Abstract
The peptide fusion inhibitor (PFI) enfuvirtide is the first of a new class of entry inhibitors to receive FDA approval. We previously determined the susceptibility of 55 PFI-naïve-patient isolates to enfuvirtide and a second peptide inhibitor, T-649. Seven of the 55 viral isolates were insusceptible to enfuvirtide, T-649, or both inhibitors in the absence of prior exposure. To determine the molecular basis of the insusceptible phenotypes, we PCR amplified and cloned five PFI-insusceptible and one PFI-susceptible, full-length, biologically functional env genes and characterized viruses pseudotyped with the Env proteins in a single-round drug sensitivity assay. Overall, the mean 50% inhibitory concentrations of enfuvirtide and T-649 for the PFI-insusceptible Env pseudotypes were 1.4 to 1.7 log10 and 1.2 to 1.8 log10 greater, respectively, than those for a PFI-susceptible lab strain, NLHX; however, all of the PFI-insusceptible Env proteins conserved the sequence of a critical enfuvirtide interaction site (residues 36 to 38 of gp41, GIV) in HR-1. In contrast, multiple amino acid changes were observed C-terminal to HR-1, many of which were located in regions of HR-2 corresponding to the PFI. Nevertheless, peptides based on patient-derived HR-2 sequences were not more potent inhibitors than enfuvirtide or T-649, arguing that the basis of PFI susceptibility is not a higher-affinity, competitive HR-1/HR-2 interaction. These results demonstrate that regions of Env outside the enfuvirtide interaction site can significantly impact the PFI susceptibility of patient-derived Env, even prior to drug exposure. We hypothesize that both gp120 gene- and gp41 gene-encoded determinants that minimize the window of opportunity for PFI to bind provide a growth advantage and possibly a predisposition to resistance to this new class of drugs in vivo.
Viral fusion and entry are mediated by two glycoproteins, one of which (gp120) interacts with CD4 and coreceptor molecules (CXCR4 and CCR5) expressed on the target cell surface (19). The specificity of the gp120-coreceptor interaction is generally dependent on sequences within the third variable (V3) loop of gp120 (reviewed in reference 10). gp120-receptor interactions trigger conformational changes within the noncovalently linked human immunodeficiency virus type 1 (HIV-1) transmembrane glycoprotein, gp41, that promote fusion between the viral and cellular membranes (3).
The ectodomain of gp41 contains a hydrophobic fusion peptide sequence at the amino terminus, followed by two leucine zipper-like heptad repeats (HR-1 and HR-2), which are connected by a disulfide-bonded loop (12). HR-1 and HR-2 are characterized by a 4,3-repeating motif that is predictive of alpha-helical secondary structure and characteristic of protein regions that form a coiled coil. Following receptor binding, conformational changes occur in gp41 that lead to the formation of an HR-1 coiled coil (the prehairpin intermediate) that, through interactions with HR-2, transitions into a stable trimer of hairpins referred to as the six-helix bundle (1, 3, 30, 31).
The peptide fusion inhibitors (PFI) enfuvirtide (formerly known as T-20) and T-649 are synthetic peptides that correspond to overlapping linear amino acid sequences within HR-2 (3, 32). They are pharmacological antagonists that are postulated to interfere with the transition from the prehairpin intermediate to the fusion-active six-helix bundle by competitively binding to targets on the HR-1 coiled-coil structure and preventing intramolecular interactions that normally occur between HR-1 and HR-2. However, these overlapping peptides target distinct regions within HR-1, evidenced by the fact that enfuvirtide-resistant viruses remain fully susceptible to T-649 (7). T-1249 (which targets the same hydrophobic pocket in the HR-1 coiled coil as T-649) has demonstrated potent dose-dependent decreases in viral load in patients harboring enfuvirtide-resistant viruses (G. D. Miralles, J. P. Lalezari, N. Bellos, G. Richmond, Y. Zhang, H. Muchison, R. Spence, C. Raskino, and R. A. DeMasi, presented at the 10th Conference on Retroviruses and Opportunistic Infections, Boston, Mass., 9 to 14 February 2003).
PFI differ from reverse transcriptase or protease inhibitors in that they target a conformational intermediate of the entry process, as opposed to interfering with enzyme-substrate interactions. Recently, enfuvirtide became the first PFI to receive Food and Drug Administration approval. In phase I/II clinical studies, optimal administration of enfuvirtide monotherapy reduced plasma viral load by an average of 1.96 log10 over an 8-day period (17). However, resistant viruses readily emerged in patients receiving suboptimal doses (29) and in patients receiving an enfuvirtide-containing salvage regimen (24; M. Mink, M. Greenberg, S. Mosier, S. Janumpalli, D. Davison, L. Jin, T. Melby, P. Sista, D. Lambert, N. Cammack, M. Salgo, and T. Matthews, presented at the XI International HIV Drug Resistance Workshop, Seville, Spain, 2 to 5 July 2002; P. Sista, T. Melby, M. L. Greenberg, D. Davison, L. Jin, S. Mosier, M. Mink, E. Nelson, L. Fang, N. Cammack, M. Salgo, and T. J. Matthews, presented at the XI Int. HIV Drug Resist. Workshop).
In vitro and in vivo resistance to enfuvirtide frequently involves substitutions in a highly conserved, contiguous three-amino-acid residue sequence within HR-1 (GIV at positions 36 to 38). Thus, this region most likely defines an important interaction site (26). Results from the phase II enfuvirtide trials demonstrated that substitutions in GIV (to DIV or GIM) led to a 3- to 45-fold decrease in susceptibility compared to the baseline variants within the patient swarm (29). Results from additional phase II trials have expanded this putative interaction site to include residues 39 to 45 of HR-1 (24, 33; Sista et al., XI Int. HIV Drug Resist. Workshop). However, approximately 20% of the resistant variants do not contain changes in this region compared to the baseline virus population (29, 33; Sista et al., XI Int. HIV Drug Resist. Workshop). This finding suggests that other residues in Env could influence the mechanism by which clinical resistance develops.
Recent studies described a mechanism of resistance to PFI in which coreceptor specificity modulates susceptibility to both enfuvirtide and T-649 (6, 7, 25). In these studies, it was demonstrated that two independent panels of chimeric viruses containing a CCR5-utilizing V3 loop were four- to eightfold less susceptible to both PFI than the CXCR4-utilizing parent strains NLHX and NL4-3. It was proposed from these studies that the window of opportunity for the PFI to interact with its viral target is governed by coreceptor affinity, and recent studies have supported this model (25). Taken together, the previous studies suggest that, although sequences within HR-1 are the target for PFI, the affinity of gp120 for the coreceptor governs the window of opportunity in which PFI can interact with their target sequences. In the present study, we extend these findings to demonstrate that sequences within gp41, but outside the predicted interaction site in HR-1, can significantly influence baseline susceptibility to this novel class of entry inhibitors.
MATERIALS AND METHODS
Preparation of genomic DNA from HIV-1-infected PBMC cultures.
Patient-derived virus isolates were produced by peripheral blood mononuclear cell (PBMC) coculture and have been described previously (7). For infections, PBMC were purified from the whole blood of a wild-type-CCR5-seronegative donor by Ficoll-Hypaque (Pharmacia, Piscataway, N.J.) density centrifugation. Cells were plated into T-75 flasks at a density of ∼3 × 106 cells/ml and activated with phytohemagglutinin (3 μg/ml) in complete RPMI (20% fetal calf serum [HyClone, Logan, Utah], l-glutamine, nonessential amino acids, and penicillin-streptomycin). Seventy-two hours later, the cells were removed from phytohemagglutinin by centrifugation and infected at a multiplicity of infection of 0.01 by adsorbing virus for 2 h at 37°C. Cells were washed in RPMI and then plated at a density of ∼2 × 106 cells/ml in T-25 tissue culture flasks in complete RPMI supplemented with interleukin 2 (20 U/ml). The production of supernatant p24 in each infected culture was monitored by enzyme-linked immunosorbent assay (Coulter, Miami, Fla.), and 5 × 106 cells were harvested at the peak of infection (day 5) for the preparation of genomic DNA. Genomic DNA was extracted from infected PBMC cultures by using the Qiagen DNeasy kit (Qiagen, Valencia, Calif.). The virus-containing supernatant was also collected at days 5 and 14 postinfection, clarified by low-speed centrifugation, and passed through a 0.45-μm-pore-size filter to produce cell-free virus stocks. Stocks were stored in 1-ml aliquots at −70°C.
Cloning and subcloning of env genes.
Full-length HIV-1 env genes were PCR amplified from genomic DNA extracted from cells harvested at day 5 postinfection by using Expand high-fidelity DNA polymerase (Roche, Indianapolis, Ind.) and the primers envA (nucleotides 5954 to 5982 of HXB2) and envN (nucleotides 9145 to 9171 of HXB2) (www.hiv.lanl.gov, sequence locator tool) (8). The amplified DNA fragments (approximately 3.2 kb) were produced by using the following amplification strategy: 1 cycle at 95°C for 2 min; 10 cycles at 95°C for 30 s, 59°C for 30 s, and 68°C for 3 min; 20 cycles at 95°C for 30 s, 59°C for 30 s, and 68°C for 3 min plus 5 s per cycle; 1 cycle at 72°C for 7 min.
The amplified HIV-1 env fragments were gel purified with a QIAquick gel extraction kit (Qiagen). The gel-purified env fragments were cloned into the bidirectional T/A cloning and expression vector pCR3.1 (Invitrogen, Carlsbad, Calif.) and transformed into maximum-efficiency competent JM109 cells (Promega, Madison, Wis.). To determine the orientation of the env gene with respect to the cytomegalovirus (CMV) promoter, a PCR screening method was performed. Briefly, colonies were inoculated simultaneously into a PCR and onto an LB-ampicillin agar plate. Amplification was carried out with Taq DNA polymerase (Eppendorf, Westbury, N.Y.) with primers that hybridized to the BGH reverse sequence within the pCR3.1 T/A cloning vector (5′-TAGAAGGCACAGTCCGAGGC-3′; nucleotides 813 to 831) and to the 3′ end of the HIV-1 genome (5′-AAAAGAAAAGGGGGGACTGGAAGGGCTA-3′; nucleotides 9069 to 9096 of HXB2). The PCR mixture was incubated at 95°C for 5 min, followed by 30 cycles at 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s and 1 cycle at 72°C for 7 min. Colonies that produced a 188-bp band were selected for growth and plasmid purification with a QIAprep spin miniprep kit (Qiagen). Recombinant plasmids were digested with EcoRI to confirm the presence of a full-length env gene insert. The nucleotide sequence of each of the patient-derived env genes was determined by using dye-terminator technology, and sequence contigs were constructed with Sequencher v3.1.1. Nucleotide sequences were translated by using DNA Strider v1.3f9 (22), and amino acid alignments were generated with Clustal W (28). All Env sequences were classified as subtype B by phylogenetic analysis (data not shown). Moreover, Env sequences derived from the same patient were significantly more closely related to each other than to sequences derived from other patients or reference strains (data not shown).
Construction of chimeric env genes.
For construction of chimeric env genes, the gp120 and gp41 coding regions of patient-isolated envs and NLHX were PCR amplified with Expand high-fidelity polymerase (Roche, Nutley, N.J.), using primers designed to introduce, by PCR mutagenesis, a silent MluI site 25 bp upstream of the gp120/gp41 cleavage site. The gp120 coding region (6310 to 7757 in HXB2) was amplified by using envA and a degenerate 3′ primer, Mlu I REV (5′-GCAAAGAGACGCGTGGTGCAGAGAGAAAA-3′; nucleotides 7725 to 7753 in HXB2 [the MluI site is underlined]). The gp41 coding region (7758 to 8789 in HXB2) was amplified by using a degenerate 5′ primer, Mlu I FOR (5′-AAGGCAAAGAGACGCGTGGT-3′; nucleotides 7722 to 7741 in HXB2), and envN. Both amplifications were performed as described above for full-length Envs except that the annealing temperature was 63°C and the extension time was 2 min. The PCR products were gel purified and cloned into the pCR3.1 T/A cloning vector. Clones containing inserts in the correct orientation for expression from the CMV promoter were digested with BssHII (located at nucleotide 1, upstream of the CMV promoter) and MluI. The 1.5-kb BssHII-MluI fragment containing the gp120 coding region was ligated into the BssHII and MluI (New England Biolabs, Beverly, Mass.)- linearized pCR-gp41-containing plasmid. DNA clones were partially sequenced in the gp120 coding region (6590 to 7200 of HXB2) and the gp41 coding region (7936 to 8260 of HXB2) to confirm the construction of plasmids containing chimeric env genes.
Drug susceptibility of patient isolates and pseudotyped virus.
The enfuvirtide and T-649 susceptibilities of patient isolates and viruses pseudotyped with proteins encoded by individual env genes were analyzed by the following method. For all patient isolates except R11, four viral populations (the original virus isolate from patient PBMC coculture, a regrown stock of the original virus isolate, and cell-free supernatant from infected PBMC cultures [collected at 5 and 14 days postinfection]) were compared to those of viruses pseudotyped with Env glycoproteins from at least five different cloned env genes. In the case of R11, the original virus isolate stock had been exhausted. Pseudotyped virus stocks were generated via transfection using Fugene (Roche). One microgram of the recombinant Env expression vector was cotransfected with 2 μg of pSG3Δenv proviral DNA into cultures of 293T cells. Culture supernatants were collected at 3 days posttransfection, aliquoted, and frozen at −70°C. The titers of the viral stocks and their susceptibility to PFI were determined by infecting TZM-bl (formerly known as JC53-BL13) HIV-1 indicator cells and measuring production of β-galactosidase or luciferase as described previously (6, 7, 29).
Relative infectivity (percent of control) was calculated by dividing the number of luciferase units at each drug concentration by the number of luciferase units in the absence of drug. Virus infectivity curves were generated by plotting the relative infectivity on the y axis against drug concentration on the x axis with Microsoft Excel vX.
Peptide synthesis and purification.
Peptides corresponding to amino acids 628 to 661 and 638 to 674 of the HXB2 gp160 coding region were synthesized with succinylated N termini and amidated C termini (SynPep, Dublin, Calif.). Crude peptides were purified to homogeneity by reversed-phase high-performance liquid chromatography using a Vyadac C18 preparative column and a linear gradient of acetonitrile containing 0.1% trifluoroacetic acid by Min Lu (Cornell University, New York, N.Y.). The lyophilized stocks were suspended in phosphate-buffered saline (pH 8.0) to a concentration of 1 mg/ml. The purity and mass of the peptides were verified with a home-built 9.4T FT-ICR mass spectrometer, and the concentration was verified by measuring the absorbance at 280 nm with a DU640 spectrophotometer (Beckman Coulter, Fullerton, Calif.).
Statistical analyses.
To determine the 50% inhibitory concentration (IC50), a linear quadratic model was constructed for each of the virus types analyzed. It was observed that the log2 IC50s were less skewed and therefore closer to the normal distribution than the values calculated by other methods. The standard two-sample t test and Wilcoxon-Rank sum analyses were applied to examine if the differences in the means of log2 IC50s for the PFI were statistically significant between reference viruses and experimental viruses. The t test and Wilcoxon-Rank sum analysis were performed once for each virus-drug combination; P values of <0.05 were considered statistically significant. The standard two-sample t test and the Mann-Whitney analyses were applied to determine if the potency differences in the mean IC50s (calculated with Microsoft Excel vX) between PFI and patient-derived PFI were statistically significant. The t test and Mann-Whitney analyses were performed once for each virus-drug combination; P values of <0.05 were considered statistically significant (calculated with InStat 3.0a).
RESULTS
Env encodes the major determinants of susceptibility to PFI.
In a previous study, the susceptibility of 29 CCR5-utilizing (R5) and 26 CXCR4-utilizing or dual tropic (collectively represented as X4) PFI-naïve patient isolates to enfuvirtide and T-649 was determined (7). The mean IC50s of enfuvirtide were 0.2 ± 0.18 and 0.1 ± 0.15 μg/ml for the R5 and X4 viruses, respectively. The mean IC50s of T-649 were 0.1 ± 0.06 and 0.07 ± 0.05 μg/ml for R5 and X4 viruses, respectively. Five of the viruses (three R5 and two X4) were markedly less susceptible to PFI than the rest of the panel. The virus isolates were 5 to 18 times less susceptible to enfuvirtide and 3 to 33 times less susceptible to T-649 relative to the mean IC50 for the corresponding panel (7). In the present study, we investigated the molecular basis for this insusceptibility by PCR amplifying and cloning individual env genes from the viral quasispecies of five insusceptible patient isolates (R11, R14, R16, X10, and X23) and, for comparison, a PFI-susceptible virus (R21).
In order to confirm that the amplified env gene of the insusceptible viral isolates defined the drug susceptibility phenotype, the PFI susceptibility of each primary isolate quasispecies was compared to that of viruses pseudotyped with glycoproteins encoded by cloned autologous env genes in a single-round drug sensitivity assay. The average IC50 for each group of pseudotyped viruses (representing a minimum of five cloned env genes) was not significantly different from that for the virus population from which they were derived in four of six cases (Fig. 1). In cases where a significant difference was observed, the pseudotyped viruses were only threefold more susceptible to enfuvirtide (R11) and twofold more susceptible to T-649 (R16). We therefore concluded that the cloned HIV-1 env genes were representative of the original viral quasispecies from which they were derived in the majority of cases and that the Env protein of these isolates defines the major determinants of PFI susceptibility to enfuvirtide and T-649.
FIG. 1.
PFI susceptibility of patient isolates and Env pseudotypes. The mean IC50 for the virus isolates includes the original patient isolate, a regrown stock, and infected PBMC culture supernatants from days 5 and 14 postinfection. The mean IC50 for the Env pseudotypes includes between 6 and 10 individual Envs produced by PCR amplification of genomic DNA from day 5 of a PBMC infection. Error bars indicate the standard error of the mean for each bar. All experiments were performed in triplicate in at least two independent experiments.
Comparison of patient-derived Env pseudotypes to the PFI-susceptible strain NLHX.
As a reference point for PFI susceptibility, we compared the patient-derived Env pseudotypes to viruses pseudotyped with Env molecules from NLHX, which is PFI susceptible, and the patient isolate Env molecules from enfuvirtide-susceptible R21 (Fig. 2). The NLHX Env contains sequences from the CXCR4-specific lab strains NL4-3 and HXB2, but the gp41 ectodomain was derived from HXB2 (11). In contrast, the NL4-3 Env protein contains a G-to-D substitution at position 36 in HR-1 and was thus used as a reference for the effect of changes within the GIV domain. Env pseudotypes derived from all five insusceptible patient isolates (R11, R14, R16, X10, and X23) were significantly less susceptible to enfuvirtide relative to NLHX using either a t test or Wilcoxon-Rank sum analysis (P < 0.00001). The IC50s ranged from 0.7 to 1.4 μg/ml, compared to 0.03 μg/ml for NLHX. Relative to the Env pseudotypes derived from fusion inhibitor-susceptible patient isolate R21, the Env pseudotypes derived from the fusion inhibitor-insusceptible viruses (R11, R14, R16, X10, and X23) were also significantly less susceptible to enfuvirtide according to either a t test or Wilcoxon-Rank sum analysis (P < 0.00001). The enfuvirtide IC50 values were 10- to 18.7-fold higher than those for R21. This result suggested that a change in the GIV codons might be present in these variants.
FIG. 2.
PFI susceptibility of patient-derived Envs relative to NLHX. The standard errors of the means ranged from 0.04 to 0.15 μg/ml for enfuvirtide and 0.12 to 0.24 μg/ml for T-649. The points on each curve represent the mean infectivity of between 6 and 10 individual Env pseudotypes at each drug concentration. Patient-derived Env pseudotypes are represented by dashed lines, while the Env pseudotypes derived from the lab strains NL4-3 and NLHX are represented by solid lines and filled symbols.
R11, R14, R16, R21, X10, and X23 Envs were also significantly less susceptible to T-649 than NLHX as determined by either a t test or Wilcoxon-Rank sum analysis (P < 0.00001). The IC50s ranged from 0.2 to 1.9 μg/ml, compared to 0.03 μg/ml for NLHX. Relative to the Env pseudotypes derived from fusion inhibitor-susceptible patient isolate R21, the Env pseudotypes derived from four of the fusion inhibitor-insusceptible viruses (R11, R14, R16, and X23) were significantly less susceptible to T-649 according to a t test and Wilcoxon-Rank sum analysis (P < 0.00001). Thus, we observed significantly lower susceptibility to PFI for the patient-derived Envs than for NLHX, whose HR-2 sequences formed the basis for the design of both enfuvirtide and T-649.
Conservation of enfuvirtide-susceptible residues in gp41.
Previous studies have demonstrated that single amino acid changes in gp41 (at positions 36 to 38) reduce virus susceptibility to enfuvirtide more than 10-fold (6, 7, 26, 29). We therefore targeted gp41 for sequence analysis to determine if the Env proteins with reduced susceptibility to enfuvirtide and/or T-649 had changes encompassing the critical GIV sequence. Interestingly, none of the gp41 sequences from the PFI-insusceptible Envs contained substitutions in the GIV motif (Fig. 3), although all X10-derived sequences contained an L-to-M variation at position 45. Previous reports have shown that this L45 M substitution generated in NL4-3, in combination with an additional change at position 40 or 43, was associated with a 300-fold decrease in enfuvirtide susceptibility relative to the baseline IC50 (Mink et al., XI Int. HIV Drug Resist. Workshop). All R21-derived Env sequences contained an N-to-S substitution at position 42, which has been associated with an enfuvirtide-hypersensitive phenotype in some studies (J. M. Whitcomb, W. Huang, S. Fransen, T. Wrin, E. Paxinos, J. Toma, M. Greenberg, P. Sista, T. Melby, T. Matthews, R. DeMasi, G. Heilek-Snyder, N. Cammack, N. Hellmann, and C. J. Petropoulos, presented at the 10th Conference on Retroviruses and Opportunistic Infections, Boston, Mass., 9 to 14 February 2003). These residues in HR-1 within the extended resistance region (residues 36 to 45) may contribute to the >10-fold difference in susceptibility between enfuvirtide-insusceptible X10 and enfuvirtide-hypersensitive R21. None of the remaining Env sequences (R11, R14, R16, and X23) contained changes in HR-1 that could be readily associated with enfuvirtide susceptibility. Since T-649 has not been evaluated in human clinical trials due to poor pharmacokinetics in animal models, resistance mapping has been limited to studies of peptide binding (9). Using the structural model proposed by Chan et al., we inspected the X23 sequence for variation in the C-terminal residues of HR-1, which compose a hydrophobic pocket, and the HR-2 residues that would be predicted to interact with the pocket (3). None of these residues differed from those in HXB2 (NLHX contains the identical sequence). In contrast to the high level of sequence conservation in the HR-1 region in these natural isolates relative to NLHX, the N-terminal half of the HR-2, in particular, exhibited multiple nonconservative amino acid substitutions, a finding that is reflected in the sequence compendium (18) (Fig. 3).
FIG. 3.
Amino acid sequence of the gp41 ectodomain of patient-derived Envs. A predicted amino acid alignment for the ectodomain of gp41 was generated for each virus isolate by using a consensus sequence representing the majority sequence for each set of patient-derived Envs. Parentheses indicate the number of individual Env clones that were sequenced and included in the consensus. The sequences of CONSENSUS_B (www.hiv.lanl.gov, 2002 HIV and SIV alignments) NL4-3, and NLHX are included for reference. The amino acid sequence of each patient-derived consensus sequence was analyzed for variation relative to NLHX. A dot indicates that the residue was conserved with NLHX, while the residues that are shown differed from NLHX in more than 50% of the patient-derived gp41 sequences. The amino acid sequence of each patient derived consensus sequence was also analyzed for variation relative to CONSENSUS_B. Letters in gray indicate residues that are conserved relative to the subtype B consensus but differ from NLHX. For consensus B, the residues in lowercase vary in subtype B viruses and the residues in uppercase are invariant in subtype B viruses. The heptad repeats (HR-1 and HR-2) are shown in bold, and the amino acids that compose the fusion peptide are underlined. The position of residues in HR-2 that corresponds to enfuvirtide and T-649 are indicated by dashed lines above the alignment. The numbering system is based on HXB2 gp41.
The gp41 coding region contains the major determinants of susceptibility to PFI.
Since we could not attribute the reduced PFI susceptibility to substitutions in the GIV motif, and other substitutions in HR-1 were limited, we investigated whether gp41 alone was sufficient to reproduce the insusceptible phenotypes. To do this, we created chimeric env genes that contained the gp41 coding region from either NLHX or one of the patient-derived Env clones, and pseudotyped these Env proteins onto the HIV-1 SG3Δenv proviral backbone. In all cases, the gp41 sequences were the major determinant of susceptibility to both PFI, and in four of the five cases (R11, R16, X10, and X23), the patient-derived gp41 region alone was sufficient to reproduce the reduced susceptibility phenotype (Fig. 4). As expected, the R21-derived gp41 region was sufficient to reproduce the PFI susceptible phenotype (data not shown). In contrast, the full insusceptibility to PFI of R14 pseudotyped viruses required the autologous, patient-derived gp120 and gp41 subunits (Fig. 4). Previous studies have shown that gp120 interactions with coreceptor modulate susceptibility to fusion inhibitors independently of gp41 (6, 7, 25). Our present data are consistent with an effect of gp120 interactions with coreceptor, since in most chimeras replacement of CCR5-binding gp120 with a CXCR4-binding gp120 resulted in a small increase in susceptibility, while the reverse substitution resulted in a more insusceptible Env. Pseudotyped viruses containing the chimeric Env combination gp120/X23-gp41/NLHX could not be produced at a sufficiently high titer to be analyzed in these studies.
FIG. 4.
PFI susceptibility of gp120-gp41 chimeric Env pseudotyped viruses. Log10 inhibitor concentration is shown on the x axis. Four types of Env pseudotypes were analyzed for each virus: the patient-derived Env, the NLHX Env, a patient-derived gp120 with the NLHX gp41 (PD gp120-NL gp41), and the NLHX gp120 with a patient-derived gp41 (NL gp120-PD gp41). Pseudotyped viruses containing the chimeric Env combination gp120/X23-gp41/NLHX could not be produced at a sufficiently high titer to be analyzed in these studies.
Patient-derived PFI are not more potent than HXB2-derived PFI.
The sequence divergence in HR-2 and the adjacent tryptophan-rich region of the insusceptible patient isolates relative to HXB2 suggested that naturally occurring variations could make the endogenous HR-2 more competitive than the HXB2-derived PFI for HR-1 binding. To test this, we synthesized a set of autologous peptides that contained each patient-derived HR-2 sequence in place of the authentic enfuvirtide and T-649 sequences. These peptides were compared to enfuvirtide and T-649 for their ability to inhibit infection by viruses pseudotyped with the autologous Env (Table 1) and heterologous patient Envs, NLHX, and NL4.3 (data not shown). The patient-derived peptides were at most 1.3 times more potent than the HXB2-derived peptides against their autologous Envs, and in five cases they were less potent (Table 1). However, only the X23-derived peptide was significantly different (7 times less potent) than the HXB2-based enfuvirtide against viruses pseudotyped with X23 Envs according to both a t test and Mann-Whitney analysis (P = 0.006 and 0.008, respectively). Similar findings were observed when the patient-derived peptides were evaluated against the viruses pseudotyped with heterologous patient Envs, NL4-3, and NLHX, with the X23-based peptide once again demonstrating 3- to 13-times-lower potency (data not shown). Interestingly, the X23-based peptide contains a valine in the first position rather than the more commonly found tyrosine. Previous studies have demonstrated that this residue is critical for enfuvirtide activity (31). These data are not consistent with a model in which higher affinity of the patient-derived HR-2 sequences for the endogenous HR-1 is responsible for baseline insusceptibility to PFI and suggest that the N-terminal tyrosine of enfuvirtide is critical for its inhibitor activity.
TABLE 1.
Comparison of the susceptibilities of Individual Envs to PFI versus patient-derived PFI
Enva | Mean IC50 (μg/ml)b of:
|
Fold differencec | Mean IC50 (μg/ml)b of:
|
Fold differencec | ||
---|---|---|---|---|---|---|
Enfuvirtide | Autologous enfuvirtide | T649 | Autologous T649 | |||
R21 | 0.04 | NDd | 0.03 | 0.03 | 1.0 | |
R16 | 0.23 | 0.35 | −1.5 | 0.09 | 0.09 | 1.0 |
R14 | 0.31 | 0.29 | 1.1 | 0.14 | 0.12 | 1.2 |
R11 | 0.33 | 0.71 | −2.1 | 0.06 | 0.06 | 1.0 |
X23 | 0.18 | 1.23 | −6.8 | 0.32 | 0.70 | −2.2 |
X10 | 0.87 | 1.95 | −2.2 | 0.08 | 0.06 | 1.3 |
Viruses pseudotyped with glycoproteins from individual Env clones were tested for sensitivity to PFI (enfuvirtide or T-649) or PFI synthesized with patient-derived (autologous) sequences.
Calculated from at least two replicate experiments performed in triplicate.
The difference between autologous PFI and PFI.
ND, not determined.
DISCUSSION
In the present study, we performed both amino acid sequence analysis and phenotypic characterization of individual Env proteins encoded by six patient-derived viruses, five of which were insusceptible to PFI in drug sensitivity assays. These viruses were derived from a larger panel of patient isolates that are likely representative of the virus strains circulating among HIV-1-infected individuals that could receive a treatment regimen containing enfuvirtide. In all cases, the major determinants of susceptibility to either of the PFI were sequences within gp41, although in one of five cases, the autologous gp120 molecule also contributed to the insusceptible phenotype. In none of the patient isolate-derived Envs could enfuvirtide or T-649 insusceptibility be explained by variations in the GIV motif. Furthermore, for the majority of Envs analyzed, only minimal changes in HR-1 were observed relative to NLHX (HXB2). This is consistent with recent studies of viruses isolated from subjects participating in phase II/III enfuvirtide trials that compared sequences in gp41 at baseline with those following development of enfuvirtide resistance. At the time of virologic failure, 20% of the patients harbored viruses lacking substitutions within the putative enfuvirtide interaction site (amino acids 36 to 45) in HR-1 (24; Mink et al., XI Int. HIV Drug Resist. Workshop; Sista et al., XI Int. HIV Drug Resist. Workshop). In the panel of viruses described here, which were isolated prior to the development of PFI, only Envs derived from the most enfuvirtide-insusceptible patient isolate (X10) contained polymorphisms in the HR-1 region (relative to NLHX). The L45 M variation present in the HR-1 sequences from X10 has been associated with the development of clinical resistance but is generally found in combination with additional changes at residues 40 and 43 (Mink et al., XI Int. HIV Drug Resist. Workshop). Moreover, the reduced susceptibility of Env pseudotypes from the remaining patient isolates clearly cannot be explained by changes in this region. In preliminary studies, the L45 M mutation in X10 appeared to contribute only a three- to fourfold resistance to enfuvirtide (data not shown). The R11, R14, and X23 Env sequences do contain a methionine residue at position 54 instead of the leucine found in NLHX, and this residue is located in a region of HR-1 that is part of the proposed enfuvirtide and T-649 binding sites. This L45 M variation is also observed in all R21 Env clones, yet these clones are quite susceptible to PFI. Furthermore, when an L54 M substitution was introduced into NL4-3 in combination with H53R (as is observed in the JRFL isolate of HIV), no change in enfuvirtide or T-649 susceptibility was observed (C. A. Derdeyn, unpublished observations). Additionally, when a panel of 29 inhibitor-naïve viruses was examined for susceptibility to fusion inhibitors, the IC50 for patient isolates that had a methionine at position 54 was shown to be equivalent to the average value (200 ng/ml) for this population (20). Therefore, this residue is unlikely to be sufficient to confer insusceptibility to either inhibitor.
Residues that influence susceptibility to T-649 have not been defined. Structural reconstructions of the T-649 peptide interacting with its target site in HR-1 have revealed a hydrophobic cavity at the base of HR-1 into which the following residues of HR-2 dock (W117, W120, and I124) (3, 21, 30). Inspection of the gp41 sequences from R11, R14, R16, and X23 revealed that the HR-1 hydrophobic pocket and the docking residues in HR-2 are conserved relative to NLHX in all of these viruses, even though the X23 Env pseudotypes are 25- to 50-fold less susceptible to T-649 than NLHX. These data thus argue that residues outside HR-1 influence baseline susceptibility to PFI, and therefore, disruption of the PFI-binding site is not the reason for the decreased baseline susceptibility.
Inspection of the Env amino acid sequences showed that there were a number of substitutions in the HR-2 and tryptophan-rich regions within the insusceptible Envs that could influence helicity of HR-2 (4, 13, 14) and thus alter the affinity of HR-2 for HR-1 (2, 21, 30). However, the patient-derived peptides that contained these substitutions were not more potent than the HXB2-based PFI against viruses pseudotyped with autologous and heterologous Envs (Table 1 and data not shown). In fact, the X23-derived T-649 was twofold less potent and the peptide based on enfuvirtide was 7 to 10 times less potent than the HXB-derived peptides, even though viruses pseudotyped with the X23 Env are significantly less susceptible to T-649 and enfuvirtide than NLHX. That the patient-derived PFI do not compete better than the HXB2-based PFI for HR-1 suggests that the molecular mechanism by which PFI are excluded from binding these insusceptible Envs involves other thermodynamic determinants, not merely HR-2 affinity for HR-1. This model is consistent with biochemical studies of small-molecule inhibitors in which HR-2 competition for inhibitor binding was not observed (5, 27). Reduced susceptibility to multiple entry inhibitors has been demonstrated for viruses that have rapid entry kinetics driven by a high affinity for the coreceptor (25). Thus, we propose a model in which the kinetics of six-helix bundle formation can be modulated by sequences in both gp120 and gp41, thereby influencing the window of opportunity for PFI to interact with the viral target in HR-1. Specifically, we hypothesize that the amino acid changes in the HR-2 region of gp41 influence the transition time from the PFI-susceptible prehairpin intermediate to the PFI-insusceptible six-helix bundle.
The baseline susceptibility to PFI of the patient-derived viruses we described previously ranged over 1.5 log10 (7). Other groups have reported an even broader range of responsiveness (20; M. L. Greenberg, T. Melby, P. Sista, R. DeMasi, N. Cammack, M. Salgo, J. Whitcomb, C. Petropoulos, and T. J. Matthews, presented at the 10th Conference on Retroviruses and Opportunistic Infections, Boston, Mass., 9 to 14 February 2003). This is an order of magnitude greater than the two- to fourfold variation in baseline susceptibility that is generally observed for reverse transcriptase and protease inhibitors (23) and may reflect the involvement of an array of viral and host factors in the fusion process (25). Indeed, the efficacy of PFI can vary widely following infection with different virus strains or use of primary cells from different individuals (7, 15, 16). Despite the broad variability, preliminary analyses of phase III clinical trials indicate that the magnitude of the decrease in viral load with initiation of enfuvirtide therapy was similar regardless of baseline susceptibility (Greenberg et al., 10th Conf. Retrovir. Opportun. Infect.). Based on these observations, we would expect that replication of the insusceptible viruses analyzed here would be suppressed during enfuvirtide therapy. Nevertheless, we would also predict that the higher baseline IC50 observed here would translate into greater residual replication of virus and therefore a shorter time to resistance, especially if plasma drug concentrations were allowed to fall below the average trough level (1 μg/ml) observed in patients treated with enfuvirtide.
Data from human trials suggest that the predominant mechanism of resistance involves disrupting the PFI target on HR-1 (29; Sista et al., XI Int. HIV Drug Resist. Workshop). However, the data presented here argue that some patient viruses may carry pre-existing mutations that will facilitate the development of drug resistance without requiring changes in the GIV sequence. The lack of natural variants in the GIV motif in all isolates sequence to date is consistent with changes in the motif being deleterious to viral fitness (18). Thus, the ability to use a GIV-independent pathway, such as exclusion of PFI by a kinetic mechanism, could facilitate development of more complete enfuvirtide resistance and preserve replication fitness during exposure to drug.
Acknowledgments
We thank Maria Salazar of the UAB Center for AIDS Research DNA Sequence Analysis Core, supported by Core grant P30-AI-27767, and Karen Zscheck of Lone Star Labs (Houston, Tex.) for excellent technical assistance with DNA sequencing; Dung-Tsa Chen, Likang Xu, and Jeannette Lee of the UAB Biostatistics and Bioinformatics Unit for statistical analysis; and Min Lu and Christopher Petropolous for helpful scientific discussions. For synthesis and verification of autologous peptides, we thank and acknowledge Min Lu, Peter Prevelige, and Jason Lanman.
This work was supported by amfAR grant 02804-30-RGT (C.A.D.), NIH R01-AI-33319 (E.H.), and the UAB CFAR Biostatistics and Central Virus Culture Cores supported by Core grant P30-AI-27767.
REFERENCES
- 1.Bernstein, H. B., S. P. Tucker, S. R. Kar, S. A. McPherson, D. T. McPherson, J. W. Dubay, J. Lebowitz, R. W. Compans, and E. Hunter. 1995. Oligomerization of the hydrophobic heptad repeat of gp41. J. Virol. 69:2745-2750. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Chan, D. C., C. T. Chutkowski, and P. S. Kim. 1998. Evidence that a prominent cavity in the coiled coil of HIV type 1 gp41 is an attractive drug target. Proc. Natl. Acad. Sci. USA 95:15613-15617. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Chan, D. C., D. Fass, J. M. Berger, and P. S. Kim. 1997. Core structure of gp41 from the HIV envelope glycoprotein. Cell 89:263-273. [DOI] [PubMed] [Google Scholar]
- 4.Chou, P. Y., and G. D. Fasman. 1978. Empirical predictions of protein conformation. Annu. Rev. Biochem. 47:251-276. [DOI] [PubMed] [Google Scholar]
- 5.Cole, J. L., and V. M. Garsky. 2001. Thermodynamics of peptide inhibitor binding to HIV-1 gp41. Biochemistry 40:5633-5641. [DOI] [PubMed] [Google Scholar]
- 6.Derdeyn, C. A., J. M. Decker, J. N. Sfakianos, X. Wu, W. A. O'Brien, L. Ratner, J. C. Kappes, G. M. Shaw, and E. Hunter. 2000. Sensitivity of human immunodeficiency virus type 1 to the fusion inhibitor T-20 is modulated by coreceptor specificity defined by the V3 loop of gp120. J. Virol. 74:8358-8367. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Derdeyn, C. A., J. M. Decker, J. N. Sfakianos, Z. Zhang, W. A. O'Brien, L. Ratner, G. M. Shaw, and E. Hunter. 2001. Sensitivity of human immunodeficiency virus type 1 to fusion inhibitors targeted to the gp41 first heptad repeat involves distinct regions of gp41 and is consistently modulated by gp120 interactions with the coreceptor. J. Virol. 75:8605-8614. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gao, F., L. Yue, S. Craig, C. L. Thornton, D. L. Robertson, F. E. McCutchan, J. A. Bradac, P. M. Sharp, B. H. Hahn, et al. 1994. Genetic variation of HIV type 1 in four World Health Organization-sponsored vaccine evaluation sites: generation of functional envelope (glycoprotein 160) clones representative of sequence subtypes A, B, C, and E. AIDS Res. Hum. Retrovir. 10:1359-1368. [DOI] [PubMed] [Google Scholar]
- 9.He, Y., R. Vassell, M. Zaitseva, N. Nguyen, Z. Yang, Y. Weng, and C. D. Weiss. 2003. Peptides trap the human immunodeficiency virus type 1 envelope glycoprotein fusion intermediate at two sites. J. Virol. 77:1666-1671. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Hoffman, T. L., and R. W. Doms. 1999. HIV-1 envelope determinants for cell tropism and chemokine receptor use. Mol. Membr. Biol. 16:57-65. [DOI] [PubMed] [Google Scholar]
- 11.Hung, C. S., N. Vander Heyden, and L. Ratner. 1999. Analysis of the critical domain in the V3 loop of human immunodeficiency virus type 1 gp120 involved in CCR5 utilization. J. Virol. 73:8216-8226. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Hunter, E. 1997. Viral entry and receptors, p. 71-121. In J. M. Coffin, S. H. Hughes, and H. E. Varmus (ed.), Retroviruses. Cold Spring Harbor Laboratory Press, Plainview, N.Y. [PubMed]
- 13.Jin, B. S., J. R. Ryu, K. Ahn, and Y. G. Yu. 2000. Design of a peptide inhibitor that blocks the cell fusion mediated by glycoprotein 41 of human immunodeficiency virus type 1. AIDS Res. Hum. Retrovir. 16:1797-1804. [DOI] [PubMed] [Google Scholar]
- 14.Judice, J. K., J. Y. Tom, W. Huang, T. Wrin, J. Vennari, C. J. Petropoulos, and R. S. McDowell. 1997. Inhibition of HIV type 1 infectivity by constrained alpha-helical peptides: implications for the viral fusion mechanism. Proc. Natl. Acad. Sci. USA 94:13426-13430. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ketas, T. J., I. Frank, P. J. Klasse, B. M. Sullivan, J. P. Gardner, C. Spenlehauer, M. Nesin, W. C. Olson, J. P. Moore, and M. Pope. 2003. Human immunodeficiency virus type 1 attachment, coreceptor, and fusion inhibitors are active against both direct and trans infection of primary cells. J. Virol. 77:2762-2767. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Ketas, T. J., P. J. Klasse, C. Spenlehauer, M. Nesin, I. Frank, M. Pope, J. M. Strizki, G. R. Reyes, B. M. Baroudy, and J. P. Moore. 2003. Entry inhibitors SCH-C, RANTES, and T-20 block HIV type 1 replication in multiple cell types. AIDS Res. Hum. Retrovir. 19:177-186. [DOI] [PubMed] [Google Scholar]
- 17.Kilby, J. M., S. Hopkins, T. M. Venetta, B. DiMassimo, G. A. Cloud, J. Y. Lee, L. Alldredge, E. Hunter, D. Lambert, D. Bolognesi, T. Matthews, M. R. Johnson, M. A. Nowak, G. M. Shaw, and M. S. Saag. 1998. Potent suppression of HIV-1 replication in humans by T-20, a peptide inhibitor of gp41-mediated virus entry. Nat. Med. 4:1302-1307. [DOI] [PubMed] [Google Scholar]
- 18.Kuiken, C., B. Foley, E. Freed, B. Hahn, P. Marx, F. McCutchan, J. W. Mellors, S. Wolinksy, and B. Korber. 2002. HIV sequence compendium. Theoretical Biology and Biophysics Group T-10, Los Alamos, N.Mex.
- 19.Kwong, P. D., R. Wyatt, J. Robinson, R. W. Sweet, J. Sodroski, and W. A. Hendrickson. 1998. Structure of an HIV gp120 envelope glycoprotein in complex with the CD4 receptor and a neutralizing human antibody. Nature 393:648-659. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Labrosse, B., J. L. Labernardiere, E. Dam, V. Trouplin, K. Skrabal, F. Clavel, and F. Mammano. 2003. Baseline susceptibility of primary human immunodeficiency virus type 1 to entry inhibitors. J. Virol. 77:1610-1613. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Lu, M., S. C. Blacklow, and P. S. Kim. 1995. A trimeric structural domain of the HIV-1 transmembrane glycoprotein. Nat. Struct. Biol. 2:1075-1082. [DOI] [PubMed] [Google Scholar]
- 22.Marck, C. 1988. ‘DNA Strider’: a ‘C’ program for the fast analysis of DNA and protein sequences on the Apple Macintosh family of computers. Nucleic Acids Res. 16:1829-1836. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Parkin, N. T., N. Hellmann, J. Whitcomb, S. Nidtha, L. Kiss, C. Chappey, and C. J. Petropoulos. 2004. Natural variation of drug susceptibility in wild-type human immunodeficiency virus type 1. Antimicrob Agents Chemother. 48:437-443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Poveda, E., B. Rodes, C. Toro, L. Martin-Carbonero, J. Gonzalez-Lahoz, and V. Soriano. 2002. Evolution of the gp41 env region in HIV-infected patients receiving T-20, a fusion inhibitor. AIDS 16:1959-1961. [DOI] [PubMed] [Google Scholar]
- 25.Reeves, J. D., S. A. Gallo, N. Ahmad, J. L. Miamidian, P. E. Harvey, M. Sharron, S. Pohlmann, J. N. Sfakianos, C. A. Derdeyn, R. Blumenthal, E. Hunter, and R. W. Doms. 2002. Sensitivity of HIV-1 to entry inhibitors correlates with envelope/coreceptor affinity, receptor density, and fusion kinetics. Proc. Natl. Acad. Sci. USA 99:16249-16254. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Rimsky, L. T., D. C. Shugars, and T. J. Matthews. 1998. Determinants of human immunodeficiency virus type 1 resistance to gp41-derived inhibitory peptides. J. Virol. 72:986-993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Sia, S. K., P. A. Carr, A. G. Cochran, V. N. Malashkevich, and P. S. Kim. 2002. Short constrained peptides that inhibit HIV-1 entry. Proc. Natl. Acad. Sci. USA 99:14664-14669. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-4680. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Wei, X., J. M. Decker, H. Liu, Z. Zhang, R. B. Arani, J. M. Kilby, M. S. Saag, X. Wu, G. M. Shaw, and J. C. Kappes. 2002. Emergence of resistant human immunodeficiency virus type 1 in patients receiving fusion inhibitor (T-20) monotherapy. Antimicrob. Agents Chemother. 46:1896-1905. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Weissenhorn, W., A. Dessen, S. C. Harrison, J. J. Skehel, and D. C. Wiley. 1997. Atomic structure of the ectodomain from HIV-1 gp41. Nature 387:426-430. [DOI] [PubMed] [Google Scholar]
- 31.Wild, C., J. W. Dubay, T. Greenwell, T. Baird, Jr., T. G. Oas, C. McDanal, E. Hunter, and T. Matthews. 1994. Propensity for a leucine zipper-like domain of human immunodeficiency virus type 1 gp41 to form oligomers correlates with a role in virus-induced fusion rather than assembly of the glycoprotein complex. Proc. Natl. Acad. Sci. USA 91:12676-12680. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Wild, C., T. Oas, C. McDanal, D. Bolognesi, and T. Matthews. 1992. A synthetic peptide inhibitor of human immunodeficiency virus replication: correlation between solution structure and viral inhibition. Proc. Natl. Acad. Sci. USA 89:10537-10541. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Xu, L., S. Hue, S. Taylor, D. Ratcliffe, J. A. Workman, S. Jackson, P. A. Cane, and D. Pillay. 2002. Minimal variation in T-20 binding domain of different HIV-1 subtypes from antiretroviral-naive and -experienced patients. AIDS 16:1684-1686. [DOI] [PubMed] [Google Scholar]