The recent emergence of an unanticipated pandemic virus has highlighted the need for influenza vaccines providing broad coverage across multiple strains and subtypes. This study demonstrates that a single dose of rAd vaccines expressing two highly conserved influenza virus antigens (NP and M2) protects from virulent influenza virus challenges with multiple widely divergent subtypes. Protection is greatly enhanced by i.n. administration, develops by 2 weeks post-immunization, and persists for at least 10 months. Intranasal rAd vaccination decreased lung virus titers and accelerated challenge virus clearance relative to i.m. vaccination. This reduction in virus titers corresponded to greatly reduced morbidity, as shown by weight loss, compared to controls and i.m. immunized animals.
Antibody and T-cell responses against both NP and M2 were seen. Previous work suggests that dominant protective mechanisms are likely M2e-specific antibodies that limit viral spread 
and mediate antibody-dependent cellular cytotoxicity 
, and NP-specific CD8+
T cells eliminating infected cells by cytolysis 
. M2e and NP-specific helper T cells probably make a lesser contribution as direct anti-viral effectors. A role for NP-specific antibody has also been proposed 
, but the significance of this in protection is debated. As with prime-boost studies 
, the A/NP+M2 rAd vector combination was superior to either component alone in providing protection, which may be due to induction of complementary effector mechanisms including virus-specific IgA and T cells in the respiratory tract.
Targeting multiple viral antigens, in this case NP and M2, clearly has advantages over using a single antigen. Immunization with either M2-rAd or A/NP-rAd alone confers less protection than using a mixture of these two rAd vectors (). The A/NP+M2-rAd combination also has potential advantages in preventing emergence of viral escape mutants, which have been reported for NP 
and M2 
. However, even under selective pressure from monoclonal antibodies, very few escape mutant sequences were observed for M2e 
suggesting that this region is biologically constrained. Combination vaccines also reduce the possibility that individuals with certain HLA types may be overall non-responders to the vaccine. The latter is a potential problem with M2 due to the small size of this protein (97 amino acids) and the limited number of potential epitopes it may contain, but less of a concern with the larger NP (498 amino acids) within which multiple epitopes for numerous MHC-I and –II alleles have been identified 
. It is possible that incorporation of additional antigens (such as M1 or PB1) into heterosubtypic vaccines may further broaden and enhance the response.
HA-expressing rAd influenza vaccines intended to induce strain-matched neutralizing antibodies have been studied in mice 
, chickens 
, and humans 
, but would require regular reformulation to accommodate antigenic variation in HA. Recent studies show that monoclonal antibodies against conserved epitopes in the HA stem region can neutralize viruses of several subtypes 
. However, such antibodies do not react with all HA subtypes, so multiple such immunogens would be needed to cover all subtypes. NP-expressing rAd5 and chimp rAd vectors given i.m. provided partial protection from both H1N1 and H5N1 challenges in mice 
, but required a higher vaccine dose than reported here. Our findings demonstrate that a single dose of the A/NP+M2-rAd combination given i.n. provided complete protection against highly virulent H1N1, H3N2, and H5N1 influenza virus challenges.
Compared to the DNA prime-rAd boost approach previously reported 
, the single-dose rAd strategy reported here has the advantage of a streamlined vaccination protocol. Significantly, it obviates the requirement for multiple vaccine doses used in prime-boost regimens (3 doses of DNA given at 2 week intervals by i.m. injection, followed by rAd given i.m. or i.n. one month later is typical). We have previously demonstrated that for NP-based immunization, DNA priming enhances protection over that of i.m. rAd alone 
. However, results reported here suggest that priming is less critical when rAd is given intranasally.
The public health applications of prime-boost and single-dose rAd vaccination are quite different. DNA priming could be administered during routine health care to establish basal immunity. It would offer partial protection in the event of an unexpected outbreak or drift, as well as priming for enhanced responses to rAd or other viral boosts. The DNA vaccines can be given repeatedly without concern about anti-vector immunity. In contrast, the single dose rAd vaccination could be used on an urgent basis early in a pandemic or unexpected outbreak.
Single-dose i.n. rAd protected within 2 weeks, with maximal protection by 3 weeks, which compares favorably with strain-matched vaccines. Immunity induced by single-dose rAd is also long-lived, with complete protection observed for at least 10 months after i.n. immunization. Although i.m. rAd provided some protection, this did not prevent weight loss and was waning at 10 months. Durable protection is important for vaccines designed to cover the gap between emergence of a new strain and the availability of matched vaccines.
Concern has been raised about the possibility of adenovirus vectors accessing the central nervous system. Although transgene expression in the olfactory bulb occurs after i.n. administration of rAd vectors to mice, this is transient, low level, and not associated with inflammation 
. Another potential limitation of rAd vaccines is host immunity to the vector which may interfere with vaccination 
. This is a particular concern for Ad5-based vectors, to which much of the human population has neutralizing antibody 
. While experimental animals mount strong immune responses against rAd vectors, this is not always the case in humans. In gene therapy trials some individuals do not make neutralizing antibody responses against rAd given i.n., even after repeated administrations 
. Other reports suggest that rAd immunization may induce de novo
T-cell responses against the transgene despite pre-existing immunity, albeit lower levels than in seronegative individuals 
. To circumvent immunity, vectors based on rare adenovirus serotypes 
, non-human primate adenoviruses 
, or chimeric viruses 
have been suggested in place of current Ad5 vectors.
The rAd doses used here are broadly similar to those used in other vaccine studies. It may be possible to use lower rAd doses. Preliminary results using A/NP-rAd given i.n. suggest that similar levels of protection are achieved with a 10-fold lower rAd dose (Figure S4
). Studies to assess minimum protective doses for the A/NP+M2-rAd combination are ongoing. An effective rAd vaccine dose for humans would have to be addressed in future clinical trials, and may not require a dose proportionate to body weight. Further vector optimization (for example by encoding both NP and M2 on a bicistronic vector) may be possible, and could allow vaccine dose to be reduced still further. It should be noted that since rAd-based vaccines targeting conserved antigens would not need to be changed frequently, vaccine manufacture could occur on an ongoing basis to produce a stockpile, rather than on a seasonal basis as is the case for current influenza vaccines.
Immune correlates of protection are needed for new vaccine types. IgA is not required for protection, but may play a role in protection when present, and could provide a useful correlate. Serum IgG responses do not correlate with protection. They are similar between i.n. and i.m. rAd immunizations which differ in protection (; Figure S1
), and develop earlier after i.m. than i.n. rAd, strengthening at 2 weeks as protection decreases (Figure S3A
). Mucosal IgG appears more promising, as i.n. rAd immunization induced higher BAL IgG responses than i.m. rAd. Interestingly, this indicates that i.n. rAd immunization likely induces IgG-secreting cells resident within the respiratory tract; if antibody reached the BAL by transudation from serum (where IgG levels are similar between i.n. and i.m. immunized mice), then BAL IgG levels would be equivalent regardless of immunization route.
Ideally, correlates of protection should be feasible to assess in humans with non-invasive sampling methods. Anatomical compartmentalization of cellular immune populations after i.n. immunization (ref. 21 and this study) complicates the matter. Cellular correlates can be identified, for example IFN-γ secreting virus-specific T cells in the lungs, but cannot be directly measured in humans. However, the frequency of IFN-γ secreting cells in blood increases between 1 and 6 months after i.n. immunization (Figure S3B
), suggesting equilibration between lung and blood T cell pools over time. If the relevant lung T cells possess a distinctive phenotype of memory, homing or activation markers, perhaps low numbers of comparable cells could be detected in the circulation.
Our studies using a non-replicating viral vector rather than productive infection are in agreement with reports that virus-specific T cells resident in the lungs after clearance of viral infection exhibit an activated phenotype in both mice 
and humans 
. While the classic paradigm is that during recall responses memory T cells activated in draining lymph nodes recirculate back to the site of infection to clear pathogen, T cells already present in tissue and re-activated locally may be able to mediate immediate effector function to control virus 
. This agrees with our observation that virus titers were significantly reduced from 2 days post-challenge, with a trend for lower titers at day 1, in i.n. but not i.m. immunized animals.
Detection of elevated cytokine levels (IFN-γ, mKC, IL-12) in BAL at both one and 10 months after i.n. rAd vaccination () is surprising. This was transgene-independent, and thus due to the rAd vector. mKC (CXCL1) is a powerful neutrophil chemoattractant and functional homolog of human IL-8/CXCL8 
. IFN-γ is immunostimulatory and secreted by various cells including activated T cells and macrophages 
. The elevated IFN-γ levels in BAL after i.n. rAd immunization could result from continued IL-12 secretion. IL-12 promotes differentiation of CD4+
T cells towards a Th1 phenotype 
and maintains CD4+
T cell effector function 
. Continued IL-12 expression in BAL after i.n. rAd immunization may maintain CD4+
T cell activation, which could sustain the strong virus-specific CD8+
T-cell responses observed in the lung.
We have not yet identified the cellular source of the cytokines seen in BAL, but rAd can infect immature dendritic cells (DC) from both mice and humans, causing them to mature and secrete IL-12 
. This occurs independently of transgene expression 
via a TLR9/MyD88 dependent pathway in vitro 
. Earlier studies demonstrated that exposure to an aerosolized antigen induced an activated CD11c+
DC subset in BAL that retained antigen presenting function for several weeks after antigen exposure 
Other studies have been interpreted as showing persistence of influenza virus antigen after infection, and have suggested that this maintains the activation of virus-specific T cells generated in response to infection 
. In contrast to the situation with natural influenza virus infection, previous studies have demonstrated persistence of both vector genome and antigen expression for at least a year following i.m. rAd immunization of mice 
. This sustained antigen expression appears important for supporting the activated phenotype of transgene-specific T cells following rAd immunization 
, although maintenance of memory T cell populations eventually becomes antigen-independent 
. Differences in expression pattern likely explain the dichotomy between the T-cell responses against the transgenes (NP and M2 which are driven by a constitutively active CMV promoter) which were sustained over the duration of the study and responses against the rAd vector (the E1, E3 deleted vector backbone driving only minimal and transient expression of Hexon and DNA-binding protein) which decline substantially over this time. Ultimately, the combination of prolonged low-level antigen presentation and Th1 cytokine production may be optimal for maintaining protective mucosal immune responses.
With their ability to induce potent innate and adaptive immune responses and to deliver antigen to intracellular processing and presentation pathways, rAd vectors may be particularly well suited for vaccination against viruses and other intracellular pathogens. Here we demonstrate their potential as an emergency, fast-acting vaccine inducing long-lasting protective immunity in the respiratory tract. Heterosubtypic rAd vaccines could be stockpiled in advance since regular reformulation would be unnecessary, and could be delivered by nasal spray, facilitating widespread administration by limited healthcare personnel. During a large scale virus outbreak or pandemic, reducing disease severity by vaccination, even while allowing mild infection, could greatly reduce the burden on healthcare facilities. Conserved antigen vaccines do not provide the type of sterilizing immunity mediated by HA-specific neutralizing antibody, but would permit only a mild, transient natural influenza virus infection. This transient infection would further boost heterosubtypic immunity and induce neutralizing antibodies against the exact virus strain circulating in the community, preventing re-infection.