Original Article Effect of Nucleoside and Nucleotide Reverse Transcriptase Inhibitors of HIV on Endogenous Nucleotide Pools

Antiviral Therapy 13:789–797

Original article Effect of nucleoside and nucleotide reverse transcriptase inhibitors of HIV on endogenous nucleotide pools

Jennifer E Vela1, Michael D Miller2, Gerald R Rhodes1 and Adrian S Ray1*

1Department of Preclinical Drug Metabolism, Gilead Sciences, Inc., Foster City, CA, USA 2Department of Clinical Virology, Gilead Sciences, Inc., Foster City, CA, USA

*Corresponding author: E-mail: [email protected]

Background: Alterations in endogenous nucleotide pools Incubation of 10 µM AZT, either alone or in combination as a result of HIV therapy with nucleoside and nucleotide with other N(t)RTIs, increased 2′-deoxyadenosine triphos- reverse transcriptase inhibitors (N[t]RTIs) is a proposed phate, 2′-deoxyguanosine triphosphate and thymidine mechanism for therapy-related adverse events and drug triphosphate levels by up to 1.44-fold the concentrations interactions resulting in treatment failure. In vitro stud- observed in untreated cells. At higher than pharmaco- ies were performed in order to understand the effect of logical concentrations of AZT, evidence for inhibition of N(t)RTIs on endogenous nucleotide pools. 2′-deoxycytidylate deaminase and enzymes involved in the Methods: The T-cell line CEM-CCRF was treated with control salvage of thymidine was also observed. Phosphorylated antimetabolites or the N(t)RTIs abacavir, didanosine, lami- metabolites of TFV are known to inhibit purine nucleoside vudine, tenofovir (TFV) and zidovudine (AZT), either alone phosphorylase (PNP). However, in contrast to a potent PNP or in combination. The levels of natural 2′-deoxynucleoside inhibitor, TFV was unable to alter intracellular dNTP pools triphosphates (dNTP) and ribonucleoside triphophosphates upon addition of exogenous 2′-deoxyguanosine. were determined by liquid chromatography coupled with Conclusions: N(t)RTIs have the potential to alter nucleotide triple quadrupole mass spectrometry. pools; however, at the pharmacologically relevant concen- Results: Antimetabolites altered nucleotide pools in a trations, tested N(t)RTI or their combinations did not have manner consistent with their known mechanisms of action. an effect on nucleotide pools with the notable exception AZT was the only N(t)RTI that significantly altered dNTP pools. of AZT.

Introduction

Nucleoside and nucleotide reverse transcriptase triphosphate levels or endogenous dNTP pools could inhibitors (N[t]RTIs), used for the treatment of HIV alter the activity and toxicity of N(t)RTIs. infection, are activated by the cellular machinery respon- Treatment with zidovudine (AZT) is associated with sible for regulating endogenous 2′-deoxynucleoside adverse events including lipodystrophy and anaemia triphosphate (dNTP) and ribonucleoside triphosphate [2,3]. Different non-mutually exclusive mechanisms (rNTP) pools (summarized in Figure 1). Once formed, for toxicities associated with AZT therapy have been the active N(t)RTI triphosphate competes with its cor- proposed including a decrease in intracellular thymidine responding natural dNTP for incorporation by the viral triphosphate (TTP) because of inhibition of thymidine reverse transcriptase or host DNA polymerases caus- salvage [4–6], direct inhibition of cellular polymerases ing suppression of viral replication or toxicity, respec- through incorporation and chain termination [7,8], tively. N(t)RTIs could cause perturbations in endog- inhibition of 3′ to 5′ exonuclease proofreading by host enous nucleotide metabolism by causing competitive DNA polymerases by AZT monophosphate (AZTMP) inhibition, allosteric modulation or altered expression [9] and formation of a highly toxic metabolite through of nucleotide metabolizing enzymes (reviewed previ- the metabolism of the 3′-azido group to an amine by ously [1]). Due to the competitive nature of their incor- cytochrome P450 and their reductases [10]. A direct poration by polymerases, changes in either N(t)RTI role for AZTMP in toxicity observed in some cell types

Ray.indd 789 5/9/08 17:35:29 JE Vela et al.

has been suggested on the basis of the correlation of Purine nucleoside phosphorylase (PNP) is a crucial AZTMP levels with cytotoxicity in cultured cells [11]. enzyme in the regulation of purine nucleotide pools AZT treatment might affect TTP levels by inhibition [13]. Results supporting inhibition of PNP by phos- of cytosolic thymidine kinase (TK1), mitochondrial phorylated metabolites of tenofovir (TFV) as the TK (TK2) and thymidylate kinase (TMPK). AZT and mechanism for increased circulating levels of didanos- AZTMP have been observed to bind with similar affinity ine (ddI) when it is coadministered with TFV diso- to TK1 and TMPK as their respective natural substrates, proxil fumarate (TDF, the oral prodrug of TFV) [14] and the build-up of high levels of AZTMP might facilitate led to the suggestion of further clinical implications for inhibition of thymidine phosphorylation [12]. PNP inhibition [15]. Based on the effects of treatment

Figure 1. Pyrimidine and purine nucleotide metabolism pathways in mammalian cells

A dCTP CTP UTP dUTP TTP

RR RR dCDP CDP UDP dUDP TDP TS & DHFR TMPK dCMP CMP UMP dUMP TMP

dCK TK dCytidine Cytidine Uridine dUridine Thymidine dCMPDA UP TP dCDA Uracil Thymine

β-Alanine and Dihydrouracil and β-Aminoisobutyrate dihydrothymine

B dATP ATP GTP dGTP

RR RR

dADP ADP GDP dGDP IMP DH dAMP AMP IMP XMP GMP dGMP

APRT ADA HGPRT dAdenosine Adenosine Inosine Guanosine dGuanosine PNP dInosine PNP ADA Hypoxanthine Guanine

Adenine XOD Xanthine Urate XOD

Schemes of (A) pyrimidine and (B) purine nucleotide metabolism pathways in mammalian cells (reviewed in a prior publication [28]). The enzymes adenine phosphoribosyltransferase (APRT), adenosine deaminase (ADA), 2′-deoxycytidine deaminase (dCDA), 2′-deoxycytidine kinase (dCK), 2-deoxycytidylate deaminase (dCMPDA), dihydrofolate reductase (DHFR), dihydropyrimidine dehydrogenase (DD), hypoxanthine-guanine phosphoribosyltransferase (HGPRT), inosine monophosphate dehydrogenase (IMPDH), purine nucleoside phosphorylase (PNP), ribonucleotide reductase (RR), thymidine kinase (TK), thymidine phosphorylase (TP), thymidylate kinase (TMPK), thymidylate synthase (TS), uridine phosphorylase (UP) and xanthine oxidase (XOD) are shown. The nucleosides 2′-deoxyadenosine (dA), 2′-deoxycytidine (dC), 2′-deoxyguanosine (dG), 2′-deoxyuridine (dU), thymidine (T) and xanthine (X) are shown. Their mono-, di- and triphosphorylated forms are abbreviated as -MP, -DP and -TP, respectively.

Ray.indd 790 5/9/08 17:35:33 Effect of nucleoside and nucleotide analogues on nucleotide pools

with highly potent PNP inhibitors, including increased and rNTP were obtained from Spectra Stable Isotopes intracellular 2′-deoxyadenosine triphosphate (dATP) (Columbia, MD, USA). The PNP inhibitor 2-amino- and 2′-deoxyguanosine triphosphate (dGTP) and inhi- 7-[1-(2-fluorophenyl)-2-(R)-hydroxyethyl]-3,5-dihy- bition of T-cell division [16], it was proposed that TFV dro-4H-pyrrolo[3,2-d]pyrimidin-4-one (PNP405) was might cause an increase in dNTP pools resulting in synthesized by NAEJA Pharmaceutical (Edmonton, antagonism of the activity of other N(t)RTIs and the AB, Canada) by methods similar to those previously observation of decreased CD4+ T-cells with the combi- reported [26]. nation of TDF and non-dose-adjusted ddI [17,18]. Triple N(t)RTI-only regimens including ddI/ Cells lamivudine (3TC)/stavudine (d4T), abacavir (ABC)/ Human T leukaemic CEM-CCRF cells were obtained ddI/d4T, ABC/3TC/AZT, ABC/ddI/TDF, ddI/3TC/TDF from the American Type Culture Collection (Manas- and ABC/3TC/TDF have shown high rates of virologi- sas, VA, USA) and were maintained in RPMI 1640 cal non-response, viral rebound and resistance muta- media supplemented with 10% heat-inactivated fetal tion selection [19–24]. Results from clinical studies bovine serum, 100 units/ml penicillin G and 100 µg/ml with triple N(t)RTI-only regimens have led to a deci- streptomycin sulfate. sion by a Department of Health and Human Services panel to not recommend therapy with ABC/3TC/AZT Effect of antimetabolites and N(t)RTIs on for treatment-naive patients and to state that other tri- nucleotide pools ple N(t)RTI-only regimens including ABC/3TC/TDF CEM-CCRF cells were seeded at 0.5 million cells/ml in and ddI/3TC/TDF should not be considered at any tissue culture flasks and treated for 24 h with the anti- time [25]. Different hypotheses have been proposed metabolites ribavirin, hydroxyurea or methotrexate at to explain the poor performance of triple N(t)RTI- concentrations of 20 µM, 3 mM or 5 µM, respectively, only therapies including antagonism of intracellular or with N(t)RTIs either alone or in combination, each N(t)RTI phosphorylation resulting in reduced triphos- incubated at 10 µM. phate analogue levels, decreased activity of N(t)RTIs Following the 24 h incubations, cells were counted caused by increases in competing dNTP pools, an using a haemocytometer and isolated from drug- intrinsic lack of potency in N(t)RTI-only regimens, containing media by spinning through a layer of oil as and the overlapping resistance profiles of N(t)RTIs described previously [14]. After removal of the oil layer, included in these regimens. cells were resuspended in 70% methanol and stored at In order to understand the effects of N(t)RTIs and -20ºC overnight to facilitate extraction of nucleotides. their combinations on nucleotide pools, including the Cellular debris was then removed by centrifugation and effect of AZT on thymidine phosphorylating enzymes the supernatant evaporated in a MiVac Duo concentra- and the inhibition of PNP by TFV, in vitro studies were tor (Genevac Inc., Valley Cottage, NY, USA). Samples carried out in CEM-CCRF cells. CEM-CCRF cells were were resuspended in 20 mM tetrabutylammonium chosen for their rapid cell division and high capacity acetate at a concentration of 1 million cells/10 µl for for de novo and salvage nucleotide synthesis. A liquid analysis by LC/MS/MS. chromatography coupled to tandem mass spectrometry method (LC/MS/MS) was developed using stable Effect of AZT and TFV on nucleotide salvage pathways isotope-labelled nucleotides to allow for accurate and To study the concentration-dependent effects of AZT precise simultaneous analysis of the effects on both on the salvage of thymidine, 24 h incubations were dNTP and rNTP endogenous pools. done in the presence of 10 µM [13C15N]thymidine and 10, 100 and 1,000 µM AZT. The appearance of Methods [13C15N]TTP was monitored as well as the levels of natural nucleotides. PNP inhibition-dependent changes in Reagents dNTP pools caused by 10 µM TFV relative to 10 µM Cell culture supplies were purchased from Invitrogen PNP405 were studied following 24 h treatment of cells (Carlsbad, CA, USA) and HyClone fetal bovine serum was with the respective compound and 10 µM exogenous purchased from Fisher Scientific (Pittsburgh, PA, USA). 2′-deoxyguanosine. Cells were prepared for analysis as ABC, TFV and 3TC were supplied by Gilead Sciences described above. Inc. (Foster City, CA, USA). AZT, 2′-deoxyguanosine, 2′-deoxyadenosine, ddI, hydroxurea, methotrexate and Measurement of endogenous dNTP and rNTP levels by ribavirin were obtained from Sigma–Aldrich (St Louis, LC/MS/MS MO, USA). The trisodium salts of the natural dNTP Previously reported methods for the quantitative and rNTP were also obtained from Sigma–Aldrich. determination of intracellular dATP using standard Stable isotope-labelled [13C15N]thymidine and dNTP curves of stable isotope-labelled dATP in cellular

Antiviral Therapy 13.6 791

Ray.indd 791 5/9/08 17:35:33 JE Vela et al.

matrices and ion-pairing reversed phase LC coupled from 1 million cells were approximately 20 fmol on the to MS/MS detection [27] were extended to include all column for each nucleotide. To ensure accuracy and dNTP and rNTP. Analyses were performed using a precision within 20% over the course of the analy- CTC Analytics Leap autosampler (Leap Technologies, sis, a quality control sample containing natural and Carrboro, NC, USA), an LC-20AD tertiary pump LC stable isotope-labelled nucleotides was injected at the system (Shimadzu Scientific Instruments, Columbia, beginning and end of each analytical batch. MD, USA) and a Sciex API 4000 triple quadrupole mass spectrometer (Applied Biosystems/MDS Sciex, Data analysis Foster City, CA, USA) running positive-ion and mul- All experiments were completed at least three indepen- tiple reaction monitoring modes. Briefly, the natu- dent times in duplicate or triplicate. The statistical sig- ral and stable isotope-labelled nucleotide pairs were nificance of changes in endogenous nucleotide pools in tuned for MS/MS detection by 10 µl/min infusions in treated relative to non-treated cells was evaluated using 0.1 mM tetrabutylammonium acetate and 20% ace- unpaired two-tailed Student’s t-tests. tonitrile to allow MS signal for each respective pair within 20% of one another. Parameters optimized Results included the mass to charge ratio of the parent and daughter ion, declustering potential, collision energy Measurement of dNTP and rNTP levels and the effect and collision exit potential. When necessary, a small of known antimetabolites on nucleotide pools correction factor was used to correct for differences in The levels of dNTP (from 51.1 to 108 pmol/million signals observed for the respective natural and stable cells) and rNTP (from 656 to 3,570 pmol/million cells) isotope nucleotide pairs. in untreated CEM-CCRF cells were similar to those Standard curves of each stable isotope-labelled reported in the literature for cultured tumour cell lines nucleotide were prepared in dried cellular extract [28]. Studies were carried out with control antime- reconstituted in 20 mM tetrabutylammonium acetate tabolites to determine the sensitivity of CEM-CCRF and samples were separated using a multistage linear cells to inhibition of nucleotide metabolizing enzymes gradient of acetonitrile in a buffer containing 0.25 mM (Table 1). Hydroxyurea, methotrexate and ribavirin tetrabutylammonium hydroxide and 4 mM ammo- are inhibitors of ribonucleotide reductase, dihydrofo- nium phosphate on a 3 µm, 1.0×100 mm Luna C18 late reductase and inosine monophosphate dehydro- reverse phase column (Phenomenex, Torrance, CA, genase (IMPDH), respectively (Figure 1). Consistent USA). The natural nucleotides in cellular samples were with their respective mechanisms of action, hydroxyu- then quantitated by comparing their peak area to that rea treatment resulted in the depletion of all dNTP observed for a standard curve of the corresponding to 0.08–0.43-fold of the levels in untreated cells and stable isotope nucleotides. Nucleotide concentrations methotrexate markedly reduced 2′-deoxyguanosine were reported in pmol/million cells on the basis of the triphosphate (dGTP) and TTP to 0.08- and 0.12- determined amount of nucleotide in each injection of fold of the levels, respectively, in untreated cells. No extract from 1 million cells. marked difference in cell number was observed fol- Standard curves including seven concentrations and lowing the 24 h incubations. General cytotoxicity was covering at least three orders of magnitude typically apparent in cells treated with these concentrations had linearity exceeding an r2 value of 0.99. Lower limits of hydroxurea and methotrexate by the decreases of quantitation for 10 µl injections containing extracts observed in almost all dNTP and rNTP pools. IMPDH

Table 1. Effect of antimetabolites on endogenous dNTP and rNTP levels in CEM-CCRF cells Fold change* dNTP rNTP Incubation dATP dGTP dCTP TTP ATP GTP CTP UTP

3 mM Hydroxyurea 0.08 ±0.03†‡ 0.24 ±0.12‡ 0.18 ±0.09‡ 0.43 ±0.19‡ 0.29 ±0.20§ 0.32 ±0.20¶ 0.44 ±0.29¶ 0.17 ±0.11‡ 5 µM Methotrexate 0.47 ±0.06‡ 0.08 ±0.01‡ 0.94 ±0.06 0.12 ±0.02‡ 0.39 ±0.10‡ 0.48 ±0.14§ 0.36 ±0.10‡ 0.25 ±0.06‡ 20 µM Ribavirin 0.70 ±0.34 0.41 ±0.14§ 1.80 ±0.50¶ 1.93 ±0.40§ 0.77 ±0.11 0.21 ±0.03‡ 0.83 ±0.17 0.86 ±0.17

*Fold change in 2′-deoxynucleoside triphosphates (dNTP) or ribonucleoside triphophosphates (rNTP) present in a side-by-side no treatment control (fold change = concentration in treated cells divided by concentration in untreated cells). Levels in untreated cells of 2′-deoxyadenosine triphosphate (dATP), 2′-deoxyguanosine triphosphate (dGTP), 2′-deoxycytidine triphosphate (dCTP) and thymidine triphosphate (TTP) were 107 ±44, 51.1 ±27.2, 53.7 ±31.0 and 108 ±27 pmol/million, respectively, and of adenosine triphosphate (ATP), guanosine triphosphate (GTP), cytidine triphosphate (CTP) and uridine triphosphate (UTP) were 3,570 ±1,840, 1,050 ±550, 656 ±427 and 1,690 ±900, respectively (mean ±s d of at least 20 independent experiments done in duplicate). †Values represent the mean ±s d of at least three independent experiments done in duplicate for each treatment condition. Significance from no treatment control on the basis of unpaired two-tailed Student’s t-test assuming equal variance ‡P<0.001, §P<0.01, ¶P<0.05.

Ray.indd 792 5/9/08 17:35:34 Effect of nucleoside and nucleotide analogues on nucleotide pools

inhibition by ribavirin resulted in decreases in dGTP the observation of thymidine kinase and thymidilate and GTP pools that were reversed by the addition of kinase-dependent formation of [13C15N]TTP. Treat- 20 µM exogenous 2′-deoxyguanosine to the media ment of cells with 10 µM [13C15N]thymidine alone (data not shown). Increased levels of 2′-deoxycytidine did not affect endogenous pyrimidine dNTP levels, triphosphate (dCTP) and TTP upon ribavirin treat- but dATP and dGTP levels were 0.7- and 0.5-fold the ment can be explained by a reduction in dGTP-medi- levels observed in untreated cells, respectively. In con- ated feedback inhibition of ribonucleotide reductase. trast to cells treated with 10 µM AZT alone, the pres- Similar results were obtained for the IMPDH inhibitor ence of [13C15N]thymidine resulted in no effect of 10 mycophenolate (data not shown). µM AZT on dNTP pools. Decreased AZT phosphorylation caused by competition for phosphorylating Effect of N(t)RTIs and their combinations on enzymes by exogenous thymidine and partial abroga- nucleotide pools tion of its activity has been reported previously [29]. In order to understand the effect of pharmacologically In the presence of [13C15N]thymidine, AZT caused relevant concentrations of N(t)RTI or their combi- a concentration-dependent increase in dCTP when nations on nucleotide pools, CEM-CCRF cells were incubated at higher than pharmacological concentra- treated for 24 h with 10 µM N(t)RTI and endog- tions of 100 and 1,000 µM. Treatment with 100 µM enous nucleotide pools were quantitated. All treat- AZT caused increased levels of endogenous TTP, by ments that included AZT caused significant increases contrast, treatment with 1,000 µM resulted in signifi- in dATP, dGTP and TTP up to 1.44-fold more than cant decreases in the levels of TTP, dATP and dGTP. those observed in untreated cells with the excep- Salvage of [13C15N]thymidine was only significantly tion of AZT alone and AZT/3TC, where increases in inhibited at the highest concentration of AZT tested dATP did not reach statistical significance (Table 2). (Figure 2B). ABC, 3TC, TFV and ddI when incubated alone or ABC/3TC, TFV/ABC, TFV/3TC, TFV/ddI, ddI/3TC, Effect of TFV on PNP activity TFV/ABC/3TC and TFV/ddI/3TC when incubated in PNP inhibition is most readily observed in cultured combination did not show a significant effect on dNTP cells by treating with exogenous 2′-deoxyguanosine (data not shown). Furthermore, ABC and 3TC did not and determining if dGTP and, to a lesser extent, dATP show any additional effect on dNTP when combined are increased in the presence of a potential inhibitor with AZT over those observed with AZT alone. No [16]. To test the relative ability of TFV to alter dNTP N(t) RTI was found to significantly alter rNTP levels pools through PNP inhibition, CEM-CCRF cells were (data not shown). treated with exogenous 2′-deoxyguanosine and either TFV or PNP405, a potent inhibitor of PNP, for 24 h Concentration-dependent effect of AZT on and dNTP pools were quantified. The addition of exog- thymidine salvage enous 2′-deoxyguanosine, TFV, or both, did not cause a To further understand the effect of AZT on thymidine significant change in dNTP pools (data not shown and salvage, the concentration-dependent effect of AZT in Figure 3, respectively). PNP405 alone had no effect on the presence of 10 µM [ 13C15N]thymidine on thymidine nucleotide pools (data not shown); however, the addi- phosphorylation and nucleotide pools was assessed tion of exogenous 2′-deoxyguanosine caused a large (Figure 2). The [13C15N]thymidine with stable isotopes increase in dGTP, a slightly smaller increase in dATP incorporated in the base and the sugar allowed for and marked decreases in dCTP and TTP.

Table 2. Effect of 10 µM AZT either alone or in combination on endogenous dNTP pools in CEM-CCRF cells Fold change* Combination N(t)RTI† dATP dGTP dCTP TTP

Single AZT 1.15 ±0.17‡ 1.38 ±0.18§ 1.24 ±0.31 1.44 ±0.22¶ Double AZT/3TC 1.13 ±0.21 1.32 ±0.20¶ 1.14 ±0.25 1.31 ±0.28# AZT/ABC 1.15 ±0.10# 1.25 ±0.22# 1.02 ±0.13 1.50 ±0.18# Triple AZT/ABC/3TC 1.23 ±0.11¶ 1.31 ±0.10§ 1.05 ±0.12 1.25 ±0.18#

*Fold change in 2′-deoxynucleoside triphosphates (dNTP) or ribonucleoside triphophosphates (rNTP) present in a side-by-side no treatment control (fold change = concentration in treated cells divided by concentration in untreated cells). Levels in untreated cells of 2′-deoxyadenosine triphosphate (dATP), 2′-deoxyguanosine triphosphate (dGTP), 2′-deoxycytidine triphosphate (dCTP) and thymidine triphosphate (TTP) were 107 ±44, 51.1 ±27.2, 53.7 ±31.0, 108 ±27 pmol/million, respectively (mean ±s d of at least 20 independent experiments done in duplicate). †Tested nucleoside and nucleotide reverse transcriptase inhibitors (N[t]RTI) or their combinations not containing AZT did not show a statistically significant effect on dNTP pools. No effect of N(t)RTI or their combinations was observed on rNTP pools. ‡Values represent the mean ±s d of at least three independent experiments done in duplicate. Significance of difference from no treatment control on the basis of unpaired two-tailed Student’s t‑test assuming equal variance §P<0.001, ¶P<0.01, #P<0.05. ABC, abacavir; AZT, ziduvudine; 3TC, lamivudine.

Antiviral Therapy 13.6 793

Ray.indd 793 5/9/08 17:35:34 JE Vela et al.

Discussion TFV [40], but TDF is not observed in the systemic circulation [41]. The resulting higher than pharma- Consistent with a previous report showing changes in cological intracellular concentrations of TFV and its dNTP levels after treatment of CEM-CCRF cells with phosphorylated metabolites were likely to have played 25 µM AZT [30], we observed modest increases in TTP, a role in their observations. In this study, we chose dATP and dGTP following a 24 h incubation with 10 µM to use CEM-CCRF cells because their rapid division AZT. Similarly, it has been observed that inhibition of and active nucleotide metabolism make them highly the major replicative DNA polymerases by an agent like sensitive to perturbations in nucleotide metabolism. aphidicolin causes an increase in intracellular dNTP pools The sensitivity of these cells was evident in the marked by uncoupling de novo synthesis from consumption dur- effects of control antimetabolites. ing DNA replication [31]. Furthermore, the extent of Various hypotheses have been proposed to explain dNTP pool increase in cells incubated with nucleotide the poor performance of triple N(t)RTI-only regimens. analogues has been related to intracellular activation and potency against polymerase α, δ and ε [32]. The incor- Figure 2. Concentration-dependent effect of AZT in CEM- poration of a chain-terminating AZTMP by cellular CCRF cells treated for 24 h DNA polymerases has been found to be relatively poor in enzymatic assays [33–36]; however, its toxicity to human A bone marrow cells in vitro has been shown to correlate 350 with its incorporation into cellular DNA [7]. One factor favouring stable chain-termination of host DNA is the 300 dCTP TTP * accumulation of high levels of intracellular AZTMP [12], dATP dGTP which has been shown to be a potent inhibitor of 3′ to 5′ 250 exonuclease proofreading [9]. Taken together, our results 200 coupled with previous observations suggest that the rea- * son for increased TTP, dATP and dGTP upon treatment 150 * of CEM-CCRF cells with AZT is due to inhibition of host 100 DNA polymerases. †

We have previously shown that TFV monophosphate Concentration, pmol/million 50 * is a competitive inhibitor of human PNP-dependent * ddI phosphorolysis with an inhibition constant (K ) of 0 i -10 100 1,000 126 nM and incubation of TFV with CEM-CCRF cells was found to cause a dose-dependent inhibition of ddI AZT concentration, µM degradation [14]. These results illustrate PNP inhibi- B 90 tion as the most plausible mechanism for the observation of increased ddI exposure when it is coadminis- 80 tered with TFV. However, based on the requirement for 70 N]TTP,

near complete inhibition of PNP to affect dNTP pools 15 60 C and T-cell function [38], TFV therapy is not expected 13 50 to affect endogenous dNTP pools. Consistent with this 40 hypothesis and unlike a more potent PNP inhibitor, in the current report we show that 10 µM TFV was pmol/million 30 unable to alter dNTP pools following treatment of cells 20 Concentration [ with exogenous 2′-deoxyguanosine. 10 * In contrast to observations made in this study, 0 another report found that ABC or TDF treatment -10 100 1,000 could increase intracellular TTP, dATP and dGTP in + primary CD4 T-cells stimulated with phytohaemag- AZT concentration, µM glutinin and interleukin-2, whereas little or no change in dNTP levels were observed in stimulated CD8+ Concentration-dependent effect of zidovudine (AZT) in the presence of T-cells [39]. Although, the difference in results from 10 µM [13C15N]thymidine on (A) 2′-deoxynucleoside triphosphate pools and the two studies might be in part due to differences in (B) salvage of [13C15N]thymidine to [13C15N]thymidine triphosphate (TTP). Values are the mean ±s d of three independent experiments carried out in duplicate. methodology including the cell types studied, it should The statistical significance of changes in relative cells treated with [13C15N] be noted that Singer et al. [39] chose to incubate cells thymidine in the absence of AZT were assessed using unpaired two-tailed Student’s t-tests (*P<0.001, †P<0.05). dATP, 2′-deoxyadenosine triphosphate; with TDF. TDF is the clinically used oral prodrug of dCTP, 2′-deoxycytidine triphosphate; dGTP, 2′-deoxyguanosine triphosphate; TFV that loads cells >1,000-fold more efficiently than TTP, thymidine triphosphate.

Ray.indd 794 5/9/08 17:35:35 Effect of nucleoside and nucleotide analogues on nucleotide pools

The results of this report suggest that pharmaco- efficacious than other triple N(t)RTI-only regimens dynamic drug interactions caused by increased [48,49]. These results might be explained by the competing dNTP pools are unlikely to play a role in orthogonal resistance profile of AZT, which does not the higher rates of virological failures observed for include M184V or K65R. these combinations. On the basis of similar triphos- Although N(t)RTIs mechanistically have the potential phate analogue formation observed for the combina- to alter nucleotide pools, the results reported here sug- tions of ABC/TFV and ddI/TFV in vitro and in clini- gest that at pharmacologically relevant concentrations cal pharmacokinetic studies [42–45], an intracellular tested, N(t)RTI or their combinations do not have an drug interaction caused by competition for N(t)RTI effect on nucleotide pools with the notable exception activation because of potentially overlapping phos- of AZT. It was also found that TFV does not have suf- phorylation pathways is also an unlikely explanation ficient potency against PNP to perturb dNTP pools. for the poor performance of triple N(t)RTI therapies. The availability of a convenient, accurate and precise In contrast, there is support for the intrinsic lack of method for measurement of nucleotides will allow for potency for triple N(t) RTI-only regimens, perhaps the confirmation of these in vitro findings in cells from caused by suboptimal distribution to certain sites of patients undergoing HIV therapy. HIV replication. Supporting this hypothesis, higher rates of treatment failure have been observed in Acknowledgements patients treated with the regimens of ABC/ddI/d4T and ABC/3TC/AZT who had viral loads ≥100,000 The authors would like to thank William J Watkins copies/ml [20,46]. The continued replication of virus (Gilead Sciences, Inc.) for overseeing the synthesis of likely allows for viral evolution through parallel PNP405 and for thoughtful discussion of the data. resistance pathways held in common by two or more of the N(t)RTI being dosed. For example, ABC/ddI/ Disclosure statement TDF therapy was associated with high rates of K65R [22], ddI/3TC/TDF and ABC/3TC/TDF selected for The authors are all employed by Gilead Sciences, Inc., M184V with or without K65R [23,24] and ABC/ the marketer of tenofovir disoproxil fumarate (Viread). ddI/d4T selected for K65R with or without S68G [47]. Recent clinical studies with 3TC/TDF/AZT and References ABC/3TC/TDF/AZT suggest that N(t)RTI-only regi- 1. Ray AS. Intracellular interactions between nucleos(t)ide mens containing AZT appear to be relatively more inhibitors of HIV reverse transcriptase. AIDS Rev 2005; 7:113–125. 2. Richman DD, Fischl MA, Grieco MH, et al. The toxicity Figure 3. Changes in dNTP levels following the addition of of azidothymidine (AZT) in the treatment of patients with AIDS and AIDS-related complex. A double-blind, placebo- 10 µM exogenous 2′-deoxyguanosine in the presence or absence controlled trial. N Engl J Med 1987; 317:192–197. of either TFV or the potent PNP inhibitor PNP405 for 24 h 3. Gallant JE, DeJesus E, Arribas JR, et al. Tenofovir DF, emtricitabine, and efavirenz vs. zidovudine, lamivudine, and efavirenz for HIV. N Engl J Med 2006; 354:251–260. * 4. McKee EE, Bentley AT, Hatch M, Gingerich J, 600 Susan‑Resiga D. Phosphorylation of thymidine and AZT in dCTP TTP heart mitochondria: elucidation of a novel mechanism of 500 dATP dGTP AZT cardiotoxicity. Cardiovasc Toxicol 2004; 4:155–167. 5. Lynx MD, McKee EE. 3′-Azido-3′-deoxythymidine (AZT) 400 * is a competitive inhibitor of thymidine phosphorylation in isolated rat heart and liver mitochondria. Biochem Pharmacol 2006; 72:239–243. 300 6. Lynx MD, Bentley AT, McKee EE. 3′-Azido-3′- deoxythymidine (AZT) inhibits thymidine phosphorylation in 200 isolated rat liver mitochondria: a possible mechanism of AZT hepatotoxicity. Biochem Pharmacol 2006; 71:1342–1348.

Concentration, pmol/million 100 7. Sommadossi JP, Carlisle R, Zhou Z. Cellular pharmacology * of 3 -azido-3 -deoxythymidine with evidence of * ′ ′ 0 incorporation into DNA of human bone marrow cells. Mol --10 µM TFV 10 µM PNP405 Pharmacol 1989; 36:9–14. 10 µM dG 8. Vazquez-Padua MA, Starnes MC, Cheng YC. Incorporation of 3′-azido-3′-deoxythymidine into cellular DNA and its removal in a human leukemic cell line. Cancer Commun 1990; 2:55–62. Values are the mean ±s d of three independent experiments done in duplicate. The statistical significance of changes relative to untreated cells were assessed 9. Bridges EG, Faraj A, Sommadossi JP. Inhibition of using unpaired two-tailed Student’s t-tests (*P<0.05). dATP, 2′-deoxyadenosine mammalian DNA polymerase-associated 3′ to 5′ triphosphate; dCTP, 2′-deoxycytidine triphosphate; dG, 2′-deoxyguanosine; dGTP, exonuclease activity by 5′-monophosphates of 3′-azido-3′- 2′-deoxyguanosine triphosphate; dNTP, 2′-deoxynucleoside triphosphate; PNP, deoxythymidine and 3′-amino-3′-deoxythymidine. Biochem purine nucleoside phosphorylase; TFV, tenofovir; TTP, thymidine triphosphate. Pharmacol 1993; 45:1571–1576.

Antiviral Therapy 13.6 795

Ray.indd 795 5/9/08 17:35:36 JE Vela et al.

10. Cretton EM, Sommadossi JP. Reduction of 3′-azido-2′,3′- 28. Traut TW. Physiological concentrations of purines and dideoxynucleosides to their 3′-amino metabolite is mediated pyrimidines. Mol Cell Biochem 1994; 140:1–22. by cytochrome P-450 and NADPH-cytochrome P-450 reductase in rat liver microsomes. Drug Metab Dispos 29. Mitsuya H, Weinhold KJ, Furman PA, et al. 3′-Azido- 1993; 21:946–950. 3′-deoxythymidine (BW A509U): an antiviral agent that inhibits the infectivity and cytopathic effect of 11. Sales SD, Hoggard PG, Sunderland D, Khoo S, Hart CA, human T-lymphotropic virus type III/lymphadenopathy- Back DJ. Zidovudine phosphorylation and mitochondrial associated virus in vitro. Proc Natl Acad Sci U S A 1985; toxicity in vitro. Toxicol Appl Pharmacol 2001; 177:54–58. 82:7096–7100. 12. Furman PA, Fyfe JA, St Clair MH, et al. Phosphorylation of 30. Fridland A, Connelly MC, Ashmun R. Relationship 3 -azido-3 -deoxythymidine and selective interaction of the ′ ′ of deoxynucleotide changes to inhibition of DNA 5′-triphosphate with human immunodeficiency virus reverse transcriptase. Proc Natl Acad Sci U S A 1986; 83:8333–8337. synthesis induced by the antiretroviral agent 3′-azido- 3′-deoxythymidine and release of its monophosphate by 13. Bzowska A, Kulikowska E, Shugar D. Purine nucleoside human lymphoid cells (CCRF-CEM). Mol Pharmacol 1990; phosphorylases: properties, functions, and clinical aspects. 37:665–670. Pharmacol Ther 2000; 88:349–425. 31. Nicander B, Reichard P. Relations between synthesis 14. Ray AS, Olson L, Fridland A. Role of purine nucleoside of deoxyribonucleotides and DNA replication in 3T6 phosphorylase in interactions between 2′,3′-dideoxyinosine fibroblasts. J Biol Chem 1985; 260:5376–5381. and allopurinol, ganciclovir, or tenofovir. Antimicrob Agents Chemother 2004; 48:1089–1095. 32. Kramata P, Downey KM. 9-(2-phosphonylmethoxyethyl) derivatives of purine nucleotide analogs: A comparison 15. Kakuda TN, Anderson PL, Becker SL. CD4 cell decline with didanosine and tenofovir and failure of triple of their metabolism and interaction with cellular DNA nucleoside/nucleotide regimens may be related. AIDS 2004; synthesis. Mol Pharmacol 1999; 56:1262–1270. 18:2442–2444. 33. Huang P, Farquhar D, Plunkett W. Selective action of 16. Bantia S, Ananth SL, Parker CD, Horn LL, Upshaw R. 3′-azido-3′-deoxythymidine 5′-triphosphate on viral reverse Mechanism of inhibition of T-acute lymphoblastic leukemia transcriptases and human DNA polymerases. J Biol Chem cells by PNP inhibitor-BCX-1777. Int Immunopharmacol 1990; 265:11914–11918. 2003; 3:879–887. 34. Copeland WC, Chen MS, Wang TS. Human DNA 17. Negredo E, Molto J, Burger D, et al. Unexpected CD4 cell polymerases alpha and beta are able to incorporate anti- count decline in patients receiving didanosine and tenofovir- HIV deoxynucleotides into DNA. J Biol Chem 1992; based regimens despite undetectable viral load. AIDS 2004; 267:21459–21464. 18:459–463. 35. Nickel W, Austermann S, Bialek G, Grosse F. Interactions 18. Barrios A, Rendon A, Negredo E, et al. Paradoxical CD4+ of azidothymidine triphosphate with the cellular DNA T-cell decline in HIV-infected patients with complete virus polymerases alpha, delta, and epsilon and with DNA suppression taking tenofovir and didanosine. AIDS 2005; primase. J Biol Chem 1992; 267:848–854. 19:569–575. 36. Johnson AA, Ray AS, Hanes J, et al. Toxicity of antiviral 19. van Leeuwen R, Katlama C, Murphy RL, et al. A nucleoside analogs and the human mitochondrial DNA randomized trial to study first-line combination therapy polymerase. J Biol Chem 2001; 276:40847–40857. with or without a protease inhibitor in HIV-1-infected patients. AIDS 2003; 17:987–999. 37. Liou JY, Krishnan P, Hsieh CC, Dutschman GE, 20. Gerstoft J, Kirk O, Obel N, et al. Low efficacy and high Cheng YC. Assessment of the effect of phosphorylated frequency of adverse events in a randomized trial of metabolites of anti-human immunodeficiency virus and the triple nucleoside regimen abacavir, stavudine and anti-hepatitis B virus pyrimidine analogs on the behavior didanosine. AIDS 2003; 17:2045–2052. of human deoxycytidylate deaminase. Mol Pharmacol 2003; 63:105–110. 21. Gulick RM, Ribaudo HJ, Shikuma CM, et al. Triple- nucleoside regimens versus efavirenz-containing regimens 38. Lewandowicz A, Tyler PC, Evans GB, Furneaux RH, for the initial treatment of HIV-1 infection. N Engl J Med Schramm VL. Achieving the ultimate physiological goal in 2004; 350:1850–1861. transition state analogue inhibitors for purine nucleoside 22. Winston A, Pozniak A, Mandalia S, Gazzard B, Pillay D, phosphorylase. J Biol Chem 2003; 278:31465–31468. Nelson M. Which nucleoside and nucleotide backbone 39. Singer S, Irlbeck D, Pruvost A, et al. Changes in levels of combinations select for the K65R mutation in HIV-1 reverse dNTP, carbovir-TP and tenofovir-DP in purified human transcriptase. AIDS 2004; 18:949–951. CD4+ and CD8+ peripheral blood T cells following 23. Jemsek J, Hutcherson P, Harper E. Poor virologic responses treatment with tenofovir +/- abacavir +/- lamivudine. XV and early emergence of resistance in treatment naive, HIV- International Drug Resistance Workshop: Basic Principles infected patients receiving a once daily triple nucleoside and Clinical Implications. 13–17 June 2006, Sitges, Spain. regimen of didanosine, lamivudine, and tenofovir DF. 11th Abstract 146. Conference on Retroviruses and Opportunistic Infections. 8–11 February 2004, San Francisco, CA, USA. Abstract 51. 40. Robbins BL, Srinivas RV, Kim C, Bischofberger N, Fridland A. Anti-human immunodeficiency virus activity 24. Gallant JE, Rodriguez AE, Weinberg WG, et al. Early and cellular metabolism of a potential prodrug of the acyclic virologic nonresponse to tenofovir, abacavir, and lamivudine nucleoside phosphonate 9-R-(2-phosphonomethoxypropyl) in HIV-infected antiretroviral-naive subjects. J Infect Dis adenine (PMPA), bis(isopropyloxymethylcarbonyl)PMPA. 2005; 192:1921–1930. Antimicrob Agents Chemother 1998; 42:612–617. 25. Panel on Antiretroviral Guidelines for Adults and Adolescents. 41. Barditch-Crovo P, Deeks SG, Collier A, et al. Phase I/ dolescents. Guidelines for the use of antiretroviral agents in II trial of the pharmacokinetics, safety, and antiretroviral HIV-1-infected adults and adolescents. Department of Health activity of tenofovir disoproxil fumarate in human and Human Services. (Updated 29 January 2008. Accessed immunodeficiency virus-infected adults. Antimicrob Agents 24 April 2008.) Available from http://aidsinfo.nih.gov/ ContentFiles/AdultandAdeloscentGL.pdf Chemother 2001; 45:2733–2739. 26. Prashad M, Har D, Chen L, Kim HY, Repic O, 42. Robbins BL, Wilcox CK, Fridland A, Rodman JH. Blacklock TJ. An efficient and large-scale enantioselective Metabolism of tenofovir and didanosine in quiescent or synthesis of PNP405: a purine nucleoside phosphorylase stimulated human peripheral blood mononuclear cells. inhibitor. J Org Chem 2002; 67:6612–6617. Pharmacotherapy 2003; 23:695–701. 27. Vela JE, Olson LY, Huang A, Fridland A, Ray AS. 43. Pruvost A, Negredo E, Benech H, et al. Measurement of Simultaneous quantitation of the nucleotide analog adefovir, intracellular didanosine and tenofovir phosphorylated its phosphorylated anabolites and 2′-deoxyadenosine metabolites and possible interaction of the two drugs triphosphate by ion-pairing LC/MS/MS. J Chromatogr B in human immunodeficiency virus-infected patients. Analyt Technol Biomed Life Sci 2007; 848:335–343. Antimicrob Agents Chemother 2005; 49:1907–1914.

Ray.indd 796 5/9/08 17:35:36 Effect of nucleoside and nucleotide analogues on nucleotide pools

44. Ray AS, Myrick F, Vela JE, et al. Lack of a metabolic and 47. Roge BT, Katzenstein TL, Obel N, et al. K65R with and antiviral drug interaction between tenofovir, abacavir and without S68: a new resistance profile in vivo detected in lamivudine. Antivir Ther 2005; 10:451–457. most patients failing abacavir, didanosine and stavudine. Antivir Ther 2003; 8:173–182. 45. Hawkins T, Veikley W, St Claire RL, III, Guyer B, Clark N, Kearney BP. Intracellular pharmacokinetics 48. DART Virology Group and Trial Team. Virological of tenofovir diphosphate, carbovir triphosphate, and response to a triple nucleoside/nucleotide analogue regimen lamivudine triphosphate in patients receiving triple- over 48 weeks in HIV-1-infected adults in Africa. AIDS nucleoside regimens. J Acquir Immune Defic Syndr 2005; 2006; 20:1391–1399. 39:406–411. 49. Moyle G, Higgs C, Teague A, et al. An open-label, 46. Staszewski S, Keiser P, Montaner J, et al. Abacavir- randomized comparative pilot study of a single-class lamivudine-zidovudine vs indinavir-lamivudine-zidovudine quadruple therapy regimen versus a 2-class triple therapy in antiretroviral-naive HIV-infected adults: A randomized regimen for individuals initiating antiretroviral therapy. equivalence trial. JAMA 2001; 285:1155–1163. Antivir Ther 2006; 11:73–78. Accepted for publication 24 June 2008

Antiviral Therapy 13.6 797

Ray.indd 797 5/9/08 17:35:36