Human neutralizing antibodies target a conserved lateral patch on H7N9 hemagglutinin head

Abstract

Avian influenza A virus H7N9 causes severe human infections with >30% fatality. Currently, there is no H7N9-specific prevention or treatment for humans. Here, from a 2013 H7N9 convalescent case in Hong Kong, we isolate four hemagglutinin (HA)-reactive monoclonal antibodies (mAbs), with three directed to the globular head domain (HA1) and one to the stalk domain (HA2). Two clonally related HA1-directed mAbs, H7.HK1 and H7.HK2, potently neutralize H7N9 and protect female mice from lethal H7N9/AH1 challenge. Cryo-EM structures reveal that H7.HK1 and H7.HK2 bind to a β14-centered surface and disrupt the 220-loop that makes hydrophobic contacts with sialic acid on an adjacent protomer, thereby blocking viral entry. Sequence analysis indicates the lateral patch targeted by H7.HK1 and H7.HK2 to be conserved among influenza subtypes. Both H7.HK1 and H7.HK2 retain HA1 binding and neutralization capacity to later H7N9 isolates from 2016-2017, consistent with structural data showing that the antigenic mutations during this timeframe occur at their epitope peripheries. The HA2-directed mAb H7.HK4 lacks neutralizing activity but when used in combination with H7.HK2 moderately augments female mouse protection. Overall, our data reveal antibodies to a conserved lateral HA1 supersite that confer neutralization, and when combined with a HA2-directed non-neutralizing mAb, augment protection.

Fig. 1. Isolation and characterization of human H7N9 mAbs in vitro.

a FACS depicting the staining and selection of H7-specific B cells from donor H7N9_HK2013 PBMCs. SSC-A, side scatter area; FSC-A, forward scatter area. b ELISA binding curves of the indicated mAbs to soluble recombinant H7N9 HA, with and without Endo H treatment, to H7N9 HA1 from 2013, 2016, and 2017, to H7N7 HA, and to 6 non-H7 HA or HA1. c Western blot of a cleaved H7 HA (molecular mass of 43 kDa for HA1 and 30 kDa for HA2) with mAb H7.HK2 or H7.HK4. d Neutralization curves of H7.HK mAbs against H7N9 2013 and 2017 pseudo viruses in MDCK cells. Data shown are mean ± SEM. Source data are provided in the Source Data file. Similar results were independently reproduced at least once.

Fig. 2. Structural analysis of H7.HK1 and H7.HK2 in complex with H7 HA trimer.

a Cryo-EM structures of H7.HK1 and H7.HK2 bound to H7 HA in the head region. b Top view of alignment of H7.HK1 and H7.HK2 complex structures. c Surface presentation of the H7.HK1 epitope (orange) on H7 HA1, with interacting CDRs shown. d H7.HK1 heavy chain forms seven hydrogen bonds and one salt bridge with H7 HA1. e H7.HK1 light chain forms one additional hydrogen bond with H7 HA1, and the interactions are stabilized by hydrophobic residues on the periphery of the light chain interface. f Modeling published structures of H7 HA1-binding antibodies (PDB: 6II4, 6II8, 6II9, 5V2A) onto the H7.HK1 bound structure, with an escape mutation R57K (green) reported for mAb 07-5F01. Competition ELISA with biotinylated H7.HK2 binding to the H7 HA monomer, in which unlabeled competing mAbs were titrated at increasing concentrations to evaluate the effect on H7.HK2 binding. Source data is provided in the Source Data file. g Sequence analysis of N = 1,483 H7 HA1s revealed a conserved lateral patch that largely overlaps with the H7.HK1 epitope. h Modeling the binding site of human receptor analogue LSTc (red) based on a previous crystal structure (PDB: 4BSE) onto H7 from the H7.HK1 complex, showing that H7.HK1 does not compete with sialic acid on the adjacent protomer (black). Alignment of the H7.HK1 complex with a previous crystal structure of H7 (PDB: 4BSE) shows that the 220-loop (pink) required for sialic acid binding (G218-G228) is disordered and would clash with the H7.HK1 light chain if it were present. Green asterisk symbol denotes the <2 Å clash between the CDR L1 N28 and the predicted location of P221 on HA1.

Fig. 3. Prophylactic and therapeutic effects of human H7N9 mAbs in mice i.n. challenged with 10 LD₅₀ of A/Anhui/1/2013 H7N9.

a Female mice were i.p. injected 100 μg (equivalent of 5 mg/kg) or 20 μg (equivalent of 1 mg/kg) of the indicated mAbs (as human IgG1 unless otherwise specified) one day before viral challenge; % survival (<20% weight loss) and % body weight of survived mice were plotted over time. b Female mice were i.p. injected 100 μg of the indicated mAbs one day after viral challenge; % survival and % body weight of survived mice were plotted over time. Arrows indicate the time when mAbs were administered. Control groups of a non-H7 placebo mAb and PBS were included. Data for each group were combined from 1 to 2 experiments and shown as mean ‒ SEM. Source data with P values from two-sided unpaired student’s t-test are provided in the Source Data file. Asterisk symbols denote P < 0.05, and # denote P < 0.1.

Fig. 4. H7.HK1 and two other lateral patch-binding antibodies define a conserved supersite of vulnerability on the HA head.

a Representation of H7.HK1, Fab6649, and 045-09-2B05 bound to their respective HAs indicates diverse angles of approach and heavy/light chain orientations towards the lateral patch supersite. b Comparison of epitopes of H7.HK1, Fab6649, and 045-09-2B05 centering on the lateral patch defines the lateral patch supersite (blue). c A subset of epitope surface, centered on the lateral patch, overlaps between H7.HK1, Fab6649, and 045-09-2B05 (magenta). Of shared epitope surface, a subset of epitope residues is conserved (positions 121, 126, 168, and 172 by H3 numbering). d The four lateral patch antibodies contact conserved residues using diverse chemistry. Table displays the type of interaction between each residue and antibody.