Profiling a large HIV-1 elite neutralizer cohort reveals remarkable CD4bs bNAb for HIV-1 prevention and therapy

Abstract

Administration of HIV-1 neutralizing antibodies can suppress viremia and prevent infection in vivo. However, clinical use is challenged by broad envelope sequence diversity and rapid emergence of viral escape^1-9. Here, we performed single B cell profiling of 32 top HIV-1 elite neutralizers to identify broadly neutralizing antibodies (bNAbs) with highest potency and breadth for clinical application. From 831 expressed monoclonal antibodies, we identified 04_A06, a new V_H1-2-encoded CD4 binding site bNAb with remarkable breadth and potency against extended multiclade pseudovirus panels (GeoMean IC₅₀ = 0.059 μg/ml, breadth = 98.5%, 332 virus strains). Moreover, 04_A06 was not susceptible to classic viral CD4bs escape variants and maintained full viral suppression in HIV-1-infected humanized mice. Structural analyses revealed that antiviral activity is mediated by an unusually long 11-amino acid heavy chain insertion. This insertion facilitates inter-protomer contacts and interactions with highly conserved residues on the adjacent gp120 protomer. Finally, 04_A06 demonstrated high activity against contemporaneously circulating viruses from the Antibody Mediated Prevention (AMP) trials (GeoMean IC₅₀ = 0.082 μg/ml, breadth = 98.4%, 191 virus strains) and in silico modeling for 04_A06LS predicted HIV-1 prevention efficacy of >93%. Thus, 04_A06 will provide unique opportunities for effective treatment and prevention strategies of HIV-1 infection.

Extended Data Fig. 1:. Gating strategy and single cell sorts of HIV-1-reactive B cell subsets

a, Individual FACS plots depicting sorting gates and frequencies of HIV-1-reactive, IgG⁺ B cells (in %) isolated from 32 donors. Numbers indicate the frequency of HIV-1 Env-reactive IgG+ B cells from the parental gate.

Extended Data Fig. 2:. Analysis of isolated heavy chain sequences

a, Clonal relationship of heavy chain sequences amplified from single HIV-1-reactive IgG⁺ B cells isolated from 32 donors. Individual clones are colored in shades of blue, gray, and white. In the center of each pie chart, numbers of productive heavy chain sequences are illustrated. Presentation of clone sizes are proportional to the total number of productive heavy chain sequences per clone. b, Dot plot bar graphs display the mean + SD.

Extended Data Fig. 3:. Neutralization and binding profile of serum, isolated and chimeric mAbs from donor EN02

a, Neutralizing activity of EN02 serum IgG against the HIV-1 global and b, f61 pseudovirus panel retrieved from. Right panels show delineation scores of f61 panel-based computational epitope mapping. c, Interference of indicated bNAbs with reference bNAbs targeting known epitopes on the HIV-1 Env trimer measured by competition ELISAs. d, Neutralization breadth (%) and potency (GeoMean IC₈₀) of representative bNAb 04_A06 (clone 9; red), 01_D03 (clone 11, blue); and 05_B08 (clone 7, green) against >245 pseudovirus strains. Samples were tested in duplicates. Neutralization data for reference bNAbs was retrieved from CATNAP database. e, The elongated heavy chain FWRH1 of 04_A06 was engrafted into clonally distinct bNAbs isolated from donor EN02 and CD4bs reference bNAbs. The antiviral activity of wildtype and chimeric versions of isolated and reference mAbs was determined against the global HIV-1 panel and common CD4bs escape pseudovirus variants. f, Breadth (%) and potency (IC₅₀) of the chimeric VRC07 antibody version against the 119 multiclade panel. g, Sequential treatment with the chimeric VRC07 antibody version (VRC07_{11aa ins}) in HIV-1_YU2-infected humanized mice following viral rebound during VRC07 monotherapy. This approach included maintaining VRC07 monotherapy while integrating VRC07_{11aa ins}. in the treatment regimen. Mice treated with a double dose of VRC07 (1mg) were included as a control group. Dashed lines in the top graphs indicate the lower limit of quantitation of the qPCR assay (LLQ) (784 copies/ml). Red lines display the average log₁₀ changes compared to baseline viral loads (day 0).

Extended Data Fig. 4:. Autoreactivity and antivital activity of isolated bNAbs from donor EN02

a, Reactivity of indicated antibodies against HEp-2 cells. Antibodies were tested at a concentration of 100 μg/ml. b, Binding profiles of isolated bNAbs from donor EN02 against indicated HIV-1 Env antigens. c, Neutralization activity of bNAb 04_A06 and reference bNAbs against the HIV-1 global pseudovirus panel. Samples were tested in duplicates. Neutralization data of reference bNAbs PGDM1400, CAP256V2LS, PGT121, 10-1074, VRC07_523-LS, N6, 1–18, 3BNC117, N49P7 and VRC01 were retrieved from CATNAP database. d, Neutralization coverage (%) and potency (IC₈₀) of 04_A06 against a panel of 50 patient-derived bulk outgrowth isolates. Neutralization data of reference bNAbs against replication competent viruses were retrieved from. Samples were tested in duplicates. e, Amino acid frequency at selected sites across 744 clade C sequences from the Los Alamos National Laboratory (LANL) HIV database is shown (top; letter height reflects frequency). The lower panels depict corresponding positions in plasma-derived single genome sequencing (SGS) env sequences from donor EN02. Red and green boxes highlight amino acid residues present in <1% and 1–5%, respectively, of LANL clade C sequences. Numbering corresponds to HIV-1 HXB2 reference strain. f, Neutralization sensitivity of selected pseudoviruses carrying EN02 env sequences shown in e.

Extended Data Fig. 5:. Cryo-EM data collection and processing

Representative micrographs, two-dimensional class averages, workflows, gold-standard Fourier shell correlation (GSFSC) curves, final density maps colored by local resolution, and particle orientation distributions for a, 05_B08 Fab / BG505_SOSIP.664, b, 01_D03 Fab / BG505_SOSIP.664, c, 04_A06 Fab / PGDM1400 Fab / BG505_SOSIP.664, and d, unbound BG505_SOSIP.664 datasets.

Extended Data Fig. 6:. Structural analysis of CD4bs antibodies from EN02 with other CD4bs antibodies and unbound BG505

a, EM density of unbound BG505_SOSIP.664 structure showing side and top views. b, Canonical interactions of CD4 and CD4bs bNAbs with Env: D368_gp120, Phe43 pocket, and N280_gp120 / R456_gp120. c, Close up view of 04_A06’s CDRH1 interactions (blue) with the primary gp120 protomer (gp120₁). d, EM density for gp120 residues 58–64 on the adjacent protomer (gp120₂) as observed in complex with 01_D03 Fab compared to unbound BG505_SOSIP.664. e, Crystal packing environment of 04_A06. f, Overlay of V_H-V_L domains in bound (from Fab-SOSIP cryo-EM structure) and unbound (from Fab crystal structure) structures.

Extended Data Fig. 7:. Structural analyses of PGDM1400

a, Sequence alignment of heavy and light chain V genes from bNAbs PGT145 and PGDM1400. Sulfated tyrosine residues located in the CDRH3 are indicated by a green “Y”. b, Structure overview showing EM density of BG505_SOSIP.664 in complex with 04_A06 and PGDM1400 Fabs. EM density for N160 glycans (top) and a Fab electrostatic surface calculation (bottom) for PGDM1400 (c) and PGT145 (d). Glycans labeled as glycan 1,2, and 3 correspond to the N160_gp120 glycans from each of the three Env protomers.

Extended Data Fig. 8:. Restriction of viral escape in vitro

a, Neutralizing activity of 04_A06 against BG505_T332N and common escape pseudovirus mutants. The panels indicate the IC₅₀ of 04_A06 against the respective pseudovirus variant. b, Logo plots illustrating the extent of neutralization escape caused by mutations in the HIV-1 Env strain BF520 for antibodies 04_A06, N6-LS, VRC07-523-LS, 1–18 and 3BNC117. The heights of individual letters represent the effect of that amino-acid mutation on antibody neutralization, with positive heights (letters above the zero line) indicating mutations that cause escape, and negative heights (letters below the zero line) indicating mutations that increase neutralization. Letters are colored by the effect of that mutation on pseudovirus cell entry, with yellow corresponding to reduced cell entry and brown corresponding to neutral effects on cell entry. The y-axis scales for each antibody are independent. Only key sites are illustrated. See https://dms-vep.org/HIV_Envelope_BF520_DMS_04-A06/htmls/04-A06_mut_effect.html, https://dms-vep.org/HIV_Envelope_BF520_DMS_04-A06/htmls/N6_LS_mut_effect.html, https://dms-vep.org/HIV_Envelope_BF520_DMS_04-A06/htmls/VRC07_523_LS_mut_effect.html, https://dms-vep.org/HIV_Envelope_BF520_DMS_04-A06/htmls/1-18_mut_effect.html, and https://dms-vep.org/HIV_Envelope_BF520_DMS_04-A06/htmls/3BNC117_mut_effect.html for interactive versions of the escape maps.

Extended Data Fig. 9:. Neutralization profile and predicted prevention efficacy against AMP trial pseudoviruses

Antiviral activity of 04_A06 and reference bNAbs against pseudoviruses generated from a, the HVTN703 and HVTN704 placebo group and b, HVTN703 intervention group, respectively. Bar graphs display the breadth (%) at the established threshold of protection (IC₈₀ ≤ 1 μg/ml). Dot plots illustrate the potency of tested bNAbs (IC₈₀) against each pseudovirus strain. Predicted HIV-1 PE of 04_A06 over time against HVTN703 intervention group pseudoviruses in comparison to VRC01. Left graph illustrates the PE of 04_A06 (red) and 04_A06LS (purple) under the assumption of 3BNC117 pharmacokinetics. The right graph displays the PE of 04_A06 and 04_A06LS under the assumption of 10–1074 pharmacokinetics. c, Illustration of the geometric mean PT₈₀ against placebo and intervention group viruses. The geometric mean PT₈₀ for each time point was determined by dividing the geometric mean of the predicted serum concentration of bNAbs in recipients at each time point during steady state (as simulated via PK modeling for each bNAb, as outlined in the methods section) by the geometric mean IC₈₀ of the bNAb against viruses present in the specified AMP trial. Calculations were performed over time after three 8-weekly infusions of 04_A06 at 30 mg/kg or a single infusion of 04_A06LS at 30 mg/kg. In the scenario involving the LS-modified version, predictions were made under the premise that 04_A06LS has a 2.5-fold higher half-life than 04_A06. Black lines indicate the PT₈₀ for VRC01. Data for reference bNAbs were retrieved from CATNAP database and.

Extended Data Fig. 10:. In vivo antiviral activity of 04_A06 in comparison to highly-active CD4bs bNAbs

a, Investigation of the antiviral activity of N49P7, N6, and 04_A06 monotherapy in HIV-1_YU2-infected humanized mice (NXG-HIS). Graphs illustrate the absolute HIV-1 RNA plasma copies/ml (top) and relative log₁₀ changes from baseline viral loads (bottom) after initiation of bNAb therapy. Dashed lines (top graphs) indicate the lower limit of quantitation of the qPCR assay (LLQ) (451 copies/ml). Red lines display the average log₁₀ changes compared to baseline viral loads (day −2). b, Alignment of plasma SGS-derived env sequences identified from day −1 (black bars) and after viral rebound from day 28 (red bars). c, Analyses of single HIV-1 plasma env sequences from HIV-1_YU2-infected humanized mice obtained on day 28 after bNAb treatment initiation. Total number of analyzed sequences is indicated in the center of each pie chart. Mice are labeled according to icon legends in a. Colored bars on the outside of the pie charts indicate mutations in Loop D.

Figure 1:. Large-scale isolation of mAbs from HIV-1 elite neutralizers

a, HIV-1 neutralizing serum activity and demographic characteristics of 32 HIV-1 elite neutralizers. Dot plots illustrate the neutralizing activity of purified serum IgG samples of each donor against the HIV-1 global pseudovirus panel. Age distribution, sex, ART status of donors and geographical origin of the samples are illustrated as pie charts and bar plots. Serum IgG samples were tested in duplicates. b, Neutralizing activity of isolated 831 mAbs against a screening pseudovirus panel of 6 viral strains from 4 different clades. Each antibody was evaluated at a concentration of 2 μg/ml. Antibodies that achieved greater than 50% neutralization were classified as neutralizing (red; those with <50% neutralization were categorized as non-neutralizing mAbs (grey). Antibodies were tested once. c, Illustration of the mAb antiviral activities. Pie charts display the proportion of nAbs, their levels of breadth, and their neutralization spectrum across HIV-1 clades. Center lines in dot plots in a, b, and c indicate means. d, VH gene segment usage and CDRH3 characteristics of 831 mAbs divided into nAbs (red) and non-nAbs (grey). V gene segments are ordered by the overall frequencies in all 831 antibodies. Bar graphs show fractions that were calculated for individual subsets. CDRH3 characteristics comprise lengths in amino acids, cumulative hydrophobicities (Eisenberg scale) and net charge. For comparison, NGS reference data from 48 healthy individuals (blue) is shown. e, VH mutational load of non-nAbs (grey) and nAbs (red). f, Correlation between VH mutations and neutralizing activity of antibodies. Pearson correlation coefficients r and two-sided p-values for nAbs (red), non-nAbs (grey) and all (black) mAbs are reported in the figure. Dashed lines show linear regressions for the respective subset. Black lines within bar graphs in d and e depict the center of combined fractions to indicate overrepresentation in either nAbs or non-nAbs. Dashed lines in d and e represent distribution means.

Figure 2:. bNAb coevolution with overlapping specificity for the CD4bs

Phylogenetic and functional analyses reveal co-existence of genetically-distinct CD4bs bNAbs in donor EN02. a, Antibody characteristics and alignment of 04_A06, 01_D03, and 05_B08 heavy chains to germline IGHV1–2 gene. b, Heatmaps illustrating the VH sequence similarities of isolated antibodies within (intradonor) and between (interdonor) donors in shades from blue to yellow. Donors of other V_H1–2-encoded CD4bs bNAbs used as comparators are displayed in the center of each grey square. Histogram outlining intra- and interdonor similarity distributions. Red line indicates intradonor similarity distribution for donor EN02; black line indications interdonor similarity distribution between donors EN02 and VH1–2-encoded CD4bs bNAbs from other donors; dashed line indicates the threshold for identification of clonally-related sequences. Shaded areas under the intradonor similarity distribution graph indicate non-clonal (blue) and clonal (yellow) sequences. c, Maximum-likelihood phylogenetic trees of isolated bNAbs encoded by IGHV1–2 (green, blue, red). Antibodies were tested for binding (mean AUC) to BG505_SOSIP.664 and neutralizing activity (GeoMean IC₅₀ and breadth) against a HIV-1 global pseudovirus panel and common CD4bs escape variants. Samples were tested in duplicates. Reference antibodies were validated for functionality in neutralization assays against the global HIV-1 pseudovirus panel, and only those with IC₅₀/IC₈₀ values deviating less than threefold from CATNAP reference data were included.

Figure 3:. 04_A06 demonstrates remarkable neutralizing potency and breadth

a, Illustration of the phylogenetic distribution of Env-pseudotyped viruses of the 12-strain global panel. Neutralizing profile of 04_A06 against the global panel compared to non-CD4bs bNAbs currently investigated in advanced clinical trials and to highly-active CD4bs bNAbs. Illustrated IC₅₀ values for each bNAb are sorted by increasing value. b, Neutralization coverage (%) and potency (IC₈₀) of 04_A06 against multiclade panels of 119 and 208 pseudoviruses, and a Clade C pseudovirus panel. Data are shown for identical virus strains from each panel for which reference neutralization data was available. c, Dot plot illustrating the neutralizing activity (IC₈₀, breadth) of 04_A06 compared with anti-HIV-1 bNAbs against 83 pseudoviruses. Neutralization data of bNAbs N6, 1–18, and 04_A06 were determined in the same laboratory. Curve graphs compare neutralizing activity of 04_A06 to current highly-active CD4bs bNAbs (N6, VRC07_523-LS, 1–18, N49P7) against panels of 332, 311, 332, or 293 pseudoviruses. d, Neutralization profile (IC₅₀, breadth) of 04_A06 compared with highly-active CD4bs bNAbs against a panel of 37 pseudoviruses resistant to CD4bs bNAbs. Data for reference bNAbs obtained from the CATNAP database if available. Otherwise bNAbs were tested in parallel to 04_A06. Reference antibodies were validated for functionality in neutralization assays against the global HIV-1 pseudovirus panel, and only those with IC₅₀/IC₈₀ values deviating less than threefold from CATNAP reference data were included. Breadth (%) was defined applying a cut-off of ≤10 μg/ml (a-c) and <5 μg/ml (d). Samples were tested in duplicates in single experiments. Geometric mean neutralization data are presented.

Figure 4:. Ultra-long eleven-amino acid insertion facilitates quaternary binding mode of 04_A06

a, EM maps showing side views of 04_A06 (blue), 01_D03 (red) and 05_B08 (gold) bNAb Fabs in complex with BG505_SOSIP.664 Env trimers. Insets illustrate a close up of bNAb interactions with the adjacent gp120 protomer (gp120₂). b, Canonical interactions of CD4 and CD4bs bNAbs with Env: D368_gp120, Phe43 pocket, and N280_gp120 / R456_gp120. See Extended data Fig. 7b for interactions with additional bNAbs including 01_D03 and 05_B08. c, Interactions between the CDRH1s of 04_A06 and 1–18, shown as blue and red cartoon representations, with the secondary (gp120₂) Env protomer, shown as a surface colored by percent conservation. d, Crystal structure of unbound 04_A06 Fab. Inset highlights the 04_A06 CDRH1, with electron density contoured at 1.5σ. e, Surface area buried by VRC01-class bNAbs on primary (gp120₁) or secondary (gp120₂) Env protomers.

Figure 5:. Maintained suppression of viremia and restriction of VRC01-class viral escape in vivo

a, Investigation of the antiviral activity of 04_A06, VRC01, and VRC07 monotherapy in HIV-1_YU2-infected humanized mice (NRG). Graphs illustrate the absolute HIV-1 RNA plasma copies/ml (top) and relative log₁₀ changes from baseline viral loads (bottom) after initiation of bNAb therapy. Dashed lines (top graphs) indicate the lower limit of quantitation of the qPCR assay (LLQ) (784 copies/ml). Red lines display the average log₁₀ changes compared to baseline viral loads (day −1). b, Analyses of single HIV-1 plasma env sequences from HIV-1_YU2-infected humanized mice obtained on day 28 after bNAb treatment initiation. Total number of analyzed sequences is indicated in the center of each pie chart. Mice are labeled according to icon legends in a. Colored bars on the outside of the pie charts indicate mutations in Loop D and beta23/V5 loop. c, Sequential treatment with 04_A06 in HIV-1_YU2-infected humanized mice following viral rebound during VRC01 monotherapy (from panel a). This approach included maintaining VRC01 monotherapy while integrating 04_A06 in the treatment regimen. Mice treated with a double dose of VRC01 (1mg) were included as a control group. Dashed lines in (top graphs) indicate the qPCR assay LLQ (784 copies/ml). Red lines display average log₁₀ changes compared to baseline viral loads (day 28). d, HIV-1 viral loads (right y-axis) and plasma bNAb concentrations (left y-axis) after treatment interruption on day 84. Blue line indicates the mean plasma bNAb concentration; red line displays geomean HIV-1 viral loads. Dashed lines indicate the qPCR assay LLQ (784 copies/ml). Mice are labeled according to icon legends in a. Grey icons refer to measured antibody plasma levels; green icons indicate viral loads. e, Alignment of plasma SGS-derived env sequences identified from day −1 (black bars), day 28 (red bars), and after viral rebound (blue bars; panel d) and sensitivity of the respective pseudovirus.

Figure 6:. Enhanced antiviral activity against AMP trial viruses and high HIV-1 prevention efficacy

a, Illustration of the neutralizing activity of 04_A06 compared to CD4bs bNAbs against representative pseudoviruses generated from the placebo arms of the two AMP trials: HVTN703/HPTN 081 in predominantly Clade B regions and HVTN704/HPTN 085 in predominantly Clade C regions^,,. Bar graphs display the breadth (%) at the established threshold of protection (IC₈₀ <1 μg/ml). Dot plots indicate the IC₈₀s of bNAbs against the individual pseudoviruses. Virus strains neutralized at the established threshold of protection (IC₈₀ <1 μg/ml) are depicted in red, whereas strains not neutralized at this threshold are shown in grey. Data for reference bNAbs were retrieved from the CATNAP database and. Antibodies were tested in duplicates. The center line in dot plots depicts the geometric mean. b, Serum concentrations of 3BNC117 and 10–1074 from a human clinical trial and in vivo pharmacokinetic profile of 04_A06 and 04_A06LS measured in hFcRn transgenic mice. For determination of the PK profile, mice received either a fixed dose of 0.5 mg of 04_A06 or 5 mg/kg of 04_A06LS. Serum concentrations (μg/ml) of 04_A06 (top), 3BNC117 (middle), and 10–1074 (bottom). Participants of the human clinical trial received either a single infusion of 3BNC117 and 10–1074 (10 mg/kg of each bNAb, blue), or three infusions of each bNAb (3 mg/kg, green; 10 mg/kg, red). Mice received a single intravenous injection of 0.5 mg of 04_A06. Serum concentrations were determined in duplicates by ELISA. Data of mice experiments are represented as mean ± SD. c, Predicted HIV-1 prevention efficacy (PE) of 04_A06 or VRC01 over time against AMP viruses. Left: PE of 04_A06 (red) 04_A06LS (purple) assuming of 3BNC117 PKs106. Right: PE of 04_A06 and 04_A06LS assuming 10–1074 PKs. PE was calculated over time after three 8-weekly infusions of 04_A06 or a single infusion of 04_A06LS (all at 30 mg/kg). For LS versions of antibodies, predictions were made assuming that 04_A06-LS has a 2.5-fold longer half-life than 04_A06. Black curves indicate the VRC01 PE in the AMP trials. Solid lines display the median and shaded areas the 95% prediction interval