Anti-Human Immunodeficiency Virus Activity, Cross-Resistance, Cytotoxicity, and Intracellular Pharmacology of the 3′-Azido-2′,3′-Dideoxypurine Nucleosides

Abstract

Although the approved nucleoside reverse transcriptase (RT) inhibitors (NRTI) are integral components of therapy for human immunodeficiency virus type 1 (HIV-1) infection, they can have significant limitations, including the selection of NRTI-resistant HIV-1 and cellular toxicity. Accordingly, there is a critical need to develop new NRTI that have excellent activity and safety profiles and exhibit little or no cross-resistance with existing drugs. In this study, we report that the 3′-azido-2′,3′-dideoxypurine nucleosides (ADPNs) 3′-azido-2′,3′-dideoxyadenosine (3′-azido-ddA) and 3′-azido-2′,3′-dideoxyguanosine (3′-azido-ddG) exert potent antiviral activity in primary human lymphocytes and HeLa and T-cell lines (50% inhibitory concentrations [IC₅₀s] range from 0.19 to 2.1 μM for 3′-azido-ddG and from 0.36 to 10 μM for 3′-azido-ddA) and that their triphosphate forms are incorporated as efficiently as the natural dGTP or dATP substrates by HIV-1 RT. Importantly, both 3′-azido-ddA and 3′-azido-ddG retain activity against viruses containing K65R, L74V, or M184V (IC₅₀ change of <2.0-fold) and against those containing three or more thymidine analog mutations (IC₅₀ change of <3.5-fold). In addition, 3′-azido-ddG does not exhibit cytotoxicity in primary lymphocytes or epithelial or T-cell lines and does not decrease the mitochondrial DNA content of HepG2 cells. Furthermore, 3′-azido-ddG is efficiently phosphorylated to 3′-azido-ddGTP in human lymphocytes, with an intracellular half-life of the nucleoside triphosphate of 9 h. The present data suggest that additional preclinical studies are warranted to assess the potential of ADPNs for treatment of HIV-1 infection.

Nucleoside reverse transcriptase (RT) inhibitors (NRTI) were the first drugs used to treat human immunodeficiency virus type 1 (HIV-1) infection, and they remain integral components of nearly all antiretroviral regimens. NRTI are analogs of 2′-deoxyribonucleosides that lack a 3′-OH group on the ribose sugar/pseudosugar. Once metabolically converted by host cell kinases to their corresponding triphosphate forms (NRTI-TPs), they inhibit HIV-1 replication by acting as chain terminators of RT-mediated DNA synthesis (8). To date, eight NRTI have been approved for clinical use, namely, didanosine (ddI), tenofovir disoproxil fumarate (TDF), zalcitabine (ddC), lamivudine (3TC), emtricitabine (FTC), abacavir sulfate (ABC), zidovudine (AZT), and stavudine (d4T). Although combination therapies that contain two or more NRTI have profoundly reduced morbidity and mortality from HIV-1 infection, the approved NRTI can have significant limitations, including acute and chronic toxicity, pharmacokinetic interactions with other antiretroviral agents, and the selection of drug-resistant variants of HIV-1 that exhibit cross-resistance to other NRTI. Thus, the discovery and development of novel NRTI with excellent activity and safety profiles and limited or no cross-resistance with currently available drugs are critical for effective therapy of HIV-1 infection.

NRTI discovery has traditionally relied on the synthesis of novel analogs that are initially evaluated for anti-HIV activity and toxicity in cell culture systems. Those with acceptable antiviral and pharmacological properties are then reevaluated for activity against drug-resistant variants of HIV-1. This empirical discovery process is usually unpredictable, and it is generally difficult to determine whether a given nucleoside analog will show activity against commonly occurring NRTI-resistant variants. Therefore, we conducted structure-activity resistance studies to specifically identify nucleoside base and sugar moieties that retain activity against drug-resistant HIV-1 (17, 23). Based on the results from these studies, the 3′-azido-2′,3′-dideoxypurine nucleosides (ADPNs) emerged as a lead class of nucleoside analogs that demonstrated excellent activity against viruses that contained the multi-NRTI resistance mutation K65R (17) or the multiple thymidine analog mutations (TAMs) D67N/K70R/T215F/K219Q (23). In the current study, we have further assessed the antiviral activity and intracellular pharmacology of the ADPNs (Fig. 1). In particular, we have compared their antiviral activities and abilities to inhibit purified HIV-1 RT with those of approved NRTI; evaluated their potential to inhibit a broad panel of drug-resistant viruses; studied the metabolism of 3′-azido-2′,3′-dideoxyguanosine (3′-azido-ddG); and investigated potential cellular cytotoxicity, including mitochondrial toxicity.

FIG. 1.

Chemical structures of NRTI investigated in this study.

MATERIALS AND METHODS

Nucleosides and nucleotides.

3′-Azido-ddG and 3′-azido-2′,3′-dideoxyadenosine (3′-azido-ddA) were purchased from Berry & Associates (Dexter, MI) or synthesized by standard approaches (1, 6, 10, 11). 3′-Azido-3′-deoxythymidine (AZT), 2′,3′-dideoxycytidine (ddC), and 2′,3′-dideoxyinosine (ddI) were purchased from Sigma-Aldrich (St. Louis, MO). (1S,4R)-4-[2-Amino-6-(cyclopropyl-amino)-9H-purin-9-yl]-2-cyclopentene-1-methanol succinate (ABC), (R)-9-(2-phosphonylmethoxypropyl)adenine (TDF), (−)-2′,3′-dideoxy-3′-thiacytidine (3TC), (−)-2′,3′-dideoxy-5-fluoro-3′-thiacytidine (FTC), and 3′-deoxy-2′,3′-didehydrothymidine (d4T) were obtained from the AIDS Research and Reference Reagent Program. AZT-TP, 3′-azido-ddGTP, 3′-azido-ddATP, ddCTP, and ddATP were purchased from TriLink Biotechnologies (San Diego, CA); 3TC-TP and d4T-TP were purchased from Sierra Bioresearch (Arizona); and TDF diphosphate and carbovir (CBV)-TP were purchased from Moravek Biochemicals (Brea, CA).

Cells and viruses.

MT-2 cells were maintained in RPMI 1640 supplemented with 10% fetal bovine serum, 10 mM HEPES buffer, 50 IU of penicillin per ml, and 50 μg of streptomycin per ml. The P4/R5 reporter cell line (provided by Ned Landau, Salk Institute, La Jolla, CA) was cultured in Dulbecco's modified Eagle's medium-phenol red-free medium supplemented with 10% fetal bovine serum, 50 IU of penicillin/ml, 50 μg of streptomycin/ml, and 0.5 μg of puromycin/ml. Human peripheral blood mononuclear cells (PBMC) were isolated by Ficoll-Hypaque discontinuous-gradient centrifugation from healthy seronegative donors, as described previously (19, 21). Cells were stimulated with phytohemagglutinin A (PHA) for 2 to 3 days prior to infection. Stock wild-type (WT) and mutant viruses were prepared as reported previously (17). The 50% cell culture infective dose for the virus stock was determined for MT-2 cells or P4/R5 cells by threefold endpoint dilution assays and calculated using the Reed and Muench equation (18).

Drug susceptibility assays.

Drug susceptibility assays were performed using P4/R5 reporter cells, MT-2 cells, or PBMC, as reported previously (17, 19, 21, 23). Inhibition of virus replication was calculated as the concentration of compound required to inhibit virus replication by 50% (IC₅₀) and was determined as described previously (4). Resistance values were determined by dividing the IC₅₀ for mutant HIV-1 by the IC₅₀ for WT HIV-1.

Cellular and mitochondrial toxicity assays.

ADPN toxicity was evaluated in human PBMC, MT-2 cells, P4/R5 cells, Vero cells (kidney epithelial cells from the African green monkey), and CEM cells (a human-T-cell-derived cell line). Briefly, log-phase cells were seeded at 5 × 10³ to 5 × 10⁴ cells/well in 96-well plates containing 10-fold serial dilutions of test compound. The cultures were incubated for 2 to 4 days, after which viability was determined by constitutive β-galactosidase production (P4/R5 cells) or staining with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) solution (Promega, Madison, WI). The median 50% cytotoxic concentration was determined from the concentration-response curve by use of the median effect method (4). Mitochondrial toxicity was evaluated by assessing the mitochondrial DNA (mtDNA) content of HepG2 cells treated with inhibitor. Low-passage-number HepG2 cells were seeded at 5,000 cells/well in collagen-coated 24-well plates and were cultured in the absence or presence of 10 μM or 100 μM NRTI for 14 days. The cell culture media and NRTI were replenished every 3 to 4 days. On day 14, the total cellular nucleic acid was extracted using a MagNA Pure liquid chromatography (LC) total nucleic acid isolation kit (Roche), and the mtDNA (cytochrome c oxidase subunit II) and β-actin DNA (Applied Biosystems) were amplified in parallel with a real-time PCR assay (9700 HT sequence detection system; Applied Biosystems), as described previously (13, 25). All samples were tested in duplicate. The amount of target (mtDNA) was normalized to the amount of an endogenous control (β-actin DNA) and was compared to the levels for untreated control cells. Lactic acid quantification was performed using a commercial d-lactic acid/l-lactic acid test kit (Boehringer Mannheim/R-Biopharma, Roche).

Intracellular pharmacology.

The accumulation and bioconversion of 3′-azido-ddG to 3′-azido-ddGTP and the intracellular half-life of 3′-azido-ddGTP in PHA-stimulated human PBMC were investigated by incubating various concentrations of 3′-azido-ddG with 24 × 10⁶ to 30 × 10⁶ cells for 12 h at 37°C in 5% CO₂-humidified air. The 3′-azido-ddG metabolites were then separated by anion-exchange chromatography using a 100- by 2.1-mm Biobasic AX instrument (Thermo Electron, Bellefonte, PA) attached to a Dionex/LC Packings Ultimate 3000 modular LC system. The mobile phase consisted of three parts: (i) 50 mM ammonium acetate buffer, (ii) aqueous ammonium hydroxide, pH 10.6, and (iii) acetonitrile. The flow rate was maintained at 400 μl/min. The eluting nucleosides/tides were detected via a TSQ Quantum Ultra triple quadrupole mass spectrometer (Thermo Electron) attached to the chromatography system. The mass spectrometer was operated in positive-ionization mode, with a spray voltage of 4.5 kV, sheath gas at 50 (arbitrary units), ion sweep gas at 5.0 (arbitrary units), auxiliary gas at 10 (arbitrary units), and a capillary temperature of 300°C. The collision cell pressure was maintained at 1.5 mtorr.

Incorporation of NRTI-TPs by HIV-1 RT.

HIV-1 RT was purified to homogeneity, as described previously (14, 15). A [γ-³²P]ATP 5′-end-labeled 20-nucleotide DNA primer (5′-TCGGGCGCCACTGCTAGAGA-3′) annealed to a 57-nucleotide DNA template (5′-CTCAGACCCTTTTAGTCAGAATGGAAAnTCTCTAGCAGTGGCGCCCGAACAGGGACA-3′) was used in all experiments. The DNA template contained A, C, T, or G at position 30 (n), which enabled us to evaluate the kinetics of different NRTI-TPs by using the same primer. Rapid quench experiments were carried out using a Kintek RQF-3 instrument (Kintek Corporation, Clarence, PA). In all experiments, 300 nM RT and 60 nM DNA template/primer were preincubated in reaction buffer (50 mM Tris-HCl, pH 7.5, 50 mM KCl) prior to being mixed with equivalent volumes of nucleotide in the same reaction buffer containing 20 mM MgCl₂. Reactions were terminated at times ranging from 10 ms to 30 min by quenching with 0.5 M EDTA, pH 8.0. The quenched samples were mixed with equal volumes of gel loading buffer (98% deionized formamide, 10 mM EDTA, and 1 mg/ml each of bromophenol blue and xylene cyanol) and denatured at 85°C for 5 min, and the products were separated from the substrates on a 7 M urea-16% polyacrylamide gel. Product formation was analyzed using a Bio-Rad GS525 molecular imager (Bio-Rad Laboratories, Inc., Hercules, CA). Data obtained from these kinetic assays were fitted by nonlinear regression using Sigma Plot software (Jandel Scientific) with the appropriate equations (12).

RESULTS

Antiviral activities of ADPNs versus approved NRTI in different cell lines.

To compare the anti-HIV activities of ADPNs and FDA-approved NRTI, we determined their inhibitory potencies in three different cell lines (Table 1). These included a single-replication-cycle drug susceptibility assay with P4/R5 cells (a HeLa cell line stably transfected with a Tat-activated β-galactosidase gene under the control of an HIV long terminal repeat promoter), a multiple-replication-cycle drug susceptibility assay with MT-2 cells, and a drug susceptibility assay with human PBMC. 3′-Azido-ddG exhibited excellent antiviral activity in all of the cell types tested. In fact, its activity was two- to fourfold more potent than that of ABC and comparable to those of ddI, ddC, d4T, and TDF. By comparison, the antiviral activity of 3′-azido-ddA was three- to fivefold less than that of 3′-azido-ddG but still comparable to the potency of ABC.

TABLE 1.

Anti-HIV activities of ADPNs and FDA-approved NRTI in different cell lines

NRTI	IC50 (μM)a
	P4/R5 cells	MT-2 cells	PBMC
3′-Azido-ddG	2.1 ± 0.9	1.7 ± 1.1	0.19 ± 0.19
3′-Azido-ddA	10.0 ± 4.9	6.4 ± 3.1	0.36 ± 0.14
ABC	6.2 ± 1.7	4.8 ± 1.6	0.91-0.98b
TDF	4.7 ± 2.3	0.54 ± 0.61	0.014 ± 0.012
ddI	2.3 ± 0.9	0.49 ± 0.32	0.1c
3TC	0.78 ± 0.48	0.60 ± 0.46	0.03-0.15b
FTC	0.17 ± 0.07	0.044 ± 0.036	0.001-0.69b
ddC	1.4 ± 0.7	0.13 ± 0.18	0.2b
AZT	0.19 ± 0.11	0.031 ± 0.024	0.01-0.09b
d4T	5.7 ± 3.3	1.9 ± 0.8	0.1c

^aData are derived from at least three independent experiments and are expressed as means ± standard deviations.

^bData are derived from reference 20 and are reported as a range.

^cData are derived from reference 22 and are reported as a range.

Activities of 3′-azido-ddATP and 3′-azido-ddGTP against recombinant HIV-1 RT.

To evaluate the ability of ADPN-TP to inhibit HIV-1 RT, we determined the enzyme's substrate selectivity for 3′-azido-ddATP and 3′-azido-ddGTP as well as all of the approved NRTI-TPs by using the pre-steady-state kinetic approach (Table 2). The results indicate that nucleotide analogs containing a 3′-azido-2′,3′-dideoxy sugar are as efficiently incorporated by WT HIV-1 RT as are the natural substrates (selectivity values range from 0.74 for 3′-azido-ddATP to 1.1 for 3′-azido-ddGTP). In contrast, HIV-1 RT exhibited greater selectivity for natural substrates than all other NRTI-TPs tested, with selectivity values ranging from 2.6 (TTP versus d4T-TP) to 17.8 (dCTP versus 3TC-TP).

TABLE 2.

Pre-steady-state kinetic parameters determined for the incorporation of clinically approved NRTI-TPs and 3′-azido- ddATP and 3′-azido-ddGTP by WT HIV-1 RT

dNTPa/NRTI-TP	kpol (s−1)b	Kd (μM)b	kpol/Kd (μM−1 s−1)	Selectivityc
dATP	24.4 ± 3.9	3.2 ± 1.2	7.6
ddATP	5.4 ± 2.3	5.3 ± 1.7	1.0	7.5
TDF-DP	6.2 ± 1.0	4.3 ± 1.0	1.4	5.3
3′-Azido-ddATP	23.2 ± 0.5	2.3 ± 0.8	10.2	0.7
dCTP	1.2 ± 0.7	1.1 ± 0.4	1.1
ddCTP	0.4 ± 0.1	2.7 ± 1.3	0.2	6.7
3TC-TP	0.023 ± 0.004	0.37 ± 0.20	0.06	17.8
dGTP	7.5 ± 1.1	4.2 ± 1.4	1.8
CBV-TP	0.6 ± 0.2	4.0 ± 1.8	0.16	11.2
3′-Azido-ddGTP	4.8 ± 1.1	2.8 ± 0.9	1.7	1.1
TTP	9.2 ± 1.3	2.2 ± 0.8	4.2
d4T-TP	4.3 ± 1.2	2.7 ± 1.1	1.6	2.6
AZT-TP	8.8 ± 0.2	2.0 ± 0.1	4.4	1.0

^adNTP, deoxynucleoside triphosphate.

^bData are derived from at least three independent experiments and are expressed as means ± standard deviations. k_pol, turnover number; K_d, apparent dissociation constant.

^cSelectivity was determined as follows: (k_pol/K_d)_dNTP/(k_pol/K_d)_NRTI-TP.

Activity of ADPNs against a panel of NRTI-resistant viruses.

To further characterize the anti-HIV-1 activity of ADPNs, we evaluated 3′-azido-ddA and 3′-azido-ddG against a panel of NRTI-resistant viruses. This panel included recombinant viruses that contained the discrimination mutation(s) K65R, L74V, M184V, or A62V/V75I/F77L/F116Y/Q151M (Q151M) (Table 3) and viruses that contained different combinations of TAMs (e.g., M41L/L210W/T215Y [AZT2], M41L/D67N/K70R/T215F/K219Q [AZT7], M41L/D67N/K70R/L210W/T215Y/K219Q [AZT9], or M41L/69SS/L210W/T215Y [69 insertion] [69SS refers to a diserine insert between residues 69 and 70]) (Table 4). The results show that both 3′-azido-ddA and 3′-azido-ddG exhibited little or no loss of activity against viruses with the K65R, L74V, or M184V mutation (Table 3). In comparison with AZT, both compounds also showed potent activity against all TAM-containing viruses (Table 4). For example, HIV-1_AZT7 was >500-fold resistant to AZT but less than 3.5-fold resistant to 3′-azido-ddA and 3′-azido-ddG. Both 3′-azido-ddA and 3′-azido-ddG, however, were less active against HIV-1_Q151M, and 3′-azido-ddG also lost activity against HIV-1_{69 insertion}.

TABLE 3.

Anti-HIV activities of 3′-azido-ddA and 3′-azido-ddG against a panel of HIV-1 viruses harboring NRTI discrimination mutations in P4/R5 cells

Virus	IC50, μM (fold resistance)a
	AZT	3′-Azido-ddA	3′-Azido-ddG	3TC	ddI	TDF
WT	0.19 ± 0.11	10.7 ± 4.9	2.1 ± 0.9	0.78 ± 0.48	2.3 ± 0.9	4.7 ± 2.3
K65R	0.21 ± 0.15 (1.1)	9.8 ± 8.4 (0.9)	5.0 ± 1.3 (2.3)	60.5 ± 66.8 (77)	6.3 ± 2.9 (2.7)	11.4 ± 3.8 (2.5)
L74V	0.21 ± 0.08 (1.1)	13.7 ± 5.7 (1.2)	2.9 ± 0.9 (1.4)	NAb	8.3 ± 1.0 (3.6)	NA
M184V	0.18 ± 0.16 (1.0)	8.9 ± 2.3 (0.8)	1.7 ± 0.1 (0.8)	>60 (>75)	4.6 ± 1.8 (2.0)	NA
Q151M	213.7 ± 12.3 (1,124)	70.1 ± 10.8 (6.5)	72.9 ± 29.5 (34.7)	NA	NA	NA

^aData are derived from at least three independent experiments and are expressed as means ± standard deviations.

^bNA, data not available.

TABLE 4.

Anti-HIV activities of 3′-azido-ddA and 3′-azido-ddG against a panel of HIV-1 viruses harboring NRTI excision mutations in P4/R5 cells

Virus	IC50, μM (fold resistance)a
	AZT	3′-Azido-ddA	3′-Azido-ddG
WT	0.19 ± 0.11	10.7 ± 4.9	2.1 ± 0.9
AZT2	10.4 ± 8.9 (54.4)	24.2 ± 3.7 (2.2)	5.2 ± 2.3 (2.5)
AZT3b	11.9 ± 11.6 (62.7)	19.4 ± 8.1 (1.8)	3.7 ± 1.4 (1.8)
AZT	96.7 ± 29.3 (507.4)	31.6 ± 3.7 (2.9)	7.6 ± 2.2 (3.5)
AZT	58.6 ± 9.2 (307.4)	37.5 ± 6.2 (3.5)	7.9 ± 4.9 (3.7)
69 Insertion	204.6 ± 18.6 (1,076)	29.9 ± 5.9 (2.8)	26.4 ± 8.8 (12.5)

^aData are derived from at least three independent experiments and are expressed as means ± standard deviations.

^bD67N/K70R/T215F/K219Q.

Intracellular metabolism of 3′-azido-ddG.

Because NRTI must be metabolically converted by host cell kinases to their corresponding triphosphate forms, pharmacological properties, such as the efficiency of NRTI cellular uptake and anabolism to the NRTI-TP and the intracellular half-life of the NRTI-TP, influence NRTI activity. To gain insight into the pharmacology of the ADPN class, we measured the accumulation and bioconversion of 3′-azido-ddG to 3′-azido-ddGTP and the intracellular half-life of 3′-azido-ddGTP in PHA-stimulated human PBMC. 3′-Azido-ddG accumulated in PBMC and was converted to 3′-azido-ddGTP in a concentration-dependent manner (Fig. 2A). At 10 μM of extracellular drug, the intracellular triphosphate levels reached 0.20 ± 0.04 pmol/10⁶ cells. In addition, the intracellular half-life of 3-azido-ddGTP was calculated to be 8.90 ± 0.14 h (Fig. 2B).

FIG. 2.

Intracellular metabolism of 3′-azido-ddG. (A) Dose response of cellular uptake of 3′-azido-ddG and conversion to 3′-azido-ddGTP in PHA-stimulated human PBMC. Data are derived from at least three independent experiments and are expressed as means ± standard deviations. (B) Decay of 3′-azido-ddGTP in PBMC after incubation with 50 μM 3′-azido-ddG for 12 h. The half-life of 3′-azido-ddGTP is calculated to be 8.9 ± 0.14 h. Data are derived from at least three independent experiments and are expressed as means ± standard deviations.

ADPN cytotoxicity.

To evaluate potential ADPN cytotoxicity, we determined the abilities of 3′-azido-ddA and 3′-azido-ddG to inhibit the growth of five different cell lines (Table 5). 3′-Azido-ddA and 3′-azido-ddG did not exhibit cytotoxicity at concentrations less than 50 μM and 100 μM, respectively. In vitro and clinical investigations have also established that NRTI cause defective mtDNA replication in cardiac and skeletal muscle, liver, adipocytes, and peripheral nerves. To investigate whether ADPNs exhibit mitochondrial toxicity, we used real-time PCR to quantify the mtDNA content in HepG2 cells (a human hepatocellular liver carcinoma cell line) that were cultured for 14 days in the presence of different concentrations of 3′-azido-ddG and 3′-azido-ddA (Table 6). In each assay, ddC and (+)-β-2′,3′-dideoxy-3′-thiacytidine [(+)-BCH-189], which are known to cause mitochondrial toxicity (25), and 3TC, which is not toxic, were included as controls. These data show that even at high concentrations of drug (100 μM), 3′-azido-ddG does not decrease the mtDNA content of HepG2 cells. 3′-Azido-ddG also did not increase lactic acid production, another important index of mitochondrial toxicity. In contrast, 3′-azido-ddA did not cause significant reductions in mtDNA or β-actin DNA levels up to 10 μM, but at high drug concentrations (e.g., 100 μM) there was a decrease in mtDNA content and an increase in lactic acid production.

TABLE 5.

Cytotoxicities of ADPNs in five different cell lines

NRTI	CC50 (μM)a
	PBMC	MT-2 cells	Vero cells	CEM cells	P4/R5 cells
AZT	>100	150	>100	86.2	>270
3′-Azido-ddA	74.3	200	50.6	50.6	>270
3′-Azido-ddG	>100	140	>100	>100	>270

^aData are expressed as the mean values from at least three independent experiments. CC₅₀, 50% cytotoxic concentration.

TABLE 6.

Effects of 3′-azido-ddG and 3′-azido-ddA on in vitro levels of mtDNA and lactic acidosis in HepG2 cells

NRTI	NRTI concn (μM)	% Inhibition (mtDNA/β-actin DNA)	% mtDNA relative to that for untreated cellsa	% Lactic acid production relative to β-actin DNAb
Untreated control			100 (99-100)	100 ± 9.8
3′-Azido-ddG	10	<1/<1	120 (112-128)	114 ± 4.8
	100	<1/<1	120 (105-137)	96 ± 6.5
3′-Azido-ddA	10	10.91/13.80	100 (100-110)	106 ± 3.6
	100	98.99/85.15	6.7 (4.8-9.3)	159 ± 20.3
3TC	10	<1/<1	119 (107-135)	105 ± 17.7
ddC	10	99.88/61.39	0.3 (0.2-0.4)	170 ± 7.1
(+)-BCH-189	100	99.96/94.30	0.6 (0.5-0.8)	416 ± 109

^aSamples were analyzed by RT-PCR following 14-day treatment of HepG2 cells with each NRTI. The data are expressed as average values from at least two independent experiments. The range of values obtained is shown in parentheses.

^bCells were cultured for 14 days in the presence of NRTI, and then lactic acid production was quantified as described in Materials and Methods. The data represent means ± standard errors from two independent experiments.

DISCUSSION

The development of novel NRTI with potent antiviral activity, good safety profiles, and limited or no cross-resistance with currently available drugs is important for the effective therapy of HIV-1 infection in treatment-naïve and -experienced patients. In this study, we describe the antiviral, biochemical, and intracellular pharmacology of the ADPNs, a lead class of compounds that were found to be active against HIV-1 containing K65R or D67N/K70R/T215F/K219Q in structure-activity resistance analyses (17, 23). It should be noted that the antiviral activity of ADPNs was first reported in the late 1980s (2, 9), but these studies did not investigate the intracellular pharmacology or activity of ADPNs against drug-resistant HIV-1.

There are two major mechanisms of HIV-1 resistance to NRTI: discrimination against NRTI-TP incorporation and excision of NRTI-MP from a terminated template primer. The next generation of NRTI should be capable of inhibiting viruses that contain common mutations from either resistance phenotype. In this regard, we show that both 3′-azido-ddA and 3′-azido-ddG have excellent activities against drug-resistant variants of HIV-1 that contain the K65R, L74V, or M184V mutation, which discriminates against NRTI-TP incorporation, and against viruses containing three or more TAMs (e.g., M41L, D67N, K70R, L210W, T215F/Y, or K219Q) associated with the NRTI-MP excision phenotype. GS-9131 {9-(R)-4′-(R)-[[[(S)-1-[(ethoxycarbonyl)ethyl] amino]phenoxyphosphonyl]methoxy]-2′-fluoro-1′-furanyladenine} is the only other nucleoside analog currently in preclinical development that exhibits a similar ability to inhibit a broad panel of NRTI-resistant viruses (5). The fact that WT HIV-1 RT incorporates nucleotide analogs containing a 3′-azido-2′,3′-dideoxy sugar as efficiently as it does the natural substrates may help to explain why the ADPNs retain activity against mutants, such as K65R, L74V, or M184V mutants, although experiments using enzymes with these mutations are required to prove this. Recently, we reported that HIV-1 selected for five mutations in RT (L74V, F77L, V106I, L214F, and K476N) after 90 passages in increasing concentrations of inhibitor in MT-2 cells that reduced susceptibility to 3′-azido-ddG by ∼10-fold (16). In contrast, we generated HIV-1 that was >16,000-fold resistant to AZT after only 65 passages under similar experimental conditions (3). Taken together, these data highlight the high genetic barrier and novel mutational pathway required for ADPN resistance. Of note, the L74V mutation alone conferred only a 1.3-fold increase in IC₅₀ compared to that of the WT.

In addition to having excellent activity against WT and NRTI-resistant HIV-1, the ADPNs exhibit favorable toxicity and intracellular-pharmacology profiles. 3′-Azido-ddA and 3′-azido-ddG did not exhibit cytotoxicity in four different cell lines at concentrations less than 50 μM and 100 μM, respectively. Furthermore, neither NRTI (3′-azido-ddG or 3′-azido-ddA) decreased the mtDNA content of HepG2 cells, suggesting that this class of NRTI may not exhibit mitochondrial toxicity in vivo. 3′-Azido-ddG is also efficiently phosphorylated to 3′-azido-ddGTP in human PBMC, with the intracellular triphosphate levels reaching 0.20 ± 0.04 pmol/10⁶ cells after exposure to 10 μM drug. By comparison, the triphosphate levels of other purine nucleosides, such as ddI (ddATP) and ABC (CBV-TP), reach concentrations of 0.05 to 0.07 pmol/10⁶ cells and 0.09 to 0.2 pmol/10⁶ cells, respectively, under identical experimental conditions (7, 24). Finally, the intracellular half-life of 3′-azido-ddGTP was calculated to be 8.90 ± 0.14 h. By comparison, the measured half-life for AZT-TP ranges from 3 to 4 h (20, 22).

Taken together, the results presented here suggest that ADPNs are attractive clinical development candidates for the treatment of patients with NRTI-resistant HIV-1. In this regard, ongoing studies in our laboratories are focused on improving the antiviral activity of this class of NRTI by introduction of novel modified purine bases and by synthesis of prodrugs.