Applying valency-based immuno-selection to generate broadly cross-reactive antibodies against influenza hemagglutinins

Abstract

Conserved epitopes shared between virus subtypes are often subdominant, making it difficult to induce broadly reactive antibodies by immunization. Here, we generate a plasmid DNA mix vaccine that encodes protein heterodimers with sixteen different influenza A virus hemagglutinins (HA) representing all HA subtypes except H1 (group 1) and H7 (group 2). Each single heterodimer expresses two different HA subtypes and is targeted to MHC class II on antigen presenting cells (APC). Female mice immunized with the plasmid mix produce antibodies not only against the 16 HA subtypes, but also against non-included H1 and H7. We demonstrate that individual antibody molecules cross-react between different HAs. Furthermore, the mix vaccine induces T cell responses to conserved HA epitopes. Immunized mice are partially protected against H1 viruses. The results show that application of valency-based immuno-selection to diversified antigens can be used to direct antibody responses towards conserved (subdominant) epitopes on viral antigens.

Fig. 1. Expression of 18 different HA subtypes in heterodimeric vaccine proteins.

a Phylogenetic tree of amino acid sequences of selected HA subtypes of group 1 (black) and group 2 (gray). Scale bar represents 10% variation in amino acid sequences. b Schematic representation of heterodimeric vaccine proteins (middle) and the units each chain is composed of (left). The dimerization unit contains either a duplicated ACID (A) or BASE (B) sequence of a modified Fos-Jun zipper, attached to a shortened Ig hinge that allows formation of covalent bonds. The targeting unit is either scFv of anti-MHCII (I-E^d) mAb (scFv^αMHCII), or scFv of anti-NIP mAb (scFv^αNIP). The antigenic unit is any HA from subtypes H1-18 of IAV. Right: The A and the B cassette plasmids encoding the A and B chain of a heterodimer. c, d HEK293E cells were transiently co-transfected with indicated pairs of A and B plasmids that together encode heterodimers with two identical MHCII-specific targeting units and two identical HA antigenic units (HA_x/HA_x), for each of H1-18. Supernatants were analyzed by: (c) Western blot under non-reducing conditions, detected with anti-A/B mAb (n = 1 experiment); (d) sandwich ELISA employing anti-A/B and anti-HA stem mAb. e HEK293E cells were co-transfected with A and B plasmids that encode heterodimers with HA expression on the A and B arms as indicated (HA on A chain is indicated before the slash, followed by HA on the B chain). Left: pairing of plasmids with an HA corresponding to each of H1-18 (scFv^αMHCII-A-H1-18) and an invariant B plasmid (scFv^αNIP-B-H1) resulting in HA_x/H1 heterodimers. Right: vice versa, resulting in H1/HA_x heterodimers. Supernatants were analyzed by sandwich ELISA using mAbs specific for A/B and H1. Shown are individual OD_405nm values plus mean of technical triplicates ± SD. (n = 1 experiment). Statistics were calculated for vaccine protein expression compared to mock, using one-way Anova with Dunnetts multiple comparisons. All heterodimeric proteins in (d) and (e) were expressed in significantly higher amounts then the mock with p < 0.0001. Amino acid sequences for included HA proteins in (a) and uncropped images for Western blots in (c) are provided as a Source Data file.

Fig. 2. MHCII-targeted heterodimeric DNA vaccines that bivalently express two identical HA induce subtype-specific antibody responses.

a Female BALB/cAnNRj mice (n = 4/group) were vaccinated i.d. with a total of 50 µg DNA (A and B plasmids) immediately followed by electroporation (EP). Plasmids encoded bivalent heterodimers with two scFv^αMHCII and two identical HAs for H1-18. HA subtype-specific IgG responses elicited by a given HA_x/HA_x (indicated left) were analyzed six weeks after vaccination against the indicated recombinant HA (top) in ELISAs. Top matrix: group 1 HAs. Bottom matrix: group 2 HAs. Scale for heat map is indicated. b–d BALB/c mice vaccinated with indicated bivalent group 1 HAs (25 µg/plasmid, n = 4/group), or with scFV^αMHCII-targeted H1/H1 (n = 7) or NaCl (n = 7) as controls, were challenged with 5xLD₅₀ PR8 (H1N1) virus ten weeks after vaccination. e Mice vaccinated with indicated bivalent group 2 HAs (25 µg/plasmid, n = 4/group), or with scFV^αMHCII-targeted H7/H7 (n = 4) or NaCl (n = 4) as controls, were challenged with 5xLD₅₀ A/turkey/Italy/3889/1999 virus (H7N1). Shown are mean weight ± SEM). Mice were euthanized when they reached ≤ 80% of their original body weight and removed from the weight curves. % survival is shown in Supplementary Fig. 4. (n = 1 experiment). The same NaCl and H1/H1 vaccinated control mice were used in (b–d). For survival after PR8 challenge, significance was calculated on day 6 (b–d) or day 4 (e) for indicated groups compared to H1/H1 or H7/H7 control vaccines using one-way ANOVA with Dunnet’s multiple comparisons test. Source data are provided as a Source Data file.

Fig. 3. The mix plasmid vaccines are translated into vaccine proteins in vitro and induce HA subtype-specific antibody responses upon immunization of mice.

a Co-transfection of nine A plasmids and nine B plasmids expressing the indicated 18 HA subtypes should theoretically yield 81 heterodimeric proteins. Each heterodimer should be monovalent for any HA subtype. b The 18 HA subtypes of group 1 and group 2 were distributed on either A or B plasmids, as indicated. In 16 HA subtype mixes, H1 and H7 (colored lightly) were excluded. c HEK293E cells were transiently co-transfected with the 18 HA subtype plasmid mixes (group 1 + 2, either with scFv^αMHCII or scFv^αNIP) and supernatants were analyzed in sandwich ELISA with the indicated coat and detection mAbs. A/B plasmid pairs expressing bivalent HA subtypes (H1/H1, H7/H7, H3/H3) or bivalent scFv from M315 myeloma protein (scFv³¹⁵/scFv³¹⁵) were included as controls. Shown are mean ± SD of technical triplicates. d Female BALB/cAnNRj mice were immunized three times i.d./EP (week 0, 5, 10) with the 16 HA subtype mix (1 µg/plasmid), in either APC-targeted (scFv^αMHCII) or non-targeted (scFv^αNIP) format. Serum collected at weeks 5, 7, and 12 were analyzed by ELISA for IgG antibodies against the indicated HA subtypes encoded within the HA mix vaccine. Shown are mean IgG titers ± SEM, n = 6 mice/group. (n = 1 experiment). Statistics in (c) were calculated for vaccine protein expression compared to scFv³¹⁵/scFv³¹⁵, H3/H3, H1/H1 or H7/H7 negative controls on 1x dilution of supernatant, n.s. = not significant, one-way Anova with Dunnett’s multiple comparisons test. There were no statistical differences between targeted and non-targeted groups in (d) (two-tailed Mann–Whitney for each timepoint). Source data are provided as a Source Data file.

Fig. 4. Vaccination with an HA plasmid mix induces antibodies against HA subtypes not included in the mix.

a Female BALB/cAnNRj mice were immunized three times (arrows) intradermally (i.d.)/EP with indicated plasmid mixes (1 µg DNA/plasmid) that encoded either 16 (H1 and H7 excluded, n = 6) or 18 (n = 5) HA subtypes in either APC-targeted (scFv^αMHCII) or non-targeted (scFv^αNIP) formats. Positive controls were scFv^αMHCII-targeted H1/H1 and H7/H7 bivalent vaccines (5 µg/plasmid, n = 4/group). Shown are H1- and H7-reactive serum IgG titers at indicated timepoints. b Mice were immunized three times (arrows) intramuscularly (i.m.)/EP with indicated mixes (1 µg DNA/plasmid, n = 7 [scFv^αNIP-group1 + 2(ΔH1ΔH7)] or n = 9 [scFv^αMHCII-group1 + 2(ΔH1ΔH7)]) or with scFv^αMHCII-H1/H1 or scFv^αMHCII-H7/H7 (5 µg/plasmid, n = 3/group) or NaCl (n = 9) controls. Shown are H1- or H7-reactive serum IgG titers at indicated timepoints. c, d Mice were vaccinated as in (b) with the indicated mix vaccines (n = 6 [scFv^αNIP-group1 + 2(ΔH1ΔH7)] or n = 9 [scFv^αMHCII-group1 + 2(ΔH1ΔH7)]), or with scFv^αMHCII-H1/H1 (n = 3), scFv^αMHCII-H7/H7 (n = 4) or NaCl (n = 10) controls. Shown are H1- and H7-reactive serum IgG1 and IgG2a (c) and IgG binding to inactivated H1N1 virus as ELSIA coat antigen (d) at week 11. e–f Mice were vaccinated as in (b) (n = 8 [mix vaccines and NaCl] or n = 4 [bivalent H1/H1 or H7/H7 control]). In vitro neutralization of PR8 virus infection of MDCK cells by week 11 sera was measured. Shown are 50% neutralizing serum titers (e). H1^PR8-specific ADCC by week 11 sera was determined by measuring in vitro activation of mouse NK cells (f). g Mice were immunized twice (arrows) with purified scFv^αMHCII-targeted ΔH1ΔH7 mix protein vaccine or scFv^αMHCII-H1/H1 protein vaccine as control (2.5 µg protein with/without AddaVax adjuvant, n = 4/group). Shown is H1-reactive total serum IgG at indicated timepoints. Mean ± SEM (a, b, and g) or titers/percentages for individual mice plus mean ± SEM (c–f) are indicated. Responses of mix vaccines (calculated for final timepoints in (a), (b), and (g)) were statistically compared to NaCl or PBS with adjuvant. In c and d, responses elicited by targeted and non-targeted vaccines were also compared. Kruskal-Wallis with Dunn’s multiple comparisons test. N.D. = non-detectable. n = 1 independent experiment, except one of three for (b). Source Data file is provided.

Fig. 5. Vaccination with HA plasmid mixes induces serum antibodies that cross-react between HA subtypes.

a, f Female BALB/cAnNRj mice were vaccinated thrice (week 0, 5 and 9) i.m./EP with plasmid mixes encoding the indicated mix vaccines (1 µg DNA/plasmid, n = 8/group). Positive controls were vaccinated with either bivalent H1/H1, H7/H7, H2/H2, H3/H3, or H5/H5, all scFv^αMHCII-targeted (5 µg DNA/plasmid, n = 4/group). Sera were harvested two weeks after the final vaccination and pooled for each experimental group. a–e Pooled sera were pre-incubated with the indicated soluble competing HA proteins (5 µg/ml, x-axis) prior to testing for binding to coat HAs in ELISA. Heatmaps show mean % inhibition of technical triplicates. f H1-reactive antibodies were affinity-purified on H1PR8-conjugated Sepharose. Purified antibodies were analyzed for binding to the indicated recombinant HA proteins in ELISA. Heatmap shows mean IgG titer of technical triplicates. Each panel presents data derived from n = 1 independent experiment. Source data are provided as a Source Data file.

Fig. 6. Vaccination with HA plasmid mixes induces antibodies against both the stem and head domains of HAs not included in the mix.

a, b Female BALB/cAnNRj mice were immunized three times (week 0, 5 and 9) i.m./EP with indicated plasmid mixtures (1 µg DNA/plasmid, n = 8/group). Controls were scFv^αMHCII-H1/H1 and scFv^αMHCII-H7/H7 (5 µg DNA/plasmid, n = 4/group) or NaCl (n = 8). We analyzed IgG titers against the H1 (a) and H7 (b) head (left panel) and stem (right panel) domains for week 11 sera. c–e Female BALB/cAnNRj mice were vaccinated twice (week 0 and 5) with the scFv^αMHCII-group1 + 2(ΔH1ΔH7) mix (1 µg DNA/plasmid). Serum IgG titers against the stem and head domain of H1 and H7 were analyzed four weeks after the second vaccination. c Ratio of IgG titers against the HA head and HA stem domain of H1 (left, n = 39) or H7 (right, n = 32) for individual mice. d IgG titers against stem (left) or head (right) domains of H1 (n = 32) and H7 (n = 32). Sera from individual mice, interconnected by lines. e IgG titers of individual mice (n = 32) against the H1 head (black bars), H1 stem (green), H7 head (gray) and H7 stem (orange). Mice were grouped (vertical dashed lines) according to their serum specificity. f Mice were vaccinated thrice as in (a, b) with the vaccines indicated on the x-axis (n = 7 [scFv^αMHCII-group1 + 2(ΔH1ΔH7)], n = 8 [scFv^αNIP-group1 + 2(ΔH1ΔH7) and scFv^αMHCII-group1 + 2], or n = 4 [scFv^αMHCII-H1/H1]). Two weeks after the third vaccination, individual sera were analyzed for IgG binding to recombinant H1^PR8 and various mutant H1^PR8 lacking defined head determinants (box) (Supplementary Fig. 14). The ratio of binding (AUC) to each mutant H1/H1^PR8 WT was calculated. Shown are titers or ratios for individual mice and mean ± SEM (a–c, f) or titers for individual mice (d, e). Statistics were calculated using Kruskal–Wallis with Dunn’s multiple comparisons test (a, b), two-tailed Mann–Whitney (c) or two-tailed Wilcoxon matched-pairs signed rank test (d). Significant differences for ratios of serum from mix vaccinated mice compared to ratios for serum from H1/H1 vaccinated mice are indicated in (f). N.D. = non-detectable. n = 1 independent experiment. Source data are provided as a Source Data file.

Fig. 7. The HA mix vaccine induces IFN-γ secreting T cells.

a Female BALB/cAnNRj mice were immunized once i.m./EP with DNA HA mix vaccines (1 µg DNA/plasmid, n = 8/group) as indicated. As controls, mice were vaccinated with scFv^αMHCII-H1/H1 (H1 from PR8, 5 µg/plasmid, n = 3) or NaCl (n = 5). Draining lymph nodes (dLN) were collected 5 days after immunization, and IFN-γ secretion was measured by ELISPOT after stimulation with either IYSTVASSL (H1) peptide, or PR8- or Cal07-overlapping peptides (both H1N1). b Female BALB/cAnNRj mice were vaccinated three times (week 0, 5 and 9) i.m./EP with DNA HA mix vaccines (1 µg DNA/plasmid, n = 8/group) as indicated. Control mice were vaccinated with scFv^αMHCII-H1/H1 (H1 from PR8, 5 µg/plasmid, n = 4) or NaCl (n = 8). Spleens were harvested two weeks after the third immunization and stimulated with HA peptides as in (a). IFN-γ secreting cells were analyzed by ELISPOT. Shown are the number of spot-forming cells (SFC) per 10⁶ cells for individual mice and mean ± SEM. Significance was calculated using Kruskal–Wallis with Dunn’s multiple comparisons test, comparing the mix-immunized groups with the NaCl control and scFv^αMHCII targeted with non-targeted control. Only differences scored as significant are indicated. n = 1 independent experiment. Source data are provided as a Source Data file.

Fig. 8. The HA mix vaccine induces partial protection against heterologous influenza virus.

a Female BALB/cAnNRj mice were immunized three times (week 0, 5 and 9) i.m./EP with the indicated DNA HA mix vaccines (1 µg DNA/plasmid, n = 7 [scFv^αNIP-group1 + 2(ΔH1ΔH7)] or n = 9 [scFv^αMHCII-group 1 + 2(ΔH1ΔH7)]). As controls, mice were immunized with scFv^αMHCII-H1/H1 (H1 from either PR8, 5 µg/plasmid, n = 3) or NaCl (n = 10). Two weeks after the third immunization, mice were infected with 2.5xLD₅₀ of H1N1 PR8 virus. b Mice were vaccinated as in (a) (n = 6 [scFv^αMHCII-H1Cal07/H1Cal07], n = 9 [NaCl] or n = 10 [scFv^αNIP- and scFv^αMHCII-targeted mix vaccines]). Two weeks after the third immunization, mice were infected with 5xLD₅₀ of H1N1 Cal07 virus. c Female BALB/cAnNRj mice were vaccinated twice (week 0 and 6) with scFv^αMHCII-targeted HA mix protein vaccine or bivalent H1/H1 (PR8) as control (2.5 µg protein with or without AddaVax as adjuvant, n = 4/group). Seven weeks after the second immunization, mice were infected with 5xLD₅₀ of H1N1 PR8 virus. d Female BALB/cAnNRj mice were immunized twice (week 0 and 5) i.m./EP with DNA HA mix vaccines (1 µg DNA/plasmid, n = 29). Controls were vaccinated with scFv^αMHCII-targeted H1/H1 (PR8) or H7/H7 (5 µg/plasmid, n = 3/group) or NaCl (n = 6). Four weeks after the second immunization, mice were challenged with 5xLD₅₀ of H1N1 PR8. Starting two days before challenge, mice vaccinated with the scFv^αMHCII mix received either anti-CD4 plus anti-CD8 mAbs (n = 16) or isotype matched control mAbs (n = 13), every other day. Weight loss (mean ± SEM, left) and survival (right) are indicated in the panels. Mice were euthanized when they reached 80% of their initial weight and removed from the weight curves. Mean ± SEM is indicated with thick colored lines and symbols. Weight of individual mice indicated in stippled and faded lines. Significance was calculated using two-tailed Mantel–Cox (survival comparing mix vaccines to NaCl) or two-way ANOVA (weight curves). For T cell-depleted mice and isotype-matched controls compared in (d), n.s. = not significant. n = 1 independent experiment. Source data are provided as a Source Data file.