Structural and mechanistic basis of neutralization by a pan-hantavirus protective antibody

Abstract

Emerging rodent-borne hantaviruses cause severe diseases in humans with no approved vaccines or therapeutics. We recently isolated a monoclonal broadly neutralizing antibody (nAb) from a Puumala virus-experienced human donor. Here, we report its structure bound to its target, the Gn/Gc glycoprotein heterodimer comprising the viral fusion complex. The structure explains the broad activity of the nAb: It recognizes conserved Gc fusion loop sequences and the main chain of variable Gn sequences, thereby straddling the Gn/Gc heterodimer and locking it in its prefusion conformation. We show that the nAb's accelerated dissociation from the divergent Andes virus Gn/Gc at endosomal acidic pH limits its potency against this highly lethal virus and correct this liability by engineering an optimized variant that sets a benchmark as a candidate pan-hantavirus therapeutic.

Fig. 1.. ADI-42898 binds a pH-sensitive quaternary epitope at the Gn/Gc interface.

(A) Schematic mechanistic model of hantavirus membrane fusion. Gn, light pink; Gc domain I, red; Gc domain II, yellow; Gc domain III, blue. TMIR, target membrane–interacting region. (B) Ribbon representation of x-ray structure of PUUV Gn^H/Gc in complex with an scFv fragment of ADI-42898. ADI-42898 heavy chain, green; ADI-42898 light chain, gray; Gc domains I to III are colored as in (A). ADI-42898’s epitope is shaped by three loops: Gn capping loop, pink; Gc bc loop, yellow; Gc cd loop, orange. (C) The ADI-42898:Gn^H/Gc contact surface in (B) was enlarged for clarity to depict amino acid residues involved in interface formation. ADI-42898 residues are numbered according to the Kabat scheme. CDR-H and CDR-L, complementarity-determining regions of the heavy and light chains, respectively. Dashed lines indicate hydrogen bonds. (D) Surface-shaded representation of the PUUV Gn^H/Gc complex with the ADI-42898 contact surface indicated as a green outline. Each amino acid residue was shaded according to the degree of sequence variability among hantavirus Gn/Gc proteins at that position based on a multiple sequence alignment (white to purple, 100 to 0% sequence conservation) (left). Amino acid side chains of Gn^H/Gc involved in ADI-42898 binding are colored according to their domain organization (right). (E) Gn^H/Gc binding to biolayer interferometry (BLI) sensors coated with ADI-42898 or ADI-42885 at the indicated pH values. Data are presented as averages ± SD, n = 3 from three independent experiments.

Fig. 2.. Mechanistic basis of viral entry blockade by ADI-42898.

(A) Capacity of Gn/Gc-specific mAbs to block rVSV-ANDV-Gn/Gc capture by a soluble protein comprising the first two extracellular cadherin repeats of the NWH cell surface receptor PCDH1 (sEC1–2) was measured by ELISA. Competitive inhibition by sEC1–2 was used as a positive control. Data are presented as averages ± SD, n = 4 to 6 from two or three independent experiments. (B) Capacity of ADI-42898 to block rVSV-ANDV-Gn/Gc attachment to human umbilical vein endothelial cells (HUVECs). Bound viral particles were enumerated by quantitative reverse transcription PCR (RT-qPCR) detecting rVSV genomic RNA (gRNA). CHS, convalescent hamster serum. Data are presented as averages ± SD, n = 4 from four independent experiments. (C) Capacity of ADI-42898 to inhibit PUUV or ANDV Gn/Gc–mediated fusion-infection. rVSV particles were preincubated with ADI-42898, followed by fusion-infection on HUVECs. Data are presented as averages ± SD, n = 6 to 12 from two to four independent experiments. (D) Heatmap of EC₅₀ (from PUUV and ANDV VLP dose-response binding curves) and IC₅₀ values (from rVSV-PUUV and -ANDV-Gn/Gc dose-response neutralization curves) derived by nonlinear regression analysis (see fig. S11). Data points are colored according to binding and neutralization potency. IgGs or Fabs with EC₅₀ or IC₅₀ values of >100 nM were designated as non-binding (NB) or non-neutralizing (NN). Also see data file S2 for EC₅₀ and IC₅₀ values. (E) Models of ADI-42898 IgG engagement of a local (Gn/Gc)₄ lattice based on the PUUV Gn/Gc^H:ADI-42898 structure (Fig. 1) are shown in orthogonal (top) and en face (bottom) views. scFv, gray; ADI-42898, green; Gn, red; Gc, yellow. (F) BLI sensorgrams for association of PUUV (top) and ANDV (bottom) Gn^H/Gc to ADI-42898 at pH 7.0, followed by dissociation of Gn^H/Gc at pH 7.0 and 5.5 (pH shift indicated by dotted line). Data are presented as averages, n = 4 from four independent experiments are shown. (G) Capacity of ADI-42898 to bind VLPs bearing PUUV (top) and ANDV (bottom) Gn/Gc at different pHs was measured by ELISA. Where indicated, Gn/Gc:ADI-42898 complexes were shifted from pH 7.0 to pH 5.5. Data are presented as averages ± SD, n = 4 from two independent experiments.

Fig. 3.. Affinity-matured mAbs, ADI-65533 and ADI-65534, show improved inhibition of virus entry.

(A) Sequence alignment of regions in ADI-42898 and its affinity-matured variants containing amino acid substitutions (green indicates Kabat numbering). (B) ADI-42898:PUUV Gn^H/Gc interface highlighting the amino acid changes in (A) and their contact residues in Gc (left). Potential interactions between the changed amino acid in CDR-H1 (T28E) with residues in the bc loops of PUUV and ANDV Gc are shown (middle). Potential interactions between the changed amino acid in CDR-H3 (G100aV) with residues in the cd loops of PUUV and ANDV Gc are shown (right). (C) Graphical representation of association (k_on at pH 7.0) and dissociation rate (k_off at pH 5.5) constants for mAb interactions with PUUV and ANDV Gn^H/Gc. Rates were isolated from BLI sensorgrams in fig. S13; data are presented as averages ± SD, n = 4 from four independent experiments are shown. (D and E) Capacity of mAbs to bind ANDV Gn/Gc–decorated VLPs at pH 7.0 (D) or in a pH 7.0-to-pH 5.5 shift regime (E) was measured by ELISA. Data are presented as averages ± SD, n = 4 to 6 from two or three independent experiments. (F) Capacity of Gn/Gc-specific mAbs to block rVSV-ANDV-Gn/Gc capture by sEC1–2 was measured by ELISA. Data are presented as averages ± SD, n = 6 from three independent experiments. (G) Capacity of ADI-42898, ADI-65533, and ADI-65534 to inhibit ANDV Gn/Gc–mediated fusion-infection. rVSV-ANDV-Gn/Gc particles were preincubated with mAbs, followed by fusion-infection of HUVECs. Data are presented as averages ± SD, n = 10 to 12 from four independent experiments.

Fig. 4.. Affinity-matured mAbs broadly neutralize hantaviruses.

(A) Potency of ADI-42898, ADI-65533, and ADI-65534 against ANDV infection of HUVECs. Data are presented as averages ± SD; n = 14 from seven independent experiments are shown. (B) Heatmap of IC₅₀ values from ANDV and PUUV dose-response neutralization curves [(A) and fig. S15] derived by nonlinear regression analysis. Data points are colored according to mAb neutralization potency. Also see data file S2 for IC₅₀ values. ^# Data from focus-reduction neutralization assays; *data from microneutralization assays.

Fig. 5.. Affinity-matured mAbs confer post-exposure protection in Syrian hamsters.

(A) Syrian golden hamsters were challenged with ANDV [200 PFU, intramuscularly (im)], followed by treatment with a single dose of mAb [2 or 0.5 mg/kg, intraperitoneally (ip)] 3 days after virus exposure. Mortality of hamsters was monitored for 28 days. Averages from one experiment, n = 6 per group. (B) In a time course study, Syrian golden hamsters were challenged with ANDV (200 PFU, im). At the indicated days after virus challenge, animals were euthanized, and virus titers in serum samples and lung tissues were assessed by plaque assay. Averages from one experiment are shown, n = 3 per group. * indicates that virus titers could not be determined. (C) Syrian golden hamsters were challenged with ANDV (200 PFU, im), followed by treatment with a single dose of mAb (25 mg/kg, ip) 6 or 7 days after virus challenge, respectively. Averages are shown from two experiments (treatment on day 6), n = 9 (mAb) and n = 11 (vehicle) or from one experiment (treatment on day 7), n = 6 (mAb) and n = 6 (vehicle). Comparisons are shown between untreated and mAb-treated animals and were analyzed by Mantel-Cox test: **P = 0.0021; ****P < 0.0001.