|Home | About | Journals | Submit | Contact Us | Français|
We previously reported that a human immunodeficiency virus type 1 (HIV-1) clade B envelope protein with a severely truncated V3 loop regained function after passage in tissue culture. The adapted virus, termed TA1, retained the V3 truncation, was exquisitely sensitive to neutralization by the CD4 binding site monoclonal antibody b12 and by HIV-positive human sera, used CCR5 to enter cells, and was completely resistant to small molecule CCR5 antagonists. To examine the mechanistic basis for these properties, we singly and in combination introduced each of the 5 mutations from the adapted clone TA1 into the unadapted envelope. We found that single amino acid changes in the C3 region, the V3 loop, and in the fusion peptide were responsible for imparting near-normal levels of envelope function to TA1. T342A, which resulted in the loss of a highly conserved glycosylation site in C3, played the primary role. The adaptive amino acid changes had no impact on CCR5 antagonist resistance but made virus more sensitive to neutralization by antibodies to the CD4 binding site, modestly enhanced affinity for CD4, and made TA1 more responsive to CD4 binding. Specifically, TA1 was triggered by soluble CD4 more readily than the parental Env and, unlike the parental Env, could mediate entry on cells that express low levels of CD4. In contrast, TA1 interacted with CCR5 less efficiently and was highly sensitive to antibodies that bind to the CCR5 N terminus and ECL2. Therefore, enhanced utilization of CD4 is one mechanism by which HIV-1 can overcome mutations in the V3 region that negatively affect CCR5 interactions.
The human immunodeficiency virus type 1 (HIV-1) envelope protein (Env) mediates sequential binding to CD4 and a coreceptor, with these interactions triggering conformational changes in Env that result in fusion between the viral and cellular membranes (2, 12, 66). The V3 loop in the gp120 subunit of the Env protein is thought to interact with the extracellular loops (ECLs) of the seven-transmembrane domain HIV-1 coreceptors, CCR5 and CXCR4 (9, 10, 28, 45, 51), while the base of the V3 loop and the bridging sheet region of gp120 are thought to engage the amino-terminal domains of the coreceptors (23). In addition, the V3 loop plays a major role in determining whether a given virus strain utilizes CCR5, CXCR4, or both coreceptors subsequent to CD4 binding (6, 7, 57). Perhaps because of its role in coreceptor engagement, the overall length of the V3 loop is highly conserved, as are specific residues that may play key roles in receptor binding (11, 33, 70). However, the V3 loop is also a target for neutralizing antibodies, making it subject to immune selection (20, 25, 26, 44, 47). In addition, the V3 loop as well as the highly variable V1/V2 region shield more conserved regions of Env that are also involved in receptor binding (16, 20, 33, 58, 59).
The importance of the V3 loop for Env function is shown by the fact that genetic deletion of residues in V3 typically results in a nonfunctional Env protein (5, 19, 67). While V3 loop-deleted Envs appear to fold normally and retain the ability to bind CD4, coreceptor interactions are apparently lost (5, 19, 27, 54, 65, 67, 69). This loss of function complicates immunogen design approaches that are predicated upon removing variable loops in gp120 in the hopes of focusing the humoral immune response on more conserved regions of Env (22). To overcome this limitation, we introduced partial V3-loop truncations into a series of HIV-1 Env proteins and identified an R5X4 HIV-1 Env, termed R3A, that could tolerate partial loss of its V3 loop (31). When 15 residues were removed from the center of the V3 loop, leaving the first 9 and last 9 residues of the region intact, the resulting virus [termed V3(9,9)] was poorly functional. However, after passage in vitro, function was enhanced via the acquisition of five mutations in the env gene. An Env cloned from the tissue culture-adapted virus, termed TA1, used CCR5 to infect cells but lost the ability to use CXCR4, was completely resistant to CCR5 antagonists by being able to recognize the drug-bound conformation of the coreceptor, and was exquisitely sensitive to neutralization by HIV-1-positive human sera and by a broadly neutralizing antibody to the CD4 binding site (31). Whether these attributes were due to the V3 loop truncation, the adaptive mutations, or some combination of the two was unclear. In addition, it is not apparent how an Env can function efficiently despite the loss of a domain that plays an important role in coreceptor engagement.
In the present study, we investigated the roles played by the adaptive mutations in TA1 function. We found that the V3 loop truncation alone accounted for resistance to CCR5 antagonists. A subset of the adaptive mutations played a major role in restoring function to V3(9,9), doing so via improved utilization of CD4. Compared to the parental R3A Env, TA1 bound to CD4 with slightly higher affinity, was more easily induced to cause membrane fusion by incubation with soluble CD4 (sCD4), and was able to mediate infection of cells expressing low levels of CD4. In contrast, utilization of CCR5 was less efficient. Thus, in this instance, loss of a region of the V3 loop that is important for coreceptor interactions was compensated for, unexpectedly, not by restoring efficient use of CCR5 but rather by enhanced use of CD4. Improved interactions with CD4 might make induction and exposure of the coreceptor binding site in the bridging sheet region of gp120 more efficient, enabling Env to overcome its impaired ability to interact with CCR5.
QT6 Japanese quail fibrosarcoma and 293T/17 human embryonic kidney cells were cultured in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum, 60 μg of penicillin/ml, and 100 μg of streptomycin/ml (complete DMEM). Complete DMEM was supplemented with 2 mM l-glutamine for QT6 cells. HEK293-derived Affinofile cells (34, 49) were cultured in complete DMEM supplemented with 50 μg of blasticidin/ml and 200 μg of G418/ml. Human glioma NP2.CD4.CCR5 cells were cultured in complete DMEM supplemented with 1 mg of G418/ml and 1 μg of puromycin/ml. SupT1.CCR5 and SupT1.CCR5.DCSIGNR cells were cultured in RPMI 1640 supplemented with 10% fetal bovine serum, 60 μg of penicillin/ml, and 100 μg of streptomycin/ml.
All viral env genes used for production of viral pseudotypes, including 400 nucleotides upstream of the env initiation codon with the complete coding sequences of tat, rev, and vpu, were cloned into the EcoRI and XhoI sites of pHSPG (41). Overlap PCR was used to construct Envs containing single or multiple adaptive mutations. The dualtropic Env R3A (42) allows entry into both macrophages and T cells, as well as into both CXCR4- and CCR5-expressing cell lines. V3(9,9) and TA1 have been described previously in detail (31). The reporter plasmid encoding luciferase under the control of a T7 promoter (pT7.luciferase) used in cell-cell fusion assays has been described previously (52). Replication-competent molecular clones were constructed by subcloning the EcoRI-XhoI fragment from pHSPG into pNL, which contains the proviral sequence of NL4-3. All viral genes are intact, with the exception of a small deletion in nef.
This assay has been previously described (52). Briefly, “target” QT6 cells cotransfected with expression plasmids encoding a luciferase reporter gene (pGEM2 T7-luc; Promega), CD4, and either CCR5 or a control expression plasmid, were incubated with QT6 “effector” cells transfected with Env expression plasmids (1). Effector cells were added to target cells ~18 h posttransfection, and the cells were allowed to interact at 37°C for 7 to 8 h. The cells were lysed, and the luciferase activity was measured with a luminometer (3). To determine the ability of sCD4 to induce cell-cell fusion, serial dilutions of four-domain sCD4 (IAVI; Neutralizing Antibody Consortium repository) were added to CD4-negative, CCR5-positive target cells immediately prior to the addition of Env-expressing effector cells.
Luciferase reporter viral pseudotypes were produced in 293T/17 cells by cotransfection of the NL4-3-based luciferase vector, pNL-luc (Env− Vpr+), with plasmids expressing the desired Env protein, as previously described (4, 8). Replication-competent viruses were produced in 293T/17 cells by transfection of a NL4-3-based proviral vector with the HIV env cloned into the XhoI and EcoRI sites. The virus titer was normalized by p24 content (Cell BioLabs, Inc., San Diego, CA). For neutralization sensitivity assays, 5 ng of virus was incubated with serial dilutions of monoclonal antibody (MAb) or sCD4 for 1 h at 37°C and then spinoculated onto SupT1.CCR5 cells (46). The cells were assayed for luciferase expression 3 days postinfection. MAbs 17b, immunoglobulin G (IgG) b12, and sCD4 were obtained from the IAVI Neutralizing Antibody Consortium repository. MAb 19 binds to CD4 and blocks HIV infection (17). For neutralization assays with MAbs 45531 and 45529 (R&D Systems, Minneapolis, MN) and 2D7 (BD Biosciences, San Jose, CA), which bind CCR5 ECL2, the target cells were NP2.CD4.CCR5. For drug inhibition assays, the CCR5 antagonist maraviroc was incubated with SupT1.CCR5 cells in the presence of 100 nM AMD3100 (NIH AIDS Research and Reference Reagent Program) for 1 h at 37°C before addition of 5 ng of virus, followed by spinoculation. Cells were incubated with virus and drug for 72 h and subsequently analyzed for luciferase expression.
For kinetic assays with CD4, CCR5, and fusion blockers, NP2.CD4.CCR5 cells were plated in a 96-well format at 1.5 × 104 cells per well. The following day, a 1:10 dilution of MAb 19 ascites, MAbs 2D7 and CTC5 (R&D Systems) at 50 μg/ml each, or enfuvirtide at 90 μg/ml was added to pretreatment wells and incubated for 1 h before 5 ng of virus/well was spinoculated onto cells for 2 h at 4°C. For all other wells, virus was first spinoculated onto cells at 4°C, followed by transfer to 37°C, with blocking agents added immediately (T = 0 min) or at 15, 30, 60, 120, or 240 min after spinoculation. Infected cells were incubated for 72 h at 37°C and then assayed for luciferase activity. Infection levels were normalized against control well infections performed in the absence of antibodies or drug.
To monitor growth of replication-competent virus, SupT1.CCR5.DCSIGNR cells were infected with 5 or 50 ng of replication-competent virus, and the reverse transcriptase activity measured by using a [3H]thymidine incorporation assay on culture supernatants pelleted by ultracentrifugation (32).
Infection of the HEK293-based CD4/CCR5 dual-inducible cell line (293-Affinofile) with luciferase reporter viral pseudotypes was used to determine the CD4 and CCR5 usage efficiencies as previously described (34, 49). Please see accompanying report (27a) for full details. Briefly, 96-well plates were seeded with 104 inducible cells per well 2 days prior to CD4 and CCR5 induction. Cells were induced in a 6×8 matrix format using twofold serial dilutions from 0.156 to 5 ng of minocycline/ml (to induce CD4) and 0.0156 to 2 μM ponasterone A (to induce CCR5) and allowed to incubate for 18 h at 37°C. Induction medium was removed, and the cells were then infected with viral pseudotypes normalized by p24 content. For dualtropic virus infection, the induced cells were treated with 10 nM AMD3100 at 37°C for 1 h prior to infection to inhibit use of endogenous CXCR4 on 293-Affinofile cells. Infection was measured by luciferase expression 3 days postinfection.
The CD4 and CCR5 surface expression levels of induced cells were analyzed by flow cytometry using phycoerythrin-conjugated mouse anti-human CD4 antibody (Invitrogen, Carlsbad, CA) or phycoerythrin-conjugated mouse anti-human CCR5 antibody (BD Biosciences) and quantified with standardized beads from QuantiBRITE system (BD Biosciences). The sensitivity vector describing the relative sensitivity of viral entry to CD4 and CCR5 surface expression levels was derived by using the VERSA computational platform (Viral Entry Receptor Sensitivity Analysis [http://versa.biomath.ucla.edu/]) (27a).
Enzyme-linked immunosorbent assay (ELISA) was used to determine the 50% effective concentration (EC50) of four-domain sCD4 binding to gp120. gp120 proteins of R3A, TA1, and V3(9,9) were expressed in 293T cells using recombinant vaccinia virus vectors and purified by Galanthus nivalis lectin chromatography (Vector Laboratory, Burlingame, CA) as previously described (56). Purified gp120 at 100 ng per well was adsorbed onto Immunolon 2HB 96-well plates (Thermo Labsystems, Waltham, MA) in ELISA capture buffer (0.15 M sodium carbonate, 0.35 M sodium bicarbonate in phosphate-buffered saline (PBS; pH 9.6) overnight at 4°C. Wells were washed with PBS-0.05% Tween and then blocked with 2% bovine serum albumin in PBS for 1 h at 4°C. Starting at 25 μg/ml, fivefold dilutions of sCD4 were made in blocking solution, added to the plate at 100 μl/well, and allowed to bind for 2 h at 4°C. After a washing step, OKT4 antibody at 1 μg/ml was added and allowed to bind for 2 h at 4°C. OKT4 was detected with horseradish peroxidase-conjugated goat anti-mouse IgG. The optical density at 450 nm was measured by using an MRX Revelation microplate reader (Dynex Technologies, Chantilly, VA). EC50s were calculated and analyzed by using a two-sided Student t test in GraphPad Prism 4.0 (GraphPad Software, Inc.).
We previously reported that replacing the central 15 amino acids from the V3 loop of the R5X4 HIV-1 strain R3A with a three-residue GAG linker resulted in a poorly functional Env, termed V3(9,9), with this designation describing a V3 loop containing its first and last nine residues in addition to the cysteine residues at the base of the loop (31). However, function was eventually restored upon passaging this virus on SupT1 cells expressing high levels of CCR5. The adapted virus, termed TA1 (for tissue culture adapted 1), retained the V3 loop truncation but exhibited greatly enhanced membrane fusion activity compared to the unpassaged V3(9,9) virus. In addition, TA1 used CCR5 but not CXCR4 to mediate infection, was highly resistant to CCR5 antagonists by virtue of the ability to utilize drug-bound conformations of CCR5, and was highly sensitive to neutralization by b12, a broadly neutralizing antibody targeting the CD4 binding site (31). Relative to the poorly functional V3(9,9) Env, TA1 contained five mutations, including deletion of residues TTKN in the V2 loop, and a single amino acid change, T342A, in C3 near the base of V3, both of which result in the loss of N-linked glycosylation sites (Table (Table1)1) . The other adaptive mutations consisted of an alanine-to-valine change at the tip of the truncated V3 loop, R254K in the C2 region, and an A509V mutation at the N terminus of the gp41 subunit. To identify the mutations responsible for the enhanced function of TA1 and, more importantly, the mechanism by which an Env protein can mediate entry despite a significant deletion in the functionally important V3 loop, we singly and in combination introduced each of the five mutations from the adapted clone TA1 into the unadapted V3(9,9) background (Table (Table1).1). In addition, we produced the reciprocal mutants and, subsequently, produced a series of combination mutants.
The panel of Env mutants was used to produce viral pseudotypes that were normalized for p24 content, titrated on SupT1 cells expressing CCR5 to define the linear range of the assay, and then used for comparative infection assays using equivalent amounts of p24. Western blot analyses indicated that all Envs were incorporated onto virions with approximately the same efficiency (data not shown). We found that the T342A mutation in the V3(9,9) background restored infection to 40% of TA1 levels (Fig. (Fig.1B),1B), whereas the GVG mutation in the V3 linker and the A509V mutation in gp41 enhanced infection efficiency by a modest amount. The ΔTTKN and R254K mutations in the V2 loop and C2, respectively, did not increase viral entry in this assay. Experiments with various combinations of TA1 mutations revealed that T342A, in combination with either the GVG mutation in V3 or the A509V mutation in gp41 restored viral entry in this single-cycle infection assay to the same levels as TA1. Conversely, restoring these mutations to wt sequences in TA1 reduced infectivity to levels near that of V3(9,9) (Fig. (Fig.1C).1C). Similar results were obtained when using human 293T or NP2 cells expressing CD4 and CCR5, though overall infection levels were not as great as those obtained with SupT1.CCR5 cells.
Based on the results of the single cycle infection assays, we expected that introduction of the T342A mutation along with either the GVG or A509V mutation into replication competent V3(9,9) virus would enhance replication to levels similar to that of the TA1 virus. However, while the double mutants T342A/GVG and T342A/A509V, as well as the triple mutant T342A/GVG/A509V, replicated efficiently on SupT1 cells that expressed CCR5 and the C-type lectin DC-SIGNR, they did so with delayed kinetics (Fig. (Fig.2).2). Consistent with this, TA1-infected cultures were characterized by extensive cytopathic effects and very large syncytia during the first 3 to 5 days postinfection, while few syncytia were observed in cultures infected with the triple mutant until more than 7 days postinfection. For all viruses, the pattern of replication was similar for virus inputs of 5 ng (Fig. (Fig.2)2) or 50 ng (data not shown), with the only difference being the time to peak RT levels. Therefore, while the T342A mutation, along with either the GVG or A509V mutations, played the greatest role in restoring function to the V3(9,9) Env, the remaining two adaptive mutations make functional contributions that are evident in the context of a spreading infection, but not in a single-cycle infection assay.
One of the most striking features of the TA1 virus is its complete resistance to CCR5 antagonists. Previous experiments using the unadapted V3(9,9) Env in cell-cell fusion assays suggested that the V3 truncation alone was able to impart some level of drug resistance. However, for TA1 it was not clear whether simple truncation of the V3 loop was solely responsible for this phenotype or whether the adaptive mutations contributed to CCR5 antagonist resistance. To examine this, we performed infection assays using viral pseudotypes bearing R3A or TA1 Envs or Envs containing combinations of the most important adaptive mutations in the V3(9,9) background. Unfortunately, infection levels obtained with the unadapted V3(9,9) Env were too low to measure in this assay.
Infection assays were performed on SupT1.CCR5 cells in the presence of increasing concentrations of the CCR5 antagonist maraviroc (Fig. (Fig.3).3). As expected, viral pseudotypes with the R3A Env were fully inhibited by maraviroc, whereas TA1 pseudotypes were completely resistant. We found that all viruses bearing the V3 loop truncation were fully resistant to maraviroc regardless of the presence or absence of any adaptive mutations. When two of the most important adaptive mutations (T342A and A509V) were introduced into the parental Env R3A, full sensitivity to maraviroc was retained (data not shown). Taken together, these results indicate that the V3 loop truncation introduced into the R3A Env protein is largely or entirely responsible for CCR5 antagonist resistance and is not significantly modified by adaptive mutations that arose during passaging in tissue culture.
TA1 is exquisitely sensitive to neutralization by the broadly neutralizing MAb b12, which engages the CD4 binding site on gp120 (31), and is even more sensitive to MAb 17b, which binds to the bridging sheet region, probably due to enhanced exposure of this functionally important domain. Although the adaptive mutations GVG, T342A and A509V did not obviously contribute to the CCR5 antagonist resistance phenotype, we found that they did play a role in the pronounced neutralization sensitivity of TA1. Viral pseudotypes were preincubated with MAb b12 prior to incubation with SupT1.CCR5 cells. Three days later, cells were lysed and the amount of luciferase activity determined. As we have reported previously, TA1 was more than 100-fold more sensitive to neutralization by b12 than parental R3A (Fig. (Fig.4A).4A). In comparison, V3(9,9) Envs containing two or three of the five adaptive mutations in TA1 exhibited intermediate neutralization phenotypes (Fig. (Fig.4A).4A). Viruses containing the unadapted V3(9,9) Env did not infect cells sufficiently well to determine accurate 50% inhibitory concentrations (IC50s). These data argue that while the V3 truncation by itself largely accounts for the resistance of TA1 to CCR5 antagonists, it is the combination of the V3 loop truncation and the adaptive changes that increase sensitivity to b12, perhaps by increasing exposure of the CD4 binding site. Indeed, TA1 was approximately two-logs more sensitive to neutralization by sCD4 than R3A (Fig. (Fig.4B).4B). However, the adaptive changes did not contribute strongly to this phenotype. Instead, the V3 truncation alone was primarily responsible for the enhanced sensitivity of TA1 to sCD4, whose binding site on gp120 overlaps but is not identical to the b12 epitope.
The adaptive mutations found in TA1 also played a role in enhancing sensitivity to neutralization by MAb 17b, which binds to the CD4-induced bridging sheet region of gp120 (30). Although R3A is resistant to MAb 17b, all V3 truncation-containing Envs were sensitive to the antibody. However, V3(9,9) Envs bearing two or three adaptive mutations were not as sensitive to neutralization by MAb 17b as was the fully adapted TA1 Env (Fig. (Fig.4C).4C). If the V3 truncation were solely responsible for enhanced sensitivity to 17b, then we would expect the double and triple mutants to be neutralized by 17b as efficiently as TA1, but this was not seen. This suggests that that the V3 truncation likely plays a central role in sensitivity to MAb 17b but that the adaptive mutations that confer greater replication capacity on the truncated Env substantially increase sensitivity to this bridging sheet-directed MAb. In summary, the adaptive mutations in TA1 result in increased sensitivity to neutralization by MAbs b12 and 17b (but not to sCD4), which bind to regions in gp120 that are involved in CD4 and coreceptor binding, respectively.
Mechanism of enhanced function via altered interactions with CD4 and CCR5. One mechanism to account for the enhanced sensitivity of TA1 to antibodies that bind to the CD4 and coreceptor binding sites would be if these epitopes on gp120 were more exposed as a result of the adaptive mutations. If so, TA1 might be able to more efficiently engage CD4 and/or CCR5 and thus be able to infect cells expressing low levels of receptor more efficiently. To investigate this, we used the dual-inducible 293-Affinofile cell line, in which the expression levels of both CD4 and CCR5 are independently modulated via separate, inducible promoters (27a, 34, 49). In this cell line, increasing concentrations of the inducers minocycline and ponasterone A result in corresponding increases in cell surface expression levels of CD4 and CCR5, respectively. By using six concentrations of minocycline and eight concentrations of ponasterone A, a matrix of 48 different expression levels of CD4 and CCR5 can be assayed. Quantitative flow cytometry was used to confirm that the expression levels of receptor and coreceptor were reproducible and consistent among experiments. Expression levels in response to graded doses of inducer varied from ~2,000 to ~110,000 copies of CD4 per cell, while CCR5 varied from ~1,200 to ~22,000 copies per cell. Primary human CD4+ T cells typically express between 50,000 and 70,000 copies of CD4 per cell, while CCR5 expression levels tend to be quite variable between individuals, though are typically less than 14,000 copies per cell (36). Thus, expression levels of both CD4 and CCR5 spanned a physiological range in response to induction.
Pseudotyped viruses bearing the R3A, TA1 or panel of V3(9,9) Envs containing key adaptive mutations were used to infect the panel of variably induced 293-Affinofile cells, with each condition being performed in duplicate within each experiment, and each experiment being replicated at least four times. We found that the parental R3A virus was able to infect cells expressing low levels of CCR5, but only if a threshold level of CD4 was expressed (Fig. (Fig.5A).5A). If CD4 levels were low, R3A was unable to infect, even if CCR5 levels were high. In contrast, TA1 was able to infect cells expressing very low levels of CD4 but was much more sensitive than R3A to CCR5 expression levels. Infection mediated by the V3(9,9) Env was both inefficient [note the scale of the y axis: V3(9,9) exhibited a signal-to-noise ratio of approximately 10:1] and required high levels of both CD4 and CCR5. In comparing the infection pattern of V3(9,9) to that of TA1, it is evident that the adaptive mutations acquired by TA1 improved its infection efficiency by greater than 1 log, enhanced the ability of the virus to utilize low levels of CD4, but had no obvious impact on CCR5 utilization. To quantify this, the complex infection data generated by the 293-Affinofile cells were mathematically fitted to three-dimensional surface plots and used to derive a sensitivity vector that describes the relative dependence of virus entry on CD4 and CCR5 levels (see accompanying report by Johnston et al. [27a]). The angle of this vector (θ) is a numerical description of whether a virus is most sensitive to changes in CD4 or CCR5 levels. The greater the angle, the greater the dependence on CCR5 levels for virus entry. Thus, TA1 had a vector angle of 69°, reflecting its greater dependence on CCR5 levels while being able to use low CD4 levels, while R3A had a vector angle of 22°, reflecting its greater dependence on CD4 levels and ability to utilize low CCR5 levels (Fig. (Fig.5B).5B). When various adaptive mutations were introduced into V3(9,9) singly and in combination, it was evident that the adaptive mutations progressively increased the vector angle, indicating that A509V, T342A, and the V3-linker GVG mutations each improved CD4 utilization, with double and triple mutants showing a combinatorial effect. For reasons that we do not understand, results with V3(9,9) containing the A509V mutation were highly variable.
The results with the Affinofile cells were confirmed in part by performing inhibition studies with antibodies to CCR5 (data not shown). We found that TA1 entry was more sensitive than R3A to antibodies that bind to the second extracellular loop (ECL2) of CCR5, a domain that is important for gp120 binding (13, 14, 60, 64), as well as by antibodies to the N-terminal domain of CCR5. The IC50 for TA1 was fivefold lower than that of R3A for MAb 2D7, whose epitope includes residues at the base of ECL2 (35, 64). Similarly, the IC50 for TA1 was 15- to 20-fold lower than that of R3A for MAbs 45529 and 45531, which recognize the C-terminal portion of ECL2 (35). Thus, antibodies that block CCR5 domains important for gp120 binding and thereby reduce the levels of available CCR5 on the target cell have a greater effect on TA1 entry, confirming that TA1 is more sensitive than R3A to levels of CCR5 on target cells.
The ability of TA1 to more efficiently utilize low levels of CD4 to infect cells could be due to enhanced affinity. To test this, monomeric R3A, V3(9,9) or TA1 gp120 proteins were captured directly onto 96-well ELISA plates. Increasing concentrations of 4-domain sCD4 were added to the captured gp120 and detected with the MAb OKT4, which recognizes a membrane-proximal epitope of CD4 (53). The EC50 for TA1 was 48.5 ng of sCD4/ml, compared to R3A at 156.1 ng/ml and V3(9,9) at 120.5 ng of sCD4/ml (Fig. (Fig.6A).6A). In addition, we found that TA1 was slightly less sensitive than R3A to inhibition by MAbs directed against CD4, suggesting that TA1 is able to bind CD4 with higher affinity (Fig. (Fig.6B6B and data not shown). Finally, these differences were relatively slight, though statistically significant and highly reproducible. Besides affinity, the ability of TA1 to utilize low levels of CD4 for infection could be due to postbinding effects, such as the induction of conformational changes needed for coreceptor engagement. One way that this can be determined is to measure the efficiency with which a viral Env protein can be triggered by sCD4 to elicit fusion with cells expressing coreceptor alone. We expressed the R3A, TA1, V3(9,9), or the triple mutant Env proteins in QT6 cells and incubated them with QT6 cells expressing high levels of CCR5 in the presence of increasing concentrations of sCD4 (Fig. (Fig.7A).7A). Under these conditions, all Env proteins were able to elicit cell-cell fusion, with R3A doing so up to threefold more efficiently than TA1 or the triple mutant. However, both TA1 and the triple mutant were triggered by lower concentrations of sCD4 than was R3A, suggesting that the adaptive mutations enhance the sensitivity of these V3-loop deleted Envs to sCD4. However, the inability of TA1 or the triple-mutant Env to achieve high levels of fusion even with high concentrations of sCD4 suggests that there are obstacles to efficient entry mediated by these Envs after CD4 has been bound in this context, most likely attributed to inefficient use of CCR5 (Fig. (Fig.55).
If the adaptive changes in TA1 enhance its ability to both bind CD4 and to undergo subsequent conformational changes more efficiently, then virions bearing the TA1 Env protein should become resistant to antibodies that bind to CD4 more quickly than virions bearing the parental R3A Env protein. However, the diminished ability of TA1 to utilize CCR5 could translate into a kinetic delay between CD4 binding and subsequent CCR5 interactions, which could account for the enhanced sensitivity of TA1 to enfuvirtide (31, 50). To test this, we modified a single-cycle replication assay such that inhibitors of CD4 binding, CCR5 binding, or virus-cell fusion were added at intervals after infection. In this assay, virus is spinoculated onto target NP2.CD4.CCR5 cells at 4°C, which allows the virus to contact the cell surface but prevents fusion and entry. NP2.CD4.CCR5 cells were used due to their adherence to substrate, making time of addition experiments with associated medium changes more efficient. At time zero, the spinoculated cells are transferred to 37°C with or without saturating concentrations of various inhibitors being added at different times. Since each inhibitor targets a specific step in viral entry, it is possible to assess the relative rates at which virus binds to CD4 or CCR5 or undergoes membrane fusion.
Under these conditions on NP2.CD4.CCR5 cells, MAbs to CD4 or CCR5, or the membrane fusion inhibitor enfuvirtide, completely prevented infection by either R3A or TA1 if added prior to the addition of virus. When the various inhibitors were added at different times after virus binding, we found that R3A became progressively resistant to the neutralizing antibody to CD4 (Fig. (Fig.7B).7B). After CD4 engagement, however, coreceptor binding and fusion occurred in relatively quick succession, as indicated by the nearly overlapping curves of R3A infection in the presence of antibodies to CCR5 or enfuvirtide. Thus, for R3A, CD4 binding appears to be the rate-limiting step in the entry process, a result that is consistent with fact that R3A required relatively high levels of CD4 to support virus infection (Fig. (Fig.5).5). In contrast, TA1 became resistant to the anti-CD4 MAb very quickly, indicating that it binds to CD4 in a functionally irreversible fashion soon after binding to the cell surface (Fig. (Fig.7C).7C). However, there was an extensive lag phase before TA1 became resistant to anti-CCR5 MAbs. Like R3A, there was a short lag between CCR5 binding and membrane fusion as judged by resistance to enfuvirtide. This result indicates that TA1 engages CD4 quickly, a finding consistent with the Affinofile cell data, as well as with the ease with which TA1 membrane fusion activity can be triggered by sCD4. However, after binding to CD4, it is evident that TA1 binds to CCR5 inefficiently, which is consistent with the finding that high levels of CCR5 are needed to support infection by this virus (Fig. (Fig.5)5) and by the fact that TA1 is about 1 log more sensitive to enfuvirtide than is R3A (31).
The central goals of this study were to examine how Env function can be restored via adaptive mutations after deletion of a significant portion of the V3 loop, and to determine whether mutations that restore Env function contribute to CCR5 inhibitor resistance or the marked sensitivity exhibited by TA1 to neutralization by the CD4 binding site MAb b12. Functionally, the V3 loop plays the major role in determining whether Env uses CCR5, CXCR4 or both coreceptors to infect cells (6, 7, 57). Structural studies indicate that in the CD4-bound conformation the V3 loop extends from the surface of gp120 by ~30 Å (24), a distance thought to be sufficient for the tip of the V3 loop to engage the ECLs of the coreceptor, while the base of the V3 loop and the adjoining bridging sheet engage the N-terminal domain of the coreceptor (9, 10, 21, 23, 24, 51). Since coreceptor binding is critically important for triggering the conformational changes needed for virus-membrane fusion, disruption of V3 loop-coreceptor interactions can prevent virus infection. Small molecule CCR5 and CXCR4 antagonists disrupt these interactions via an allosteric mechanism in which coreceptor conformation is altered (15, 39, 55, 61, 62), while genetic truncation of the V3 loop directly impacts Env-coreceptor interactions. TA1 provides a useful model to study the plasticity of Env-coreceptor interactions as it efficiently utilizes CCR5 even in the presence of saturating concentrations of potent CCR5 antagonists.
Theoretically, the adaptive changes in TA1 Env could enhance function via several mechanisms. The most obvious, perhaps, would be for mutations in TA1 to enhance CCR5 interactions by increasing binding affinity or by making Env more susceptible to the conformational changes that are induced by CCR5 binding. Our results are not consistent with either of these models, although the assays available to assess these functions are indirect and not as quantitative as we would like. That TA1 interacts with CCR5 less efficiently than R3A is suggested by the fact that its fusion activity is reduced in most contexts and by the finding that while R3A can mediate infection of cells expressing low levels of CCR5, TA1 cannot. Similarly, our time of addition studies with various HIV entry inhibitors revealed that after binding CD4, R3A quickly engages CCR5. In contrast, a lag phase of ~60 min follows CD4 binding for TA1 before it irreversibly engages CCR5. However, do the mutations present in TA1 enable it to interact with CCR5 more efficiently than the unadapted V3(9,9) Env? Due to the poor function exhibited by V3(9,9), our data on this point are not conclusive. If the adaptive changes present in TA1 improved its ability to utilize CCR5, we would anticipate that TA1 would be able to infect cells that express low levels of CCR5 more efficiently than the unadapted V3(9,9) Env. Our results with the Affinofile cell system suggest that this is not the case: TA1 and V3(9,9) were similarly dependent upon CCR5 expression levels under conditions in which CD4 was not limiting, with neither being able to mediate infection of cells expressing low levels of CCR5 regardless of CD4 expression levels.
While our results argue that the adaptive changes in TA1 do not enhance CCR5 utilization, we found clear evidence that, relative to R3A and V3(9,9), TA1 exhibits enhanced ability to bind CD4 and to undergo CD4-induced conformational changes. This conclusion is supported by several lines of evidence. First, provided that sufficiently high levels of CCR5 were expressed, TA1 was able to mediate infection of cells expressing low levels of CD4 more efficiently than either R3A or V3(9,9). Second, TA1 gp120 bound to sCD4 with higher affinity than R3A or V3(9,9) gp120, although this does not necessarily mean that trimeric TA1 Env binds CD4 with greater affinity than R3A. However, time of addition studies with antibodies that bind CD4 and block infection indicated that virions expressing the TA1 Env engaged cell surface CD4 more quickly than did the parental R3A virus, which is consistent with enhanced affinity. Finally, the membrane fusion activity of TA1 could be triggered by low levels of sCD4. Taken together, these results indicate that the adaptive mutations that arose in V3(9,9) upon passaging on SupT1 cells, ultimately giving rise to TA1, did so at least in part by not only enhancing the ability of Env to bind to CD4, but to undergo CD4-induced conformational changes.
In addition to the fact that TA1 functions relatively well despite lacking much of its V3 loop, it exhibits two other unusual phenotypes: complete resistance to CCR5 inhibitors and markedly enhanced sensitivity to neutralization by the CD4-binding site MAb b12. In the case of CCR5 inhibitor resistance, our results suggest that the V3 loop truncation alone accounts for this phenotype. If so, by what mechanism does this occur, and might V3 truncation always result in resistance to coreceptor inhibitors? The model we propose for this phenotype has as its basis the well-known plasticity of Env-coreceptor interactions. A large number of structure-function studies are consistent with a two-step binding model, in which the V3 loop engages the ECLs of CCR5 (with the second ECL being particularly important) while the N terminus of CCR5 binds to the base of the V3 loop and the bridging sheet, with sulfated tyrosine residues in CCR5 playing a central role (9, 10, 18, 23, 24, 38, 45). Nonetheless, this model is subject to variability: some Envs are more sensitive to mutations in the N terminus of CCR5, while others are more sensitive to mutations in the ECLs (14). The picture that emerges is a two-step binding model in which the contributions of each binding event (N terminus versus ECLs) to inducing membrane fusion vary to some degree between virus strains. We propose that R3A is better able to utilize the N-terminal domain of CCR5 than many other viruses. Thus, loss of the distal portion of the V3 loop, with the concomitant loss of interactions with the CCR5 ECLs, is tolerated to some degree. In contrast, similar deletions in other virus strains usually result in a complete loss of function (19, 54, 65, 67, 69). Since small molecule CCR5 inhibitors bind to a hydrophobic pocket in the transmembrane helices of the coreceptor, they may preferentially disrupt the conformation of the ECLs (15, 39, 55, 61-63). If this model is correct, then viruses that can engage the N-terminal domain of the coreceptor more efficiently than other virus strains may continue to engage CCR5 even in its drug-bound conformation. Our limited structure-function studies with CCR5 mutants and TA1 are consistent with this model. Thus, provided that a virus can tolerate loss of the distal portion of the V3 loop, it may very well exhibit resistance to CCR5 inhibitors in the context of reduced overall function. Consistent with this, an HIV-2 strain that functions well despite loss of its V3 loop also exhibits complete resistance to CCR5 inhibitors (37).
Although the adaptive mutations in TA1 made little or no contribution to CCR5 inhibitor resistance, they clearly contributed to the sensitivity of TA1 to neutralization by the CD4 binding site MAb b12. It is tempting to speculate that the enhanced ability of TA1 to both bind and respond to CD4 is linked to its enhanced sensitivity to a CD4 binding site antibody. If the CD4 binding site becomes more exposed as a consequence of partial V3 loop truncation coupled with adaptive mutations, including the loss of two N-linked glycosylation sites, then enhanced sensitivity to neutralization would likely result. A crystal structure of an unliganded HIV-1 gp120 would be needed to more fully understand the contributions of the adaptive mutations to increased sensitivity to MAb b12.
In summary, we find that deletions in the V3 loop that abrogate important interactions with CCR5 are partially compensated for by mutations that enhance interactions with CD4, the step in the virus entry pathway that is immediately upstream of coreceptor binding. How enhanced CD4 binding can compensate for less efficient CCR5 interactions is not clear, but might be related to the cooperative nature of the membrane fusion reaction. It is likely that for membrane fusion to occur, an Env trimer will have to bind multiple CD4 and coreceptor molecules (29, 43, 48). In addition, membrane fusion may also require the concerted action of several Env trimers (40), though some models suggest that a single trimer might be sufficient to form a fusion pore (68). By binding to CD4 more quickly, the coreceptor binding sites in TA1 Env trimers should be induced more efficiently, making subsequent interactions with CCR5 possible. Associated with this phenotype is exquisite sensitivity to the broadly neutralizing MAb b12. If this property is due to enhanced exposure of the CD4 binding site, V3-truncated and adapted Envs could be explored for their abilities to induce antibodies against this conserved and functionally important region.
C.A.-G. and M.M.L. were supported in part by NIH T32 AI007632. R.W.D. was supported by NIH R01 AI 040880 and by the International AIDS Vaccine Initiative. R.G.C. was supported by NIH R01 AI 035572. J.A.H. and R.W.D. were also supported by NIH RO1 AI45378.
Published ahead of print on 19 August 2009.