An inactivated multivalent influenza A virus vaccine is broadly protective in mice and ferrets

Abstract

Influenza A viruses (IAVs) present major public health threats from annual seasonal epidemics and pandemics and from viruses adapted to a variety of animals including poultry, pigs, and horses. Vaccines that broadly protect against all such IAVs, so-called “universal” influenza vaccines, do not currently exist but are urgently needed. Here, we demonstrated that an inactivated, multivalent whole-virus vaccine, delivered intramuscularly or intranasally, was broadly protective against challenges with multiple IAV hemagglutinin and neuraminidase subtypes in both mice and ferrets. The vaccine is composed of four β-propiolactone–inactivated low-pathogenicity avian IAV subtypes of H1N9, H3N8, H5N1, and H7N3. Vaccinated mice and ferrets demonstrated substantial protection against a variety of IAVs, including the 1918 H1N1 strain, the highly pathogenic avian H5N8 strain, and H7N9. We also observed protection against challenge with antigenically variable and heterosubtypic avian, swine, and human viruses. Compared to control animals, vaccinated mice and ferrets demonstrated marked reductions in viral titers, lung pathology, and host inflammatory responses. This vaccine approach indicates the feasibility of eliciting broad, heterosubtypic IAV protection and identifies a promising candidate for influenza vaccine clinical development.

INTRODUCTION

Influenza A viruses (IAVs) pose a continual major public health threat. Globally, endemic or “seasonal” influenza results in 3 to 5 million severe illnesses and up to 650,000 deaths each year (1). Influenza pandemics, in which human-adapted IAVs unpredictably emerge from the enormous IAV reservoir of wild waterfowl and domestic animals (2) and against which most humans lack protective immunity, can have even larger global impacts (3), such as the 1918 influenza pandemic, which resulted in at least 50 million deaths (4). In addition, IAVs adapted to nonhuman hosts emerge sporadically to cause severe human zoonotic infections (such as poultry-associated H5N1 and H7N9) (5) or even pandemically (such as pandemic H1N1 “swine” influenza in 2009). The fact that IAVs are permanently adapted to, or repeatedly infect, a wide variety of warm-blooded animal hosts such as horses, dogs, seals and more than 100 avian species indicates that IAV risks to humans are widely distributed in nature (6, 7). Furthermore, these viruses are composed of a broad array of different subtypes and genotypes of variable and often unpredictable human pathogenicity. Currently, the only IAV vaccines licensed for human use are made each year to match specific circulating human influenza virus strains in both Northern and Southern Hemispheres (8). Such vaccines do not protect against antigenically variant annual IAV strains, new pandemic viruses, poultry-associated viruses, or viruses adapted to or frequently infecting other mammalian hosts. There is a critical need for influenza vaccines that broadly protect against all such IAVs, a so-called “universal” vaccine (9, 10).

IAVs are enveloped, negative-sense, single-stranded RNA viruses with eight genome segments (11). IAVs express two major surface glycoproteins, hemagglutinin (HA) and neuraminidase (NA), and are subtyped by antigenic characterization of the HA and NA glycoproteins. Sixteen HA and nine NA subtypes are consistently found in avian hosts in various combinations (for example, A/H1N1 or A/H3N2), and these wild bird viruses are thought to be the ultimate source of human pandemic influenza viruses (6). IAV genome segmentation allows for viral genetic reassortment, and because HA and NA are encoded on separate gene segments, IAVs of any of the 144 possible subtype combinations can theoretically be generated after mixed infections in a host, in a process called “antigenic shift.” IAVs are also evolutionarily dynamic RNA viruses with high mutation rates. Mutations that change amino acids in the antigenic regions of HA and NA proteins, or in other regions that affect antigenicity, may allow human-adapted strains to evade human population immunity, a process termed “antigenic drift.” Despite enhanced surveillance, future pandemics cannot yet be predicted, including when and where a pandemic virus strain will emerge, what the viral subtype will be, how pathogenic it will be in humans, or whether there will be some immunologic cross-reactivity with prior circulating IAVs.

An effective universal vaccine would ideally provide broad protection against all IAV subtypes with pandemic or zoonotic potential that are circulating in animals (in particular, birds and domestic mammals). A universal influenza vaccine could also be used as a superseasonal vaccine that provides protection against antigenically variable seasonal strains, minimizing the need for annual predictive vaccine strain antigenic matching (10). Efforts to develop such broadly protective vaccines have been underway for decades (12) and have included experimental vaccines specifically targeting the M2 ectodomain (13, 14) or NA (15, 16) proteins to stimulate the development of protective antibody responses, vaccines based on antigens that stimulate the development of T cell responses (17), and most recently, a variety of vaccine approaches targeting antigenically conserved epitopes on the HA head and stalk, as recently reviewed (18, 19). Although several vaccine candidates have advanced into early clinical development (20–22), it has not yet been determined whether these strategies will provide broad protection in humans. Moreover, a practical vaccine inducing broad heterosubtypic protection has not yet been demonstrated with any of the above approaches.

Previously, we reported a vaccine consisting of a cocktail of four viral-like particles expressing HA proteins that showed heterosubtypic protection against mismatched and completely heterosubtypic challenges in mice (23). In the present study, the protective efficacy of an inactivated multivalent IAV vaccine without antigenic matching to challenge viruses was evaluated in mice and ferrets. The vaccine cocktail included four β-propiolactone (BPL)–inactivated wild-type, low-pathogenicity avian influenza viruses: H1N9, H3N8, H5N1, and H7N3. The four HA subtypes were chosen to reflect the subtypes of currently circulating annual IAV strains (H1 and H3) or recent consequential epizootic IAV infections (H5 and H7) and represent both major phylogenetic HA groups, clade 1 (H1 and H5) and clade 2 (H3 and H7), as well as four different NA subtypes also chosen to represent the two major NA clades (2). Challenge viruses included homosubtypic viruses (viral strains expressing the same subtypes but are antigenically different from HA and NA in the vaccine viruses), as examples of protection against endemic influenza, and heterosubtypic viruses (viral strains with HA or NA subtypes different from those in the vaccine, as examples of potential zoonotic and pandemic IAVs). Immunized mice were broadly protected against mismatched lethal challenges with pathogenic 1918 pandemic H1N1, H7N9, highly pathogenic avian influenza (HPAI) H5N8, and avian H7N1 virus challenges. Complete protection was afforded after challenge with the H6N1 virus, expressing a heterosubtypic HA (H6) not contained in the vaccine; likewise, complete protection was observed after challenge with the doubly heterosubtypic H10N7 virus, expressing both HA and NA subtypes not contained in the vaccine. Protection from the doubly heterosubtypic H10N7 challenge was associated with the increase of granzyme B and perforin expressing cytotoxic T and natural killer (NK) cells in the lungs of infected mice but not with serum neutralizing antibodies, suggesting that broadly protective pulmonary cellular immune responses were elicited by the vaccine. Similarly, broad protection was observed in ferrets after antigenically mismatched homosubtypic (H1N1), partially heterosubtypic (H3N2), and completely heterosubtypic (H2N7 and H10N7) challenges.

RESULTS

Vaccination elicits antibody responses and confers protection against homosubtypic and heterosubtypic viral challenge in mice

The multivalent vaccine was prepared using BPL inactivation of avian IAV H1N9, H3N8, H5N1, and H7N3 subtypes (table S1), grown in Madin-Darby canine kidney (MDCK) cells, and purified by sucrose density gradient. Mice were primed on day 0 and boosted 28 days later by intramuscular (IM) or intranasal (IN) immunization with 6 μg of total protein (1.5 μg per subtype) (fig. S1). Both IN and IM immunization elicited serum immunoglobulin G (IgG) antibody responses as measured by enzyme-linked immunosorbent assay (ELISA) to homologous HAs (H1, H3, H5, and H7) and NAs (N1, N3, N8, and N9) and hemagglutination inhibition (HAI) antibodies (Fig. 1, A and B, and fig. S2A). Although IM immunization induced generally higher serum IgG and HAI antibody responses than IN immunization (Fig. 1, A and B), IN immunization induced a more pronounced IgA response in bronchoalveolar lavage (BAL) fluid (fig. S2, B and C). A statistically significant induction of IgA response was observed in IN-immunized mice compared to phosphate-buffered saline (PBS)–immunized mice (anti-H1, P = 0.04; anti-H3, P = 0.004; anti-H5, P = 0.01; anti-H7, P = 0.002; and anti-N3, P = 0.01); however, the concentration of BAL IgA detected was low regardless of the antigen specificity. In addition, antibodies to the conserved stalk (or stem) regions of both group 1 and 2 HAs were generated by IM or IN immunization (fig. S2, D and E).

Fig. 1. Multivalent IAV vaccination elicits antibody responses and confers protection against viral challenge in mice.

Serum IgG concentrations against (A) the four vaccine hemagglutinin (HA) antigens and (B) the four vaccine neuraminidase (NA) antigens were measured by ELISA and shown as area under the curve (AUC) 3 weeks after the boost immunization in PBS-, IM-, or IN-immunized mice. Welch ANOVA test and post hoc Dunnett’s T3 multiple comparison test were used to compare antibody concentrations between groups; *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. (C to F) Percent survival and percent weight loss in PBS-, IM-, or IN-immunized mice after lethal challenge (10× LD₅₀) with six different IAV challenge strains (n = 5 per group except H5N8 challenge): (C) 1918 pandemic H1N1, (D) H7N9, (E) HPAI H5N8 (n = 10 per group), (F) chimeric avian H7N1, (G) chimeric avian H6N1, and (H) chimeric H10N7 virus. Error bars represent SD.

To evaluate vaccine efficacy against viral challenge, cohorts of animals immunized with the multivalent vaccine were challenged with six IAV strains at 10 times the murine median lethal dose (LD₅₀) (Table 1). PBS-immunized animals steadily lost weight and showed 100% mortality after challenge with each of these viruses (Fig. 1, C to H). IM and IN immunization provided 100% protection against lethal 10× LD₅₀ challenge with the fully reconstructed 1918 pandemic H1N1 and zoonotic H7N9 viruses (Fig. 1, C and D), with little associated weight loss. Lethal 10× LD₅₀ challenge with an HPAI H5N8 virus (Fig. 1E) showed 90% protective efficacy for this antigenically variant, systemically replicating HPAI virus after IM immunization. Protective efficacy of IN immunization against this HPAI virus was less than that observed for IM immunization, with 70% survival after lethal H5N8 challenge.

Table 1.

Comparison of challenge virus HA and NA identity to vaccine virus components.

Mouse challenge IAV (10× LD50) and identity to vaccine viruses
Subtype (strain)	Percent nucleotide (and amino acid) identities
	HA	NA
H1N1 (A/Brevig Mission/1/1918)	79.1% (92.8%)	85.5% (91.5%)
H5N8 (A/chicken/Netherlands/EMC-3/2014)	79.7% (85.6%)	76.8% (84.8%)
H6N1 (recombinant Avian)*	51.0% (63.1%) [versus H1]†	92.1% (96.2%)
H7N1 (recombinant Avian)*	100.0% (100%)	92.1% (96.2)
H7N9 (A/Shanghai/1/2013)	76.9% (84.8%)	91.0% (95.5%)
H10N7 (recombinant Avian)*	48.5% (65.8%) [versus H7]†	53.9% (58.7%) [versus N9]†
Ferret challenge IAV (10xLD50) and identity to vaccine viruses
Subtype (strain)	Percent nucleotide (and amino acid) identities
	HA	NA
H1N1 (A/swine/Iowa/1931)	77.6% (89.9%)	85.5% (89.8%)
H2N7 (recombinant with 1957 pandemic HA)*	44.9% (42.5%) [versus H1]†	53.9% (58.7) [versus N9]†
H3N2 (A/Port Chalmers/1973)	83.8% (94.2%)	43.5% (50.4) [versus N3]†
H10N7 (recombinant Avian)*	48.5% (65.8%) [versus H7]†	53.9% (58.7) [versus N9]†

^*Recombinant avian IAV made as 2:6 recombinants with above HA and NA gene segments with remaining 6 gene segments from A/green winged teal/Ohio/175/1986 (H2N1) (24).

^†Heterosubtypic HA and NA challenge virus homologies were compared to phylogenetically closest vaccine sequence as noted.

To study the protective efficacy of the multivalent vaccine against lethal challenge with a homosubtypic, partially heterosubtypic (challenge with a virus with a different HA subtype), and completely heterosubtypic (challenge with a virus with both different HA and NA subtypes), immunized mice were challenged with chimeric avian H7N1, H6N1, and H10N7 viruses (23, 24), respectively. These viruses were chosen to represent different antigenic distances from the vaccine antigens (Table 1). The H7N1 virus HA matched the vaccine H7 HA, along with a minor mismatch in N1. In contrast, the H6N1 challenge used the same NA as in the H7N1 challenge but, in this case, with an HA subtype not contained in the vaccine. The H10N7 virus expressed no HA and NA subtypes contained in the vaccine. In all three challenge experiments, both IM- and IN-immunized mice showed 100% survival (Fig. 1, F to H).

IM- and IN-immunized mice lost very little weight, and all survived against a 10× LD₅₀ dose of H7N1 challenge, in contrast to PBS-immunized mice, which showed a rapid decline in body weight with 100% fatality occurring between days 6 and 8 after challenge (Fig. 1F). A second cohort of mice was challenged with a 10× LD₅₀ dose of the H6N1 virus. IM- and IN-immunized mice lost less weight and had 100% survival after lethal challenge, in contrast to PBS-immunized mice, which showed a rapid decline in body weight and 100% fatality by days 6 to 7 after challenge (Fig. 1G). IM-immunized mice showed more early weight loss than IN-immunized mice, with weight loss on days 1 to 4 similar to PBS-immunized groups, but with recovery and 100% survival. In a third cohort, mice were challenged with a 10× LD₅₀ dose of the H10N7 virus to examine protection against a doubly heterosubtypic viral infection (both different HA and NA subtypes). Both IM- and IN-immunized mice lost weight through day 4 but then rapidly recovered and had 100% survival after lethal challenge, in contrast to PBS-immunized mice, which showed a steady decline in body weight and 100% fatality by days 6 to 7 after challenge (Fig. 1H). Together, these results showed that the multivalent vaccine provided broad protection from a variety of lethal challenges with viruses of varying degrees of HA and NA antigenic distances in mice. In five of the six different lethal viral challenge experiments, both IM and IN mice showed 100% survival, with 0% survival of PBS-immunized animals. For HPAI H5N8 challenge, IM and IN immunization afforded 90 and 70% protection, respectively.

Vaccination results in distinct lung gene expression responses during mismatched and heterosubtypic challenge in mice

To determine the effects of immunization on pulmonary inflammatory and immune responses during viral infection, expression microarray analysis was performed on lung tissue collected from mice on day 6 after challenge with H7N1, H6N1, and H10N7 chimeric viruses. Viral replication in whole lung tissue measured by quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) did not detect matrix (M) gene viral RNA by 3 to 6 days after challenge with H7N1 and H6N1 viruses and showed nearly 90% reduction in viral RNA by 6 days after H10N7 challenge (Fig. 2A). Analysis of variance (ANOVA) identified significantly differentially expressed genes (greater than twofold difference in median expression, P < 0.05) between PBS-, IM-, and IN-immunized animals. Expression of genes associated with type I interferon (IFN) responses, lymphocyte activation, reactive oxygen species (ROS) responses and DNA damage, and programmed cell death were reduced in IM- and IN-immunized mice challenged with H7N1, H6N1, and H10N7 compared to PBS-immunized, challenged animals (Fig. 2A). Higher expression of these genes in PBS-immunized animals correlated with more viral replication. Concordant with the weight loss in IM- and IN-immunized animals observed after completely heterosubtypic H10N7 lethal challenge compared to either H7N1 and H6N1, the IM- and IN-immunized animals showed slightly stronger inflammatory response-related gene expression after H10N7 challenge (Fig. 2A). These animals also showed slightly higher degrees of viral replication.

Fig. 2. Immunized mice have reduced IAV RNA, inflammatory responses, and evidence of lung pathology.

(A) Top: bar graph showing relative expression of IAV M gene mRNA in immunized mouse lung compared to PBS-immunized animals as measured by qRT-PCR. Bottom, differences in lung gene expression in lungs of PBS-, IM-, and IN-immunized mice identified by ANOVA (greater than twofold difference in median expression, P < 0.01) on day 6 after challenge; heatmaps show the relative expression of type I IFN response genes, lymphocyte activation genes, ROS response and DNA damage genes, and programmed cell death genes. Genes with increased expression are shown in red, genes with no change as black, and genes showing decreased expression in blue. (B) Lung histopathology of PBS-, IM-, and IN-immunized mice lethally challenged with chimeric avian H6N1, H7N1, or H7N10 viruses (10× LD₅₀ dose), and analyzed at day 5 after infection. In each case, PBS-immunized animals showed widespread, severe viral pneumonia with necrotizing bronchitis and bronchiolitis as well as alveolitis. In contrast, IM- or IN-immunized animals showed an absence of pneumonia, with no bronchitis, bronchiolitis, or alveolitis. Aggregates of lymphoid tissue were observed in peribronchiolar and peribronchiolar spaces in immunized animals. Original magnifications are ×20; scale bar, 500 μM. See fig. S4 to S6 for additional pathological analyses.

IN-immunized animals showed more robust pulmonary gene expression responses compared to IM-immunized mice. Direct comparison (t test, twofold difference in median expression, P < 0.05) of IM- and IN-immunized animals revealed subtle but significant differences in gene expression (fig. S3A) that showed enrichment of pathways for neutrophil adhesion, IFN signaling, cytokine and chemokine signaling, dendritic cell maturation, and other pathways involved in innate response to viral infection (fig. S3B).

Lung pathology and immune cell infiltrates are reduced in vaccinated mice during mismatched and heterosubtypic IAV challenge

Histopathological examination was performed on mouse lung sections at day 5 after viral challenge for H7N1, H6N1, and H10N7 experiments (Fig. 2B and figs. S4 to S6). Lung sections of PBS-immunized mice challenged with H7N1, H6N1, and H10N7 viruses showed marked pathological changes involving over 50% of the lung parenchyma, including multifocal, moderate-to-severe, necrotizing bronchitis and bronchiolitis and moderate-to-severe alveolitis with pulmonary edema and fibrinous exudates [Fig. 2B and figs. S4A (first row, left), S5A (first row, left), and S6 (first row, left)]. Influenza viral antigen staining showed widespread positivity in respiratory epithelial cells, in alveolar epithelial cells, and in alveolar macrophages [figs. S4A (second row, left), S5A (second row, left), and S6 (second row, left)]. In contrast, lung sections from IM- and IN-immunized animals challenged with H7N1 showed minimal histopathological changes, an absence of alveolitis, and no viral antigen in alveolar epithelial cells or alveolar macrophages. The respiratory epithelium of bronchi and bronchioles was intact [fig. S4A (first row, middle and right)]. Similarly, vaccinated animals challenged with partially heterosubtypic H6N1 and completely heterosubtypic H10N7 viruses showed minimal histopathological changes [Fig. 2B and figs. S5A (first row, middle and right) S6A (first row, middle and right)], rapidly reproliferating respiratory epithelium in bronchi and bronchioles characterized by abundant mitotic figures, and an absence of alveolitis. No viral antigen [influenza nucleoprotein (NP)] was observed in alveolar epithelial cells or macrophages in vaccinated animals challenged with H7N1 [fig. S4A (second row, middle and right)], whereas viral antigen staining was observed in occasional alveolar macrophages in lungs of vaccinated mice challenged with H6N1 and H10N7 [figs. S5A (second row, middle and right) and S6 (second row, middle and right)].

After H7N1 challenge, sections from PBS-immunized mouse lungs showed occasional interstitial CD19⁺ B cells and CD3⁺ T cells and abundant Ly6G⁺ neutrophils throughout the lung parenchyma [fig. S4A (third to fifth rows, left)]. In contrast, IM- and IN-immunized mouse lungs showed increased CD19⁺ and CD3⁺ lymphocyte aggregates that were especially prominent in perivascular and peribronchiolar locations and a marked reduction in lung parenchymal Ly6G⁺ neutrophils (fig. S4 (third to fifth rows, middle and right)]. Large aggregates of CD19⁻ plasma cells were observed focally in the lungs of immunized mice (fig. S4B). Similar changes were observed in sections of mock-, IM-, and IN-immunized mouse lungs after H6N1 challenge (fig. S5) and H10N7 challenge (fig. S6). These results demonstrated that both IM and IN immunization in mice resulted in protection from lethal challenge with a partially heterosubtypic lethal H6N1 and completely heterosubtypic H10N7 viral challenge, associated with reductions in viral titer, pathologic changes, and host immune and inflammatory responses in lung, as well as a marked increase in B and T cell aggregates in the lungs of immunized mice.

Protection against IAV challenge is conferred by passive serum transfer in mice

Passive serum transfer experiments in mice, in which serum from immunized animals was injected intraperitoneally 1 day before challenge in naive recipient mice with homosubtypic H7N1, provided 100% protection with serum from IM-immunized, but not from IN-immunized animals (fig. S7A), consistent with the lower concentrations of serum antibody observed in IN-immunized animals (Fig. 1, A and B). Serum transfer experiments followed by H6N1 challenge, in which the NA subtype, but not the HA subtype, was contained in the multivalent vaccine produced analogous results to homosubtypic H7N1 challenge, in that serum from IM-immunized mice protected H6N1-challenged naive recipient mice, whereas serum from IN-immunized mice did not (fig. S7B). In this case, protection was likely afforded by anti-NA antibodies in IM-immunized serum or anti–group 1 HA stalk antibody. In contrast, passive serum transfer from either IM- or IN-immunized mice provided no protection against doubly heterosubtypic H10N7 challenge in naive recipient mice (fig. S7C). No neutralizing antibody activity was detected against the H10N7 virus in serum from vaccinated mice (fig. S8A).

Cytotoxic T and NK cells respond during heterosubtypic challenge in mice

Flow cytometry was performed on immune cells isolated from lungs 5 days after the lethal doubly heterosubtypic H10N7 viral challenge in mice using an exploratory antibody panel focusing on cytotoxic lymphocytes (table S2). The frequencies of lung cytotoxic CD8 T cells were significantly increased in both IM- and IN-immunized mice compared to the PBS-immunized mice in terms of granzyme B (P < 0.001 for IM and P = 0.02 for IN; Fig. 3A) and perforin (P = 0.001 for IM and IN; Fig. 3B) production. Although at lower frequencies, cytotoxic CD4 T cells were significantly induced in IN-immunized mice compared to IM- and PBS-immunized mice as shown in Fig. 3C (P = 0.002 versus mock and P = 0.004 versus IM) and Fig. 3D (P = 0.001 versus mock and P = 0.002 versus IM). IN-immunized mice also showed the highest frequency of cytotoxic NK cells (Fig. 3, E and F). Although the frequency of cytotoxic NK cells was not increased in IM-immunized mice, the overall expression of granzyme B and perforin was greater in cytotoxic NK cells in IM-immunized mice compared to the PBS-immunized animals (Fig. 3, G and H). The increased abundance of cytotoxic molecules in NK cells from IM-immunized mice could be linked to the higher virus-specific serum IgG concentrations in IM-immunized mice (fig. S8B), which could induce antibody-dependent cell-mediated cytotoxicity. Although cytotoxic T and NK cells in the lung were induced in both IM- and IN-immunized mice upon challenge, the frequencies of lung cytotoxic T and NK cells were low in all three groups before challenge (fig. S9), suggesting a rapid induction and expansion of cytotoxic lymphocytes in immunized animals after viral challenge. Combined with the lack of heterosubtypic protection after passive serum transfer (fig. S7C), these data suggest that the heterosubtypic protection observed in mice (Figs. 1H and 2 and fig. S6) is more likely mediated by cytotoxic cellular immune responses rather than by serum antibodies. In addition, IM-immunized mice showed the highest frequency of CD8 and CD4 lung resident memory T (T_RM) cells (Fig. 3, I and J). IM-immunized mice also showed a higher frequency of mature NK cells compared to the PBS-immunized mice (fig. S10A). γδ T cells were the most frequent in the PBS-immunized mice (fig. S10B). There were no differences in the frequency of regulatory T cells between groups (fig. S10C).

Fig. 3. Cytotoxic T cell and NK cell responses are elicited during heterosubtypic challenge in mice.

Flow cytometry was performed on immune cells isolated from lungs 5 days after a lethal heterosubtypic H10N7 challenge. Among lung leukocytes (intravenous CD45 staining negative), CD8 T cells, CD4 T cells, and NK cells were analyzed for two representative cytotoxic molecules: granzyme B and perforin. Fluorescence minus one (FMO) controls were used to gate granzyme B or perforin positive populations. The frequencies of (A) granzyme B– or (B) perforin-positive CD8 T cells relative to the total CD8 T cells are shown. BUV, brilliant ultraviolet; AF, Alexa Fluor; PE, phycoerythrin. The frequencies of (C) granzyme B– or (D) perforin-positive CD4 T cells relative to the total CD4 T cells are shown. Similarly, the frequencies of (E) granzyme B– or (F) perforin-positive NK cells relative to the total NK cells are shown. APC, allophycocyanin. The differences in the abundance of (G) granzyme B or (H) perforin in NK cells between groups were measured by comparing sum of fluorescent intensities (sum FI). The frequencies of (I) CD8 resident memory T (T_RM) cells and (J) CD4 T_RM relative to the total CD8 and CD4 T cells, respectively, are shown. Animal experiments were done twice independently; red and blue symbols indicate each independent experiment (four to five mice per group). Each symbol represents an individual animal. Contour plots are representative of the frequency of granzyme B– or perforin-positive CD8 T cells, CD4 T cells, and NK cells. Gating strategies to identify lung CD8, CD4, and NK cells are shown in fig. S20. Welch ANOVA test and post hoc Dunnett’s T3 multiple comparison test were used to compare between groups; *P < 0.05, **P < 0.01, and ***P < 0.001.

Vaccination elicits antibody responses and confers protection against mismatched and heterosubtypic viral challenge in ferrets

Ferrets were primed and boosted 28 days later by IM or IN immunization with 400 μg of total protein (100 μg per subtype). IN immunization was performed without adjuvant, and IM immunization was performed using a squalene-based adjuvant (fig. S1B). PBS-immunized control animals were intranasally inoculated with PBS or intramuscularly inoculated with PBS including adjuvant. The sequence similarities between the vaccine strains and the challenge strains are summarized in Table 1. The vaccine elicited serum IgG antibody responses to homologous HAs (H1, H3, H5, and H7) and NAs (N1, N3, N8, and N9) (Fig. 4, A and B). Vaccination also elicited serum group 1 and 2 HA stalk antibodies and HAI antibodies in ferrets (fig. S11, A and B). Similar to what was observed in mice, IM immunization induced generally higher serum IgG antibody responses than IN immunization in ferrets (Fig. 4, A and B).

Fig. 4. Quadrivalent vaccination is immunogenic and confers protection against viral challenge in immunized ferrets.

Serum IgG concentrations against (A) the four vaccine HA antigens and (B) the four vaccine NA antigens were measured 3 weeks after the boost immunization in PBS-, IM-, or IN-immunized mice using ELISA. Welch ANOVA test and post hoc Dunnett’s T3 multiple comparison test were used to compare antibody concentrations between groups. *P < 0.05, **P < 0.01, ***P < 0.001. (C to F) Reductions in nasal wash and lung viral titers were measured in IM- or IN-immunized ferrets as compared to PBS-immunized ferrets (n = 6 per group). Nasal wash titers were measured at days 1, 3, 5, and 7 after challenge, and lung titers were determined on day 5 after challenge. (C) Reductions in titers after A/swine/1931 (H1N1) challenge. (D) Reductions in titers after A/Port Chalmers/1/1973 (H3N2) challenge. (E) Reductions in titers after chimeric 1957 pandemic H2N7 challenge. (F) Reductions in titers after chimeric avian H10N7 challenge. For (C) and (D), two of six ferrets per group were used for lung viral titration. For (E) and (F), four of six ferrets per group were used for lung viral titration. Unpaired t test (C and D; n = 2) and ordinary ANOVA test and post hoc Dunnett’s multiple comparison test (E and F; n = 4) were used to compare viral titers in IM- or IN-immunized ferrets to PBS-immunized ferrets at the indicated time points. Error bars represent geometric SD. Horizontal dashed lines indicate the limit of detection.

The protective efficacy against mismatched, homosubtypic A/H1N1 viral challenge was next evaluated using A/swine/1931 H1N1 virus (Table 1). HAI assay results demonstrated a statistically significant (P = 0.0119), about fourfold difference in HAI titers between the vaccine H1 HA and the challenge H1 viruses (fig. S12A), demonstrating that the challenge virus has substantial antigenic difference compared to vaccine antigens. The N1 NA shared only 85.5% nucleotide identity with the N1 NA in the vaccine. Viral titers in ferret nasal wash in IM- and IN-immunized ferrets were reduced to near undetectable concentrations by day 5 after challenge as compared to PBS-immunized ferrets (Fig. 4C), and lung titers collected from IM- and IN-immunized ferrets at day 5 after challenge showed no detectable viral replication.

Similarly, IM-, IN-, or PBS-immunized ferrets were challenged with a partially heterosubtypic human seasonal IAV, A/Port Chalmers/1973 (H3N2) (Table 1). In this case, the H3 was antigenically mismatched to the avian H3 HA in the vaccine, as supported by cross-HAI evaluation (fig. S12B). Sequence identity to the vaccine H3 HA was 83.8%. Furthermore, the challenge virus expressed an N2 subtype NA not contained in the vaccine. The closest vaccine NA sequence shared only 43.5% identity with the challenge virus N2 subtype. Post-challenge viral titers in nasal wash in IM- and IN-immunized ferrets were reduced to near undetectable concentrations by day 5 after challenge (Fig. 4D), and lung titers collected at day 5 after challenge showed no detectable viral replication in the IN-immunized group.

To measure vaccine protective efficacy in ferrets against doubly heterosubtypic IAV challenge, where the challenge viruses have HA and NA subtypes different from the vaccine viruses, two cohorts of immunized and PBS-immunized ferrets were challenged with either A/H2N7 or A/H10N7. Sequence identity of the challenge HA and NA subtypes as compared to the vaccine components ranged from 44.9 to 53.9% (Table 1). Viral titers in nasal wash in IM- and IN-immunized A/H2N7-challenged ferrets were significantly reduced by day 5 after challenge (P = 0.005 for IM and P < 0.001 for IN; Fig. 4E), and lung titers collected at day 5 after challenge showed no detectable viral replication in both the IM- and IN-immunized groups. Similarly, challenge viral titers in ferret nasal wash in IM- and IN-immunized A/H10N7-challenged ferrets were significantly reduced by day 5 after challenge (P = 0.005 for IM and P = 0.004 for IN; Fig. 4F). Lung titers collected at day 5 after challenge showed no detectable viral replication in IN-immunized ferrets, whereas IM-immunized ferrets had viral titers that were not reduced compared to PBS-immunized ferrets. The lack of detectable viral replication in the lungs of IN-immunized ferrets could be associated with the earlier recovery from the initial weight loss in IN-immunized ferrets compared to IM-immunized ferrets (fig. S13A). Other challenge studies with A/swine/1931 (H1N1), A/Port Chalmers/1/1973 (H3N2), and 1957 chimeric H2N7 viruses did not show noticeable differences in weight loss between groups (fig. S13, B to D).

Host gene expression and viral RNA differ during mismatched and heterosubtypic challenge in vaccinated ferrets

To characterize the host gene expression response to viral challenge in immunized ferrets, RNA expression microarray analysis and IAV M gene qRT-PCR were performed on lung tissue collected on day 5 after challenge. Viral M gene RNA in lung tissue measured by qRT-PCR showed a greater than 97% reduction in viral RNA at 5 days after challenge with H1N1, H3N2, H2N7, and H10N7 viruses (Fig. 5A). ANOVA was performed to identify differentially expressed genes (greater than twofold difference in median expression, P < 0.05) between PBS-, IM-, and IN-immunized animals. Because of the small number of animals, the analysis was performed with all four viruses in each group (PBS-, IM-, and IN-immunized). These sequences were enriched for pathways associated with the innate antiviral response, including IFN signaling, cytokine signaling, lymphocyte activation, oxidative damage, and DNA repair responses (Fig, 5A), which were highly expressed in control animals but lower in the IM- and IN-immunized animals (Fig. 5A), consistent with little viral replication in the lungs of these animals. No differences in lung gene expression responses were identified between IM- and IN-immunized, challenged ferrets.

Fig. 5. Reduced influenza viral RNA, lung inflammatory responses, and lung pathology were observed in vaccinated ferrets.

(A) Top, bar graph showing relative expression of IAV M gene mRNA in immunized ferret lung compared to PBS-immunized animals as measured by qRT-PCR. Bottom, differences in lung gene expression responses in lungs of PBS-, IM-, and IN-immunized ferrets identified by ANOVA (greater than twofold difference in median expression, P < 0.01) on day 5 after challenge; the heatmaps show the relative expression of type I IFN response genes, cytokine signaling genes, lymphocyte activation genes, and DNA repair genes. (B) Lung histopathology is shown for PBS-, IM-, and IN-immunized mice lethally challenged with A/swine/1931 (H1N1), A/Port Chalmers/1/1973 (H3N2), or completely heterosubtypic challenge with chimeric H2N7 or H10N7 challenge viruses. In each case, PBS-immunized animals showed widespread, severe viral pneumonia with necrotizing bronchitis and bronchiolitis as well as alveolitis. In contrast, IM- or IN-immunized animals showed an absence of pneumonia, with no bronchitis, bronchiolitis, or alveolitis. Aggregates of lymphoid tissue were observed in peribronchiolar and peribronchiolar spaces in immunized animals. Original magnifications are ×20; scale bar, 500 μM. See fig. S14 to S17 for additional pathological analyses.

Lung pathology was reduced in vaccinated ferrets during mismatched and heterosubtypic viral challenge

Histopathological analysis was performed on ferret lung sections at day 5 after viral challenge (Fig. 5B and figs. S14 to S17). Lung sections of PBS-immunized ferrets challenged with the mismatched A/swine/Iowa/1931 (H1N1) virus showed marked pathological changes involving over 50% of the lung parenchyma, including multifocal, moderate-to-severe, necrotizing bronchitis and bronchiolitis (Fig. 5B and fig. S14, A to C), along with moderate-to-severe alveolitis with a neutrophil-predominant, mixed inflammatory cell infiltrates, pulmonary edema, and fibrinous exudates, a pathology remarkably similar to that seen with ferret infection with the 1918 pandemic H1N1 virus (25). Influenza viral antigen staining showed widespread positivity in respiratory epithelial cells, in alveolar epithelial cells, and in alveolar macrophages ([fig. S14D). In contrast, lung sections from IM- and IN-immunized animals showed minimal histopathological changes, including mild, focal bronchiolitis, an absence of alveolitis, and no viral antigen in alveolar epithelial cells with only occasional staining in alveolar macrophages (fig. S14, E to L).

Histopathological analysis was performed on ferret lung sections with the partially heterosubtypic A/Port Chalmers/1973 (A/H3N2) virus (expressing an antigenically variant HA and a different NA as compared to the vaccine viruses), as well as doubly heterosubtypic viral challenges with A/H2N7 and A/H10N7 viruses. In each of these cases, PBS-immunized ferrets showed marked pathological changes involving over 50% of the lung parenchyma, including multifocal, moderate-to-severe, necrotizing bronchitis and bronchiolitis, along with moderate-to-severe alveolitis with a mixed inflammatory cell infiltrate, and focal pulmonary edema and fibrinous exudates (Fig. 5B and fig. S15 to S17). In PBS-immunized animals, influenza viral antigen staining showed widespread positivity in respiratory epithelial cells, alveolar epithelial cells, and alveolar macrophages (figs. S15D to S17D). In contrast, lung sections from IM- and IN-immunized animals showed minimal histopathological changes (fig. 5B), including mild, focal bronchiolitis, an absence of alveolitis, and no viral antigen in alveolar epithelial cells with only occasional staining in alveolar macrophages (figs. S15 to S17, E to L). Some viral antigen was detected in bronchiolar respiratory epithelial cells in the absence of inflammation or histopathologic changes in H10N7 intramuscularly challenged animals, which may correspond to the detectable viral titers in some of those animals (Fig. 4F).

Low-dose immunization confers protection against heterosubtypic IAV challenge

Immunization with lower doses ($\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$4$}\right.$ of the antigen) of the same vaccine provided 100% protection against 10× LD₅₀ lethal challenge with completely heterosubtypic H10N7 and partially heterosubtypic H6N1 viruses in mice (fig. S18, A and B). Similarly, immunization with lower doses ($\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$4$}\right.$ antigen) provided protection against partially heterosubtypic H3N2 and mismatched H1N1 challenge in ferrets (fig. S18, C and D). These results suggest that a lower dose of the multivalent vaccine could still provide a high degree of protection.

Good manufacturing practice, production, toxicology, and immunogenicity studies

The systemic toxicity, local tolerance, and immunogenicity of the BPL-inactivated influenza vaccine were evaluated in New Zealand white rabbits. No mortality was observed after administration of the vaccine IN or IM. There were no clinical observations, injection or instillation site observations, changes in body weights, changes in food consumption, changes in body temperatures, systemic toxicity, local tolerance, or ocular effects attributed to administration of the vaccine by either route. Rabbits that received the vaccine intranasally or intramuscularly mounted serum antibody responses against the vaccine HA and NA antigens on day 45 (fig. S19). Next-generation sequencing of the four viral stocks used for good manufacturing practice (GMP) production showed no evidence of mixed influenza virus infection or presence of adventitial avian viruses (GenBank SRA accession number PRJNA847201).

DISCUSSION

Both the rapid evolution of IAV and its extensive animal reservoir, which allows for zoonotic infections that can lead to pandemic development, necessitate the need for new vaccine strategies to protect at least against all IAV or, ideally, against all influenza A and B viruses (26). Because IAVs are very diverse genetically and antigenically, there are only a few highly conserved epitopes shared among them. Therefore, most of the strategies to develop a broadly protective influenza vaccine have sought to induce protective immunity to these highly conserved universal epitopes (19). In theory, these vaccine-induced immune responses could at least minimize severe clinical disease even in the case of an antigenically variant influenza strain. A limited number of highly conserved regions found in various IAV have been identified, and strategies for producing broadly protective influenza vaccines have gained new traction in the past decade, including vaccines inducing antibodies against conserved epitopes in HA (especially the HA stalk) (27), NA (28), the extracellular domain of M2 (M2e) (29), or other viral proteins and vaccines that promote enhanced protective T cell responses (30). Although vaccines targeting the HA stalk, NA, and M2e have been generally designed to induce broadly reactive antibodies that would confer protection, vaccine strategies inducing T cell responses could have advantages over antibody-based vaccines. This is because T cells recognize linear epitopes presented by host major histocompatibility complex molecules and many conserved peptides from influenza proteins could be targeted. In contrast to the more focused vaccine approaches described above, the major advantage of the BPL-inactivated IAV vaccine approach described here is that the whole-virus vaccine induced broad immune responses, including antibody responses to HA and NA, as well as T cell responses, making vaccine escape less likely.

Low-pathogenicity, BPL-inactivated, whole avian IAVs were combined in this study to develop a broadly protective vaccine. An advantage of this approach is the incorporation of multiple, structurally intact full-length HA and NA proteins that represent a broad consensus of the cladal HA and NA distribution of IAV. The avian IAV HA proteins included in the vaccine also have low degrees of glycosylation on the HA head as compared to human viruses, which could allow for the development of antibody responses against protein antigens that might be masked in seasonal IAVs. Moreover, the inclusion of internal viral proteins in the vaccine is likely to provide diverse T cell epitopes and generate broadly protective T cell immunity against antigenically distinct emerging strains (30), as indicated in this study using flow cytometry. Compared to the vaccine strategies using purified antigens, the multivalent strategy deployed in this study is expected to generate diverse antibody and cellular responses by providing various B and T cell epitopes rather than narrowly targeting conserved regions of a single viral protein. It will be important to investigate whether the diverse immune responses generated by the multivalent strategy could effectively suppress the emergence and selection of escape mutation(s) that were suggested to be associated with strategies narrowly targeting conserved regions of influenza viruses (31).

Moreover, the approach taken in this study to generate the vaccine uses procedures that are currently used for producing seasonal influenza vaccines. This is a substantial advantage because manufacturing such a vaccine would be relatively simple and inexpensive. In addition, the vaccine, delivered either intramuscularly or intranasally, is likely to be safe for humans because no toxicity was observed in mice or ferrets or in a rabbit toxicity study conducted as part of GMP manufacturing.

In both IM- and IN-immunized mice and ferrets, peribronchiolar and perivascular lymphoid aggregates were observed. In mice, it was possible to perform immunostaining to show that these aggregates contained both CD19⁺ B and CD3⁺ T lymphocytes, suggesting their potential role in providing protection. The fact that IN-immunized mice and ferrets demonstrated potent protective efficacy even against doubly heterosubtypic challenge despite lower concentrations of serum neutralizing antibody responses suggests that mucosal immunity plays an important role in anti-influenza virus responses. In mice, both IM and IN immunization induced robust cytotoxic CD8 T cell responses in the lungs after the fully heterosubtypic H10N7 challenge. Increases in cytotoxic NK cells were also positively associated with heterosubtypic protection regardless of the immunization routes. Antibody-dependent cell-mediated cytotoxicity could explain the increase of granzyme B and perforin in NK cells in IM-immunized mice, which had higher concentrations of serum antibodies. Although not the focus of this study, it will be important to further investigate the cellular immune responses correlated to the heterosubtypic protection observed. The analysis of pre- and post-challenge cytotoxic T and NK cells indicated a rapid induction of cytotoxic activities in the lungs of immunized animals after the viral challenge. However, it is unknown whether these cytotoxic lymphocytes are lung resident or rapidly recruited to the lung upon viral challenge. In subsequent studies, isolation and characterization of subpopulations of T and NK cells from different tissues, such as the lung, lymph node, and peripheral blood both before and after challenge will elucidate the detailed cellular mechanisms of protection provided by this multivalent vaccine approach.

As another metric of protective efficacy, both IM and IN immunization was associated with a reduction in lung pathology and inflammatory gene expression responses in both mice and ferrets. The reduction in lung neutrophils during lethal challenge resulting from vaccination is likely critical to protection against lethal infection because neutrophils and the ROS that they produce and excrete are major contributors to severe pulmonary pathology during fatal IAV infections (24, 32–34). Also concordant with reductions in viral replication and lung neutrophils, expression microarray analysis showed reductions in expression of genes in type I IFN, ROS damage, and cell death pathways compared to PBS-immunized and challenged mice and ferrets. Unexpectedly, the route of immunization affected pulmonary gene expression responses and T and NK cell profiles in mice. Although both IM and IN immunization were protective against lethal challenge in mice, lower inflammatory gene expression response pathways correlated with IM immunization as compared to IN immunization. IM immunization induced stronger cytotoxic T cell responses but weaker cytotoxic NK cell responses as compared to IN immunization. This immunization route–dependent effect suggests that mucosal or systemic vaccination affects aspects of the pulmonary immune responses, which could arise from differences in effector functions or epithelial transport of IgA and IgG antibodies (35) or may be due to different pools of adaptive immune cells responding to each route of vaccination.

There are some limitations of the animal studies described here. Vaccinations were performed on influenza virus–naïve mice and ferrets, which does not reflect the more complex exposure histories of humans to prior influenza viruses and vaccinations. Specifically, immunological imprinting has been reported to be an important determinant in the observed low vaccine efficacy in humans, making predictions from naive animal models difficult (36). In addition, T cell responses likely differ between naive mice and humans with complex preexisting cellular IAV immunity (37). To overcome these complexities, clinical studies of vaccine efficacy will be needed to evaluate the breadth of vaccine efficacy in humans in comparison to efficacy afforded by vaccination with the commercial annual influenza vaccines. Similarly, evaluation of breadth of protective efficacy in humans could be initially studied in a challenge model (38) using antigenically variant human H1N1 and H3N2 viruses (39, 40). The durability of responses will need to be studied both in animal model and, subsequently, in humans. Although preliminary vaccine dose studies were performed here, additional dose reduction studies are warranted in both animals and subsequently in humans.

The BPL-inactivated whole-virus vaccine described here shares desired characteristics of a broadly protective influenza A vaccine. The vaccine includes multiple viral proteins such as HA, NA, M2e, and NP, which elicit broadly protective B and T cell responses. Initial phase 1 safety and immunogenicity studies of the vaccine in humans after IN and IM immunization will begin in June 2022 at the National Institutes of Health (NIH) Clinical Center (under Food and Drug Administration-approved IND 27691). This vaccine approach could have broad application as both a prepandemic and “superseasonal” IAV vaccine with low reactogenicity that would be inexpensive, easily distributed worldwide, and could be of value in human influenza virus control.

Study design

This study was designed for preclinical evaluation of a broadly protective IAV vaccine in mouse and ferret models. The protective efficacy of the vaccine candidate was evaluated by comparing survival, clinical signs, viral load, histopathology, and gene expression profiling between immunized and control animals upon various heterologous viral challenges. Randomization of animals was achieved by randomly assigning animals to cages upon arrival and predefined sequential assignment of study groups based on the cage numbers. Blinding was achieved by assigning alphanumeric codes to individual animals and samples during data collection and unblinding upon the completion of the experiment. No data were excluded from reporting. Sample size calculation (power analysis) was not performed.

Quadrivalent vaccine design and construction

Low-pathogenicity avian viruses A/mallard/Ohio/265/1987 (H1N9), A/pintail/Ohio/339/1987 (H3N8), A/mallard/Maryland/802/2007 (H5N1), and A/Environment/Maryland/261/2006 (H7N3) were grown in MDCK cells (CCL-34, American Type Culture Collection) followed by inactivation using BPL (P5648, MilliporeSigma). Viral culture supernatant was buffered with Hepes (15630, Thermo Fisher Scientific) at a final concentration of 0.1 M, and BPL was added (0.1% final). After overnight incubation at 4°C for the viral inactivation, BPL was hydrolyzed at 37°C for 90 min. Inactivated viruses were concentrated by ultracentrifugation at 50,000g for 2 hours and purified using a 20 to 60% (w/v) discontinuous sucrose density gradient purification (100,000g for 2 hours). A band at the 20 to 60% sucrose interface containing purified viruses was collected, and the sucrose was removed by pelleting viruses (50,000g for 2 hours) followed by virus resuspension in PBS. The total protein amount of the purified viruses was quantified using a bicinchoninic acid (BCA) protein assay kit (23225, Thermo Fisher Scientific). For IN immunization in mice, one dose of vaccine was formulated to contain 1.5 μg of each antigen (6 μg total) in 50 μl of PBS delivered in both nostrils. For IM immunization in mice, one dose of vaccine was formulated to contain 1.5 μg of each antigen (6 μg total) in 25 μl of PBS and supplemented with 25 μl of adjuvant AddaVax (vac-adx-10, InvivoGen). For the ferret study, 100 μg of each antigen (400 μg total) in 1 μl of PBS was used for one dose of IN immunization (500 μl per nostril). For IM immunization in ferrets, 100 μg of each antigen (400 μg total) in 250 μl of PBS was supplemented with 250 μl of AddaVax. Low-dose vaccines were prepared as described above but with $\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$4$}\right.$ antigen (total of 1.5 μg for mice and total of 100 μg for ferrets).

Challenge viruses

Fully reconstructed 1918 pandemic H1N1 virus was generated using a 12-plasmid reverse genetics system as previously described (41, 42) and passaged in MDCK cells. Avian influenza viruses (H6N1, H7N1, and H10N7) and recombinant H2N7 virus containing the 1957 H2 pandemic HA were generated as previously described (24, 41). A/swine/Iowa/1931 (H1N1), A/Port Chalmers/1/1973 (H3N2), A/chicken/Netherlands/EMC-3/2014 (H5N8), and A/Shanghai/1/2013 (H7N9) viruses were passaged in embryonated specific pathogen-free chicken eggs (catalog no. 10100329, Charles River). Table 1 summarizes the challenge viruses used in this study and the HA and NA identity to vaccine virus components. All viruses and infectious samples were handled under enhanced biosafety level 3 (BSL-3) laboratory conditions except for A/Swine/Iowa/1931 (H1N1) and A/Port Chalmers/1/1973 (H3N2), which were handled in BSL-2 laboratory conditions. Experiments with the fully reconstructed 1918 pandemic virus and the HPAI H5N8 virus were conducted in accordance with the select agent guidelines of the NIH, the Centers for Disease Control and Prevention, and the U.S Department of Agriculture, under the supervision of the NIH Select Agent and Biosurety Programs and the NIH Department of Health and Safety.

Mouse studies

Seven- to 8-week-old female BALB/C mice (the Jackson Laboratories) were lightly anesthetized with isoflurane supplemented with O₂ (1.5 liter/min) before immunization or virus challenge. Mice were IN- or IM-immunized twice 4 weeks apart with the quadrivalent vaccines prepared as described above. PBS-immunized control mice received PBS or adjuvant without antigen. Serum samples were collected 3 weeks after the boost immunization to measure antibody responses elicited by the immunization. Lethal challenge infections (10× LD₅₀ dose in 50 μl of inoculum per animal) were performed 4 weeks after the boost immunization. Post-challenge body weight and survival were monitored for 14 days from five mice per experimental condition. Ten mice, instead of five, were used for the HPAI H5N8 challenge study. Mice were humanely euthanized if more than 25% of initial body weight was lost. For measuring viral loads and transcriptomics from H7N1, H6N1, and H10N7 challenge groups, lungs were harvested at day 6 (n = 4) after infection and immediately frozen in dry ice and stored at −80°C until processed. For histopathology, lungs were harvested at day 5 after infection (n = 2) followed by inflation and fixation using 10% neutral buffered formalin (NBF).

Flow cytometry

Mice were immunized twice followed by 10× LD₅₀ H10N7 challenge as described above (four to five mice per group: PBS-, IM-, and IN-immunized). Mice were humanely euthanized at day 5 after infection or before challenge (4 weeks after second immunization), and lungs were collected in MACS Tissue Storage Solution (130-100-008, Miltenyi Biotec) in ice. To discriminate circulating (intravascular) fraction of lymphocytes from tissue (extravascular) lymphocytes, anesthetized mice were intravenously administered 2 μg of anti-CD45 antibody (table S2) in 100 μl of saline through retro-orbital route 5 min before euthanasia. Cells stained with the IV-CD45 antibody were excluded from analysis (see fig. S20 for gating strategy). Single cells were prepared using Lung Dissociation Kit (130-095-927, Miltenyi Biotec) and gentleMACS Dissociator (130-096-427, Miltenyi Biotec) following the manufacturer’s instruction and counted using Countess 3 (AMQAF2000, Thermo Fisher Scientific). Four million cells were blocked with 4 μl of Mouse BD Fc Block (553141, BD Biosciences) and simultaneously stained with 2 μl of LIVE/DEAD Fixable dye (L23105, Thermo Fisher Scientific) in 100 μl of PBS supplemented with 2% fetal bovine serum for 45 min at 4°C. Cells were then stained for 15 surface markers (table S2) in 50 μl of Stain Buffer (554657, BD Biosciences) for 45 min at 4°C, followed by formaldehyde fixation and permeabilization using the Transcription Factor Buffer Set (562574, BD Biosciences) following the manufacturer’s instruction. Fixed/permeabilized cells were stained for FoxP3, perforin, granzyme B, and CD3 (table S2) for 45 min at 4°C. Flow cytometry was performed on a FACSymphony A5 Cell Analyzer (BD Biosciences). Data were analyzed with FlowJo v10.8.0 analysis software (BD Biosciences) after the removal of low-quality events using flowAI (43). Fluorescence compensation was performed using single-stained controls prepared with UltraComp eBeads Plus (01-3333-42, Thermo Fisher Scientific). All antibody titers have been established in digested lungs prepared as described above. Gating strategies are provided in fig. S20 and were based on the use of fluorescence minus one controls. To compare the abundance of granzyme B perforin in NK cells between groups, 2000 NK cells were randomly selected from each sample, granzyme B or perforin positive events were gated (fig. S20), and the fluorescent intensities (FI) of the gated events were summed (sum FI). Sum FI values were used to compare the granzyme B or perforin abundance in NK cells between groups. Mouse studies for flow cytometry were performed twice independently.

Ferret studies

Five- to 7-month-old female ferrets (Triple F Farms) were lightly anesthetized with isoflurane supplemented with O₂ (1.5 liter/min) before immunization or virus challenge. Ferrets were IN- or IM-immunized twice 4 weeks apart with the quadrivalent vaccines prepared as described above (500 μl per nostril). PBS-immunized control ferrets received PBS (IN) or PBS with adjuvant (IM). Serum samples were collected 3 weeks after the boost immunization to measure antibody responses elicited by the immunization. Challenge infections (1 ml of inoculum per animal) were performed 4 weeks after the boost immunization. For A/swine/1931 (H1N1) and A/Port Chalmers/1973 (H3N2) challenge, 1 × 10⁷ plaque-forming units (PFU) of each virus was used. For chimeric H2N7 and H10N7 challenge, 2 × 10⁵ PFU of each virus was used. To measure viral shedding in the upper respiratory tract after infection, nasal wash samples were collected at days 1, 3, 5, and 7 after infection, using 1 ml of PBS. To measure viral shedding, transcriptomics, and histopathology in the lower respiratory tract, ferrets were euthanized at day 5 after challenge, and lungs were harvested. Left cranial lobes of the harvested lungs were immediately frozen in dry ice and stored at −80°C until they are processed for measuring viral loads and transcriptomics. For histopathology, the remaining lungs were inflated and fixed using 10% NBF. All experimental animal work was performed in accordance with U.S. Public Health Service Policy on Humane Care and Use of Laboratory Animals in an animal BSL-2 (ABSL-2) laboratory or an enhanced ABSL-3 laboratory (for the viral challenge) as necessary at the National Institute of Allergy and Infectious Diseases (NIAID) of the NIH following approval of animal safety protocols by the NIAID Animal Care and Use Committee.

RNA isolation and expression microarray analysis

Frozen lungs, collected as described above, were lightly defrosted, homogenized in TRIzol (15596018, Thermo Fisher Scientific), and total RNA was isolated following the manufacturer’s protocol. Isolated total RNA was purified using RNeasy Mini Kit (74106, QIAGEN). Gene expression profiling experiments were performed using Agilent Mouse Whole Genome 44K microarrays (G4122F, Agilent). Ferret expression microarray analysis was performed using custom microarrays from Agilent Technologies (44). Fluorescent probes were prepared using Agilent Quick Amp Labeling Kit (5190-2305, Agilent) according to the manufacturer’s instructions. Each RNA sample was labeled and hybridized to individual arrays. Spot quantitation was performed using Agilent’s Feature Extractor software, and all data were uploaded into Genedata Analyst 9.0 (Genedata). Data normalization was performed in Genedata Analyst 9.0 using central tendency followed by relative normalization using pooled RNA from PBS-immunized mouse lung (n = 4) or ferret lung (n = 3) as a reference. Transcripts showing differential expression (twofold, P < 0.01) between infected and control animals were identified by standard t test. The Benjamini-Hochberg procedure was used to correct for false positive rate in multiple comparisons. Panther and Ingenuity Pathway Analysis was used for gene ontology and pathway classification. The complete Minimum Information About a Microarray Experiment (MIAME)–compliant microarray dataset has been deposited in National Center for Biotechnology Information’s (NCBI) Gene Expression Omnibus (GEO) (www.ncbi.nlm.nih.gov/geo) and is accessible through GEO Series accession number GSE180226.

Viral load determination in mouse and ferret lungs and in ferret nasal wash

Influenza viral titers in mouse and ferret lungs were quantified using qRT-PCR from the RNA samples prepared as described above. Reverse transcription of total RNA was performed using the SuperScript III First-Strand cDNA synthesis kit (18080051, Thermo Fisher Scientific) primed with an equal mix of oligo(dT) and the Uni12 influenza A specific primer: 5′ AGCRAAAGCAGG 3′. The IAV matrix (M) gene amplicon was quantified using the following primers and probe sequences: forward primer, 5′-ARATGAGTCTTCTRACCGAGGTCG-3′; reverse primer, 5′-TGCAAAGACATCYTCAAGYYTCTG-3′; and probe, 5′-[6-FAM] TCAGGCCCCCTCAAAGCCGA [BHQ1]-3′ (45, 46). Real-time PCR was performed on a Bio-Rad CFX384 Touch Real-Time PCR Detection System with TaqMan 2X PCR Universal Master Mix using a 10-μl total reaction volume in duplicate. Ct values were normalized to the calibrator mouse gene, Gapdh (4352932E, Thermo Fisher Scientific). Viral loads of ferret nasal wash samples were measured using 50% tissue culture infectious dose (TCID₅₀) assay in MDCK cells. The Reed and Muench method (47) was used for the TCID₅₀ calculation.

Immunogenicity

Serum samples collected about 3 weeks after boost immunization were used to investigate the immunogenicity of the quadrivalent vaccine. HAI assays were performed as previously described (48). ELISAs were also used to measure antibodies recognizing homologous HAs (H1, H3, H5, and H7) and NAs (N1, N3, N8, and N9) using recombinant HA and NA proteins. Antibodies recognizing group 1 and 2 HA stalks were also measured. Recombinant proteins for the ELISAs were designed on the basis of previously published HA (49), NA (50), group 1 HA stalk (51), and group 2 HA stalk (52) constructs with a Strep-Tag II affinity tag. In particular, for the group 2 HA stalk, a chimeric HA consisting of a globular head of H4 HA and a stalk of H3 HA (cH4/3) was used. The recombinant proteins were expressed in insect cells, purified using Strep-Tactin Sepharose (2-1201, IBA GmbH), and quantified using BCA protein assay kit (23225, Thermo Fisher Scientific) as previously described (51). Purified proteins were diluted in PBS (1 μg/ml) and added to 96-well ELISA plates (50 μl per well) (456537, Thermo Fisher Scientific). The plates were incubated overnight at 4°C followed by the addition of blocking buffer consisting of 1% bovine serum albumin (BSA) in PBS (100 μl per well). After 30 min at room temperature, the plates were washed three times with wash buffer (0.05% Tween 20 in PBS). Serum samples were serially diluted in antibody diluent (1% BSA and 0.05% Tween 20 in PBS) and added to the washed plates (50 μl per well). After incubation at room temperature for 2 hours, the plates were washed three times and 1:10,000 diluted horseradish peroxidase (HRP)–conjugated anti-mouse IgG antibody (A28177, Thermo Fisher Scientific) or anti-ferret IgG antibody (ab112770, Abcam) was added (100 μl per well). After incubation for 1 hour at room temperature, the plates were washed six times followed by a 30-min room temperature incubation with HRP substrate solution (100 μl per well) prepared by adding a 10-mg o-phenylenediamine dihydrochloride tablet (P8287, MilliporeSigma) to 20 ml of phosphate-citrate buffer preparation (P4922, MilliporeSigma). The reaction was stopped by adding 1 M sulfuric acid (100 μl per well), and the optical density was measured at 492 nm (OD₄₉₂). Area-under-the-curve (AUC) values were calculated using Prism9 software v.9.2.0 (GraphPad Software). The baseline for AUC calculations was set as 0.1 to exclude nonspecific signals from the AUC calculation. The OD₄₉₂ of 0.1 is about two times the OD₄₉₂ value from the control wells that were treated equally but without the addition of diluted serum. Reciprocal dilutions (dilution factors) of the serum were used as x values for the AUC calculation. In addition, the concentration of secretory IgA in mice was also measured from BAL fluid collected about 3 weeks after boost immunization. The BAL fluids were serially diluted, and the concentration of IgA was measured as described above. HRP-conjugated anti-mouse IgA antibody (ab97235, Abcam) was used.

Histopathology and immunohistochemistry

NBF-fixed mouse and ferret lungs were processed for histopathology and immunohistochemistry as previously described (53). Hematoxylin and eosin–stained slides were examined from two mice or four ferrets per virus group at 5 days after infection. Immunohistochemistry was done on the same sets of fixed tissues as the histopathology. For the mouse lung tissues, IAV and immune cells (neutrophils, B cells, and T cells) were evaluated by immunohistochemistry. A goat polyclonal primary anti-IAV (1:200; ab20841, Abcam, USA) was used to stain influenza NP proteins, followed by an HRP-conjugated rabbit anti-goat IgG secondary antibody (1:2000; ab6741, Abcam). Anti-CD19 antibody (1:800; 90176S, Cell Signaling Technology, USA), anti-CD3 antibody (1:150; ab16669, Abcam), and anti-Ly6G antibody (1:2,000; ab210204, Abcam) were used to stain B cells, T cells, and neutrophils, respectively. For each of these three primary antibodies, primary antibody staining was followed by an HRP-conjugated goat-anti rabbit IgG secondary antibody (1:10,000, 1:1000, and 1:5000, respectively; ab205718, Abcam). For the ferret lung tissues, only IAV distribution was measured by immunohistochemistry, as above for mouse tissues. All slides were scanned on an Aperio ScanScope XT system, enabling whole-slide analysis.

Plaque reduction assay

About 50 PFU of H10N7 or H1N9 was incubated in serial twofold dilutions of serum from IM- or PBS-immunized mice in Dulbecco’s modified Eagle’s medium, starting at a 1:10 dilution. The virus-serum mixtures were incubated at 37°C for 30 min in a total volume of 250 μl and added onto MDCK cells confluently grown in 12-well tissue culture plates. After incubation (37°C, 45 min with periodic agitation), 1.5 ml of 1% Avicel (CL-661, FMC BioPolymer) in 2× Eagle’s minimum essential medium (115-073-101, Quality Biological), supplemented with 1× antimycotics-antibiotics (15240096, Thermo Fisher Scientific) and TPCK-treated trypsin (1 μg/ml; T1426, MilliporeSigma), was added to the wells. The plates were incubated for 48 hours at 37°C in a 5% CO₂ atmosphere with 80% relative humidity and subsequently fixed with 10% NBF for 15 min at room temperature. Cells were stained with 2% crystal violet. Plaques were manually counted for each dilution and normalized to the plaque counts of a no-serum control. The experiment was performed twice with duplicates for each dilution.

GMP manufacture and toxicology studies

The four vaccine virus (table S1) seed stocks were used for GMP manufacturing of the vaccine components in certified Vero cells (World Health Organization Vero reference cell bank 10-87). Rabbit toxicology and immunogenicity studies were performed under contract by Batelle. Thirty 4- to 7-month-old New Zealand white rabbits (2 to 4 kg) were used for the study. A group of 10 animals received saline administered intranasally (240 μl; 120 μl per naris) and intramuscularly (n = 10; 5 male and 5 female); another group of 10 received the GMP-manufactured vaccine intranasally (20 μg of each antigen; 80 μg total), and the final group of 10 received the GMP-manufactured vaccine intramuscularly (20 μg of each antigen; 80 μg total). Each group consists of five male and five female rabbits. First doses of vaccine were delivered on day 1, and boosters were delivered on day 29. Body weight measurements were made throughout the study. Body temperatures were measured before immunization and at 6 and 24 hours after immunization on days 1 and 29. Blood collection was performed before the study and on days 1, 8, 15, 30, and 45. Urine was collected before the study and on days 1, 8, 15, 30, and 45. Animals were euthanized on day 45 (n = 30) for histopathological analyses from harvested tissues. Good Laboratory Practice was followed for these experiments. Body weight and temperature were evaluated for any differences occurring during the study. Collected blood was evaluated for hematology (cell morphology, erythrocyte count, hemoglobin, hematocrit, leukocyte count, platelet count, mean corpuscular hemoglobin, mean corpuscular volume, and reticulocyte count) using the Advia 120 Hematology Analyzer, for serum chemistry [alanine aminotransferase, rabbit albumin, albumin/globulin ratio, alkaline phosphatase, bilirubin, blood urea nitrogen, calcium, cholesterol, creatine kinase, creatinine C-reactive protein, electrolytes (sodium, potassium, and chloride), globulin, glucose, lactate dehydrogenase, phosphorous, total protein, and triglycerides] by the Roche cobas c501 Chemistry Analyzer, and for immunogenicity. Urinalysis [appearance, bilirubin, glucose, ketones, leukocytes, nitrite, occult blood, pH, protein, specific gravity, total volume, urobilinogen, and microscopic examination of sediment (white blood cells, red blood cells, casts, epithelial cells, mucus, sperm, bacteria, yeast, or crystals)] was performed on collected urine samples over the course of the study using the Roche cobas u411 urine analyzer. Harvested tissues (adrenal glands, bone marrow, brain, cervix, colon, eyes, gallbladder, small and large intestines, heart, kidneys, liver, lungs, lymph nodes, sciatic nerve, nasal turbinates, ovaries, pancreas, pituitary gland, prostate gland, rectum, skeletal muscle, skin, spinal cord, spleen, stomach, testes, trachea, urinary bladder, and uterus) underwent histopathologic evaluation for signs of toxicity by a board-certified veterinary pathologist after fixation in 10% NBF and hematoxylin and eosin staining.

RNA isolation was performed using InnuPure C16 touch system (catalog no. 845-00020-2, Analytik Jena) with innuPREP Virus DNA/RNA Kit-IPC16 (catalog no. 845-IPP-7016096, Analytik Jena). RNA was extracted from 200 μl of GMP viral culture and resuspended in 100 μl of elution buffer. For each sample, 5 μl of total RNA was amplified using the Ovation RNA-Seq System V2 from NuGEN (NuGEN, San Carlos, CA). The amplified total cDNAs were analyzed by an Agilent 2100 Bioanalyzer using the Agilent High Sensitivity DNA Kit (Agilent Technologies, Santa Clara, CA). Amplified cDNA (4 μl) was used as input to generate Illumina sequencing library using Takara ThruPLEX DNA-Seq HV PLUS (Takara Bio USA, San Jose, CA) according to the manufacturer’s instruction. Final Illumina sequencing libraries were analyzed with the Agilent 2100 Bioanalyzer using the Agilent High Sensitivity DNA Kit. Libraries were then pooled and sequenced as 150–base pair (bp) single-read and 8-bp index on an Illumina NextSeq sequencer (Illumina, San Diego, CA). Here, more than 32-Gb sequences were generated. All sequences generated were deposited into the GenBank SRA database (PRJNA847201).

Statistical analysis

Prism 9 software v. 9.2.0 (GraphPad Software) was used for statistical evaluations of antibody concentrations and viral titers by ordinary ANOVA test. Tukey’s multiple comparison test was used as post hoc test to compare antibody concentrations between groups. Viral titers of the IM and IN groups were log-transformed and compared to PBS group using Dunnett’s multiple comparison test as post hoc test. Hierarchical clustering and additional analyses were performed using TIBCO Spotfire Analyst 7.6.0 (TIBCO Software).

Data and materials availability:

All data associated with this study are present in the paper or the Supplementary Materials. The complete MIAME-compliant microarray dataset has been deposited in NCBI’s Gene Expression Omnibus (GEO) (www.ncbi.nlm.nih.gov/geo) and is accessible through GEO Series accession number GSE180226. All sequences have been deposited as a series with accession no. PRJNA847201 at the GenBank SRA database (www.ncbi.nlm.nih.gov/sra/PRJNA847201).