Characterization of a Novel Human Immunodeficiency Virus Type 1 Protease Inhibitor, A-790742^▿

Generation of HIV-1 resistant to A-790742 by in vitro selection.

The generation of resistant virus by in vitro passages has been described previously (2, 4, 9, 25, 27, 35). Briefly, 2 × 10⁶ MT-4 cells were infected with WT HIV-1 strain pNL4-3 or RF separately at a multiplicity of infection of 0.003 for 2 h, washed, and then cultured in the presence of A-790742 at an initial concentration of 1.4 nM. Viral replication was monitored by the determination of p24 antigen levels in the culture supernatants by enzyme-linked immunosorbent assay every 3 to 4 days. When p24 antigen levels exceeded 10 ng/ml, the viral supernatants were filtered and used to infect fresh MT-4 cells in the presence of an increasing concentration of A-790742, leading to the generation of virus with increased resistance to A-790742. Samples of viral supernatants from each passage were frozen at −80°C for later analyses.

Titration and 50% effective concentration (EC₅₀) and 50% toxic dose (TD₅₀) determinations using the MTT assay.

The titers of all viruses used for the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay were determined by infecting 2 × 10⁵ MT-4 cells for 1 h at 37°C with serial half-log dilutions of virus. Following infection, cells were washed and plated onto a 96-well plate at a concentration of 10⁴ cells per well in the presence of 0.5% dimethyl sulfoxide (DMSO). Five days postinfection, the level of virus-induced cytopathic effect was measured using an MTT colorimetric assay (2, 12, 16, 31), and the 50% tissue culture infectious dose of each virus was calculated using the Sperman-Karber method.

For EC₅₀ determinations, 10⁶ MT-4 cells were infected with viral stock at a multiplicity of infection of 0.003 (2, 16, 27, 29, 31). One hour postinfection, cells were washed and plated onto 96-well plates at concentrations of 10⁴ cells per well in the presence of eight serial dilutions of each tested compound in triplicate. The final concentration of DMSO in all wells was 0.5%. Five days postinfection, an MTT assay was done and the EC₅₀ for each compound was calculated using Prism software (version 4) as described previously (2, 16, 27, 29, 31).

The TD₅₀ determination was done with MT-4 cells. Cells were plated onto 96-well plates at a concentration of 10⁴ cells per well in the presence of eight serial dilutions of each tested compound in duplicate. The final concentration of DMSO in all wells was 0.5%. Five days later, an MTT assay was done and the TD₅₀ for each compound was calculated using Prism software (version 4) as described before (2, 16, 27, 29, 31).

DNA sequence analyses and construction of recombinant HIV-1 molecular clones.

Culture supernatants from virus passages and plasma samples from patients who failed LPV/RTV clinical trials were analyzed by DNA sequence analysis. RNA from these samples was isolated using the QIAamp RNA extraction kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Subsequently, it was processed by reverse transcription-PCR (RT-PCR) and nested PCR to amplify the HIV-1 protease genes. The amplified samples were sequenced using the ABI Prism dye terminator cycle sequencing ready reaction kit and were analyzed on an Applied Biosystems 3100 genetic analyzer.

HIV-1 molecular clones containing mutations in the protease gene were constructed for study in single-cycle assays. For LPV-resistant HIV-1 patient samples, the protease coding region was first RT-PCR amplified with primers that incorporated ApaI and XmaI restriction sites and then was cloned into a pNL4-3 replication-defective viral vector with a luciferase gene (pNL4-3-Luc), which contains an ApaI site upstream and an XmaI site downstream of the protease coding region (27, 28). In addition, HIV-1 molecular clones with mutations selected by A-790742 in the protease gene were constructed by generating PCR fragments from the protease region of different mutant viruses derived from passage in the presence of A-790742, cloning the fragments into the pNL4-3-Luc vector as ApaI-XmaI fragments, and selecting individual clones containing the mutation(s) of interest. The RF-Luc constructs contained the WT or mutant RF protease gene inserted into the pNL4-3-Luc vector as an ApaI-XmaI fragment.

Antiviral activity of PIs against HIV-1 molecular clones with mutations in the protease gene.

Antiviral activities of HIV-1 PIs against HIV-1 molecular clones with and without mutations in the protease gene were measured using a single-cycle assay. Human embryonic kidney 293 (HEK-293) cells were cotransfected with HIV-1 molecular clones containing a luciferase reporter gene plus a vesicular stomatitis virus G envelope protein expression vector by using Lipofectamine 2000 and Plus reagent (Invitrogen). The transfection conditions were optimized to achieve >90% transfection efficiency according to the manufacturer's instructions. Four hours after transfection, cells were trypsinized and 2 × 10⁴ cells per well were plated onto 96-well plates containing serial dilutions of each tested PI in triplicate. Forty-eight hours posttransfection, pseudotyped viral stocks generated in the presence of inhibitors were used to infect fresh HEK-293 cells plated on 96-well plates at a density of 10⁴ cells per well. Replication in each well was monitored by measuring luciferase expression in infected target cells 48 h after infection. The percent inhibition of luciferase signal in cells treated with inhibitors relative to that in the untreated cells was calculated. EC₅₀s were determined using Prism software (11, 30, 33). Replication capacities (RCs) of mutant constructs were calculated by comparing the luciferase activity generated by a mutant to that generated by the WT. The RCs were expressed as percentages of the replication of the WT construct.

(This work was presented in part at the 45th Interscience Conference on Antimicrobial Agents and Chemotherapy, 2005 [28a].)

Antiviral activity of A-790742 against WT HIV-1 and LPV-resistant HIV-1 mutant viruses.

The anti-HIV activity of A-790742 (Fig. 1) against WT HIV-1 and HIV-1 mutant viruses with LPV-resistant mutations was compared to those of LPV, ATV, darunavir (DRV), and tipranavir (TPV). A-790742 had an EC₅₀ of 3 nM against WT pNL4-3 and 5 nM against WT RF in the absence of HS (Table 1). Thus, its activity was similar to or slightly better than that of ATV against pNL4-3 and RF. It was six times more active than LPV against both viruses and four times better than DRV against pNL4-3. A-790742 was 61 and 41 times more active than TPV against pNL4-3 and RF, respectively. In the presence of 50% HS, A-790742 was approximately seven times less potent than without HS, with an EC₅₀ of 20 nM. It was 7.5 and 392 times more potent than LPV and TPV, respectively, under the same conditions (Table 1).

The activities of A-790742, ATV, LPV, DRV, and TPV were evaluated against a panel of passaged viruses derived from in vitro selection with LPV. These viruses had key mutations in protease at the following positions: 32, 46, 47, and 84 for A17; 46, 50, 54, and 82 for B26; and 32, 46, 47, and 84 for P25 (Table 2). The EC₅₀s of A-790742 against A17-, B26-, and P25-passaged viruses were 3- to 15-fold higher than that of the WT pNL4-3. A-790742 was the least active against the P25 virus, with a 15-fold change in its EC₅₀. ATV, on the other hand, was 13 to 89 times less active against the same panel of viruses than against WT pNL4-3. It was more potent against the B26 virus and less potent against the A17 and P25 viruses. DRV was 28- to 71-fold less potent against this panel of viruses than against WT pNL4-3. It was more active against the A17 virus than the B26 and P25 viruses. TPV was tested against A17 only and was two times less potent than it was against the WT. As reported before, all of these mutant viruses were highly resistant to LPV, with 59- to more than 500-fold reductions in susceptibility relative to that of the WT virus (2, 27). Taking these results together, A-790742 displayed anti-HIV activity superior to those of LPV, ATV, DRV, and TPV for WT pNL4-3 and RF as well as for LPV-resistant HIV-1 mutants.

Antiviral activity of A-790742 against HIV-1 molecular clones with protease genes derived from patient isolates.

To determine if A-790742 could effectively inhibit different PI-resistant viruses, it was tested along with LPV and ATV in single-cycle assays against HIV molecular clones containing protease genes derived from HIV-1 isolates from patients who failed different PI treatments (Table 3). The first three constructs in Table 3 were derived from samples from patients who failed LPV treatment. The rest of the constructs were from a panel of molecular clones displaying reduced susceptibility to more than one PI. Each of these tested isolates contained a mixture of primary and secondary mutations as designated by the International AIDS Society-USA resistance panel (11, 13). In general, all clones were less resistant to A-790742 than to LPV and ATV. A-790742 was active against clones with V82A/F/T mutations. However, combined mutations at positions 71 and 84 appeared to confer some degree of resistance to A-790742. Of note, mutations at positions 71 and 84 were selected during the passage of pNL4-3 in the presence of A-790742 as described below. Other secondary mutations also contributed to the generation of resistance to all tested PIs.

Selection of A-790742-resistant virus by in vitro passages.

In order to understand the importance of the genetic background in the selection of viruses resistant to A-790742, two different strains of HIV-1 (pNL4-3 and RF) were passaged in the presence of increasing concentrations of A-790742 (Tables 4 and 5) . Viral replication was monitored every 3 to 4 days by the detection of viral antigen (p24) before increasing the concentration of the inhibitor. pNL4-3 and RF initially were grown in the presence of 1.4 nM A-790742 (i.e., one-half of the EC₅₀ for pNL4-3 virus and one-quarter of the EC₅₀ for RF virus). There also was an attempt to grow RF virus at an initial concentration of 3 nM A-790742 (one-half of the EC₅₀), but viral replication was very low (data not shown). During the selection process, the A-790742 concentration was increased gradually to a final concentration of 2 μM (pNL4-3) (Table 4) or 0.5 μM (RF) (Table 5). It took approximately 3.5 months to complete the selection. The selection with pNL4-3 progressed faster than that with RF virus. A total of 21 passages were performed for pNL4-3, but only 15 passages were performed for RF. This difference might be due to the different genetic backgrounds of the two viruses and suggests that resistance in the RF background developed less rapidly than in the pNL4-3 background.

Sequence analyses of the protease-coding region from HIV-1 passaged in A-790742.

To monitor the protease mutations selected by A-790742, RNA was isolated from the culture supernatant of passages 5 to 10 and 13, 15, 18, and 21 for pNL4-3 selection (Table 4) and from those of passages 6, 8, 10, 13, and 15 for RF selection (Table 5). The RNA was amplified by RT-PCR and nested PCR, and then the population sequences of the PCR products were determined. In the selection with pNL4-3, two primary mutations initially were observed: I84V (passages 5 to 10 and 13) and V82L (passages 8, 9, 10, and 13). These two mutations also were detected in passages 15, 18, and 21, suggesting that they were important for the generation of viral resistance. In passages 15, 18, and 21, the additional mutations L23I, L33F, K45I, A71V, and V77I also were detected. The A-790742 resistance selection using RF virus produced a different pattern of mutations. The mutation V75I was found in passage 8, but this mutation became undetectable in passage 10. Instead, mutations A71V and V82G were found in passage 10. In passages 13 and 15, the additional mutations M46I and L63P were observed. The cleavage sites of viruses derived from passage 21 of pNL4-3 selection and passage 15 of RF selection were sequenced, and no mutations were detected.

Susceptibility of HIV-1 selected by A-790742 to different PIs.

To determine the susceptibility of HIV-1 mutants selected by A-790742 to different PIs, the titers of several pNL4-3 passage viruses (passages 9, 15, and 21) and one RF passage virus (passage 13) were determined, and their levels of resistance to A-790742 and a number of commercially available PIs, such as ATV, LPV, saquinavir (SQV), RTV, amprenavir (APV), IDV, and nelfinavir (NFV), were determined using an MTT assay (Table 6). Virus from pNL4-3 passage 9 was 12.7-fold less susceptible to A-790742 than the WT pNL4-3 virus. This passage was the most resistant to RTV, with a 26.7-fold increase in its EC₅₀ compared to that of the WT pNL4-3 virus. ATV, LPV, SQV, APV, IDV, and NFV demonstrated three- to ninefold increases in their EC₅₀s at passage 9 compared to that of the WT. Virus selected in pNL4-3 passage 15 had similar levels of resistance to A-790742 and RTV, with a 28-fold change in the EC₅₀. In addition, ATV and APV had 13.6- and 10.5-fold changes in their respective EC₅₀s for virus from passage 15. LPV, SQV, IDV, and NFV had about five- to sixfold changes in their EC₅₀s with this passaged virus. pNL4-3 passage 21 virus had the highest degree of resistance to all tested PIs. It was highly resistant to A-790742 (>2,500-fold resistance), RTV (>141-fold resistance), ATV (145-fold resistance), and NFV (>93-fold resistance). However, it had only a low to moderate level of resistance to the other tested PIs. In particular, it retained most of its sensitivity to SQV, LPV, and IDV.

Virus from passage 13 selected in the RF background had similar levels of resistance to A-790742, RTV, and NFV, with 72-, 79-, and 67-fold changes in EC₅₀s, respectively, compared to that of the WT RF virus. APV, ATV, and IDV had about 31- to 41-fold changes in their EC₅₀s with this passaged virus. In addition, the passage 13 virus displayed the greatest resistance to LPV, with a >301-fold increase in its EC₅₀ compared to that of the WT RF virus, and it was the least resistant to SQV, with a 21-fold increase in its EC₅₀ compared to that of the WT RF virus.

Phenotypic analyses of HIV-1 molecular clones resistant to A-790742.

To determine the effect of genotypic changes found in the passaged variants on the susceptibility of HIV-1 to A-790742 and commercially available PIs, HIV-1 molecular clones with protease mutations selected by A-790742 in pNL4-3 and RF were engineered in the pNL4-3-Luc vector (Table 7). These constructs permitted the rapid and sensitive quantification of drug susceptibility during a single round of viral replication in the presence of increasing concentrations of drugs (30, 33).

Molecular clones containing the single mutation V82L or I84V in the pNL4-3 background displayed resistance only to A-790742 and RTV (Table 7). In particular, clone 1, which contained an I84V mutation, was two- and threefold less susceptible to A-790742 and RTV, respectively, than WT pNL4-3-Luc. In contrast, clone 2, which harbored a V82L mutation, displayed 22-fold more resistance to A-790742 but only 7-fold more resistance to RTV. Mutant clone 3, which had secondary mutations L33F and K45I in addition to primary mutations V82L and I84V, was 23-fold less susceptible to A-790742. Clone 3 was more resistant than clone 2 to all other tested PIs except SQV (Table 7). Clone 4 contained secondary mutations L33F, A71V, G73S, and V77I in addition to primary mutations V82L and I84V. Mutations L33F, A71V, and V77I were secondary mutations detected by population sequencing in the passaged variants (Table 4). However, G73S was not detected by population sequencing; it was identified by sequencing individual TA clones derived from pNL4-3 passaged in the presence of A-790742. Clone 4 was more resistant to A-790742 (158-fold change in the EC₅₀) but less resistant to all tested PIs than clone 3 (Table 7). The difference in the resistance profiles of clones 3 and 4 suggests that K45I is important for the resistance to all PIs except A-790742 and SQV. Furthermore, individual or combined A71V, G73S, and V77I mutations contributed to the increase in resistance to A-790742. Among these three mutations, A71V was detected in both pNL4-3 and RF after selection using A-790742 and probably contributed most of the additional resistance to A-790742. It is of note that molecular clone 3, which contained most of the mutations of passaged HIV-1 P21 (Table 6), showed the same trend of resistance as P21. Both clone 3 and P21 were more resistant to A-790742, ATV, and RTV than the other tested PIs.

Clone 5, which contained A-790742-selected PI mutations in the RF background, also was evaluated in the single-cycle assay (Table 7). This clone contained the primary mutation V82G and the other mutations L63P and A71V. It was 72-fold less susceptible to A-790742 than the WT construct and was highly resistant to LPV and RTV, with more than 1,300-fold resistance to these two PIs. However, it had only moderate levels of resistance to the other tested PIs. Importantly, it was only 28 times more resistant to SQV than the WT construct. Interestingly, overall this construct conferred a higher level of resistance to all tested PIs than all of the molecular clones in the pNL4-3 background (clones 1, 2, 3, and 4) (Table 7).

During phenotypic testing, significant differences in replication efficiencies between molecular clones with WT and mutant protease genes in the RF background were observed (Fig. 2). Constructs with the combined mutations L63P, A71V, and V82G or M46I, A71V, and V82G replicated 99% less efficiently than the WT HIV-RF-luc construct based on the difference in luciferase signal. The luciferase signal of the RF molecular clone harboring the mutations M46I, A71V, and V82G was so low that its phenotype could not be determined in the single-cycle assay. One of the possible reasons for the lower replication capacity of these constructs with the V82G mutation is the negative impact of this mutation on the fitness of the constructs. The presence of this mutation also might explain the slow growth of virus during the selection with A-790742 in the RF background. All other clones with mutations in the protease gene replicated at the same level as the virus with WT protease (pNL4-3-Luc).