± 0.67	4.8 × 10−2	220

^aCalculated as (k_p/K_d)_dGTP/(k_p/K_d)_ETV-TP.

^bKinetic parameters are from reference.¹⁴

^cKinetic parameters are from reference.⁵⁸

Incorporation of acyclic adenine analogs

Tenofovir (PMPA) and adefovir (PMEA) both have an acyclic moiety and a phosphonate group, but PMPA has an additional methyl group (Figure 1). In general, the selection factors were lower for PMEA-DP (44 to 250) than PMPA-DP (40 to >110,000) (Table 2 and Figure 3). The X-family DNA polymerases β and λ exhibited the least amount of discrimination for both analogs, whereby the k_p value dropped by an average of 14-fold and the K_d value increased by an average of 5-fold. In contrast, the additional methyl group in PMPA-DP led to a greater degree of discrimination for the Y-family DNA polymerases, although, different mechanisms of discrimination were observed for the four enzymes. For Pol η, the rate of PMPA-DP and PMEA-DP incorporation dropped by an average of 420-fold while the binding was slightly tighter than dATP. For Pols κ, ι, and Rev1, the rate of PMEA-DP incorporation decreased by 60-, 30-, and 4-fold, and the ground-state binding affinity (1/K_d) was 3-fold weaker, unchanged, and 60-fold weaker, respectively. As noted previously, PMPA-DP is a poor substrate for Pols κ, ι, and Rev1.³¹ The kinetic parameters could only be measured for Pol κ because PMPA-DP incorporation was too inefficient to be observed with Pol ι and Rev1.

Figure 3.

Comparison of selection factors. The selection factors from Tables 2–5 are plotted for each X- and Y-family DNA polymerase examined herein and the corresponding nucleotide analog. The selection factors defined as “high” are graphed as being greater than 1,000,000. The bars are color-coded for each nucleotide analog as follows: PMPA-DP is gray, PMEA-DP is red, L-3TC-TP is green, L-TBV-TP is blue, and ETV-TP is black.

Incorporation of nucleotides with L-stereochemistry

Previously,³¹ the selection factors were determined for Pols β, λ, η, ι, κ, and Rev1 inserting L-3TC-TP, and the following general trend emerged: the k_p drops significantly (6,100-fold on average) while the K_d decreases (6-fold on average) compared to dCTP (Table 3). Telbivudine (L-TBV) is the L-isomer of thymidine (Figure 1). No observable L-TBV-TP incorporation was detected for Rev1, likely because it is a deoxycytidyl transferase and prefers dCTP regardless of template base.²¹ For the remaining DNA polymerases, relatively large selection factors of greater than 8,000 were determined (Table 4 and Figure 3). Unlike L-3TC-TP, the binding of L-TBV-TP to the Pol•DNA complex was weakened by 10-fold on average but the rate of incorporation remained slow compared to dTTP. These kinetic results suggested that (i) the oxathiolane ring of L-3TC-TP is important for the tight ground-state binding affinity of L-3TC-TP and (ii) that the polymerase active sites of these Y-family polymerases could accommodate L-dNTPs, albeit inefficiently.

Table 3.

Kinetic parameters for nucleotide incorporation opposite template dG at 37 °C.

Human DNA Polymerase	dNTP	kp (s−1)	Kd (µM)	kp/Kd (µM−1s−1)	Selection Factora
Pol β	dCTPb	5.02 ± 0.07	0.71 ± 0.04	7.1
	L-3TC-TPb	0.00390 ± 0.00010	0.18 ± 0.02	2.2 × 10−2	325
Pol λ	dCTPc	1.57 ± 0.04	0.9 ± 0.1	1.7
	L-3TC-TPb	0.00402 ± 0.00008	0.106 ± 0.010	3.8 × 10−2	46
Pol η	dCTPb	49 ± 2	25 ± 4	2.0
	L-3TC-TPb	0.0296 ± 0.0009	3.3 ± 0.3	9.0 × 10−3	220
Pol κ	dCTPb	11.8 ± 0.6	32 ± 5	3.7 × 10−1
	L-3TC-TPb	0.00045 ± 0.00001	4.0 ± 0.5	1.1 × 10−4	3,300
Pol ι	dCTPb	0.075 ± 0.002	50 ± 5	1.5 × 10−3
	L-3TC-TPb	0.00356 ± 0.00009	96 ± 8	3.7 × 10−5	40
Rev1	dCTPd	22.4 ± 0.9	2.2 ± 0.3	10
	L-3TC-TPb	0.0236 ± 0.0006	1.04 ± 0.10	2.3 × 10−2	450
Pol γ	dCTPe	44 ± 2	1.1 ± 0.1	40
	L-3TC-TPe	0.125 ± 0.005	9.2 ± 0.9	1.4 × 10−2	2,900

^aCalculated as (k_p/K_d)_dCTP/(k_p/K_d)_L-3TC-TP.

^bKinetic parameters are from reference.³¹

^cKinetic parameters are from reference.¹⁴

^dKinetic parameters are from reference.²¹

^eKinetic parameters are from reference.⁶¹

Table 4.

Kinetic parameters for nucleotide incorporation opposite template dA at 37 °C.

Human DNA Polymerase	dNTP	kp (s−1)	Kd (µM)	kp/Kd (µM−1s−1)	Selection Factora
Pol β	dTTPc	12.3 ± 0.5	3.5 ± 0.5	3.5
	L-TBV-TP	0.00055 ± 0.00001	11 ± 1	5.0 × 10−5	70,000
Pol λ	dTTPb	3.9 ± 0.2	2.6 ± 0.4	1.5
	L-TBV-TP	0.00072 ± 0.00002	53 ± 6	1.4 × 10−5	11,000
Pol η	dTTPc	35 ± 1	41 ± 5	8.5 × 10−1
	L-TBV-TP	0.0028 ± 0.0003	400 ± 100	7.0 × 10−6	122,000
Pol κ	dTTPc	4.35 ± 0.04	11.0 ± 0.5	4.0 × 10−1
	L-TBV-TP	0.00065 ± 0.00006	90 ± 20	7.2 × 10−6	55,000
Pol ι	dTTPc	0.75 ± 0.02	13 ± 1	5.8 × 10−2
	L-TBV-TP	0.00071 ± 0.00006	100 ± 30	7.1 × 10−6	8,100
Rev1	dTTPc	0.0038 ± 0.0003	6 ± 1	6.3 × 10−4
	L-TBV-TP	No observed incorporation			High

^aCalculated as (k_p/K_d)_dTTP/(k_p/K_d)_L-TBV-TP.

^bKinetic parameters are from reference.¹⁴

^cKinetic parameters are from reference.³¹

Incorporation of entecavir 5′-triphosphate

Entecavir (ETV) is a deoxyguanosine analog with a cyclopentyl sugar ring (Figure 1). Interestingly, Pols β, λ, η, ι, κ, and Rev1 incorporated ETV-TP opposite dC with much lower selection factors (Table 5 and Figure 3) compared to the other anti-HBV analogs (Tables 2–4). Except for Rev1, the lack of discrimination between ETV-TP and dGTP was due to the binding affinity of ETV-TP being approximately 20-fold tighter on average. However, the rate of ETV-TP incorporation was on average 120-fold slower than dGTP.

Qualitative analysis of non-chain terminators being embedded into DNA

ETV and L-TBV are not obligate chain terminators like PMEA, PMPA, and L-3TC (Figure 1), since these two analogs possess the required 3′-hydroxyl group for downstream enzymatic steps such as primer extension or DNA ligation if incorporated into a gapped DNA substrate. Therefore, it is possible that these analogs could become embedded into the human genome. To examine the extension possibility, we pre-incubated a solution of the enzyme and the appropriate 5′-[³²P]-radiolabeled 21-41mer DNA substrate before initiating the reaction with either ETV-TP or L-TBV-TP (see Experimental Procedures). After allowing sufficient time for the analog to be incorporated, the four natural dNTPs were added to the reaction mixture. Representative gel images are shown in Supplemental Figures 1 and 2 for an X-family (Pol β) and a Y-family (Pol η) DNA polymerase, respectively. Both Pol β and Pol η efficiently catalyzed full-length product (41mer) for the control and ETV-TP as early as 10 s after the addition of the natural dNTPs. In contrast, minimal extension of an L-TBV-MP terminated primer was observed after 1 h, and the small amount of 41mer may be partially due to the extension of the unreacted 21mer substrate rather than the 22mer terminated with the analog. Similar extension activities of both analogs were detected for Pols λ, ι, and κ (data not shown). For Rev1, the main product after ETV-TP incorporation was the 23mer. Subsequent incorporations likely did not occur because Rev1 can only efficiently catalyze the transfer of dCTP when dG is the template.²¹ Nonetheless, ETV-TP may be a masked chain terminator for Rev1. Together, these preliminary data suggested L-TBV-TP was a chain terminator while ETV-TP was not for Pols β, λ, η, ι, κ, and Rev1.

Next, we determined whether single-nucleotide gap DNA substrates having primers terminated with ETV-MP or L-TBV-MP are substrates for human DNA ligase I. A pre-incubated solution of Pol β and the appropriate 5′-[³²P]-radiolabeled 21-19/41mer DNA were treated with either ETV-TP or L-TBV-TP before initiating the ligation reaction with human Δ235 DNA ligase I and ATP (see Experimental Procedures). As shown in Supplemental Figure 3, the ligated DNA product (41mer) was detected for ETV-TP and the control reactions but not L-TBV-TP. However, the ligation of ETV-MP was less efficient than the dGMP control, for the first visible appearance of 41mer was at 10 s and 300 s for dGMP and ETV-MP, respectively. Based on the extension and ligation assays, these results suggested that ETV-TP can be incorporated and embedded into the genome via primer extension or subsequent ligation while L-TBV-TP would act as a chain terminator that may lead to single-strand DNA breaks. To eliminate ambiguity in interpreting the likelihood of extension and ligation, we are currently preparing primers with L-TBV-MP or ETV-MP at the 3′-terminus and will directly examine these reactions.

DISCUSSION

Overall trends of anti-HBV drugs by non-canonical human DNA polymerases

Ranking the selection factors of the five anti-HBV analogs for each enzyme (Tables 2–5 and Figure 3) generated the following profiles: the order is identical for Pols λ, η, and Rev1 (ETV-TP < PMEA-DP < L-3TC-TP < PMPA-DP < L-TBV-TP) and Pols κ and ι (ETV-TP < PMEA-DP < L-3TC-TP < L-TBV-TP < PMPA-DP) while Pol β is unique (ETV-TP < PMPA-DP < PMEA-DP < L-3TC-TP < L-TBV-TP). Consistently, the lowest level of discrimination occurred with ETV-TP whereas PMPA-DP and/or L-TBV-TP had the highest selection factors for the six human pols examined in this work. The different chemical modifications resulted in unique mechanisms of incorporation. Some modifications (e.g. L-stereochemistry and acyclic ribose) were unfavorable for incorporation while others (e.g. oxathiolane and methylene moieties in the sugar ring) enhanced the incorporation efficiency by forming a more stable ternary complex through tighter binding. The structural basis for the wide range of incorporation efficiencies exhibited by the human DNA polymerases with these anti-HBV drugs remains uncertain because structures have not been determined for a human DNA polymerase in complex with an anti-HBV nucleotide analog.

Potential relationship between clinical toxicity and drug incorporation by host DNA pols

Nucleoside analog usage can be limited by drug resistance and unwanted side effects.¹ Unfortunately, a striking 15–22% of HBV patients report moderate to severe side effects (e.g. myopathy and nephrotoxicity).^{32, 33} These adverse events may not be derived from mitochondrial toxicity^34–36 because assays indicate that anti-HBV nucleoside analogs are weak inhibitors of Pol γ^37–40 and/or induce relatively low mitochondrial toxicity.^40–44 Since human replicative polymerases (e.g. Pols α, δ, and ε) employ stringent mechanisms of nucleotide selection, these enzymes are unlikely to incorporate PMEA-DP,⁴⁵ PMPA-DP,⁴⁶ ETV-TP,⁴⁷ L-TBV-TP,⁴⁴ or L-3TC-TP,⁴⁸ indicating these drugs are relatively weak inhibitors. The selection factors measured for exonuclease-deficient Pol γ^{38, 49} and exonuclease-deficient T7 DNA polymerase,⁵⁰ two replicative A-family enzymes, incorporating PMPA-DP are 10,800 and 54,400, respectively, which are significantly larger than the values determined for Pol β (40), Pol λ (65), and Pol η (1,800) (Table 2). Similarly, Pol γ discriminates between dCTP and L-3TC-TP by 2,900-fold which is larger than most of the selection factors calculated for the non-canonical human pols (46 to 3,300) in Table 3. Overall, the X- and Y-family DNA polymerases exhibit a lower degree of discrimination than a replicative enzyme like Pol γ.

To better evaluate the potential of nucleotide analog incorporation in vivo, it is important to consider the intracellular concentrations of natural nucleotides relative to the nucleotide analogs. Using a liver cell line such as HepG2, the dNTP:nucleotide analog ratios have been determined to be or are predicted to be as follows: 6:1 for dCTP:L-3TC-TP,⁵¹ 1:0.4 for dATP:PMPA-DP,⁵² 1:1 for dATP:PMEA-DP,⁵² 25:1 for dGTP:ETV-TP,⁵³ and 1,600:1 for dTTP:L-TBV-TP.⁵⁴ Please note, the concentrations for dATP, dGTP, and dTTP were obtained from Table 1 in reference,⁵⁵ therefore, the ratios predicted for PMPA-DP, PMEA-DP, ETV-TP, and L-TBV-TP may be different in vivo. Nonetheless, these ratios suggest intracellular concentrations of nucleotide analogs relative to natural dNTPs can approach 1:1 so that the selection factors (Tables 2–5) reflect insertion frequencies. For example, the selection factor for Pol η incorporating PMEA-DP is 210, thereby estimating that Pol η can incorporate 1 PMEA-DP molecule per 210 incorporations of dATP. Another important consideration is that Pols β, λ, η, ι, κ, and Rev1 are expressed at the mRNA level in most human tissues, including those afflicted with side effects.^{9, 10} Thus, based on the relatively low selection factors measured herein (Tables 2–5 and Figure 3), human X- and Y-family DNA polymerases are likely to incorporate some anti-HBV nucleotide analogs in vivo which would inhibit base excision repair, nonhomologous end joining, translesion DNA synthesis, V(D)J recombination, and somatic hypermutation pathways. These events would induce a cascade of cellular events associated with nucleoside analog toxicity: genomic instability → apoptosis/cell death → side effects.^{5, 56} However, the efficacy and toxicity of a drug are a complex function of drug uptake, transport, and metabolism.

Kinetic results support entecavir as a potential carcinogen

According to the US prescribing information sheet for entecavir, solid tumors were detected in rodents that were exposed to high doses of entecavir.⁵⁷ Entecavir possesses a 3′-hydroxyl group, therefore, the carcinogenic activity may arise when this drug is embedded into genomic DNA. It has been reported that entecavir functions as a masked chain terminator for HBV pol and cellular Pols α, β, γ, δ, and ε.⁴⁷ However, our efficient incorporation (Table 5), extension (Supplementary Figures 1B and 2B), and ligation (Supplementary Figure 3B) results suggested that there is a high potential for entecavir to become embedded into DNA. Importantly, entecavir is not a masked chain terminator for Pols β, λ, η, ι, and κ, therefore, these polymerases can likely rescue a stalled replication fork due to entecavir incorporation. Together, these processes are likely to contribute to a putative mechanism of carcinogenicity, especially if the embedded drug induces higher error rates during subsequent rounds of replication. Langley et al.⁴⁷ did not provide information on their assays with Pol β, therefore, we are unsure of the discrepancy in our incorporation and extension results. Also, ETV-TP has been shown to be a substrate for HIV-1 reverse transcriptase (RT), although, the enzyme discriminates between ETV-TP and dGTP by 220-fold,⁵⁸ a selection factor that is greater than those measured for the human pols (2 to 32 in Table 5). HIV-1 RT is able to bypass a site-specific entecavir in a DNA template, although, pausing is detected near the entecavir site.⁵⁹ These kinetic results suggested that entecavir would not be a good drug candidate for treating HIV-1 infected patients.

Telbivudine is unlikely to become embedded in DNA

Unlike entecavir, telbivudine exhibited an extremely low potential to become embedded in the human genome, since the selection factors were large (8,100 to >122,000) and an L-TBV-MP-terminated primer was poorly extended (Supplementary Figures 1C and 2C) and was not a substrate for human DNA ligase I (Supplementary Figure 3C). The unnatural L-stereochemistry of telbivudine is likely problematic for extension and ligation because of the unfavorable alignment of the 3′-hydroxyl group for catalysis. Thus, L-stereochemistry is less problematic for the incorporation step than subsequent enzymatic steps. The incorporation and extension of L-dCTP or L-dTTP (i.e. L-TBV-TP) at low concentrations was not observed for Pols α, β, and ε,⁶⁰ which is likely due to the weak binding affinity and slow rate of incorporation for unmodified L-nucleotides as observed for the non-canonical pols (Table 4). Interestingly, the oxathiolane ring of L-3TC and L-FTC, an anti-HIV nucleoside analog which has a fluoro group at the C5 position of the L-3TC chemical structure,³¹ improves the incorporation efficiency for L-dCTP analogs (Table 3).

Concluding remarks

Unfortunately, similar transient state kinetic approaches have not been used to determine the selection factors for the HBV polymerase incorporating nucleotide analogs, so it is difficult to discern how selective the anti-HBV drugs are for the viral enzyme versus the human DNA polymerases. Nonetheless, these findings highlight the importance of initiating an in vivo investigation to confirm whether the non-canonical human DNA polymerases contribute to drug toxicity. In addition, this work established structure-function relationships that will be important in designing nucleoside analogs to overcome the limitations of clinical toxicity and drug resistance associated with current FDA-approved drugs.

Glossary

BSA

bovine serum albumin

dNTP

2′-deoxyribonucleotide 5′-triphosphate

DP

diphosphate

ETV

entecavir

FDA

Food and Drug Administration

HBV

hepatitis B virus

HIV-1

human immunodeficiency virus type 1

L-3TC

β-L-2′,3′-dideoxy-3′-thiacytidine or lamivudine

L-FTC

β-L-2′,3′-dideoxy-5-fluoro-3′-thiacytidine or emtricitabine

L-TBV

telbivudine or β-L-2′-deoxythymidine 5′-triphosphate

MP

monophosphate

PMEA

adefovir or 9-[2-(phosphonomethoxy)ethyl]adenine

PMPA

tenofovir or 9-[2-(phosphonomethoxy)propyl]adenine

Pol

DNA polymerase

Pol β

DNA polymerase beta

Pol γ

DNA polymerase gamma

Pol η

DNA polymerase eta

Pol ι

DNA polymerase iota

Pol κ

DNA polymerase kappa

Pol λ

DNA polymerase lambda

RT

reverse transcriptase

TP

triphosphate