Human naïve B cells recognize prepandemic influenza virus hemagglutinins

Abstract

Understanding the naïve B cell repertoire and its specificity for potential zoonotic threats, such as the highly pathogenic avian influenza (HPAI) H5Nx viruses, may allow prediction of infection- or vaccine-specific responses. However, this naïve repertoire and the possibility to respond to emerging, prepandemic viruses are largely undetermined. Here, we profiled naïve B cell reactivity against a prototypical HPAI H5 hemagglutinin (HA), the major target of antibody responses. We found that the frequency of H5-specific human naïve B cells targeting the HA "head" domain was increased relative to cross-reactive B cells to a circulating seasonal H1N1 strain. We classified the isolated monoclonal antibodies (mAbs) by the HA epitopes engaged and found that selected mAbs neutralized H5N1 at germline. We determined a cryo-electron microscopic structure of one mAb in complex with H5 HA to define its epitope. Our study defines the naïve human B cell repertoire recognizing a potentially zoonotic HPAI.

Fig. 1.. H5 influenza-specific naive B cells isolation and characterization.

(A) HA structure with head, stem, and full-length soluble ectodomain (FLsE) marked; phylogenetic analyses of influenza HA with potential pre-pandemic influenza viruses marked (red asterisks) (B) H5 VN04 HA-specific naive B cells from healthy human donors were isolated by fluorescence-activated cell sorting (FACS) gated on CD19⁺IgD⁺IgG⁻CD27⁻; a representative plot from healthy donor 17 (HD17) is shown (see figs. S1C and S2 for detailed gating). H5 FLsE double-positive naive B cells (left) defined head and H1/H5 binding naive B cells (right) were single-cell sorted. (C) Flow plot of H5-specific class-switched (swIg) B cells from the same representative donor as in panel (B), gated on CD19+IgD−IgM− swIg B cells, which were analyzed but not sorted. Frequency of head- and H1/H5-specific events among (D) naïve B cells (%Naive [left panel], p = 0.002; %H5++ [right panel], p=0.0015) and (E) swIg B cells (%swIg [left panel], p= 0.0099); %H5++ [right panel], p=0.0009) from each donor (n = 7). ** p < 0.01; *** p < 0.001. (F) Variable heavy (left) and light (right) chain gene usage for H5 head (n=53; blue) or H1/H5 (n=35; grey) specific paired B cell sequences defined by index sort analysis from 7 donors. (G) H5 FLsE, head, and H1/H5-specific frequencies among total naive B cells screened for each donor.

Fig. 2.. Binding and neutralization summary of naive H5-specific antibodies.

(A) ELISA EC₅₀ (μg/ml) binding heat map of 26 naive mAbs expressed as IgGs to H5 VN04 HA FLsE, H5 VN04 head, H1 MI15 HA FLsE. IgGs are sorted for head and H1/H5 specificity assigned by index sort analysis. N.B., no binding. (B) Naive mAb binding to representative HAs from diverse H5 clades. Positive binding was defined as >0.5 OD₄₅₀ at 100μg/ml IgG. (C) Polyreactive ELISAs were performed at a single mAb concentration (15 μg/ml) in duplicate and polyreactivity was defined by an OD₄₅₀ ≥ 0.40 for two or more polyreactive antigens (dsDNA, cardiolipin, or human insulin).

Fig. 3.. Epitope mapping of naive antibodies by ELISA competition.

(A) H5 HA vulnerable sites (VSs): VS1 (green), VS2 (blue), VS3 (orange), and VS4 (red) (40), defined by H5-specific antibodies, 13D4 (PDB 6A0Z), 65C6 (PDB 5DUM), 100F4 (PDB 5DUR), and H5M9 (PDB 4MMH). (B) ELISA-based competition for naive H5 antibodies against previously isolated H5 antibodies from memory B cells. Data are presented as means of two technical replicates. (C) Epitope mapping of naive mAbs using yeast display with a heat map showing loss of binding to H5 VN04 mutants; darker shades indicate greater loss of binding. (D-F) Mutations enriched for loss of binding to (C) HD15_F05 mAb (blue), (D) HD16_D07 mAb (purple), and (E) HD16_D04 mAb (pink) are mapped onto an H5 VN04 HA protomer (PDB: 2FK0). Yeast display binding data are representative of two independent experiments.

Fig. 4.. Cryo-EM structure of HD16_D04 in complex with H5 HA.

(A) Overall, global reconstruction map at 3.1Å nominal resolution defines the binding epitope of HD16_D04 at the vestigial esterase domain. HA1 is colored light and HA2 dark grey on one protomer. One molecule of HD16_D04 variable light and heavy chains is colored blue and green, respectively. The Fab constant domain volume was removed for clarity. The inset shows the same HA:Fab complex along the 3-fold axis of the HA timer. (B) Close-up of the HA:Fab interface is shown using a derived atomic model. The interacting amino-acid side chains are displayed as sticks; the residue numbers are indicated and the ones on HA underlined. (C) Amino-acid sequence alignment of H5 HAs belonging to the 10 clades tested. Dots indicates sequence conservation. The consensus sequence and the conservation index among the sequences used is shown above the alignment. Blue and green spheres indicate the points of contact of the HD16_D04 light and heavy chains, respectively.

Fig. 5.. H5 reporter virus neutralization of naive antibodies.

(A) Percent neutralization of H5 VN04 reporter virus by isolated naive mAbs. Data represent mean percentage of two technical replicates at 100μg/ml mAb. (B) H5 VN04 FLsE ELISA EC₅₀ values plotted against percent neutralization (IgG at 100μg/ml) for a matched VN04 reporter virus strain. Head and H1/H5 mAbs are shown in grey and red, respectively, plotted with a line of identity (C) H5 VN04 reporter virus neutralization for selected head-directed naive mAbs with detectable neutralization at 100μg/ml with respective IC₅₀ and EC₅₀ values listed. Neutralization data represent means ± SD of four technical replicates.