|Home | About | Journals | Submit | Contact Us | Français|
New approaches for vaccination to prevent influenza virus infection are needed. Emerging viruses, such as the H5N1 highly pathogenic avian influenza (HPAI) virus, pose not only pandemic threats but also challenges in vaccine development and production. Parainfluenza virus 5 (PIV5) is an appealing vector for vaccine development, and we have previously shown that intranasal immunization with PIV5 expressing the hemagglutinin from influenza virus was protective against influenza virus challenge (S. M. Tompkins, Y. Lin, G. P. Leser, K. A. Kramer, D. L. Haas, E. W. Howerth, J. Xu, M. J. Kennett, J. E. Durbin, R. A. Tripp, R. A. Lamb, and B. He, Virology 362:139–150, 2007). While intranasal immunization is an appealing approach, PIV5 may have the potential to be utilized in other formats, prompting us to test the efficacy of rPIV5-H5, which encodes the HA from H5N1 HPAI virus, in different vaccination schemes. In the BALB/c mouse model, a single intramuscular or intranasal immunization with a live rPIV5-H5 (ZL46) rapidly induced robust neutralizing serum antibody responses and protected against HPAI challenge, although mucosal IgA responses primed by intranasal immunization more effectively controlled virus replication in the lung. The rPIV5-H5 vaccine incorporated the H5 HA into the virion, so we tested the efficacy of an inactivated form of the vaccine. Inactivated rPIV5-H5 primed neutralizing serum antibody responses and controlled H5N1 virus replication; however, similar to other H5 antigen vaccines, it required a booster immunization to prime protective immune responses. Taken together, these results suggest that rPIV5-HA vaccines and H5-specific vaccines in particular can be utilized in multiple formats and by multiple routes of administration. This could avoid potential contraindications based on intranasal administration alone and provide opportunities for broader applications with the use of a single vaccine vector.
Influenza virus is a negative-sense, segmented RNA virus in the family Orthomyxoviridae. It is classified into subtypes based on the major antigenic surface glycoproteins, hemagglutinin (HA) and neuraminidase (NA). Thus far, there are 17 different HA subtypes and 9 different NA subtypes (1, 2), all containing segments of avian origin (3). Influenza virus has the capacity to reassort, whereby gene segments are exchanged, creating a new influenza virus to which the population is immunologically naïve. It was believed that reassortment or an intermediary species (e.g., swine) was necessary for human infection until, in 1997, 18 humans in Hong Kong were infected with highly pathogenic avian influenza (HPAI) A virus subtype H5N1, resulting in 6 fatalities (4). It is now believed that the 1918 so-called Spanish flu, the deadliest influenza pandemic in recorded history, was generated by a similar mechanism (5). HPAI viruses are now considered a potential pandemic threat, and since their emergence in humans, there has been a reported total of 608 cases causing 359 deaths (as of August 2012) (http://www.who.int/influenza/human_animal_interface/H5N1_cumulative_table_archives/en/index.html).
Vaccination is considered the most effective approach of controlling seasonal influenza virus as well as potentially pandemic viruses in humans. Inactivated vaccines grown in embryonated chicken eggs remain the standard and are the most widely used for prevention of seasonal influenza (6–9). There are a number of limitations with this production strategy for HPAI H5N1, including safety concerns related to growing live influenza virus during production, lethality of H5 viruses in embryonated eggs reducing vaccine yields, and contraindications of egg-based vaccines, all of which could severely limit the response against an emerging pandemic. Currently, there are a number of candidate vaccines in clinical trials at this time (reviewed in reference 10), including inactivated viruses formulated with a variety of adjuvants, such as oil-in-water and live-attenuated influenza virus vaccines (11, 12). These approaches do not, however, address the concerns of using an egg-based vaccine for pandemic preparedness.
We have shown previously that parainfluenza virus 5 (PIV5), a nonsegmented negative-stranded RNA virus, expressing the HA from an H3 influenza virus (recombinant PIV5-H3 [rPIV5-H3]), is a safe and effective vaccine against influenza virus infection in mice (13). Similarly, we recently described an rPIV5 vaccine expressing the HA of an H5N1 influenza virus and showed it to be effective at protecting against HPAI H5N1 (A/VN/1203/04) virus challenge when delivered as a live, intranasal vaccine in mice (14). While intranasal (i.n.) immunization is appealing, there are certain drawbacks, including potential contraindications regarding the use of live, intranasal virus as a vaccine in immunocompromised populations (6). An injectable vaccine may avoid this issue and provide the opportunity for mass vaccination in agricultural applications. Here, we extend our previous work by analyzing the efficacy of an rPIV5-H5 vaccine (ZL46) delivered by alternate routes and show that rPIV5-H5 is protective against HPAI H5N1 challenge when administered not only intranasally (i.n.) but also intramuscularly (i.m.). Moreover, inactivated rPIV5-H5 vaccine was also effective at protecting against H5N1 infection following a booster immunization.
Influenza A viruses used include VNH5N1-PR8/CDC-RG, a reverse-genetic H5N1 vaccine strain (termed rgVN-PR8; kindly provided by Ruben Donis, CDC, Atlanta, GA), and A/Vietnam/1203/04 (H5N1; kindly provided by Richard Webby, St. Jude Children's Research Hospital, Memphis, TN). A/VN-PR8 was propagated in the allantoic cavity of embryonated hen eggs at 37°C for 48 to 72 h. β-Propiolactone (BPL)-inactivated A/Vietnam/1203/04 was provided by Richard Webby from St. Jude Children's Research Hospital (Memphis, TN). A/Vietnam/1203/04 was propagated in the allantoic cavity of embryonated hen eggs at 37°C for 24 h. All viruses were aliquoted and stored at −80°C. All experiments using live, highly pathogenic avian influenza viruses were reviewed and approved by the institutional biosafety program at the University of Georgia and were conducted in biosafety level 3 (BSL3) enhanced containment by following guidelines for use of select agents approved by the CDC.
Female 6- to 8-week-old BALB/c mice (Charles River Laboratories, Frederick, MD) were used for all studies. Mouse immunizations and studies with BSL2 viruses were performed in BSL2 and animal BSL2 (ABSL2) facilities. Mouse HPAI infections were performed in enhanced BSL3 and ABSL3 facilities by following guidelines approved by the institutional biosafety program at the University of Georgia and for use of select agents approved by the CDC. All animal studies were conducted under guidelines approved by the Animal Care and Use Committee of the University of Georgia.
Madin-Darby canine kidney (MDCK) and Madin-Darby bovine kidney (MDBK) cells were cultured in Dulbecco's modified Eagle medium (DMEM) with 5% fetal bovine serum (FBS), 5% l-glutamine, and an antibiotic/antimycotic solution (10,000 IU/ml penicillin, 10,000 μg/ml streptomycin, and 25 μg/ml amphotericin B; Cellgro Mediatech, Inc.). Vero cells were cultured in minimum essential medium (MEM; Thermo/HyClone) supplemented with 10% FBS and antibiotic/antimycotic. All cells were incubated at 37°C in 5% CO2.
rPIV5-H5-SH-HN (ZL46) and rPIV5-H5-HN-L were generated as described in the companion article (14). For ZL46, the HA gene from A/VN/1203/04 was synthesized as a modified cDNA with the polybasic cleavage site removed and inserted into the plasmid BH276 containing the full-length genome of PIV5 (rPIV5-H5-SH-HN). The gene end (GE), intergenic region, and gene start (GS) sequences between the small hydrophobic (SH) and hemagglutinin-neuraminidase (HN) genes were added to the primer to stop HA gene transcription and start HN gene transcription. ZL48 was made in a similar fashion, except HA was inserted between HN and large RNA polymerase (L;rPIV5-H5-HN-L). ZL46 and ZL48 viruses were rescued and sequenced as described elsewhere (13–15). To easily differentiate the rPIV5-H5 insertions, we will refer to them as ZL46 and ZL48.
PIV5 and rPIV5 virus stocks were grown in MDBK cells (maintained for <20 passages) for 5 to 7 days in DMEM containing 2% FBS until their hema-adsorption titers plateaued. Media were collected and clarified by centrifugation at 3,000 rpm for 10 min in an Eppendorf tabletop centrifuge (5810 R). Bovine serum albumin (BSA) was added to the clarified supernatant to bring the total solution to 1% BSA. The virus stocks were aliquoted and frozen quickly in dry ice and stored at −80°C. Virus titers were determined by plaque assay on Vero cells as described below.
PIV5 titers were determined by plaque assay using Vero cells. Cells were incubated with serial dilutions of virus samples using DMEM with 1% BSA and antibiotic/antimycotic. The cells were removed and overlaid with 1:1 low-melt-point agarose and DMEM with 2% FBS and antibiotic/antimycotic and incubated at 37°C for 5 to 6 days. Following incubation, cell monolayers were fixed with 10% buffered formalin and then permeabilized using 1× phosphate-buffered saline (PBS) with 2% FBS, 0.1% sodium azide, and 0.5% saponin (permeabilization buffer). PIV5 was detected using antibodies specific to the shared region of the V protein and phosphoprotein (P) of PIV5 (V/P) for 1 h. Horseradish peroxidase (HRP)-tagged goat-anti-mouse IgG (H&L) secondary antibody (Invitrogen) then was added and incubated for 30 min. To visualize plaques, 3,3',5,5'-tetramethylbenzidine (TMB) peroxidase substrate (prepared according to the manufacturer's instructions) was added (Vector Laboratories, Inc.). The plates were then washed and allowed to dry, and the numbers of plaques were counted.
Influenza virus titers were determined either by 50% tissue culture infectious dose (TCID50) assays as previously described (16) or by plaque assays on MDCK cells. For plaque assays, MDCK cells were incubated for 2 h at 37°C with serial dilutions of virus samples using MEM supplemented with 1 μg/ml tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-treated trypsin (Worthington Biochemical). Diluted virus samples were then removed, and monolayers were overlaid with 1.2% microcrystalline cellulose Avicel (17) containing 1 μg/ml TPCK-treated trypsin. Plates were incubated for 72 h, gently washed with PBS to remove overlay, fixed with cold methanol-acetone (40:60%), air dried, and counterstained with crystal violet. Plaques then were visualized.
MDBK cells in T-150 flasks were infected with PIV5, ZL48, or ZL46 at a multiplicity of infection (MOI) of 0.1. The media were collected after 72 h and centrifuged at 3,000 rpm for 10 min to remove cell debris. The clarified media were overlaid on 20% sucrose in NTE buffer (0.1 M NaCl, 0.01 M Tris-HCl, 0.001 M EDTA, pH 7.4). Samples were centrifuged at 40,000 rpm for 1.5 h at 4°C. Pellets were resuspended in 0.5 ml of PBS and mixed with 1.3 ml of 80% sucrose in NTE buffer. A gradient sucrose solution was generated by adding 1.8 ml 50% sucrose and 0.6 ml 10% sucrose, respectively. The gradient sucrose solution was then centrifuged at 45,000 rpm for 3 h at 4°C. The white bands formed by virions at the interface between 50 and 10% sucrose were collected and pelleted by centrifugation at 40,000 rpm for 1.5 h at 4°C. The pellets were resuspended in 0.5 ml PBS. The purified virions were analyzed by 10% SDS-PAGE and stained with Coomassie blue.
Vero cells were infected at an MOI of 5 using PIV5 or ZL46, or they were mock infected. Cells were lysed using PBS with 2 mM EDTA, Roche Complete Mini protease inhibitor (Roche Applied Science), and 1% Triton X-100 (octyl phenoxy polyethoxyethanol; Sigma) 24 h postinfection. Separation and Western blotting were performed as previously described (18). Hyperimmune serum from mice infected with rgA/VN-PR8 was used as a primary antibody to detect HA, and V/P-specific monoclonal antibodies (MAb) were used to detect PIV5 V/P. Precision Plus Protein Western C (Bio-Rad) was used as a standard.
Vero cells were grown in 24-well plates and infected with PIV5 or ZL46 at an MOI of 5 or were mock infected. At 24 h postinfection, cells were fixed with 5% buffered formalin for 10 min at room temperature. Cells were then permeabilized with permeabilization buffer and incubated for 1 h with an anti-HA (H5) A/VN/1203/04 MAb (BEI Resources). Phycoerythrin-labeled goat anti-mouse IgG (BD Pharmingen) was applied for 45 min to detect HA. To detect PIV5, V/P-specific antibodies were added and incubated for 1 h. To visualize PIV5, an Alexa Fluor-488-labeled secondary antibody (Invitrogen) was added and incubated for 30 min. PBS (0.5 ml) was added to each well, and fluorescence was examined using an AMS EVOS fluorescence microscope. Cells were washed with PBS extensively between each step.
For vaccination, 106 PFU of PIV5 or ZL46 in 50 μl PBS was administered intranasally (i.n.) to mice anesthetized with 2,2,2-tribromoethanol in tert-amyl alcohol (Avertin; Aldrich Chemical Co). For sublethal rgA/VN-PR8 infection, 2,000 PFU in 50 μl PBS was administered as described for PIV5 vaccination. For rgA/VN-PR8 intramuscular (i.m.) vaccination, 2,000 PFU rgA/VN-PR8 was administered in 50 μl PBS in the caudal thigh muscle. For inactivated A/VN/1203/04 immunization, BPL-inactivated virus was resuspended at 256 hemagglutination units (HAU)/ml in 50 μl PBS and injected into the caudal thigh muscle of both legs (100 μl; total of 25 HAU). Blood was collected on day 21 postimmunization. Nasal wash and bronchoalveolar lavage (BAL) procedures were performed on days 14 or 21 postvaccination using 0.5 or 1 ml PBS, respectively.
H5-specific serum antibody titers were measured using an IgG enzyme-linked immunosorbent assay (ELISA). Immulon 2 HB 96-well microtiter plates (ThermoLabSystems) were coated with 2 μg/ml recombinant H5 protein and incubated at 4°C overnight. Plates were then washed with KPL wash solution (KPL, Inc.) and the wells blocked using KPL wash solution with 5% nonfat dry milk and 0.5% BSA (blocking buffer) for 1 h at room temperature. Serial dilutions of serum samples were made (in blocking buffer), transferred to the coated plate, and incubated for 1 h at room temperature. To detect bound serum antibodies, alkaline phosphatase-labeled goat anti-mouse IgG (KPL, Inc.) in blocking buffer was added to each well and incubated for 1 h at room temperature. Plates were developed using a pNPP phosphatase substrate (KPL, Inc.), and the reaction was allowed to develop at room temperature. Optical density (OD) was measured at 405 nm on a Bio-Tek Powerwave XS plate reader. The IgG titer was determined to be the lowest serum dilution with an OD greater than the mean OD of naïve serum plus 2 standard deviations.
Influenza neutralizing antibody titers were measured in serum by a microneutralization assay with an ELISA endpoint. Heat-inactivated serum was serially diluted in DMEM with 1% BSA, antibiotic/antimycotic, and 1 μg/ml TPCK trypsin. Diluted serum was then incubated with 1,000 TCID50 of rgA/VN-PR8 for 2 h at 37°C. Following incubation, MDCK cells were added and plates incubated at 37°C for 18 to 24 h. After incubation, wells were fixed with ice-cold methanol-acetone (80:20, respectively), and an ELISA was performed as described above using an anti-nucleoprotein (NP) MAb (clone H16-L10; kindly provided by Jon Yewdell, NIH, Bethesda, MD). The neutralization titer was determined to be the lowest serum dilution capable of neutralizing 1,000 TCID50 rgA/VN-PR8, as determined by an OD readout two times above the background OD.
Twelve days postvaccination with PIV5, ZL46, or rgA/VN-PR8, mediastinal lymph nodes (MLN) from mice were harvested, pooled, and homogenized. Lymphocytes were depleted of erythrocytes using Gey's balanced salt solution (Sigma-Aldrich) for 5 min at room temperature and debris removed. Cells were then counted using a Z2 Coulter particle count and size analyzer (Beckman Coulter). Gamma interferon (IFN-γ) enzyme-linked immunosorbent spot (ELISpot) assays to detect T-cell responses in lymphocytes to inactivated A/VN/1203/04 were performed as previously described (19). Cells were restimulated with inactivated A/VN/1203/04 (the equivalent of 10 HAU per well), Ebola GP P2 EYLFEVDNL as an irrelevant peptide (1 μg/ml), and concanavalin A (2 μg/ml) in 50 μl complete tumor medium (CTM). Spots were counted using an AID ViruSpot reader (Cell Technology, Inc.).
BALB/c mice were first vaccinated as described above and then bled as indicated. For challenge, mice were anesthetized and inoculated i.n. with 10 50% lethal doses (LD50) A/Vietnam/1203/04 (20 PFU) or 20,000 PFU of rgA/VN-PR8 diluted in 50 μl PBS. Mice were monitored daily for morbidity and mortality with body weights measured every other day. On day 3 postinfection (p.i.), groups of mice were humanely euthanized and lungs collected into 1.0 ml PBS, homogenized, and clarified by centrifugation. A TCID50 assay was then used to determine virus titers from clarified homogenates as previously described (16).
Statistical differences in survival were determined by log-rank analysis. Differences in lung virus titers were determined by analysis of variance (ANOVA), followed by a Dunnett's multiple comparison test. P ≤ 0.05 was considered significant. Statistical analyses were performed using GraphPad Prism.
We have previously shown that PIV5-expressed recombinant H3 incorporated the influenza virus HA protein into the PIV5 virion surface (13). However, the ZL46 virus has the HA gene inserted closer to the PIV5 leader than the rPIV5-H3 virus (between SH and HN or HN and L, respectively; Fig. 1A). Moreover, modification of the cleavage site of the H5 HA may have hindered expression of the glycoprotein. To test for normal expression and packaging of recombinant PIV5, MDBK cells were infected with PIV5, ZL48, or ZL46 or were mock infected. ZL48 has the H5 gene inserted between HN and L (Fig. 1A) (14), similar to PIV5-H3 (13), so it was included as a control comparable to the previously published virus. Supernatants were collected, purified over sucrose, separated by SDS-PAGE, and Coomassie stained to visualize protein bands. Protein bands at sizes appropriate for PIV5 HN, NP, F, P, and M proteins were readily visible in all samples, while a band at a size appropriate for influenza virus HA was visible in ZL46 and ZL48 samples but not PIV5 (Fig. 1B). Identities of these bands were confirmed by Western blot analysis (data not shown).
To confirm that H5 HA was incorporated into the rPIV5-H5 virion, we utilized dynamic light scattering (DLS) and gold nanoparticle (AuNP) labels to detect HA on the virion surface of ZL46 compared to rPIV5 virions as previously described (20). Cleared virus culture supernatants of PIV5, ZL46, or rgA/VN-PR8 were incubated with AuNP-labeled anti-HA (H5) antibodies (A/VN/1203/04 MAb; BEI Resources) and then measured for aggregation of the AuNP probes. The degree of AuNP aggregation correlates with the presence of virus containing specific HA with increases in virus increasing aggregation, causing a shift in Z average. An increase in the mean hydrodynamic diameter (z average) of 8 nm was observed for ZL46 compared to that for PIV5 (90.41 ± 1.316 versus 82.08 ± 0.605 nm, respectively), indicating that there was antigen-specific aggregation of the AuNP probes upon introduction of the viruses. This suggests that HA is present on the surface of the virion. The mean diameter observed for PIV5 was approximately the same size as that of culture supernatant or allantoic fluid alone (77.06 ± 0.609 and 81.25 ± 1.287 nm, respectively). The positive control, rgA/VN-PR8 virus, had a mean diameter of 113.67 ± 1.475 nm.
While HA incorporation into the virion was measurable by protein stain and DLS, the HA antigen was reduced compared to HN, and HN was reduced in ZL46 and ZL48 compared to the level in PIV5 (Fig. 1B). Thus, the amount of HA incorporated into the PIV5 virion was likely less than that incorporated into an influenza virion. However, we had previously demonstrated that rPIV5 expressed H3 HA in infected cells (13). To confirm that native HA was being expressed during PIV5-H5 infection, Vero cells were infected with PIV5 or ZL46 (MOI, 5) or were mock infected, and 24 h later they were lysed and analyzed by Western blotting. Detection with polyclonal antiserum raised against the rgA/VN-PR8 (H5N1) virus visualized a 75-kDa protein, the size of the influenza HA0 monomer in the ZL46 cell lysate but not in the PIV5 or mock lysate (Fig. 1C). A 46-kDa band detected with the PIV5 V/P-specific MAb is present in all infected lysates. To confirm that H5 HA is expressed on the surface of infected cells, immunofluorescence staining was performed. Vero cells infected with PIV5 or ZL46 (MOI, 5) or mock infected were stained with an anti-HA (H5) MAb or a monoclonal antibody specific for the V/P proteins of PIV5 (anti-V/P). Robust, equivalent expression of V/P was detected in both ZL46- and PIV5-infected cells, while H5 was detected only in ZL46-infected cells, confirming that HA is being expressed in cells infected with rPIV5-H5 (Fig. 1D).
Intranasal immunization with recombinant PIV5 constructs expressing the HA of an H3 virus or an H5 from HPAI H5N1 were shown to be protective against influenza virus challenge in mice (13, 14). However, intramuscular immunization, a route of immunization widely used for vaccination, has not been tested. Moreover, since we detected the H5 HA in the rPIV5-H5 virion, it was possible that inactivated virus induced protective immune responses similar to the current inactivated influenza virus vaccines. To determine the immunogenicity of rPIV5-H5 administered by various routes, mice were vaccinated i.n. with live ZL46 or i.m. with either live or inactivated (iZL46) virus. A last group of mice was vaccinated i.m. with inactivated A/VN/1203/04 (iA/VN/1203/04) as a positive control. All groups were compared to mice given PIV5 i.n. Mice were bled on days 7, 14, and 21, and their sera were assessed for HA-specific IgG and H5N1 influenza neutralizing antibodies. Mice vaccinated intranasally or intramuscularly with ZL46 produced high levels of IgG as early as 7 or 14 days postimmunization, respectively (Fig. 2A), with titers comparable to those of mice immunized with inactivated whole influenza virus (iA/VN/1203/04). This high level of H5-specific serum antibodies primed by ZL46 immunization i.m. or i.n. also correlated with robust neutralizing antibody titers (Fig. 2B). In contrast, mice vaccinated with inactivated ZL46 (iZL46) i.m. produced limited IgG and neutralizing antibody, suggesting that the amounts of HA antigen incorporated into the virion are insufficient to prime potent humoral responses or that PIV5 replication is required to induce immunity. As expected, PIV5-vaccinated mice produced no detectable HA-specific IgG antibodies or rgA/VN-PR8 neutralizing antibodies (Fig. 2A and andBB).
One advantage of intranasal immunization is the potential to induce a mucosal immune response. To assess differences in the IgA responses in mice vaccinated with rPIV5-H5 i.m. versus i.n., mice were vaccinated with PIV5, ZL46 i.m. or i.n., or a sublethal dose of rgA/VN-PR8, and nasal wash and BAL fluid examinations were performed on days 14 or 21 postimmunization. No IgA was detectable in the nasal lavage or BAL fluid of mice vaccinated i.m. with ZL46 or rgA/VN-PR8 (Fig. 2C and andD,D, respectively). In contrast, intranasal administration of ZL46 induced robust IgA responses in both the nasal passages and lungs of immunized mice (Fig. 2C and andD,D, respectively). IgA levels were comparable to those of rgA/VN-PR8-infected mice on day 14; however, the mucosal IgA response in influenza virus-inoculated mice continued to rise after day 14, likely due to the longer period of virus replication with the influenza virus on the A/PR8/34 backbone before clearance compared to that of rPIV5 (data not shown).
To assess differences in T cell priming according to route of administration, groups of mice vaccinated i.m. or i.n. with PIV5, ZL46, or rgA/VN-PR8 were euthanized on day 12 postinfection, and lymph node lymphocytes were assayed for influenza virus-specific IFN-γ-producing T cells. Intranasal vaccination with rPIV5-H5 or influenza virus primed robust influenza-specific T cell responses in draining lymph nodes but also had increased nonspecific responses compared to i.m.-immunized mice (Fig. 2E). i.m. vaccination with ZL46 primed an A/VN/1203/04-specific T cell response more effectively that i.m. administration of rgA/VN-PR8, possibly due to improved replication of PIV5 in muscle tissue compared to influenza virus. This would be an advantage for i.m. vaccination with PIV5, as T cells can play a role in protection against influenza virus infection.
To determine if this mucosal response is necessary for protection against influenza virus infection, i.e., if i.m. immunization with rPIV5-H5 could protect against challenge, mice were vaccinated with PIV5, rPIV5-H5, or rgA/VN-PR8 delivered i.n. or i.m. A group of mice was also vaccinated with inactivated PIV5-H5 i.m. (iZL46) to determine if the weak IgG responses detected (Fig. 2A) were protective. On day 28 postimmunization, mice were challenged with 10 LD50 HPAI A/VN/1203/04. Consistent with previous results and observed antibody titers, mice vaccinated with rPIV5-H5 i.n. were protected from weight loss and mortality associated with HPAI H5N1 challenge (Fig. 3A and andB).B). Mice vaccinated with ZL46 i.m. were also protected from H5N1 challenge; however, there was associated mortality and weight loss late in infection, suggesting that the mucosal antibody responses are important for complete protection. Virus titers were assessed in a subset of mice 3 days postchallenge. H5N1 virus was undetectable in mice immunized with ZL46 or rgA/VN-PR8 i.n. In contrast, mice immunized with ZL46 i.m. had no reduction in challenge virus titer compared to PIV5 i.m.-immunized mice, whereas rgA/VN-PR8 i.m.-immunized mice also had no detectable virus (Fig. 3C). This suggests again that mucosal IgA primed by rPIV5-H5 i.n. immunization is important for complete protection. Alternatively, priming of immune responses to multiple influenza antigens, as occurs with rgA/VN-PR8 immunization (data not shown), may overcome the need for mucosal IgA responses. Protection was not observed in mice vaccinated with inactivated rPIV5-H5 (iZL46), confirming that live virus, and presumably replication, is required for the induction of protective immunity. So, while both routes of administration are protective (as measured by weight loss and survival), induction of mucosal IgA responses and/or the increased IFN-γ T cell numbers associated with i.n. immunization are limiting infection and virus replication in the lung.
A single i.m. immunization with inactivated rPIV5-HA induced limited HA-specific serum IgG (Fig. 1A) but no neutralizing antibodies (Fig. 1B), and it provided no protection from infection (Fig. 3). Similar results have been seen with other H5 vaccine antigens (21). To test whether a boost could increase the immune response to protective levels, ZL46-primed mice were bled weekly and boosted on day 28 postpriming, and sera were collected 7 and 14 days later. Mice immunized with live ZL46 i.n. or i.m. were compared to mice immunized i.m. with inactivated ZL46 (iZL46) or inactivated A/VN/1203/04 (iA/VN/1203/04). Once again, live ZL46 induced robust IgG and neutralizing antibody responses by 21 days postpriming with more modest IgG and no neutralizing antibody responses from iZL46. However, 1 week after boosting, all three ZL46 immunization methods (live i.n., live i.m., and inactivated i.m.) had robust HA-specific IgG and H5 neutralizing serum antibody responses (Fig. 4A and andB).B). Six weeks after boosting, mice were challenged with the rgA/VN-PR8 virus, and 3 days later they were euthanized to assess lung virus titers. Strikingly, all vaccinated mice (ZL46 i.n. or i.m., iZL46 i.m., and iA/VN/1203/04 i.m.) had significantly reduced lung virus titers compared to the PIV5-immunized control mice (P < 0.05 by ANOVA, followed by a Dunnett's multiple-comparison test) (Fig. 4C), whereas none of the HA-vaccinated groups were significantly different from each other. While i.n. immunization with ZL46 continued to provide the best protection (3/5 mice had undetectable virus titers while all of the other ZL46-immunized mice all had detectable virus on day 3 postchallenge), i.m. immunization with live or inactivated PIV5-HA was effective at inducing protective neutralizing serum antibody responses that were comparable to those of whole, inactivated, wild-type influenza virus.
Inactivated vaccines against HPAI H5N1 viruses are generally poorly immunogenic in mammals, requiring high antigen doses, multiple immunizations, and adjuvants in some cases (21–24). Both live and inactivated HPAI vaccines have issues with production, including reduced yields of vaccine during production, difficulty generating seed strains, and safety concerns (25). We have developed a novel recombinant paramyxovirus vaccine vector, PIV5 that expresses the HA from H5N1 influenza virus and induces protective immunity against HPAI virus infection (Fig. 1 to to3)3) (14). Here, we assessed the efficacy of this vaccine by utilizing different immunization methods and comparing live and inactivated vaccines.
Intranasal immunization with live rPIV5-H5 induces robust serum and mucosal antibody responses specific for the HA transgene. The HA-specific response induced by a single immunization with the avirulent rPIV5-H5 vaccine is comparable to responses induced by sublethal infection with rgA/VN-PR8 influenza virus, where the influenza virus-immunized mice had clinical signs of illness and lost up to 15% of their original body weight (Fig. 2 and data not shown). Thus, PIV5 provides an appealing method for intranasal immunization without the concerns for reassortment posed by live attenuated influenza virus vaccines.
Unlike influenza virus, which generally replicates in airway or gut epithelial cells due to receptor and protease requirements (26), PIV5 has the potential for broader cellular tropism. This feature makes it an appealing candidate for use as a live intramuscular vaccine. While this also presents the possibility that a PIV5 vaccine could disseminate to other tissues, previous studies found no evidence of pathology in other tissues after intranasal PIV5 infection (13), suggesting intramuscular immunization with an rPIV5 vector would be safe as well.
Intramuscular immunization with rPIV5-H5 induces robust HA-specific and neutralizing serum antibody responses, comparable to intranasal immunization with rPIV5-H5 or intramuscular immunization with whole inactivated H5N1 virus (Fig. 2). The intramuscular immunization with the live rPIV5-H5 has the added advantage of priming robust HA-specific T cell responses, where intramuscular immunization with live influenza viruses elicited limited increases in antigen-specific IFN-γ-producing lymphocytes (Fig. 2E). This is a benefit over inactivated vaccines, as inactivated influenza is a poor inducer of cellular immune responses (27, 28). Intramuscular immunization with live rPIV5-H5 still requires further optimization, however, as protection from challenge was incomplete (Fig. 3). This may be due to the lack of mucosal antibody induced through intranasal immunization and/or the lower serum neutralizing antibody titers compared to those of the high-dose inactivated H5N1 vaccine immunization. While the rPIV5-H5 intramuscular immunization was imperfect, it does provide a potential route of immunization that may be more appropriate for individuals with asthma or other contraindications for intranasal immunization. Moreover, it provides an opportunity for combining this vaccine with other injectable vaccines, as well as an injectable vaccine formulation which may be appealing for agricultural applications.
Mucosal antibodies have been associated with protection from both homologous and heterosubtypic immunity (29–35). Here, we found that similar to inactivated influenza virus vaccines, intramuscular administration of a live rPIV5-H5 vaccine failed to induce mucosal IgA responses but protected against lethal H5N1 challenge (Fig. 2 and and3).3). Mucosal (intranasal) immunization with the same vaccine primed comparable virus neutralizing serum antibody titers but also induced virus-specific lung and nasal IgA (Fig. 2). These mice were also protected from mortality associated with a lethal H5N1 infection and had no detectable virus in the lung on day 3 postinfection. In contrast, mice without detectable IgA (intramuscular immunized groups) had virus titers similar to those of control animals. Thus, while intramuscular immunization may have some advantages over intranasal immunization, particularly for individuals with chronic respiratory disease, intranasal immunization with this vaccine is likely to be the most effective route of administration. Inactivated rPIV5-H5 was not efficacious following a single i.m. administration. It is possible that the amount of HA incorporated into the virion is insufficient to effectively prime a protective antibody response against influenza infection; however, inactivated, whole-virus influenza vaccines have also failed to induce virus neutralizing serum antibody responses after a single immunization (21, 22). Here, the inactivated H5N1 whole virus did induce neutralizing antibodies, possibly due to the high antigen dose (25 HAU, equivalent to 106 PFU of infectious virus prior to inactivation); moreover, the whole virus has the advantage of priming responses to multiple influenza antigens, which can contribute to protection from challenge. Alternatively, the influenza virus may contain other antigens or pathogen-associated molecular patterns (PAMPs) that more effectively prime the response (i.e., act as an adjuvant), and the PIV5 virus lacks these stimulatory molecules. This is reflected in the reactogenicity seen with whole-virion vaccines (36). In either case, replication-competent rPIV5-H5 overcame this deficiency, priming both T cell and neutralizing antibody responses that protected against homologous HPAI challenge. Moreover, boosting with the inactivated rPIV5-H5 vaccine (iZL46) was successful at priming influenza virus neutralizing antibody responses that reduced H5N1 virus titers after challenge comparably to live rPIV5-H5 and inactivated H5N1 vaccine (Fig. 4). As PIV5 vaccines can be readily grown in vaccine-approved cell lines (e.g., Vero cells), rPIV5-HA may provide an avenue for rapid, safe production of a traditional HA-specific inactivated pandemic influenza virus vaccine without the challenges associated with identification and development of influenza vaccine seed strains (7, 9). Formulation of inactivated rPIV5-H5 vaccines with an adjuvant may enable priming of neutralizing antibody titers with a single immunization, similar to other H5 HA-based vaccines (25); however, concerns associated with adjuvant use would still need to be considered. Similarly, a PIV5-vectored influenza vaccine could have other issues similar to those of licensed influenza vaccines. While production in Vero cells would eliminate issues with optimizing vaccine growth in embryonated eggs (9), insertion of distinct hemagglutinins could result in altered growth characteristics of the rPIV5-HA vaccine. We have inserted HA genes from group 1 and group 2 HAs, including H1, H3, and H5, and have not seen changes in the growth of these viruses (data not shown, reference 14, and reference 13, respectively), but additional constructs need to be tested to confirm this. Shelf life and cold-chain requirements are also relevant to vaccine development; these are aspects of PIV5-vectored vaccine development that still need to be considered.
One of the advantages of PIV5 as a vaccine vector is the potential to manipulate gene expression (14, 37), immune responses to the virus (38, 39), and apoptosis (40), as well as the potential incorporation of antigens into the virions. The HN protein of PIV5 is essential for budding of PIV5 virions (41, 42). Protein concentrations of HN were reduced in ZL46 and ZL48 compared to those of wild-type PIV5 (Fig. 1B), suggesting that the HA was competing for incorporation. The limited incorporation of HA or competition with HN could explain the reduced immunogenicity of the inactivated ZL46 and the requirement of a boost to induce neutralizing antibodies (Fig. 2B and and4B).4B). HA concentration incorporation could be increased by generating an HN/HA chimera with the HA stalk and head attached to the HN cytoplasmic and transmembrane domains. If successful, the HN might be dispensable in the vaccine, further increasing HA incorporation into the vaccine virions.
The availability of alternative immunization methods presents advantages for agricultural applications. Vaccination of chickens against avian influenza virus and other avian pathogens is currently accomplished by a variety of methods, including subcutaneous injection, eye drop inoculation, spray (aerosol) delivery, delivery in drinking water, and in ovo inoculation (43). We have not tested avian species for susceptibility to PIV5 immunization; however, PIV5 does grow in embryonated chicken eggs (44) and the V protein has been shown to block mda-5 activity in avian cells, limiting IFN-β production (45). Thus, PIV5 may be effective as a vaccine for avian species.
Although we have previously shown intranasal administration of rPIV5-HA to be safe in mice (13), respiratory delivery of live replicating virus-vectored vaccines can be of concern for asthmatic or immunocompromised patients (6, 7). The option of intramuscular administration of live vaccine or the inactivated vaccine would provide appealing alternatives to intranasal immunization without a modification of the vaccine platform. This is in contrast to the current licensed alternatives, where either a live attenuated influenza virus vaccine or a split, inactivated wild-type virus vaccine is delivered intranasally or intramuscularly, respectively (6, 7). Intranasal and intramuscular administration of inactivated influenza vaccines has been analyzed for induction of serum antibody and mucosal IgA responses in humans. Intranasal administration of inactivated vaccine was associated with increases in IgA in nasal secretions, with increased rates of IgA being associated with i.n. antigen dose (46). In humans, both serum antibody and nasal IgG or IgA have been associated with protection from influenza virus infection (30). Thus, there are opportunities for traditional, inactivated vaccines to be delivered by multiple routes. An inactivated intranasal influenza virus vaccine was licensed for use in Switzerland in 2001. Clinical trials showed that the vaccine elicited the greatest IgA response when formulated with heat-labile toxin (47). Subsequently, the vaccine was associated with increases in Bell's palsy and is no longer in use (48), so while inactivated vaccines can be delivered intranasally, caution will have to be used with adjuvants. Similarly, use of an inactivated paramyxovirus-vectored vaccine would have to be carefully tested, as a formalin-inactivated vaccine for respiratory syncytial virus (RSV) led to more severe disease in vaccinated children upon subsequent natural infection. PIV5 vaccination should not result in a similar phenotype, however, as the G protein of RSV and a CX3C chemokine motif located as a highly conserved structure in the RSV G protein are associated with the enhanced disease phenotype (49), and PIV5 does not encode a G protein or similar structure (15, 50). Moreover, intramuscular immunization with a live or inactivated rPIV5-HA vaccine could enable coformulation with existing live or antigen-based (inactivated) vaccines, improving potential utility in mass vaccination campaigns. Finally, the availability of multiple administration and formulation options would be useful for agricultural vaccination programs, where a universal production platform for diverse applications could provide a safer and more cost-effective option and improve vaccination (51). Thus, an rPIV5-HA vaccine is a versatile vaccine platform for production of vaccines to protect against emerging or pandemic influenza viruses and H5N1 HPAI viruses in particular.
We thank the Animal Resources personnel at the College of Veterinary Medicine, University of Georgia, for excellent animal husbandry and the Office of Biosafety personnel at the University of Georgia for strong support of biocontainment and select agent research. Additional thanks go to Dan Dlugolenski, Scott Johnson, and Jeff Hogan for superior assistance with ABSL3 studies and Cheryl Jones for excellent technical support at the College of Veterinary Medicine, University of Georgia.
This work was supported by grants from the National Institute of Allergy and Infectious Disease (R01AI070847) to B.H.
Published ahead of print 17 October 2012