A Prominent Site of Antibody Vulnerability on HIV Envelope Incorporates a Motif Associated with CCR5 Binding and Its Camouflaging Glycans

Abstract

The dense patch of high-mannose-type glycans surrounding the N332 glycan on the HIV envelope glycoprotein (Env) is targeted by multiple broadly neutralizing antibodies (bnAbs). This region is relatively conserved, implying functional importance, the origins of which are not well understood. Here we describe the isolation of new bnAbs targeting this region. Examination of these and previously described antibodies to Env revealed that four different bnAb families targeted the (324)GDIR(327) peptide stretch at the base of the gp120 V3 loop and its nearby glycans. We found that this peptide stretch constitutes part of the CCR5 co-receptor binding site, with the high-mannose patch glycans serving to camouflage it from most antibodies. GDIR-glycan bnAbs, in contrast, bound both (324)GDIR(327) peptide residues and high-mannose patch glycans, which enabled broad reactivity against diverse HIV isolates. Thus, as for the CD4 binding site, bnAb effectiveness relies on circumventing the defenses of a critical functional region on Env.

Figure 1. Diversity of antibodies targeting the high-mannose patch epitope

(A) IAVI Protocol G donors (14, 82, 26) were screened for neutralization breadth and potency on an 6-virus indicator panel and compared to previously identified donors (36 and 17). Serum neutralization IC₅₀ values are listed for each donor (above). Donor sera were screened on both wild-type (WT) and N332A pseudotype virus and the percent loss of neutralization values for each isolate are listed for the new donors (below). Percent loss of neutralization was calculated using the formula 100*((WT_IC50 − N332A_IC50)/WT_IC50). (B) WT and N332A 92BR020 gp120 were conjugated onto magnetic beads and used to adsorb neutralizing antibodies in donor sera before testing in heterologous neutralization assays with isolate IAVI C22. Isolates that demonstrated the largest differential in neutralization between WT and N332A constructs were advanced as baits for antigen-specific cell sorting. (C) Single IgG+ memory B cells were antigen-selected by flow cytometry into lysis buffer. B cells were selected for phenotype CD3−/CD8−/C14−/IgM−/CD19+/CD20+/CD27+/IgG+. Sorted cells were then reverse transcribed into cDNA followed by PCR with gene-specific primers. (D) Variable heavy and light chain gene information, percent mutation from germline, and CDRH3 sequence, and insertions/deletions are tabulated for isolated antibodies (PGDM11, PGDM14, PGDM21, and PGDM31) and compared to previously isolated antibodies (2G12, PGT121, PGT128, PGT130, and PGT135). Antibody families are differentiated by color.

Figure 2. High-mannose patch bnAbs isolated from different donors show differences in neutralization breadth and potency

Antibodies isolated from donor 14 (PGDM11, 12, 13, 14), donor 82 (PGDM21) and donor 26 (PGDM31) were evaluated for breadth and potency on a 6-virus panel with the N332 glycan site intact (A) and on the corresponding panel of viruses with the glycan site removed by alanine mutagenesis (B). Antibodies that produced in sufficient yield (PGDM11, PGDM12, PGDM21, and PGDM31) were evaluated for neutralization breadth (C) and potency (D) on a 106-virus panel.

Figure 3. Differences in epitope recognition of glycans and the C-terminal end of V3 loop

(A) Antibodies PGDM12 (donor 14) and PGDM21 (donor 82) were evaluated for neutralization activity against double glycan site mutants at or near the high-mannose patch epitope. Listed values show fold difference in neutralization IC₅₀. LT indicates low viral titers and ‘enhance’ denotes enhancement of neutralization IC₅₀ values compared to WT. (B) PGDM12 and PGDM21 were evaluated for binding to a glycan array. PGDM21 shows binding to Man₉GlcNAc₂ and various complex-type glycans. PGDM12 shows binding only to Man₉GlcNAc₂. (C) Single alanine mutants of the ³²²IIGDIRQAH³³⁰ residues at the C-terminal base of the V3 loop were produced for isolates 92BR020 and JR-CSF and tested for neutralization by the listed antibodies. ND denotes not determined.

Figure 4. Mapping overlapping antibody epitope footprints

Pairwise alanine mutants were created for N137, N156, N301, and N332 glycans and the ³²⁴GDIRQAH³³⁰ residues of the V3 loop. Virus mutants were then tested for neutralization by PGT121, PGT124, PGT128, PGT130, PGDM12, PGDM21 and, as a control, the CD4 binding site antibody 12A12. Epitope summaries for all of the tested high-mannose patch antibodies are shown and are derived from neutralization data shown in Figure S1 and Table S5. Dependence on residues are highlighted by color and degree of dependence is shown by relative letter sizes for residues and glycan sites.

Figure 5. Functional conservation of the ³³⁴GDIRQAH³³⁰ peptide site

(A) CXCR4-tropic viruses are more resistant than CCR5-tropic viruses to neutralization by high-mannose patch antibodies. All viruses were grown in HIV-naïve PBMCs and tested for neutralization in a TZM-bl luciferase assay. For each virus, missing glycan sites at N301 or N332 and mutations from the canonical GDIRQAH sequence are shown. V2-apex antibodies PGT145, PG9, and PG16 as well as the CD4 binding site antibody VRC01 were included for comparison. ND = not determined. IC₅₀ values are reported in μg/ml (B) A CCR5 N-terminal peptide-Fc construct was tested for binding to gp120 alanine mutants and showed dependence on the N301 glycan sequon and the ³²⁵DIR³²⁷ residues of gp120. Gp120 mutants were prepared from pseudovirions and captured via an anti-C5 antibody on ELISA plates. Due to low affinity, the CCR5 peptide-Fc was biotinylated and tetramerized with streptavidin before binding to captured gp120 mutants. The CD4i antibody 17b was included for comparison. All values are normalized to the V3-specific antibody F425 (B4e8). Soluble CD4 (sCD4) was added to a final concentration of 5 μg/mL where indicated. End-point OD405 values that showed less than 50% binding relative to wild-type (WT) are highlighted in red.

Figure 6. V3-glycan antibodies allosterically inhibit CD4 binding

(A) Sera from 21 donors of the IAVI protocol G cohort were tested for neutralization on a 6-virus cross-clade panel of pseudoviruses with and without the glycan at the N332 position. Listed values are serum neutralization IC₅₀. Neutralization titers that showed higher potency in the absence of the N332 glycan compared to the wild-type virus are highlighted with a bold box, and the percentages of viruses that resulted in enhanced neutralization potency in the absence of the N332 glycan site are listed. Serum from the PGT121 donor (N332 glycan dependent) was included as a control. (B) Among 42,715 HIV Env sequences in the Los Alamos database, 23,158 (54%) have the “GDIRQAH” sequence, while 19,557 (46%) deviate at D325, R327, or H330 residues. Using 46% as a baseline measurement of mutation at these residues by chance, every glycan site on Env was then evaluated for greater mutation (> 0) or greater conservation (< 0) of these residues in the absence of individual glycan sites. (C) Key contacts to residues ³²⁴GDIRQAH³³⁰ on gp120 by GDIR-bnAbs and CD4i antibodies. Overlaid liganded structures of PGT122, PGT124 and PGT128 (shades of blue) with key contacts to the ³²⁴GDIRQAH³³⁰ residues on gp120 highlighted in red. These structures are overlaid with liganded structures of CD4i antibodies 17b, X5, and 412d (shades of yellow) with key contacts to the ³²⁴GDIRQAH³³⁰ residues on gp120 highlighted in green. Arg327 on gp120 for all structures is shown as sticks. All structures are aligned on gp120. (D) High-mannose patch antibodies were tested for competition with sCD4 on isolate JR-FL E168K/N192A Env displayed on the surface of 293T cells. The CD4 binding site antibody 12A12 was included as a positive control and the V3-specific antibody F425 was included as a negative control.

Figure 7. Epitope recognition on native Env by V3-glycan antibodies

(A) Negative-stain EM of antibody Fabs on BG505 SOSIP.664. X-ray structures of gp120 liganded with PGT124 (PDB 4R2G), PGT122 (PDB 4TVP) and PGT128 (PDB 4TYG) were superimposed onto gp120 in the unliganded trimer volume using the ‘MatchMaker’ function in Chimera. All glycans shown are part of the 4TVP X–ray structure. For PGDM14 and PGDM21, the relative orientations of both Fabs were mapped onto a low resolution surface volume generated from the BG505 SOSIP.664 X-ray structure (PDB 4TVP). The reconstructions of PGDM14 (EMD-8181) and PGDM21 (EMD-8182) were superimposed onto this trimer volume using the ‘Fit in Map’ function in Chimera (https://www.cgl.ucsf.edu/chimera/). (B) Summary table of all GDIR-glycan bnAb families. Donor ID refers to the IAVI Protocol G donor ID (Simek et al., 2009), ND indicates not determined. (C) Proposed recognition of overlapping epitopes between V3-glycan antibodies and CD4i antibodies. In addition to making contacts to the ³²⁴GDIRQAH³³⁰ peptide region, CD4i antibodies require formation of the bridging sheet on gp120 following CD4 engagement in order to bind and therefore can only bind open conformations of the Env trimer. V3-glycan antibodies are capable of binding the immunogenic ³²⁴GDIRQAH³³⁰ peptide region on native Env in the absence of CD4 by penetrating the glycan shield and also directly binding the surrounding glycans.