Relationship between 3′-Azido-3′-Deoxythymidine Resistance and Primer Unblocking Activity in Foscarnet-Resistant Mutants of Human Immunodeficiency Virus Type 1 Reverse Transcriptase

Abstract

Phosphonoformate (foscarnet) is a pyrophosphate (PP_i) analogue and a potent inhibitor of human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT), acting through the PP_i binding site on the enzyme. HIV-1 RT can unblock a chain-terminated DNA primer by phosphorolytic transfer of the terminal residue to an acceptor substrate (PP_i or a nucleotide such as ATP) which also interacts with the PP_i binding site. Primer-unblocking activity is increased in mutants of HIV-1 that are resistant to the chain-terminating nucleoside inhibitor 3′-azido-3′-deoxythymidine (AZT). We have compared the primer-unblocking activity for HIV-1 RT containing various foscarnet resistance mutations (K65R, W88G, W88S, E89K, S117T, Q161L, M164I, and the double mutant Q161L/H208Y) alone or in combination with AZT resistance mutations. The level of primer-unblocking activity varied over a 150-fold range for these enzymes and was inversely correlated with foscarnet resistance and directly correlated with AZT resistance. Based on published crystal structures of HIV-1 RT, many of the foscarnet resistance mutations affect residues that do not make direct contact with the catalytic residues of RT, the incoming deoxynucleoside triphosphate (dNTP), or the primer-template. These mutations may confer foscarnet resistance and reduce primer unblocking by indirectly decreasing the binding and retention of foscarnet, PP_i, and ATP. Alternatively, the binding position or orientation of PP_i, ATP, or the primer-template may be changed in the mutant enzyme complex so that molecular interactions required for the unblocking reaction are impaired while dNTP binding and incorporation are not.

Phosphonoformate (foscarnet) inhibits a wide variety of DNA and RNA polymerases including human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) (40, 43). Foscarnet competes with pyrophosphate (PP_i) for binding in PP_i exchange reactions, suggesting that foscarnet interacts with the site on the enzyme where PP_i is formed (7, 38). On structural grounds, the site of interaction for PP_i (and, presumably, for foscarnet) should lie within the deoxynucleoside triphosphate (dNTP) binding site since PP_i is formed from the β and γ phosphates of the dNTP substrate, but foscarnet inhibition of DNA synthesis in vitro is noncompetitive with dNTP substrates (58), indicating that foscarnet and dNTP bind to distinguishable forms of RT. In addition, median-effect analysis comparing foscarnet and 3′-azido-3′-deoxythymidine (AZT) triphosphate (AZTTP) alone or as mixtures has shown that inhibition by these compounds is mutually exclusive (10, 22, 50), implying that binding of either inhibitor prevents binding of the other to the same enzyme molecule.

In intact cells, the inhibition of HIV-1 by foscarnet and AZT is synergistic (10, 22). The mechanism of this synergy is unclear, but it has recently been demonstrated that AZT monophosphate (AZTMP) incorporation can be reversed by the primer-unblocking activity of RT, in which the chain-terminating residue is transferred to an acceptor substrate such as PP_i or an NDP or NTP (1, 4, 32, 34). Hence, the level of AZT inhibition observed in vivo is dependent both on AZTMP incorporation by the polymerase activity of RT and on AZTMP removal by the primer-unblocking reaction. The primer-unblocking reaction is dependent on the catalytic aspartate residues of the polymerase active site (34) and is equivalent to the reversal of the polymerization reaction (pyrophosphorolysis); however, there is evidence that ATP, rather than PP_i, may be the most physiologically relevant substrate (3, 32, 46). Primer unblocking is inhibited by low concentrations of foscarnet (P. R. Meyer et al., unpublished data), which would explain the synergistic interaction between foscarnet and AZT. Treatment of HIV-1-infected patients with AZT leads to selection of mutations at positions 41, 67, 70, 210, 215, and 219 in the RT coding region (45). RTs containing the AZT resistance mutations show little increase in their ability to discriminate against AZTTP incorporation (5, 20, 23-25); however, the mutant enzymes have increased ability to remove AZTMP after it has been incorporated (1, 32).

Numerous foscarnet resistance mutations have been identified in HIV-1 RT either by site-directed mutagenesis or by isolation from virus culture or clinical specimens. Mutations introduced by site-directed mutagenesis at codons 72, 89, 90, 113, 114, 115, 151, 154, 183, and 190 in RT were shown to decrease sensitivity to foscarnet (17, 26, 27, 29, 39, 42, 49). Most of these residues are directly involved in dNTP or primer-template binding, and the mutations are encountered only rarely in vivo, presumably because they have detrimental effects on polymerase activity. The mutations W88S, W88G, Q161L, and H208Y were originally identified in clinical isolates of HIV-1 from foscarnet-treated patients (31). W88G, E89K, L92I, S156A, and Q161L/H208Y were obtained by serial passage of HIV in MT-2 cells in the presence of escalating concentrations of foscarnet (13, 31, 52). Alkyglycerol conjugates of foscarnet have been developed to improve cellular uptake and reduce toxicity (14, 15, 21). Serial passage of HIV-1 in the presence of these conjugated forms of foscarnet yielded additional mutations including S117T, F160Y, and M164I (usually in combination with the L214F mutation) (13), which also conferred resistance to unconjugated foscarnet in virus culture assays. Finally, the K65R mutation was originally isolated from patients treated with 3′-dideoxycytidine or 3′-dideoxyinosine (12, 59) and subsequently shown to confer resistance to other chain terminators (45) and also to foscarnet (13, 48).

Foscarnet resistance mutations confer hypersensitivity to AZT and suppress AZT resistance mutations (31, 52, 54), suggesting an in vivo relationship between foscarnet and AZT resistance; this is supported by earlier studies showing synergistic inhibition of HIV-1 by these two compounds (10, 22). Arion et al. (2) have recently shown that the A114S mutation, which confers foscarnet resistance in vitro (2, 26), is deficient in primer-unblocking activity with either PP_i or ATP as the acceptor substrate and that the enhanced primer-unblocking activity of AZT-resistant HIV-1 RT is greatly diminished when the A114S mutation is also present. Therefore, the interactions observed in phenotypic resistance assays, as well as the biochemical evidence, suggest that AZT and foscarnet resistance mutations have opposite effects on a common biochemical reaction.

In this study we have focused on the foscarnet resistance mutations that were selected in virus culture or from clinical isolates. We show that quantitative differences in unblocking activity are inversely correlated with foscarnet resistance and directly correlated with AZT resistance. From structural considerations, many of the residues altered by the foscarnet resistance mutations are unlikely to interact directly with the polymerase active site, the incoming dNTP, or the primer terminus. Hence, indirect effects on the structure of RT must be invoked to explain foscarnet resistance. Noncompetitive inhibition of dNTP incorporation by foscarnet can be explained if foscarnet binding occurs when the primer terminus occupies the dNTP binding site on the enzyme (i.e., prior to translocation of the enzyme to the next position on the template), and the mutually exclusive binding of foscarnet and AZTMP can be explained if foscarnet can bind to RT only prior to translocation and AZTTP can bind only to an RT molecule that has already translocated to the next position. We suggest that foscarnet inhibits HIV replication by forming a dead-end complex at the PP_i binding site on RT that resembles the “closed” complex induced by dNTP binding and that the foscarnet resistance mutations act directly or indirectly to alter the interactions of RT with foscarnet and/or primer-template.

MATERIALS AND METHODS

Cells and viruses.

MT-2 cells (obtained from the AIDS Research and Reference Reagent Program, National Institute of Allergy and Infectious Diseases, National Institutes of Health; contributed by D. Richman) 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. Wild-type (WT) and mutant virus stocks were prepared in MT-2 cells and subjected to titer determination as previously described (31, 37).

Antiviral susceptibility assays.

Antiviral activity was determined by inoculating MT-2 cells with WT or mutant virus stocks (multiplicity of infection, 0.01 50% tissue culture infective dose/cell) followed by incubation in the presence of two- or threefold serial dilutions of foscarnet or AZT. The amount of HIV-1 p24 antigen released into the culture supernatant was determined 7 days after infection and used to determine the concentration of each drug required to reduce p24 antigen production by 50% (IC₅₀) (31).

Expression and purification of HIV-1 RT.

His-tagged HIV-1 RT was prepared as previously described (34). Mutations were introduced by the megaprimer PCR method (44) using Pwo polymerase (Boehringer Mannheim). The DNA sequence of the mutant clones was determined by the dideoxy method. The RNA-dependent DNA polymerase assay was described previously (55). One unit is the amount of enzyme required for incorporation of 1.0 nmol of [³H]dTMP in 10 min at 37°C using poly(rA)-oligo(dT) as the primer-template. The specific activities of the WT and mutant enzymes were as follows: WT RT, 20,000 U/mg; W88G RT, 19,000 U/mg; D67N/K70R/T215F/K219Q RT, 10,000 U/mg; D67N/K70R/W88G/T215F/K219Q RT, 5,000 U/mg; M41L/T215Y RT, 27,000 U/mg; M41L/W88G/T215Y RT, 13,000 U/mg; Q161L RT, 8000 U/mg; Q161L/H208Y RT, 10,000 U/mg; D67N/K70R/T215Y/K219Q RT, 7,600 U/mg; D67N/K70R/Q161L/T215Y/K219Q RT, 3,700 U/mg; D67N/K70R/Q161L/H208Y/T215Y/K219Q RT, 5,400 U/mg; D67N/K70R/E89K/T215Y/K219Q RT, 3,000 U/mg; K65R RT, 16,000 U/mg; K65R/D67N/K70R/T215Y/K219Q RT, 12,000 U/mg; W88S RT, 20,000 U/mg; D67N/K70R/W88S/T215Y/K219Q RT, 7,000 U/mg; S117T RT, 12,000 U/mg; D67N/K70R/S117T/T215Y/K219Q RT, 6,600 U/mg; M164I RT, 19,000 U/mg; and D67N/K70R/M164I/T215Y/K219Q RT, 10,000 U/mg.

3′- and 5′-labeled chain-terminated oligonucleotide primer-templates.

L32 primer (5′-CTACTAGTTTTCTCCATCTAGACGATACCAGA-3′) was annealed with excess WL50 template (5′-GAGTGCTGAGGTCTTCATTCTGGTATCGTCTAGATGGAGAAAACTAGTAG-3′), and 16 pmol of primer-template was incubated with 50 μCi (15 pmol) of [α-³²P]ddATP (Amersham) and 40 pmol of HIV-1 RT in 300 μl of RB buffer (40 mM HEPES [pH 7.5], 20 mM MgCl₂, 60 mM KCl, 1 mM dithiothreitol, 2.5 % glycerol, 80 μg of bovine serum albumin per ml) for 30 min at 37°C to add a labeled chain-terminating nucleotide to the primer 3′ terminus. Oligonucleotide primer L33 (5′-CTACTAGTTTTCTCCATCTAGACGATACCAGAA-3′) was 5′-³²P labeled with T₄ polynucleotide kinase as described previously (34), annealed with excess template WL50, and chain terminated with AZTMP.

Removal of AZTMP or ddAMP from blocked primer-templates.

Primer rescue assays to measure the formation of unblocked primer were performed as previously described (32, 34). In brief, 200 nM WT or mutant HIV-1 RT was incubated with 5 nM AZTMP-terminated, 5′-³²P-labeled L33 primer-WL50 template and 3.2 mM ATP in RB buffer at 37°C for the times indicated for each experiment. The RT was inactivated by heat treatment, and the unblocked primer was extended by incubation with the exonuclease-free Klenow fragment of Escherichia coli DNA polymerase I (0.3 U; USB Corp.) and all four dNTPs (100 μM each). The products were heated at 90°C and fractionated on a 20% polyacrylamide gel containing 8 M urea.

Removal of ddAMP from ddAMP-terminated primer-template was monitored by detection of [³²P]Ap₄ddA as previously described (34). WT or mutant HIV-1 RT (200 nM) was incubated with 5 nM [³²P]ddAMP-terminated L32 primer-WL50 template and 3.2 mM ATP in reaction buffer at 37°C for the times indicated for each experiment. The incubation mixture was heated and fractionated on a 20% denaturing polyacrylamide gel as described above. For all experiments, ATP solutions were treated with thermostable pyrophosphatase (Roche Molecular Systems) to remove contaminating PP_i (34). A 100-μl volume of 64 mM ATP was incubated with 1 U of pyrophosphatase at 75°C for 10 min. If PP_i contamination remained after the first treatment, the procedure was repeated.

Products were quantitated with a Molecular Dynamics Storm 840 PhosphorImager. Apparent kinetic constants for dinucleoside polyphosphate synthesis or primer rescue were obtained under conditions of saturating amounts of RT so that the rate was limited by the concentration of the RT-primer-template complex, which was assumed to be equal to the primer concentration. The apparent k_cat and K_m were determined from the rate of product formation as a function of the nucleotide substrate by fitting the data to the Michaelis-Menten equation using Sigmaplot 4.0.

Electrophoretic mobility shift assays to detect RT-primer-template complexes with foscarnet.

Electrophoretic mobility shift assays were carried out as previously described (32, 56). Briefly, 5′-³²P-labeled AZTMP-terminated primer-template or 3′-[³²P]ddAMP-terminated primer-template (5 nM) was incubated in RB buffer with HIV-1 RT (200 nM) in the presence or absence of foscarnet and 1 μM unlabeled AZTTP (for reactions with AZTMP-terminated primer-template) or 1 μM unlabeled ddATP (for reactions with ddAMP-terminated primer-template) in a total volume of 10 μl. After incubation for 15 min at 37°C followed by 5 min on ice, 3 μl of heparin loading buffer (0.01 U/ml of heparin, 30% glycerol, 0.25% bromphenol blue, 0.25% xylene cyanol) was added. After an additional 7 min on ice, the samples were loaded onto a nondenaturing 6% polyacrylamide gel and electrophoresis was carried out in Tris-taurine buffer (90 mM Tris, 30 mM taurine) at 4°C.

RESULTS

Effect of the W88G mutation on the removal of AZTMP from blocked primer-templates.

The W88G mutation was selected for detailed studies because it confers a high level of foscarnet resistance in vivo (31, 54) and has been found in clinical isolates (31). Removal of AZTMP was assessed by measuring the formation of unblocked primer from a previously blocked primer-template (Fig. 1) (32). After incubation of the 5′-labeled primer-template with WT or mutant HIV-1 RT, the RT was heat inactivated and unblocked primer was extended by a second incubation with an exonuclease-free Klenow fragment of E. coli DNA polymerase I (Fig. 1A). Primer extension in the second incubation step depended on incubation with ATP and WT or mutant HIV-1 RT in the first incubation step. The results in Fig. 1A were used to determine the incubation times during which the reaction rates for ATP-dependent primer rescue were approximately linear for each mutant enzyme (Fig. 1B). Then rates of primer rescue were determined as a function of ATP concentration (Fig. 1C), and apparent catalytic efficiencies (k_cat/K_m,ATP) were calculated for each enzyme (see Fig. 3A). As previously reported (32, 35), RT containing the AZT resistance mutations had increased rates of the ATP-dependent removal of AZTMP—a fivefold increase for D67N/K70R/T215F/K219Q RT in comparison to the WT enzyme and a threefold increase for RT containing the M41L/T215Y mutations. Removal activity for RT containing the foscarnet resistance mutation W88G was decreased 5.3-fold relative to that of WT RT. When the W88G mutation was introduced into AZT-resistant backgrounds, M41L/T215Y and D67N/K70R/T215F/K219Q, the removal activity was decreased eight- and fourfold, respectively. PP_i-dependent removal of AZTMP was also decreased by the W88G mutation but was not increased by inclusion of the AZT resistance mutations (Fig. 1D), in agreement with previous results (32, 35). In summary, the decrease in ATP-dependent removal of AZTMP conferred by the W88G mutation suppressed the elevated AZTMP removal activity seen with M41L/T215Y or D67N/K70R/T215F/K219Q RT to levels near or below that of WT RT.

FIG. 1.

Effect of the W88G mutation on removal of AZTMP from blocked primer-template. (A) AZTMP-terminated [5′-³²P]L33 primer-WL50 template was incubated with the indicated WT or mutant RT in the absence (−) or presence (+) of 3.2 mM ATP for the indicated times at 37°C. The RT was inactivated by heat treatment, and the unblocked primer was extended by incubation with an exonuclease-free Klenow fragment of E. coli DNA polymerase I. The products were separated on a 20% denaturing polyacrylamide gel. The positions of unextended primer (primer) and of products formed after elongation to the end of the template (ext. primer) are shown to the left of the figure. (B) Radioactivity in products longer than 34 nucleotides (rescued primers) from experiments whose results are shown in panel A were quantitated by PhosphorImager analysis, expressed as a percentage of total radioactivity for each lane, and plotted against time. (C) Experiments were performed as described for panel A, except that the ATP concentration was varied from 0.2 to 6.4 mM and the time of incubation (2 to 90 min) was chosen for each RT to allow a maximum of 40% of the primer to be rescued. (D) Rescue experiments were performed as described for panel A, except that 50 μM PP_i was used instead of ATP. For panels B, C, and D, the symbols represent data points obtained in a typical experiment with the RTs indicated at the bottom of the figure, and the lines represent the best fit of the data to a hyperbola.

FIG. 3.

ATP-dependent removal of AZTMP, synthesis of Ap₄ddA, and foscarnet inhibition of dTMP incorporation by WT and mutant RTs. (A) Rates of removal of AZTMP from AZTMP-terminated [5′-³²P]L33 primer-WL50 template were determined as a function of ATP concentration as described in the legend to Fig. 1 for WT and mutant RTs (identified at the bottom of the figure). (B) Rates of [³²P]Ap₄ddA synthesis from [³²P]L32-ddAMP primer-WL50 template were determined as a function of ATP concentration as described in the legend to Fig. 2. For panels A and B, specificity constants (k_cat/K_m) were determined as described in Materials and Methods. (C) In vitro sensitivity to foscarnet inhibition of [³H]dTMP incorporation directed by a poly(rA)-oligo(dT) primer-template for WT and mutant RTs. Mean and standard deviation are shown for two or more experiments.

Effect of the W88G mutation on the synthesis of Ap₄ddA by ATP-dependent removal of ddAMP from blocked primer-templates.

We have previously shown (34) that ATP-dependent removal of ddAMP from a [³²P]ddAMP-terminated primer-template could be detected by the appearance of a new compound that corresponded to [³²P]Ap₄ddA. Figure 2A shows that the amount of labeled primer decreased and the amount of labeled Ap₄ddA increased with increasing incubation time. The amounts of radioactivity in primer and Ap₄ddA were quantitated using a PhosphorImager, and the percentage of total counts in Ap₄ddA was plotted against time (Fig. 2B). The rates of Ap₄ddA synthesis were determined as a function of ATP concentration (Fig. 2C), and catalytic efficiencies (k_cat/K_m,ATP) for Ap₄ddA synthesis were calculated for each enzyme (Fig. 3B). Ap₄ddA synthesis activities paralleled the activities for removal of AZTMP shown in Figure 3A. Increased Ap₄ddA synthesis by RT containing the AZT resistance mutations was reported previously (32, 35). The W88G mutation decreased Ap₄ddA synthesis by more than ninefold relative to WT RT, and addition of W88G to the AZT resistance background reduced Ap₄ddA synthesis to a similar extent. PP_i-dependent removal of ddAMP to form [³²P]ddATP was also greatly reduced by the W88G mutation, but there was no increase in [³²P]ddATP formation by the AZT-resistant RTs in comparison with that by WT RT (Fig. 2D). The rates of pyrophosphorolysis (estimated from the linear portion of the time course) were decreased by 90, 96, and 93%, for W88G, M41L/W88G/T215Y, and D67N/K70R/W88G/T215F/K219Q RT, respectively, in comparison with the enzymes lacking the W88G mutation. The corresponding ATP-dependent unblocking activities (determined from k_cat/K_m) were decreased by 89, 93, and 91%, respectively (Fig. 3B). By contrast, the forward reaction (as reflected in the specific activity for the purified enzymes as measured by [³H]dTMP incorporation) was reduced by only 5, 52 and 50%, respectively.

FIG. 2.

Effect of the W88G and AZT resistance mutations on removal of ddAMP from blocked primer-template through formation of Ap₄ddA or ddATP. (A) [³²P]ddAMP-terminated L32 primer-WL50 template (5 nM) was incubated with the indicated RT and 3.2 mM ATP for 2.5, 5, 10, or 20 min at 37°C. Products were separated on a 20% denaturing polyacrylamide gel. (B) The radioactivity from experiments whose results are shown in panel A was quantitated by PhosphorImager analysis, and the percentage of radioactivity recovered as Ap₄ddA was plotted against time. (C) Reactions were carried out as described for panel A but with 0.2 to 6.4 mM ATP and 1 to 90 min of incubation, depending on the enzyme, to allow a maximum of ∼35% Ap₄ddA formation. The rate of Ap₄ddA formation was calculated and plotted against ATP concentration. (D) Experiments were performed as described for panel A but with 5 μM PP_i instead of ATP. The radioactivity was quantitated by PhosphorImager analysis, and the percentage recovered as ddATP was plotted against time. For panels B, C, and D, the symbols represent data points obtained with the RTs indicated at the bottom of the figure and the lines are the best fit to a hyperbola.

As summarized in Fig. 3, the W88G mutation conferred decreased ATP-dependent unblocking activity on both ddAMP- and AZTMP-terminated primer-templates, and in both cases the decrease was partially or completely reversed by the presence of M41L/T215Y or D67N/K70R/T215F/K219Q AZT resistance mutations. Figure 3C shows that the presence W88G mutation resulted in a 19- to 28-fold increase in IC₅₀ for foscarnet inhibition of dTMP incorporation on a poly(rA)-oligo(dT) primer-template. The sensitivity to foscarnet inhibition was decreased to a similar extent by the W88G mutation in the presence or absence of the AZT resistance mutations.

Ap₄ddA synthesis by RTs containing other foscarnet resistance mutations.

The formation of labeled Ap₄ddA is shown for WT and mutant RTs in Table 1. Similar results were observed for the removal of the block to elongation from AZTMP-terminated primer-templates by the primer rescue assay (data not shown). Primer-unblocking activity was substantially reduced for K65R, W88G, Q161L, and Q161L/H208Y mutants in comparison with WT RT. The activities were not significantly reduced for W88S, S117T, and M164I RT. The effects of the foscarnet resistance mutations were similar in the presence and absence of the AZT resistance mutations at codons 67, 70, 215 and 219; i.e., the foscarnet resistance mutations and the AZT resistance mutations had opposite effects on the primer-unblocking reaction.

TABLE 1.

Primer-unblocking activity of WT and mutant HIV-1 RT^a

Resistance mutation(s)b	Ap4ddA formation (kcat/Km [M−1 s−1])c
	None (WT)	K65R	W88G	W88S	E89K	S117T	Q161L	M164I	Q161L/H208Y
None (WT)	0.23 ± 0.05 (1.0)	0.085 ± 0.003 (0.37)	0.025 ± 0.011 (0.11)	0.47 ± 0.03 (2.0)	NDd	0.27 ± 0.02 (1.2)	0.045 ± 0.001 (0.20)	0.17 ± 0.08 (0.74)	0.027 ± 0.001 (0.12)
67/70/215/219	3.3 ± 1.2 (14)	0.33 ± 0.03 (1.4)	0.19 ± 0.02e (0.83)	3.8 ± 1.0 (17)	0.26 ± 0.04 (1.1)	1.0 ± 0.1 (4.3)	0.37 ± 0.01 (1.6)	1.68 ± 0.03 (7.3)	0.40 ± 0.03 (1.7)

^aPrimer-unblocking activity is measured by ATP-dependent formation of [³²P]Ap₄ddA from [³²P]ddAMP-labeled primer-template as described in the text.

^bRTs expressed with the amino acid substitutions indicated at the top of each column in the absence or presence of D67N/K70R/T215Y/K219Q substitutions (except as indicated [see footnote e]).

^cValues are the mean and standard deviation of two or more determinations. Numbers in parentheses show the fold change relative to WT RT.

^dND, not determined.

^eD67N/K70R/W88G/T215F/K219Q RT.

Foscarnet sensitivity of mutant HIV and RTs containing foscarnet resistance mutations.

Virus containing foscarnet mutations alone or in combination with the AZT resistance mutations D67N/K70R/T215Y or F/K219Q were tested for foscarnet sensitivity in MT-2 cells (Table 2). Purified mutant or WT RT was also tested for foscarnet sensitivity in a dTMP incorporation assay (Fig. 3C) (see Materials and Methods). The results are also shown in Table 2. In vivo resistance to foscarnet ranged from 2.7- to 8.7-fold for these mutants in comparison to that of WT virus and was suppressed by the presence of the AZT resistance mutations for some foscarnet mutants (e.g., W88G, W88S, Q161L, M164I, and Q161L/H208Y) but not for others (e.g., K65R and E89K). IC₅₀s were 12- to 166-fold lower when sensitivity to foscarnet was tested in vitro. In comparison with WT RT, the relative resistance was similar in vivo and in vitro for mutants that were moderately or minimally resistant to foscarnet (e.g., K65R, W88S, S1117T, and M164I). By contrast, the mutants with the greatest foscarnet resistance (W88G, Q161L, and Q161L/H208Y) were two to three times more resistant to foscarnet by the in vitro assay than by the MT-2 culture assay. In contrast to the results in the virus replication assays, little or no suppression of foscarnet resistance by the AZT resistance mutations was evident in the in vitro assays.

TABLE 2.

Foscarnet inhibition of virus replication and in vitro DNA synthesis by WT and mutants of HIV-1

Resistance mutation(s)a	IC50 (μM) for foscarnetb
	None (WT)	K65R	W88G	W88S	E89K	S117T	Q161L	M164I	Q161L/H208Y
In vivoc
None (WT)	14.5 ± 6.6 (1.0)	45 ± 23 (3.1)	126 ± 48 (8.7)	39 ± 21 (2.7)	66 ± 32 (4.5)	50 ± 33 (3.4)	98 ± 25 (6.7)	50 ± 11 (3.4)	98 ± 40 (6.7)
67/70/215/219	10.2 ± 6.0 (0.70)	38 ± 22 (2.6)	59 ± 28 (4.0)	22 ± 6 (1.5)	134 ± 88 (9.3)	34 ± 9 (2.4)	34 ± 12 (2.3)	21 ± 5 (1.5)	37 ± 14 (2.6)
In vitrod
None (WT)	0.21 ± 0.01 (1.0)	0.96 ± 0.26 (4.6)	4.0 ± 0.7 (19)	0.30 ± 0.04 (1.4)	NDe	0.38 ± 0.03 (1.8)	3.1 ± 1.9 (15)	0.30 ± 0.01 (1.4)	3.7 ± 2.3 (18)
67/70/215/219	0.16 ± 0.08 (0.76)	0.41 ± 0.09 (2.0)	4.5 ± 1.2f (21)	0.33 ± 0.12 (1.6)	2.1 ± 0.6 (10)	0.87 ± 0.60 (4.1)	2.9 ± 1.0 (14)	0.28 ± 0.11 (1.3)	2.1 ± 0.5 (10)

^aMutations present in each virus or purified RT include foscarnet resistance mutations listed at the top of the column in the absence or presence of D67N/K70R/T215Y/K219Q mutations listed for each row (except as indicated [see footnote f].)

^bResults are given as mean and standard deviation. Numbers in parentheses show the fold changes relative to WT virus or WT RT.

^cSusceptibility to foscarnet inhibition of virus replication measured in MT-2 cells after 7 days as described in the text (mean of three or more determinations).

^dSusceptibility to foscarnet inhibition of [³H]dTMP incorporation into poly(rA)-oligo (dT) (mean of two or more determinations).

^eND, not determined.

^fD67N/K70R/W88G/T215F/K219Q mutations.

Relationship of primer-unblocking activity to foscarnet resistance and AZT resistance.

The primer-unblocking activity for WT and mutant RTs and the IC₅₀ for foscarnet inhibition of replication of the viruses containing these mutations showed a strong inverse correlation (Fig. 4A). By contrast, the IC₅₀ for AZT inhibition of recombinant viruses (Table 3) was directly correlated with the primer-unblocking activity for the corresponding WT or mutant RTs (Fig. 4B). Tachedjian et al. (54) showed that the foscarnet and AZT susceptibility of recombinant strains of HIV-1 are inversely correlated, and our results provide a biochemical rationale for this observation. Shao et al. (47) have shown a correlation between AZT resistance and AZTTP inhibition of DNA chain elongation by HIV-1 RT in the presence of 0.5 mM GTP.

FIG. 4.

Relationship between primer-unblocking activity for 16 enzymes and susceptibility to foscarnet inhibition (A) or AZT inhibition (B) of replication of the corresponding viruses. Data are from Tables 1 to 3. Correlation coefficients for the log transformed data are −0.77 for panel A and +0.84 for panel B. A similar test of the relationship between primer-unblocking activity and sensitivity to foscarnet of in vitro [³H]dTMP incorporation gave a correlation coefficient of −0.67.

TABLE 3.

AZT inhibition of WT and mutant HIV-1 replication

Resistance mutation(s)a	IC50 (μM) for AZTb
	None (WT)	K65R	W88G	W88S	E89K	S117T	Q161L	M164I	Q161L/H208Y
None (WT)	0.024 ± 0.018 (1.0)	0.026 ± 0.007 (1.1)	0.016 ± 0.010 (0.7)	0.027 ± 0.020 (1.1)	0.007 ± 0.005 (0.3)	0.019 ± 0.017 (0.8)	0.016 ± 0.008 (0.7)	0.016 ± 0.008 (0.7)	0.008 ± 0.003 (0.3)
67/70/215/219	0.36 ± 0.28 (15)	0.067 ± 0.041 (2.8)	0.027 ± 0.009 (1.1)	0.17 ± 0.13 (7.1)	0.027 ± 0.009 (1.1)	0.057 ± 0.029 (2.4)	0.064 ± 0.029 (2.7)	0.044 ± 0.029 (1.8)	0.019 ± 0.006 (0.8)

^aMutant virus contained the foscarnet resistance mutations listed at the top of each column in the absence or presence of the D67N/K70R/T215Y/K219Q AZT resistance mutation.

^bResults are given as mean and standard deviation. Concentration of AZT required to inhibit WT or mutant HIV-1 replication by 50% (IC₅₀), measured in MT-2 cells after 7 days as described in the text. Numbers in parentheses are fold changes relative to WT RT.

Formation of foscarnet-RT-primer-template complexes.

Upon binding the dNTP complementary to the next position on the template, RT adopts a “closed” configuration in which the “fingers” domain moves closer to the dNTP binding site (16). The “closed” complex can also be detected by electrophoretic mobility shift assays (11, 18, 32, 56) carried out in the presence of excess unlabeled homopolymer primer-templates or heparin to disrupt the binary complex. Figure 5 shows a similar complex formed by RT in the presence of foscarnet and chain-terminated primer-template. High concentrations of the next complementary dNTP usually resulted in more than 80% of the labeled primer-template migrating as stable complexes (32, 56); however, at most, 20% of the labeled primer-template was present in heparin-stable complexes at the highest concentration of foscarnet tested, suggesting that the recovery of foscarnet-RT-primer-template complexes may be suboptimal in these assays. Similar results were obtained with ddAMP-terminated (Fig. 5A) and AZTMP-terminated (Fig. 5B) primer-template.

FIG. 5.

Ability of foscarnet to induce the formation of a heparin-stable complex between HIV-1 RT and chain-terminated primer-templates. (A) Electrophoretic mobility shift assays were carried out as described in Materials and Methods with [³²P]ddAMP-terminated L32 primer-WL50 template (5 nM) without or with 200 nM WT HIV-1 RT and foscarnet (PFA) at the concentrations indicated. (B) Assays were carried out as in panel with 5 nM 5′-³²P-labeled AZTMP-terminated L33 primer-WL50 template. Positions of the nucleoprotein complex (Complex) and free DNA (P/T) are shown to the right of the figure.

DISCUSSION

Resistance of HIV-1 to foscarnet is achieved in vivo and in vitro by mutations that reduce the ability of RT to catalyze the primer-unblocking reaction. Diminished unblocking activity has previously been reported for foscarnet-resistant enzymes including A114S and K65R RT (2, 48). On the other hand, resistance to AZT is achieved by mutations that increase the primer-unblocking activity, allowing the mutant enzymes to remove AZTMP as well as other chain-terminating nucleotides that block DNA chain elongation (1, 3, 28, 30, 32, 33, 35, 36, 46). RTs containing both foscarnet and AZT resistance mutations had levels of primer-unblocking activity resulting from the combination of these two opposing effects. This explains why foscarnet-resistant mutants of HIV-1 are often hypersensitive to AZT (31, 52) and why AZT resistance mutations are suppressed by foscarnet resistance mutations and visa versa (2, 13, 54).

In vitro resistance to foscarnet by WT and mutant RTs, as measured by effects on [³H]dTMP incorporation, did not match the in vivo resistance measured in cell culture assays (Table 2). The IC₅₀s observed in vitro were 12- to 166-fold lower than those obtained in the culture assays, in agreement with previous reports (38, 43). This discrepancy is probably due to the limited ability of foscarnet to penetrate into cells, so that high concentrations of extracellular foscarnet are needed to reach inhibitory levels intracellularly. The permeability barrier can be circumvented by the use of alkylglycerol-conjugated derivatives of foscarnet, which are 10- to 35-fold more potent inhibitors of HIV-1 replication than is free foscarnet because they readily penetrate into cells and are then cleaved to form free intracellular foscarnet (13-15, 21). These findings probably explain the differences in foscarnet sensitivity between our cell-based and enzyme assays.

Crystal structures have been reported for the binary complexes of HIV-1 RT and primer-template (9, 19, 41) and for a ternary complex containing HIV-1 RT, primer-template, and the incoming dNTP (16). Figure 6 shows the positions of the residues affected by foscarnet resistance mutations in the p66 subunit of the ternary complex (Fig. 6B) and the binary complex (Fig. 6A) in which AZTMP-terminated DNA primer-template was captured prior to translocation (termed the N-complex because the terminal nucleotide on the primer occupies the dNTP binding site) (41). In the ternary complex, the enzyme is in a “closed” configuration and residue K65 of the p66 subunit interacts directly with the γ phosphate of the incoming dNTP. In the binary complex, the enzyme is in an “open” configuration and K65 is at the tip of the fingers domain far from the dNTP binding site (9, 19, 41). The residues affected by the other foscarnet resistance mutations in this study occupy the same positions in both structures and do not make direct contact with the catalytic residues of RT or the incoming dNTP. Of course, the amino acid substitutions that occur at these positions may have additional structural consequences; however, information is not currently available to evaluate this possibility.

FIG. 6.

Locations of residues K65, W88, E89, S117, Q161, M164, and H208 in the structures of binary and ternary complexes of HIV-1 RT. (A) Binary N-complex of HIV-1 RT with AZTMP-terminated primer-template and a monoclonal antibody Fab fragment (41). (B) Ternary complex of HIV-1 RT with dTTP and ddAMP-terminated primer-template (16). The template (T) is shown in blue, and the primer strand (P) is shown in green. The nucleotide occupying the N site in each complex is shown as a space-filling model with the atoms indicated by the Corey-Pauling-Koltun color scheme (AZTMP in panel A and dTTP in panel B). The seven residues that are substituted in the foscarnet resistance mutants in this study are shown as space-filling models. The figure was prepared with Protein Explorer (http://molvis.sdsc.edu/protexpl/frntdoor.htm) using PDB structure coordinates 1N6Q (A) and 1RTD (B) retrieved from the Brookhaven Protein Data Base.

To carry out the primer-unblocking reaction, the primer terminus must be in the untranslocated position relative to the polymerase active-site residues. So far, no crystal structure has been reported for RT that contains bound PP_i, ATP, or foscarnet; however, since PP_i is derived from the β and γ phosphates of the incoming dNTP, it is possible to infer which residues are likely to interact with PP_i from the crystal structure of the RT-dTTP-primer-template complex (3, 6, 16, 32, 46). If this inference is correct, then RT must adopt a closed configuration to allow residue 65 to interact with PP_i and, by extension, with ATP or foscarnet (compare Fig. 6A and B). The fact that the K65R mutation confers foscarnet resistance and decreased primer-unblocking activity suggests that a closed conformation of RT plays a role in the primer-unblocking reaction and in foscarnet sensitivity and that the substitution of arginine for lysine at this position could reposition a positive charge which would affect interactions with PP_i, ATP, and foscarnet. A closed conformation of the foscarnet-RT-primer-template complex can also be inferred from Fig. 5, which shows foscarnet-dependent formation of a stable complex that can be detected by an electrophoretic mobility shift assay which is similar to the assay used to detect a closed complex formed with incoming dNTP (11, 18, 32, 56); however, it is not clear why a higher concentration of foscarnet is needed to form the complex compared with the concentration needed to inhibit polymerase activity.

PP_i-dependent unblocking activity was decreased to roughly the same extent as the ATP-dependent activity in several foscarnet-resistant RTs (Fig. 1D and 2D and data not shown); however, the PP_i-dependent activity did not parallel the increase in the ATP-dependent reaction resulting from the AZT resistance mutations. This suggests that the PP_i- and ATP-dependent pathways include a common step(s) (identified by the foscarnet resistance mutations) and that there is at least one additional step that distinguishes between these two substrates (identified by the AZT resistance mutations). In addition, the foscarnet resistance mutations have little effect on dNTP incorporation and preferentially reduce the reverse pathway.

Several possible explanations can be considered for the strong association observed between foscarnet resistance and decreased primer unblocking: (i) foscarnet resistance mutations may reduce binding and retention of foscarnet, PP_i, and ATP in the RT-primer-template complex; (ii) the mutations may affect the position or orientation of PP_i and ATP, resulting in a less favorable interaction with the α-phosphate linkage between the last two nucleotides of the terminated primer; or (iii) the mutations may affect the position of the primer-template in the enzyme complex so that interaction with foscarnet or the acceptor substrates for excision is impaired. Experimental approaches are being developed to differentiate among these possibilities.

We have shown that binding of the next complementary dNTP and primer unblocking are mutually exclusive activities of RT and that, as a result, the primer-unblocking reaction is very sensitive to inhibition by the next complementary dNTP (32, 33). This occurs because the dNTP can bind only after RT has translocated to the next position on the template and removed the primer terminus from the dNTP binding site, whereas the unblocking reaction can occur only when the enzyme is in the position prior to translocation. Boyer et al. (3) have proposed the terms “N site” and “P site” for the positions of the primer terminus before and after translocation, respectively. The fact that foscarnet inhibition of DNA synthesis is noncompetitive with the nucleotide substrate (58) implies that foscarnet binds to a different form of RT from that which binds to dNTP and is consistent with foscarnet binding to the enzyme when the primer terminus is in the N site. The studies that show that in vitro inhibition of RT by foscarnet and AZTTP cannot occur simultaneously (10, 22, 50) would also be explained if foscarnet binds only when the primer terminus is in the N site and AZTTP can bind only when the primer terminus is in the P site.

Mutations (including K65, K72, D113, A114, Y115, and Q151) at residues in RT that interact with the incoming dNTP confer foscarnet resistance, presumably through changes in the PP_i binding site, which involves the same residues that interact with the β and γ phosphates of the dNTP. These mutations may also influence the position of the terminated primer in the complex through interaction with the terminal nucleotide when the primer is in the N site. The other foscarnet resistance mutations in this study affect residues whose positions are the same in the binary and ternary complexes (Fig. 6) and are far from the dNTP binding site. Except for residue 89, the mutations occur at sites that are unlikely to interact directly with the template, the primer, or any component of the polymerase active site unless they are accompanied by major structural rearrangements of the enzyme.

The E89 residue appears to interact with the template strand (the distance to the phosphate backbone is ∼3.3 Å) (9, 16, 19, 41), and a mutation at this site could influence the positioning of the primer terminus in either the N or the P site. The W88 residue is >7 Å from the nearest atoms in either the primer or the template; nonetheless, the W88G mutation is the most frequently observed mutation in clinical isolates from foscarnet-treated patients (31) and conferred the highest level of foscarnet resistance and the greatest defect in primer-unblocking activity in the present studies. The mechanism of the effects of W88G is obscure but may be mediated through effects on the adjacent E89 residue. The Q161 residue lies in the “palm” domain (>9 Å from the active-site aspartate residues and >6 Å from the closest atoms in either the primer or the template); however, altered dNTP binding and divalent cation specificity have been reported for Q161L RT (57), suggesting that the effects of this mutation are mediated through conformational rearrangements in the enzyme. The S117T, M164I, and H208Y mutations also lie in the palm domain between the fingers domain and the template strand. These mutations may identify a region of RT that participates in conformational changes related to translocation of the primer terminus and/or binding of foscarnet, PP_i, and ATP. Further elucidation of the biochemical mechanisms of resistance by these mutations may shed light on pathways of conformational communication within RT.

Therapies combining foscarnet or foscarnet derivatives with nucleoside antiretroviral drugs have been proposed (13) because of the ability of foscarnet to potentiate the antiviral activity of AZT and to suppress the AZT resistance phenotype or prevent the appearance of AZT resistance mutations. Dual resistance to foscarnet and AZT is selected only with difficulty (51, 53, 54) and appears to require multiple mutations. Mutations that enhance the primer-unblocking activity confer increased resistance to other nucleoside RT inhibitors (8, 33, 46), suggesting that this activity plays a role in determining the potency of a variety of chain-terminating drugs and that understanding of the molecular basis of resistance to agents that inhibit this activity will play an important role in future chemotherapeutic strategies against HIV.