A universal influenza vaccine capable of inducing cross-protection among heterosubtypic influenza strains is critically needed to prevent seasonal and pandemic flu outbreaks. The highly conserved NP and M2 of influenza A virus have been used as target antigens in the development of universal influenza vaccines. In mice, previous studies have shown that gene- and/or vector-based NP+M2 vaccines induced strong humoral and cellular responses, and protected against lethal influenza virus challenges, the combination of NP and M2 is superior to the sole one when the immune response and protection efficacy were considered 
. Here, we describe a vaccine based on a fusion protein of NP and M2e expressed in E. coli
. In mice, the NM2e fusion protein elicited robust antibody responses and T cell responses against NP and M2e. More importantly, mice immunized with NM2e formulated with Al(OH)3
adjuvant were protected against lethal challenge with high doses of influenza A virus PR8 (20LD50) compared with the relatively low challenge dose (1 LD90 
or 4 LD50 
) in other studies with M2-based vaccines. It is noticeable here that NM2e did contain the character of both NP and M2e and induced stronger protective immune response than NP protein. The present results suggest that E. coli
-expressed NM2e fusion protein is immunogenic in mice and may be a suitable candidate for a universal influenza vaccine.
It is necessary to understand the immune correlates of protection for new vaccine types. Here, NM2e immunization induced a substantial antibody response to M2e in mice, and an analysis showed that the protection was closely correlated with the anti-M2e antibody titer. Previous experiments have shown similar results 
. Anti-M2e antibody is unable to neutralize the virus to prevent infectivity, but it is able to disrupt the viral life cycle and kill infected cells by the mechanism of antibody-dependent cell-mediated cytotoxicity (ADCC) 
. M2e-specific IgG2a isotype has been proposed as an effective inducer of the ADCC response 
. Jegerlehner et al
suggested that IgG2a levels were correlated with protection against influenza infection in mice. However, Denis et al
reported that low levels of anti-M2e IgG2a induced by PapMV-CP-M2e immunization did not efficiently protect mice against a challenge with 4 LD50
of influenza A virus; thus, they considered that anti-M2e IgG1 may also play a role in M2e-mediated protection. Our results demonstrated that protection was highly correlated with not only IgG2a but also IgG1. The M2e peptide contains an MHC class II-restricted epitope 
, and studies have documented that influenza-specific CD4+ T cells are involved in immune protection 
. In the present study, M2e-specific IFN-γ-, IL-4-, and IL-10-secreting SMNCs (mainly CD4+ T cells; data not shown) were significantly correlated with protection. It was reported that following influenza infection, antigen-presenting cells secrete IL-10, which contributes to the differentiation of Th0 cells into Th2 cells; subsequently, Th2 cells secrete IL-4, IL-5, and IL-6, which help to preferentially drive IgG1, IgA, and IgE antibody production by antibody-secreting plasma cells 
. Th1 cells secrete IFN-γ, which helps to produce IgG2a antibodies 
. Thus, our data indicate that both M2e-specific antibodies and CD4+ T cells contribute to the protection induced by NM2e protein.
Several studies have reported that NP plays a role in the elimination of influenza virus-infected cells via specific CD8+ killer T cells 
. However, recent studies have suggested that antibodies against NP are necessary for NP immunization to confer protection and that NP-immune serum can transfer protection 
. Although NP-specific IgG antibodies had no effect on neutralization and failed to block viral infection of cells, antibodies to NP may nevertheless provide an unexpected yet important mechanism of protection against influenza 
. In the present study, protection against virus challenge was closely correlated with the presence of anti-NP antibodies, including IgG1 and IgG2a isotypes. Furthermore, the protective effect was significantly correlated with SMNCs specific for NP55–69 (H-2d
-restricted Th epitope), but not SMNCs specific for NP147–155 (H-2d
-restricted CTL epitope). It is not surprising that NP-specific CD8+ T cells were not elicited in response to NM2e immunization because although gene- and vector-based NP vaccines easily induce CD8+ T cells, protein vaccines might not, even with CpG as adjuvant. Further research is needed to identify a suitable adjuvant that favors the induction of CD8+ T cells in response to NM2e protein vaccination. Meanwhile, it is essential to identify the mechanism of action of NM2e immunity and to establish in vitro
assays for measuring immunity.
Adjuvant is required for protein subunit vaccines to elicit effective and long-lasting immune protection 
. In this study, inclusion of the Al(OH)3
adjuvant, which is widely and safely used in human vaccines 
, enhanced both humoral and cellular immune responses elicited by NM2e. However, it should be noted that alum boosts mainly the Th2 immune response, while it inhibits the Th1 and CTL response elicited by many antigens 
, which was also shown in this study. Different from the immune mechanism of alum adjuvant, CpG efficiently promotes Th1 and CTL responses 
. In the present study, inclusion of CpG alone considerably enhanced the humoral immune response elicited by NM2e, however it did not enhance the CTL response markedly. The inclusion of CpG with NM2e improved the anti-M2e IgG1 titer, but not the IgG2a titer (, right), which was different from the NP-specific IgG1/IgG2a pattern (anti-NP IgG1 titer decreased and anti-NP IgG2a titer increased). McCluskie MJ et al 
once reported CpG together with the E. coli
heat-labile enterotoxin (LT) strengthened the humoral response against some antigen, while others not, so adjuvanticity may have also depended on the particular antigen. NP is one of the internal protein of influenza virus, while M2e is the extracellular domain of the membrane protein 2, thus we infer CpG showed different adjuvanticity when two different antigens were used. In addition, Shim B-S 
once prepared the construct expressing three tandem copies of M2e conjugated to C-terminus sequence of M2 protein (3M2eC), their research showed that immunization with 3M2eC induced predominantly IgG1 as compared to IgG2a subclass, similar to the result in this study.
CpG was expected to complement the immune effect induced by Al(OH)3
when used together in the formulation of NM2e protein. However, CpG plus Al(OH)3
did not show a clear synergistic effect on the immunogenicity of NM2e. It is possible that remnant endotoxin (about 2000 EU/mg) in the NM2e formulation expressed in E. coli
might have interfered with the effect of CpG. Our recent data indicated that once remnant endotoxin was removed further from the NM2e protein formulation, CpG and Al(OH)3
used together did show a clear synergistic effect on the immune response (Fig. S1
) and protective efficacy (Fig. S2
) of NM2e, and inclusion of Al(OH)3
alone still markedly improved the immune efficacy of the NM2e vaccine.
In conclusion, we have described a potential universal influenza vaccine that provides cross-protective immunity against influenza A virus PR8 in mice. The vaccine is based on the recombinant fusion protein NM2e expressed in E. coli, which consists of the 23-amino acid external domain of M2 protein attached to the C-terminus of NP. When administered with Al(OH)3 as adjuvant, NM2e provided almost complete protection against a lethal-dose challenge of A/PR8 in mice. The combination of Al(OH)3 and NM2e offers promising prospects for further vaccine development.