A broadly neutralizing anti-influenza antibody reveals ongoing capacity of haemagglutinin-specific memory B cells to evolve

Abstract

Understanding the natural evolution and structural changes involved in broadly neutralizing antibody (bnAb) development holds great promise for improving the design of prophylactic influenza vaccines. Here we report an haemagglutinin (HA) stem-directed bnAb, 3I14, isolated from human memory B cells, that utilizes a heavy chain encoded by the IGHV3-30 germline gene. MAb 3I14 binds and neutralizes groups 1 and 2 influenza A viruses and protects mice from lethal challenge. Analysis of VH and VL germline back-mutants reveals binding to H3 and H1 but not H5, which supports the critical role of somatic hypermutation in broadening the bnAb response. Moreover, a single VLD94N mutation improves the affinity of 3I14 to H5 by nearly 10-fold. These data provide evidence that memory B cell evolution can expand the HA subtype specificity. Our results further suggest that establishing an optimized memory B cell pool should be an aim of 'universal' influenza vaccine strategies.

Figure 1. Reactivity of 3I14 against group 1 and group 2 influenza A viruses.

(a) Phylogenetic tree of the 18 HA subtypes of influenza A viruses based on amino acid sequences. Group 1 subtypes are listed in red and group 2 subtypes in blue. The amino acid distance scale bar denotes a distance of 0.1. (b) FACS analysis of 3I14 binding to a broad range of group 1 and group 2 HAs. 293T cells were transiently transfected with different HA-expressing plasmids, followed by staining with the purified scFvFc antibodies and APC-labelled mouse anti-human Fc antibody. Binding of 3I14 (red line), F10 (group 1-specific, green line), CR8020 (group 2-specific, blue line), FI6v3 (groups 1 and 2 specific, purple line), CR9114 (group 1 and 2 specific, orange line), and irrelevant mAb Fm-6 (anti-SARS virus, grey filled histogram) were analysed by flow cytometry. Data represent a representative experiment from three independent experiments. (c) 3I14 scFvFc Ab binding (K_d values) to recombinant HAs representative of group 1 (red) or group 2 (blue) subtypes. Data represent a representative experiment from three independent experiments. (d) 3I14 scFvFc Ab neutralization (IC₅₀ values) of infectious viruses group 1 (red) or group 2 (blue) subtypes. 3I14 was represented by squares; anti-group 1 mAb F10 was represented by triangles. Graphs used for IC₅₀ values determined by averaging neutralization titre of two to three independent experiments. (e) 3I14 scFvFc neutralization (IC₅₀ values) of pseudoviruses representative of group 1 (red) or group 2 (blue) subtypes. These data represent average neutralization titres of two to three independent experiments. Anti-group 1 mAb F10 scFvFc was used for reference. FACS, fluorescence-activated cell sorting.

Figure 2. 3I14 cross-competes with binding by other anti-stem bnAbs, FI6v3, CR9114, 39.29, F10 and CR8020 to H3 or H5.

5 μg ml⁻¹ H3-BR07 or H5-VN04 protein immobilized on ELISA plates were incubated with twofold serial dilution of 3I14 Fab from 80 to 0.3 nM mixing with other scFvFc Abs at 5 nM. The binding of scFvFc Abs was detected using HRP-conjugated mouse anti-human CH₂ antibodies. 3I14 Fab strongly inhibits the binding of CR8020, CR9114, FI6v3 and 39.29 to H3-BR07 but not with E730 (a–d). 3I14 partially inhibits the binding of F10 and 39.29 to H5-VN04 but does not inhibit the binding of 2 A (e,f). HRP, horseradish peroxidase

Figure 3. Protective efficacy of 3I14 in mice.

Group of five mice were treated with 25 or 5 mg kg⁻¹ doses of purified IgGs given intraperitoneally 24 h before lethal challenge by i.n. inoculation with H7N7-NL219, H7N9-AH13, H3N2-BR07-ma or H5N1-VN04 influenza viruses (∼10 LD₅₀). (a) Survival (%) and (b) body weight change (%) of mice that treated with bnAb 3I14 and group 1 control mAb F10. Data represent mean change in body weight of five mice per group compared with their baseline body weight.

Figure 4. 3I14 blocks trypsin-mediated HA maturation and pH-dependent conformational changes.

(a) Trypsin Cleavage Inhibition Assay. 0.4 μg recombinant H3-histidine (H3-BR07) was incubated in the presence of 2.5 μg 3I14 or Fm-6 IgG1, or in the absence of antibody in Tris-HCl buffer at pH 8.0 containing 2 μg ml⁻¹ Trypsin at 37 °C. Trypsin digestion was stopped at several time-points by boiling the sample in a 100 °C water bath. Samples were run on 10% reduced SDS–PAGE and blotted using a HisProbe-HRP Abs. Data represent a representative experiment from three independent experiments. (b) 3I14 IgG1 prevented by low-pH triggered conformational rearrangements on the surface-expressed H3-A2/68 and H3-BR07. Upper panels show four various conformations of HA: uncleaved precursor (HA0, left); trypsin in neutral pH cleaved (mature HA, left middle); fusion pH cleaved (mature HA, right middle) and trimeric HA2 (HA2, right). The conformation rearrangements of surface-expressed H3 were detected by FACS staining of 3I14 (solid bars) and the head binding control mAb E730 (open bars). Binding is expressed as the percentage of binding to untreated HA (HA0). For this antibody inhibition assay, H3 was pretreated without mAb, with 3I14, or with control Ab, Fm-6 IgG1 before exposure of the cleaved HAs to pH 4.9. Data represent mean+s.d. of three independent experiments. SDS–PAGE, SDS–polyacrylamide electrophoresis.

Figure 5. 3I14 mediates ADCC.

3I14 and other anti-stem bnAbs, FI6v3, CR9114, 39.29, F10 and CR8020 induced antibody-dependent cellular cytotoxicity (ADCC) in H3- and H5-expressed 293T cells. Upper panel, 2 × 10⁴/well H3-expressed 293T cells were attached to the plates before assay, and the medium was then replaced with 0.5% FBS low IgG serum assay buffer. Different bnAbs at concentration of 10, 5, 2.5 and 1.25 μg ml⁻¹ were added to each well. After 1 h, PBMCs were added at 1.2 × 10⁵ cells/well to assay plates and incubated for 6 h. The supernatants were harvested and detected using LDH cytotoxic kit by ELISA. Lower panel, H5-expressed 293T cells were seeded as target cells. Bars represent mean±s.e.m. P value was calculated using two-way ANOVA, compared with F10 (upper) and CR8020 (lower). ‘*' represents P value for each comparison <0.0001. Data represent a representative experiment from three independent experiments. ANOVA, analysis of variance.

Figure 6. Modelling of 3I14 and docking with H3/H5.

(a) The superimposition of 3I14 model with FI6v3, 39.29 and MAb 3.1. The bnAbs are displayed in ribbon representations. The heavy chain is in blue and the light chain is in cyan. The HCDR3s and LCDR1s are indicated by the hatched ovals. The residues in HCDR3 of 3I14, FI6v3, 39.29 and MAb 3.1 are coloured in red, magenta, yellow and green, respectively. (b) The complex structures of IGHV3-30-derived bnAbs with HAs. The epitope residues on the HAs are displayed in surface representation and the CDR loops of bnAbs are shown are shown as ribbons. HA1 of HA monomer is in wheat, HA2 is in salmon and epitope residues are labelled as red. The key residues L38 and K39 are coloured in yellow. Heavy chain CDRs of bnAbs are in blue and light-chain CDRs are in cyan. 3I14 was homology modelled using the antibody-modelling module in BioLuminate and the model was superimposed to H3/39.29 complex structure before docking with RosettaDock. Other three IGHV3-30 bnAbs, FI6v3, 39.29 and MAb 3.1 were downloaded from Protein Data Bank. (c) The interaction of D94 in 3I14 with H3/H5. H3 is shown in cyan with K39 shown as stick; H5 is shown in green with E39 shown as stick; 3I14 is shown in orange in H3/3I14 model and in yellow in H5/3I14 model with D94 shown as stick.

Figure 7. 3I14 and VLD94N variant neutralized HA pseudotyped virus H5N1-VN04 and infectious virus H3N2-BR07.

The 3I14 (black) and VLD94N variant (red) neutralized pseudotyped virus H5N1-VN04 (a) and H3N2-BR07 virus (b). These data represent average neutralization titres of two to three independent experiments.