Simultaneous Targeting of Multiple Hemagglutinins to APCs for Induction of Broad Immunity against Influenza

This article is featured in In This Issue, p.2057

Influenza viruses are constantly evolving, thus evading the potential neutralization by Abs formed during previous virus exposures. The evolution is mostly driven by antigenic drift: nonsynonymous point mutations induced during virus replication cause amino acid changes that abrogate recognition by neutralizing Abs. On a more sporadic basis, more drastic antigenic shifts may occur as a result of genetic reassortments, potentially causing outbreaks of pandemic influenza. The major causes of influenza-related morbidity and mortality in humans have been the H1, H2, and H3 subtypes of influenza A, as illustrated by the three pandemics of the twentieth century: the 1918 Spanish flu (H1), the 1957 Asian flu (H2), and the 1968 Hong Kong flu (H3). However, recently emerging subtypes of traditionally zoonotic (primarily avian) strains, such as H5, H6, H7, H9, and H10 influenza viruses, have caused severe, but isolated, disease outbreaks in humans (1–5) and may cause future influenza pandemics. Because current vaccines predominantly induce strain-specific Abs (6, 7) and require a substantial amount of time for production (8–10), novel vaccine strategies conferring broader protection are needed.

DNA vaccines are in vogue because of their potential for rapid insertion of novel Ags and a speedy vaccine production. However, DNA vaccines are typically hampered by low immunogenicity, particularly in larger animals and humans. To remedy this shortcoming, better methods of DNA delivery have been developed (11–13). However, the improvements have typically been restricted to more efficient DNA uptake by cells at the site of delivery. An alternative method for increasing immunogenicity is to administer genes encoding a natural adjuvant together with the vaccine Ag. This could be chemokines that attract and activate leukocytes at the site of vaccine delivery (14) or perforin, which induces necrotic death (15). In another strategy, DNA has been constructed so that it encodes secreted fusion proteins that target Ags to APCs for enhanced immune responses (16–21). This approach combines the attractiveness of DNA immunization with the well-known principle of APC targeting of Ag to increase Ag immunogenicity (22–24). Interestingly, targeting of Ag to different APC surface molecules may skew immune responses toward different arms of immunity (19, 25, 26). Thus, the technology may allow matching of vaccine-induced immune responses to the type of immunity needed for protection against a particular pathogen.

Previously, we have demonstrated that a single DNA immunization that causes targeting of influenza hemagglutinin (HA) to MHC class II (MHCII) molecules can confer long-lasting Ab-mediated protection against homotypic influenza challenges in mice (18). The increased efficacy conferred by MHCII targeting of Ag has recently been translated to influenza vaccination of ferrets and pigs (21). As a mechanism for enhanced Ab induction, it has been suggested that the DNA-encoded anti–MHCII-HA vaccine proteins bridge B cells and APCs in an APC–B cell synapse (16, 27, 28). In this article, we show that the same MHCII-targeted DNA vaccine format can induce strong Ab responses against H5, H6, H8, H9, H11, and H13 subtypes of group 1 influenza viruses. We further demonstrate that the strong Ab responses are maintained after delivery of a mixture containing the six DNA constructs within a single bolus. Importantly, the mixture containing the six HAs could also induce cross-protective Abs against group 1 strains of influenza that had not been included in the vaccination mix [A/Puerto Rico/8/1934 (H1N1) (PR8) and A/California/07/2009 (H1N1) (Cal07)].

Vaccine plasmids were constructed as previously described, with Ag linked to an MHCII-specific scFv via a dimerization unit containing the C_H3 domain of human IgG3 (16, 18). Briefly, HAs from A/Hong Kong/483/97 (H5N1), A/northern shoveler/California/HKWF115/07 (H6N1), A/pintail duck/Alberta/114/1979 (H8N4), A/Hong Kong/1073/99 (H9N2), A/duck/Yangzhou/906/2002 (H11N2), and A/black-headed gull/Netherlands/1/00 (H13N8) were picked up by PCR from cDNA (VG11689-C, VG11723-C, VG11722-C, VG11229-C, VG11705-C, and VG11721-C, all from Sino Biological, Beijing, China) and cloned into SfiI sites in the CMV-based pLNOH2 vector (29). Primers used were: H5₁₇5′: GGC CTC GGT GGC CTG GAC CAG ATT TG, H5₅₃₂3′: ACC GGC CCT GCA GGC CTC ACT GGT AGG TGC CCA TAC TCT C, H6₂₃5′: GGC CTC GGT GGC CTG GAC AAG ATT TG, H6₅₄₇3′: ACC GGC CCT GCA GGC CTC ACA GGG AGG AG, H8₂₄5′: GGC CTC GGT GGC CTG GAC AGG ATT TG, H8₅₄₇3′: ACC GGC CCT GCA GGC CTC ACA GGG AGG C, H11₂₃5′: GGC CTC GGT GGC CTG GAT GAG ATT TG, H11₅₄₆3′: ACC GGC CCT GCA GGC CTC ACA GGG AGG AG, H13₂₅5′: GC CTC GGT GGC CTG GAC AGG ATT TG, and H13₅₄₆-3′: ACC GGC CCT GCA GGC CTC AAA TGC TGG AG. The cDNA of HA from H9 (VG11229-C) was delivered without aa 530–541. To insert the remaining C-terminal amino acids, and, as such, generate a gene that would resemble the other HAs with respect to inclusion of the transmembrane region, we used the following primers to insert an extension to the 3′ end: H9₁₉5′: GGC CTC GGT GGC CTG GAC AAG ATT TG and H9₅₂₉-3′: CAG GGA GGA CAC TGT GCT GTA GAT GGT CAG AAT CTT GTA GG TG, followed by H9₁₉5′ and H9₅₄₁3′: ACC GGC CCT GCA GGC CTC ACA GGG AGG AC.

Vaccine plasmids were transiently transfected into HEK 293E cells by the addition of 1 μg of DNA and Lipofectamine (11668-019; Invitrogen, Life Technologies, CA), and supernatants were assayed by ELISA, as previously described (18). Briefly, 96-well plates (Costar 3590; Sigma-Aldrich, St. Louis, MO) were coated with anti-human IgG (C_H3 domain; 1 μg/ml; MCA878G; AbD Serotec, Hercules, CA) or 4-hydroxy-3-iodo-5-nitrophenylacetic acid (NIP) conjugated to BSA (1:1000), and vaccine proteins in transfected supernatants were detected with biotinylated anti-human IgG (C_H3 domain; HP6017, B3773-0.2ML; 1:3000; Sigma-Aldrich) and alkaline phosphatase–conjugated Streptavidin (RPN1234V; 1:3000; GE Healthcare, Buckinghamshire, U.K.). The plates were developed with Phosphatase Substrate (P4744-10G; Sigma-Aldrich) and read with a Tecan reader (Tecan, Mannedorf, Switzerland) using Magellan version 5.03.

Six- to twelve-week-old BALB/cAnN mice were used (Taconic, Ejby, Denmark). The animals were housed under minimal disease conditions at Oslo University Hospital. Experiments were approved by the Norwegian Animal Research Authority.

PR8 and Cal07 were kindly provided by A.G. Hauge (National Veterinary Institute, Oslo, Norway). The viruses were propagated by inoculating virus into the allantoic cavity of 10-d-old embryonated chicken eggs. Allantoic fluid was harvested and confirmed negative for bacterial contamination. TCID₅₀ was determined.

Mice were anesthetized by s.c. injection of Hypnorm/Dormicum (0.05 ml working solution/10 g), and the lower back region was shaved. Twenty-five microliters of plasmids (purified from an EndoFree kit; QIAGEN), dissolved in NaCl, were injected intradermally on each flank of the mouse, followed immediately by skin electroporation (EP) with an AgilePulse (Harvard Apparatus BTX, Holliston, MA). For viral challenge, anesthetized mice were infected intranasally (i.n.) with 5 × LD₅₀ of PR8 (2.0 × 10⁴ TCID) in 20 μl (10 μl per nostril). Mice were monitored for weight loss, with an end point of 20% weight reduction, as required by the National Committee for Animal Experiments.

ELISA 96-well plates (Costar 3590) were coated with one of the following recombinant influenza HA proteins: PR8 (11684-V08H; Sino Biological), A/Hong Kong/483/97 (H5N1) (11689-V08H; Sino Biological), A/northern shoveler/California/HKWF115/07 (H6N1) (MBS434125; MyBioSource, San Diego, CA, or 11723-V08H; Sino Biological), A/pintail duck/Alberta/114/1979 (H8N4) (11722-V08H; Sino Biological), A/Hong Kong/1073/99 (H9N2) (11229-V08H; Sino Biological), A/duck/Yangzhou/906/2002 (H11N2) (11705-V08H; Sino Biological), A/black-headed gull/Netherlands/1/00 (H13N8) (11721-V08H; Sino Biological), A/Shanghai/1/2013 (H7N9) (40104-V08B; Sino Biological), or A/Hong Kong/1/1968 (H3N2) (40116-V08B; Sino Biological). Alternatively, plates were coated with inactivated virus: PR8 (1:1600; Charles River, Wilmington, MA) or Cal07 (Pandemrix; 1:50; GlaxoSmithKline, Brentford, U.K.). The coated plates were blocked, and serial dilutions of mouse sera were added to the ELISA plates. Next, alkaline phosphatase–conjugated anti-mouse IgG (100M4816; 1:5000, Sigma-Aldrich) was added, and plates were developed as described above.

BALB/c mice (n = 6) were vaccinated three times with DNA, followed immediately by skin EP (weeks 0, 4, and 12). Blood was harvested by saphenous vein puncture, and serum was separated by centrifugation. The microneutralization assay was set up as previously described (18). Briefly, equal amounts of serum from individual mice were pooled by group and incubated with receptor-destroying enzyme (Denka Seiken, Tokyo, Japan). The enzyme was then deactivated at 56°C for 1.5 h, and serial serum dilutions were added in duplicates to 96-well plates (Costar 3590). Diluted virus and controls were added, and plates were incubated at 37°C for 1 h prior to the addition of 20,000 MDCK cells per well. Plates were incubated at 37°C overnight. Next, cells were fixed with 80% acetone solution and left to air dry before washing with 0. 3% Tween washing buffer. Biotinylated H16-L10-4R5 (American Type Culture Collection [ATCC], Manassas, VA) was added (1:1000), and plates were developed with alkaline phosphatase–conjugated Streptavidin (RPN1234V; GE Healthcare) and phosphatase substrate (P4744-10G; Sigma-Aldrich). Plates were read with a Tecan reader using Magellan version 5.03.

BALB/c mice (n = 7 per group for the vaccine containing a mixture of six MHCII-targeted HAs [anti–MHCII-MIX] and for the vaccine containing a mixture of six NIP-targeted HAs [anti–NIP-MIX]; n = 6 per group for PR8 and NaCl controls) were vaccinated three times with 25 μg of DNA/EP (weeks 0, 4, and 8). Two weeks after the last vaccination, blood was harvested by cardiac puncture under full sedation, and serum was collected by centrifugation. A new cohort of naive BALB/c mice (n = 10 per group) was inoculated i.v. with 200 μl of pooled serum. A lethal PR8 challenge (5 × LD₅₀) was administered i.n. 24 h later, and weight loss was monitored.

BALB/c mice (n = 10 per group) were vaccinated three times with DNA/EP (weeks 0, 4, and 8). Starting on day 12 after the last vaccination, and then every other day until completion of the experiment, mice received i.p. injections of 200 μg of anti-CD4 mAb (GK1.5) and anti-CD8 mAb (TIB105; both from ATCC) or isotype controls (SRF8-B6 and Y13-238). At day 14 after the last vaccination, mice were challenged i.n. with PR8 influenza (5 × LD₅₀) and monitored for weight loss.

After termination, spleens were harvested, stained, and analyzed by FACS for the extent of T cell depletion. Briefly, single-cell suspensions of splenocytes were blocked (30% inactivated rat serum and 0.1 μg/ml HB197; ATCC) and stained with FITC-conjugated mAb against CD19 (35-0193-9500; Tonbo Biosciences, San Diego, CA) (dump gate), Pacific Blue–conjugated mAb against CD8a (558106; BD Pharmingen, San Jose, CA), and PerCP/Cy5.5-conjugated mAb against CD4 (100434; BioLegend, San Diego, CA). Samples were run on an LSR II flow cytometer (Becton Dickinson, Franklin Lakes, NJ), and data were analyzed with FlowJo software (v7.6) (TreeStar, Ashland, OR).

Statistical calculations were performed using two-way ANOVA, with the Bonferroni multiple-comparison test (GraphPad Prism 7).

HA from six subtypes of group 1 influenza viruses [A/Hong Kong/483/97 (H5N1), A/Northern shoveler/California/HKWF115/07 (H6N1), A/pintail duck/Alberta/114/1979 (H8N4), A/Hong Kong/1073/99 (H9N2), A/duck/Yangzhou/906/2002 (H11N2), and A/black-headed gull/Netherlands/1/00 (H13N8)] were inserted separately as antigenic units into plasmids encoding an MHCII-targeted DNA vaccine format (16, 18). Each plasmid encodes, under the same promoter, a targeting unit (scFv) that directs the vaccine toward MHCII molecules, a dimerization unit derived from the hinge and C_H3 exons of human IgG3, and one of the six HAs as an antigenic unit. The dimerization unit induces formation of homodimeric vaccine proteins following transcription and translation (Fig. 1A). The new vaccines are denoted anti–MHCII-H5, anti–MHCII-H6, anti–MHCII-H8, anti–MHCII-H9, anti–MHCII-H11, and anti–MHCII-H13. To assess the potential effect of MHCII targeting, we also constructed equivalent nontargeted control vaccines in which the targeting unit was replaced with a scFv specific for the synthetic hapten NIP (16, 18), presumably not expressed in mice. The nontargeted vaccines are denoted anti–NIP-H5, anti–NIP-H6, anti–NIP-H8, anti–NIP-H9, anti–NIP-H11, and anti–NIP-H13.

To generate comparable results with the different vaccine constructs, the inserted HA genes were generally selected in the same way with respect to the included sequences (Supplemental Fig. 1). The intrinsic HA signaling peptides were removed, because the targeting unit, dimerization unit, and HA were transcribed as one chain with an Ig-derived signaling peptide located upstream of the targeting unit (29), and intracellular domains were removed to secure efficient and similar secretion of the different vaccine proteins. The transmembrane region was only partially truncated to maintain a well-conserved MHC class I T cell epitope in BALB/c mice (IYSTVASSL); such partial inclusion of the transmembrane region did not impede secretion of the vaccine fusion protein containing H1 in previous experiments (18). Using this strategy, H6, H8, H9, H11, and H13 were secreted. The exception was HA from H5 influenza; the inclusion of the IYSTVASSL sequence severely reduced secretion efficacy after in vitro transfection. Upon removal of the complete transmembrane region of H5 HA, the construct was secreted to similar levels as the other HA vaccines into the supernatants of transiently transfected HEK 293E cells (Fig. 1B). Hence, for the H5 vaccines, we used the shortened sequence devoid of the transmembrane region. As positive controls in some of the experiments, we included the previously published MHCII-targeted vaccine encoding HA from PR8 [anti–MHCII-H1(PR8)] or the MHCII-targeted vaccine encoding HA from Cal07 [anti–MHCII-H1(Cal07)] (18).

BALB/c mice were immunized once with 25 μg of plasmid encoding anti–MHCII-H5, anti–MHCII-H6, anti–MHCII-H8, anti–MHCII-H9, anti–MHCII-H11, or anti–MHCII-H13. In addition, a positive control group was vaccinated with anti–MHCII-H1(PR8) (18). The vaccine plasmids were delivered intradermally in combination with EP to secure efficient cellular uptake of DNA (11, 30). Following vaccination, sera were monitored for the development of Abs against recombinant HA from H5, H6, H8, H9, H11, H13, and H1(PR8) influenza viruses. Importantly, vaccination with each of the MHCII-targeted vaccines encoding a particular HA induced strong strain-specific Abs (Fig. 2A). Thus, as an example, vaccination with anti–MHCII-H6 induced high IgG titers against recombinant HA from influenza H6 but failed to induce Abs against HA from influenza H5, H8, H9, H11, H13, or H1 (Fig. 2A, top middle panel). Similarly, vaccination with each of the MHCII-targeted vaccines induced comparable high levels of strain-specific Abs but failed to induce Abs against other types of influenza. There is a slight deviation for anti–MHCII-H5, which induced somewhat lower titers against the homologous HA (top left panel). This is consistent with the observation by other investigators that H5 has a low immunogenicity compared with other HAs (31).

Four weeks after a single DNA immunization, mice were infected with a lethal dose of influenza PR8 and monitored for weight loss. Corresponding with the lack of cross-reactive Abs (Fig. 2A), vaccination with anti–MHCII-H5, anti–MHCII-H6, anti–MHCII-H8, anti–MHCII-H9, anti–MHCII-H11, or anti–MHCII-H13 did not protect mice against a challenge with PR8 influenza virus (Fig. 2B). In contrast, mice immunized with anti–MHCII-H1(PR8) were completely protected, consistent with previous results (18).

Conventional influenza vaccines contain three strains of influenza. In this study, we wanted to examine whether anti–MHCII-MIX could induce high levels of Abs. We chose to keep the total amount of DNA delivered constant (25 μg per mouse) to avoid a synergistic effect from simultaneous delivery of several HAs (25 μg each). Therefore, we mixed equal amounts (4.2 μg DNA) of anti–MHCII-H5, anti–MHCII-H6, anti–MHCII-H8, anti–MHCII-H9, anti–MHCII-H11, and anti–MHCII-H13 totaling 25 μg. To assess the potential enhancement of immune responses resulting from targeting of HA to MHCII molecules, we prepared a corresponding nontargeted control (anti–NIP-MIX).

BALB/c mice were immunized three times with anti–MHCII-MIX or anti–NIP-MIX (Fig. 3). Ab responses were detected against recombinant HAs of all subtypes included in the vaccine mixture (H5, H6, H8, H9, H11, and H13). Importantly, a single vaccination was sufficient for induction of detectable Abs against all HA subtypes (Fig. 3), even though titers were lower than those obtained with immunizations using 25 μg for each DNA vaccine separately (Fig. 2). The first and second boost markedly enhanced Ab levels (Fig. 3); they exceeded those seen after a single immunization with 25 μg of DNA (Fig. 2). In all instances, responses were increased by immunization with anti–MHCII-MIX compared with anti–NIP-MIX. More specifically, vaccination with anti–MHCII-MIX statistically increased Ab responses against HA from H8, H9, and H13 influenza viruses at weeks 7 and 14, whereas responses against HA from H5 and H6 influenza viruses were significantly increased first at week 14. With regard to H11, anti–MHCII-MIX did not significantly increase Ab responses above that of the nontargeted control vaccine. The differences in responses obtained after vaccination with anti–NIP-HA could reflect inherent differences in immunogenicity between different HA subtypes. In conclusion, strong anti-HA IgG Ab responses against HAs of each of six very different strains of influenza were induced by their simultaneous delivery in the context of MHCII-targeted DNA vaccines.

We next tested whether the mixture could also promote cross-reactive immune responses against subtypes not included in the mixture. To this end, we examined the Ab responses induced after vaccination with anti–MHCII-MIX or anti–NIP-MIX against inactivated PR8 and Cal07 in ELISAs (neither of these H1N1 influenza viruses had been included in the vaccine mixtures). Interestingly, a small, but persistent, IgG response was detectable against both viruses (Fig. 4A, 4B). However, the titers after vaccination with anti–MHCII-MIX were not significantly elevated above those after vaccination with anti–NIP-MIX, and sera from both of these vaccinations could not neutralize the H1 viruses in microneutralization assays (Fig. 4C, 4D). As positive controls in these experiments, we included anti–MHCII-H1(PR8) and anti–MHCII-H1(Cal07). As reported previously, vaccination with these two constructs elicited strain-specific Abs with neutralizing activity (18). The cross-reactive Abs observed after vaccination with anti–MHCII-MIX and anti–NIP-MIX are likely to be specific for common epitopes in the head region of HA, because an ELISA designed to detect stem-reactive Abs could not demonstrate significant increases after vaccination (Supplemental Fig. 2).

Similar to H5, H6, H8, H9, H11, and H13, H1 is a subtype belonging to group 1 influenza viruses. The group is classified based on shared stem structures (32), and we wanted to test whether vaccination with anti–MHCII-MIX and anti–NIP-MIX could induce Abs against subtypes classified as group 2 members. Thus, sera were examined for reactivity against HA from H3 and H7 subtypes of influenza. No significant H3- or H7-reactive Ab responses were detectable by ELISA compared with vaccination with NaCl, although sporadic responses were observed in a few mice (Supplemental Fig. 3).

Despite the absence of H1-reactive neutralizing Abs in sera (Fig. 4C, 4D), we decided to challenge mice with a lethal dose of Cal07 (Fig. 4E, 4F) or PR8 (Fig. 4G, 4H). As positive controls, and consistent with previous results, mice immunized with anti–MHCII-H1(PR8) or anti–MHCII-H1(Cal07) were protected against both of these H1 viruses by cross-protective T cells (18). More surprisingly, the mice that received anti–MHCII-MIX were protected against weight loss and death after challenge with Cal07 (p < 0.01) (Fig. 4E, 4F) and PR8 (p = 0.01) (Fig. 4G, 4H). In the Cal07 challenge, there was also significantly less weight loss and better survival in the anti–MHCII-MIX group compared with the nontargeted controls. In conclusion, vaccination with anti–MHCII-MIX confers a cross-reactive immune response that protects mice against heterosubtypical challenge with two H1 subtypes (Cal07 and PR8).

To test whether Abs contributed to the observed protection, sera collected 2 wk after three vaccinations (weeks 0, 4, and 8) were transferred i.v. to BALB/c mice, followed by challenge with H1 (PR8). Mice receiving sera from animals vaccinated with anti–MHCII-MIX had a significant delay in weight loss and death compared with mice vaccinated with anti–NIP-MIX or NaCl (p < 0.001) (Fig. 5A, 5B). On average, disease and death were delayed by 24 h in the mice that received sera from anti–MHCII-MIX–vaccinated mice compared with treatment with anti–NIP-MIX or NaCl. Thus, it appears that vaccination with anti–MHCII-MIX induced Abs that weakly protected against the group 1 virus H1 (PR8), a subgroup of HA not included in the vaccine mixture (Fig. 6).

The transfer of pooled sera from vaccinated mice constitutes a substantial dilution of donor-derived Abs in recipients. Thus, to more directly test the efficiency of Abs in the vaccinated mice, we depleted T cells in mice immunized three times prior to a challenge with PR8 virus. The depletion procedure removed >99% of all CD4⁺ and CD8⁺ T cells in the spleen (Supplemental Fig. 4). The results showed that depletion of T cells prior to influenza challenge had no effect on protection against PR8 (Fig. 5C, 5D). These data indicate that anti–MHCII-MIX induced cross-reactive protective Abs, in addition to strong Ab responses against each of the six subtypes of influenza HA included in the vaccine mixture. These experiments do not rule out that cross-protective T cells could also play a role in mice immunized with anti–MHCII-MIX.

The possible emergence of pandemic influenza viruses calls for novel and effective vaccination strategies. We have previously demonstrated that targeting of HA from H1N1 influenza viruses to MHCII molecules rapidly induced protective levels of neutralizing Abs in mice (18) and larger animals (21) and that APC-targeted DNA vaccines could be produced within 1 mo after identification of the viral sequence (28). In this article, we have demonstrated that this versatile and flexible format also contributed to improved Ab responses when extended to a mixture of six subtypes of influenza (H5, H6, H8, H9, H11, and H13). Moreover, when the six HA vaccines were mixed and injected simultaneously, Ab responses against the six HAs were maintained.

The levels of Abs elicited by each of the six HAs in the DNA vaccine mixture were clearly enhanced if the proteins were targeted to MHCII molecules. We have hypothesized that this may be due to the formation of APC–B cell synapses in which MHCII-targeted vaccine proteins bridge the two cell types (16, 27, 28). Indeed, APC–B cell synapses have been demonstrated in vitro (33); however, in vivo evidence is lacking. In such an APC–B cell synapse, HA could bind the BCR of HA-specific B cells, whereas the MHCII-specific scFv could bind an MHC⁺ APC. As such, the APC and B cell could both try to endocytose the vaccine protein, and the identical display of Ag/peptide MHCII molecules on APCs and B cells could lead to enhanced activation of CD4⁺ T cells. In sum, targeting of HA to MHCII molecules accelerates formation of protective immune responses.

In previous experiments, we have demonstrated that the MHCII-targeted vaccine proteins are secreted as dimers and that the two monomeric HAs in the antigenic units of a single dimeric vaccine molecule do not appear to interact with each other (18). Presumably, the hinge/CH₃-dependent dimerization occurs in the endoplasmic reticulum (ER), similar to the assembly of Ig H chains. Further, when vaccinating with a mixture, single cells are likely transfected with most, if not all, of the different plasmids (Fig. 6A). If so, in the ER of each cell, the different vaccine polypeptides are likely to randomly associate into dimeric vaccine proteins expressing two identical or two different HAs. The number of different possible combinations from n chains expressing n different HAs can be calculated using the formula [n (n + 1)]/2 (Fig. 6B). Thus, cells transfected with six vaccine plasmids, as used in this study, should yield 21 different vaccine dimers, of which 17% will express two identical HA monomers of H5, H6, H8, H9, H11, or H13. Although not known, one might speculate that the bivalent vaccine proteins are more potent than the heterodimeric univalent vaccine proteins, because the bivalent form may have an increased ability to cross-link BCRs of HA-specific B cells. If true, increasing the number of plasmids in the mix, while keeping the total DNA amount constant, will reduce the production of each of the bivalent homodimeric proteins (Fig. 6B), thereby likely reducing their immunogenicity. This effect might be detrimental to the induction of strain-specific Abs that can confer neutralizing activity against each HA in the vaccine mixture. In contrast, increasing the number of plasmids could increase the likelihood of eliciting Ab responses to shared subdominant epitopes, such as the HA stem, thus favoring induction of broadly cross-reactive Abs at the expense of strain-specific Abs against HA.

In addition to the development of Ab responses specific for each included HA, vaccination with anti–MHCII-MIX conferred some Ab-mediated protection against influenza subtypes that were not included in the vaccine mixture (H1N1 PR8 and Cal07). Abs were of relatively low titers, and reactivity with the stem of HA from H1 and H7 influenza viruses appeared to differ between individual mice. The cross-reactive Abs were not neutralizing in in vitro microneutralization assays. However, Abs might confer protection by complement-dependent lysis (34) or by Ab-dependent cell-mediated cytotoxicity (35–38). Further studies will be needed to assess this.

In addition to Abs, T cells could have contributed to the observed protection. Indeed, cytotoxic T cells play an important role in the clearance of influenza-infected cells (39, 40), and with the exception of H5 HA, the different HAs included in the vaccine mix share at least one known H-2K^d–restricted CD8⁺ T cell epitope that could have induced cross-reactive cytotoxic T cell responses against influenza PR8 (41). In addition to CD8⁺ T cells, influenza-specific CD4⁺ T cells have been shown to correlate with protection against disease (42). Relevant to this, a partially conserved BALB/c CD4⁺ T cell epitope is found among the various HAs in the vaccine mix and PR8 (43) (Supplemental Fig. 1). Most likely, Abs and T cells may have acted in concert to mediate the observed cross-protection against influenza PR8 that was observed in mice vaccinated with anti–MHCII-MIX.

Although conventional influenza vaccines typically contain two influenza A subtypes and one influenza B strain, in this study we injected mice simultaneously with MHCII-targeted HAs from six subtypes of influenza A. Importantly, homologous Ab responses were not impaired by the simultaneous delivery, and strong Ab responses were developed against each of the HAs included in the vaccine. This suggests that it would be possible to combine the most relevant seasonal and potential pandemic strains into a single vaccine mixture to confer increased breadth of protection in the recipient. In addition, we observed that MHCII-targeted vaccination with the HA mixture could raise low levels of cross-reactive Abs against a subtype that was not included in the vaccine mixture. Thus, the vaccine strategy could also confer some basic protection against an unexpected influenza emergence. To translate these results to humans, we have recently developed a targeting unit that is pan-specific for HLA class II molecules; this targeting unit should allow delivery of vaccine Ag to HLA class II molecules on APCs in all humans (21). Thus, the strategy presented in this article could be adapted to a clinical setting.

We thank Harvard Apparatus BTX for providing the skin electroporator. The technical help of Elisabeth Vikse is gratefully acknowledged.

B.B. and G.G. are inventors on patent applications filed on the vaccine molecules by Inven2 at the University of Oslo and Oslo University Hospital, according to institutional rules. B.B. is head of the scientific panel of Vaccibody AS and holds shares in the company. The other authors have no financial conflicts of interest.