|Home | About | Journals | Submit | Contact Us | Français|
Recently, several broadly neutralizing monoclonal antibodies (bnMAbs) directed to the CD4-binding site (CD4bs) of gp120 have been isolated from HIV-1-positive donors. These include VRC01, 3BNC117, and NIH45-46, all of which are capable of neutralizing about 90% of circulating HIV-1 isolates and all of which induce conformational changes in the HIV-1 gp120 monomer similar to those induced by the CD4 receptor. In this study, we characterize PGV04 (also known as VRC-PG04), a MAb with potency and breadth that rivals those of the prototypic VRC01 and 3BNC117. When screened on a large panel of viruses, the neutralizing profile of PGV04 was distinct from those of CD4, b12, and VRC01. Furthermore, the ability of PGV04 to neutralize pseudovirus containing single alanine substitutions exhibited a pattern distinct from those of the other CD4bs MAbs. In particular, substitutions D279A, I420A, and I423A were found to abrogate PGV04 neutralization. In contrast to VRC01, PGV04 did not enhance the binding of 17b or X5 to their epitopes (the CD4-induced [CD4i] site) in the coreceptor region on the gp120 monomer. Furthermore, in contrast to CD4, none of the anti-CD4bs MAbs induced the expression of the 17b epitope on cell surface-expressed cleaved Env trimers. We conclude that potent CD4bs bnMAbs can display differences in the way they recognize and access the CD4bs and that mimicry of CD4, as assessed by inducing conformational changes in monomeric gp120 that lead to enhanced exposure of the CD4i site, is not uniquely correlated with effective neutralization at the site of CD4 binding on HIV-1.
A study (Protocol G) that screened 1,800 HIV-1 donors infected with viruses of different clades revealed that a significant fraction of donors developed broad and potent neutralizing serum responses, in agreement with studies from several laboratories (5–8, 20, 23). The top 1% of Protocol G donors that exhibited the most broad and potent serum neutralizing responses were designated “elite neutralizers.” A significant proportion of Protocol G donors, who ranked within the top 5%, had an overall broad and/or potent serum neutralization activity that was mediated by antibodies to a conserved region on the primary entry receptor of the virus, the CD4 binding site (CD4bs) (26). The CD4bs is of particular interest as a potential vaccine target since it is a conserved region whose accessibility, at least to CD4, must be maintained.
The first potent broadly neutralizing monoclonal antibody (bnMAb) to this region, MAb b12, was isolated from a phage display library utilizing RNA from an HIV-1-seropositive individual (presumed clade B virus) (1, 2) and neutralized 35% of a 162-virus cross-clade panel (25). However, the observation that b12 interacts with gp120 apparently solely through its heavy chain (34) and the inability, despite extensive efforts, to isolate further anti-CD4bs bnMAbs led to doubts as to whether such Abs could be elicited through immunization. An advance came when MAb HJ16 was isolated by immortalization of memory B cells from a clade C virus-infected donor and shown to exhibit breadth similar to that of b12 (3). A breakthrough was achieved when two bnMAbs, VRC01, which neutralized 91% of a panel of 190 pseudoviruses, and VRC03, which neutralized 57% of pseudoviruses on this panel, were then isolated from a clade B virus-infected donor (27). Furthermore, several new, potent anti-CD4bs bnMAbs have been isolated very recently (22, 24, 28). Of these, 3BNC117 (22), NIH45-46 (4, 22), and PGV04 (also known as VRC-PG04) (22, 24, 28) compete in breadth and potency with VRC01. Furthermore, an engineered variant of NIH45-46, NIH45-46W, shows even greater potency (4).
Here, we focus on characterization of PGV04 to determine whether there is a common mechanism of neutralization used by bnMAbs targeting the CD4bs. PGV04 was isolated from single memory B cells in peripheral blood mononuclear cells (PBMC) of an elite neutralizer, using the RSC3 protein and a CD4bs-defective version for selective isolation of potent CD4bs MAbs (28). Interestingly, the circulating virus in the PGV04 donor was subtyped as clade A1/D recombinant, in contrast to the VRC01 donor virus (clade B) and also the donors of 3BNC117 (clade B virus) and b12 (presumed clade B virus). Here, we show that PGV04 exhibits neutralization breadth and potency similar to those of the much-studied MAbs PG9 and VRC01. Moreover, the neutralizing activity of PGV04 largely recapitulated the neutralization profile of the corresponding donor serum. PGV04 was distinguished from CD4, VRC01, and b12 by its pattern of sensitivity to single alanine substitutions on the background of the JR-CSF pseudovirus. Furthermore, in contrast to VRC01, PGV04 did not enhance binding of the CD4-induced (CD4i) MAbs 17b or X5 to their epitopes colocalized within the coreceptor binding site on monomeric gp120. Notably, none of the CD4bs bnMAbs induced the CD4i site on functional trimers.
The following Abs and reagents were procured by the IAVI Neutralizing Antibody Consortium: 2G12 (Polymun Scientific, Vienna, Austria), X5 and 17b (Strategic Biosolutions), soluble CD4 and CD4-IgG (Progenics Pharmaceuticals, Tarrytown, NY), F425 (provided by Lisa Cavacini, Beth Israel Deaconess Medical Center), JR-CSF gp120 and BaL gp120 (provided by Guillaume Stewart-Jones, MRC Human Immunology Unit, Oxford), JR-FL gp120 (Progenics, Tarrytown, NY), and YU2 gp120s (provided by Robert Doms, University of Pennsylvania).
The donor from whom PGV04 was isolated, donor 74 (26), was selected from the IAVI-sponsored study Protocol G (23). Protocol G enrollment included individuals at least 18 years of age with documented HIV infection for at least 3 years, clinically asymptomatic at the time of enrollment, and not currently receiving antiretroviral therapy. The volunteer was identified as an elite neutralizer based on broad and potent serum neutralizing activity against a cross-clade pseudovirus panel (23).
Mock-treated or endo-β-N-acetylglucosaminidase H (endo-H)- and endo-F-treated BaL gp120 was diluted in phosphate-buffered saline (PBS) and coated at 5 μg/ml overnight at 4°C as previously described (9). Briefly, 5-fold serial dilutions of the MAbs, in 1% bovine serum albumin (BSA) in PBS, were added at a starting concentration of 10 μg/ml. The plates were incubated for 1 h at 37°C and then washed 4 times before the secondary MAb, goat anti-human IgG F(ab′)2 conjugated to alkaline phosphatase (AP; Jackson), was added for 1 h at 37°C. The signal was detected using phosphatase substrate, and the optical density was read at 405 nm. For PGV04 binding to gp120 isolated from JR-CSF pseudovirus, virus was collected 3 days posttransfection, supernatants were spun down at 300 × g for 5 min, and virus was lysed with 1% NP-40 at room temperature (RT) for 30 min. Enzyme-linked immunosorbent assay (ELISA) plates were coated overnight at 4°C with anti-gp120 Ab D7324 (International Enzymes, Inc.) at a concentration of 5 μg/ml in PBS. Plates were washed 4 times, and lysed virus was added at 50 μl/well and incubated for 2 h at 37°C. The remainder of the experiment was conducted as described above. For enhancement of the coreceptor binding site on gp120, 10 μg/ml of PGV04, CD4-IgG, b12, 2G12, VRC01, or VRC03 was added to JR-FL or YU2 gp120-coated plates for 30 min at RT. Then, 5-fold serial dilutions of biotinylated X5 or 17b (50 μl/well) were added, starting at 50 μg/ml and 100 μg/ml, respectively. The plates were washed 4 times and 50 μl/well of streptavidin conjugated to AP was added at 1:1,000 for 1 h. The plates were washed and developed as described above. The polyreactivity assay was done as previously published (25).
Isothermal titration calorimetry (ITC) measurements were performed using the ITC200 microcalorimeter system (MicroCal, Inc.). All reactions were carried out at 37°C as previously described (27). The concentration of gp120 in the sample well was approximately 8 μM, and that of PGV04 in the syringe was approximately 84 μM. The molar concentrations of the proteins were calculated using the following molar extinction coefficients: gp120, 80,110 M−1 cm−1, and PGV04, 229,720 M−1 cm−1. Data were analyzed with MicroCal Origin software using a single-site binding model.
The effects of CD4 or CD4-binding site antibodies on binding of YU2 core gp120 to antibody 17b were determined on Biacore T-200 (GE Healthcare) at 25°C with buffer HBS-EP+ (10 mM HEPES, pH 7.4, 150 mM NaCl, 3 mM EDTA, and 0.05% surfactant P-20). Anti-human Fc antibody was immobilized on a CM5 sensor chip with the standard amine coupling method. IgG 17b was then captured by immobilized anti-human Fc antibody to about 250 response units. YU2 core gp120 solutions at 62.5 nM in the absence or presence of a 10-fold molar excess of two-domain CD4 or Fabs of CD4-binding site antibodies was injected over the 17b and blank channels. The binding reactions were allowed to continue for 2 min at a flow rate of 30 μl/min and then were followed by a dissociation phase for another 4 min. After each run, the channels were regenerated with two 25-μl injections of 3.0 M MgCl2 at a flow rate of 50 μl/ml. T-200 Biacore Evaluation software was used to subtract appropriate blank references.
Saturating amounts of PGV04, b12, 2G12, sCD4, VRC01, VRC03, or 17b were added at 20 μg/ml to JR-FL- or YU2ΔCT-transfected 293T cells and incubated for 30 min at 4°C. Then, a 5-fold serial dilution of biotinylated 17b, starting at 20 μg/ml, was added to each well containing the competitor MAb for 1 h at 4°C. The plate was washed 2 times and stained with a 1:200 dilution of NeutrAvidin conjugated to R-phycoerythrin (PE) (Invitrogen). Binding was analyzed using flow cytometry as previously described (25).
Neutralization assays for MAbs and patient sera were performed by Monogram Biosciences as previously described, using a single round of replication pseudovirus and measuring entry into U87 cells expressing either CCR5 or CXCR4 by luciferase activity (19). The generation of pseudoviruses incorporating HIV-1 JR-CSF single alanine substitutions is fully described elsewhere (16). The neutralization activity of PGV04 against HIV-1 JR-CSF pseudovirus containing single alanine substitutions in the Env protein was measured using a TZM-bl assay as previously described (12).
Statistical analyses were done with Prism 5.0 for Mac (GraphPad, La Jolla, CA).
We tested PGV04 for neutralizing activity against a multiclade 162-pseudovirus panel using U87 target cells (Fig. 1A and B). The neutralization profile of PGV04 contrasted with the previously published neutralization profiles of b12 and CD4 on the same 162-pseudovirus panel (25). PGV04 was compared to PG9, a well-characterized MAb with exceptional breadth and potency, used as a reference to gauge the breadth and potency of PGV04 (24, 25). PGV04 neutralized 142 of the 162 pseudoviruses (88%) at a 50% inhibitory concentration (IC50) of <50 μg/ml, while PG9 neutralized 122 pseudoviruses (75%) (Fig. 1A; also see Table S1 in the supplemental material). Hence, PGV04 exhibited significantly enhanced breadth compared to that of PG9 (Fisher's exact test, P = 0.0063) (Fig. 1B). However, the median IC50 of viruses neutralized at an IC50 of <50 μg/ml was comparable for PGV04 (0.20 μg/ml) and PG9 (0.27 μg/ml), indicating that PGV04 and PG9 display similar potency. Of note, there were 7 viral isolates that both MAbs were unable to neutralize, although these isolates are neutralized by at least one of the bnMAbs PG16, b12, 4E10, 2G12, and 2F5 (25).
Next, we compared the breath and potency of PGV04, PG9, and VRC01 on a second, 97-pseudovirus panel using TZM-bl target cells (Fig. 1A and B; also see Table S2 in the supplemental material). The results for the two virus panels were similar, although in the latter panel, the median IC50 of the viruses that were neutralized at <50 μg/ml was significantly lower for PG9 (0.06 μg/ml) than for VRC01 (0.17 μg/ml) and PGV04 (0.12 μg/ml) (Mann-Whitney test, P < 0.0001 and P = 0.0137, respectively), indicating that PG9 is more potent on this panel than VRC01 and PGV04. On this panel, PG9, PGV04, and VRC01 neutralized 82%, 87%, and 93% of viruses, respectively. Hence, VRC01 exhibited significantly greater breadth than PG9 in this assay (Fisher's exact test, P = 0.0479) but had only slightly broader neutralization than PGV04. The median IC50 for the viruses neutralized with an IC50 of <50 μg/ml was slightly lower for PGV04 (0.12 μg/ml) than for VRC01 (0.17 μg/ml) (Mann-Whitney test, P value = 0.0324), revealing that PGV04 is marginally more potent than VRC01 on this panel.
We next investigated the neutralization breadth and potency of PGV04 compared to those of the donor serum from which PGV04 was isolated. The IC50 of PGV04 correlated highly with the 50% neutralization titers (NT50) of the donor serum (Mann-Whitney test, P < 0.0001) (Fig. 1C). However, there were certain viruses that were neutralized by the donor serum but not by PGV04 (5 of 162 viruses or 3% of the viruses) (see Table S1 in the supplemental material). In these cases, neutralizing antibodies distinct from PGV04 appear to be responsible for the serum's neutralizing activity. Surprisingly, in a few cases, PGV04 neutralized a particular isolate with an IC50 of <1 μg/ml while the donor serum did not neutralize this same isolate (6 of 162 or 4% of the viruses). The neutralization curves for the 6 viruses that the donor serum did not neutralize were trending positive at the limit of detection of the assay. Therefore, the neutralization experiments for these viruses were repeated, and serum neutralization was indeed noted, removing the apparent anomaly (see Table S1).
It has been previously shown that soluble CD4 (sCD4) enhances the binding affinity of the CD4i MAb 17b to monomeric recombinant gp120 (31, 32). Unlike b12, VRC01 has also been shown to enhance 17b binding, although not to the same extent as CD4-IgG (13, 27). In addition to VRC01, several other CD4bs bnMAbs have been shown to induce the CD4i site, suggesting that the ability to mimic CD4 may be a common feature of the most potent bnMAbs (22). CD4bs MAbs such as b12 do not induce the CD4i site but are not as broad and potent as the recently described CD4bs bnMAbs. To determine whether the ability to mimic CD4 in inducing the CD4i site on gp120 is necessary for a MAb to be classified with the most potent and broad CD4bs MAbs, we added saturating amounts of PGV04, CD4-IgG, b12, 2G12, VRC01, VRC03, or no MAb to wells coated with monomeric gp120 and then titrated in either biotinylated 17b or another CD4i MAb, X5 (Fig. 2A and B). PGV04 did not enhance 17b or X5 binding to either JR-FL or YU2 gp120, while CD4-IgG and VRC01 enhanced the binding of both CD4i MAbs, indicating that the breadth and potency of the most broad and potent CD4bs MAbs is not necessarily correlated with CD4 mimicry. Consistent with previously published data, b12 decreased 17b and X5 binding, suggesting that it partially competes with the CD4i MAbs (14, 27, 28).
Recent structural studies have shown that VRC01 and PGV04 display very similar epitope footprints on core gp120 (28). Therefore, to gain insight into how VRC01 enhances CD4i MAb binding to gp120 while PGV04 does not, we measured PGV04 binding to YU2 gp120 using isothermal titration calorimetry (ITC) for comparison with earlier measurements on VRC01 and CD4 (13). CD4, VRC01, and PGV04 have similar apparent binding constants (Ka) to gp120 as estimated by ELISA (data not shown) and, as expected, have similar free energies (ΔG) of binding. The thermodynamic parameters of PGV04 binding to YU2 gp120 were as follows: ΔH was −27.97 kcal/mol (standard error [SE], 2.18), −TΔS was 18.87 kcal/mol (SE, 2.19), and ΔG was −9.06 kcal/mol (SE, 0.23). The entropies for the binding of VRC01 (38.7 kcal/mol) and CD4-IgG (40.0 kcal/mol) to gp120 (13) are about twice those of PGV04 (19.9 kcal/mol), suggesting that gp120 may undergo more extensive rearrangements than PGV04 on binding to VRC01 or CD4-IgG, consistent with the exposure/generation of the CD4i site with the former molecules. The high entropy associated with binding of CD4-IgG and VRC01 binding to gp120 is compensated by high enthalpies of binding, ~50 kcal/mol (13), to generate the observed ΔG values. PGV04 binding to gp120 is correspondingly associated with a much lower enthalpy change of −28 kcal/mol.
Surface plasmon resonance (SPR) was also used to study the enhancement of 17b binding to YU2 gp120 core by CD4-IgG or CD4bs MAbs. Core gp120 was flowed over immobilized 17b in the absence or presence of a 10-fold molar excess of CD4-IgG or Fabs of CD4bs MAbs. A low concentration of gp120 core, 62.5 nM, was chosen whereby binding to 17b was barely detected. CD4-IgG, VRC01, and NIH45-46 readily induced the 17b epitope upon binding to gp120, while PGV04, b12, and VRC03 produced no significant enhancement of 17b binding (Fig. 3).
We next investigated whether PGV04 could induce 17b binding to Env trimers. 293T cells were cotransfected with JR-FL Env DNA and pSG3, a plasmid containing the HIV-1 backbone. Forty-eight hours posttransfection, b12, 2G12, PGV04, sCD4, VRC01, VRC03, or 17b MAbs were prebound at saturating concentrations to the surface-expressed Env trimers. Such JR-FL trimers have been shown to be predominantly cleaved and to possess an antigenic profile expected for functional Env complexes (15). Biotinylated 17b was then titrated onto the prebound complex, and binding was detected using flow cytometry. Binding of sCD4 to cell surface-expressed Env created a structural environment that promoted 17b binding (Fig. 4A). However, none of the MAbs tested induced detectable levels of 17b bound to the functional trimer. Notably, 2G12 produced similar binding curves in the presence and absence of each of the prebound MAbs, demonstrating that none of the CD4bs MAbs induced a significant degree of gp120 shedding from the cell surface trimers (data not shown). These results suggest that there are structural constraints in the functional trimer that prevent the exposure of the coreceptor site on gp120 upon binding of CD4bs MAbs that do not exist in the gp120 monomer. A summary of the enhancement of 17b binding to monomeric gp120 and induction of trimeric Env is summarized in Table 1.
Scheid et al. reported that VRC01 and several novel CD4bs MAbs (3BNC60, 3BNC117, 12A12, 12A21, 1NC9, and NIH45-46) facilitated the binding of CD4i MAb 3-67 (21) to cell surface-expressed Envs YU2ΔCT and/or BaLΔCT (22). One MAb, 8ANC195, did not induce CD4i MAb binding in this study, which was attributed to 8ANC195 not being a traditional CD4bs MAb. This MAb was equally sensitive to binding to gp120 monomers containing the D368R and I420R substitutions, as well as having a different neutralization pattern from the other MAbs. Based on this finding, the authors concluded that the ability to induce CD4i MAb binding is a shared feature of the most potent MAbs. We therefore determined whether PGV04 would induce CD4i MAb binding to cell surface-expressed YU2ΔCT. However, we found that, in contrast to 3-67, 17b alone bound strongly to the YU2ΔCT-expressed Env and this binding was at a higher level than the binding of 17b after saturating amounts of PGV04, VRC01, and PG16 were prebound to the Env (Fig. 4B). As for YU2 gp120 monomer, b12 (and a 17b control) competed with 17b binding, and CD4-IgG had a slightly higher level of binding to YU2ΔCT Env than the background binding of 17b to YU2ΔCT Env. We therefore concluded that none of the CD4bs MAbs we measured, b12, PGV04, VRC01, or b6, nor our control MAbs PG16 and 17b, enhanced 17b binding to YU2. To control for cleaved trimer expression, we determined the levels of b6 and b12 binding to the YU2 Env (15). b6 bound to the YU2 Env to the same degree as 17b, while b12 binding was slightly higher, indicating that most of the YU2 trimers expressed on the cell surface are uncleaved (Fig. 4C). These results are consistent with previous observations that, in contrast to JR-FL, surface-expressed YU2 Env is mostly expressed uncleaved (15). Therefore, we hypothesize that 17b either mostly or only bound YU2 uncleaved spikes and that the most potent CD4bs MAbs do not induce the CD4i site as recognized by 17b on JR-FL or YU2 Env cleaved trimers.
To better understand how PGV04 interacts with the functional trimer, we assessed PGV04's neutralizing activity against a panel of JR-CSF pseudovirus variants containing single alanine substitutions in the gp120 protein (see Table S3 in the supplemental material). Substitutions at D279, I420, and I423 greatly compromised PGV04 neutralization, decreasing the neutralization potency to 1%, 3%, and 5% of that of wild-type JR-CSF, respectively (Table 2). The D279A substitution also greatly compromised CD4 and VRC01 neutralization potency, and yet it had little effect on b12 activity. The I420A and I423A substitutions likewise decreased VRC01's neutralization potency, although not to the same extent as for PGV04. Conversely, these two substitutions increased the neutralization potency of b12 and CD4-IgG. Interestingly, the isoleucines at positions 420 and 423 are 97% and 88% conserved, respectively, among HIV-1 isolates in the Los Alamos database. The aspartic acid at position 279 is only conserved among 45% of isolates, whereas asparagine is located at this position in 51% of isolates. However, the presence of an asparagine at position 279 does not affect the PGV04 epitope, as evidenced by the ability of PGV04 to neutralize isolates containing this substitution (e.g., JR-FL).
Other alanine substitutions also abrogated PGV04 neutralization but not to the same extent as the three residues mentioned above. The N276A substitution decreased PGV04's neutralization potency to 13% of that of wild-type virus. Notably, the N-acetylglucosamine from the N-linked glycan at this residue has previously been determined to contact VRC01 through resolution of the crystal structure of VRC01 bound to gp120 core (33). Surprisingly, removal of this glycan resulted in a 4-fold increase in VRC01's neutralization potency while having no impact on CD4-IgG or b12 neutralization. Additionally, the I307A, I309A, F317A, and Y318A substitutions decreased PGV04 neutralization to 9 to 36% of that of wild-type virus. These residues are found on the tip of the V3 loop and have been shown to be important in maintaining the interaction between gp120 and gp41 (29). PGV04 and VRC01 show an opposite trend compared to CD4-IgG and b12 for reasons that are unclear.
Figure 5A shows the alanine substitutions that affected PGV04 neutralization mapped on the gp120 core in its PGV04-bound conformation. For some of the predicted PGV04 contact residues in the crystal structure, individual alanine substitutions did not yield a decrease in neutralization. This phenomenon was also seen in VRC01 neutralization (13) and is attributed to the inability of single residue substitutions that do not contribute substantial binding energy to the ligand-ligand interaction to be able to significantly disrupt high-affinity antibody-ligand binding.
To further study the PGV04 interaction with full-length monomeric gp120, we evaluated the binding activity of PGV04 to a panel of gp120 proteins containing single alanine substitutions that were captured from lysed JR-CSF pseudovirus (see Table S4 in the supplemental material). The N276A (0%), D279A (0%), N280A (1%), D457A (4%), and R469A (3%) gp120 variants showed severely decreased PGV04 binding relative to that of wild-type gp120, confirming the importance of the corresponding residues in forming the PGV04 epitope (Table 3). The result for D279A gp120 is consistent with the neutralization data found for the D279A variant virus (Table 2) and the published PGV04 crystal structure (33). The other three residues that are important for CD4 binding could not be tested in neutralization experiments because the corresponding variant virus infectivity values were below the detection level of the assay. The D457A substitution has previously been shown to decrease b12 and CD4 binding to below 10% of that of wild-type gp120 (16). Also, a D368A substitution decreases PGV04 binding to 12% of the binding of wild-type gp120. D368 has previously been shown to be part of the CD4, b12, and VRC01 epitopes but not of the HJ16 epitope (3, 26, 33).
In our alanine-scanning studies, we found that certain N-linked glycans affected PGV04 binding and neutralization. Therefore, we determined whether the removal of glycans by endo-H, endo-F1, endo-F2, and endo-F3 from BaL gp120 protein produced in 293T cells would affect the binding of PGV04. PGV04 and the other CD4bs MAbs, b12, VRC01, VRC03, and b6, retained similar levels of binding to both mock and deglycosylated forms of the protein, suggesting that CD4bs MAbs in general do not have a strong dependence on glycans for binding to gp120 (Fig. 6 and similar results for which data are not shown).
PGV04 was tested for binding to a panel of antigens to investigate polyreactivity (Fig. 7). PGV04 bound the JR-FL gp120 positive control and did not bind to any of the other antigens, i.e., histone, human placental DNA, apo-transferrin, ORI, human IgG Fc, transferrin, disialoganglioside G0, ovalbumin (OVA), and gp41, therefore showing no evidence of polyreactivity.
In this study, we characterized PGV04, a newly identified broad and potent CD4bs-directed HIV-1 neutralizing MAb. PGV04 neutralization was determined on two different pseudovirus panels. For the 162-pseudovirus panel, PGV04 and PG9 displayed comparable potency while PGV04 exhibited greater breadth. For the 97-pseudovirus panel, PG9 exhibited similar breadth but increased potency compared to that of PGV04. PG9 neutralized the viruses in the latter panel with greater potency than the viruses in the 162-virus panel, while PGV04 displayed similar neutralization potency on both panels. This may be due to the isolates chosen on both panels, since 7.4% of the viral isolates on the 162-virus panel were from acute/early infection while 44.3% of the viral isolates on the 97-virus panel were from acute/early infection (see Tables S1 and S2 in the supplemental material). Also, the target cell lines used were different: TZM-bl cells were used for the 97-pseudovirus panel and U87 cells for the 162-virus panel.
Overall, PG9 and -16, VRC01, and PGV04 neutralized more than 70% of circulating viruses with a mean IC50 demonstrating approximately 10-fold greater potency than the earlier established HIV-1 bnMAbs b12, 2G12, 2F5, and 4E10. Interestingly, the virus isolated from the PGV04 donor was subtyped as clade A1/D recombinant, which differs from the donors associated with the other CD4bs bnMAbs, 3BNC117 (clade B virus), VRC01 (clade B virus), HJ16 (clade C virus), and b12 (presumed clade B virus). Therefore, it seems that elicitation of broadly neutralizing CD4bs-directed MAbs is not dependent on the clade of the infecting isolate. Of note, the PGV04 donor virus does not appear to be of any known circulating recombinant form (CRF) listed in the Los Alamos HIV databases.
Our results show significant differences between CD4bs bnMAbs in their mode of recognition of the CD4bs region on monomeric gp120. CD4-IgG, VRC01, and NIH45-46 enhance exposure of the coreceptor binding site as assessed by 17b binding, while PGV04, VRC03, and b12 do not. Furthermore, the entropy change associated with PGV04 binding to gp120 is much less than that associated with VRC01 binding, consistent with a smaller conformational change in gp120 on PGV04 binding. Nevertheless, PGV04 binding to gp120 is associated with relatively large entropic and compensating enthalpic changes that are suggestive of protein conformational changes or extensive solvent rearrangements. Further insight is provided by the structures of CD4 (11), VRC01 (33), NIH45-46 (4), VRC03 (28), and PGV04 (28) complexed with gp120 core. These structures show that the conformational changes induced by these CD4bs ligands in gp120 core are similar but have critical differences. In particular, while VRC01 and NIH45-46 induce a conformation of the bridging sheet that is similar to that induced by CD4 and therefore compatible with 17b and X5 recognition, PGV04 and VRC03 induce a different conformation between residues 428 and 431 in strands β20/21. Additionally, a critical 17b-interacting residue, Met434, on β21 in both PGV04- and VRC03-bound forms assumes a position that clashes with CDRH2 of 17b, while that in the VRC01-, NIH45-46-, and CD4-bound forms adopts a nonclashing conformation.
In contrast to the enhancement of binding on monomeric gp120, only CD4 induced the coreceptor site as assessed by 17b binding on JR-FL functional trimers; the MAbs VRC01, VRC03, b12, and PGV04 were ineffective. These observations suggest that the CD4bs is presented differently in the context of recombinant gp120 and the functional trimer; that is, the conformation and/or flexibility of gp120 around the CD4bs is different, allowing a change in the monomer but not the trimer. Furthermore, the finding that none of the CD4bs bnMAbs induced CD4i MAb binding to surface-expressed trimers contrasts with the results found earlier (22). The difference in the two findings may stem, in part, from the use in the previous study (22) of surface-expressed YU2 and BaL Envs, which are known to be a mixture of cleaved and uncleaved Envs (15). Using YU2 and BaL Envs, one cannot differentiate between the expression and induction of the CD4i site on cleaved and uncleaved trimers, the latter of which are not used in viral entry. In contrast, JR-FL surface-expressed trimers are fully cleaved (15) and competent for viral entry. The induction of the CD4i site in this system therefore represents changes occurring on functional Env spikes. A further difference between the two studies is that here we used the CD4i MAb 17b, whereas the earlier study used MAb 3-67, and the two MAbs may bind somewhat different residues and sense somewhat different conformational changes.
We found that several alanine substitutions affected CD4bs MAb neutralization and binding differently, again illustrating that PGV04, VRC01, b12, and CD4-IgG recognize the CD4bs in somewhat different ways. For example, alanine substitutions at positions D279, I420, and I423 greatly decreased neutralization by PGV04 but varied in their effects on VRC01, CD4-IgG, and b12. The D279A substitution decreased neutralization by VRC01 and CD4-IgG but did not substantially affect neutralization by b12. The I420A and I423A substitutions decreased VRC01 neutralization but increased both CD4-IgG and b12 neutralization. In addition, certain substitutions in the V3 loop decreased neutralization by PGV04 and VRC01 but increased neutralization by CD4 and b12. The highly conserved nature of the residues important for PGV04 recognition probably explains how PGV04 is able to achieve broad neutralization. Additionally, mapping the residues found to be important in neutralization and binding onto the gp120 core in its PGV04-bound state allowed us to gain insight into the contribution of each residue in binding to gp120 in the context of the trimer (Fig. 5A and B).
In summary, the recent isolation of several new CD4bs MAbs, notably, VRC01 (27) and 3BNC117 (22), has led to the idea that the success of a CD4bs MAb in broad and potent neutralization may depend on its ability to precisely induce changes in gp120 that mimic those induced by CD4. Indeed, certain bnMAbs, such as VRC01, do enhance binding of the CD4i MAb 17b to monomeric gp120. However, we show here that the equally broad and potent bnMAb PGV04 does not enhance such binding. Furthermore and crucially, in contrast to CD4, none of the CD4bs MAbs tested induced the 17b site on trimeric cleaved Env. Thus, a certain degree of mimicry of CD4 by anti-CD4bs bnMAbs may simply be a consequence of binding to the CD4 epitope on monomeric gp120 rather than a neutralization mechanism. As argued previously (10, 17, 18, 30), the binding of an IgG molecule to an envelope spike may be sufficient to lead to neutralization irrespective of the precise epitope. Finally, the description of differences in recognition of the CD4bs by different bnMAbs is encouraging for immunogen design in relaxing somewhat the stringency of the types of Abs that should be elicited.
We thank the research staff members of the Monogram Clinical reference Laboratory, all of the study participants and research staff at each of the Protocol G clinical centers, and all of the Protocol G team members. In addition, we thank all of the IAVI Protocol G project, clinical, and site team members, the IAVI Human Immunology Laboratory, and all of the Protocol G clinical investigators, specifically, George Miiro, Anton Pozniak, Dale McPhee, Olivier Manigart, Etienne Karita, Andre Inwoley, Walter Jaoko, Jack DeHovitz, Linda-Gail Bekker, Punnee Pitisuttithum, Robert Paris, Jennifer Serwanga, and Susan Allen.
This work was supported by the International AIDS Vaccine Initiative (IAVI), NIAID (grants AI33292 to D.R.B. and AI084817 to I.A.W.), the Ragon Institute, and the United States Agency for International Development (USAID).
The neutralization experiments on the 97-virus panel were done by Michael Seaman's group (Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA), and this work was funded through the Bill and Melinda Gates foundation (grant no. 38619).
The contents are the responsibility of the authors and do not necessarily reflect the views of USAID or the United States Government.
Some of the authors have a patent pending for the MAb PGV04.
Published ahead of print 15 February 2012
Supplemental material for this article may be found at http://jvi.asm.org/.