|Home | About | Journals | Submit | Contact Us | Français|
The region of the human immunodeficiency virus type 1 (HIV-1) envelope glycoprotein gp120 that engages its primary cellular receptor CD4 forms a site of vulnerability to neutralizing antibodies. The monoclonal antibody b12 exploits the conservation and accessibility of the CD4-binding site to neutralize many, though not all, HIV-1 isolates. To understand the basis of viral resistance to b12, we used the atomic-level definition of b12-gp120 contact sites to study a panel of diverse circulating viruses. A combination of sequence analysis, computational modeling, and site-directed mutagenesis was used to determine the influence of amino acid variants on binding and neutralization by b12. We found that several substitutions within the dominant b12 contact surface, called the CD4-binding loop, mediated b12 resistance, and that these substitutions resided just proximal to the known CD4 contact surface. Hence, viruses varied in key b12 contact residues that are proximal to, but not part of, the CD4 contact surface. This explained how viral isolates were able to evade b12 neutralization while maintaining functional binding to CD4. In addition, some viruses were resistant to b12 despite minimal sequence variation at b12 contact sites. Such neutralization resistance usually could be reversed by alterations at residues thought to influence the quaternary configuration of the viral envelope spike. To design immunogens that elicit neutralizing antibodies directed to the CD4-binding site, researchers need to address the antigenic variation within this region of gp120 and the restricted access to the CD4-binding site imposed by the native configuration of the trimeric viral envelope spike.
The human immunodeficiency virus type 1 (HIV-1) envelope glycoprotein spike (Env), which mediates viral entry into host cells, is composed of three gp120 surface glycoproteins attached noncovalently to three gp41 transmembrane molecules (58). Globally, HIV-1 consists of numerous genetic subtypes and circulating recombinant forms. The Env protein sequence from distinct subtypes can differ by more than 30%, and amino acid variation in exposed gp120 loops can exceed 5% per year in a single individual (21, 45). During the course of HIV-1 infection, viral mechanisms of variation and immune evasion respond to the adaptive immune response to preserve viral replication. With regard to humoral immunity, longitudinal studies reveal a continually evolving virus, which stays a few antigenic steps ahead of the host neutralizing antibody response (1, 31, 32, 42, 46, 52-56). Since neutralizing antibodies against HIV-1 are thought to be a key component of an effective preventive immune response (16, 29, 36), vaccine-elicited antibodies will have to contend with viral immune evasion and sequence diversity to effectively neutralize circulating HIV-1 strains.
One potential solution for eliciting a broadly reactive neutralizing antibody response lays in the constraints that Env function places on viral variation. HIV-1 gp120 first binds to its primary host cell receptor CD4, followed by binding to a coreceptor, generally CCR5 (58). Therefore, the regions of gp120 that engage CD4 and CCR5 are potential sites of vulnerability for broadly neutralizing antibodies (reviewed in references 16 and 36). Recent analyses of sera from HIV-1-infected individuals demonstrate that serum neutralizing antibodies can target the CD4-binding site (CD4bs) of gp120 and are capable of neutralizing diverse strains of HIV-1 (4, 10, 26, 27, 48). Thus, the humoral immune system can target a functionally conserved region of the HIV-1 Env. One specific example is the monoclonal antibody b12 that binds to the CD4bs and is able to neutralize many strains of HIV-1 (5, 7, 43). Antibody b12 was isolated from a phage display library derived from a clade B HIV-1-infected individual (2), and the crystal structures of b12 alone and in complex with the clade B isolate HXBc2 provide an atomic-level definition for the interaction of antibody and virus (35, 60). A major goal of vaccine researchers is to use a combination of functional, structural, and virological information to design vaccine immunogens that could generate neutralizing antibodies that are similar to monoclonal antibody b12.
A potential limitation to such vaccine design efforts is the prevalence of neutralization resistance to b12 among circulating strains of HIV-1. Resistance to b12 has been observed in about 25% of clade B viruses and greater than 50% of non-clade B viruses (5, 23, 24, 33, 39). While a number of studies described changes in Env that affect b12 neutralization sensitivity (3, 11-13, 17, 19, 22, 25, 30, 33, 34, 37, 38, 41, 50, 59), these studies generally focused on prototype viral strains or strains that are not necessarily representative of the circulating primary isolates. In addition, prior studies did not have the benefit of the atomic-level structure of b12 bound to the core of gp120. To investigate the mechanistic basis for b12 resistance among circulating strains of HIV-1, we studied panels of clade B and C reference Env pseudoviruses that were derived from the early stage of HIV-1 infection. Within these two viral panels, 7 of 19 clade B and 7 of 18 clade C viruses were highly resistant to b12 neutralization (23, 24). To understand how natural HIV-1 variation might generate resistance to neutralization, HIV-1 amino acid variation among the reference viruses was mapped onto the atomic-level structure of b12 in complex with the HXBc2 gp120 core. We then modeled each natural amino acid variant within the b12-gp120 contact surface and estimated structural compatibility. The relative binding of b12 to monomeric gp120s derived from each reference virus was measured, and the impact of contact surface variants on b12 neutralization sensitivity was examined. Site-directed mutagenesis was used to further study the effect of specific amino acid residues within the b12-binding surface on b12 binding and neutralization. To determine the impact of viral spike conformation on b12 resistance, we also examined the impact of Env mutations that reside outside of the b12 contact surface and were thought to affect the quaternary conformation of the viral Env spike (6, 19, 20, 25). Taken together, the results show that antigenic variation within the b12-gp120 contact surface and quaternary factors (in the context of the viral Env spike) combine to affect b12 resistance in natural isolates of HIV-1. These data inform vaccine design by explaining how HIV-1 varies specific residues of the CD4-binding region to evade neutralization yet retains binding to its primary receptor, CD4. Understanding the structural basis of b12 resistance may allow modifications of vaccine immunogens to better elicit neutralizing antibodies to vulnerable regions of HIV-1.
The codon-aligned HIV-1 gp120 DNA sequences of group M (n = 776) were obtained from the Los Alamos HIV Database (http://hiv-web.lanl.gov/content/hiv-db/mainpage.html). The alignment included 209 clade B and 390 clade C sequences. The sequences of 19 clade B and 18 clade C reference viruses were added to the alignment, codon aligned, and manually adjusted using BioEdit (15). Maximum-likelihood trees for group M, clade B, and clade C were inferred using the genetic algorithm for rapid likelihood inference (GARLI) as well as a general time-reversible model and by assuming the gamma distribution in-site rate for both heterogeneous and invariant sites (62). GARLI computation processes were carried out on the CIPRES cluster at the San Diego Supercomputing Center with default parameters (http://www.phylo.org/portal/). Trees were displayed with Dendroscope (18).
For the analysis of protein sequence variation from HXBc2, sequences of HIV-1 gp120 were downloaded from the Los Alamos HIV Database. The gp120 amino acid sequences of HIV-1 group M (n = 1250), clade B (n = 551), and clade C (n = 488), including the clade B and C reference panels, were aligned using MUSCLE on the NIAID Biocluster (https://niaid-biocluster.niaid.nih.gov/) and then manually adjusted and trimmed to the gp120 core (from residues 88 to 492 with V1, V2, and V3 deleted) using BioEdit. Positional frequency summaries of the aligned gp120 core sequences of group M, clade B, and clade C were generated in BioEdit. The frequencies of non-HXBc2 residues at each position were summed and tabulated along the linear HXBc2 sequence. The non-HXBc2 fraction for each position then was plotted on the three-dimensional core gp120 structure or against the primary sequence of HXBc2 for visualization.
To map protein sequence variation onto the three-dimensional structure of gp120, the B-factor field of the Protein Data Bank file of the gp120 core was replaced with the value for non-HXBc2 fraction data obtained as described above, and the structural elements were colored according to a variation scale, with a white-to-red gradient representing 0 to 100% variation from the HXBc2 sequence. The gp120 core sequence of each b12-resistant clade B and C reference virus was modeled on the b12-bound gp120 core structure (Protein Data Bank code 2NY7, chain G) using SwissModel (http://swissmodel.expasy.org/SWISS-MODEL.html) in alignment mode. Models were analyzed for potential incompatibilities with b12, assuming an invariant backbone for both gp120 and b12 and side chains with energetically favorable rotomers. Substitution incompatibility was concluded only if no energetically favorable side chain orientation could be identified that did not result in a significant steric clash between gp120 and b12. Main-chain and side-chain contacts between b12 and the gp120 core were defined by the Molecular Surface Program (8). Structure-related figures were generated with Pymol (http://www.delanoscientific.com/).
The CD4bs antibodies b12 and b6 were provided by Dennis Burton and Ralph Pantophlet (Scripps Research Institute, La Jolla, CA), F105 was from Marshall Posner (Dana-Farber Cancer Institute, Boston, MA), and m18 was from Dimiter Dimitrov (National Cancer Institute, Frederick, MD). The coreceptor binding site antibody 17b was from James Robinson (Tulane University, New Orleans, LA). Two-domain soluble CD4 (sCD4-183) was obtained through the NIH AIDS Research and Reference Reagent Program or purchased from Pharmacia. TZM-bl cells were obtained from the NIH AIDS Research and Reference Reagent Program, as contributed by John Kappes and Xiaoyun Wu. The full-length HIV-1 rev/env expression plasmids of the reference panel viruses were described previously (23, 24) and were obtained from the NIH AIDS Research and Reference Reagent Program. Env pseudovirus was prepared as previously described and stored at −80°C or in the vapor phase of liquid nitrogen (23).
Site-directed mutagenesis was performed according to the instructions for the QuikChange XL site-directed mutagenesis kit (Stratagene, La Jolla, CA). We amplified the plasmid template with each forward and reverse primer using Pfu Turbo, digested the reaction mixture with DpnI, and transformed it into Top10 chemical-competent cells (Invitrogen, Carlsbad, CA). Colonies were screened for the presence of the desired mutation by DNA sequencing. The entire HIV-1 rev/env region was sequenced for the final plasmid preparation (Qiagen Inc., Valencia, CA). To generate double mutations, plasmids containing a single mutation were used as templates.
The relative binding capacity of monomeric gp120 and b12 was measured by a gp120 capture enzyme-linked immunosorbent assay (ELISA) as previously described (33). Briefly, 96-well ELISA plates were coated with sheep anti-gp120 C5 antibody D7324 at 1 μg/ml in phosphate-buffered saline (PBS) overnight at 4°C, followed by incubation with Env pseudovirus treated with 0.5% Triton X-100 for 1 h at 37°C. Plates were blocked with 1% bovine serum albumin in PBS for 1 h at 37°C. Serial dilutions of b12, starting from 50 μg/ml (338 nM) for b12-resistant strains and 10 μg/ml (68 nM) for b12-sensitive strains, was added for 1 h at 37°C, along with other control antibodies. The antibody binding was detected with horseradish peroxidase-conjugated goat anti-human immunoglobulin G (Jackson ImmunoResearch Laboratories Inc., West Grove, PA). Plates were read on a Spectramax microplate spectrophotometer (Molecular Devices, Sunnyvale, CA). The binding capacity was reported as the 50% effective concentration (EC50) calculated by curve fitting using the sigmoidal dose-response model with a variable slope in GraphPad Prism, version 5.0 (GraphPad Software Inc., La Jolla, CA).
Neutralization was measured using HIV-1 Env pseudoviruses to infect TZM-bl cells as described previously (23, 49). Briefly, 40 μl of virus was incubated for 30 min at 37°C with 10 μl of serially diluted test antibody in duplicate wells of a 96-well flat-bottom culture plate. To keep assay conditions constant, sham medium was used in place of antibody in specified control wells. The virus input was set at a multiplicity of infection of approximately 0.1, which generally results in 100,000 to 400,000 relative light units in the luciferase assay. Neutralization curves were fit by nonlinear regression using a four-parameter hill slope equation programmed into JMP statistical software (JMP 5.1; SAS Institute Inc., Cary, NC). The 50% inhibitory concentrations (IC50) were reported as the antibody concentrations required to inhibit infection by 50%.
Statistical analyses were performed using GraphPad Prism, version 5.0 (GraphPad Software Inc.).
Monoclonal antibody b12 inhibits HIV-1 entry by binding to a region of gp120 that overlaps the functionally conserved site of gp120 attachment to CD4. The known contact sites for b12 were defined in a structure derived from the clade B virus HXBc2 (60). Therefore, we analyzed the total variation of b12 contact sites relative to the HXBc2 sequence among group M isolates and among clade B and C isolates of HIV-1. By DNA sequence analysis, clades B and C represent distinct branches of group M diversity (Fig. (Fig.1A,1A, left). Among clade B viruses, isolate HXBc2 resides close to the central branching point (Fig. (Fig.1A,1A, middle), while among clade C viruses, it is a distant outlier (Fig. (Fig.1A,1A, right). The variation of clade C viruses, or of all group M viruses, from the HXBc2 protein sequence can be displayed directly on the gp120 surface structure (Fig. (Fig.1B)1B) or as a histogram in the context of HXBc2 primary amino acid sequence (Fig. (Fig.2).2). While some regions of the gp120 core structure show significant variation, most of the b12 contact surface is conserved (Fig. (Fig.1B,1B, top row). However, one conspicuous site of divergence within the b12 contact surface is at the CD4-binding loop (Fig. (Fig.1B,1B, insets). This gp120 region of 10 continuous amino acids (Fig. (Fig.2)2) is composed of a short α3-helix and a β15-strand, which hydrogen bonds in an antiparallel manner with the C strand of CD4 and is essential to gp120-CD4 recognition. The CD4-binding loop is a central and critical contact site for b12 binding, yet the α3-helix shows significant variation, especially among clade C viruses (Fig. (Fig.1B,1B, insets). To investigate how such variation affects b12 recognition, we analyzed a recently defined reference panel of primary clade B and clade C Env pseudoviruses.
A well-characterized panel of 19 clade B and 18 clade C Env pseudoviruses, all derived during the acute or early stage of HIV-1 infection, was used to study b12 resistance (23, 24). Sequence analysis revealed that the reference viruses were spread relatively evenly within the evolutionary dendrogram (Fig. (Fig.1A).1A). Seven clade B and seven clade C viruses were highly resistant to neutralization by b12. We aligned the amino acid sequences of the clade B and C reference viruses to the b12-sensitive clade B virus HXBc2 at the b12 contact residues defined by the atomic-level structure of the HXBc2 gp120 core bound with b12 (60) (Table (Table1).1). Among the 33 sites of b12-gp120 contact, 15 were invariant between the clade B and C reference viruses and HXBc2. The median number of sites that showed variation from HXBc2 was three among clade B viruses and six among clade C viruses. The greatest level of amino acid variation occurred in the CD4-binding loop, which constitutes almost 50% of the contact surface between the gp120 core and b12. Amino acid variation was observed at 6 of the 10 continuous amino acids that make up the CD4-binding loop.
To assess the impact of sequence variation on the gp120-binding surface for antibody b12, we computationally modeled each site of amino acid variation observed among the b12-resistant viral strains. Modeling preserved the backbone conformation from the b12-HXBc2 gp120 core structure, with potential clashes analyzed for all likely side chain rotomers. This analysis revealed that variants predicted to interfere with b12 binding were concentrated within the dominant b12 contact surface, the CD4-binding loop, at residues 364 to 373. These variants, underlined in Table Table1,1, occurred at just three sites and included S364H, P369L/T/Q, and T373M. Interestingly, each of these three b12 contact residues was identified as a site of high variation in the sequence analysis described above, and each residue was located just proximal to the CD4 contact area of gp120 (Fig. (Fig.3).3). Therefore, variation at these b12 contact sites would not be expected to significantly affect the gp120-CD4 binding interaction. As expected, some of the observed amino acid variants occurred solely among b12-resistant viruses. For example, the combination of P369L/T/Q together with T373M was seen among only three clade B b12-resistant viruses. However, other variants predicted to deleteriously affect b12 binding occurred commonly among sensitive and resistant viruses. For example, T373M alone was common among both b12-sensitive and -resistant clade B viruses. In addition, P369L was completely conserved in the panel of 18 clade C viruses. Hence, the one clade B virus with a leucine residue at position 369 was b12 resistant, as predicted by structural modeling. But 369L was present in all 18 clade C viruses and did not distinguish b12-sensitive and -resistant viruses. Thus, while the HXBc2 gp120-b12 crystal structure generally served as a useful guide for allowed variation, side-chain incompatibilities that were predicted from modeling structures that utilized a rigid backbone for the b12-gp120 crystal structure did not always apply to the b12 interaction with the Env of other viruses, particularly the clade C viruses.
Guided by sequence analysis and computational modeling, we performed site-directed mutagenesis to restore wild-type phenotypes to key contacts predicted to mediate b12 resistance. We hypothesized that these amino acid substitutions would enhance b12 binding to gp120 and increase viral sensitivity to b12 neutralization. To assess the impact of specific amino acid changes on b12 binding, we measured the relative binding of b12 to monomeric gp120 by ELISA using captured gp120 that had been gently detergent dissociated from Env pseudoviruses. In this manner, gp120 binding could be assessed using the same virus stock that was used to measure virus neutralization sensitivity. Data for b12 binding and neutralization activity are summarized for each of the wild-type and the mutant Env pseudoviruses in Table Table2.2. As described below, we could readily account for the resistance of some viruses to monoclonal antibody b12 based solely on alterations at known b12 contact sites. Two b12-resistant viruses, the clade B virus 7165.18 and the clade C virus ZM109F.PB4, carried a serine-to-histidine mutation at position 364 within the CD4-binding loop, which was predicted to interfere with b12 binding. We made a single mutation at residue 364 (H364S) in each viral Env. In both cases, this resulted in a more-than 40-fold increase in b12 binding to gp120 (Table (Table22 and Fig. 4A and C), and both viruses became fully sensitive to b12 neutralization (Fig. 4B and D). Note that this mutation had little or no effect on viral neutralization sensitivity to sCD4. The clade C Du123.6 virus containing 364H already was sensitive to b12 neutralization (IC50, 1.1 μg/ml), but the introduction of an H364S mutation into this virus increased b12 binding by more than 10-fold and neutralization sensitivity by 36-fold (Table (Table2).2). Of note, the 364H variant occurs in only about 1% of clade B isolates, but it is relatively common among clade C viruses, with 18% of viruses containing this sequence according to the Los Alamos HIV Database. Therefore, 364H may be an important mechanism contributing to b12 resistance among clade C viruses.
To further elucidate the role of the 364H mutation, we introduced this histidine point mutation into eight b12-sensitive strains, five clade B (including HXBc2) and three clade C. The introduction of the 364H sequence completely abolished b12 binding to gp120s of YU2 and SS1196.1, and it reduced b12 binding to HXBc2 gp120 by 40-fold (data not shown). The YU2 and SS1196.1 pseudoviruses with this mutation displayed a 50-fold or greater level of b12 resistance compared to those of the respective wild-type viruses. However, the 40-fold reduction in b12 binding to HXBc2 gp120 resulted in only a twofold increase in resistance to b12 neutralization, probably because the highly b12-sensitive HXB2 pseudovirus maintains a highly open Env configuration. In contrast, the S364H mutation introduced into viruses SC422661.8 and QH0692.42 caused only a modest reduction in b12 binding to gp120, less than sevenfold. As a result, the S364H mutation did not significantly change the b12 neutralization sensitivity of these two viruses (data not shown). The introduction of the 364H sequence into three b12-sensitive clade C viruses, Du151.2, ZM197M.PB7, and CAP210.2.00.E8, resulted in a significant reduction of b12 binding to each viral gp120, and all three viruses became resistant to b12 neutralization (data not shown).
Alterations at residues 369 and 373 within the CD4-binding loop also were predicted to interfere with b12 binding. We performed site-directed mutagenesis to restore these contact sites in three clade B b12-resistant viruses, 6101.10, TRO.11, and CAAN5342.A2. The single mutation L369P for 6101.10, T369P for TRO.11, and Q369P for CAAN5342.A2 increased b12 binding for each of the respective gp120 molecules to less than 1 nM, yet all three viruses remained resistant to b12 neutralization (Table (Table2).2). By combining a second mutation, M373T, viruses 6101.10 and TRO.11 became modestly sensitive to b12 neutralization, with IC50s of 16.1 and 13.1 μg/ml, respectively (Table (Table22 and Fig. Fig.4F).4F). However, CAAN5342.A2 remained completely resistant to b12 neutralization despite reversion to the HXBc2 amino acid sequence at both positions 369 and 373 and despite strong binding by b12 to this gp120. Also, the single mutation M373T in virus BG1168.1 increased b12 binding by 12-fold, yet the virus remains resistant to b12 neutralization. Overall, these data showed that alterations at positions 369 and 373 can affect b12 binding and neutralization. However, in several cases b12 binding to gp120 was substantially increased with little or no impact on virus neutralization. This suggested that b12 has restricted access to its epitope on the functional native Env trimer of some isolates. A further example of this occurrence is demonstrated by the wild-type sequence of virus PVO.4, which is highly resistant to b12 while containing b12-binding site residues that are identical to those of HXBc2.
With the exception of the serine-to-histidine alteration at position 364 for virus ZM109F.PB4, we could explain only a limited amount of the b12 resistance of clade C viruses based on the analysis and modeling of known b12 contact sites. Most clade C sequences in the Los Alamos HIV Database, and all clade C viruses in our reference panel, have a leucine instead of proline at residue 369, regardless of b12 neutralization sensitivity. Because 369L in the HXBc2-based structure predicted interference with b12 binding, we performed site-directed mutagenesis to restore this contact site to proline in eight clade C viruses: four that were highly resistant to b12 neutralization and four that were moderately sensitive (Table (Table2).2). The single mutation L369P had no substantial impact on b12 binding to each of the respective gp120 molecules, and all four b12-resistant viruses remained highly resistant to b12 neutralization. However, the four moderately sensitive viruses became 4- to 20-fold more sensitive to b12 neutralization, demonstrating that while the computational modeling for clade C viruses was not absolutely predictive, 369P does play a role in b12 neutralization.
To further evaluate the overall relationship between b12 binding and neutralization, we analyzed the relative capacity of b12 to bind to gp120 and neutralization sensitivity for all 37 clade B and C Env pseudoviruses. The b12-binding EC50 for all 23 b12 neutralization-sensitive viruses was less than 1 nM (Table (Table22 and Fig. Fig.5A),5A), suggesting that high-affinity b12 binding is necessary for potent neutralization. Additionally, b12 demonstrated poor binding (EC50 > 338 nM) to the gp120s of 5 of 14 neutralization-resistant viruses, 4 of which were clade C. However, b12 demonstrated modest to high relative binding (EC50 < 5 nM) to the gp120s of five b12-resistant viruses, suggesting that the resistance of these viruses was the result of limited antibody access to its epitope on the viral Env spike and not from variation at sites of direct b12-gp120 contact. Thus, even though linear regression indicates a statistically significant association between antigenic recognition (as measured by b12 binding in the context of monomeric gp120) and neutralization sensitivity (Fig. (Fig.5A),5A), it is clear that b12 binding affinity for viral gp120 does not fully explain viral neutralization sensitivity. We also found a general association between viral neutralization sensitivity to b12 and that to sCD4 (Fig. (Fig.5B).5B). Lastly, relative b12 binding to monomeric gp120 and overall neutralization sensitivity to b12 and sCD4 were comparable for the clade B and C reference strains studied here (Fig. (Fig.5C5C).
As mentioned above, examples of b12 neutralization resistance in the setting of strong b12 binding to monomeric gp120 suggest that the quaternary configuration of viral spike limits Env conformational changes required for b12 access. Published electron tomograms indicate that b12 binding requires significant movement of gp120 in the viral spike (28), and a number of Env modifications have been described that are thought to increase quaternary spike accessibility and result in increased viral sensitivity to neutralization. For example, the removal of an N-linked glycan at residue 197 (Δ197) has been reported to increase viral neutralization sensitivity to both sCD4 and b12 (19, 25). We therefore tested the effect of Δ197 on b12 sensitivity by introducing the point mutation S199A, which disrupts the glycosylation sequon and prevents N-linked glycosylation at residue 197 at the base of the V2 loop. Among the 14 b12-resistant viruses tested, 9 became sensitive to neutralization by b12 after the introduction of this mutation (Table (Table3),3), and an example of these data is shown in Fig. 6A and B. This increase in b12 sensitivity generally was accompanied by an increase in sCD4 sensitivity, suggesting that the removal of this glycan eases quaternary restraints, which otherwise would reduce CD4 binding. Interestingly, the Δ197 glycan removal increased sCD4 sensitivity in four viruses (6101.10, BG1168.1, TRO.11, and CAP244.2.00.D3), but these viruses remained resistant to b12 neutralization. The sCD4 and b12 neutralization profiles of virus CAAN5342.A2 were not affected by the Δ197 glycan removal.
We also tested other mutations that have been reported to affect the quaternary conformation of the viral Env spike (6, 11, 12, 19, 47). The removal of a glycan at the base of the V3 loop at position 301 (Δ301) by point mutation T303A rendered virus PVO.4 sensitive not only to b12 and sCD4 neutralization but also to neutralization by antibodies b6, F105, and m18 (Table (Table33 and Fig. 6C and D). Since PVO.4 has a wild-type (HXBc2) sequence at all known b12 contact sites, these data are consistent with a model of the quaternary restriction of b12 binding in the native viral PVO.4 spike. Of note, the Δ301 mutation did not alter the b12 resistance of 6101.10 and BG1168.1, suggesting some strain specificity to the effect of particular glycan alterations. Several single or double mutations in the C-terminal half of gp41 recently have been shown to affect viral sensitivity to neutralization by b12 (6). We therefore tested the point mutation T569A or I675V or the double mutation T569A and I675V. As expected, these gp41 mutations did not affect the relative binding of b12 to monomeric gp120, yet they rendered both CAAN5342.A2 and BG1168.1 sensitive to b12 and sCD4 neutralization (Table (Table33 and Fig. 6E and F). The increase in viral sensitivity to sCD4 as well as to the weakly neutralizing CD4bs antibodies such as b6, F105, and m18 indicates that these gp41 mutations have a major effect on viral spike accessibility. Of note, although the length of variable loops and the number of putative N-linked glycosylation sites on gp120 have been reported to affect antibody neutralization (9, 14, 46, 57), we did not observe an association between b12 neutralization sensitivity and the length of variable loops or the number of putative N-linked glycosylation sites (data not shown).
HIV-1 Env exhibits a high degree of antigenic diversity and appears to have limited sites of vulnerability to neutralizing antibodies. However, the region of gp120 that contacts its primary receptor, CD4, is functionally conserved and is vulnerable to neutralizing antibodies. Some HIV-1-positive sera contain broadly reactive neutralizing antibodies directed to the CD4bs, and there is one well-characterized neutralizing monoclonal antibody, b12, that targets the CD4bs. Hence, the CD4bs has become a focus of vaccine design efforts to elicit neutralizing antibodies that would be effective against most circulating strains of HIV-1 (16, 36, 40). In this regard, the atomic-level structure of the core of gp120 bound to monoclonal antibody b12 provides detailed architectural information that can form the basis of immunogens design efforts. One potential limitation to the success of these efforts is the natural occurrence of b12-resistant viruses. If the atomic-level structure of the b12 contact surface is to be used for vaccine design, we have to understand the structural basis of b12 resistance to prospectively design immunogens that have the capacity to generate neutralizing antibodies to both b12-sensitive and b12-resistant viruses. The analysis described here reveals specific antigenic and phenotypic alternations that can confer b12 resistance in a set of naturally occurring clade B and C viruses.
Prior to the elucidation of the atomic-level structure of b12 bound to the gp120 core and the knowledge of b12 contact sites, numerous studies described amino acid alterations that account for b12 resistance in some viral isolates. Sodroski and colleagues studied the emergence of b12 resistance for two strains of chimeric simian-human immunodeficiency virus, SHIV-HXBc2 and SHIV-89.6P, after passage in monkeys. They documented specific amino acid changes in regions V2 (SHIV-89.6P) or V1, V2, V3, and gp41 (SHIV-HXBc2) that were associated with b12 resistance. These mutations did not influence binding to monomeric gp120 but did affect binding to cell surface-expressed Env, suggesting that b12 resistance was the result of alterations in Env conformation that restricted the access of b12 to its epitope (13, 50). Pantophlet and colleagues performed a comprehensive alanine mutagenesis of the JR-CSF isolate and described numerous alterations that affected b12 binding to gp120 or the b12 neutralization of the resulting Env pseudoviruses. In some cases, these mutations resulted in decreased b12 binding to gp120 and were in positions that we now know are specific b12 contact sites. In other cases, mutations conferred b12 resistance without affecting b12-gp120 binding (37). Similarly, Parren and colleagues and Moore and colleagues studied primary viral isolates and found that the lack of binding to the relevant gp120 could account for only some of the observed resistance (17, 33, 38). Additional cell culture passage and mutagenesis studies have described amino acid alterations that can increase or decrease viral neutralization sensitivity to b12 (3, 11, 12, 19, 22, 25, 30, 34, 41, 59). In summary, these studies revealed that b12 resistance could result either from poor epitope recognition on gp120 or from steric constraints that limit antibody access to its cognate epitope.
Our study provides a novel analysis of the mechanism of b12 resistance among relevant circulating strains of HIV-1 clades B and C. Guided by the known HXBc2 gp120 contact sites for b12 and the precise Env sequence of each virus studied, we used sequence analysis, molecular modeling, and site-directed mutagenesis to analyze the structural basis of b12 resistance. To assess the total level of amino acid variation for b12 contact sites on gp120, we mapped HIV-1 sequence variations for all group M as well as clade B and clade C viruses onto the surface structure of gp120. The majority of b12 contacts were highly conserved, but there was substantial variation at several sites within the CD4-binding loop that forms the dominant binding surface for b12. Monoclonal antibody b12 interacts with all 10 continuous amino acids that make up the CD4-binding loop (positions 364 to 373), while CD4 interacts with only 6 core residues. Interestingly, the highest variation occurred among the four b12 contact residues that were not sites of CD4 contact. These were amino acid positions 364, 369, 372, and 373 (Fig. (Fig.2).2). We next used the HXBc2 structural data to computationally model each site of amino acid variation observed among the b12-resistant viral strains. This analysis predicted that amino acid variants at three specific sites would result in clashes that interfere with b12 binding. Each of these variants (S364H, P369L/T/Q, and T373M) was within the CD4-binding loop, and each residue was identified by our sequence analysis to be subject to high levels of variation. In addition, computational modeling showed that each of these three b12 contact sites were just adjacent to the known CD4 contact surface. Hence, among the b12-resistant viruses, the contact sites predicted to interfere with b12 binding were concentrated within the CD4-binding loop just proximal to the CD4 contact surface and were sites of significant overall sequence variation. A somewhat similar mechanism of viral escape has been proposed for human rhinovirus 14, which is a nonenveloped picornavirus. The canyon hypothesis suggested that conserved residues within the floor of the viral capsid bind to the cellular receptor ICAM-1. The small dimensions of the canyon limit accessibility to antibodies, and hence neutralizing antibody contacts were at residues adjacent to the canyon that could vary and permit escape from neutralization (44, 51).
To further assess the impact of amino acid variants on b12 binding and neutralization activity, we performed site-directed mutagenesis on b12-resistant viruses to revert specific contact residues back to the wild-type (HXBc2) sequence. As predicted by computational modeling, alterations at residues 364, 369, and 373 could decrease b12 binding and render a virus neutralization resistant. As an example, the clade B viruses 6101.10 and TRO.11 became sensitive to b12 when amino acids at positions 369 and 373 were reverted to the wild-type (HXBc2) sequence. Similarly, the single residue 364H in viruses 7165.18 and ZM109F.PB4 accounted for both poor b12-gp120 binding and a lack of the neutralization of these viruses. Because the 364H sequence is relatively common in clade C, 364H may be an important mechanism contributing to b12 resistance in clade C viruses. It is interesting that some broadly neutralizing clade C sera contain neutralizing antibodies directed to the CD4bs (4), and that these clade C sera can neutralize viruses with the 364H alteration. This suggests the existence of anti-CD4bs antibodies that have an epitope sequence different from that of b12.
While computational modeling did predict the importance of residues 364, 369, and 373 for b12 recognition, our modeling of potential b12 clashes was not always predictive of poor b12 binding and neutralization, particularly among clade C viruses. Although a proline at residue 369 was critical for high-affinity b12 binding and the neutralization of clade B viruses, a P369L alteration was present in all 18 clade C reference viruses regardless of b12 sensitivity. This suggests that computational modeling based on the clade B HXBc2 sequence is less predictive of the b12 interaction with the CD4bs of clade C viruses. Likewise, the T373M mutation among clade B viruses did contribute to neutralization resistance, but this single- amino-acid mutation also was present on several neutralization-sensitive viruses. It is important to note that our computational modeling makes some simplifying assumptions related to protein dynamics. To reduce the complexity of computation, we assumed a rigid backbone of the HXBc2 template and allowed only side-chain movements. In reality, certain degrees of flexibility, including backbone variation, likely are present on gp120, and the antigen-combining site of an antibody often adapts to changes on the contact surface of its ligand. This induced fit may allow b12 to tolerate some amino acid alterations, such as the single P369L or T373M alteration, with only small changes in binding affinity. Finally, the molecular modeling based on the atomic-level structure of the core of gp120 does not take into consideration the interaction of variable loops, the structure of gp41, or the interaction of the three gp120 surface unit molecules (quaternary factors) on the overall structural stability and conformation of the viral spike.
The prior data on the role of Env conformation in restricting b12 access (3, 6, 11-13, 17, 19, 25, 30, 33, 37, 38, 41, 50, 59, 61), and our own data demonstrating clear examples of b12 resistance despite a wild-type (HXBc2) sequence at known b12 contact sites and despite strong binding to monomeric gp120, prompted further investigation to address the potential impact of Env conformation on b12 neutralization. We therefore introduced several previously described mutations that alter the quaternary conformation of HIV-1 Env but lie outside the b12-binding surface of gp120. In one example, the clade B virus PVO.4 was fully resistant to b12 despite having a fully wild-type (HXBc2) sequence at all known b12 contact sites and despite a high relative binding capacity of b12 to the PVO.4 gp120 monomer. The removal of a single glycosylation site, position 301 at the base of the V3 loop (19), improved b12 neutralization potency from an IC50 of >50 μg/ml to 2.0 μg/ml. Several similar examples of this effect for clade B and C viruses, including the effect of mutations in gp41 (6), are shown in Table Table3.3. In some cases, the conformational restriction to b12 access was quite dramatic. Virus CAAN5342.A2 was b12 resistant and remained resistant despite reversions to a wild-type (HXBc2) sequence that increased b12-gp120 binding by 200-fold (the EC50 decreased from 28 to 0.14 nM). However, the wild-type CAAN5342.A2 was rendered highly sensitive to b12 neutralization by one or two mutations in gp41 that had no effect on b12 binding. This suggests a strong interplay between the relative b12 affinity for its target surface on gp120 and the level of steric restriction by the native trimer. If the trimer is rendered highly flexible or open, even modest levels of binding affinity will result in a high level of virus neutralization. In almost every case, increased access by b12 was accompanied by an increase in the neutralization activity of sCD4. Without atomic-level structural data of the native trimer, we can only speculate on the interaction of the gp120 and gp41 molecules that make up each viral spike. One possibility is that the quaternary packing of the V2 and V3 regions of each protomer are affected by the removal of glycans at the base of the V2 or V3 loop. The result is a more flexible native configuration that no longer restricts access to the CD4-binding region of gp120 (Fig. (Fig.6G6G).
We can attempt to explain the mechanism of b12 resistance among the entire set of b12-resistant viruses studied. However, we caution that mutations altering the overall quaternary packing of the viral spike are likely to create Env conformations that are unusually flexible or open and not commonly found among naturally occurring viral isolates. With this caveat in mind, we reviewed the impact of sequence and conformational alterations on all 14 b12-resistant viruses. Of six resistant clade B viruses with mutations at b12 contact sites, reversion to the wild-type (HXBc2) sequence resulted in three viruses that became b12 sensitive (viruses 7165.18, 6101.10, and TRO.11). The gp120 of the remaining three viruses (BG1168.1, CAAN5342.A2, and TRJO4551.58) each demonstrated increased gp120-b12 binding but remained neutralization resistant. Each of these three viruses could be rendered neutralization sensitive by conformation-altering mutations. One clade B virus, PVO.4, contained wild-type (HXBc2) sequence at all b12 contact sites and became sensitive to b12 upon specific gp120 glycan removal distal from the actual contact surface. We also made b12 contact site mutations in five of seven resistant clade C viruses, but only one (ZM109F.PB4) became b12 sensitive. In addition, the reverted clade C viruses did not demonstrate an increase in b12-gp120 binding affinity. As noted previously, this suggests a limitation in modeling, based on the HXBc2 sequence, to predict b12 clashes for clade C viruses. As observed for clade B viruses, all but one of the resistant clade C viruses (CAP244.2.00.D3) could be rendered b12 neutralization sensitive by mutations that altered the Env conformation. Therefore, we can explain the precise structural basis for neutralization resistance for some viruses tested, particularly clade B viruses, and we can demonstrate the interaction of antigenic variation and conformational restriction on viral resistance to the anti-CD4bs monoclonal antibody b12.
In summary, HIV-1 is vulnerable to neutralizing antibodies that target the site of attachment to its primary receptor, CD4. Our data provide a novel mechanistic explanation for how resistance to the anti-CD4bs monoclonal antibody b12 is achieved among a relevant sampling of circulating strains of HIV-1. Using the known HXBc2 gp120 contact sites for b12 and the precise Env sequence of each virus studied, we were able to precisely define the structural context of key amino acid variants that diminish b12 binding and neutralization. These mutations were clustered within the CD4-binding loop of gp120, which forms the dominant binding surface to b12. Notably, amino acid variants conferring resistance resided just proximal to the known contacts of CD4, suggesting a preferential tolerance for variation surrounding the CD4 contact sites. In addition to alterations at b12 contact sites, resistance was influenced by the conformation of the native envelope trimer that restricts access to CD4-binding regions of gp120. To design vaccine immunogens that elicit anti-CD4bs antibodies similar to b12, researchers need to consider the antigenic variation within contact sites on gp120 and the restricted access to the CD4bs imposed by the native configuration of the trimeric viral Env spike.
This work was supported by the Intramural Research Program of the Vaccine Research Center, NIAID, NIH.
We thank Gary Nabel for helpful advice, Diane Wycuff for technical support and helpful discussions, and the Structural Biology Section personnel for helpful comments. We thank Dennis Burton, Marshall Posner, James Robinson, and Dimiter Dimitrov for gp120-reactive antibodies and the NIH AIDS Research and Reference Reagent Program for sCD4.
Published ahead of print on 19 August 2009.