A computationally optimized hemagglutinin virus-like particle vaccine elicits broadly reactive antibodies that protect nonhuman primates from H5N1 infection

Abstract

Background: Highly pathogenic H5N1 avian influenza viruses continue to spread via waterfowl, causing lethal infections in humans. Vaccines can prevent the morbidity and mortality associated with pandemic influenza isolates. Predicting the specific isolate that may emerge from the 10 different H5N1 clades is a tremendous challenge for vaccine design.

Methods: In this study, we generated a synthetic hemagglutinin (HA) on the basis of a new method, computationally optimized broadly reactive antigen (COBRA), which uses worldwide sequencing and surveillance efforts that are specifically focused on sequences from H5N1 clade 2 human isolates.

Results: Cynomolgus macaques vaccinated with COBRA clade 2 HA H5N1 virus-like particles (VLPs) had hemagglutination-inhibition antibody titers that recognized a broader number of representative isolates from divergent clades as compared to nonhuman primates vaccinated with clade 2.2 HA VLPs. Furthermore, all vaccinated animals were protected from A/Whooper Swan/Mongolia/244/2005 (WS/05) clade 2.2 challenge, with no virus detected in the nasal or tracheal washes. However, COBRA VLP-vaccinated nonhuman primates had reduced lung inflammation and pathologic effects as compared to those that received WS/05 VLP vaccines.

Conclusions: The COBRA clade 2 HA H5N1 VLP elicits broad humoral immunity against multiple H5N1 isolates from different clades. In addition, the COBRA VLP vaccine is more effective than a homologous vaccine against a highly pathogenic avian influenza virus challenge.

Figure 1.

Phylogenetic diversity of H5N1 influenza virus. The unrooted phylogenetic tree was inferred from hemagglutinin (HA) amino acid sequences derived from 8–10 representative isolates in all clades and subclades and also the computationally optimized broadly reactive antigen (COBRA) HA, using the maximum likelihood method. Clade/subclade clusters were identified and are indicated in the shaded ovals. The star identifies where the COBRA antigen is located relative to the various representative isolates. Sequences were aligned with MUSCLE 3.7 software, and the alignment was refined by Gblocks 0.91b software. Phylogeny was determined using the maximum likelihood method with PhyML software. Trees were rendered using TreeDyn 198.3 software [44]. The National Center for Biotechnology Information accession numbers for the HA sequences used in phylogeny inference were obtained through the Influenza Virus Resource [45].

Figure 2.

Serological findings. Total immunoglobulin G (IgG) levels at week 3 (A) and week 6 (B) were determined via enzyme-linked immunosorbant assay for each vaccine group. Each collected antiserum was assayed for antibody binding to representative hemagglutinin (HA) molecules from clade 2.1 (A/Indonesia/5/2005), clade 2.2 (A/Whooper Swan/Mongolia/244/2005), and clade 2.3 (A/Anhui/1/2005). Hemagglutination inhibition titers for each vaccine group were determined at week 3 (C) and week 6 (D), using representative H5N1 influenza viruses: clade 2.1 (A/Indonesia/5/2005), clade 2.2 (A/Whooper swan/Mongolia/244/2005), and clade 2.3 (A/Anhui/1/2005). The number of nonhuman primates that responded with a titer >1:40 is listed above each bar. None of the mock-vaccinated animals responded, and no value is listed. Neutralizing antibody at week 3 (E) and week 6 (F) was determined via mVN for each vaccine group. All values represent the geometric mean titer (95% confidence interval). Statistical significance of the antibody titer data was determined using 2-way analysis of variance followed by the Bonferroni posttest to analyze differences between each vaccine group for each of the different antigens that were tested (multiparametric). Significance was defined as P < .05. Statistical analyses were performed with GraphPad Prism software. Abbreviations: COBRA, computationally optimized broadly reactive antigen; VLP, virus-like particle.

Figure 3.

Hemagglutination inhibition (HAI) serum antibody titers from vaccinated nonhuman primates against a panel of clade 0, 1, 2, 4, and 7 isolates. HAI titer for each vaccine group was determined at week 6 by use of H5N1 influenza viruses. Values represent the geometric mean titer (95% confidence interval) of log₂-transformed titers Significant differences were determined by 2-way analysis of variance with the Bonferroni posttest to evaluate differences between the vaccine formulations for each test antigen. A P value of <.05 was considered significant as described in Figure 2. Abbreviations: COBRA, computationally optimized broadly reactive antigen; GMT, geometric mean titer; VLP, virus-like particle.

Figure 4.

Histopathologic analysis of H5N1-infected lung. Hematoxylin-eosin–stained sections of lungs from day 3 after infection: unvaccinated (A), WS/05 virus-like particle (VLP; B), computationally optimized broadly reactive antigen (COBRA) VLP (C), and unvaccinated, noninfected (D). Influenza in situ hybridization from day 1 after infection: unvaccinated (E), WS/05 VLP (F), and COBRA VLP (G).

Figure 5.

Footprint of hemagglutination inhibition (HAI) reactivity. The unrooted phylogenetic tree was inferred from hemagglutinin amino acid sequences derived from both vaccines and test antigens, using the maximum likelihood method. Phylogenetic analysis is the same as described in the legend for Figure 1. Numbers above each branch identify the clade/subclade of the respective cluster. The stars identify where the 2 vaccine antigens (computationally optimized broadly reactive antigen [COBRA] and WS/05) are located relative to the test antigens. The shading indicates the footprint of HAI reactivity for the COBRA vaccine (light gray) and WS/05 vaccine (dark gray).