Induction of broadly reactive anti-hemagglutinin stalk antibodies by an H5N1 vaccine in humans

Abstract

Influenza virus infections are a major public health concern and cause significant morbidity and mortality worldwide. Current vaccines are effective but strain specific due to their focus on the immunodominant globular head domain of the hemagglutinin (HA). It has been hypothesized that sequential exposure of humans to hemagglutinins with divergent globular head domains but conserved stalk domains could refocus the immune response to broadly neutralizing epitopes in the stalk. Humans have preexisting immunity against H1 (group 1 hemagglutinin), and vaccination with H5 HA (also group 1)--which has a divergent globular head domain but a similar stalk domain--represents one such sequential-exposure scenario. To test this hypothesis, we used novel reagents based on chimeric hemagglutinins to screen sera from an H5N1 clinical trial for induction of stalk-specific antibodies by quantitative enzyme-linked immunosorbent assay (ELISA) and neutralization assays. Importantly, we also investigated the biological activity of these antibodies in a passive transfer in a mouse challenge model. We found that the H5N1 vaccine induced high titers of stalk-reactive antibodies which were biologically active and protective in the passive-transfer experiment. The induced response showed exceptional breadth toward divergent group 1 hemagglutinins but did not extend to group 2 hemagglutinins. These data provide evidence for the hypothesis that sequential exposure to hemagglutinins with divergent globular head domains but conserved stalk domains can refocus the immune response toward the conserved stalk domain. Furthermore, the results support the concept of a chimeric hemagglutinin universal influenza virus vaccine strategy that is based on the same principle.

Importance: Influenza virus vaccines have to be reformulated and readministered on an annual basis. The development of a universal influenza virus vaccine could abolish the need for this cumbersome and costly process and would also enhance our pandemic preparedness. This study addressed the following questions, which are essential for the development of a hemagglutinin stalk-based universal influenza virus vaccine. (i) Can stalk-reactive antibodies be boosted by vaccination with divergent HAs that share conserved epitopes? (ii) How long-lived are these vaccine-induced stalk-reactive antibody responses? (iii) What is the breadth of this reactivity? (iv) Are these antibodies functional and protective? Our results further strengthen the concept of induction of stalk-reactive antibodies by sequential exposure to hemagglutinin immunogens with conserved stalk and divergent head domains. A universal influenza virus vaccine based on the same principles seems possible and might have a significant impact on global human health.

Copyright © 2014, American Society for Microbiology. All Rights Reserved.

Figures

FIG 1

H5N1 vaccination induces stalk-reactive antibodies in humans. (A) Study design. Participants received the H5N1 vaccine intramuscularly on day 0 and day 21. Blood was drawn on days 0, 21, and 42 postpriming, as well as after 6 and 12 months. (B) Stalk-reactive antibodies were measured in a cH6/1 ELISA. Reciprocal geometric mean titers on day 0 were detected at 1:2,722 and showed a high, significant increase for day 21 (1:1,1943; P < 0.0001) and day 42 (1:15,267; P < 0.0001). The levels of stalk-reactive antibodies declined after 6 months (1:4,143; P = 0.032) and remained at a similar level up to 12 months postvaccination (1:4,087; P = 0.0051), still significantly higher than baseline titers. (C) Induction of stalk-reactive antibodies over baseline. Stalk-reactive antibody titers showed a 4.4-fold induction over baseline after the first round of vaccination, with a further boost to 5.6-fold after the second round. After 6 months, the stalk-reactive antibodies declined to a level of 1.5-fold over baseline and were retained at that level for up to 12 months (1.6-fold). (D) Induction of stalk-reactive antibodies for individual vaccine groups. There was no statistical significant difference in induction of stalk-reactive antibodies for the various HA contents of the vaccines, as well as adjuvant, for any time point (21 days, P = 0.3391; 42 days, P = 0.6267; 6 months, P = 0.3068; 12 months, P = 0.9807). (E) Biological activity of stalk-reactive antibodies was measured in a microneutralization assay with a cH9/1 N3 virus. Using an HA head domain and neuraminidase to which humans are naive ensured measurement of stalk-reactive antibodies only. Baseline titers were relatively high, at 1:569, but showed a significant increase 42 days postpriming, to 1:1,092 (P < 0.0001). (F) Correlation of stalk-reactive antibodies measured by ELISA and microneutralization assay. The levels of stalk-reactive antibodies measured by ELISA significantly correlated with their measured neutralizing ability in vitro (Spearman r = 0.5225; P < 0.0001). n.s., not significant. In Fig. 1 to 4, * indicates a P value of ≤0.05, ** indicates a P value of ≤0.01, and *** indicates a P value of ≤0.001.

FIG 2

H5N1 vaccination induces cross-reactive group 1 HA antibodies in humans. (A) Titers of antibody against group 1 HAs H2 and H18 and group 2 HA H3 were measured on day 0 and day 42 postpriming. Titers of antibody against H2 increased from 1:1,118 to 1:4,095 (P < 0.0001) and titers of antibody against H18 from 1:791 to 1:2,024 (P < 0.0001). Titers of antibody against H3 remained at similar levels, 1:2,186 and 1:2,599 (P = 0.66). (B) Induction over baseline of antibodies against group 1 HAs H2 and H18 and group 2 HA H3 was calculated for day 42 postpriming. Antibodies against H2 show a 3.6-fold induction over baseline and induction of antibodies against H18 was 2.5-fold, confirming group cross-reactivity for group 1 HAs. Cross-group antibodies against H3 showed only a low induction, at 1.19-fold over baseline. (C) Correlation of stalk-reactive and H2 cross-reactive antibodies. Antibodies against the HA stalk domain and H2 showed a significant positive correlation (Spearman r = 0.5656; P < 0.0001). This indicates that the cross-reactivity is mostly caused by antibodies against shared epitopes in the stalk domain. All samples were analyzed, and overlapping results are presented as single points.

FIG 3

H5N1 vaccination elicits a weak antibody response against NA in humans. (A) Titers of antibody against homologous N1 (A/Vietnam/1203/2004 [VN]) and heterologous N1 (A/California/04/2009 [Cal09]) were measured on day 0 and day 42 postpriming. Titers of antibody against VN N1 increased slightly, from 1:755 to 1:1,131 (P = 0.0031), but the increase for Cal09 N1 from 1:1,817 to 1:2,072 was not significant (P = 0.7231). (B) Induction over baseline of antibodies against homologous N1 (VN) and heterologous N1 (Cal09) was calculated for day 42 postpriming. The induction for VN N1 was 1.5-fold over baseline, while induction for Cal09 N1 was only 1.1-fold.

FIG 4

Protective activity of stalk-reactive antibodies. (A) All participants with an 8-fold or higher induction (n = 27) of stalk-reactive antibody titers 42 days postpriming were selected. Their day 0 and day 42 sera were pooled and intraperitoneally injected into mice. After 2 h, mice were challenged intranasally with cH9/1 N3 virus. Lungs of 5 mice for each group were harvested 3 days and 6 days after challenge. (B) An ELISA against the H9 head domain confirmed that there was no significant increase in antibodies against the H9 head domain 42 days postpriming (1:1,131 on day 0 versus 1:1,275 on day 42; P = 0.4901). (C) Virus titers in the lung were assessed in a plaque assay 3 days and 6 days postchallenge. Virus titers in the lungs of mice that received postvaccination sera were significantly lower on day 3 postchallenge (9.6-fold difference; P = 0.0036). On day 6 after virus challenge, 3 of the mice that received postvaccination sera had lung virus titers below the detection limit of 10 PFU per ml; the other 2 showed titers of 0.5 × 102, which was significantly lower than the mean titer of 1.3 × 104 for the mice that received prevaccination sera (P = 0.0243). (D) Mice that received postvaccination sera had significantly lower weight loss 3 days postchallenge (1.6% versus 3.7% weight loss; P = 0.0144) and even regained weight 6 days postchallenge, while mice that received prevaccination sera continued to lose weight (4.6% weight loss). (E) The neutralizing activity of stalk-reactive antibodies for the sera used in the passive-transfer experiment was assessed in a cH9/1 N3 microneutralization assay. The titer 42 days postpriming was significantly higher, at 1:1,225, than that on day 0, 1:582 (P = 0.0022).

FIG 5

Model for induction of stalk-reactive antibodies by sequential exposure to HAs with conserved epitopes in the stalk but divergent globular head domains. Most humans had prior contact with H1 (A) and are therefore primed against the H1 stalk domain (dark green). The H5 (B) stalk domain (light green) shares conserved epitopes with the H1 stalk domain. This study shows that vaccination against H5N1 can boost antibodies against those shared epitopes. A vaccination strategy based on chimeric HAs (C) would combine an H1 stalk and an H5 head domain and might induce even higher titers of antibody against the HA stalk domain.