A modified vaccinia Ankara vaccine vector expressing a mosaic H5 hemagglutinin reduces viral shedding in rhesus macaques

Abstract

The rapid antigenic evolution of influenza viruses requires frequent vaccine reformulations. Due to the economic burden of continuous vaccine reformulation and the threat of new pandemics, there is intense interest in developing vaccines capable of eliciting broadly cross-reactive immunity to influenza viruses. We recently constructed a "mosaic" hemagglutinin (HA) based on subtype 5 HA (H5) and designed to stimulate cellular and humoral immunity to multiple influenza virus subtypes. Modified vaccinia Ankara (MVA) expressing this H5 mosaic (MVA-H5M) protected mice against multiple homosubtypic H5N1 strains and a heterosubtypic H1N1 virus. To assess its potential as a human vaccine we evaluated the ability of MVA-H5M to provide heterosubtypic immunity to influenza viruses in a non-human primate model. Rhesus macaques received an initial dose of either MVA-H5M or plasmid DNA encoding H5M, followed by a boost of MVA-H5M, and then were challenged, together with naïve controls, with the heterosubtypic virus A/California/04/2009 (H1N1pdm). Macaques receiving either vaccine regimen cleared H1N1pdm challenge faster than naïve controls. Vaccination with H5M elicited antibodies that bound H1N1pdm HA, but did not neutralize the H1N1pdm challenge virus. Plasma from vaccinated macaques activated NK cells in the presence of H1N1pdm HA, suggesting that vaccination elicited cross-reactive antibodies capable of mediating antibody-dependent cell-mediated cytotoxicity (ADCC). Although HA-specific T cell responses to the MVA-H5M vaccine were weak, responses after challenge were stronger in vaccinated macaques than in control animals. Together these data suggest that mosaic HA antigens may provide a means for inducing broadly cross-reactive immunity to influenza viruses.

Fig 1. Early viral clearance from lower respiratory tract in monkeys vaccinated with an H5 mosaic MVA vaccine.

Macaques were given a mosaic H5 HA antigen expressed on either plasmid DNA or in MVA and then boosted with MVA-H5M. All animals were challenged with the heterosubtypic virus A/California/04/2009 (H1N1pdm). Viral titers in (A) bronchoalveolar lavage (BAL) and nasal wash (B) were determined by standard plaque assays on MDCK cells. Control group includes 6 additional historical controls we previously published that were challenged with the same dose of the same viral stock by the same route [29]. The 2 control animals inoculated during this study are indicated by a star symbol. At day 4 post infection there was a statistically significant difference between DNA prime vaccinated and control animals as determined by a Kruskal-Wallis test combined with permutation. Bars indicate geometric mean and error bars indicate the 95% confidence interval of the geometric mean.

Fig 2. The H5 mosaic MVA vaccine stimulates antibodies against both H1N1 and H5N1 HA subtypes after prime and boost.

Antibody binding to HA was measured using enzyme-linked immunosorbent assay (ELISA). Purified HA protein from A/California/04/2009 (H1N1pdm) (A) or A/Vietnam/1203/2004 (H5N1) (B) was used to capture HA antibodies in an ELISA assay. Symbols indicate mean OD; error bars indicate standard deviation.

Fig 3. Vaccination with mosaic H5 MVA stimulates neutralizing and HI antibodies against H5N1 only.

HI antibody titer was determined using a standard HI assay against both H1N1pdm A/California/04/2009 and H5N1 A/Vietnam/1203/2004. HI antibodies against H1N1pdm were undetectable until 30 days after challenge (A). HI antibodies against H5N1 were detected as early as 7 days after prime (B). Neutralization against the challenge H1N1pdm A/California/04/2009 virus was measured using plaque reduction neutralization test, where neutralization was only detected 14 days post challenge (C). Data points represent individual monkeys and report the IgG concentration where a 50% reduction in plaque formation is observed, bars indicate geometric mean and error bars indicated the 95% confidence interval of the geometric mean.

Fig 4. Mosaic H5 MVA vaccination stimulates antibodies capable of antibody dependent cell mediated cytotoxicity (ADCC), which are maintained throughout infection.

We determined ADCC antibody titer against the H1N1pdm A/California/04/2009 using CD107a to detect degranulation. ADCC antibodies were detected at high magnitude after the MVA boost and were then maintained at a lower level throughout challenge (A). Data points indicate individual monkeys with the bar and error bar’s indicating median and standard deviation. Regressing viral titers on ADCC antibodies at 2 days post challenge shows only 5% of the variability in viral titer can be explained by ADCC antibodies and the association is not significant, R² = 0.0452, p = 0.4467 (B). At 4 days post challenge 20% of the variability in viral titer can be explained by ADCC antibodies and the association is not significant, R² = 0.1966, p = 0.0979 (C). The dashed line represents the 95% confidence interval of the linear regression model.

Fig 5. Animals mount T cell responses against the H1N1pdm challenge after vaccination with H5 mosaic MVA vaccine.

We detected T cell responses in peripheral blood mononuclear cells (PBMC) using an IFN-γ elispot assay. PBMCs were stimulated with 7 pools of overlapping synthetic peptides from HA. We detected no responses 7 days after prime (A) or 14 days after boost (B). T cell responses were detected 7 days after challenge, with most responses detected in PBMCs stimulated with peptide pool 6 (C). The H5 mosaic HA used in vaccination had the most amino acid sequence identity shared with the challenge strain A/California/04/2009 (H1N1pdm) in the HA 2 domain, with the highest sequence identity of 82.4% occurring within peptide pool 6 (identical amino acid residues indicated by a black stripe; D). Error bars in A, B, and C indicate standard deviation.