Influenza virus is a segmented negative-sense RNA virus, belonging to the
Orthomyxoviridae family. The virion is surrounded by a lipid membrane containing two major glycoproteins, the hemagglutinin (HA) and neuraminidase (NA). The HA protein is the most abundant viral surface glycoprotein and is responsible for the attachment of virus to terminal sialic acid residues on host cell receptors (
4) and mediating fusion between viral and cellular membranes (
6). There are 16 distinct antigenic subtypes of influenza A viruses that are recognized on the basis of antigenic properties of the HA protein (
15). Influenza A virus causes a highly contagious and acute viral respiratory disease, which can pose serious public health problems resulting in significant morbidity and mortality worldwide each year (
36,
40). A more serious concern is that avian influenza viruses are a source for a diverse mix of antigenic subtypes representing a large reservoir of novel viruses to which the human population is naïve (
32,
37,
38,
41). Genetic reassortment of RNAs between avian and human influenza viruses or mutations affecting host range could enable avian viruses to transmit among the human population, which may lead to a global pandemic with high mortality.
Vaccination has been an effective way to reduce the disease resulting from an influenza virus infection. The major supply of influenza virus vaccine is currently produced using embryonated chicken eggs. However, manufacturing problems experienced in recent years illustrate that the current methods of production are fragile in ensuring an adequate and timely supply of influenza virus vaccine. More importantly, the egg-based technology may not be suitable to respond to a pandemic crisis. The H5 avian influenza virus strains responsible for recent epizoonotic outbreaks in Asia are lethal to chicken eggs (
25,
29,
31). Also, due to the high pathogenicity of avian influenza virus strains, the conventional production of avian influenza virus vaccines would require biosafety level 3 containment facilities. In addition, vaccine development and production take several months following the identification of new potential strains and typically require reassortment with a high-yield strain. Therefore, a strategy that can rapidly produce new influenza virus vaccines is needed as a priority for pandemic preparedness.
Virus-like particles (VLPs) have been generated and tested as vaccine candidates for a variety of viruses (
2,
11,
12,
19,
22). It was recently reported that immunization with influenza VLPs (H3N2 and H9N2) reduced challenge virus replication and conferred protection against an influenza virus challenge (
8,
24). However, the immune responses induced by influenza VLPs are not well characterized, and the memory responses and cross-protective immunity are unknown for VLP immunization. In this study, we developed VLPs for influenza virus A/PR8 (H1N1), for which the challenge system and immune epitopes are well-defined in a mouse model. Intranasal immunization of mice with these influenza VLPs induced mucosal and systemic immune responses, including both humoral and cellular immune components. We observed that immune responses induced by the influenza VLPs conferred cross-protection against lethal challenge with homologous or heterologous strains. We further analyzed protective memory immune responses induced by VLP immunization.
MATERIALS AND METHODS
Virus and cells.
Spodoptera frugiperda Sf9 cells were maintained in suspension in serum-free SF900II medium (GIBCO-BRL) at 27°C in spinner flasks at a speed of 70 to 80 rpm. CV-1 and Madin-Darby canine kidney (MDCK) cells were grown and maintained in Dulbecco's modified Eagle's medium (DMEM). Mouse-adapted influenza A/PR8/34 (provided by Huan Nguyen, University of Alabama at Birmingham, Birmingham, AL) and A/WSN/33 (from Yumiko Matsuoka, CDC, Atlanta, GA) viruses were prepared as lung homogenates from intranasally infected mice.
Preparation of influenza VLPs.
A cDNA for influenza virus M1 (A/PR8) was obtained from Yumiko Matsuoka (CDC, Atlanta, GA). The M1 gene was PCR amplified with primers containing flanking restriction enzyme sites for cloning into the pSP72 plasmid expression vector under the T7 promoter (forward primer, 5′ TCC
CCCGGG CCACC ATG AGC CTT CTG ACC GAG GTC 3′; reverse primer, 5′ TTA CT
TCTAGA TTA CTT GAA CCG TTG CAT CTG 3′; SmaI and XbaI sites are underlined). The pSP72 clone containing the M1 gene was confirmed by DNA sequencing, and the expression of the M1 protein was confirmed by Western blot analysis of CV-1 cells transfected with pSP72 containing the M1 gene following infection with a recombinant vaccinia virus expressing T7 polymerase. The M1 gene was subcloned into the SmaI and XbaI site in the baculovirus transfer vector pc/pS1 containing a hybrid capsid-polyhedrin promoter. To produce a recombinant baculovirus (rBV) expressing M1, Sf9 insect cells were cotransfected with Baculogold DNA (BD/PharMingen) and the pc/pS1-M1 transfer vector by following the manufacturer's instructions. The supernatant was harvested 5 days after transfection, and recombinant plaques expressing M1 were selected by plaque assay and expanded. A rBV expressing influenza virus A/PR8 HA (H1N1) was previously described (
10). For Western blot analysis to determine the expression of M1 and HA, transfected or infected cells were dissolved in sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) sample buffer (50 mM Tris, 3% β-mercaptoethanol, 2% SDS, 10% glycerol), separated by SDS-PAGE, and then probed with mouse anti-M1 antibody (1:4,000; Serotec) and sera from PR8 virus-infected mice (1:1,000). The virus titer was determined with a Fast Plax titration kit according to the manufacturer's instructions (Novagen, Madison, WI).
To produce VLPs containing influenza virus M1 and HA, Sf9 cells were coinfected with rBVs expressing HA and M1 at multiplicities of infection of 4 and 2, respectively. Culture supernatants were harvested at 3 days postinfection, cleared by low-speed centrifugation (2,000 ×
g for 20 min at 4°C) to remove cells, and VLPs in the supernatants were pelleted by ultracentrifugation (100,000 ×
g for 60 min). The sedimented particles were resuspended in phosphate-buffered saline (PBS) at 4°C overnight and further purified through a 20%-30%-60% discontinuous sucrose gradient at 100,000 ×
g for 1 h at 4°C. The VLP bands were collected and analyzed by using Western blots probed with anti-M1 antibody and mouse anti-PR8 sera for detecting M1 and HA, respectively. The level of residual rBV in the purified VLPs was determined by plaque assay, and equivalent titers of HA-expressing rBVs were estimated to contribute less than 5% of HA in VLPs as determined by Western blotting (data not shown). The functionality of HA incorporated into VLPs was assessed by hemagglutination activity using chicken red blood cells as described previously (
26). Also, cleavability of HA into HA1 and HA2 subunits was determined by using increasing concentrations of trypsin (treated with
l-1-tosylamide-2-phenylethyl chloromethyl ketone [TPCK]; Sigma) as previously described (
18).
Electron microscopy.
To examine budding of VLPs, Sf9 cells infected with rBVs expressing M1 and HA were fixed with 0.25% glutaraldehyde and 1% osmium tetraoxide, dehydrated with ethanol, and then embedded in Epon resin. Thin sections were stained with lead citrate and uranyl acetate and observed by electron microscopy. For negative staining of VLPs, sucrose gradient-purified VLPs (1 to 5 μg) were applied to a carbon-coated Formvar grid for 30 seconds. Excess VLP suspension was removed by blotting with filter paper, and the grid was immediately stained with 1% phosphotungstic acid for 30 seconds. Excess stain was removed by filter paper, and the samples were examined using a transmission electron microscope.
Immunization and challenge.
Female inbred BALB/c mice (Charles River) aged 6 to 8 weeks were used. Mice (24 mice per group) were intranasally immunized with 40 μg of VLPs two times (weeks 0 and 3) and 10 μg of VLPs three times (weeks 0, 3, and 6) in 50 μl of PBS at 3-week intervals. To determine the effect of VLP integrity on its immunogenicity, VLPs were heat treated at 95°C for 5 min and used to immunize mice as a control group. For virus challenge, isoflurane-anesthetized mice were intranasally infected with 2,000 PFU of A/PR8 virus (10× the 50% lethal dose [LD50]) or 750 PFU WSN (10× LD50) in 50 μl of PBS per mouse 4 or 21 weeks after the final immunization. For measurement of immune response parameters, six mice from each group were sacrificed prior to challenge or on day 4 postchallenge. Mice were observed daily to monitor changes in body weight and to record death.
Sample collection.
Blood samples were collected by retro-orbital plexus puncture before immunization, at 2 weeks after boost immunization, and at different time points (weeks 4, 8, and 21) after the last immunization. After the blood samples were allowed to clot and centrifuged, serum samples were collected and stored at −20°C prior to antibody titration. Nasal and tracheal washes and lung samples were collected from individual mice at week 4.5 after the last immunization or on day 4 after a challenge infection (
3,
33). The whole-lung extracts prepared as homogenates using frosted glass slides were centrifuged at 1,000 rpm for 10 min to collect supernatants. The lung supernatants were frozen and kept at −70°C until used for immunoglobulin and virus titers and cytokine assays. Cells from bone marrow were harvested from individual mice 21 weeks after the last immunization, prepared as previously described (
14), and used for detection of influenza virus-specific immunoglobulin G (IgG) and IgA antibody-secreting plasma cells. Lymphocytes from spleen samples were collected from sacrificed mice and used for enzyme-linked immunospot (ELISPOT) analysis.
Evaluation of humoral immune responses.
Influenza virus-specific antibodies of different subtypes (IgG, IgG1, IgG2a, IgG2b, IgG3, and IgA) were determined in sera, wash samples of nose and trachea, and lung extracts by enzyme-linked immunosorbent assay (ELISA) as described previously (
26). Briefly, 96-well microtiter plates (Nunc-Immuno Plate MaxiSorp; Nunc Life Technologies, Basel, Switzerland) were coated with 100 μl of inactivated PR8 (or WSN or heat-treated VLPs) at a concentration of 4 μg/ml in coating buffer (0.1 M sodium carbonate, pH 9.5) at 4°C overnight. The plates were then incubated with horseradish peroxidase-labeled goat anti-mouse IgG, IgG1, IgG2a, IgG2b, IgG3, or IgA (Southern Biotechnology) at 37°C for 1.5 h, and then, the substrate
O-phenylenediamine (Zymed, San Francisco, Calif.) in citrate-phosphate buffer (pH 5.0) containing 0.03% H
2O
2 (Sigma) was used to develop color. The optical density at 450 nm was read using an ELISA reader (model 680; Bio-Rad).
Determination of influenza virus-specific antibody-secreting cells from bone marrow.
Inactivated PR8 viral antigen or anti-mouse IgA and IgG antibodies as capture antibodies were used to coat Multiscreen 96-well filtration plates (Millipore). Freshly isolated cells from bone marrow (1 × 10
6 cells) were added to each well and incubated for 48 h at 37°C with 5% CO
2. Using horseradish peroxidase-conjugated anti-mouse immunoglobulin antibodies and the ELISPOT assay substrate diaminobenzidine (Research Genetics), color was developed following the manufacturer's instructions, and counting of ELISPOTs was performed as described previously (
13).
Lung viral titers and virus neutralization assay.
Lung viral titers and neutralization assays were performed using MDCK cells as previously described (
26). Briefly, serum samples were serially diluted in DMEM, and a final volume of 190 μl was mixed with 10 μl of diluted virus stock containing approximately 100 infectious particles. The virus-serum mixtures were incubated at 37°C for 1 h and then added to six-well plates containing confluent MDCK cell monolayers. The plates were incubated at 37°C for 1 h, overlay medium containing DEAE dextran, nonessential amino acids, glutamine, and trypsin was added, and incubated for 2 or 3 days. The cells were then fixed with 0.25% glutaraldehyde and stained with 1% crystal violet.
Cytokine assays.
All antibodies against mouse cytokines used in cytokine ELISPOT assays were purchased from BD/PharMingen (San Diego, Calif.). Anti-mouse gamma interferon (IFN-γ), interleukin-2 (IL-2), IL-4, and IL-5 antibodies (3 μg/ml in coating buffer) were used to coat Multiscreen 96-well filtration plates (Millipore). Freshly isolated splenocytes (1.5 × 10
6 cells) were added to each well and stimulated with a mixture of two major histocompatibility complex class I (MHC-I) peptides (IYSTVASSL and LYEKVKSQL) or a pool of five MHC-II peptides (SFERFEIFPKE, HNTNGVTAACSH, CPKYVRSAKLRM, KLKNSYVNKKGK, and NAYVSVVTSNYNRRF) at a concentration of 10 μg/ml (
7,
23). The plates were incubated for 36 h at 37°C with 5% CO
2. Development and counting of cytokine ELISPOTs were performed as described previously (
13). Cytokine ELISA was performed as described previously (
26). Ready-Set-Go IL-6 and IFN-γ kits (eBioscience, San Diego, CA) were used for detecting cytokine levels in lung extracts following the manufacturer's procedures.
Passive immunization.
Sera from influenza VLP-immunized mice or from naïve mice were heated for 30 min at 56°C to inactivate complement. Serum was administered intranasally (50 μl per mouse) to naïve mice. After 2 h, mice were challenged with a lethal dose of live influenza PR8 virus (2,000 PFU per mouse, 10× LD50), and morbidity and mortality were monitored daily.
Statistics.
All parameters were recorded for individual mice within all groups. Statistical comparisons of data were carried out using the analysis of variance and Npar1-way Kruskal-Wallis test of the PC-SAS system. A P value of <0.05 was considered significant.
DISCUSSION
We have investigated detailed immune responses induced by influenza VLPs, including antibody isotypes, neutralizing activity, cellular immune responses, and induction of memory responses, which have not been investigated previously. In addition, our findings provide evidence that influenza VLPs containing HA, but not HA-negative M1 VLPs, can induce protective immunity against a lethal virus challenge with homologous as well as heterologous virus strains.
Immunization with recombinant influenza virus HA proteins was previously demonstrated to afford protection in chickens against challenge infection (
5). However, preparing HA proteins with high purity as a vaccine candidate on a large scale may require a high-cost manufacturing process. In this regard, the production and purification processes of influenza VLPs in insect cells can be relatively simple and easily scalable. Insect cells do not add sialic acids to the N-glycans during posttranslational modifications (
16), which explains why VLPs with HA are effectively released from the insect cell surfaces in this and other studies (
8). Nonetheless, it will be interesting to determine the effect of neuraminidase coexpression on VLP budding and yield in the insect cells. In addition, incorporating an additional component, neuraminidase, into VLPs would be an advantage for an influenza virus vaccine, although VLPs containing influenza virus HA and M were found to be effective in inducing protective immune responses in the absence of adjuvants.
Maintaining the VLP structure and functionality of HA are expected to be important for inducing protective immunity. It is likely that the HA molecules on the surfaces of the VLPs maintain the native-like conformation as evidenced by hemagglutination activity and cleavability of HA in VLPs. Disrupting the intact VLP structure and inactivating the hemagglutination activity of HA abrogated the humoral immune responses against A/PR8 virus and did not induce protective immunity. Therefore, the particulate nature and intactness of VLPs are critically important in inducing protective immunity and may be necessary in facilitating interaction with antigen-presenting cells leading to strong immune responses. In support of this notion, HIV VLPs were found to preferentially interact with CD11b+ monocyte/macrophage and B220+ B-cell populations in vitro (present study).
The current, parenterally administered influenza virus vaccine is considered to provide protective immunity against circulating viruses by inducing neutralizing antibodies directed against HA, although it is relatively less effective against antigenic variants within a subtype (
1). Serum antibodies induced by intranasal immunization with VLPs were found to have the capability to neutralize virus infectivity in vitro. We demonstrated that intranasal immunization with PR8 VLPs can confer 100% protection against PR8 as well as WSN strains using a 10× LD
50 dose without any clinical symptoms, and VLP-immunized mice also survived a lethal dose of both strains as high as 200× LD
50 with some weight loss (data not shown). The PR8 HA has approximately 91% amino acid homology with WSN HA on the basis of sequence analysis (GenBank accession numbers NC_004521 and ABF47955 for PR8 HA and WSN HA, respectively). Reflecting serological differences between A/PR8 and A/WSN, in addition to differences in lung viral titers, hemagglutination inhibition titers and neutralizing activity of PR8 VLP immune sera against A/WSN were two- to threefold less than those against A/PR8. Also, sera of mice infected with sublethal doses of WSN showed four- to eightfold differences in binding antibody titers against PR8 compared to those of the homologous antigen WSN, and this serologic difference was similarly observed when sera of mice infected with sublethal doses of PR8 were tested (data not shown). Nonetheless, we observed cross-reactive binding antibodies against A/WSN in VLP immune sera, and humoral and cellular immune responses were rapidly expanded upon lethal virus challenge with PR8 or WSN. Therefore, our studies demonstrate that influenza VLPs can be developed as a candidate vaccine. It will be of interest to determine whether the immune responses induced by influenza VLPs are cross-reactive with more distantly related strains within the same subtype. Influenza M1 VLPs can incorporate different subtype HAs, resulting in mixed influenza VLPs, and experiments to determine whether such phenotypically mixed VLPs can provide protection against influenza viruses of different subtypes are in progress.
An important goal of vaccination is to induce memory immune responses, which can provide long-term protective immunity. The cells responsible for memory response are T and B lymphocytes that can persist for long periods of time and can quickly be reactivated following infection. Induction of memory cells has been mostly investigated following live virus infection (
30,
39), but not much is known about memory responses after immunization with nonreplicating VLP vaccines. A fraction of memory B lymphocytes developed in the secondary lymphoid organs is routed to the bone marrow, resides there as long-lived plasma cells, and secretes antibodies, maintaining long-term serum antibody levels. We observed the presence of influenza virus-specific antibody-secreting plasma cells in the bone marrow of the VLP-immunized mice and found that VLP-immunized mice were protected equally well 4 weeks or 5 months after the final immunization. In addition, naïve mice that received intranasal administration of heat-treated immune sera collected 5 months postvaccination were completely protected from lethal virus challenge with either homologous or heterologous strains (Table
2), demonstrating the protective role of antibodies induced by VLPs. Taken together, these results suggest that influenza VLPs can induce the differentiation of B cells to long-lived plasma cells secreting antibodies, which may play a role in maintaining long-term protective immunity.
Lung cytokine-mediated immunoinflammatory reactions as well as infiltration of activated lymphocytes may be a cause of the morbidity and mortality associated with influenza virus infections (
21,
27,
35). We observed that high levels of IL-6 and IFN-γ were detected in naïve or HA-negative M1 VLP-immunized mouse lungs after challenge, whereas little or no proinflammatory cytokines were present in the lungs of the influenza HA VLP-immunized mice. Also, there seems to be a correlation between lung viral titers and the levels of inflammatory cytokines. This is consistent with a previous study demonstrating that high levels of lung viral titers and proinflammatory cytokines (IFN-α and IL-6) were found in the lungs of pigs with swine influenza virus infection (
35). Also, lymphocytes expressing CD69, an activation marker, were lower in influenza HA VLP-immunized mice than in naïve mice after challenge (data not shown). Thus, influenza VLP immunization can prevent immunopathologic lung inflammation upon influenza virus infection.
In summary, our results demonstrate that influenza VLPs can induce neutralizing antibodies and cellular immune responses, which can confer protection against lethal virus infection by homologous or heterologous strains within the same subtype. In addition, mucosal antibody and cellular immune responses induced by influenza VLPs were rapidly expanded upon challenge virus infection, inhibiting viral replication and lung inflammatory cytokine production. These results provide insight for developing effective prophylactic vaccines based on VLPs to fight pathogenic influenza viruses that pose a pandemic threat.