A/Puerto Rico/8/34 (H1N1/1934), A/New Caledonia/20/99 (H1N1/1999), A/California/7/09 (H1N1/2009), and A/Brisbane/02/18 (H1N1/2018) were sourced from the National Institute of Biological Standards and Controls (NIBSC, Potters Bar, United Kingdom) and propagated in the allantoic cavity of embryonated chicken eggs (19). After 72 hours of incubation at 37°C, the virus strains were isolated, purified from allantoic fluid, and stored at -80°C until further use. The egg-cultured H1N1/2018 strain was gifted by Mark Pronk from Erasmus University in Rotterdam, The Netherlands.

WIV vaccines of H1N1/1934, H1N1/1999, and H1N1/2009 influenza virus strains were prepared by overnight incubation of the virus strains with 0.1% v/v β propiolactone (Acros Organics, Geel, Belgium) under continuous rotation at 4°C. The WIV was then dialyzed against 4- (2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)-buffered saline (Thermo Fisher Scientific, Bleiswijk, the Netherlands) at 4°C overnight to remove β-propiolactone. Inactivation of the viruses was confirmed by inoculating Madin-Darby canine kidney (MDCK) cells and assessing virus replication with hemagglutination assay (20). Subunit vaccines of H1N1/1934, H1N1/1999, and H1N1/2009 were prepared from their respective WIV as described before (21). The preparations, consisting mainly of hemagglutinin (HA), were used for coating enzyme-linked immunosorbent assay (ELISA) plates. Neuraminidase (NA) and cH9/1 proteins were prepared using the baculovirus expression system. The cH9/1 protein contains the HA stalk from H1N1/2009 and the HA head from the unrelated A/guinea fowl/Hong Kong/WF10/99 (H9N2) virus (22, 23). The total protein concentration of WIV and surface proteins was determined with the micro-Lowry assay (19).

The Central Committee for Animal Experimentation of the Netherlands approved the animal procedures of this study (CCD application number AVD105002016599). Before the start of experiments, female CB6F1 (C57Bl/6 x BALB/c F1) mice of 6-8 weeks (Envigo, Horst, the Netherlands) were acclimatized for 1.5 weeks. Mice were co-housed in open cages and had ad libitum access to sterilized water and a standard diet. Mice were divided into five experimental groups, each containing six mice ( Figure 1 ). For the negative control, untreated mice were used. The mice from the other four groups were vaccinated with a total dose of 5 µg of WIV (≈ 1.6 µg of HA) of the H1N1/2009 influenza virus strain. For extended antigen delivery, the 5 µg dose of WIV was administered over extended periods of 14 (0.36 µg/day), 21 (0.24 µg/day), or 28 days (0.18 µg/day) by daily subcutaneous injections. The immunization schedules were designed to have the last injection for each group on day 28. Furthermore, positive control mice were immunized through an intramuscular prime on day 0 and boost injection on day 28 (each 2.5 µg). Blood was drawn weekly by cheek puncture to follow total IgG levels during the experiment. On day 42, all mice were euthanized, and blood was collected through heart puncture. Blood was clotted overnight at room temperature, centrifuged for 10 min at 3000 x g, and serum was used for ELISA and hemagglutination inhibition assays (HAI). Spleens, inguinal lymph nodes, and axillary lymph nodes from mice were collected in complete Iscove’s Modified Dulbecco’s Medium (cIMDM; Thermo Fisher Scientific) containing 10% v/v deactivated fetal bovine serum (dFBS, Lonza, Basel, Switzerland), 100 µg/mL streptomycin (Invitrogen, Breda, the Netherlands), 100 U/mL penicillin (Invitrogen) and 50 µM 2-mercaptoethanol (Invitrogen) Lymph nodes and spleens were used for flow cytometry and Enzyme-Linked Immunosorbent Spot (ELISPOT), respectively.

Axillary and inguinal lymph nodes of each mouse were pooled, and cells were isolated by mechanically crushing the tissues between two microscopy slides in 3 mL ice-cold cIMDM (24). Falcon tubes with cell strainer caps (35 µm, Corning, Amsterdam, the Netherlands) were used to remove clumps of cells prior to cell counting and flow cytometry staining. For staining, 1 x 10⁶ cells were transferred to a 96-well plate and centrifuged at 600 x g for 5 min. at 4°C. The cells were washed with phosphate-buffered saline (PBS, Thermo Fisher Scientific) and incubated with ZombieNIR-labeled viability dye for 15 min. ( Supplementary Table 1 ). Next, cells were washed and blocked for 15 min. with 0.05 mL blocking buffer [98% v/v FACS buffer [PBS + 2% v/v dFBS (Lonza)] and 2% v/v Fc receptor binding inhibitor (Thermo Fisher Scientific)]. Then, cells were washed, incubated for 30 min. with 0.05 mL of antibody mix in blocking buffer ( Supplementary Table 1 ) washed again, and fixed by incubating for 10 min. with 0.15 mL fixation buffer (Thermo Fisher Scientific). After washing the cells, samples were analyzed on a BD FACSymphony A5. In all washing steps, cells were washed twice with FACS buffer and centrifuged at 600 x g for 5 min. at 4°C. Data analysis was performed by the Kaluza Analysis Software version 2.1. Unstained cells were used to set the optimal flow cytometry voltage for each fluorophore and Fluorescence Minus One (FMO) controls were used to set the gates ( Supplementary Figure 1 ). Absolute numbers of immune cells were determined from 1 x 10⁶ live cells.

To determine the level of influenza-specific interferon (IFN)-γ and interleukin (IL)-4 secreting splenocytes, murine ELISpot kits (MABTEC, the Netherlands) were used according to the manufacturer’s instructions. Briefly, spleens were homogenized using a GentleMACS dissociator (Miltenyi Biotec, Leiden, the Netherlands) and treated with Ammonium-Chloride-Potassium lysis buffer (0.83% NH₄Cl, 10 mM KHCO₃, 0.1 mM ethylenediaminetetraacetic acid (EDTA)) to lyse erythrocytes. Splenocytes (5 × 10⁵/well) were incubated in cIMDM medium and stimulated with or without 10 µg/mL of H1N1/2009 WIV. After 16h of incubation, plates were incubated with alkaline phosphatase-conjugated anti-mouse IFN-γ or IL-4 antibodies. Spots were developed using nitro-blue tetrazolium chloride/5-bromo-4-chloro-3′-indolyphosphate p-toluidine substrate and counted with an AID ELISpot reader (Autoimmune Diagnostika GmbH, Strassberg, Germany). To assess influenza-specific cellular responses, the number of IFN-γ or IL-4 secreting splenocytes was determined by subtracting the number of spots in the unstimulated wells from the number of spots in the WIV-stimulated wells.

ELISA was performed to measure WIV-, HA subunit-, HA-stalk-, and NA-specific IgG, or WIV-specific IgG1 and IgG2a levels, as previously described (20, 21). Briefly, to determine IgG titers, plates were coated with WIV (0.3 µg/well), HA subunit (0.1 µg/well), cH9/1 (HA-stalk, 0.1 µg/well), or NA (0.2 µg/well) proteins. Serum samples were prediluted 1:200 to determine WIV- and HA subunit-specific IgG or diluted 1:100 for HA stalk- and NA-specific IgG and applied in a 2-fold serial dilution on the plates. To measure IgG1 and IgG2a levels, serum samples were applied in a 1:100 dilution and concentrations were determined from an IgG1 or IgG2a standard curve (Southern Biotech, Birmingham, USA). Horseradish peroxidase-linked goat anti-mouse IgG, IgG1, or IgG2a antibodies (Southern Biotech, Birmingham, USA) were used to detect bound IgG, IgG1 or IgG2a, respectively. IgG titers were calculated as log₁₀ of the reciprocal of the serum sample dilution corresponding to an absorbance of 0.2 at a wavelength of 492 nm.

To assess the avidity of the influenza-specific IgG, the ELISA was performed as above but after incubation of the plates with the diluted serum samples, 8M urea (Sigma, St. Louis, MO, USA) was added to the wells for 1h followed by washing and incubation with horseradish peroxidase-linked goat anti-mouse IgG secondary antibody (Southern Biotech). The avidity index was calculated as: (OD₄₉₂ urea treated samples/OD₄₉₂ untreated samples) x 100%.

HAI assay was performed following a previously described method (25). Briefly, hemagglutination units (HAU) of each influenza virus strain (H1N1/1934, H1N1/1999, H1N1/2009, or H1N1/2018) were determined by measuring the hemagglutination activity of virus suspensions that were serially diluted in the presence of 1% guinea pig erythrocytes obtained from Envigo. HAU were calculated as the reciprocal of the highest virus dilution causing complete hemagglutination of erythrocytes. Next, serum samples were diluted 1:3 with receptor-destroying enzyme (Seiken, Japan) and incubated for 16 hours at 37°C. The following day, the receptor-destroying enzyme was inactivated by treating the serum samples for 30 min. at 56°C. The serum samples were then applied in a 2-fold serial dilution in PBS (Thermo Fisher Scientific) and incubated for 40 minutes with 4HAU of the different live influenza virus strains. Next, 1% of guinea pig erythrocytes was added and incubated for 2 hours at room temperature (RT). Serum titers are expressed as the reciprocal of the highest dilution of serum that prevented erythrocyte agglutination.

Statistical analysis of results was performed using GraphPad Prism version 9.5.1 (La Jolla, CA, USA). Normal distribution of data was confirmed using the Kolmogorov-Smirnov test. Normally distributed data were assessed with one-way ANOVA and Tukey post-hoc test. Corrections for multiple comparison was done with Benjamini, Krieger and Yekutieli corrections. p < 0.05 was considered as significant; * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

To investigate whether extended delivery of influenza vaccine quantitatively enhances the induction of cellular and humoral immune responses, mice were immunized with H1N1/2009-derived WIV vaccine by daily subcutaneous injections for periods of 14, 21, or 28 days. A prime-boost immunization group was used as a control ( Figure 1 ).

Among the immunizations for extended periods of time, 28 days of extended antigen delivery induced the highest levels of cellular immune responses, followed by 21 and 14 days of delivery ( Figure 2 ). This trend was evident for the level of total CD4+ T cells in the lymph nodes ( Figure 2A ) and influenza virus-specific IL-4-secreting splenocytes ( Figure 2C ). The mice of the 28-days group also showed the highest level of influenza virus-specific IFN-γ-secreting splenocytes ( Figure 2B ), but differences with the other groups did not reach statistical significance. Furthermore, mice of the 28-days group showed slightly higher and more consistent numbers of IFN-γ and IL-4 secreting splenocytes than mice that received prime-boost injections. Therefore, prolonging the delivery of WIV vaccine to 28 days led to more robust cellular immune responses than prime-boost immunization.

Next, we determined influenza virus-specific antibody levels in the serum ( Figure 3 ) and germinal center responses in the lymph nodes ( Figure 4 ). Influenza virus-specific IgG antibodies were followed weekly up to 6 weeks ( Figure 3A ). In mice of the 14-, 21-, and 28-days groups, IgG titers increased gradually over the administration period and leveled off on the day of the last injection (day 28) without further increase until day 42. In contrast, in prime-boost immunized mice antibody titers peaked at day 21 after the first immunization and increased further after the booster on day 28. On day 42, we determined endpoint titers ( Figure 3B ) and found the highest antibody titer in the 28-days group, followed by the 21- and 14-days groups. Extended antigen delivery for 28 days induced a significantly higher antibody titer than prime-boost immunization, whereas extended antigen delivery for 14 or 21 days induced similar levels. Regarding HAI antibodies ( Figure 3C ), the 28-days group showed the highest level of HAI antibodies, followed by the 21-days group. Only the mice of the 21 days (5/6) and 28 days (6/6) groups seroconverted by inducing a HAI titer of ≥1:40, whereas the 14 days and prime-boost groups did not induce any detectable HAI antibodies.

To investigate whether the enhanced antibody response after extended antigen delivery was associated with the induction of a more robust germinal center (GC) reaction, we quantified the cells of the GC in the lymph nodes of mice two weeks after the last injection ( Figure 4 ; Supplementary Figure 2 ). Our data show that delivery of WIV for 14 days or 21 days, and prime-boost immunization induced higher numbers of GC B (B_GC) cells than found in control mice. To our surprise, extended antigen delivery for 28 days did not increase the frequency nor the absolute number of B_GC cells in the lymph nodes ( Figures 4B, C ). Within the B_GC cell population, we measured the frequency of light zone (LZ) and dark zone (DZ) B_GC cells to detect an increase in affinity selection or in proliferation and somatic hypermutation, respectively (26). Our data show that prolonging the antigen delivery increased the frequency of LZ B_GC cells ( Figure 4C ), whereas it reduced the frequency of DZ B_GC cells ( Figure 4D ). However, the absolute number of LZ B_GC cells and DZ B_GC cells decreased after prolonged antigen delivery ( Supplementary Figures 2A, B ). Furthermore, the LZ/DZ ratio increased in the lymph nodes of these mice ( Figure 4E ; Supplementary Figure 2C ), as was also observed in the prime-boost injected mice. Therefore, our data imply that the GC reaction in mice of the 28-days group and the prime-boost group is dominated by affinity selection rather than by proliferation and somatic hypermutation two weeks after the last injection.

In addition to B_GC cells, the frequencies of follicular T helper (T_FH) cells and CD138⁺ plasmablasts, the precursors of long-lived plasma cells, were measured in the lymph nodes of the immunized mice. As shown in Figure 4F , the 28-days group showed the lowest frequency of T_FH cells in lymph nodes, followed by the 21- and 14-days groups. In contrast, the frequency of T_FH cells in the prime-boost immunized group remained at similar levels to the control group. The reduction in B_GC cells and T_FH cells frequencies coincided with an increased frequency of plasmablasts ( Figure 4H ). Only the 28-days group showed a statistically significant increase in the frequency of plasmablasts as compared to the control group. Therefore, our data indicates that prolongation of the vaccine delivery induced a robust GC reaction (elevated levels of plasmablasts in lymph nodes and serum IgG antibodies) that had gone into termination (small numbers of T_FH cells and B_GC cells, high frequencies of LZ B_GC cells) two weeks after the last injection.

In addition to the abovementioned quantitative measures, we assessed the quality of the induced antibody responses by measuring the antibody avidity and the levels of IgG isotypes ( Figure 5 ). Our data demonstrate that the IgG antibodies of mice from different immunization groups did not differ significantly in the avidity index, only the 14-days group showed a slightly but not significantly higher avidity index than the other immunization groups ( Figure 5A ). Nevertheless, the total level of high avidity IgG was highest in the mice of the 28-days group ( Figure 5B ) reflecting the high amount of influenza-specific IgG in this experimental group and this level was significantly higher than in the prime-boost group.

We also followed the titers and kinetics of antiviral IgG1 ( Figure 5C ) and IgG2a (27, 28) ( Figure 5D ) in the mice during the study. The levels of influenza virus-specific IgG1 started increasing seven days after the first injection in the 14-, 21-, and 28-days groups. However, IgG1 levels in the prime-boost group only started increasing at 14 days post-immunization. The 28-days group displayed the highest level of IgG1, followed by the 21-days, 14-days, and prime-boost groups ( Figure 5E ). The kinetics of influenza virus specific IgG2a were quite different compared to IgG1 kinetics. In the mice of the 14-days, 21-days, and the prime-boost groups, IgG2a levels were minorly increased only on day 42. However, in the 28-days group, the levels of IgG2a already started rising seven days after the first injection and further increased very significantly after antigen delivery stopped. Consequently, only the 28-days group showed a high endpoint IgG2a level ( Figure 5F ), resulting in balanced amounts of IgG1 and IgG2a ( Figure 5G ). In contrast, the IgG2a/IgG1 ratio in mice of the other immunization groups was skewed towards IgG1. These results suggest that prolonging the delivery of WIV to 28 days enhanced the quality of the induced antibody response.

In addition to the abovementioned qualitative measures of the induced immune response, we measured cross-reactive antibodies against antigenically distinct influenza viruses and antibodies specific for the HA stalk and for NA ( Figure 6 ). Firstly, antibodies were assessed against the historic strains H1N1/1934 ( Figure 6A ) and H1N1/1999 ( Figure 6B ), and against H1N1/2018, a drift variant of the H1N1/2009 strain ( Figure 3C ) with the HAI assay. We observed that HAI antibodies were only induced against the H1N1/2018 strain. As with the homologous virus ( Figure 3C ), the highest levels of HAI antibodies against the H1N1/2018 strain were found in the 28-days group, followed by the 21-days group. The 14-days and prime-boost groups did not display measurable HAI antibodies against the drifted strain. These data show that prolonging antigen delivery enhanced the induction of antibodies specific for the head domain of the homologous and drifted strain.

Secondly, we measured antibodies against the HA protein of the historic strains H1N1/1934 ( Figure 6D ) and H1N1/1999 ( Figure 6E ) with ELISA, using subunit vaccines of the respective strains as coating. Although HAI antibodies against the historic strains were below detection limit, we did find antibodies binding to HA of those strains. Of the different immunization groups, the 28-day group showed the highest titers, followed by the 21-days, 14-days, and the prime-boost group. Thus, our data demonstrate that, other than HAI antibodies, cross-reactive antibodies are induced against the HA of historic strains.

Thirdly, the antibodies specific for the conserved HA-stalk ( Figure 6F ) and the conserved NA protein ( Figure 6G ) were measured with ELISA. Our data show that all immunization regimens induced antibodies reacting with the H1 stalk. Moreover, 28 days of extended antigen delivery resulted in the highest levels of HA-stalk-specific antibodies, followed by 21 days and 14 days. Levels of HA-stalk specific antibodies in the 28-days group were significantly higher than in the prime-boost group. On the other hand, NA-specific antibodies were not effectively induced by any of the regimens. Thus, these data show that prolonging delivery of a WIV influenza vaccine enhanced the induction of cross-reactive antibodies specific for the conserved HA-stalk region.