Precise targeting of HIV broadly neutralizing antibody precursors in humans

Abstract

A protective HIV vaccine will need to induce broadly neutralizing antibodies (bnAbs) in humans, but priming rare bnAb precursor B cells has been challenging. In a double-blinded, placebo-controlled phase 1 human clinical trial, the recombinant, germline-targeting envelope glycoprotein (Env) trimer BG505 SOSIP.v4.1-GT1.1, adjuvanted with AS01_B, induced bnAb precursors of the VRC01-class at a high frequency in the majority of vaccine recipients. These bnAb precursors, which target the CD4 receptor binding site, had undergone somatic hypermutation characteristic of the VRC01-class. A subset of isolated VRC01-class monoclonal antibodies neutralized wild-type pseudoviruses and was structurally extremely similar to bnAb VRC01. These results further support germline-targeting approaches for human HIV vaccine design and demonstrate atomic-level manipulation of B cell responses with rational vaccine design.

Fig. 1.. Immunogenicity of GT1.1 in adults without HIV.

All statistical tests: Wilcoxon rank-sum test: ns, not significant (p > 0.05); *, p < 0.05; **, p < 0.01; ****, p < 0.0001. (A) Overview of vaccination schedule and sample collection. (B) Example of BAMA data that were used to generate area under the curve (AUC) calculations. (C) GT1.1 binding AUC for placebo (p), low dose (LD) and high dose (HD) recipients at the indicated time points. (D) Design of knockout (KO) proteins to measure epitope-specific responses towards the apex and CD4-binding site (CD4bs). The colored region schematically represents the epitope that is knocked out; the mutations are changes from GT1.1 to knock out respective epitopes. (E) Example of binding (in background-removed, i.e., net, MFI) to GT1.1, GT1.1 apex KO and GT1.1 CD4bs KO proteins, which were used to calculate the binding ratios. (F) Differential binding ratios (GT1.1/GT1.1 apex KO, top panel; GT1.1/GT1.1 CD4bs KO, bottom panel) calculated based on net MFI values (see Methods). A ratio >2.5 (shaded area) was considered a significantly positive CD4bs-specific response. The numbers in the graphs represent the number of participants with a ratio >2.5 at each time point. (G) Serum neutralizing antibody (nAb) titers (50% inhibitory dilutions; ID50) at week 26 to GT1.1 pseudovirus. The numbers on the x-axis indicate the participants with a positive titer (ID50 > 10). (H) Serum nAb titers (ID50) to GT1.1, 426c.TM and CH505.gly4 viruses and their CD4bs-KO mutants that were designed to reduce or eliminate neutralization mediated by CD4bs-directed antibodies. The pie charts summarize the number of participants in each dose group that have a >2.5-fold decrease in nAb activity against the CD4bs-KO mutant viruses. (I) Electron microscopy-based polyclonal epitope mapping (EMPEM) analysis of week 26 samples from low and high dose recipients. Each horizontal line depicts one low dose recipient (left) or high dose recipient (right). The bars represent the frequency of participants with a certain polyclonal epitope response. The 3D reconstructions on the right show epitopes targeted in representative participants.

Fig. 2.. Induction of epitope-specific IgG memory B cells after GT1.1 vaccination.

Each dot represents one trial participant in all panels; all statistical tests: Wilcoxon rank-sum test: ns, not significant (p > 0.05); *, p < 0.05; **, p < 0.01; ****, p < 0.0001. (A-D) Frequency of GT1.1-specific IgG memory B cells (A), apex-specific IgG memory B cells (B), CD4bs-specific IgG memory B cells (C) and non-apex, non-CD4bs but GT1.1-specific IgG memory B cells (“other specificities”, (D)) present in peripheral blood mononuclear cells (PBMCs) from the placebo (p), low dose (LD) and high dose (HD) groups. The numbers on top of the graphs represent the median frequency of the indicated population in the total IgG memory B cell compartment. Wilcoxon rank-sum test: ns, not significant (p > 0.05); ***, p < 0.001; ****, p < 0.0001. (E) Number of apex-specific (yellow/orange) and CD4bs-specific (blue) B cell receptor (BCR) sequences obtained from high dose recipients at weeks 10 and 26. (F) Mean ± SEM immunoglobulin heavy chain variable gene (IGHV) usage for apex-specific B cells in both groups and timepoints combined. Frequency of gene usage was determined for each vaccine recipient; the bars represent the mean gene usage across individuals. IGHV genes were included when present in at least 10 vaccine recipients, and only the top 10 expressed IGHV genes are shown. (G) Mean ± SEM immunoglobulin heavy chain variable gene (IGHV) usage for CD4bs-specific B cells in both groups and timepoints combined as in (F). (H) Frequency of VRC01-class B cells, defined as a BCR signature of IGHV1–2 with a five residue CDRL3, in CD4bs-specific B cells that use IGHV1–2. (I) Frequency of VRC01-class memory B cells in total memory B cells. The numbers on top of the graph represent the median frequency of the VRC01-class B cells in the total IgG memory B cell compartment. Boxplots were constructed using positive VRC01-class responders. (J) Number of participants with a VRC01-class response, defined as having detected ≥1 VRC01-class IgG memory B cell. (K) Frequency of B cells in an expanded lineage. Expanded lineages are defined as consisting of ≥2 members. (L) Percent somatic hypermutation (SHM) expressed as a frequency of nucleotides mutated compared to germline sequences.

Fig. 3.. Properties of GT1.1-induced VRC01-class BCRs.

(A) Left and middle panels: total and VRC01-class amino acid mutations in IGHV1–2 of VRC01-class B cells isolated at week-10 (left) and week-26 (right) for low and high dose recipients combined. The values represent the number of B cells at a particular coordinate. The staggered black line shows the expected level of VRC01-class mutations expected to be introduced by random SHM in IGHV1–2. On track VRC01-class mutations are defined as mutations that are shared with bnAbs VRC01, PGV04, PGV20, VRC-CH31, 3BNC60 or 12A12 (41). Right panel: number of VRC01-class for each dose group. The coordinates of minimally mutated VRC01 (minVRC01) are shown in purple. minVRC01 is a synthetic bnAb that has the minimal number of SHM necessary for broad and potent neutralization (99). (B) Proportion of BCRs using light chain variable genes that are associated with VRC01-class bnAbs (colored), from this study, VRC01-class bnAbs and a control data set encompassing >500,000 paired HC/LC antibody sequences (OAS; see Methods) (42). (C) Sequence logos for five-residue CDRL3s, non-VRC01-class five-residue CDRL3s from OAS (42), VRC01-class BCRs from this study per group and time point, and VRC01-class bnAbs. The amino acids are colored by physiochemical characteristics and the logos are separated into kappa (left) and lambda (right) light chains. (D) Sequence logos for the conserved motif at CDRH3 position 103–5, for each of the groups in panel C. (E) Sequence logos for the conserved acidic motif at CDRH3 positions 98–100, for each of the groups in panel C. (F) Left panel: frequency of CDRL1 glycine substitutions in non-VRC01-class BCRs (grey) and VRC01-class BCRs (green). Each dot represents one individual. Triangles represent participants without CDRL1 glycine substitutions. Wilcoxon rank-sum test: ns, not significant (p > 0.05); **, p < 0.01. Right panel: example of a lineage showing a high frequency of glycine substitutions (red). (G) Left panel: frequency of CDRL1 deletions in non-VRC01-class BCRs (grey) and VRC01-class BCRs (green). Each dot represents one individual. Wilcoxon rank-sum test: **, p < 0.01. Inset: example lineage from one participant showing CDRL1 deletions (red dashes) of various sizes.

Fig. 4.. Binding and neutralization by VRC01-class mAbs after GT1.1 vaccination.

Each dot represents a single tested mAb. All statistical tests: Wilcoxon rank-sum test: ns, not significant (p > 0.05); *, p < 0.05; ***, p < 0.001. (A) Schematic of GT1.1 (grey) highlighting the CD4bs (yellow), surrounding potential N-linked glycosylation sites (white) and GT mutations T278R and G471S. GT1.1 lacks all four CD4bs glycans (absence indicated in white) but does have two GT mutations T278R and G471S. BG505 N276D has three CD4bs glycans (presence indicated in purple), and two GT mutations that are reverted to WT (indicated in purple). GT1.1 N276 has the N276 glycan restored and GT mutation R278T, indicated in purple, but lacks the other CD4bs-proximal glycans. BG505 SOSIP is a wild-type, fully glycosylated trimer. (B) Pie charts representing the proportion of mAbs that showed binding to the indicated Env trimer at least ≥3-fold above background at the highest concentration. (C) Half-maximal binding concentrations (EC₅₀s in μg/mL) of 276 VRC01-class mAbs against the trimers indicated on top of the graphs and in panel A. The dashed lines indicate the maximum mAb concentration tested. (D-F) As in panel A-C, but for clade B trimer AMC008. The triangles in panels E indicate mAbs that do not bind the indicated trimer. (G-H) As in panel B-C, but depicting the half-maximal inhibitory concentrations (IC₅₀ in μg/mL) in a pseudovirus (pseudovirus) neutralization assay against the indicated pseudoviruses in panel A. (I) Pie charts representing the proportion of mAbs that showed a ≥2.5-fold decrease in neutralizing concentration (IC₅₀) of pseudovirus 426c.TM versus its CD4bs KO 426.TM.KO. (J) Half-maximal inhibitory concentrations (IC₅₀ in μg/mL) in a pseudovirus neutralization assay against 426c.TM and its CD4bs KO 426c.TM.KO.

Fig. 5.. GT1.1-induced VRC01-class mAbs structurally resemble VRC01-class bnAbs and accommodate CD4bs glycans.

(A) Segmented cryo-EM maps (top) and atomic models (bottom) of five mAbs in complex with either GT1.1 (left) or fully glycosylated BG505 SOSIP (right) compared to a published cryo-EM map and model of VRC01-bound BG505 SOSIP (EMD 21111, PDB 6V8X). The resolution of each structure is shown below the mAb name. For the models, Env is shown as surface representation, and antibody chains are shown as ribbons. (B) Comparison of overall approach angle and orientation of GT1.1-elicited VRC01-class mAbs and VRC01 (PDB 6V8X). (C) Key molecular interactions between CD4bs residues N279/N280 and each VRC01-class mAb. Putative hydrogen bonds (atomic distance < 3.4 Å) are shown as dashed lines. Antibody numbering follows the Kabat scheme. (D) Comparison of CDRL3 loops and their loop D contacts, including the conserved Y₉₅E₉₆ motif (see also fig. 3C). (E) Overview of CDRL1 interactions with Env loop D. For GT1.1 complexes, which lack the N276 glycan, a semitransparent glycan is shown based on alignment of a VRC01-bound structure (PDB 6V8X).