Influenza Virus-like Particles Containing HA, NA, and M1 Induced Protection in Chickens against a Lethal Challenge with the Highly Pathogenic H5N1 Avian Influenza Virus

Background: Intermittent outbreaks of highly pathogenic avian influenza virus (flu) infections illustrate the potential for pandemic spread of this deadly disease, thus making the development of sufficient supplies of safe vaccines a necessity. Influenza virus-like particles (Vlps) have been suggested as a promising non-egg or nonmammalian cell culture-based candidate for vaccination against flu. Vlps containing hemagglutinin have previously been shown to promote protection against homologous viral strains. In this report, we describe the development of an H5N1 flu Vlp vaccine involving only three flu viral structural proteins, (i.e., HA, NA, and M1), which were derived from an avian flu A/duck/Hokkaido/vac-1/2004 (H5N1) virus. The H5N1 Vlps produced from insect cells exhibited hemagglutination and neuraminidase activities and generated an immune response in BALB/c mice. We additionally performed viral challenge studies using chickens. Methodology and results: Vlps consisting of the hemagglutinin (HA), neuraminidase (NA), and matrix 1 (M1) proteins of A/duck/Hokkaido/vac-1/2004 (H5N1) were transferred using baculovirus within Spodoptera frugiperda (Sf9) cells. Mice were first immunized with Vlps, and the immune response was compared between animals vaccinated with HA and NA-HA-negative M1 Vlps. The IgG levels of HA-M1 Vlpand NA-M1 Vlp-treated groups were observed, and 5-fold higher levels of H5N1-specific antibodies were induced in the groups of mice immunized with HA-NA-M1 Vlp. The HA-NA and HA-NA-M1 Vlps vaccines induced IgG2a and IgG2b antibodies as well as IgG1 antibodies, indicating that both Th1 and Th2 immune responses were induced. Furthermore, NA-M1 immunization induced IgG and IgG1 isotype antibodies and led to low levels of IgG2a and IgG2b. Additionally, all chickens immunized with HA-NA-M1 Vlp were protected against deadly infections with the highly pathogenic avian flu virus A/ chicken/Yamaguchi/4/2004 (H5N1). Conclusions: Intramuscular administration of H5 Vlps conferred immunity against a deadly viral challenge. The H5 Vlp vaccine was more successful at raising Th1-biased protective responses, including IgG2a production. Thus, flu Vlps offer an important approach for immunization, especially if a pandemic occurs. Recognizing the current state of vaccination strategies, it is imperative to investigate the associated immunogenicity and defensive capabilities of H5 Vlps in comparison with inactivated whole virus and attenuated live H5N1 viruses. These results support the continued development of Vlps as a formulation for a vaccine against flu infection. Journal of Vaccines & Vaccination J o ur al of Va ccines & Vainati o n


Introduction
Human infection with the highly pathogenic avian influenza A H5N1 remains an indisputable and remarkably complex global health challenge despite the current focus on the pandemic outbreaks initiated by the rise of influenza A H1N1/09 virus [1][2][3]. The highest incidence of human infections caused by H5N1 occurred in Hong Kong in 1997 with eighteen cases and six deaths [4,5], and the virus has continued to be associated with a high case-fatality rate [3]. Recently, human infection has only occurred through close contact with animals (mainly poultry) infected with H5N1 virus. H5N1 infection does not currently satisfy the third pandemic criterion, sustained human-tohuman transmission, despite reports of transmission between humans [6,7]. Vaccine development has been a key component of pandemic preparedness plans [8][9][10][11]. Vaccination is an essential countermeasure to combat seasonal and pandemic flu. Flu vaccines have been prepared as live, inactivated whole virus generated in embryonated hens' eggs, using a strategy developed more than 60 years ago, or as split virions. Recently, mammalian cells have been utilized to produce flu vaccines in Europe. The major immunogens in these vaccines are the viral proteins hemagglutinin (HA) and neuraminidase (NA). Although production of these vaccines is feasible, it is also inefficient with respect to time and manufacturing constraints. Therefore, when a flu pandemic arises, it takes nearly 4-6 months after the identification of the virus to prepare the first distributable quantities of vaccine. Early studies with inactivated split virion vaccines that comprise H5N1 strains demonstrated that such vaccines could be safely administered to human subjects [12][13][14][15]. In fact, many countries wish to assemble a stockpile of H5N1 vaccine. However, H5N1 vaccine stockpiles will be severely limited by the absence of sufficient H5 vaccine antigen because of limited worldwide production. Recently, a few noteworthy strategies have been used to increase antigen production. One promising engineering technique uses recombinant noninfectious virus-like particles (Vlps) that present structurally native, immunologically relevant viral antigens. Various studies have demonstrated the behavior, immunogenicity, and defensive capability of flu Vlps. These Vlps comprised the HA, NA, and matrix 1 (M1) proteins, which are processed by the synchronous expression of these genes in an insect cell baculovirus expression system [16][17][18][19][20][21][22][23]. We have previously shown that intranasal immunization with a recombinant baculovirus containing the flu virion HA protein protects mice against a lethal challenge with influenza A and B viruses [24]. In this report, we describe the development of an H5N1 flu Vlp vaccine that contained three flu infection structural proteins, HA, NA, and M1, derived from the avian flu A/duck/Hokkaido/vac-1/2004 (H5N1) virion [25]. The H5N1 Vlps derived from insect cells showed hemagglutination and neuraminidase activities and generated immune responses in BALB/c mice. We similarly performed experimental studies utilizing chickens.

Generation of recombinant baculoviruses
The M1 gene was cloned into the pAcDB3 bacmid exchange vector (BD Biosciences) (which had previously been processed with XbaI) downstream of the polyhedrin promoter, as an XbaI DNA fragment (1.0 kb). The resultant baculovirus exchange plasmid containing influenza genes was designated pAcDB3-M1. The HA and M1 genes were cloned into a bacmid exchange vector (SmaI-processed AcDB3-M1 plasmid DNA) downstream of the p10 promoter, as SmaI DNA fragments (1.8 kb) (Figure 1). The NA gene was then cloned into the same vector (after EcoRI-BglII processing of the pAcDB3-M1 plasmid DNA) downstream of the p10 promoter, as EcoRI-BglII DNA pieces (1.4 kb) ( Figure 1). The two resultant baculovirus exchange plasmids containing the flu virus genes were designated pAcDB3-HA-M1 and pAcDB3-NA-M1.
Finally, a bacmid exchange vector that allowed the expression of each of the three flu genes was constructed, with the genes cloned into the SmaI-processed pAcDB3-NA-M1 plasmid DNA downstream of the p10 promoter, as SmaI DNA sections (1.8 kb) ( Figure 1). This process resulted in a plasmid, pAcDB3-HA-NA-M1, that encoded the HA, NA, and M1 genes, each within its own expression cassette. The vector also incorporated a p10 and polyhedrin promoters and translation termination signal. The DNA fragment from pNVAX1250 containing the NA, M1, and HA expression cassettes was then transferred into bacmids, and Spodoptera frugiperda (Sf9) insect cells (ATCC CRL-1711) were transfected with the bacmids to generate recombinant baculovirus [17]. Titers of the recombinant baculovirus stocks were determined by agarose plaque tests in Sf9 cells. Recombinant baculoviruses were utilized to infect Sf9 cells. Vlps were harvested at 72 h posttransfection. Recombinant Vlps were purified by a 20-60% (w/v) discontinuous sucrose density gradient and sedimented by ultracentrifugation for 16 h at 27,000 rpm and 4°C. Fractions (1 ml) were collected from the top of each gradient using a 1-ml syringe. Expression of HA, NA and M1 proteins in the purified Vlp preparation was confirmed by Western blot analysis, as described by Pushko et al. [17]. Hemagglutination titration assays were used to measure HA content, as described in the general procedure by Donald and Isaacs [26]. The purified Vlps were stored under refrigeration (4-8°C).

Hemagglutinin (HA) and hemagglutination inhibition (HI) tests
Two-fold serial dilutions of flu Vlps in PBS were prepared and incubated at 25°C for 2 h with 0.5% (v/v) chicken red blood cells. The degree of hemagglutination was assessed, and the best dilution for agglutinating 0.5% (v/v) chicken red blood cells was determined [26]. An HI test was completed according to previously described methods [27], utilizing four hemagglutination units of β-propiolactoneinactivated avian flu A/duck/Hokkaido/vac-1/2004 (H5N1) combined with serial dilutions of antibodies from Vlp-vaccinated BALB/c mice and then assessing the agglutination of chick red blood cells.

Serologic test for the determination of antibody responses
Female BALB/c mice (6 weeks old) were purchased from Nippon SLC (Hamamatsu, Japan) and housed under humane conditions consistent with the guidelines and regulations of our institutional committee. Mice (7 per group) were intramuscularly immunized with 10 µg of Vlps in 50 µl of PBS twice at 3-week intervals (weeks 0 and 3). Blood samples were collected retro-orbitally prior to vaccination and at various time points after the final immunization (weeks 1 and 4). The blood samples were allowed to clot prior to centrifugation, then serum was isolated and stored at -20°C until titration.
Peripheral blood samples were collected before and after immunization at various time points. The H5N1 virion-specific antibody titers were determined by ELISA, as previously described [19]. Briefly, an inactivated flu H5N1 virion was used as a coating antigen on 96-well microtiter plates (Nunc, Thermo Fisher Scientific, MA, USA), which were then incubated overnight at 4°C. The wells were then washed three times with PBS containing 0.05% Tween 20 (PBST) and blocked with PBST containing 3% BSA for 1 h at 37°C. Serially diluted serum samples were included and incubated for 2 h at 37°C. Finally, HRP-conjugated anti-rodent goat IgG (GE Healthcare, London, UK), IgG1 (Southern Biotechnology Associates, AL, US), IgG2a (Invitrogen, California, US) and IgG2b (Invitrogen) were utilized as secondary antibodies. The ELISA endpoint titers were determined as the most notable dilutions that yielded optical densities greater than the mean optical density plus 3 standard deviations of analogously diluted negative control sera. To confirm the antibody titers, absorbance was read at 450 nm.

Animals and experimental design
The vaccine was prepared by emulsifying 250 µg of the Vlp antigen product with Freund's Complete Adjuvant (FCA, BD Difco, Michigan, USA) at a ratio of 30:70 (v/v). To assess the vaccine effectiveness, chickens were challenged intranasally with a highly pathogenic avian flu virus, A/chicken/Yamaguchi/7/2004 (H5N1) [28]. Four-week-old White Leghorn SPF chickens were divided into two groups (3 per group). Chickens were immunized three times at 2-week intervals (weeks 0, 2, and 4) with 250 µg of Vlp antigen plus adjuvant via intramuscular injections. As a non-immunized control group, 3 additional White Leghorn chickens were injected with an emulsion of distilled water in FCA in the same proportion as the Vlp antigen. Three weeks after the final round of immunization, under Animal Biosafety Level 3 conditions, the chickens were challenged intranasally with the 1,000 CLD50 of A/chicken/Yamaguchi/7/2004 (H5N1). To confirm the reproducibility of the challenge at day 2 post-challenge (p.c.), intraoral and cloacal swabs were gathered from each chicken, as previously described [13]. The clinical data were computed as the mean score for each fowl for each observation. Blood specimens were gathered at 0, 2, 4, 6, and 7 weeks post-vaccination. Intraoral and cloacal swabs were performed on day 2 p.c. to assess viral shedding. Animal experiments were approved by the Institutional Animal Care and Use Committee of the Graduate School of Veterinary Medicine, Hokkaido University (endorsement number: 10-0007), and all experiments were performed in accordance with the guidelines of this committee.

Production of the Vlp vaccine for A/duck/Hokkaido/Vac-1/2004 (H5N1)
Recombinant baculoviruses were generated for individual expression of HA, NA, or M1 proteins as well as for the co-expression of these proteins (Figure 1 Figure 2). The HA titers of HA-NA-M1 Vlps and HA-M1 Vlps were 6.8×108 HA and 3.4×108 HA, respectively. Additionally, the size and morphology of influenza Vlps were examined by electron microscopy ( Figure 2B). The morphology of Vlps resembled that of influenza virus particles with spikes on their surfaces, which are characteristic of flu HA proteins on virions. The particle sizes ranged from approximately 80 to 120 nm. Taken together, these results demonstrate that Sf9 cells infected with rBVs expressing HA, NA and M1 generated particles resembling influenza virions in terms of morphology and size.

The immune response elicited by Vlp vaccination
To determine whether influenza Vlps induce immune responses specific to flu HA, mice were intramuscularly immunized with 10 µg of flu Vlps containing HA-M1, NA-M1, HA-NA-M1 or an HAnegative Vlp containing M1 alone (referred to as HA-M1 Vlp, HA-NA-M1 Vlp, and M1 Vlp, respectively) at 3-week intervals. The serum immune response stimulated after two rounds of immunization was assessed at various time points (weeks 0, 4, and 8). Sera diluted 100 to 1000-fold was utilized to verify the A/duck/Hokkaido/vac-1/2004 (H5N1)-specific total IgG by ELISA. At week 8 post-vaccination, the total IgG levels of groups treated with HA-M1-Vlp or NA-M1 Vlp were observed, and 5-fold higher levels of H5N1-specific antibodies were detected in HA-NA-M1 Vlp-immunized mice ( Figure 3A). At weeks 4 and 8 post-immunization, the levels of antibodies were unaltered. No significant levels of antibodies specific for A/duck/Hokkaido/vac-1/2004 (H5N1) were detected in the groups of mice vaccinated with HA or HA-NA-negative M1 Vlps. Limited data are available on the protective role of neuraminidase (NA), the second major glycoprotein. A subsequent study reported that flu Vlps containing M1 and NA are capable of inducing immunity to homologous as well as antigenically dissimilar flu A strains [29]. The mice immunized with NA Vlps were 100% protected against lethal infections by the homologous A/ pr/8/1934 (H1N1) and heterosubtypic A/Philippines/1982 (H3N2) infections. In this study, antibodies were analyzed in HA-M1-Vlpimmunized groups.
To further characterize the immune responses, we assessed the pattern of isotypes induced by immunization with Vlps. IgG1, which is indicative of a T helper type 2 (Th2)-biased response, was discovered to be the prevailing antibody subclass raised in the vaccinated groups at week 8 post-immunization ( Figure 3B). Moreover, HA-M1 and HA-NA-M1 Vlps groups induced IgG2a and IgG2b antibodies at week 8 post-immunization, showing Th1-biased responses ( Figure 3C). The HA-M1 and HA-NA-M1 Vlp vaccines raised IgG2a and IgG2b antibodies as well as the IgG1 subclass, indicating that both Th1 and Th2 immune responses were elicited (Figures 3B and 3C). However, higher levels of H5N1-specific antibodies were induced in HA-NA-M1 Vlp-immunized mice than in HA-M1 Vlp-vaccinated mice. Finally, considerable levels of total IgG and IgG1 isotype antibodies, but low levels of IgG2a and IgG2b antibodies, were observed after immunization with NA-M1 Vlps ( Figure 3C).

Immunization with HA-NA-M1 Vlps protects chickens against an H5N1 challenge
To evaluate the immunogenicity of HA-NA-M1 Vlps, fourweek-old White Leghorn SPF chickens were placed into two groups. Chickens were intramuscularly vaccinated three times (weeks 0, 2, and 4) at 2-week intervals with 250 µg of Vlp vaccine with adjuvant. At 2, 4, 6, and 7 weeks after priming, we determined the serum levels of HI titers ( Table 1). The serum HI antibody titers to the immunization strain were 4 at 2 weeks after the first immunization and reached 32-64 at 3 weeks after the third vaccination ( Table 1). The survival rates of the chickens challenged with A/chicken/Yamaguchi/7/2004 (H5N1) are shown in Figure 4. None of the immunized chickens displayed any observable disease after viral challenge. All non-immunized chickens challenged with A/chicken/Yamaguchi/7/2004 (H5N1) died at 2 days p.c. (Figure 4). To assess the potential of the Vlp vaccine to induce immunity and counteract viral shedding, we attempted to retrieve the virus from swabs of the immunized and non-immunized chickens at 2 days p.c. with A/chicken/Yamaguchi/7/2004 (H5N1) ( Table 2). The infectivity titers of the recovered viruses from immunized chickens were fundamentally lower than those of non-immunized chickens.

Discussion
Vlps provide a promising platform as vaccine candidates as they present an antigen in an immunogenic form. One favorable element of the Vlp vaccine approach is that, as a virus-mimicking molecules with various viral antigens and epitopes, Vlps enable a diverse set of protective responses [27,30]. Flu Vlp vaccines have been demonstrated to be exceptionally immunogenic and protective in various animal models [27,[29][30][31][32]. Recently, Prel et al., in addition to Tao et al., presented a Vlp vaccine utilizing a triple baculovirus recombinant that encoded HA, NA, and M1 proteins from H5N3 AIV and demonstrated the protection imparted by these Vlps in Muscovy ducks [33,34]. In the present study, we assessed an H5N1 flu Vlp vaccine containing only three flu virus structural proteins, i.e., HA, NA, and M1, in chickens.
The application of a Vlp vaccine might alleviate the health concerns associated with live-attenuated and inactivated whole virus vaccines. For example, there is evidence that the appearance of the highly pathogenic H5N2 of American lineage in Japan resulted from the exchange of the live virus from an incompletely inactivated or contaminated vaccine [35]. In light of the noninfectious nature of Vlps and their lack of viral genomic material, the Vlp vaccine presents an attractive option as a vaccine candidate [31]. Moreover, throughout AIV movement in the poultry population, HA and NA surface antigens undergo systematic amino acid substitutions, which could lead to evasion of the viral resistance conferred by prior vaccination. Because the Vlp vaccine approach permits rapid vaccine preparation, this approach could be utilized to easily redesign the vaccine to reflect transformed or newly arisen strains. The neutralizing reactions afforded after two rounds of immunization were assessed at various time points (weeks 0, 4, and 8). At week 8 post-immunization, total IgG was observed in groups that received HA-M1-Vlp and NA-M1 Vlps, and 5-fold higher levels of H5N1-specific antibodies were induced in the groups of mice immunized with HA-NA-M1 Vlps ( Figure 3A). No significant levels of antibodies specific for A/duck/Hokkaido/vac-1/2004 (H5N1) were detected in groups of mice immunized with HA or HA-NA-negative M1 Vlps. In contrast to HA, the role of NA in vaccine-induced protection has not been investigated, even though both HA and NA are glycoproteins expressed on the surfaces of flu virions. Among the factors that have restricted the investigation of NA-specific vaccines is the trend of immune response towards HA after flu vaccination. The overwhelming impact of HA as an antigen was realized when HA and NA were presented in similar viral particles [36,37]. Flu Vlps are a remarkable vehicle for presenting NA as the major immunogen in the coating of a benign particle. In this time-course study, we explored the effectiveness and immunogenicity of Vlps containing NA. Flu M1 Vlps containing NA were immunogenic and induced functional NA-specific antibodies ( Figure 3A). Regardless of the induction of M1-specific antibodies, the contribution of flu M1 to flu resistance is insignificant [38,39].
We assessed the isotype classes induced by Vlp immunization. The IgG1 antibody, which reflects a Th2-biased response, was discovered to be the predominant subclass in the immunized groups ( Figure  3B). In addition, the groups receiving HA-M1 and HA-NA-M1 Vlps predominantly exhibited IgG2a and IgG2b antibodies, indicating Th1-biased responses ( Figure 3C). The HA-M1 and HA-NA-M1 Vlps vaccines elicited IgG2a and IgG2b antibodies as well as the IgG1 subclass, characteristic of both Th1 and Th2 immune responses ( Figures  3B and 3C). Nevertheless, higher levels of H5N1-specific antibodies were induced in the groups of HA-NA-M1 Vlp-immunized mice than HA-M1 Vlp-immunized mice. Furthermore, substantial levels of total IgG and IgG1 isotype antibodies were detected in the NA-M1 Vlp group, whereas only low levels of IgG2a and IgG2b antibodies were detected ( Figure 3C).
An analysis of antibody isotypes in immune sera revealed that the immunization of mice (BALB/c) with NA-M1 Vlps predominantly induced IgG1 antibodies. Soluble recombinant HA protein antigens are also known to induce IgG1-dominant antibody responses compared to whole inactivated virus or HA-M1 Vlp antigens, which predominantly induce IgG2a isotype antibodies [21,22]. It is probable that NA Vlp without HA, which therefore lacks hemagglutination activity, will be less effective in inducing IgG2a-dominant immune responses than HA Vlp, which has hemagglutination activity.
In the lethal homologous HPAI virus challenge, the group immunized with HA-NA-M1 Vlp demonstrated 100% protection. In contrast, all chickens in the control group died at 2 days p.c. ( Figure  4). The challenge virus was detected in several of the intraoral swab samples of chickens immunized with HA-NA-M1 Vlp but not in the cloacal swabs (Table 2). This result indicated that the present vaccine did not completely eliminate the virus, but caused a critical decrease in excessive viral shedding to the chickens, as the administration of an oil-based adjuvanted vaccine has comparable efficacy [40]. Overall, the baculovirus-based Vlp vaccine, HA-NA-M1 Vlp, provided protection to chickens sufficient to prevent observable disease and decreased the amount of viral shedding induced by infection with currently common H5 HPAI viruses.
In conclusion, an intramuscular injection of H5 Vlps provided excellent protection against a lethal viral challenge. The H5 Vlp antibody was more successful in promoting protective Th1-type responses, including IgG2a production. Therefore, flu Vlps could become a significant focus for vaccine strategies, especially if a pandemic occurs. Recognizing the present state of the available vaccines, it is important to carefully examine the relative immunogenicity and protective efficacy of H5 Vlps in comparison with inactivated whole virus and attenuated live H5N1 immunizations.    a Blood samples were collected at 0, 2, 4, 6, and 7 weeks after initial inoculation