Correlation between Viral Resistance to Zidovudine and Resistance at the Reverse Transcriptase Level for a Panel of Human Immunodeficiency Virus Type 1 Mutants

Abstract

Using a large panel of human immunodeficiency virus type 1 site-directed mutants, we have observed a higher correlation than has previously been demonstrated between zidovudine (AZT)-triphosphate resistance data at the reverse transcriptase (RT) level and corresponding viral AZT resistance. This enhanced-resistance effect at the RT level was seen with ATP and to a lesser extent with PP_i when ATP was added at physiological concentrations. The ATP-dependent mechanism (analogous to pyrophosphorolysis) appears to be dominant in the mutants bearing the D67N and K70R or 69 insertion mutations, whereas the Q151M mutation seems independent of ATP for decreased binding to AZT-triphosphate.

Resistance to zidovudine (AZT) commonly develops during human immunodeficiency virus type 1 (HIV-1) therapy, because there is rapid virus turnover and the HIV-1 reverse transcriptase (RT) is an error-prone polymerase (3, 13, 15). The selection of resistant virus is seen for all HIV-1 inhibitors currently in clinical use (25), and in addition, multidrug resistance has been documented during combination therapy (14, 21). Analysis of AZT-resistant HIV-1 strains from patients has resulted in the identification of several mutations in the RT coding sequence (6, 9, 15). For example, changes at codons 41, 67, 70, 210, 215, and 219 result in specific resistance to AZT (6, 9, 15). These mutations can cause the virus to become highly resistant to AZT (in some combinations, they produce a >100-fold increase in the 50% inhibitory concentration [IC₅₀]). However, the biochemical mechanism of resistance conferred by these mutations has remained a puzzle for a long period. This is because only minor or no differences were seen by comparing the kinetic properties of mutant enzymes with those of the wild-type enzyme in the presence of different AZT-triphosphate (AZT-TP) concentrations (4, 10, 11, 17). However, a different set of RT mutations, i.e., A62V, V75I, F77L, F116Y, and Q151M, which confer resistance to multiple nucleoside analogs (where Q151M is the key mutation), did show decreased binding of AZT-TP that correlated with virus data (24).

Recently, novel biochemical mechanisms for AZT resistance have been proposed. Arion et al. (1) showed that removal of AZT-monophosphate (AZT-MP) from blocked primers was enhanced for an AZT-resistant mutant compared to what occurred with wild-type RT and that this was due to pyrophosphate (PP_i)-dependent pyrophosphorolysis, the reverse reaction of DNA synthesis. Likewise, Meyer et al. (19) have demonstrated a ribonucleotide ATP- or GTP-dependent primer-unblocking reaction that produces dinucleoside polyphosphate. Both of these studies were performed with HIV-1 RT having mutations at codons D67N, K70R, T215F, and K219Q. An increase in AZT-TP resistance of up to fivefold was demonstrated with this mutant using physiological concentrations of either PP_i (1) or ATP (19). Thus, it is likely that a nucleoside triphosphate (NTP), PP_i, or both are involved in the AZT resistance mechanism.

In the present study, an RT assay was established to enable discrimination of AZT-TP resistance when ATP, GTP, or PP_i was added in physiological concentrations. The aim was also to identify the particular mutations involved, by comparing the level of binding of AZT-TP to those of a variety of mutant RTs containing combinations of different mutations. The choice of mutations was based on those seen in patient samples treated with AZT alone or together with other nucleoside analogs (14). Further, mutations were chosen to study the potential involvement of the residues in the RT finger domain as well as those causing resistance to multiple nucleoside drugs.

We performed site-directed mutagenesis with single- or multiple-nucleotide changes being introduced into the RT coding region of a wild-type HXB2-D EcoRI-NdeI restriction enzyme fragment cloned into the expression vector pKK233-2 (Pharmacia Biotech) (16). Mutations were generated using commercial site-directed mutagenesis kits: ExSite and QuikChange (Stratagene). The mutated cloned expression vectors were transformed into Escherichia coli strain XL1-Blue, and the genotypes were verified by DNA sequence analysis (12). Using the HIV-1 drug susceptibility assay, the phenotypic susceptibilities of these viable recombinant viruses to AZT was determined as described previously (7). Briefly, the proviral molecular clone pHIV?RTBstEII (8) was cotransfected with the RT-PCR product into MT-4 cells (5, 7). The fold level of resistance was calculated by dividing the mean IC₅₀ for a recombinant virus from a mutant clone by the mean IC₅₀ for the recombinant wild-type virus (HXB2). Expression and purification of RT have been described previously (22). Briefly, RT was expressed in E. coli XL1-Blue grown with 50 mg of kanamycin per liter and ampicillin in 2× TY medium (1.6% Bacto Tryptone, 1% yeast extract, 0.5% NaCl) for 18 h at 36°C and in presence of 100 μg of isopropyl-β-d-thiogalactopyranoside (IPTG) per liter for 6 h. After sonication of extracts containing the enzyme, purification was performed by ion-exchange chromatography in several steps using both 50 HQ and 50 HS columns (PerSeptive Biosystems). The final RT products, in a heterodimeric form (p66-p51), had a purity of >90% by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (data not shown). The purified RTs were assayed for inhibition of AZT-TP using separate kit components from the Lenti RT assay (Cavidi Tech, Uppsala, Sweden). The principle, performance, and composition of this nonradioactive microtiter plate RT assay have been described previously (4, 20). The assay components used in this study for the RT activity step were (i) RT reaction buffer [HEPES, 10 mM, pH 7.6; oligo(dT)₂₂, 40 pM; MgCl₂, 10 mM; dextran sulfate, 0.05 g/liter; spermine, 2 mM; Triton X-100, 0.5% (vol/vol); EGTA, 0.2 mM; bovine serum albumin, 0.5 mg/ml], (ii) the template [i.e., 96-well microtiter plates were filled with solid-phase-conjugated poly(A) at 40 ng/well], and (iii) the substrate deoxy-NTP, namely, separately lyophilized bromodeoxyuridine TP (BrdUTP). Extra chemicals for novel use in this assay system were ATP and GTP (Pharmacia Biotech) and sodium PP_i (Sigma). The RT reaction was started by addition of purified RT (mutant or wild-type enzyme) at 20 to 60 pM (1 to 4 mU/well). The RT reaction proceeded for 3 h at 33°C and was terminated by washing; in the next step, alkaline phosphatase-conjugated anti-BrdU antibody was incubated for 90 min at 33°C and the reaction was stopped with a new wash. The RT activity was measured by adding alkaline phosphatase substrate and reading the color change at 405 nm. The K_m values were determined using BrdUTP at six different concentrations from 30 nM to 16 μM. Within this substrate range, the enzyme reaction was linear for the RT assay step (3 h) and, thereby, steady-state kinetics were assumed. The K_i values were determined for AZT-TP (Moravek, Brea, Calif.) using three different concentrations (ranging from 1.5 to 100 nM) of the nucleotide, adjusted optimally for the expected K_i value of each mutant. Competitive inhibition can be assumed since the V_max values did not change with these different AZT-TP concentrations (data not shown). This finding makes it relevant to use the mutant K_i/K_m ratio relative to the wild-type ratio as a measurement of fold resistance. The K_m and K_i values were obtained by fitting the data to the Michaelis-Menten competitive inhibition equation using the Grafit 4.10 program (Erithacus Software).

First we studied the kinetics of AZT-TP inhibition with wild-type RT toward one mutant RT (Table 1). The degree of resistance shown by mutant 41/67/70/215 (bearing the mutations M41L, D67N, K70R, and T215Y) is expressed as an increase in the K_i/K_m value compared to the corresponding wild-type ratio. In the absence of added ATP or PP_i, mutant 41/67/70/215's K_i/K_m ratio was 0.6 when divided by the ratio derived for wild-type RT (Table 1). This apparent lack of biochemical resistance is consistent with the results of a previous study of AZT-resistant mutants using conventional RT assays (11). However, the inclusion of physiological concentrations of ATP (3.2 mM) dramatically increased this resistance 10.3-fold. This degree of resistance was within a factor of 2 of the 24-fold resistance seen with the HIV drug susceptibility assay (Table 2). We also tested the effect of PP_i (150 μM) on AZT-TP inhibition. The level of resistance of mutant 41/67/70/215 relative to that of the wild type was 4.4-fold (Table 1). A control experiment which ensured that the previous effect seen by ATP was not influenced by PP_i contamination was performed. This was done by pretreating the ATP reagent with thermostable pyrophosphatase (Boehringer Mannheim) (data not shown). The ATP effect was, however, dependent on the concentration used (see results with 1.5 and 6 mM ATP in Table 1). The highest resistance level of mutant 41/67/70/215 was found with 3.2 mM ATP. When ATP (3.2 mM) combined with PP_i (150 μM) was studied, both the wild type and mutant 41/67/70/215 showed an increase in K_i and K_m values, but the level of resistance of mutant 41/67/70/215 (6.0-fold) was lower than the level with 3.2 mM ATP alone (10.3-fold). However, ATP per se was slightly inhibiting, as seen by decreased relative V_max values in Table 1. Also, in a separate experiment with this assay system, ATP seemed to behave as a noncompetitive inhibitor (data not shown).

TABLE 1.

Comparison of the AZT susceptibilities of a strain with wild-type RT and the RT mutant M41L/D67N/K70R/T215Y in AZT-TP inhibition assay systems^c

Extra additions (concn) in AZT-TP–RT assay	HXB2				Fold resistance of mutant 41/67/70/21				Fold resistance of mutant 41/67/70/215 in RT assay relative to the level of resistance of HXB2a
	Ki (nM)	Km (nM)	Ki/Km	Relative Vmax (%)b	Ki (nM)	Km (nM)	Ki/Km	Relative Vmax (%)
None	0.8	150	0.0053	100	0.6	170	0.0035	100	0.7 (0.1)
ATP (1.5 mM)	1.2	160	0.0075	89	7.3	220	0.033	98	4.4 (1.3)
ATP (3.2 mM)	1.4	150	0.0093	78	33.4	350	0.095	94	19.3 (1.2)
ATP (6.0 mM)	4.2	280	0.015	58	59.2	390	0.152	86	10.1 (3.0)
GTP (0.5 mM)	1.2	170	0.0071	91	3.0	230	0.013	97	1.8 (0.4)
PPi (150 μM)	4.2	170	0.025	93	24.2	220	0.11	96	4.4 (2.0)
ATP (3.2 mM) + PPi (150 μM)	9.3	310	0.03	55	81	450	0.18	80	6.0 (2.8)

^aFold resistance is the ratio of the K_i/K_m values for the mutant divided by the equivalent ratios for the strain with wild-type RT. Standard errors are indicated in parentheses. 

^bV_max values were calculated as percentages of the V_max of either HXB2 or the mutant without extra additions in the assay system. 

^cAll values are means calculated from results of at least two separate experiments. 

TABLE 2.

AZT resistance of RT mutants and the wild type measured by the AZT-TP–RT inhibition assay with or without 3.2 mM ATP and by the HIV-1 drug susceptibility assay^a

Strain	AZT-TP–RT assay without ATP				AZT-TP–RT assay with 3.2 mM ATP				Virus assay fold resistance
	Ki (nM)	Km (nM)	Ki/Km	Fold resistance	Ki (nM)	Km (nM)	Ki/Km	Fold resistance with ATP
Wild type (HXB2)	0.8	150	0.0053	1.0 (0.1)	1.4	150	0.0093	1.0 (0.1)	1.0
41L/215Y	1.4	230	0.0061	1.1 (0.1)	7.4	310	0.024	2.6 (0.2)	26.2 (6.0)
41L/184V/215Y	1.7	200	0.0085	1.6 (0.2)	8.7	190	0.046	4.9 (1.0)	5.2 (1.2)
41L/67N/70R/215Y	0.6	170	0.0035	0.7 (0.1)	33.4	350	0.095	10.3 (1.2)	24.1 (7.7)
41L/67N/70R/184V/215Y	0.8	160	0.005	0.9 (0.5)	12.0	300	0.04	4.3 (0.5)	6.4 (1.5)
41L/67N/70R/184V/210W/211K/214F/215Y	0.8	220	0.0036	0.7 (0.1)	19.4	380	0.051	5.5 (1.9)	12.0 (2.6)
69S-SS	0.9	120	0.0075	1.4 (0.3)	5.2	180	0.029	3.1 (0.1)	2.0 (1.0)
41L/69S-SS/215Y	1.0	180	0.0055	1.0 (0.3)	40.9	290	0.141	15.2 (1.8)	12.0 (1.2)
69S-SG	1.9	250	0.0076	1.4 (0.4)	25.4	410	0.062	6.7 (1.7)	8.1 (3.1)
69S-AG	0.7	100	0.007	1.3 (0.2)	11.8	120	0.098	10.5 (2.1)	7.0 (2.0)
41L/69S-AG/210W/211K/214F/215Y	0.6	130	0.0046	0.9 (0.0)	33.8	230	0.147	15.8 (4.0)	>36
151M	2.6	120	0.021	4.0 (0.2)	6.7	190	0.035	3.8 (0.4)	2.1 (0.4)

^aThe AZT-TP–RT assay fold resistance values are the ratios of the K_i/K_m values for the mutants to the K_i/K_m values for the wild type. The virus AZT assay fold resistance values are based on IC₅₀s relative to that of the wild-type HXB2 virus control. The mean IC₅₀ of AZT for the wild-type control virus was 0.012 μM. All values are the averages calculated from results of at least two separate experiments. Standard errors are indicated in parentheses. 

As ATP appeared predominant for the AZT-TP resistance mechanism, the system of AZT-TP inhibition of RT was next used to investigate which mutation sites were involved and the results were compared with results for the same mutations in an HIV-1 drug susceptibility assay. Ten site-directed mutant RTs together with mutant 41/67/70/215 and wild-type RT were studied with and without the addition of ATP. These data are illustrated in Table 2. In the absence of ATP, all the mutants except the Q151M mutant (mutant 151) behaved in a way similar to that of the strain with wild-type RT (0.7- to 1.6-fold resistance), and no correlation with the virus inhibition data was seen. By contrast, mutant 151 showed similar fourfold levels of resistance both with and without added ATP. For the remaining mutants, a reasonable correlation was observed between results of the virus assay and results of the AZT-TP inhibition assay when 3.2 mM ATP was included. However, there was one apparent exception. Only 2.6-fold resistance was seen for mutant 41/215 compared to its level of resistance in the drug susceptibility assay (26.2-fold). The levels of resistance of mutants with T69S-XX insertions, known to be associated with multinucleoside resistance (14, 26), all matched well their levels of resistance in the virus assay, and in some cases the biochemical resistance data were even higher than the corresponding viral resistance data. The behavior of the mutants having the M184V mutation indicated in certain cases that there was a reduction of resistance at the enzyme level that matched the virus data. When intracellular concentrations of GTP (0.5 mM) and PP_i (150 μM) were studied for two other mutants, 41/215 and 69S-AG, similar or lower-fold resistance data were found (data not shown) as seen in Table 1 for mutant 41/67/70/215 using GTP or PP_i. These data further confirmed that ATP has the dominant effect when it is used in this assay system.

Our aim was to set up a simple AZT-TP–RT inhibition assay that better reflected the AZT sensitivities of mutants and therefore reproduced the cell culture data more accurately. As ATP- and PP_i-dependent excision of the incorporated AZT-MP has been shown by other investigators (1, 19), an alternative explanation for how ATP or PP_i stimulates excision for specific mutants in our assay system seems less likely. Thus, we have demonstrated using our assay system a PP_i-dependent pyrophosphorolysis and, even more, an ATP-dependent mechanism for AZT-TP resistance in a wide range of RT mutants. The reason that our assay seems to yield higher magnitudes of resistance may be that it enables the measurement of multiple chain termination events. This should reflect the in vivo situation better and may explain why Meyer et al. (19) did not find a similar magnitude of resistance with 3.2 mM ATP and no effect with PP_i (although only the effects of a 50 μM concentration were examined) using their mono-chain-terminating system. However, the real influence of PP_i (together with that of ATP) in vivo obviously depends on intracellular concentrations. In this regard, the intracellular concentrations of NTPs are well established at 3.2 ± 1.5 mM for ATP and 0.5 ± 0.2 mM for GTP (23). The intracellular concentration of PP_i has been reported to be 130 μM (2).

It is of interest to determine which mutations are involved in the ATP-mediated effect. The most important mutations appear to be located in the finger domain of RT, specifically in the β2-β3 loop, as alterations at positions 67, 69, and 70 were dominant. With respect to mutants with insertions at position 69, this is the first demonstration of resistance at the enzyme level for mutants bearing only the 69S-SG or 69S-AG changes. In contrast, the mutant with the multi-nucleoside drug-resistant mutation Q151M alone appeared to be involved in decreased binding to AZT-TP independently of ATP, which is in line with a previous report (24), although only modest reduced AZT sensitivity was seen with this key mutation by itself. However, there was an incomplete correlation between results of the in vitro and cell culture assays with mutants lacking either mutations at positions 67 and 70 or insertions at position 69. Thus, AZT resistance at the enzyme level for mutant 41/215 was very low, even with the addition of ATP, when compared to the susceptibility of the virus with these mutations (Table 2). It may be possible that different combinations of mutations may produce differing mechanisms of resistance, some of which are not analyzable in this assay system. There is clearly a need for additional studies to explain the mechanism of the “non-finger β2-β3 loop” mutations (e.g., at codons 41, 215, and 219) in AZT resistance, since it is known that they can contribute considerable AZT resistance. Using a heteropolymer template primer instead of poly(A)-oligo(dT) in this assay system might give some insight into this effect. Although our assay system did not use the natural substrate or sequence of the template primer, this did not appear important for demonstrating a significant effect on AZT resistance. In our study, the thymidine analog BrdUTP was used as the deoxy-NTP substrate instead of dTTP. However, this has been shown to give similar K_m and V_max values for the wild-type RT enzyme in the Lenti RT assay (4). Although it is possible that there are differences in the kinetic properties of BrdUTP for different RT mutants, we believe that these effects are likely to be small since the ATP-free assay data gave rates similar to that for the wild-type enzyme (with the exception of that for mutant 151). This finding is understandable since ATP-dependent pyrophosphorolysis involves the removal of AZT-MP from a terminated primer and thus does not directly have an impact on the BrdUTP substrate. In addition, our findings related to the ATP-dependent mechanism of resistance for 69S-SS insertion in an AZT mutant background have recently been confirmed (18).

In conclusion, we have shown a higher correlation than previously has been demonstrated between AZT resistance in the virus and resistance to AZT-TP at the RT enzyme level. This is the first time that this correlation has been studied with a large group of AZT-resistant mutants. With some modifications of our in vitro assay system, one could investigate this potential resistance mechanism with other nucleoside analogs and potentially screen for small molecules that block this mechanism. Furthermore, the study of mutant RT crystals in complex with the primer template and ATP or PP_i with an incorporated chain terminator will be of significant interest to confirm this mechanism structurally. These data will be of potential use for the design of novel drugs with greater resilience to common resistance mutations.