In Vitro Hypersusceptibility of Human Immunodeficiency Virus Type 1 Subtype C Protease to Lopinavir

Abstract

In order to characterize the impact of genetic polymorphisms on the susceptibility of subtype C strains of human immunodeficiency virus type 1 to protease inhibitors (PIs), a subtype B protease that originated from an infectious clone was modified through site-directed mutagenesis to include the amino acid residue signatures of subtype C viruses (I15V, M36I, R41K, H69K, L89 M) with (clone C6) or without (clone C5) an I93L polymorphism present as a molecular signature of the worldwide subtype C protease. Their susceptibilities to commercially available PIs were measured by a recombinant virus phenotyping assay. We could not detect any differences in the 50% inhibitory concentration (IC₅₀s) of amprenavir, indinavir, ritonavir, saquinavir, and nelfinavir for the clones analyzed. However, we did observe hypersusceptibility to lopinavir solely in clone C6, which includes the I93L substitution (a 2.6-fold decrease in the IC₅₀ compared to that for the subtype B reference strain). The same phenotypic behavior was observed for 11 Brazilian and South African clinical isolates tested, in which only subtype C isolates carrying the I93L mutation presented significant hypersusceptibility to lopinavir.

Initially characterized in developed countries such as the United States and countries in Western Europe, subtype B of human immunodeficiency virus (HIV) type 1 (HIV-1) was considered the major variant outside Africa and in the rest of the world. As the HIV-1 pandemic has extended around the globe, strain C was shown to be the most prevalent subtype, accounting for 56% of the infections worldwide (4). Subtype C was first detected in South Africa and Ethiopia in retrospectively analyzed samples from 1984 and 1986, respectively (9, 36); and it has been found in the majority of sub-Saharan African countries, such as South Africa, Botswana, Tanzania, and Kenya (3, 15, 16, 19, 20, 31, 32). Recombinants of subtype C strains and stains of subtypes previously prevalent in these countries have also been documented in Zambia (23) and Tanzania (11). Outside Africa, India has the largest population infected with subtype C viruses (25), and it is estimated that by 2010 India will have the highest number of HIV-1-infected people in the world (2). Finally, China also has a high prevalence of subtype C strains, and notoriously, these strains have subtype B and subtype C recombinant genomes (21).

Drugs targeting the reverse transcriptase (RT) and protease (PR) of HIV-1, such as nucleoside RT inhibitors, nonnucleoside RT inhibitors, and PR inhibitors (PIs), have revolutionized the treatment of HIV-1-infected individuals (5, 17). However, not all patients respond to highly active antiretroviral therapy, and in many patients the virus develops drug resistance, one of the most serious obstacles to sustained suppression of HIV-1 (12, 22, 34). The emergence of amino acid mutations associated with resistance to RT inhibitors and PIs has been extensively characterized (8, 24). The primary mutations found in the PR gene lead to a severalfold decrease in sensitivity to one or more PIs (8, 24). Compensatory mutations may not result in a significant decrease in drug sensitivity but are associated with restoration of the original viral fitness in the presence of existing inhibitors (8, 24). The PR genes of the non-subtype B isolates found in drug-naïve patients have amino acid signatures different from those found in their subtype B counterparts. The high prevalence of the L10I, K20R, M36I, and L89M mutations in non-subtype B strains is largely known (18, 29). Likewise, the L63P and V77I substitutions are more frequent in subtype B isolates (18). Some of these substitutions are reported to be implicated in drug resistance in subtype B strains. Furthermore, almost all studies of HIV-1 drug susceptibility have been performed in developed countries, where subtype B still dominates the epidemic. Similarly, the resistance mutations selected during antiretroviral treatment were mostly studied with subtype B-infected patients.

The biochemical and phenotypic roles of the natural polymorphisms found in the PRs of non-subtype B strains are poorly understood. Velazquez-Campoy and colleagues (33) observed a 2.7-fold increased fitness of the clade C PR compared to that of the clade B PR. Whether this PR kinetics parameter influences the efficiency of virus replication still needs to be addressed. Given the increasing genetic heterogeneity of HIV-1 isolates worldwide and the intention to expand the use of PIs and other antiretroviral drugs to developing countries, it is important to characterize the PI susceptibility profiles of the subtype C isolates occurring among drug-naïve populations. It is important to check for the specific characteristics of these isolates to guide better, longer-lasting, and efficient drug regimens. This work describes the results of a phenotypic study of HIV-1 subtype C PR genetically reconstructed from a prototypic clade B PR backbone (infectious clone pNL4-3 PR) into which the molecular signatures of clade C were introduced. These polymorphisms of clade C PRs are commonly found in isolates from Brazilian and South African drug-naïve individuals. Also, some clinical isolates of both clades from Brazil and South Africa were phenotypically tested.

MATERIALS AND METHODS

Clones and clinical samples.

The PR gene from infectious clone pNL4-3 (1), a subtype B strain, was used as the prototypic clade B PR backbone for most of the site-direct mutagenesis studies.

Plasma was obtained from 11 HIV-1-positive antiretroviral drug-naïve individuals, as confirmed by serology, at different Brazilian and South African AIDS clinics and voluntary counseling and testing centers. These samples were generated from a study previously approved by a Brazilian institutional review board (project 526-CONEP).

RNA isolation, PCR amplification, and sequencing.

Virus RNA was isolated from plasma by using QIAmp Viral RNA kit (QIAGEN, Hilden, Germany), as described previously (27). Following cDNA generation with random primers, nested PCR was conducted for amplification of individual PRs (whole region) by using outer primers RPV5 (5′-GGGAAGATCTGGCCTTCCTGCAAGGG-3′) and RVP3 (5′-GGCAAATACTGGAGTATTGTATGG-3′) and inner primers K1 (5′-CAGAGCCAACAGCCCCACCA-3′) and MOP2 (5′-GGTCCATCCATTCCTGGTTT-3′) with the PCR conditions described elsewhere (27). The PCR fragments were sequenced in both directions with an ABI 310 automated sequencer (Applied Biosystems, Foster City, Calif.) with the same primers used in the second round of the amplifications.

Sequence analysis and subtyping.

The sequences generated were aligned and manually edited with the ClustalW program (30) by using the Genetic Data Environment package for data exchange (26). A reference set for HIV-1 subtyping analysis from the Los Alamos HIV sequence database (http://hiv-web.lanl.gov/) was included in the alignment. Phylogenetic inferences were deduced by the neighbor-joining method with the F84 model of substitution in the PAUP package (version 4.0b2a) (28).

A subtype C consensus sequence was deduced by using all subtype C sequences deposited in the Los Alamos HIV sequence database; and multiple sequences from Brazil (n = 22) (27), South Africa, Tanzania, and Zambia (n = 53), and eastern India (n = 8) (25) were included in the alignment to generate country-based consensus sequences by using the VESPA program (10).

Site-directed mutagenesis.

In order to introduce the molecular signatures of the subtype C PR into the PR of prototypic infectious clone pNL4-3, a PR fragment was generated by PCR with the outer primers described above and cloned into the pCR4 TOPO vector with the TOPO TA cloning kit (Invitrogen, Carlsbad, Calif.) according to the protocol of the manufacturer. This gene was then subjected to site-directed mutagenesis in order to modify the target codons by a PCR mutagenesis procedure described by the manufacturer (Quick Change kit; Stratagene, La Jolla, Calif.). Briefly, the procedure induces the polymerization of a new strand from the original target plasmid by use of a set of complementary oligonucleotides as primers for the reaction for each strand. Those primers anneal at the same position in the plasmid and introduce the desired point mutation (or mutations) by mismatching. A Pfu DNA polymerase (Invitrogen) is used to complete the strands. To optimize the efficiency of mutagenesis, the parental methylated strands are digested with DpnI, an endonuclease specific for methylated DNA that allows the new mutagenized strands to reanneal before being transformed into competent bacterial cell (Escherichia coli DH5α). Table 1 shows the primers used for mutagenesis. Six differences were found between the subtype C and subtype B consensus sequences (I15V, M36I, R41K, H69K, L89M, and I93L). However, some investigators have excluded the I93L substitution from the subtype C consensus sequence (33). In order to explore the importance of this specific signature in the subtype C PR drug resistance phenotype, we have generated two clones, clones C5 and C6. Clone C6 contains all six substitutions, and clone C5 lacks the I93L polymorphism.

TABLE 1.

Primers used for site-directed mutagenesis of pNL4-3

Primer	Sequence (5′-3′)a	Mutation induced
Pro I15V-F	C ACA ATA AAG GTA GGG GGG CAA TT	I15V
Pro I15V-R	AA TTG CCC CCC TAC CTT TAT TGT G	I15V
Pro M36I-F	TA TTA GAA GAA ATC AAT TTG CCA GG	M36I
Pro M36I-R	CC TGG CAA ATT GAT TTC TTC TAA TA	M36I
Pro R41K-F	AAT TTG CCA GGA AAA TGG AAA CCA A	R41K
Pro R41K-R	TGG TTT CCA TTT TCC TGG CAA ATT	R41K
Pro H69K-F	AA ATC TGC GGA AAA AAA GCT ATA GG	H69K
Pro H69K-R	CC TAT AGC TTT TTT TCC GCA GAT TT	H69K
Pro L89M-F	TT GGA AGA AAT ATG TTG ACT CAG AT	L89M
Pro L89M-R	AT CTG AGT CAA CAT ATT TCT TCC AA	L89M
Pro I93L-F	CTG TTG ACT CAG CTT GGC TGC ACT	I93L
Pro I93L-R	AGT GCA GCC AAG CTG AGT CAA CAG	I93L

^aMutations induced in the pNL4-3 sequence by each set of primers are highlighted in boldface.

HIV-1 PR phenotyping.

The detection of HIV-1 phenotypic resistance to U.S. Food and Drug Administration-approved antiretroviral compounds was performed by using a recombinant virus assay technology, as described previously (7), with slight modifications. The HIV-1 RNA was extracted from plasma samples, and a 640-bp fragment containing the entire HIV-1 PR-coding sequence was amplified by nested RT-PCR with primers RVP3 and RVP5 (outer primers) and primers K1 and MOP2 (inner primers), as described previously (13). Only the inner primers were used when clones C5 and C6 were amplified. The pool of PR-encoding fragments was then cotransfected into CD4⁺ T lymphocytes (MT-4 cells) with plasmid pGEMT3ΔPR, which carries defective HIV HXB2 genomic cDNA that lacks the gene for PR (ΔPR) (13). Homologous recombination led to the generation of a chimeric virus containing PR sequences derived from mutagenized clones or patient viruses. A chimeric virus containing the NL4-3 PR (HXB2/NL4-3-PR) was also generated as a subtype B control virus to obtain the reference 50% inhibitory concentrations (IC₅₀s) during the phenotyping assay. The susceptibilities of chimeric virus to all U.S. Food and Drug Administration-approved PI compounds (amprenavir, indinavir, lopinavir, ritonavir, saquinavir, and nelfinavir) were determined by three independent 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)-based cell viability assays with MT-4 cells, each of which was performed in duplicate, as described previously (7). All statistical treatments for calculation of the IC₅₀s for the clones and isolates were performed by using the Analyze-it program (version 1.62) for the Microsoft Excel statistics package and Sigmaplot software. Briefly, the percentage of viable cells obtained by the MTT assay after 7 days of infection was plotted semilogarithmically against the concentrations of the drugs tested. A Hill's three-parameter nonlinear regression was performed to obtain the sigmoid curve of viable cells and the IC₅₀ of each drug for each virus tested. The mean IC₅₀s are always shown together with the standard deviations for the replicates.

RESULTS

In order to characterize the genetic profiles of the PRs from subtype C viruses, the consensus PR amino acid sequences of subtype C isolates from Brazil, India, and Africa were obtained by use of the subtype C sequence from the Los Alamos HIV sequence database and were aligned with the subtype B PR consensus sequence (Fig. 1) Six differences were found between the subtype C and the subtype B consensus sequences (I15V, M36I, R41K, H69K, L89M, and I93L). However, some investigators have excluded the I93L substitution from the subtype C consensus sequence (33). These six substitutions were then inserted into a prototypic subtype B infectious clone, pNL4-3, by site-directed mutagenesis. In order to specifically study the role of the I93L substitution in the subtype C backbone, two different clones were generated: clone C6 included all six signatures found in the subtype C PR consensus sequence, and clone C5 lacked the I93L substitution (see the Fig. 1 legend for details).

FIG. 1.

VESPA program analyses of PR consensus sequences from HIV-1 isolates from Brazil, India, and Africa (ZM, Zambia; TZ, Tanzania; ZA, South Africa; BW, Botswana) showing signature sequences for each of these groups. Consensus subtype C PR sequences for each group were compared to the consensus subtype C PR sequence for isolates worldwide from the Los Alamos HIV sequence database. Signatures present in less than 50% of the isolates are not shown, with the exception of L63P in the PR of Brazilian subtype B isolates (47.5%). Clones C5 and C6 originated in this study, and their PR sequences are also included in the alignment. The 10 sequences listed after clones C5 and C6 are related to the Brazilian and South African clinical subtype C isolates studied in this work. At the bottom of the alignment a consensus subtype B PR sequence is compared with the PR sequences of strains Br-RS2094 and HXB2/NL4-3-PR. Molecular signatures marked in gray differentiate the subtype C PR from the subtype B PR. The isolate subtype assignment was generated as described in Materials and Methods and is depicted in parentheses after the isolate designation.

The susceptibilities of these clones to all commercially available PIs were determined, and the results are depicted in Table 2 The IC₅₀ of lopinavir was 2.9 times lower for subtype C clone C6 than for reference strain HXB2/NL4-3-PR (P < 0.001 by the two-tailed t test). Interestingly, this effect was not observed when clone C5 was tested (P < 0.4 by the two-tailed t test). This hypersusceptibility was observed only when lopinavir was used in the phenotypic test.

TABLE 2.

Phenotypic analysis of clones C5 and C6 and reference strain HXB2/NL4-3-PR

Isolate	IC50 (nM)a
	APV	SQV	IDV	RTV	NFV	LPV
pNL43	72.8 ± 6.2 (1)	3.4 ± 0.45 (1)	7.3 ± 0.45 (1)	44.0 ± 3.5 (1)	48.3 ± 5.8 (1)	27.13 ± 6.9 (1)
Clone C5	68.5 ± 5.16 (0.94)	2.9 ± 0.4 (0.85)	6.6 ± 0.74 (0.9)	46.2 ± 3.4 (1)	39.2 ± 2.9 (0.81)	19.0 ± 5.2 (0.69)
Clone C6	70.7 ± 4.3 (0.97)	2.6 ± 0.37 (0.75)	6.4 ± 0.37 (0.87)	42.0 ± 3.5 (0.95)	32.2 ± 4.2 (0.66)	9.4 ± 0.52 (0,34)

^aThe values represent the means ± standard deviations of six independent experiments. The values in parentheses are the fold resistance compared to that of strain HXB2/NL4-3-PR abbreviations: APV, amprenavir; SQV, saquinavir; IDV, indinavir; RTV, ritonavir; NFV, nelfinavir; LPV, lopinavir.

In order to see if this hypersusceptibility phenotype observed in these two artificial clones could be found in HIV-positive drug-naïve patients infected with subtype C isolates, the PCR products directly generated from the PR genes of five isolates from Brazil and six isolates from South Africa by RT-PCR were selected. After phylogenetic analysis, eight clones with a clear assignment to clade C were selected (see the Fig. 1 legend for details). Isolate Br-SC1812 belonged to subtype C and did not carry the I93L mutation, and isolate Br-RS2172 belonged to subtype B and carried the I93L substitution. All remaining isolates belonged to subtype C and carried the I93L mutation. Two subtype B isolates (isolates Br-RS2099 and Br-RS2094) from the same geographical area were used as controls in the phenotyping assay. A 600-bp pol fragment was amplified from these 11 isolates and was recombined with pGEMT3ΔPR to generate chimeras. The recombinant viruses were phenotyped by using lopinavir. The fold resistance generated in these analyses is depicted in Fig. 2 There was a clear correlation of the results obtained with the infectious clone and the clinical isolates. Isolate Br-SC1812, which belongs to the subtype C clade and which lacks the I93L mutation, was not hypersusceptible to lopinavir. Similarly, the IC₅₀ for isolate Br-RS2172, which clustered with clade B and which carried the I93L substitution, was not lower than that for reference strain HXB2/NL4-3-PR. All the other clade C isolates showed clear hypersusceptibility to lopinavir, with the IC₅₀s for those isolates being from 16.2 to 2.6 times lower than that for reference strain HXB2/NL4-3-PR. The majority of the African isolates had higher levels of hypersusceptibility to lopinavir than their Brazilian counterparts. The results of phenotypic analysis of four selected South African isolates and reference strain HXB2/NL4-3-PR for susceptibility to each commercially available PI except lopinavir is depicted in Table 3 Hypersensitivity was observed only when lopinavir was tested.

FIG. 2.

Histogram showing the phenotypic resistance to lopinavir found for 11 clinical isolates recovered in Brazil (Br) and South Africa (ZA) (for sequence details, see the alignments depicted in Fig. 1). The results indicate the fold resistance of HXB2/NL4-3-PR to lopinavir (IC₅₀ = 27 nM for virus in MT-4 cells) deduced from the same set of experiments. Each bar represent the average of six replicates, and the lines at the top of each bar represent the standard deviations.

TABLE 3.

Phenotypic analysis of resistance of South African isolates and reference strain HXB2/NL4-3-PR to NL4-3 each commercially available PI

Isolatea	Fold resistance
	IDV (3.0)c	RTV (3.5)	NFV (4.0)	SQV (2.5)	APV (2.5)
L45	1b	0.5	0.5	0.8	0.9
MD	1	0.7	0.7	0.6	0.68
PT02	1.3	1.2	0.5	0.8	0.8
L14	1	0.5	0.3	0.4	1

^aAll isolates had the I93L mutation.

^bFold resistance compared to that of pNL4-3 (IC₅₀s for pNL4-3, indinavir [IDV], 12 nM; ritonavir [RTV], 47 nM; nelfinavir [NFV], 62 nM; saquinavir [SQV], 12 nM; amprenavir [APV], 100 nM).

^cThe values in parentheses are the biological cutoffs generated by Harrigan et al. (6) by a similar phenotyping technology. The values are expressed as fold resistance compared to that of pNL4-3. The cutoff for lopinavir was not determined in this study.

DISCUSSION

Hypersusceptibility to antiretroviral compounds is found in several virus isolates from treated and untreated individuals. This feature is biologically defined by a lower IC₅₀ of a certain drug compared to the IC₅₀ for a reference strain (usually less than two times lower). It is well known that the N88S substitution selected by nelfinavir therapy confers hypersusceptibility to amprenavir (38). This in vitro characteristic can also influence the clinical outcome when a large decrease in the viral load is observed in patients whose isolates carry the N88S mutation following amprenavir treatment (37). Other reports have also shown hypersusceptibility to nonnucleoside RT inhibitors in strains carrying nucleotide- or thymidine-associated mutations (35). In our study, we have observed a lopinavir IC₅₀ 2.6 times lower for clone C6 carrying subtype C signatures than for reference strain HXB2/NL4-3-PR. Interestingly, this effect was not observed when clone C5 was tested, implicating an association of the I93L mutation with this phenotype. It is notable that this mutation was previously reported to be associated with isolates from patients who had failed lopinavir therapy in clinical trials in which individuals infected with subtype B were studied (14). We could not observe any difference in the IC₅₀s of the other commercially available PIs tested when the IC₅₀s for these two clones were compared to those for strain HXB2/NL4-3-PR.

The same phenotypic behavior was observed when clinical isolates were phenotyped by a similar methodology. In this case, the IC₅₀s for all isolates belonging to subtype C and carrying the I93L mutation, regardless of the country from which they were isolated, ranged from 16.2 to 2.6 times lower than that for the reference strain (P < 0.0001 by the two-tailed t test). However, when subtype B or C isolates missing this specific substitution were analyzed, we observed IC₅₀s that were from 1.2 to 0.7 times lower than that for reference strain HXB2/NL4-3-PR (P < 0.77 by the two-tailed t test). Interestingly, it was not possible to observe the same phenotypic effect when I93L was naturally present in subtype B (isolate Br-RS2172). This hypersensitivity was notable only when lopinavir was tested with our clinical isolates and clone C6. These findings provide strong evidence for the role of the I93L mutation in the hypersusceptibility to lopinavir in the presence of a subtype C PR backbone. The molecular interactions between the specific substitutions of PR in the different clades may account for the different behaviors of subtype variants carrying the same I93L substitution. Harrigan et al. (6) have used a similar phenotyping technique to test several isolates obtained from antiretroviral drug-naïve individuals, including 358 South African patients. They have not found any significant differences in the IC₅₀s for subtype B and subtype C isolates with any drug tested. Other investigators have not included lopinavir among the antiretroviral compounds tested in phenotypic assays (35). However, in consonance with our results, those investigators could not show any differences in the IC₅₀s for subtype B and subtype C strains when the other PIs were tested. In another study, Velazquez-Campoy et al. (33) expressed HIV-1 protease in E. coli and purified recombinant PR enzymes from subtype B and subtype C isolates to determine their kinetic parameters by in vitro enzymatic activity assays with fluorogenic peptides. They demonstrated that the K_is for all PIs tested (amprenavir, ritonavir, saquinavir, indinavir, and nelfinavir) were similar. However, again, lopinavir was not included in that analysis, and they did not include the I93L substitution in the subtype C recombinant PR enzyme. They also demonstrated that the vitality (k_cat/V_max) of the PRs from subtype C isolates was 2.5 times higher than that of the PRs from subtype B isolates.

HIV infection and AIDS is the leading cause of death in sub-Saharan Africa and the fourth leading cause of death worldwide. A dramatic drop in the mortality rate was observed in AIDS patients in developed countries after the implementation of highly active antiretroviral therapy. However, few people from countries with limited resources have access to antiretroviral drugs. In fact, the World Health Organization conservatively estimated that in 2002, some 6 million people in resource-limited settings are in need of antiretroviral therapy (http://www.unaids.org). The introduction of innovative and affordable antiretroviral combinations in these countries is urgently needed. As subtype C strains are responsible for the majority of new infections in the HIV infection and AIDS epidemic, it is critical that the first drug treatment regimen be potent and tailored to yield long-lasting results without therapeutic failures. The findings reported here indicate that subtype C isolates are hypersusceptible to lopinavir in vitro. We do not know if this in vitro phenotypic hypersensitivity to this PI would result in different clinical outcomes. It is known that the combination of lopinavir and ritonavir found in Kaletra has a high inhibitory quotient, but this hypersensitivity may not influence clinical outcomes. However, these findings must be further tested in controlled clinical trials comparing the virological responses of patients infected with subtype B HIV isolates and those infected with subtype C HIV isolates receiving lopinavir.