Mucosal Immunization with a Candidate Universal Influenza Vaccine Reduces Virus Transmission in a Mouse Model

ABSTRACT

Pandemic influenza is a major public health concern, but conventional strain-matched vaccines are unavailable early in a pandemic. Candidate “universal” vaccines targeting the viral antigens nucleoprotein (NP) and matrix 2 (M2), which are conserved among all influenza A virus strains and subtypes, could be manufactured in advance for use at the onset of a pandemic. These vaccines do not prevent infection but can reduce disease severity, deaths, and virus titers in the respiratory tract. We hypothesized that such immunization may reduce virus transmission from vaccinated, infected animals. To investigate this hypothesis, we studied mouse models for direct-contact and airborne transmission of H1N1 and H3N2 influenza viruses. We established conditions under which virus transmission occurs and showed that transmission efficiency is determined in part at the level of host susceptibility to infection. Our findings indicate that virus transmission between mice has both airborne and direct-contact components. Finally, we demonstrated that immunization with recombinant adenovirus vectors expressing NP and M2 significantly reduced the transmission of virus to cohoused, unimmunized mice in comparison to controls. These findings have broad implications for the impact of conserved-antigen vaccines, not only in protecting the vaccinated individual but also in protecting others by limiting influenza virus transmission and potentially reducing the size of epidemics.

IMPORTANCE Using a mouse model of influenza A virus transmission, we demonstrate that a candidate “universal” influenza vaccine both protects vaccinated animals from lethal infection and reduces the transmission of virus from vaccinated to nonvaccinated mice. This vaccine induces immunity against proteins conserved among all known influenza A virus strains and subtypes, so it could be used early in a pandemic before conventional strain-matched vaccines are available and could potentially reduce the spread of infection in the community.

INTRODUCTION

The possibility that newly emergent influenza viruses may become efficiently transmissible among humans is a major public health concern. Thus, studies of influenza virus transmission and the development of medical countermeasures, including antiviral drugs and vaccines, are of high priority. Current influenza vaccines generate strain-specific antibodies against the highly variable viral hemagglutinin (HA) and neuraminidase. However, the production of strain-matched vaccines requires several months, limiting their utility against new, rapidly spreading viruses. In contrast, cross-protective vaccines targeting conserved antigens such as nucleoprotein (NP) and matrix 2 (M2) do not rely on strain identification and could be used off the shelf early in an outbreak. They protect animals from lethal challenge with a broad range of influenza A virus strains and subtypes, including H1N1, H3N2, and H5N1 (1–8). While these vaccines do not prevent infection, they significantly reduce morbidity, mortality, and viral titers in the respiratory tract (6, 7). A potential concern about infection-permissive vaccines is that although they would protect the vaccinee, they might not control transmission to others.

A recent epidemiological modeling study suggests that reducing virus transmission could dramatically reduce the size of influenza pandemics or seasonal epidemics and slow the antigenic evolution of seasonal influenza viruses (9). Thus, a critical question to test empirically is whether cross-protective vaccines provide this reduction in transmission.

Animal models for influenza virus transmission include ferrets and guinea pigs (10–13), which efficiently transmit infection either to cohoused animals (direct contact) or across a perforated barrier (either aerosol or respiratory droplet transmission). While experiments with these species have elucidated viral and environmental factors affecting host range and transmission (14–17), studies have been limited by space requirements, expense, and a lack of reagents for immunological studies. For practical reasons, it is generally not possible to use sufficient numbers of ferrets or guinea pigs for statistical analyses in transmission experiments (18).

In contrast, mice have been critical for defining host defenses against influenza viruses, and many immunological reagents are available. A mouse influenza transmission system was developed by Schulman and Kilbourne 50 years ago (19–24). Mice have since been seldom used to study influenza virus transmission, although a recent report described contact-dependent transmission in mice (25). Here we have established a model allowing transmission experiments with H1N1 and H3N2 influenza viruses and accommodating group sizes that permit statistical analyses. We demonstrate that transmission efficiency is in part dependent on the susceptibility of contact animals and show that limiting exposure to the airborne route alone results in a significantly lower frequency of transmission than when mice are cohoused. We then use this model to demonstrate that immunization with a combination of recombinant adenovirus (rAd) vectors expressing NP and M2, a candidate universal influenza vaccine, reduces viral transmission from vaccinated animals to cohoused, unvaccinated animals in the absence of neutralizing antibodies.

(This work was presented in part at the Gordon Research Conference: Biology of Acute Respiratory Infection, Ventura, CA, March 2012.)

MATERIALS AND METHODS

Experimental animals and animal housing.

CFW mice [Crl:CFW(SW)] were obtained from Charles River Laboratories, DBA/2J mice were purchased from the Jackson Laboratory (Bar Harbor, ME), and BALB/cAnNCR mice were purchased from the National Cancer Institute (Frederick, MD). All animals were females aged 8 to 15 weeks at the time of use. For experiments using mechanically ventilated housing, mice were housed 5 per cage in standard shoebox cages with microisolator lids (191 by 292 by 127 mm) in a Smart Bio-Pak rack (model SB4100; Allentown Caging Co., Allentown, NJ), with water supplied by an automated system and a positive-pressure ventilation rate set at a factory-defined 60 air changes per hour. For nonventilated housing, mice were housed in the same-sized shoebox cages with water bottles and microisolator lids on an open-sided shelf with no exhaust system, just room ventilation.

For experiments examining airborne transmission (defined here as transmission by either aerosols or droplets), contact and donor mice were separated by using dividers manufactured to our specifications by Ancare (Bellmore, NY) and further modified to ensure a tight fit to the sides of the cage (D. P. Cook Construction, Gaithersburg, MD). Each divider consisted of two perforated stainless steel sheets held 5 mm apart by spacers, with angled flanges at either end of the divider to prevent direct contact between mice at the corners. Each sheet had 532 perforations, 3.5 mm in diameter, and 71 perforations, 7 mm in diameter, allowing air to exchange between the two compartments. The 7-mm-diameter perforations were arranged so as to be inaccessible to mice when the cages were assembled. The divider runs the length of a standard mouse cage, separating it into two compartments of ∼90 by 292 mm. Dividers were held in place with a wire grid (Ancare) providing water bottles and feed for each of the compartments. Cages were topped with standard microisolator lids and further secured with elastic bands.

As a humane endpoint, mice losing >25% of their body weight were euthanized. All experiments used standard mouse bedding (Paperchip; Shepherd Specialty Papers, Watertown, TN), which was not changed during the transmission period. Food and water were provided ad libitum. All animal experiments were conducted under temperature- and humidity-controlled (mean daily values of 22.3°C ± 0.1°C and 51.02% ± 0.1% humidity) animal biosafety level 2 (ABSL-2) conditions in animal facilities accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. All animal protocols and procedures were approved by Institutional Animal Care and Use Committees at the Center for Biologics Evaluation and Research.

Influenza viruses.

Mouse-adapted A/FM/1/47-ma (H1N1) (A/FM), originally provided by Earl Brown (University of Ottawa, Canada) (26); the mouse-adapted A/Philippines/2/82 × A/PR/8/34 (H3N2) reassortant (X-79) (27); A/PR/8/34 (H1N1) (PR8); and A/Udorn/307/72 (H3N2) (A/Udorn) were prepared as previously described (28). Mice were infected intranasally (i.n.) under isoflurane anesthesia in a 50-μl volume with a dose of 10⁴ 50% tissue culture infectious doses (TCID₅₀), unless stated otherwise.

rAd vaccines.

The production and titration of recombinant adenovirus (rAd) vectors expressing influenza A virus nucleoprotein (NP), influenza B virus nucleoprotein (B/NP), or consensus M2 were described previously (3, 4). For immunization experiments, a dose of 5 × 10⁹ particles each of NP-rAd and M2-rAd or 1 × 10¹⁰ particles of B/NP-rAd in a 50-μl volume was given i.n. under isoflurane anesthesia.

Lung sampling, BAL, and nasal washing.

Mice were euthanized with ketamine-xylazine. Nasal washes were obtained by inserting an intravenous catheter (24 gauge by 3/4 in.; Terumo Medical Corp., Elkton, MD) into the nasopharynx via a slit in the trachea and injecting phosphate-buffered saline (PBS) into the nasal cavity until 200 μl of flowthrough was collected from the nares. Lungs were then harvested. To prevent cross-contamination at the harvest stage, contact animals in each cage were dissected before donor animals. Instruments were cleaned and sterilized by immersion of contact surfaces in 70% ethanol between use on each animal. Mucosal sampling for lung cells and bronchoalveolar lavage (BAL) fluid was otherwise performed as described previously (6).

Virologic analyses.

Virus titers in lung tissues and nasal wash samples were determined by a TCID₅₀ assay on MDCK cells (29), read out by hemagglutination with 0.5% chicken erythrocytes as described previously (6). Titers were calculated by using the method of moving averages and Weil's tables (30, 31). Assay limits of detection were 10^2.68 TCID₅₀/organ and 10^2.19 TCID₅₀/ml for lung and nasal wash specimens, respectively. Contact animals were considered positive for virus transmission if virus was detected by a TCID₅₀ assay in either lung homogenates, nasal wash specimens, or both.

MID₅₀ determination.

Groups of 5 mice were intranasally inoculated with each dilution of virus in a 50-μl volume under isoflurane anesthesia and euthanized 3 days later for collection of lung and nasal wash samples. Animals were considered infected if virus was detected by a TCID₅₀ assay in either lung homogenates, nasal wash specimens, or both. The 50% mouse infectious doses (MID₅₀) with 95% confidence intervals were calculated by using the method of moving averages, as described above.

Immunologic analyses.

Enzyme-linked immunosorbent assays (ELISAs) for antibody against the M2 ectodomain (M2e) and recombinant NP were previously described (5, 6). Quantitation of gamma interferon (IFN-γ)-secreting cells by an enzyme-linked immunosorbent spot (ELISPOT) assay was performed as described previously (4), except that peptide pools covering the M2 and NP sequences of PR8 were used. The M2 peptide pool was a series of 13- to 15-mer peptides with an 11-amino-acid overlap. The NP peptides (15- to 17-mer with an 11- to 13-amino-acid overlap) were divided into 3 pools of 27 to 29 peptides each. Peptides were synthesized by Genscript (Piscataway, NJ).

Hemagglutination inhibition assay.

Sera were incubated overnight at 37°C with 3 volumes of receptor-destroying enzyme (RDE; Denka Seiken Co. Ltd., Tokyo, Japan) and heat inactivated at 56°C for 30 min before the addition of 6 volumes of 0.85% saline. Twofold dilutions of sera were made in microtiter plates for a final volume of 25 μl, and an equal volume of A/FM diluted to 4 hemagglutinating units (HAU)/25 μl was added to each well and mixed thoroughly. One hour later, 50 μl of 0.5% chicken erythrocytes was added and mixed well. Plates were scored after 45 min, and hemagglutination inhibition (HAI) titers were calculated. Polyclonal anti-A/Fort Monmouth/1/47 (H1N1) antiserum (NR-3099), obtained through the Biodefense and Emerging Infections Research Resources Repository, NIAID, NIH, was used as a positive control.

Microneutralization assay.

Serial 2-fold dilutions of RDE-treated sera in PBS were made in a 96-well U-bottom plate. An equal volume (20 μl) of A/FM at 1 × 10⁴ TCID₅₀/ml was added to each well, gently mixed, and incubated at 37°C. After 2 h, 20 μl of the virus-antibody mixture was transferred onto plates of MDCK cells plus virus growth medium. Plates were then incubated for 48 h and read out by hemagglutination with 0.5% chicken erythrocytes.

Statistical analyses.

SigmaPlot 12 (Systat Software, Point Richmond, CA) was used for statistical analyses as follows: survival analysis was done by using the log rank method, virus titer comparisons were done by using the t test, and transmission rate comparisons were performed by using the chi-square test.

RESULTS

Establishment of the transmission system.

In preliminary experiments, CFW outbred mice (as used by Schulman and Kilbourne [21]) were infected with a virulent mouse-adapted influenza virus H1N1 strain, A/FM. Twenty-four hours later, 2 infected donor mice were placed into the same cage as 3 uninfected contact animals. Three days later, all mice were euthanized, and lungs and nasal wash specimens were collected for virus titration. During the contact period, donors but not contacts progressively lost weight. While all donors had virus recoverable from lungs and most had virus recovered from nasal wash specimens, A/FM was detectable in only 2/18 (11%) contacts when cages were held on a mechanically ventilated cage rack (data not shown), but this rose to 6/18 (33%) contacts when cages were housed on an open-sided shelf with room ventilation only (Table 1). There was no obvious correlation in either experiment between individual donor lung or nasal wash virus titers on day 4 and the number of cohoused contacts that became infected. This suggests that virus shedding from donors earlier than day 4 is more important for virus transmission and/or that other factors are involved. Because of the higher rate of transmission observed, all further experiments used cages housed under room ventilation conditions.

TABLE 1.

Effect of mouse strain on A/FM transmission

Donor mouse strain	No. of infected donors/no. of donors surviving to sampling (mean titer ± SEM)a		Contact mouse strain	No. of infected contacts/no. of contacts surviving to sampling (mean titer ± SEM)
	Lung (log10 TCID50/organ)	Nasal wash (log10 TCID50/ml)		Lung (log10 TCID50/organ)	Nasal wash (log10 TCID50/ml)	Total (%)
CFWb	12/12 (6.83 ± 0.21)	8/11c (3.10 ± 0.28)	CFW	6/18 (4.38 ± 0.39)	2/18 (2.70 ± 0.25)	6/18 (33.3)
DBA/2Jb	NT	NT	DBA/2J	5/18 (5.73 ± 0.20)	0/18	5/18 (27.7)
BALB/cb	11/11 (7.78 ± 0.15)	11/11 (4.51 ± 0.30)	BALB/c	0/18	0/18	0/18 (0)
BALB/cd	10/10 (8.01 ± 0.09)	10/10 (3.68 ± 0.23)	CFW	12/18 (7.22 ± 0.19)	8/18 (3.48 ± 0.26)	12/18f (66.7)
CFWd	4/4 (8.18 ± 0.23)	4/4 (4.38 ± 0.12)	BALB/c	2/18 (5.43 ± 1.00)	0/18	2/18f (11.1)
DBA/2Je	NT	NT	CFW	8/18 (5.71 ± 0.23)	1/18 (4.69)	8/18 (44.4)
CFWe	6/6 (6.85 ± 0.08)	6/6 (4.98 ± 0.42)	DBA/2J	8/18 (5.81 ± 0.39)	3/18 (3.86 ± 0.17)	8/18 (44.4)

^aDonor groups initially consisted of 12 animals; denominators show the number of donors surviving to sampling at 4 days postinfection/3 days postcontact. NT, not tested; no DBA/2J donors survived to day 4 postinfection.

^bExperiments were conducted on separate occasions.

^cNasal wash specimens could not be collected from one animal for technical reasons.

^dExperiments were conducted simultaneously.

^eExperiments were conducted simultaneously.

^fSignificantly different (P = 0.002).

In a further experiment using CFW mice, contacts were monitored for weight loss and survival. All A/FM-infected donors died at between days 3 and 5 (Fig. 1). Significant weight loss was seen in 2 contacts (both in the same cage), but all survived. At 1 month postcontact, 4/18 contacts (including the 2 that lost weight) had A/FM-specific antibody responses with HAI titers of 160 to 640. Using seroconversion as a marker for transmission, only half the transmitted infections caused clinically apparent disease, and insufficient virus was transmitted to initiate a lethal infection. This experiment also suggests that secondary transmission (i.e., between infected and uninfected contacts) is either inefficient or absent in the case of A/FM.

FIG 1.

Weight loss and seroconversion of contact mice. Donor CFW mice (8 weeks old) were infected with 10⁴ TCID₅₀ of A/FM, and 24 h later, 2 infected (donor) mice were placed into the same cage as 3 uninfected (contact) mice. Cages were held on open-sided racks, and mice were monitored daily for body weight. At 1 month postcontact, all surviving animals were bled, and individual serum samples were assessed by an HAI assay. Each panel represents one cage. Open symbols denote donor mice; closed symbols indicate contacts. HAI-positive contacts are indicated by gray symbols, with HAI titers indicated for each positive animal.

Effect of mouse and virus strains.

As different mouse strains vary in their susceptibility to influenza viruses, we extended our studies to investigate direct-contact transmission in two additional mouse strains: DBA/2J, which is highly susceptible to lethal influenza virus infection, and BALB/c, which is relatively resistant (32, 33). Transmission experiments in these two mouse strains using 10⁴ TCID₅₀ of A/FM were conducted as described above. Transmission of A/FM between DBA/2J mice occurred to a similar extent as that seen between CFW mice, but transmission between BALB/c mice was not seen (Table 1).

To investigate whether the lack of transmission of A/FM in BALB/c mice was due to decreased susceptibility to infection or because of an impaired ability of BALB/c mice to transmit the virus, a “crisscross” transmission experiment was performed, in which BALB/c mice were used as donors with CFW mice as recipients, and vice versa. Limited transmission of A/FM was seen for CFW donors with BALB/c contacts, but significantly higher levels of transmission were seen for BALB/c donors with CFW contacts (Table 1). A similar experiment examining transmission between the two susceptible stains DBA/2J and CFW showed similar rates of transmission with both donor/contact combinations (Table 1). Taken together, these findings suggest that the establishment of infection in contact animals is dependent at least in part on the relative sensitivity of the contact animal. However, donor traits also seem to be involved, as transmission of A/FM from BALB/c donors to CFW contacts was more efficient than transmission from CFW donors to CFW contacts. Interestingly, virus titers in lungs and nasal wash specimens of infected BALB/c and CFW mice were very similar over the first 4 days of infection (data not shown).

Limited transmission (with 2/18 contacts becoming infected [data not shown]) between BALB/c mice occurred for two different mouse-adapted viruses (PR8 and X-79). These viruses were also inefficiently transmitted between CFW mice (with 1/18 contacts becoming infected [data not shown]).

Taken in combination, our data suggest that transmission efficiency is multifactorial and dependent on the sensitivity of contacts, donor characteristics, and biological properties of the specific virus strain.

Mode of transmission.

To investigate whether transmission was via a direct-contact-dependent route or via the airborne route (by either droplet or aerosol), we developed a removable perforated divider running the length of the cage, as described in Materials and Methods. This prevents direct contact between donor and contact mice while permitting air exchange between the two cage compartments. Using CFW mice and the non-mouse-adapted strain A/Udorn, which has been reported to efficiently transmit between BALB/c mice by direct contact but not via the airborne route (25), we found that direct contact resulted in efficient transmission (experiment 1) (Table 2). However, when donor and contact mice were physically separated by the perforated divider, a significantly smaller proportion of contacts became infected (approximately 6-fold; P < 0.001 versus direct contact). A similar experiment with A/FM resulted in one-third of animals in direct contact with donors becoming infected, compared to only 1 of 26 contacts that were physically separated from the donors (data not shown).

TABLE 2.

Direct contact versus airborne transmission of A/Udorn

Caging	Donor mouse strain	No. of infected donors/no. of donors surviving to sampling (mean titer ± SEM)		Contact mouse strain	No. of infected contacts/no. of contacts surviving to sampling (mean titer ± SEM)
		Lung (log10 TCID50/organ)	Nasal wash (log10 TCID50/ml)		Lung (log10 TCID50/organ)	Nasal wash (log10 TCID50/ml)	Total (%)
Expt 1
Direct	CFW	20/20 (6.73 ± 0.09)	18/20 (3.37 ± 0.13)	CFW	13/30a (6.12 ± 0.37)	24/30b (4.10 ± 0.23)	25/30 (83.3)c
Airborne	CFW	20/20 (6.77 ± 0.10)	20/20 (3.54 ± 0.12)	CFW	4/30a (6.45 ± 0.45)	3/30b (3.78 ± 0.55)	4/30 (13.3)c
Expt 2
Direct	BALB/c	20/20 (6.64 ± 0.06)	20/20 (3.53 ± 0.11)	CFW	15/30 (5.24 ± 0.44)	26/30d (3.91 ± 0.18)	28/30 (93.3)e
Airborne	BALB/c	20/20 (6.63 ± 0.07)	20/20 (3.54 ± 0.14)	CFW	7/30 (5.86 ± 0.64)	8/30d (3.47 ± 0.31)	9/30 (30)e

^aSignificantly different (P = 0.022).

^bSignificantly different (P < 0.001).

^cSignificantly different (P < 0.001).

^dSignificantly different (P < 0.001).

^eSignificantly different (P < 0.001).

Because higher levels of A/FM transmission were seen with BALB/c donors and CFW contacts, a comparison of direct-contact versus airborne transmission was performed as described above, using this donor/contact combination with A/Udorn. This resulted in more effective transmission by both routes, with 93.3% of CFW contacts becoming infected by direct contact and 30% becoming infected when the donors and contacts were physically separated (experiment 2) (Table 2). Again, the proportion of infected contacts was significantly lower when physical contact was prevented (P < 0.001). Thus, these data consistently indicate that when a perforated barrier limits virus transmission between mice to the airborne route alone, transmission is significantly less efficient than when mice are cohoused and transmission could result from direct (including nose-nose or nose-fomite) contact and/or via the airborne route.

To estimate the approximate quantity of virus required to establish an infection in each mouse strain, 50% mouse infectious dose (MID₅₀) titrations were performed for A/FM in CFW, DBA/2J, and BALB/c mice and for A/Udorn in CFW and BALB/c mice (Table 3). Interestingly, while the MID₅₀ for A/FM in CFW mice was found to be 10-fold lower than that in BALB/c mice and 4-fold lower than that in DBA/2J mice, the MID₅₀s for A/Udorn were broadly similar in CFW and BALB/c mice.

TABLE 3.

MID₅₀ determination

Mouse strain	MID50 (95% CI)a
	A/FM	A/Udorn
CFW	0.079 (0.34–0.02)	2.37 (4.87–1.16)
BALB/c	0.79 (3.4–0.19)	5.01 (7.94–3.16)
DBA/2J	0.32 (1.16–0.09)	NT

^aThe MID₅₀ for each mouse strain/virus is reported as the TCID₅₀, with 95% confidence intervals (CI) shown in parentheses. NT, not tested.

Conserved-antigen vaccines protect CFW mice and reduce virus transmission.

Intranasal immunization with a mixture of recombinant adenoviruses expressing NP and M2 confers robust protection from challenge with multiple subtypes of influenza A virus, lasting at least 10 months in BALB/c mice (7). NP is a target for helper and cytotoxic T cells in mice and humans (34, 35), and NP-specific antibodies may also contribute to protection (36). M2-specific T cells contribute to protection in mice (4), and M2-specific antibodies limit viral spread and mediate antibody-dependent cellular cytotoxicity (reviewed in reference 37). We wanted to confirm the effectiveness of this vaccination in CFW mice, since there are genetic differences in vaccine responsiveness between mouse strains, especially for M2 (38). CFW mice were immunized once intranasally with NP+M2-rAd or with B/NP-rAd as a specificity control. Antibody and T cell responses were assessed 1 month later.

Weak serum IgG responses against M2e were seen in NP+M2-rAd-immunized CFW mice compared with BALB/c mice (strong responders to M2), with no detectable M2e-specific IgG in the BAL fluid (Fig. 2A). Strong IgG responses against NP were observed (Fig. 2B). No IgA against M2e was observed, and IgA levels against NP were low and detectable only in BAL fluid (data not shown). As expected, no neutralizing antibody responses against A/FM were seen in serum from mice immunized with NP+M2-rAd or B/NP-rAd or naive mice (Fig. 2C). Strong T cell responses to NP but not to M2 were observed in lungs of all NP+M2-rAd-immunized mice by an IFN-γ ELISPOT assay (Fig. 2D). Individual animals varied greatly in their responses to the 3 NP peptide pools, presumably reflecting the outbred nature of CFW mice. B/NP-rAd-immunized or naive mice did not respond to NP or M2 peptides.

FIG 2.

rAd immunization induces mucosal immune responses and protects CFW mice from A/FM challenge. (A and B) Antibody responses against M2e (A) and NP (B) were assessed by an ELISA 1 month after immunization of CFW mice (n = 4 to 5) with NP+M2-rAd or B/NP-rAd or in naive mice. Serum from NP+M2-rAd-immunized BALB/c mice was used as a control. Results show means ± standard errors of the means. (C) Neutralizing antibody responses against A/FM were determined by a microneutralization assay on sera from immunized (n = 10), naive (n = 4), and A/FM convalescent (n = 3) CFW mice; the positive control was polyclonal anti-A/FM serum, and the negative control was PBS. Bars show means ± standard errors of the means. The dashed line shows the limit of detection of the assay. (D) An IFN-γ ELISPOT assay was used to assess lung T cell responses in CFW mice immunized with NP+M2-rAd (n = 5) or B/NP-rAd (n = 5) or in naive animals (n = 4). Results following restimulation with the indicated peptide pools are shown. Bars show the mean total numbers of cells per organ for each animal ± standard errors of the means. (E and F) Survival (E) and weight loss (F) following challenge of CFW mice (n = 10) with 10⁴ TCID₅₀ A/FM at 1 month postimmunization with NP+M2-rAd or B/NP-rAd. Weight loss curves show means ± standard errors of the means. After day 4, values for B/NP-rAd do not represent a valid group mean, since some animals had died. The one survivor regained weight.

One month after immunization, CFW mice were challenged with A/FM (Fig. 2E and F). All NP+M2-rAd-immunized mice survived, compared to 10% survival in the B/NP-rAd group. Additionally, the B/NP-rAd group had greater weight loss. In parallel groups of mice, virus was eliminated from nasal wash specimens of NP+M2-rAd-immunized animals within 5 days and from the lung at between 7 and 9 days postchallenge (data not shown). Importantly, protection was observed despite both a weak M2-specific response in CFW mice compared to the response reported in previous studies with BALB/c mice and the much greater susceptibility of CFW mice to influenza virus. The dose of A/FM used in these experiments (10⁴ TCID₅₀/mouse) corresponds to 100 50% lethal doses (LD₅₀) in BALB/c mice but 2,000 LD₅₀ in CFW mice (data not shown).

We next addressed the key question of whether the reduction in virus shedding by vaccination against conserved antigens reduces transmission. CFW donors were immunized intranasally with NP+M2-rAd or B/NP-rAd and challenged with A/FM 1 month later. The following day, 2 donors were placed into cages with 3 naive CFW contact animals. Direct-contact conditions were chosen to maximize the potential for transmission. Lung and nasal wash virus titers were assessed 4 days after infection (3 days after contact), as described above. Transmission to contacts was seen for a substantial fraction of B/NP-rAd-immunized donors (50.0% and 40.0% for experiments 3 and 4, respectively) (Table 4). Strikingly, transmission from NP+M2-rAd-immunized donors occurred at much lower rates (16.7% and 6.8%). This represents a large reduction in transmission (66.7% and 83.3% reduction for experiments 3 and 4, respectively) compared to B/NP-rAd-immunized mice, which is statistically significant in each case. Individual lung and nasal wash virus titers from experiment 4 are shown in Fig. 3.

TABLE 4.

Immunization against conserved antigens reduces transmission of A/FM from infected, vaccinated animals to unvaccinated contacts

Donor immunization	No. of infected donors/no. of donors surviving to sampling (mean titer ± SEM)a		No. of infected contacts/no. of contacts surviving to sampling (mean titer ± SEM)			% reduction in transmission (P value determined by chi-square test)
	Lung (log10 TCID50/organ)	Nasal wash (log10 TCID50/ml)	Lung (log10 TCID50/organ)	Nasal wash (log10 TCID50/ml)	Total (%)
Expt 3b
B/NP-rAd	15/15 (7.66 ± 0.09)	15/15 (4.15 ± 0.20)	14/36 (5.47 ± 0.39)	7/36 (3.29 ± 0.38)	18/36 (50.0)
A/NP+M2-rAd	22/22 (7.01 ± 0.15)	13/22 (3.13 ± 0.20)	6/36 (6.35 ± 0.17)	2/36 (3.20 ± 0.25)	6/36 (16.7)	66.7 (0.006)
Expt 4c
B/NP-rAd	23/23 (7.95 ± 0.07)	20/23 (3.59 ± 0.26)	17/45 (5.99 ± 0.35)	6/45 (3.61 ± 0.40)	18/45 (40.0)
A/NP+M2-rAd	28/29 (6.53 ± 0.18)	2/29 (2.82 ± 0.37)	2/44d (4.68 ± 1.75)	1/44d (2.95 ± 0.00)	3/44 (6.8)	83.3 (<0.001)

^aDenominators show the number of donors surviving to sampling at 4 days postinfection/3 days postcontact.

^bGroups initially consisted of 24 donors and 36 contacts.

^cGroups initially consisted of 30 donors and 45 contacts.

^dOne contact mouse was euthanized prior to sampling for humane reasons unrelated to infection.

FIG 3.

Virus titers from donor and contact mice in transmission reduction experiment 4. Shown are nasal wash (A and C) and lung (B and D) virus titers in donor (open bars) and contact (filled bars) CFW mice at 3 days postcontact and 4 days post-donor infection with 10⁴ TCID₅₀ A/FM. Donor mice were immunized with NP+M2-rAd (A and B) or B/NP-rAd (C and D) 1 month prior to challenge. Cages were housed in unventilated racks. Each panel represents one cage (2 donors plus 3 contacts). Gray bars show infected contact animals. Dashed lines indicate the limit of detection. DEAD indicates that the donor was dead at day 4 and not sampled. SAC indicates 1 contact that was euthanized for reasons unrelated to infection. These data are from experiment 4 (Table 3).

Virus replication kinetics were examined in parallel groups of mice. No significant differences in lung virus titers between NP+M2-rAd- and B/NP-rAd-immunized groups were observed until day 4 postchallenge (Fig. 4A and C). Nasal wash virus titers were significantly lower in the NP+M2-rAd-immunized group at earlier times: from day 2 in experiment 3 and from day 3 in experiment 4 (Fig. 4B and D). These differences in nasal shedding at early times may be related to the differences in transmission.

FIG 4.

Effect of rAd immunization on nasal and lung challenge virus titers. CFW mice (n ≥ 5) were immunized with NP+M2-rAd (black bars) or B/NP-rAd (gray bars) and challenged with 10⁴ TCID₅₀ A/FM 1 month later. Virus titers were measured in lungs (A and C) and nasal wash specimens (B and D) at days 1, 2, 3, and 4 postchallenge in 2 independent experiments (A and B, experiment 3; C and D, experiment 4). Dashed lines show the limit of detection of the assay. Error bars indicate standard errors of the means. Statistical significance was assessed by a t test.

DISCUSSION

Studies of candidate “universal” influenza vaccines targeting conserved influenza virus antigens have demonstrated that such vaccines do not prevent infection but can protect from lethal infection and reduce virus titers in the respiratory tract. We hypothesized that these reductions in virus titers could limit virus transmission from vaccinated, infected animals to nonvaccinated contacts, which could greatly magnify the public health impact of these vaccines. Investigation of this hypothesis using ferrets or guinea pigs would have been impractical, so we decided to address this question about conserved-antigen vaccines in a mouse transmission model that would be amenable to statistical analyses.

Recent attempts to develop mouse models for influenza virus transmission have met with various levels of success. One study in BALB/c mice failed to demonstrate direct-contact transmission of a range of influenza viruses, including the reconstructed 1918 pandemic virus and the virulent A/Vietnam/1203/2004 (H5N1) strain (11). However, other studies reported direct-contact transmission for H5N1 and H7N7 avian influenza virus strains in BALB/c mice (39) and for H9N2 viruses in outbred Kunming mice (40). Previous results showed that different influenza virus strains vary in transmissibility in mice, with H2N2 strains being more transmissible than H1N1 strains (24). More recently, Edenborough et al. showed that human H3N2 strains (including A/Udorn) but not H1N1 strains were transmissible in BALB/c mice (25). Interestingly, an H3N1 reassortant was transmitted, suggesting an important role for HA as a determinant of transmissibility (25). In our direct-contact experiments, we found no transmission of H1N1 strain A/FM among BALB/c mice and limited transmission of PR8 and a PR8-based H3N2 reassortant (X-79) in both BALB/c and CFW mice. However, effective direct-contact transmission of A/FM and A/Udorn was seen in CFW mice. The present study shows that the efficiency of transmission is determined not only by the transmitting animal and virus strain but also at least in part by the susceptibility of the contact animal, with BALB/c mice being less susceptible to A/FM infection than either DBA2/J or CFW mice (Table 3).

Experimentally determined MID₅₀ values are useful for comparing mouse strains, but the amount of virus sufficient to establish an infection when given under anesthesia may differ from that needed to initiate an infection by natural exposure. Factors such as aerosol versus intranasal bolus administration, the volume of the inoculum, and the anesthetic used affect sites of virus deposition in the respiratory tract and can have dramatic effects on viral pathogenesis, LD₅₀, and MID₅₀ (41–45).

Despite virus titers in the lungs and nasal wash specimens of infected contacts being much greater than the experimentally determined MID₅₀ values, secondary transmission (i.e., from one infected contact to other contact animals) was not seen. A possible explanation for this lies in the kinetics of transmissibility. As the peak period of transmission from donor mice is 24 to 48 h postinfection (20, 25), infected contact mice might not reach peak transmissibility for at least another 24 to 48 h. In experiments terminated 3 days after the initiation of contact, there may have been insufficient time for secondary transmission between contact animals to occur. In addition, the limit of detection in the TCID₅₀ assay is substantially greater than the experimentally determined MID₅₀ values for the viruses used (Table 3), suggesting that titers in mice recently infected with low virus doses may not be detected by the TCID₅₀ assay. However, the frequency of A/FM transmission was similar when assessed by seroconversion at 1 month (Fig. 1) or by virus titration at 3 days postcontact (Table 1). Thus, it appears that neither early harvest nor the TCID₅₀ assay results in missed transmission events. Overall, these data indicate that secondary transmission is a rare event in mice under these conditions. The reasons for this are unclear.

The ability of influenza viruses to be transmitted by the airborne route (via either respiratory droplets of ≥20 μm or bioaerosol particles of ≤5 μm) is thought to be important for the rapid spread of infection in humans. Unlike humans and ferrets, infected mice do not sneeze, but sneezing may not be critical for airborne transmission. Infected guinea pigs also do not sneeze, despite their capacity for efficient airborne transmission (11). Virus is recoverable from infected ferrets during both sneezing and normal breathing (13), and virus has been isolated from the air surrounding infected mice during the window of infectivity (22). Although sneezing produces copious aerosol, normal breathing may account for more of the total daily bioaerosol produced (46). In human influenza patients, viral RNA was detected in exhaled breath in the absence of coughing, with >99% of exhaled particles from these patients being <5 μm in diameter (47), which is small enough to penetrate the alveoli (48). While we were unable to distinguish between droplet and aerosol transmission, our results suggest that some form of airborne transmission between mice occurred in our experiments. First, the lower rate of transmission in mechanically ventilated cages is suggestive of an airborne component to virus transmission, although alternative explanations (such as increased desiccation of virus-laden fomites) are also possible. Second, experiments where donors were separated from contacts by a perforated barrier showed virus transmission (Table 2 and data not shown) but at a significantly reduced frequency compared with animals in direct contact.

This relatively inefficient airborne transmission contrasts with studies by Schulman and Kilbourne (19, 22). However, these early studies infected donors via whole-body aerosol exposure, depositing virus onto the fur and potentially contributing to a more efficient spread of virus via contaminated fomites, and were conducted in an exposure chamber constructed from a modified autoclave where air was pumped through the chamber (19). In agreement with our findings in mechanically ventilated cages, high-velocity airflow decreased the rate of transmission seen in these previous studies, but low-velocity forced airflow may have carried virus from infected to contact animals, possibly increasing the likelihood of airborne transmission.

An additional factor which may impact the efficiency of airborne transmission is humidity. Previous studies have demonstrated that high humidity decreases the rate of transmission (14, 19). Although the cages in our study were held in a temperature- and humidity-controlled animal room, we were unable to monitor the humidity within individual cages. However, it would be expected that the limited air circulation in unventilated cages would result a more humid environment and, thus, decreased transmission.

The relative importance of direct-contact versus airborne transmission in humans remains unclear. A recent attempt to correlate nasal virus titers with virus recovered from fomites and the air surrounding naturally infected patients found virus detectable by PCR in air samples (49). A higher rate of detection was seen in air samples collected closer (∼3 ft) to infected individuals, suggesting that aerosol transmission over a short range is possible (48, 49), but virus was rarely detected on swabbed fomites (49). This contrasts with the more frequent detection on fomites in studies with experimentally infected adult volunteers (50) or naturally infected children in child care centers and households (51). Influenza viruses remain viable for several days when dried onto nonporous surfaces such as glass, stainless steel, or plastic (52, 53) but survive for shorter periods on porous objects such as tissues and fabric (52, 54) or on contaminated hands (54). Because both airborne and nose-nose/nose-fomite transmission are possible in cohoused (direct-contact) animals, it is not possible to precisely assign the contribution of each route of transmission in these experiments.

Regardless of the route of transmission, the emergence of new influenza viruses able to spread in a population remains a major public health challenge. Interventions to minimize the transmission of newly emergent influenza viruses include antiviral drugs and social-distancing measures (55). However, these are expensive and short-term solutions with practical difficulties. Immunization is likely more feasible. Current strain-matched vaccines block transmission by preventing infection of hosts but would be unavailable at the outset of a pandemic. For example, during the 2009 pandemic, vaccine was not available until 6 months after the identification of the virus strain, by which time the pandemic wave had peaked (56). In contrast, conserved-antigen vaccines require no advance prediction of which virus strains will circulate and could be available for use early in a pandemic. Conserved-antigen vaccines are therefore attractive for the control of unexpected outbreaks or pandemic influenza, and several are currently in clinical trials (57).

One perceived problem with cross-protective vaccines is that although they reduce virus titers in the respiratory tract and provide protection from severe consequences of infection, they would not necessarily prevent infection of vaccinated individuals who might then spread infection to others. Therefore, it is critical to assess whether immunity directed against conserved antigens is able to limit transmission from infected individuals. Previous studies have suggested that this may be the case, as the transmission rate is lower from animals that have been previously infected with a virus bearing heterologous surface antigens (23, 25, 58). Here we have used the mouse transmission model to demonstrate that a conserved-antigen vaccine inducing immunity against NP and M2 in the absence of neutralizing antibodies is sufficient both to protect vaccinees and to reduce the further spread of disease to others not vaccinated (Table 4).

The results are particularly impressive as CFW mice make relatively poor M2-specific antibody responses compared to those seen in high responders such as BALB/c mice. Indeed, in NP+M2-rAd-immunized BALB/c mice, lung virus titers are ∼2 logs lower than those in B/NP-rAd-immunized controls as early as day 1 following a similar A/FM challenge (7), whereas in NP+M2-rAd-vaccinated CFW mice, lung titers did not drop significantly until day 4 after challenge (Fig. 4). Regardless, a significant reduction in transmission from NP+M2-rAd-immunized CFW mice was seen, despite relatively small reductions in lung and nasal wash virus titers (∼1 to 1.5 logs lower than those in controls) (Fig. 4). Reducing the duration of virus shedding may be a critical goal for vaccination. In both mouse and ferret models, vaccination against NP and M2 results in accelerated clearance of challenge virus compared to controls (6, 7, 59), further supporting the potential for conserved-antigen vaccines to limit influenza virus transmission.

Mathematical modeling suggests that a vaccine able to limit influenza virus transmission from vaccinated humans to a similar degree to that reported here could significantly mitigate the size of an influenza pandemic, even under conditions of incomplete vaccination coverage (9). It might be possible to enhance the transmission reduction effect of cross-protective vaccines still further by targeted vaccination of populations with limited prior exposure to influenza virus. Relatively low transmission rates (with 25% of contacts becoming infected) have been observed in an adult human volunteer study (50), possibly due to cross-reactive cellular immune responses, which have been shown to reduce virus shedding in human challenge experiments (9, 60–62). However, children shed influenza virus at higher titers and for a longer period than adults (63), perhaps due to their lack of cross-protective immunity from limited prior experience of influenza virus infection. Although annual vaccination of children with strain-matched vaccines would prevent infection via neutralizing antibodies, this may hamper the development of cross-protective responses that could reduce the level and duration of shedding in the event of a pandemic for which strain-matched vaccines are unavailable (64). Vaccines able to provide cross-protective immune responses, perhaps used in addition to strain-matched vaccines, may thus be able fill this public health gap.

In summary, we have developed mouse models for both direct-contact and airborne influenza virus transmission and used them to address one of the key questions about the potential impact of universal influenza vaccines that do not generate neutralizing antibodies. Although such vaccines permit some degree of infection, our results demonstrate that NP+M2 vaccination not only protects the recipient against morbidity and mortality but also greatly reduces the transmission of infection to others. Thus, cross-protective vaccines may provide benefit to both vaccinees and the nonvaccinated population.