Broadly Cross-Reactive, Nonneutralizing Antibodies against Influenza B Virus Hemagglutinin Demonstrate Effector Function-Dependent Protection against Lethal Viral Challenge in Mice

Abstract

Protection from influenza virus infection is canonically associated with antibodies that neutralize the virus by blocking the interaction between the viral hemagglutinin and host cell receptors. However, protection can also be conferred by other mechanisms, including antibody-mediated effector functions. Here, we report the characterization of 22 broadly cross-reactive, nonneutralizing antibodies specific for influenza B virus hemagglutinin. The majority of these antibodies recognized influenza B viruses isolated over the period of 73 years and bind the conserved stalk domain of the hemagglutinin. A proportion of the characterized antibodies protected mice from both morbidity and mortality after challenge with a lethal dose of influenza B virus. Activity in an antibody-dependent cell-mediated cytotoxicity reporter assay correlated strongly with protection, suggesting that Fc-dependent effector function determines protective efficacy. The information regarding mechanism of action and epitope location stemming from our characterization of these antibodies will inform the design of urgently needed vaccines that could induce broad protection against influenza B viruses.IMPORTANCE While broadly protective antibodies against the influenza A virus hemagglutinin have been well studied, very limited information is available for antibodies that broadly recognize influenza B viruses. Similarly, the development of a universal or broadly protective influenza B virus vaccine lags behind the development of such a vaccine for influenza A virus. More information about epitope location and mechanism of action of broadly protective influenza B virus antibodies is required to inform vaccine development. In addition, protective antibodies could be a useful tool to treat or prevent influenza B virus infection in pediatric cohorts or in a therapeutic setting in immunocompromised individuals in conjugation with existing treatment avenues.

Keywords: ADCC; HA; MAb; influenza B.

FIG 1

Phylogenetic tree of influenza B virus HA and immunization strategy. (A) Influenza B virus HA amino acid sequences were aligned and rooted to the ancestral influenza B/Lee/1940 virus HA. The ancestral strains (orange) prior to divergence, antigenically distinct B/Victoria/2/1987-like strains (green), and B/Yamagata/16/1988-like strains (purple) are indicated. Stars indicate recombinant HAs or purified viruses used to test the broad binding capabilities of our MAbs. The scale bar represents a 1% difference in amino acid sequence identity. Sequences were obtained on FluDB or GISAID, and the tree was generated in Clustal Omega and visualized in FigTree. (B) Generation of broadly reactive MAbs against influenza B virus HA. This schematic highlights the strategy used to generate anti-influenza B virus HA MAbs through hybridoma technology. PEG, polyethylene glycol.

FIG 2

Anti-influenza B virus HA MAbs are broadly cross-reactive and nonneutralizing in vitro. (A) Binding profiles of anti-influenza B virus HA MAbs. The isolated MAbs display broad binding profiles for ancestral (orange), B/Victoria/2/1987-like (green), and B/Yamagata/16/1988-like (purple) lineage viruses as tested by ELISAs against purified viruses. Shown are the quantitative endpoint titers. The negative-control (Neg. Ctrl) antibody used is a MAb directed to influenza A virus H6 HA (MAb 8H9), and the positive control (Pos. Ctrl) is a MAb specific for the influenza B virus NA (MAb 4F11). (B) Neutralization activity of anti-influenza B virus HA MAbs. The MAbs were found to be nonneutralizing in plaque reduction neutralization assays. The IC₅₀ values were determined using Prism (GraphPad) based on the reduction in plaque number. A neutralizing anti-influenza B HA MAb was used as a positive control, and MAb 8H9 (anti-H6) was used as a negative control.

FIG 3

Binding of MAbs to cells expressing cH8/B HA. The binding of MAbs to MDCK cells stably expressing cH8/B (H8 head, influenza B virus HA stalk) was tested by immunofluorescence. An anti-H8 MAb, KL-H8-1A7, against the H8 head of the cH8/B construct was used as a positive control, and KL-BNA-4F11, an anti-influenza B NA MAb, was used as a negative control; a secondary-only stain is shown. (A and B) Binding to cells expressing cH8/B_ala (two alanines at the head-stalk interface, as found in influenza B virus HA) (A) and binding to cells expressing cH8/B_cys (two cysteines at the head-stalk interface, as found in influenza A virus HA) (B).

FIG 4

The majority of the MAbs bind to linear epitopes on the influenza B virus HA, most of which are located in the stalk domain. (A) Binding to full-length influenza B/Lee/1940 virus recombinant HA under reducing denaturing conditions. MAbs that bound to full-length B/Lee/1940 HA in a Western blot analysis following a reducing SDS-PAGE likely recognize linear or microconformational epitopes on the HA, while an absence of binding indicates that the target epitope is likely conformational in nature and was denatured during the assay. Fifteen of 22 MAbs appear to bind to nonconformational epitopes. An irrelevant H7 HA was used to test for nonspecific binding, and an anti-hexahistidine antibody was used as a positive control. (B) Determination of the binding region for MAbs targeting nonconformational epitopes. MAbs that bound to linear or microconformational epitopes were used to probe against different fragments of the influenza B/Lee/1940 virus HA, as indicated. For reference, the variability across different influenza B virus HAs used in this study is also shown to highlight the highly conserved nature of the stalk domain. The fragments were expressed in HEK293Ts using pCAGGS or pEGFP-C1 expression vectors and tagged with a hexahistidine tag or GFP, respectively, and the cell lysates were used for Western blot analyses. For fragments expressed in pCAGGS (head and stalk III), an anti-hexahistidine tag antibody was used as a positive control, and the control lysate was from HEK 293Ts transfected with pCAGGS empty plasmids. An anti-EGFP antibody was used as a positive control for fragments expressed in the pEGFP-C1 backbone (stalk I, stalk II, and long alpha-helix), and the control lysate was from cells transfected with pEGFP empty plasmids which expressed EGFP protein. (C) Binding to HA pretreated under various pH conditions and/or with a reducing agent. The MAbs displayed different binding profile changes in the ELISA depending on the condition under which the viruses were pretreated: low pH or with DTT, a reducing agent. An increase in binding post-low-pH pretreatment implies that the target epitope is more exposed in the postfusion conformation, while a loss of binding under reducing conditions implies that the target epitope was completely denatured or was located on HA1 which is removed under these conditions. The control MAb CR8059 binds to the head domain, while MAb CR9114 binds to a conformational epitope on the stalk domain. The percent change in area under the curve (from the binding curves generated from absorbance at 490 nm) relative to neutral pH (blue) is charted for each MAb, and the various conditions are indicated (pH 4.4, red; pH 4.4 with 0.2 M DTT, green).

FIG 5

Prophylactic protection from viral challenge in the mouse model and corresponding in vitro ADCC induction. (A) Five mice per group were given 5 mg/kg of the MAb intraperitoneally 2 h prior to virus challenge with 5 mLD₅₀ of B/Malaysia/2506/04 (V). The MAbs were classified as fully protective, partially protective, or nonprotective based on survival, and the weight loss and Kaplan-Meier survival plots are shown for each group. The dashed line indicates a cutoff of 75% of the initial weight, the humane endpoint. Twelve of 22 MAbs were fully protective in this challenge model. MAb 8H9 (anti-H6 MAb) was used as a negative control at the same dose and is commonly indicated in each group (the control group was shared between the panels). (B) Two MAbs from each category of protective efficacy were selected and administered to mice in a similar prophylactic setting. These mice were challenged with 5 mLD₅₀ of B/Florida/04/06 (Y) virus, and relatively similar weight loss and survival profiles were observed in these mice. (C) In vitro antibody-dependent cell-mediated cytotoxicity reporter assay. The MAbs were tested in an antigen-specific ADCC reporter assay (Promega), and the fold induction of a luciferase-based readout for each MAb was indicated over the negative control (MAb 8H9). The MAbs are divided into the same three groups of fully (filled shapes; left), partially (half-filled shapes; middle), and nonprotective (nonfilled shapes; right) and are indicated with the same symbols as in panel A.

FIG 6

Correlation between antibody characteristics and in vivo protection. (A) Comparison of ADCC reporter activity in vitro and protective efficacy against challenge in vivo in mice. ADCC activity in vitro (y axis) is graphed against the degree of protection conferred by these MAb prechallenge from the data displayed in Fig. 4. Overall, MAbs that are completely protective in vivo in the prophylactic setting prior to influenza B virus challenge appear to robustly induce ADCC in vitro (**, P = 0.0022). ns, nonsignificant. (B) Correlation of ADCC induction in vitro and weight loss in vivo. The correlation between the in vitro ADCC activity and average (Avg.) maximal weight loss (x axis, in percentage with a 25% cutoff) is shown. There is a strong correlation between a MAb’s ADCC activity in vitro and its ability to protect against weight loss postchallenge. The isotype of the MAbs are indicated: IgG1, blue; IgG2a, red; IgG2b, green. (R² = 0.7567; pattern recognition receptor [PRR] = 0.8699; ****, P < 0.0001). (C and D) Antibody isotype plotted against average maximal weight loss or ADCC induction. On plotting the in vitro ADCC induction and in vivo weight loss for each MAb on the basis of isotype, we find that IgG2a MAbs induce the highest levels of ADCC in vitro, followed by IgG2b and IgG1 MAb(s), and correspondingly also lead to the lowest weight loss observed in our in vivo experiment. (C) **, P = 0.0016. (D) **, P = 0.0019. (E and F) Region of antibody binding compared in context of weight loss or ADCC reporter activity. Instead of antibody isotype, the data sets are represented on the basis of their binding to the head or the stalk of the HA. There is no significant trend in terms of whether the MAbs bind to the head or the stalk of the HA. (G and H) Nature of epitope and its relationship to weight loss or ADCC induction. There is a statistically significant difference between MAbs targeting conformation epitopes and linear/microconformational epitopes in context of inducing higher ADCC in vitro and leading to lower overall weight loss in vivo in the context of viral challenge. MAbs binding conformational epitopes show higher ADCC reporter activity and better in vivo efficacy. (G) *, P = 0.0457. (H) *, P = 0.0476.