Female BALB/c mice aged 6-8 weeks (Janvier, le Genest-Saint-Isle, France) were used in all experiments. All experiments involving research animals were pre-approved for ethics by the Norwegian Food Safety Authority. Cell work was performed with human embryonic kidney 293E cells purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA).

Anesthetized mice [0.1mg/10g: cocktail of Zoletil Forte (250mg/ml; Virbac France), Rompun (20mg/ml; Bayer Animal Health GmbH), and fentanyl (50µg/ml; Actavis, Germany)] were vaccinated intra muscularly (i.m.) with 25µg DNA (αMHCII-HA) into each quadriceps femoris, immediately followed by electroporation over the injection site (Elgen; Inovio Biomedical Co., Blue Bell, PA). The αMHCII-HA plasmid encodes HA from influenza A/California/07/2009 (H1N1), aa 18-541, linked to the MHCII-specific scFv via a dimerization unit consisting of the C_H3 domain of human IgG3 (22). All DNA vaccines were purified by using an EndoFree Plasmid Mega kit (catalog no. 12381; Qiagen, Hilden, Germany) and dissolved in sterile injection fluid (0.9% NaCl). Alternatively, anaesthetized mice were vaccinated i.m. with 1/10 human dose of Pandemrix with AS03 adjuvant (GlaxoSmithKline, Belgium), or a non-adjuvanted trivalent inactivated seasonal influenza vaccine [strains: A/Michigan/45/2015 (H1N1)pdm09-like virus, A/Singapore/INFIMH-16-0019/2016(H3N2)-like virus B/Colorado/06/2017-like virus (B/Victoria/2/87 lineage)].

Mice were anaesthetized as described above, and a 5xLD₅₀ dose of A/California/07/2009(H1N1) delivered in 10µl into each nostril. Mice were monitored daily for weight loss and euthanized at 80% of the original body weight. In figures, euthanized mice are scored as 80% for the remaining experimental time.

Draining LNs (iliac) were harvested and single cell suspensions prepared by GentleMACS dissociator (Miltenyi Biotech, Germany). Cells were stained with anti-CD3 (75-0032, Tonbo biosciences, San Diego, CA, USA), anti-GL7 (144603, Tonbo), anti-CD38 (102718, Tonbo), and anti-B220 (552771, BD Biosciences, Franklin Lakes, NJ, USA). HA reactivity was evaluated by binding to a His-tagged recombinant HA (Cal07) protein with an Y98F substitution (29) (rec.HA^Y98F), detected by anti-6xHis mAb (ab133714, Abcam, Cambridge, England). All samples were analyzed using an Attune NxT flow cytometer (Thermo Fisher Scientific, Waltham, MA, USA) and FlowJo software (ver.10).

Draining LNs were embedded in OCT mounting medium (00411243, Q Path, VWR, Radnor, PA, USA), immediately frozen on dry ice and stored at −80°C. Six-micrometer sections were collected on glass slides, air dried, fixed in room temperature acetone for 5 min, air dried, and blocked in 30% normal rat serum with FcRγ blocking reagent (10 μg/ml, HB-197). Sections were then incubated with 2µg/ml rec.HA^Y98F, followed by rabbit anti-HA(Cal07) pAb (11085-T54, Sino Biological, Inc) and anti-GL7-PE (144608, BioLegend, San Diego, CA, USA). Finally, colors were amplified using anti-FITC-Alexa Fluor 488 (A-11090, Thermo Fisher Scientific) and anti-R Phycoerythrin-Texas Red (ab34734, Abcam), and counterstained with DAPI. Sections were mounted with ProLong Diamond Antifade Mountant (P36970, Thermo Fisher Scientific). Images were acquired in a Nikon Eclipse Ti microscope using a Nikon S Plan Fluor 20x objective with a 0.60 numerical aperture and a Nikon Digital Sight Camera. All micrographs were analyzed and processed using ImageJ Version: 2.0.0-rc-69/1.52p, Build: 269a0ad53f.

Blood was harvested by puncture of the saphenous vein, and sera collected by centrifugation. ELISA plates (Costar 5390, Corning, Corning, NY) were coated with 0.5µg/ml rec. HA from A/California/07/2009 (11085-V08H, Sino Biological, Inc., Wayne, PA, USA), blocked with 2% BSA/PBS, and incubated with serially diluted serum samples assayed for individual mice. Captured serum antibodies were detected with anti-mouse IgG1-bio (553500, BD Pharmingen, San Diego, CA, USA), or anti-mouse IgG2a- bio (553502, BD Pharmingen), and streptavidin-alkaline phosphatase (RPN1234, GE Healthcare, Buckinghamshire, UK), or alkaline phosphatase conjugated goat anti-mouse IgG (A2429, Saint-Louis, MO, USA). Plates were developed with phosphatase substrate (P4744, Sigma-Aldrich).

Resistance to UREA wash was used to calculate avidity index. Captured serum antibodies were incubated for 10min with 2M UREA or PBS before detecting remaining serum antibodies with alkaline phosphatase conjugated goat anti-mouse IgG (A9316, Sigma Aldrich). AUC was calculated for the dilution curves and baseline for AUC was calculated based on NaCl serum levels. Avidity index is defined as AUC for samples treated with UREA divided by AUC for the corresponding PBS treated sample.

Bone marrow was harvested from femur and tibia. Single cell suspensions were prepared and seeded on MultiScreen HTS filter plates (MSIPS45, Merck Millipore Ltd., Tullagreen, Ireland) pre-coated overnight at 4°C with 0.5 μg/well of rec.HA (Cal07) (11085-V08H, Sino Biological), and incubated for 20h. Spots were detected with anti-mouse IgG (A1418, Sigma-Aldrich), developed with phosphatase substrate (P4744, Sigma-Aldrich) and analyzed in CTL- ImmunoSpot^® analyzer (CTL, Shaker Heights, OH, USA).

In vivo cellular cytotoxicity assay were adapted from Durward et al. (30). In brief, splenocytes were harvested and single cell suspensions prepared. Splenocytes were incubated with the MHC class I restricted influenza HA (Cal07) peptide IYSTVASSL, the NP peptide (RLIQNSLTIERMVLS), or a negative control peptide, at a density of 5x10⁷ cells/mL for 1 h at 4°C followed by 30 min incubation at 37°C. Peptide‐loaded cells were washed twice in PBS and subsequently stained with 5 μM CellTrace Violet (CTV) (C34557, Life Technologies) (HA peptide loaded cells), or 1 μM CellTrace Far Red (CTFR) (C34564, Life Technologies) (NP peptide loaded), or double stain (CTV and CTFR) (negative control) at a density of 5x10⁷ cells/mL for 20 min at 37°C. Cells were mixed in equal ratios (1:1:1), and a total of 15x10⁶ cells injected i.v. in a 100µl volume to vaccinated mice. Spleens were harvested 16h later, single cell suspensions prepared, and the presence of peptide loaded cells investigated by flow cytometry. The ratio of CTV to CTV/CTFR or CTFR to CTV/CTFR cells were calculated as % specific lysis = [1 − (average ratio in group with NaCl vaccinated mice/experimental ratio)].

Spleens from mice were collected 9 and 21 days after vaccination and homogenized through a wire mesh to get a single cell suspension by Lympholyte M (Cedarlane, Burlington, US) gradient centrifugation, and thereafter kept frozen in Fetal Bovine Serum (FBS, Sigma/Merck) with 10% DMSO (Sigma/Merck) at -150°C. Splenocytes were thawed, washed in RPMI 1640 (Gibco, Thermo Fischer Scientific) with 10% FBS (Sigma/Merck) and rested for 24 hours prior to stimulation with 5.6 µg HA/ml (400HAU) of A/California/07/2009 (H1N1) for 4 hours at 37°C in the presence of 2.5 µg/ml brefeldin A (BFA, Sigma/Merck). Positive control was stimulated with 50 ng/ml phorbol myristate acetate (PMA, Sigma/Merck) and 1 µg/ml ionomycin (Sigma/Merck). Cells were stained for viability (Live/dead aqua, Molecular Probes, Thermo Fischer Scientific) and extracellular markers in Brilliant staining buffer (BD Biosciences, San Antonia, CA, US) (each staining for 30 minutes at room temperature), then fixed and permeabilized (45 minutes at 4°C using Foxp3 fixation and permeabilization kit, eBioscience) prior to intracellular staining (1 hour 4°C). Percentage of positive CD3, CD4, CD8, CD44, CD62L, CD25, CD19, CD49b, Foxp3, CD107a, IFNγ, TNFα, IL-2 and IL-17A cells was analysed on a ZE5 flow cytometer (Bio-Rad, CA, US). Fractions of memory T cells; T effector memory (TEM) CD44⁺CD62L^-, and T central memory (TCM) CD44⁺CD62L⁺, and naïve T cells CD44^-CD62L⁺ as well as NK and NKT cells were also assessed using FlowJo_V10 (Tree Star, San Carlos, CA, US).

Splenocytes were harvested 21 days after vaccination, and single cell suspensions rested for 24 hours prior to stimulation with 5.6 µg HA/ml (400HAU) A/California/07/2009 (H1N1) for 4 hours at 37°C in the presence of 2.5 µg/ml brefeldin A (BFA, Sigma). Splenocytes were then stained for identification of IFNγ, IL-2, and TNFα positive CD4 and CD8 T-cells.

The p-values represent exact values calculated by unpaired non-parametric two-tailed Mann-Whitney tests. Weight curves were analyzed with two-way ANOVA, and survival curves with the Gehan-Breslow-Wilcoxon test. Statistical analysis of flow cytometry was performed using one-way ANOVA with the Holm-Sidiak multiple-comparison test. All analysis was performed using GraphPad Prim 9 software.

In order to compare the antibody kinetics of different vaccine strategies, we vaccinated mice once i.m. with either TIV, Pandemrix, or the MHCII-targeted DNA vaccine (αMHCII-HA). Both Pandemrix and αMHCII-HA are monovalent vaccines, designed to protect against A/California/07/2009 (H1N1), whereas TIV in addition to a pdm09 like strain contains an H3N2 strain and an influenza B strain. The plasmids encoding αMHCII-HA were formulated in a physiological saline solution (NaCl), while Pandemrix was formulated with the adjuvant solution AS03. Thus, both saline and AS03 were used as experimental negative controls.

Following a single vaccination with the different vaccines, serum samples were collected and monitored for HA specific antibodies over 180 days. Both αMHCII-HA and Pandemrix rapidly generated high titers of Cal07 HA specific IgG that were maintained over time. A peak was observed between 42 and 92 days post vaccination, and where the mice vaccinated with Pandemrix had significantly higher total IgG levels as compared to αMHCII-HA ( Figure 1A ). However, αMHCII-HA raised significantly higher antibody responses as compared to TIV, which only induced modest antibody responses in this system. After the peak, the responses seemed to reach a plateau from about day 106, and that were maintained for at least 180 days.

The antibody responses were highly strain specific, and only mild reactivity against the serologically different strain A/Puerto Rico/8/1934 (H1N1) (PR8) was detected after vaccination with Pandemrix or αMHCII-HA ( Figure 1B ). Interestingly, the difference between Pandemrix and αMHCII-HA seemed to be mostly due to a significantly higher amount of IgG1 antibodies following vaccination with Pandemrix ( Figure 1C ), while there were no significant differences in IgG2a levels ( Figure 1D ).

At day 180, mice were challenged with a lethal dose of influenza virus Cal07. Weight was monitored and used as an objective indicator of morbidity. As expected from the measured antibody responses, both αMHCII-HA and Pandemrix vaccination induced significantly improved protection as compared to TIV, characterized by near sterile protection and minimal weight loss after challenge ( Figure 1E ). In accordance with ethical requirements, mice that reached a 20% weight loss during the infection were euthanized. Importantly, none of the mice vaccinated with αMHCII-HA or Pandemrix reached this threshold ( Figure 1F ). Based on the low antibody levels associated with TIV, it may, however, be surprising that only 2/8 mice in this group lost 20% or more of their weight, as opposed to the negative control groups where 8/8 had to be euthanized. The mice receiving TIV lost weight until day 6 after infection, but from then on stabilized and regained weight ( Figure 1F ).

In sum, we observed that a single vaccination with either αMHCII-HA or Pandemrix could raise strong and long-lasting strain specific protective antibody responses against HA.

Time is essential during a pandemic outbreak, with respect to both production time and the generation of protective immunity. Thus, we investigated how fast the different vaccines were able to induce protective immunity. BALB/c mice were vaccinated once with αMHCII-HA, Pandemrix, TIV, or controls, and sera examined for antibody responses at day 7 after vaccination. Importantly, a majority of the mice vaccinated with either αMHCII-HA or Pandemrix had detectable levels of HA specific IgG, but there were also some that had not yet seroconverted ( Figure 2A ). TIV immunized mice did not display any serum antibodies at day 7. Interestingly, αMHCII-HA was the only vaccine that could induced any antibody responses against the heterologous strain PR8, albeit only in 2 out of 16 mice in the group ( Figure 2B ). For homologous antibody responses against HA from Cal07, IgG1 and IgG2a levels were similar for Pandemrix and αMHCII-HA, but as was observed for total IgG, not all mice had seroconverted at this early time point ( Figures 2C, D ).

At day 8 after a single vaccination, mice were given a lethal dose of influenza Cal07 virus and weight was monitored. Interestingly, mice receiving αMHCII-HA lost significantly less weight as compared to mice vaccinated with Pandemrix, but there was a smaller initial weight loss also in this group ( Figure 2E ). The trend also held when assessing survival as defined by a 20% weight loss, with αMHCII-HA displaying a 25% relative improved survival rate, albeit not statistically significant, when compared to Pandemrix ( Figure 2F ). Mice vaccinated with TIV were not protected 7 days after challenge. None of the vaccines induced protection against the heterologous PR8 virus at day 8 post vaccination ( Figures 2G, H ).

Taken together, we found that a single vaccination with either Pandemrix or αMHCII-HA could rapidly lead to seroconversion that translated into protection already one week after vaccination. In addition, vaccination with the MHCII-targeted DNA vaccine significantly improved morbidity as compared to Pandemrix.

To characterize the antibody response induced after vaccination with the different vaccines in detail, we first investigated the presence of plasma cells in bone marrow following vaccination. Thus, bone marrow was harvested from mice that had been vaccinated once with αMHCII-HA or Pandemrix. Single cell suspensions were prepared, and the number of anti-HA (Cal07) secreting cells assayed by ELISpot. At day 9, we detected no anti-HA secreting cells in bone marrow, but by day 14 anti-HA secreting cells had formed for both these vaccines. Although low, at day 21 the levels had doubled, indicating a steady rise in anti-HA secreting cells in response to vaccination. The development was similar for both αMHCII-HA and Pandemrix, but there was a tendency that Pandemrix had slightly higher numbers of plasma cells at day 14 and 21 ( Figure 3A ).

To evaluate long-term responses, we also examined plasma cells at day 180 in bone marrow following vaccination with TIV, αMHCII-HA, and Pandemrix. Interestingly, mice vaccinated with αMHCII-HA had significantly higher levels of plasma cells in the bone marrow as compared to mice receiving Pandemrix. Mice vaccinated with TIV did not show any plasma cells in response to vaccination after 180 days ( Figure 3B ).

Next, we wanted to investigate the avidity of the vaccine induced antibodies and set up an assay measuring the resistance to UREA wash as an indication of antibody binding avidity. In this assay, relative signals between washing with UREA or PBS in ELISA were compared, and an avidity index of 0 or 1 indicated no resistance or absolute resistance to UREA wash, respectively. Sera collected from mice at day 7, 14, and 21 post vaccination were assayed. In accordance with the above results from ELISA ( Figures 1 , 2 ), we did not observe significant differences in antibody titers for Pandemrix and αMHCII-HA. However, mice vaccinated with Pandemrix demonstrated a steady increase in serum antibodies with higher avidity towards HA from Cal07 ( Figure 3C ), mimicking the tendency observed for plasma cells in bone marrow ( Figure 3A ). At day 21, mice vaccinated with Pandemrix had significantly more serum antibodies with increased avidity as compared to mice vaccinated with αMHCII-HA, even though serum antibody levels were similar between the two vaccines ( Figure 3C ).

In sum, we found that Pandemrix induced antibodies with higher avidity more rapidly than αMHCII-HA (day 21). However, with time (6 months) the level of plasma cells in response to αMHCII-HA vaccination was significantly higher than after vaccination with Pandemrix.

The early presence of plasma cells in bone marrow could indicate a strong germinal center (GC) reaction to the vaccine antigen ( Figure 3A ). Thus, we wanted to investigate the formation of GCs, as well as the presence of HA reactive B cells with a GC phenotype.

Mice were vaccinated once and draining lymph nodes (LN) (iliac) harvested on days 9, 14, and 21. Cells from the prepared single cell suspensions were stained for GC B cells (B220⁺ CD38^lo GL7⁺), identified by a recombinant HA probe ( Figure 4A ). A steady rise in HA reactivity among B cells was observed from day 9 through day 21 post vaccination ( Figure 4B ). Mice receiving αMHCII-HA had GC B cells with a significantly elevated HA reactivity. However, Pandemrix induced a stronger GC reaction, and although the percentage of GC HA reactivity was lower in these mice, the total number of HA reactive GC B cells was significantly higher than after vaccination with αMHCII-HA ( Figure 4C ).

The increased levels of GC B cells observed after vaccination with Pandemrix was likely augmented by the adjuvant AS03. Pandemrix is a split vaccine and also contains other antigens than HA. Thus, a new experiment was performed for day 21 post vaccination to also include an adjuvant control group and the non-adjuvanted split vaccine TIV. As expected, vaccination with αMHCII-HA again induced the highest percentages of HA reactive B cells with a GC phenotype in the LNs. The control groups vaccinated with NaCl or AS03 defined the background, and we observed that TIV induced low, but significant, levels of HA reactive GC B cells ( Figure 4D ). When looking at the total number of HA reactive GC B cells, however, we again observed that Pandemrix induced the highest absolute numbers ( Figure 4E ).

The formation of GC was also evaluated by microscopic imaging of LNs harvested at day 21 after vaccination. The cryopreserved LN sections were stained with the GC activation marker GL7, the recombinant HA probe, and DAPI ( Figure 4F ). Multiple GC structures with HA reactivity were observed for the mice receiving αMHCII-HA or Pandemrix, in accordance with the flow cytometry data.

Taken together, the data demonstrate that vaccination with αMHCII-HA induced a response where the reactivity of GC B cells was focused on HA. Pandemrix induce a stronger immune response in total, and as such had a higher total number of GC B cells and HA reactive GC B cells.

The fairly similar B cell activation and antibody levels observed after vaccination with Pandemrix or αMHCII-HA led to the question of whether differences in induction of cytotoxic T cells could explain the improved survival that was observed in mice at day 8 after vaccination with αMHCII-HA ( Figure 2 ). Thus, single cell suspensions of splenocytes from naïve mice were loaded with MHC class I restricted peptides from HA, influenza nucleoprotein (NP), or an irrelevant peptide as negative control, and stained with cell trace dyes. Next, the peptide loaded splenocytes were injected i.p. into mice immunized 9 days or 8 weeks prior with TIV, Pandemrix, αMHCII-HA, or NaCl. Following a 16h incubation, spleens were harvested and the presence of transferred cells investigated by flow cytometry ( Figure 5A ). The ratios of HA or NP peptide loaded splenocytes to the irrelevant peptide negative control in NaCl treated mice was used as a reference to calculate the relative specific lysis of HA or NP peptide loaded cells in the vaccinated groups.

Importantly, mice vaccinated 9 days earlier with αMHCII-HA displayed a strong cytotoxic response towards the HA peptide that was about 10-fold higher than that observed in mice vaccinated with Pandemrix ( Figure 5B ). Vaccination with Pandemrix raised a similar cytotoxic response against both NP and HA, while αMHCII-HA, as expected, did not induce any cytotoxic activity towards the NP peptide ( Figure 5C ). Interestingly, TIV induced similar responses as Pandemrix against both groups of peptides. In order to also examine the long-term cytotoxic potential, we set up a similar experiment 8 weeks post vaccination. The cytotoxic responses were markedly reduced in all the vaccine groups as compared to day 9, but the group vaccinated with αMHCII-HA maintained a significant and strong cytotoxic response ( Figure 5D ).

In sum, the data clearly demonstrated that αMHCII-HA induced a superior cytotoxic response both at early and later time points as compared to Pandemrix. The early induction of cytotoxic immunity could explain the improved protection observed after vaccination with αMHCII-HA and the viral challenge at 8 days post vaccination ( Figure 2 ).

T cells with an enhanced effector function have often been characterized by the dual secretion of two or more key cytokines (31), and the significant difference observed for cytotoxic responses between the vaccine groups ( Figure 5 ) points towards different functional T cell profiles. Thus, we investigated the secretion of IFNγ, IL-2, and TNFα in T cells from splenocytes harvested 21 days after a single vaccination, and stimulated ex vivo with HA (Cal07). In accordance with the improved cytotoxic response following vaccination with αMHCII-HA ( Figure 5 ), the highest numbers of IFNγ, IL-2, and TNFα secreting CD4 T-cells were observed for this group ( Figure 6A ). TIV induced somewhat higher numbers of cytokine secreting cells as compared to Pandemrix. The trend held also for double secreting CD4 T-cells. Triple secreting CD4 T-cells were not observed for any vaccine groups ( Figure 6A ). A similar trend was observed for the CD8 T-cells, but with somewhat higher levels ( Figure 6B ).

While the cytokine profiles indicated an increased level of effector T-cells, analysis of cells with effector memory (T_EM) and central memory (T_CM) phenotype was performed for CD4 and CD8 T-cells after ex vivo antigen stimulation. T_EM was identified as CD4/CD8⁺ CD44⁺ CD64L^- cells and T_CM was identified as CD4/CD8⁺ CD44⁺ CD64L⁺ cells. The T_EM subsets were slightly elevated in mice receiving αMHCII-HA for both CD4 and CD8 T-cells, whereas the other vaccine groups were similar to background ( Figures 6C, D ). T_CM were not clearly elevated above background levels (NaCl and AS03), although the percentage of CD8 T_CM seemed to be higher in mice receiving TIV compared to αMHCII-HA ( Figures 6E, F ).

Taken together, the data confirms an increased potential for activation of multifunctional T cells following vaccination with αMHCII-HA, as compared to Pandemrix and TIV. Further, immunization with αMHCII-HA increased the T_EM levels of both CD4 and CD8 T cells, indicating a strong protective effect. This is reflected in the highest survival rate after viral challenge compared across the vaccine platforms tested.