H3N2 influenza virus infection induces broadly reactive hemagglutinin stalk antibodies in humans and mice

Abstract

Broadly neutralizing antibodies directed against the conserved stalk domain of the viral hemagglutinin have attracted increasing attention in recent years. However, only a limited number of stalk antibodies directed against group 2 influenza hemagglutinins have been isolated so far. Also, little is known about the general level of induction of these antibodies by influenza virus vaccination or infection. To characterize the anti-stalk humoral response in the mouse model as well as in humans, chimeric hemagglutinin constructs previously developed in our group were employed in serological assays. Whereas influenza virus infection induced high titers of stalk-reactive antibodies, immunization with inactivated influenza virus vaccines failed to do so in the mouse model. Analysis of serum samples collected from human individuals who were infected by influenza viruses also revealed the induction of stalk-reactive antibodies. Finally, we show that the hemagglutinin stalk-directed antibodies induced in mice and humans have broad reactivity and neutralizing activity in vitro and in vivo. The results of the study point toward the existence of highly conserved epitopes in the stalk domains of group 2 hemagglutinins, which can be targeted for the development of a universal influenza virus vaccine in humans.

Fig 1

Infection with influenza virus induces a stalk-specific humoral response in mice. (A) Mice (n = 5) were sublethally infected with 10⁴ PFU of Phil82 (H3 Phil82 10⁴), X-31 (H3 X31 10⁴), or RheaH7 (H7 Rhea 10⁴) virus. Additional groups of X-31-infected mice were infected 4 weeks later with 10⁶ or 10⁵ PFU of the Phil82 drifted strain (H3 X31 10⁴ + H3 Phil82 10⁶ or H3 X31 10⁴ + H3 Phil82 10⁵, respectively). Sera were collected from all animals 4 weeks after the last infection. Stalk-reactive antibodies were detected by measuring binding of sera to cH4/3 HA protein (Perth09 H3 stalk domain and H4 head domain) in an ELISA. Serum from naive mice was used as a negative control (naive sera). (B) ELISA stalk reactivity (cH4/3 substrate) of mice that were vaccinated intramuscularly (i.m.) with either an inactivated Phil82 virus preparation (inact. virus Phil82; 1 μg total protein) or seasonal TIV vaccine (TIV vaccine; 1 μg of the H3 component) or with DNA encoding Wisc05 HA (DNA Wisconsin/05 H3 HA; 80 μg, electroporation). Values from X-31-infected animals from panel A are shown as positive controls. (C) Reactivity of the elicited antibodies to H4 hemagglutinin (from A/duck/Czech/1956). (D) Reactivity of the elicited antibodies to H7 hemagglutinin (from A/chicken/Jalisco/12283/2012). (E) Plaque reduction neutralization assay with cH7/3N3 virus and purified IgG from sera shown in panel A. The virus (100 PFU) was preincubated with 5-fold dilutions of purified IgG preparations (starting concentration, 600 μg/ml) for 1 h at RT prior to infection of a monolayer of MDCK cells. Agar overlay contained the respective dilutions of IgG. Polyclonal sera raised against an H7N7 virus were used as positive controls. (F) Passive transfer experiment with sera from sequentially infected mice (H3 X31 10⁴ + H3 Phil82 10⁵) or naive mice (naive). Animals were challenged 2 h after transfer with a cH5/3N1 virus, and weight was monitored for 11 days. Numbers on the side of the legend indicate numbers of surviving mice out of the numbers of mice per group.

Fig 2

Reactivity of mouse sera to homologous full-length HAs. Sera from groups of mice shown in Fig. 1 were analyzed for reactivity to homologous HA protein. Mice (n = 5) were sublethally infected with 10⁴ PFU of Phil82 (H3 Phil82 10⁴) and X-31 (H3 X31 10⁴) viruses. Additional groups of X-31-infected mice were infected 4 weeks later with 10⁶ or 10⁵ PFU of the Phil82 drifted strain (H3 X31 10⁴ + H3 Phil82 10⁶ or H3 X31 10⁴ + H3 Phil82 10⁵, respectively). Groups of mice were also vaccinated with either an inactivated Phil82 virus preparation (IV H3 Phil82; 1 μg total protein) or seasonal TIV vaccine (TIV vaccine; 1 μg of the H3 component) or with DNA encoding Wisc05 HA (DNA Wisconsin/05 H3 HA; 80 μg, electroporation). Serum from naive mice was used as a negative control (naive sera). (A) ELISA reactivity of X-31-infected animals to homologous HA (from A/Hong Kong/1/1968). (B) ELISA reactivity of Phil82-infected or -vaccinated animals to homologous HA (from A/Philippines/2/1982). (C) ELISA reactivity of TIV-vaccinated animals to homologous HA (from A/Perth/16/2009). (D) ELISA reactivity of Wisc05 DNA-vaccinated animals to homologous HA (from A/Wisconsin/67/2005).

Fig 3

Infection with seasonal H3N2 virus induces stalk-directed antibodies in humans. (A and B) Sera from patients who presented with influenza-like illness in the 2010 to 2011 season were analyzed for seroconversion by ELISA with recombinant Perth09 HA as the substrate. For each patient, reactivity of samples collected before the influenza season, immediately after illness, or after the season was analyzed. The two negative-control data sets are grouped on the left side of the panel (influenza B infected). Data shown are OD values from samples diluted 1:200 (A) or endpoint titers from the same assay (B). (C and D) The same set of samples was analyzed for HA stalk reactivity by using a chimeric HA construct expressing a group 2 stalk (Perth09) and an H5 head (cH5/3). The two negative-control data sets are grouped on the left side of the panel (influenza B infected). Data shown are OD values from samples diluted 1:200 (C) or endpoint titers from the same assay (D). (E) The induction of stalk-directed antibodies correlates with the level of seroconversion as measured by reactivity to the full-length HA protein (R² = 0.6653, P < 0.0001).

Fig 4

Vaccination with seasonal inactivated trivalent vaccines does not induce an anti-HA stalk humoral response in humans. (A) Serum samples obtained from patients who received the inactivated trivalent vaccine in the 2009 to 2010 season were analyzed by ELISA for stalk reactivity using recombinant cH5/3 protein as the substrate. For each patient, matched samples obtained before vaccination and 4 weeks after were analyzed in the same assay. Data shown are from samples diluted 1:200. (B) Data in panel A are represented as fold inductions by plotting the ratio obtained by dividing the signal in the postvaccination sample by the signal obtained for the sample collected before vaccination.

Fig 5

The anti-stalk response boosted by seasonal H3N2 infection in humans has neutralizing activity. The five highest responders were chosen for this analysis (as determined by ELISA). For each individual, total IgG was purified from the sample collected before onset of the influenza season as well as from the postillness sample. To measure stalk-directed neutralizing activity, we used a virus that expresses an HA with an H3 stalk and an H7 head as well as an N3 NA (cH7/3N3). The virus (100 PFU) was incubated with IgG dilutions for 1 h at RT before infection of monolayers of MDCK cells. For all samples, the starting IgG concentration used in the assay was 600 μg/ml, and 5 5-fold dilutions were considered. The respective IgG concentrations were maintained in the agar overlay. Serum from an uninfected individual (Fam157) was used as an uninfected negative control. Polyclonal sera raised against an H7N7 virus were used as positive controls, and PBS was used as a negative control. (B) Correlation between ELISA reactivity and neutralizing activity. OD values from the 1:200 dilution in ELISA using cH5/3 protein were plotted against neutralizing activity at a 125-μg/ml IgG dilution, and a correlation analysis was performed (R² = 0.6764, P = 0.001).