Inflammasome-independent role of the apoptosis-associated speck-like protein containing CARD (ASC) in the adjuvant effect of MF59

Abstract

Clinical studies have indicated that subvirion inactivated vaccines against avian influenza viruses, particularly H5N1, are poorly immunogenic in humans. As a consequence, the use of adjuvants has been championed for the efficient vaccination of a naïve population against avian influenza. Aluminum salts (alum) and the oil-in-water emulsion MF59 are safe and effective adjuvants that are being used with influenza vaccines, but the mechanism underlying their stimulation of the immune system remains poorly understood. It was shown recently that activation of a cytosolic innate immune-sensing complex known as “NLR-Pyrin domain containing 3” (NLRP3) inflammasome, also known as “cryopyrin,” “cold-induced autoinflammatory syndrome 1” (CIAS1), or nacht domain-, leucine-rich repeat-, and PYD-containing protein 3 (Nalp3), is essential for the adjuvant effect of alum. Here we show that the inflammasome component apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC), an adapter protein within the NLRP3 inflammasome, is a crucial element in the adjuvant effect of MF59 when combined with H5N1 subunit vaccines. In the absence of ASC, H5-specific IgG antibody responses are significantly reduced, whereas the responses are intact in NLRP3^−/− and caspase-1^−/− mice. This defect is caused mainly by the failure of antigen-specific B cells to switch from IgM to IgG production. We conclude that ASC plays an inflammasome-independent role in the induction of antigen-specific humoral immunity after vaccination with MF59-adjuvanted influenza vaccines. These findings have important implications for the rational design of next-generation adjuvants.

The primary means of infection- and vaccine-mediated immunologic protection against influenza A viruses is the development of antibodies that bind and neutralize the receptor-binding function of HA. A number of avian influenza A viruses, such as H5N1, have infected humans during the past decade. In response to the recent H5N1 outbreak, several governments, including that of the United States, produced and stockpiled H5N1 subvirion inactivated vaccines (http://www.who.int/csr/resources/publications/WHO_HSE_EPR_GIP_2008_1d.pdf). The primary purpose of these stockpiles is for deployment during the early phases of a pandemic while a matching vaccine is developed. The results from the first clinical trials of unadjuvanted H5N1 vaccines were disappointing in terms of immunogenicity (1). However, substantial improvements have been seen in trials with the adjuvanted formulations specifically using the new-generation oil-in-water emulsion-based adjuvants, which have induced significantly higher antibody responses to low antigen-content vaccines (2, 3).

The most common adjuvants in clinical use with influenza vaccines are the insoluble aluminum salts, generically referred to as “alum,” and the oil-in-water emulsion adjuvant, MF59 (4). MF59 has been tested with H5 and H9 influenza vaccines and showed promising results; an ability to potentiate and increase the breadth of the immune response induced by low-antigen-content vaccines (2, 5). It has been shown that alum and MF59 induce the secretion of a range of cytokines associated with recruitment of innate immune cells to the injection site (6). Among the latter cells are dendritic cells (DCs), which are essential for antigen presentation and the subsequent activation of naïve T cells. The immunostimulatory activity of alum has been attributed recently to a family of innate immune sensors known as “NOD-like receptors” (NLRs) (7). NLRs are a large family of intracellular proteins that are believed to be involved primarily in the innate immune response to microbial pathogens through the recognition of a conserved pathogen-associated molecular pattern (8–10). However, they also contribute by sensing “danger signals,” i.e., endogenous molecules that are produced during tissue damage or inflammation (9, 11). Specifically, in mice deficient in NLR-Pyrin domain containing 3 (NLRP3), decreased IL-1β secretion and antigen-specific humoral immune responses to immunization with alum-adsorbed antigens have been observed (7, 12).

NLRP3 allows the recruitment and autocatalytic activation of the cysteine protease caspase-1 in a large cytosolic protein complex named the “inflammasome” (8). Once activated, the inflammasome mediates caspase-1 cleavage of the inactive precursor of the proinflammatory cytokine IL-1β resulting in release of mature IL-1β. The adapter protein apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC) bridges the interaction between NLRP3 and caspase-1, making it essential for activation of the inflammasome (8). Whether any of the NLRP3 inflammasome components (NLRP3, ASC, and caspase-1) play a role in MF59-mediated immune stimulation has not been addressed.

The goal of this study was to understand the roles of the different NLRP3 inflammasome components (NLRP3, ASC, and caspase-1) in MF59-mediated immune stimulation when combined with H5N1 subvirion inactivated vaccines. We found that in ASC^−/− mice, unlike NLRP3^−/− and caspase-1^−/− mice, the production of H5-specific IgG antibodies was significantly reduced after immunization with MF59-adjuvanted H5N1 vaccines in comparison with WT mice. This effect was adjuvant specific, because ASC^−/− mice elicited an H5-specific IgG antibody response comparable with that of WT mice when immunized with unadjuvanted H5N1 vaccines. In addition, we found that the development of germinal center (GC) B cells in ASC^−/− draining lymph node cells was reduced. These defects extended to memory, because antigen-specific recall antibody responses were impaired significantly in ASC^−/− mice. Finally, inflammatory cytokine production by ASC^−/− bone marrow-derived dendritic cells (BMDCs) was lower than that of WT cells. Therefore, ASC plays a caspase-1 (and thus inflammasome)-independent role in the induction of antigen-specific humoral immunity in response to MF59-adjuvanted vaccines.

Results

MF59 Is Superior to Alum in Priming for Robust H5-Specific Antibody Responses.

HA-specific antibodies are considered the most important correlate of protection elicited by inactivated influenza vaccines. It was shown previously that in mice MF59 is superior to alum and several other adjuvants in priming for a robust antigen-specific humoral immune response when combined with seasonal influenza vaccines (13). To confirm this observation, we primed four groups of WT C57BL/6 mice with H5N1 subvirion vaccines (MF59- or alum-adjuvanted or unadjuvanted) or with sublethal H5N1 influenza virus (homologous to the vaccine strain) infection. After boosting all groups with the unadjuvanted H5N1 subvirion vaccine in PBS, we first determined the serum levels of H5-specific IgG antibodies induced. We found that mice primed with the MF59-adjuvanted vaccine elicited significantly higher antibody titers than the mice primed with alum-adjuvanted or unadjuvanted vaccine (Fig. 1A). The H5-specific IgG titers in the MF59-adjuvanted group were comparable to those elicited in the infection-primed mice (Fig. 1A). We next determined the frequency of H5-specific antibody-forming cells (AFCs) in the bone marrow and spleen after the boosting immunization. For the mice primed with the unadjuvanted vaccine, the number of AFCs was below the assay detection limit (Fig. 1 B and C). Consistent with the serum IgG results, however, the number of H5-specific IgG-forming cells was comparably elevated in the MF59- and infection-primed mice and was significantly higher than that observed in alum-primed mice (Fig. 1 B and C).

Fig. 1.

Superior priming with MF59-adjuvanted H5N1 vaccines in comparison with the alum-adjuvant or the unadjuvanted formulations. Four groups of C57BL/6 WT mice (n = 5) were primed i.m. with 1 μg of H5N1 subvirion vaccine (MF59- or alum-adjuvanted or unadjuvanted) or with sublethal A/vietnam/1203/2004 (H5N1) influenza virus infection. All groups were boosted with 3 μg of H5N1 subvirion vaccine in PBS, and sera and other organs were harvested 5 d after the boost. (A) Mean circulating levels of H5-specific IgG antibodies. (B and C) Mean number of H5-specific IgG-forming cells in the bone marrow (B) and spleen (C). The results are representative of at least two separate experiments. Statistically significant differences between WT and mutant groups are indicated by asterisks; *P < 0.05. ns, not significant. nd, not detected. Error bars show SE.

Humoral Immune Response to MF59-Adjuvanted Vaccines Requires ASC but Not NLRP3 or Caspase-1.

The NLRP3 inflammasome has been implicated in the adjuvant effect of alum. Therefore, we asked whether the NLRP3 inflammasome plays any role in mediating the adjuvant activity of MF59 (7, 12). We first determined the HA inhibition (HI) titers of WT, NLRP3^−/−, ASC^−/−, and caspase-1^−/− mice after vaccination with MF59-adjuvanted H5N1 subvirion vaccine. We found that ASC^−/− mice, but not NLRP3^−/− or caspase-1^−/− mice, elicited significantly lower HI titers than WT mice (Fig. 2A). We next determined the serum levels of the different H5-specific antibody isotypes in immunized mice 3 wk after a booster immunization. Consistent with the HI results, circulating levels of H5-specific IgM, IgG, IgG1, IgG2b, and IgG2c antibodies were comparably elevated in the serum of WT, NLRP3^−/−, and caspase-1^−/− mice (Fig. 2 B–F). Immunized ASC^−/− mice, however, elicited significantly less IgG, IgG1, IgG2b, and IgG2c antibodies than WT mice (Fig. 2 C–F). Interestingly, circulating levels of H5-specific IgM antibodies in ASC^−/− mice were comparable to those of WT mice (Fig. 2B). As a control, we examined the H5-specific antibody response to the unadjuvanted vaccine in WT, NLRP3^−/−, ASC^−/−, and caspase-1^−/− mice (Fig. S1). ASC^−/− mice responded robustly to the unadjuvanted vaccine, indicating that the defective response observed with the MF59-adjuvanted vaccine was not caused by a general defect in these mice in response to this particular inactivated influenza vaccine. We also tested the antibody response to H5N1 vaccine adjuvanted with incomplete Freund's adjuvant (IFA) in WT and ASC^−/− mice. Again, there were no significant differences in H5-specific IgG1 or IgG2c antibody titers (Fig. S2). In addition, infection with a sublethal dose of an attenuated H5N1 virus resulted in comparable anti-H5 IgG titers in WT and ASC^−/− mice when measured 3 wk after infection (Fig. S3). These results suggested that ASC is specifically required for the optimal generation of antigen-specific antibodies in response to MF59-adjuvanted H5 vaccines. In addition, class switching anti-H5 antibodies from IgM to IgG, rather than antibody production itself was impaired in ASC^−/− mice.

Fig. 2.

Defective humoral immune response to MF59-adjuvanted H5N1 vaccines in ASC-KO mice. C57BL/6 WT (n = 5), Nlrp3^−/− (n = 4), ASC^−/− (n = 5), and caspase-1^−/− (n = 5) mice were immunized twice, 3 wk apart, with 1 μg of H5N1 subvirion vaccine adjuvanted with MF59. Sera were collected 3 wk after the boosting immunization. (A) Mean serum HI titers. (B–F) Mean circulating levels of H5-specific IgM (B), IgG (C), IgG1 (D) IgG2b (E), and (F) IgG2c antibodies. Results are representative of at least two separate experiments. Statistically significant differences between WT and mutant groups are indicated by asterisks; *P < 0.05. Error bars show SE.

ASC Is Required for GC B-Cell Formation After Immunization with MF59-Adjuvanted Vaccines.

Antibody class switching after vaccination or infection takes place mainly within the GC reaction in the draining lymph nodes (14). The observed decrease in IgG production in ASC^−/− mice after MF59-adjuvanted H5N1 vaccination suggested a defect in the development of the GC reaction in these mice. To investigate this possibility, we immunized WT, NLRP3^−/−, ASC^−/−, and caspase-1^−/− mice with the MF59-adjuvanted H5N1 vaccine and harvested the draining (inguinal) lymph nodes as well as sera 7 d later. H5-specific IgM and IgG antibodies were undetectable in the sera of ASC^−/− mice, whereas both isotypes were detectable in the other three groups (Fig. 3 A and B), indicating a delay in the induction of the antibody response in the ASC^−/− mice. In agreement with this observation, WT, NLRP3^−/−, and caspase-1^−/− mice all had significantly larger lymph nodes than ASC^−/− mice, as evidenced by the difference in cellularity (Fig. 3C). We next looked at the frequency of GC B cells in the draining ipsilateral lymph nodes. We found that in WT, NLRP3^−/−, and caspase-1^−/− mice, GC B cells were comparably developed, accounting for 30–50% of the B220⁺ B-cell population (Fig. 3 D and E). On the other hand, GC B cell frequencies in the ASC^−/− mice were significantly lower (Fig. 3 D and E). The generation of a GC B-cell reaction requires the help of CD4⁺ T cells, so we asked whether they exhibited similar defects. Indeed, we found that absolute numbers of CD4⁺ T cells in the ASC^−/− mice were significantly lower than in WT mice (Fig. 3F). These results show that ASC, independent of the caspase-1 inflammasome, is required for the rapid induction of the antibody response to MF59-adjuvanted H5N1 vaccines. Moreover, the development of a GC reaction in the draining lymph nodes of ASC^−/− mice is impaired, explaining the defect observed in antibody isotype switching in these mice. This defect was not B-cell specific, because CD4 T-cell frequencies also were significantly lower in ASC^−/− mice, potentially contributing to the GC phenotype.

Fig. 3.

Analysis of GC B-cell development after immunization with MF59-adjuvanted H5N1 subvirion vaccine. C57BL/6 WT (n = 5), Nlrp3^−/− (n = 4), ASC^−/− (n = 5), and caspase-1^−/− (n = 3) mice were immunized with 5 μg of H5N1 subvirion vaccine adjuvanted with MF59. Sera and draining (inguinal) lymph nodes were collected 7 d after immunization. (A and B) Mean circulating levels of H5-specific IgM (A) and IgG (B) antibodies were determined as in Fig. 1. (C) Mean number of live cells in the ipsilateral draining lymph node. (D) Representative FACS plots for lymphocytes from the draining lymph nodes stained with monoclonal antibodies to B220 and CD95 (Fas) on day 7 after immunization with 5 μg of H5N1 subvirion vaccine adjuvanted with MF59. The gated population (circled in red) from each plot represents GC B cells, which were defined as cells that were negative for surface CD4, CD8, and CD11b, with low expression of surface IgD and high expression of surface B220 and FAS (CD195). (E) Total number of GC B cells as determined by FACS analysis of the cells with the phenotype defined in D. (F) Total number of CD4⁺ lymphocytes. Statistically significant differences between WT and mutant groups are indicated by asterisks; *P < 0.05. Error bars show SE.

ASC Is Required for Robust Antigen-Specific Recall B-Cell Responses.

The ultimate outcome of a GC B-cell reaction is the generation of high-affinity, antigen-specific memory B cells and long-lived plasma cells (14). Upon antigen reexposure, memory B cells expand rapidly and differentiate into AFCs. We therefore investigated whether the defect in GC B-cell development in ASC^−/− mice after priming would, in turn, result in a defective H5-specific memory B-cell compartment. To test that notion, we primed WT, NLRP3^−/−, ASC^−/−, and caspase-1^−/− mice with one dose of MF59-adjuvanted H5N1 vaccine, and all mice were given a booster with the same vaccine in PBS 3 wk later. Similar to our observations with the primary response, 5 d after antigen reexposure, WT, NLRP3^−/−, and caspase-1^−/− mice had significantly larger draining lymph nodes than ASC^−/− mice, and that response was mirrored by a greater number of live cells (Fig. 4A). Also, GC B-cell numbers were significantly lower in the draining lymph nodes of ASC^−/− mice than in the other groups (Fig. 4B). Moreover, the number of CD138⁺ plasma cells was significantly lower in ASC^−/− mice than in WT, NLRP3^−/−, and caspase-1^−/− mice (Fig. 4C). In agreement with the decreased number of CD138⁺ cells in ASC^−/− mice after the boost, the number of H5-specific AFCs in the bone marrow was similarly lower (Fig. 4D). As expected, when the H5-specific IgG antibody titers in mice sera before and 5 d after the boost were compared, ASC^−/− mice showed a slight (twofold) increase in antibody titer (Fig. 4E). In contrast, WT, NLRP3^−/−, and caspase-1^−/− mice showed a significantly larger (five- to sixfold) increase in H5-specific IgG titers after the boost, signifying a robust memory response in these mice (Fig. 4E). These results demonstrate that ASC is required for the optimal priming of antigen-specific B-cell memory responses with MF59-adjuvanted vaccines.

Fig. 4.

Poor H5-specific antibody recall responses in immunized ASC^−/− but not in WT, NLRP3^−/−, or caspase-1^−/− mice after antigen reexposure. C57BL/6 WT (n = 5), Nlrp3^−/− (n = 4), ASC^−/− (n = 3), and caspase-1^−/− (n = 3) mice were primed with 1 μg of H5N1 subvirion vaccine adjuvanted with MF59 and were boosted 3 wk later with 5 μg of H5N1 subvirion vaccine in PBS. Sera were collected before and 5 d after the boost. Ipsilateral draining lymph nodes and bone marrow were collected 5 d after the boost. (A) Mean number of live cells in the draining ipsilateral draining lymph nodes. (B) Mean number of plasma cells in the draining lymph nodes as determined by FACS analysis of cells that were negative for surface CD4, CD8, and CD11b, with low expression of surface IgD and high surface expression of CD138. (C) Mean number of GC B cells determined as in Fig. 2E. (D) Mean frequencies of H5-specific AFCs in the bone marrow of WT, Nlrp3^−/−, ASC^−/−, and caspase-1^−/− mice 5 d after antigen reexposure. (E) Fold increase in H5-specific IgG titers 5 d after antigen reexposure. Statistically significant differences between WT and mutant groups are indicated by asterisks; *P < 0.05. Error bars show SE.

ASC Expression in DCs Is Required for Adequate Production of Inflammatory Chemokines.

We have shown that ASC deficiency results in poor B-cell and CD4 T-cell responses to MF59-adjuvanted H5N1 subvirion vaccines. To elucidate further the mechanism by which ASC acts, we started from the knowledge that the adjuvant effect of MF59 and alum depends at least partially on their ability to induce the secretion of chemokines responsible for recruiting inflammatory cells to the site of injection (6). DCs are the most potent antigen-presenting cells capable of stimulating naïve T cells for the subsequent induction of antigen-specific T-cell and B-cell responses. It has been shown that DCs internalize most of the adjuvant within 48 h of MF59 i.m. injection, (15). We therefore investigated whether the defective antigen-specific antibody response in immunized ASC^−/− mice was caused by a defective inflammatory response by BMDCs. To examine this possibility, BMDCs from WT, NLRP3^−/−, ASC^−/−, and caspase-1^−/− mice were incubated with MF59, and the secretion of proinflammatory chemokines in the culture supernatant was determined. WT BMDCs secreted significantly more proinflammatory chemokines such as macrophage inflammatory protein (MIP)1b and MIP2 (P = 0.002 and 0.005, respectively) than did BMDCs derived from ASC^−/− mice but not significantly more than BMDCs derived from NLRP3^−/− and caspase-1^−/− mice (Fig. 5 A and B). MF59 was responsible for the chemokine stimulation, because the differences in MIP1b secretion were seen in cells treated with MF59 alone but not in cells treated with vaccine antigen alone (Fig. S4). These results suggest that the inflammasome-independent role of ASC in generating the proper inflammatory environment after immunization may be the underlying cause of the defective induction of humoral immunity in the ASC-deficient mice.

Fig. 5.

Defective inflammatory response to MF59in ASC^−/− BMDCs. BMDCs were stimulated for 24 h with MF59 in complete medium (1:100 vol/vol), and the quantities of (A) MIP1β and (B) MIP2 were determined in cell-culture supernatants. Tests were assayed in triplicate or quadruplicate. Statistically significant differences between WT and mutant groups are indicated by asterisks; *P < 0.05. Error bars show SE.

Discussion

Since their emergence in 1996, highly pathogenic H5N1 influenza viruses have posed a threat of becoming the causative agent of the next influenza pandemic. This situation has been the catalyst for an increase in the amount of resources applied to preparedness activities for such a pandemic. Vaccine development has been one of the foci of these activities, and a major challenge that has emerged has been the formulation of conventional influenza vaccines so that they can effectively prime a population immunologically naïve to H5N1 antigens. Unadjuvanted split and subunit H5N1 vaccines have been shown reproducibly to be poorly immunogenic in humans (1, 16). The addition of alum adjuvant induced only marginal improvements (17). In contrast, oil-in-water–based adjuvants (e.g., MF59) have been shown to prime for a robust and broadly cross-reactive antibody response to subvirion H5N1 vaccines, with greatly reduced antigen requirements (18). MF59 has shown encouraging results in several influenza formulations and is licensed for use in seasonal influenza vaccines in Europe (5, 19–21). However, the mechanisms underlying the ability of MF59 and other adjuvants to stimulate the immune system remain poorly understood. To develop more defined next-generation adjuvants, it is critical to determine the immunologic processes that are stimulated by MF59 and other efficacious adjuvants. Recent studies have shown that induction of antigen-specific antibody responses by commonly used vaccine adjuvants, such as alum and Freund's complete and incomplete adjuvants, did not require signaling through Toll-like receptors (22). However, the ability of alum and other particulate adjuvants to enhance IL-1 secretion via NLRP3 has been described (7, 12, 23). In addition, some studies have shown that enhancement of antigen-specific humoral immunity by alum also was NLRP3 dependent (7, 12), although later reports suggested that NLRP3 was not required (24, 25). Here, we found that the antibody response to H5N1 subunit influenza vaccines adjuvanted with MF59 critically depend on the presence of the adapter molecule ASC. After immunization with MF59-adjuvanted H5N1 subvirion vaccines, the amount of H5-specific antibodies of IgG1, IgG2b, and IgG2c subtypes were all significantly lower in ASC^−/− mice than in WT mice. In contrast to studies with alum, we found that neither NLRP3 nor caspase-1 was required for the antigen-specific antibody responses to MF59-adjuvanted H5N1 influenza subvirion vaccines, indicating that the proposed role of ASC is independent of the caspase-1 inflammasome. It is interesting, however, that the murine antibody response to antigens emulsified in Freund's complete adjuvant has been shown to be unaffected by the absence of ASC (7). This result suggests that, even among emulsion-based adjuvants, the molecular requirement for the proper induction of an adaptive immune response varies. It also is possible that the complex nature of Freund's complete adjuvant affords immune stimulation through a number of redundant mechanisms.

Complexities of the response to adjuvants aside, our data support recent findings in the murine model of collagen-induced arthritis as well as in the development of experimental autoimmune encephalomyelitis that show an inflammasome-independent role for ASC in the induction of various forms of immunity and immunopathology (26, 27). There are several molecular mechanisms by which ASC may control the antigen-specific antibody response to MF59-adjuvanted vaccines. A key underlying feature in the ASC^−/− mice was the significantly reduced production of key inflammatory cytokines by BMDCs compared with WT-derived cells after stimulation with MF59. Consistent with a defect in antigen-presenting cells, it has been suggested that antigen-specific activation of T cells by ASC-deficient BMDCs was significantly impaired (26). In line with this notion, after a single immunization with MF59-adjuvanted H5N1 vaccine, we observed a delayed antigen-specific antibody response and significantly lower numbers of CD4⁺ T cells in the draining lymph nodes in ASC^−/− but not NLRP3^−/− or caspase-1^−/− mice in comparison with their WT counterparts. In humans, the early development of an antigen-specific CD4⁺ T-cell response was shown to be a successful predictor of the robustness and persistence of an antibody response following MF59-adjuvanted H5N1 vaccination (28). We also demonstrated that after immunization the development of GC B cells in the draining lymph nodes was significantly impaired in ASC^−/− mice in comparison with that seen in WT, NLRP3^−/−, and caspase-1^−/− mice. In addition, we showed that these defects resulted in a poor memory B-cell response in ASC^−/− mice. Because the aggregate evidence demonstrates a critical role for ASC independent of inflammasome, an entire body of literature linking chronic models of inflammation and infection to inflammasome activation by experiments performed in ASC^−/− mice should be reassessed.

The promising nature of MF59 use in humans and the identification of ASC as a critical factor for the proper induction of the adaptive immune response to MF59-adjuvanted vaccines would have important implications for the formulation of future influenza vaccines.

Materials and Methods

Mice and Immunization.

Cryopyrin/Nlrp3^-/-, ASC^-/-, and caspase-1^-/- mice were backcrossed to a C57BL/6 background for at least 10 generations. Mice were housed in a pathogen-free facility, and the animal studies were conducted under protocols approved by the St. Jude Children's Research Hospital Committee on Use and Care of Animals. Animals were immunized intramuscularly twice with 1, 3, or 5 μg of monovalent H5N1 influenza (rgA/Vietnam/1203/2004) subvirion vaccine (Aventis Pasteur Inc., Swiftwater, PA) either unadjuvanted (in PBS), mixed with MF59 (1:1) or adsorbed to aluminium hydroxide at 14.1 mg/ml Alum, 3 μg/ml antigen and 5 mM Histidine buffer pH 6.5 at 4 °C overnight. Incomplete Freund's adjuvant or IFA (Sigma, St. Louis, MO) was mixed with the vaccine solution in 1:1 (vol/vol) ratio and emulsified just before injection. Each mouse dose contained 500 μg Alum. The intranasal infections were performed with 100 egg infectious dose 50 of the rgA/Vietnam/1203/2004 (ΔHA) diluted in PBS to a final volume of 30 μl.

Standard procedures and methods such as hemagglutination inhibition assay, flow cytometry, antigen-specific ELISA and ELISPOT assays, cytokine analysis and statistical analyses are described in SI Materials and Methods.