Antibodies from a Human Survivor Define Sites of Vulnerability for Broad Protection against Ebolaviruses

Abstract

Experimental monoclonal antibody (mAb) therapies have shown promise for treatment of lethal Ebola virus (EBOV) infections, but their species-specific recognition of the viral glycoprotein (GP) has limited their use against other divergent ebolaviruses associated with human disease. Here, we mined the human immune response to natural EBOV infection and identified mAbs with exceptionally potent pan-ebolavirus neutralizing activity and protective efficacy against three virulent ebolaviruses. These mAbs recognize an inter-protomer epitope in the GP fusion loop, a critical and conserved element of the viral membrane fusion machinery, and neutralize viral entry by targeting a proteolytically primed, fusion-competent GP intermediate (GP_CL) generated in host cell endosomes. Only a few somatic hypermutations are required for broad antiviral activity, and germline-approximating variants display enhanced GP_CL recognition, suggesting that such antibodies could be elicited more efficiently with suitably optimized GP immunogens. Our findings inform the development of both broadly effective immunotherapeutics and vaccines against filoviruses.

Keywords: 15878; EBOV; Ebola virus; ebolavirus; ferret; human broadly neutralizing antibodies; infection; mouse; neutralization; protection.

Figure 1. Identification of bNAbs from a Human Survivor of EBOV Infection

(A) Heatmap of rVSV-SUDV GP neutralization activity by cross-reactive mAbs from a human EVD survivor mAb library. Activity at 330 nM (top) and 33 nM (bottom) of mAb is shown. mAbs with activity at both concentrations are highlighted in red. Unk., unknown competition group. See also Figure S1. (B–P) Neutralization of rVSV-GPs by mAbs with SUDV GP-specific neutralizing activity in (A). Means ± SD for three replicates are shown. See also Table S1.

Figure 2. Potency and Breadth of Authentic Ebolavirus Neutralization by Human bNAbs

(A–C) Neutralization of authentic EBOV (A), BDBV (B), and SUDV (C) by lead bNAbs identified in Figure 1. Means ± SD for three replicates are shown. (D) Summary of authentic ebolavirus neutralization. IC₅₀, mAb concentration that affords half-maximal neutralization of viral infectivity. Hyphens indicate no detectable neutralizing activity.

Figure 3. Negative-Stain Electron Microscopy of Fab: EBOV GPΔTM Complexes

(A–D) 3D reconstructions of four Fab:EBOV GPΔTM complexes are shown in transparent surface representation (gray) with a structure-based model of the GPΔTM trimer fitted into the density. Structural subdomains of the GP trimer are labeled as follows: GP1 glycan cap (aqua green), GP1 core (blue), GP2 (light blue), GP2 internal fusion loop region (white), and the stalk/HR2 region (yellow). The conserved glycan on Asn 563 in GP2 is indicated in orange. GP-bound Fabs are shaded in lilac. Top and side views are shown at left and right, respectively, in (A)–(D) and (I). A bottom view of ADI-16061:GP is shown. Orange stars, approximate termini of the β13- β14 loop, connecting the GP1 base and glycan cap. (E–H) Magnified views of (A)–(D) indicating the putative Fab contact sites on the GPΔTM trimer (lilac). Highlighted residues are located within 5 Å of the Fab model. Black outline, N-linked glycan on Asn 563. Red, residues conferring viral neutralization escape. See also Figures S2 and S3. (I) 3D reconstructions of ADI-15878 and ADI-15946 are superimposed with a structure-based model of the KZ52 Fab:GPΔTM complex (PDB: 3CSY) (Lee et al., 2008a) to highlight the angle of approach of each base binder. See also Figure S4.

Figure 4. Predicted bNAb Epitopes

The predicted GP:NAb contact surface is indicated as a yellow outline on a surface-shaded representation of the GPΔTM trimer model (PDB: 5JQ3). Each amino acid residue in GP1 (blue) and GP2 (red) is shaded according to the degree of sequence conservation among ebolavirus GP proteins at that position, based on a multiple sequence alignment (light to dark, 0% to 100% sequence identity) (see the STAR Methods for details). In (A)–(C), the N–linked glycan on GP2 residue Asn 563 is indicated as a translucent envelope and green balls-and-sticks. Antibody contact sites in (C), (E), (G), and (H) are derived from published structures of EBOV GP:Fab complexes.

Figure 5. Mechanistic Basis of Human NAb-Dependent Blockade of Viral Entry

(A) Capacity of NAbs to block EBOV GP_CL:NPC1 binding in an ELISA. Immobilized rVSV-EBOV GP_CL particles were incubated with increasing concentrations of each NAb and then with a pre-titrated concentration of purified NPC1 domain C-flag. Binding of NPC1 to GP_CL was detected with an anti-flag antibody. Means ± SD for three replicates are shown. (B) Capacity of NAbs to block exposure of the RBS mediated by GP → GP_CL cleavage. rVSV-EBOV GP particles were incubated with cathepsin L (CatL) at pH 5.5 in the presence of increasing concentrations of each mAb. GP:NPC1 binding was detected by NPC1 ELISA as in (A). Means ± SD for six replicates from two pooled experiments are shown. (C) Capacity of the NAbs to block EBOV GP → GP_CL cleavage. Reactions from (B) were resolved by SDS-PAGE, and GP1 was visualized by western blotting. Open triangle, uncleaved GP1. Filled triangle, cleaved GP1. Asterisk, partially cleaved GP1 species. Blot image has been cropped to show only relevant lanes. (D) Capacity of NAbs to neutralize viruses bearing uncleaved or cleaved GP (WT or N563A) in a single-round infection assay. Heatmaps of IC₅₀ values derived from NAb neutralization dose-curves of VSV-EBOV GP and CatL-generated VSV-EBOV GP_CL. Particles bearing EBOV GP(N563A) lack the conserved glycan at that residue. See also Figure S5. (E) Kinetic binding constants and apparent binding affinities (K_D^app) for NAb:GP and NAb:GP_CL interaction were determined by biolayer interferometry. See also Table S2. (F) Schematic model for steps in ebolavirus entry targeted by NAbs.

Figure 6. Germline Origin of ADI-15878 and Sequence Determinants of Its Neutralizing Activity

(A) Alignment of mature VH and VL sequences (ADI-15878^WT and ADI-15742^WT) with their closest human germline V and J gene segments and reconstruction of an inferred germline ancestor (ADI-15878^IGL) bearing mature CDR-H3 (Kabat numbering). Amino acid residues divergent from the inferred germline sequence are shaded in pink; substitutions shared by ADI-15878^WT and ADI-15742^WT are shaded in maroon. Open boxes, primer-induced mutations. CDR-H3 residues critical for viral neutralization are in red. (B) Comparison of K_D^app s for NAb binding to GP and GP_CL versus neutralization of rVSV-EBOV GP infection (IC₅₀). Blue and orange circles, ADI-15878^WT, ADI-15878^IGL, and their HC:LC chimeras versus EBOV GP and GP_CL, respectively. Green squares, ADI-15878 bearing mutations in CDR-H3. Pink lines, high and low limits of detection for binding K_D^app and neutralization IC₅₀, respectively. See also (C) and Figures S5 and S6. (C) Heatmaps for neutralization of rVSVs bearing ebolavirus GP and GP_CL by ADI-15878 variants. See also (B) and Figure S6. See also Figure S7.

Figure 7. In Vivo Protective Efficacy of Broadly Neutralizing Human mAbs

(A) BALB/c mice were challenged with mouse-adapted EBOV (EBOV-MA) and then treated with a single dose of each NAb or vehicle (PBS). (B) Weight loss curves for control and NAb-treated groups in (A). (C) Type 1 IFNα/β R^−/− mice were challenged with WT SUDV and then treated with two doses of each NAb or PBS. (D) Weight loss curves for control and NAb-treated groups in (C). (E) Ferrets were challenged with WT BDBV and then treated with two doses of each NAb or PBS. (F) Measurements of blood viremia in infected ferrets in (E) (GEQ, viral genome equivalents). †, deceased animal. See also Table S3. n, number of animals per group. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. All other comparisons to the PBS group were not significant.