All-in-One Bacmids: an Efficient Reverse Genetics Strategy for Influenza A Virus Vaccines

ABSTRACT

Vaccination is the first line of defense against influenza virus infection, yet influenza vaccine production methods are slow, antiquated, and expensive as a means to effectively reduce the virus burden during epidemic or pandemic periods. There is a great need for alternative influenza vaccines and vaccination methods with a global scale of impact. We demonstrate here a strategy to generate influenza A virus in vivo by using bacmid DNAs. Compared to the classical reverse genetics system, the “eight-in-one” bacmids (bcmd-RGFlu) showed higher efficiency of virus rescue in various cell types. Using a transfection-based inoculation (TBI) system, intranasal delivery to DBA/2J and BALB/c mice of bcmd-RGFlu plus 293T cells led to the generation of lethal PR8 virus in vivo. A prime-boost intranasal vaccination strategy using TBI in the context of a bcmd-RGFlu carrying a temperature-sensitive H1N1 virus resulted in protection of mice against lethal challenge with the PR8 strain. Taken together, these studies provide proof of principle to highlight the potential of vaccination against influenza virus by using in vivo reverse genetics.

IMPORTANCE Vaccination is the first line of defense against influenza virus infections. A major drawback in the preparation of influenza vaccines is that production relies on a heavily time-consuming process of growing the viruses in eggs. We propose a radical change in the way influenza vaccination is approached, in which a recombinant bacmid, a shuttle vector that can be propagated in both Escherichia coli and insect cells, carries an influenza virus infectious clone (bcmd-RGFlu). Using a surrogate cell system, we found that intranasal delivery of bcmd-RGFlu resulted in generation of influenza virus in mice. Furthermore, mice vaccinated with this system were protected against lethal influenza virus challenge. The study serves as a proof of principle of a potentially universal vaccine platform against influenza virus and other pathogens.

INTRODUCTION

Influenza epidemics continue to be a major disease burden in humans, with approximately 5 to 15% of the population infected on an annual basis (1). Severe influenza disease is estimated to affect 3 to 5 million people worldwide and is associated with 250,000 to 500,000 deaths annually (2). Efforts to predict the emergence of a pandemic virus through surveillance did little to curtail the spread of the swine-origin influenza pandemic in 2009 (3).

Vaccination currently remains the most effective protection against influenza virus infection, and it is considered the first line of defense against illness. Human influenza vaccines are produced as either split virion-inactivated (killed virus [KV]) vaccines or live attenuated influenza virus (LAIV) vaccines (4). These vaccines, however, are reformulated often, due to the virus' ability to undergo antigenic drift that escapes the immunological pressure developed against previous strains (5). A major drawback in the preparation of LAIV and KV vaccines is that production relies on a heavily time-consuming process of growing the viruses in eggs or tissue culture cells. Additionally, since most human influenza virus strains grow poorly in these systems, vaccine strains are produced from reassortants that generally carry the surface gene segments from the candidate virus and other segments from a high-growth donor virus. Reverse genetics (RG) has certainly improved the ability to generate such high-growth reassortants (6, 7); however, growing influenza virus in eggs or tissue culture may result in adaptive changes on the viral surface proteins, resulting in antigenic mismatch. LAIV vaccines have an advantage over KV vaccines because they produce broader responses by stimulating both the humoral and the T-cell arms of the immune system (8, 9). The 2009 pandemic H1N1 virus (pH1N1), however, highlighted the fact that these traditional vaccine production systems are too slow to significantly ameliorate or alter the impact of a pandemic, given that the initial pH1N1 vaccine candidates were not well suited for growth in eggs (10). Furthermore, LAIV vaccines are not approved for children under 2 years old and are not as effective in the elderly, and these represent two groups at high risk of influenza virus infection. LAIV vaccines are contraindicated for those with egg allergies. Alternatively, egg-free influenza vaccine strategies have been investigated, including strategies using recombinant viral proteins, recombinant viruses, and virus-like particles (VLPs) (11). FluBlok, a baculovirus-based recombinant hemagglutinin (HA) influenza vaccine, is the only influenza vaccine approved for human use that does not rely on traditional production systems, but it must also undergo reformulation as a result of antigenic drift.

DNA vaccines are undoubtedly the cheapest vaccines to produce. Despite its simplicity and solid safety history, DNA vaccination has been hampered by its low immunogenicity in humans (12). The limitations of the immunogenicity of DNA vaccines could be overcome by providing alternative machinery capable of amplifying the antigen of interest. In this report, we demonstrate the de novo generation of influenza viruses in vivo (mice) by using a transfection-based inoculation (TBI) method and a recombinant bacmid as the delivery vehicle for the influenza virus RG clone. These studies could pave the way for refocusing DNA vaccine efforts with enhanced antigen production in vivo.

MATERIALS AND METHODS

Viruses and plasmids.

RG expression elements for the generation of influenza virus mRNA and virus-like RNAs by using RNA polymerase II (pol II; pol2) and human RNA polymerase I (hpol1) promoters present in the pDP2002 vector (a derivative of pHW2000) have been previously described (13, 14). The RG 8-plasmid system for A/Puerto Rico/8/34/Mount Sinai (H1N1; PR8), which is based on the pDZ vector, was a kind gift from Peter Palese, Mount Sinai School of Medicine, New York, NY (15). The pDZ-based PR8 gene segments were subcloned into the pDP2002 vector in order to maintain consistency with other RG clones described below. The pDP2002-based RG 8-plasmid system for the temperature-sensitive A/turkey/Ohio/313053/2004 (H3N2) virus with and without a C-terminal HA tag in PB1 (Ty04att and Ty04ts, respectively) has been previously described (16). The pDP2002-based surface gene segments from A/Vietnam/1203/2004 (H5N1) with an HA gene without the polybasic cleavage site (∂H5N1) have also been previously described (16, 17). Viruses were amplified in 10-day-old specific-pathogen-free (SPF) embryonated chicken eggs following standard techniques for growth of influenza viruses. The attenuated H1N1Ty04att vaccine virus carrying the HA and NA segments from the PR8 virus in the Ty04att backbone was prepared following a previously described protocol (16). Virus stocks were prepared and frozen at −80°C until use. The 50% tissue culture infective dose (TCID₅₀) titers were determined in MDCK cells by the Reed and Muench method as described previously (18).

Construction of recombinant bacmids.

The plasmid pFastBac1 (Life Technologies, Grand Island, NY) was modified by deletion of the polyhedrin promoter (plh) to generate the cloning vector p∂FB1. The RG competent internal gene cassettes from the PR8 strain were sequentially subcloned from the pDP2002 vector into p∂FB1 by using appropriate restriction sites to produce p∂FP6PR8 (∼21.9 kb). p∂FP6PR8 was further modified with insertion of the thymidine kinase gene (tk) flanked by lambda phage attR recombination elements (attTK) with a size of 2,247 bp, produced from an overlapping PCR from the Baculo-Direct vector (Life Technologies). The resulting plasmid was designated p∂FP6attTK_PR8. A similar strategy was used to generate a shuttle vector carrying the internal gene cassettes from Ty/04ts and the attTK element in order to generate p∂FP6attTK_Ty04ts. For cloning, the Quick ligation kit (New England BioLabs, Ipswich, MA) or DNA ligation kit for long fragments (TaKaRa Bio Inc., Japan) was used along with either One Shot TOP10 chemically competent E. coli cells (Life Technologies) or One Shot ccdB Survival 2 T1R competent E. coli cells (Life Technologies). At each cloning step, plasmids were verified by sequencing and gene functionality analysis via reverse genetics or minireplicon assays (data not shown). Sequencing of recombinant plasmids was performed with a combination of universal primers (19), custom primers, and the BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems, Foster City, CA) on a 3500 Genetic Analyzer (Applied Biosystems) according to the manufacturer's instructions. Sequence analysis was performed using software available through the Lasergene package (DNAstar Inc., Madison, WI). For minireplicon assays, plasmids pcDNA774PB1, pcDNA762PB2, pcDNA787PA, and pcDNA693NP (20) were used, depending on the gene to be tested in the p∂FB1 vector. The pHW72-EGFP plasmid carrying the influenza virus enhanced green fluorescent protein (EGFP) reporter replicon was a gift from Robert Webster, St. Jude Children's Research Hospital, Memphis, TN (21).

The plasmid pENTR-1A dual vector was used to generate multiple plasmids carrying the RG competent surface (HA and NA) gene cassettes (Table 1 and Fig. 1). In addition, one of these constructs was further modified to carry an EGFP gene cassette under the control of a cytomegalovirus (CMV) promoter. The following plasmids were constructed: pE64P (H1N1_PR8), pE46VL (∂H5N1_VN1203), and pE4E6VL (∂H5N1_VN1203/EGFP). Plasmids were verified by sequencing and gene functionality by reverse genetics as described above, prior to further genetic manipulation.

TABLE 1.

Recombinant bcmd-RGFlu DNAs generated in this study

DNA segment	Acronym	Subtype	Donor vector	Entry plasmid	Eight-in-one plasmid (segment order)a	Six- or eight-in-one bacmidb
Internal gene backbone
A/Puerto Rico/8/1934 (H1N1)	PR8	H1N1	p∂FP6PR8	N/Ac	N/A	bcmd-P6PR8
A/Turkey/Ohio/313053/04 (H3N2) ts	Ty04ts	H3N2	p∂FP6Ty04ts	N/A	N/A	bcmd-P6Ty04ts
Surface gene segments
A/Puerto Rico/8/1934 (H1N1)	PR8	H1N1	p∂FP6attTKPR8	pE64P	p∂FP8P (12643578)	bcmd-P8PR8 (H1N1PR8)
A/Vietnam/1203/2004 (H5N1) mod.	VN1203	∂H5N1	p∂FP6attTKPR8	pE46VL	p∂FP8LP (12643578)	bcmd-P8LP (∂H5N1VN1203)
A/Vietnam/1203/2004 (H5N1) mod.	VN1203	∂H5N1	p∂FP6attTKPR8	pE4E6VL	p∂FP8LPE (126E43578)	bcmd-P8LPE (∂H5N1VN1203/EGFP)
A/Puerto Rico/8/1934 (H1N1)	PR8	H1N1	p∂FP6attTKTy04ts	pE64P	p∂FP8T11 (35642817)	bcmd-P8Ty04ts (H1N1PR8)

^aUnderlined segment numbers indicate surface gene segment order and, where indicated, presence of EGFP gene (E).

^bSix-in-one bacmids were constructed from the internal gene backbone segments, and eight-in-one bacmids were constructed from surface gene segments.

^cN/A, not applicable.

FIG 1.

Schematic representation of the construction of a bcmd-RGFlu vector. (A) A prototypical strategy for construction of a bcmd-RGFlu vector starts with the cloning of RG competent units (bidirectional RNA pol1 and pol2 promoters flanking a cDNA copy) of the internal gene segments of a given influenza virus strain and further modification with the incorporation of lambda phage attR recombination units flanking the thymidine kinase gene, p∂FP6attTK. (B) In a separate set of reactions, RG competent units for the HA and NA gene segments are cloned into the pENTR-1A vector to generate the pE46 vector. (C) Using the Gateway LR recombination system and the plasmids p∂FP6attTKand pE46, the plasmid p∂FP8 is obtained; this plasmid contains 8 RG influenza virus competent units. (D) The p∂FP8 vector contains Tn7 recombination signals flanking the RG influenza virus clone that are used for the transposition of the RG units into a bacmid in DH10bac E. coli cells. Picking white colonies and three rounds of antibiotic selection are used to identify bacmids containing an RG influenza virus infectious clone. The resulting bcmd-RGFlu is approximately 170 kb long. Transfection of bcmd-RGFlu into appropriate cells leads to de novo influenza virus synthesis.

The Gateway LR cloning method (Life Technologies) was used to transfer the gene cassettes cloned in the pENTR-1A vector into the p∂FP6attTK_PR8 and p∂FP6attTK_Ty04ts vectors in order to generate plasmids with a full set of reverse genetics competent influenza virus gene segments (Table 1 and Fig. 1). Subsequently, recombinant bacmids were produced by using MAX Efficiency DH10Bac competent E. coli cells (Life Technologies) following the manufacturer's directions. Transposition of the above recombinant shuttle plasmids was selected on 1.2% LB–agar plates containing Bluo-Gal (Sigma-Aldrich, St. Louis, MO), isopropylthio-β-galactoside (IPTG; Sigma), gentamicin (Sigma), kanamycin (Sigma), and tetracycline hydrochloride (Sigma) according to the instructions for the Bac-to-Bac expression system (Life Technologies). After three rounds of negative selection with antibiotics, colorless colonies were picked and bacmid DNAs were extracted and identified by PCR using appropriate primer sets (sequences are available from the authors upon request). The positive recombinant bacmids were designated bcmd-P6PR8, bcmd-P8PR8 (H1N1_PR8), bcmd-P8LPE (∂H5N1_VN1203), bcmd-P8LPE (∂H5N1_VN1203/EGFP), and bcmd-P8Ty04ts (H1N1_PR8). A full list of plasmids and primers used and generated during cloning is available from the authors upon request. Recombinant bacmid DNAs were extracted by using the PureLink HiPure plasmid maxiprep kit (Life Technologies) and then purified with the UltraClean endotoxin removal kit (MO Bio Laboratories Inc., Carlsbad, CA).

Eukaryotic cells.

Madin-Darby canine kidney epithelial (MDCK) cells, African green monkey kidney Vero cells, human lung carcinoma A549 cells, and swine kidney PK-15 cells were maintained in modified Eagle's medium (MEM; Sigma) containing 10% fetal bovine serum (FBS; Sigma). Mouse embryonic fibroblast NIH 3T3 cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Gibco, Gaithersburg, MD) with 10% FBS. Human embryonic kidney 293T cells (HEK293T) were cultured in Opti-MEM I (Gibco) containing 5% FBS. Mus musculus lung adenoma LA-4 cells were maintained in Kaighn's modification of Ham's F-12 medium (ATCC, Manassas, VA). Abelson murine leukemia virus-transformed RAW264.7 macrophages were maintained in DMEM containing 10% FBS. Normal human bronchial primary epithelial (HBEM) cells (catalog number 502K-05a; Cell Applications Inc., San Diego, CA) were grown in HBEM growth medium (Cell Applications) as described previously (22).

Transfection of tissue culture cells.

Transfections of reverse genetics plasmids were performed essentially as described previously (23). Transfections with recombinant bacmid DNAs were optimized based on the parameters discussed in detail in the paragraphs below. Cell suspensions were seeded into 6-well tissue culture plates and incubated at 37°C overnight before transfection. For the 293T/MDCK cocultured cells (10:1), 1 × 10⁶ cells were transfected with 5 μg of bacmid DNA or 1 μg of each plasmid in the RG 8-plasmid set. At 12 h posttransfection (hpt), the transfection medium was removed and replaced with 1 ml of Opti-MEM medium supplemented with l-(tosylamido-2-phenyl)ethyl chloromethyl ketone-treated bovine trypsin (TPCK-trypsin; 1 μg/ml; Worthington, Lakewood, NJ), and cultures were incubated for up to 144 hpt (24). Transfection and posttransfection steps for other cells (Table 2) proceeded similarly except that different concentrations of TPCK-trypsin were used, as follows. For HBEC, LA-4, RAW264.7, NIH 3T3, and A549 cells, 1 × 10⁶ cells in 6-well plates were supplemented at 12 hpt with serum-free medium containing 0.5 μg/ml TPCK-trypsin. Vero cells were incubated with Opti-MEM containing 0.5 μg/ml TPCK-trypsin, which was replenished daily. PK-15 cells were supplemented with TPCK-trypsin at a concentration of 0.25 μg/ml. Tissue culture supernatants were collected at the times indicated and with blind passage, as shown, in either 10-day-old SPF embryonated chicken eggs or MDCK cells or both, to monitor for the presence of rescued viruses.

TABLE 2.

Rescue efficiency of bcmd-P8LPE (∂H5N1_VN1203/EGFP) in various cell lines^a

Cell line	Origin	Virus titer at 72 hpt (TCID50/ml)
A549	Human lung carcinoma	1.0 × 106
PK-15	Swine kidney	1.0 × 105b
HBEC	Human primary respiratory epithelium	1.0 × 107b
LA-4	Mouse lung adenoma	Negative
RAW264.7	Mouse macrophage	Negative
NIH 3T3	Mouse embryonic fibroblasts	Negative

^abcmd-P8LPE (∂H5N1_VN1203/EGFP) was transfected into the indicated cells as described in Materials and Methods. Virus titers in tissue culture supernatants from transfected cells at 72 hpt were analyzed in MDCK cells.

^bBlind passage of 72-hpt supernatants was needed in order to detect live virus from transfected cells.

IFAs.

Recombinant bcmd-RGFlu DNAs were verified for influenza virus protein expression by using specific antibodies in immunofluorescence assays (IFAs) as described previously (24). Briefly, cells were transfected with 5 μg bcmd-RGFlu DNA in a 6-well plate for 24 h and then fixed with 100% cold methanol for 10 min. After washing the cells three times with phosphate-buffered saline (PBS), IgA-type monoclonal antibody (MAb) DPJY01 against the HA of H5N1 viruses (diluted 1:500) (25) was added and the mixture was incubated for 30 min at room temperature (RT). The cells were then washed three times with PBS, and then a second antibody, goat anti-mouse Ig (H+L)–fluorescein isothiocyanate (FITC; diluted 1:2,500; Southern Biotech, Birmingham, AL) was added and this mixture was incubated for 30 min at RT. Finally, cells were washed again three times with PBS and examined under an Axiophot Photomicroscope produced by Carl Zeiss (excitation wavelengths [λEx] of 488/543 nm and emission wavelengths [λEm] of 522/590 nm for 100 ms).

Western blot analysis.

At 24 hpt, transfected cells were lysed with Laemmli sample buffer and processed for Western blotting as described previously (26). Antibodies used included rabbit anti-NP antibody (1:1,500; Novus Biologicals LLC, Littleton, CO), goat anti-PB1 VK-20 (N terminal [N-ter]) (1:150; Santa Cruz Biotechnology Inc., Dallas, TX), goat anti-PB2 VN-19 (N ter) (1:200; Santa Cruz Biotechnology), mouse anti-NS1–23-1 (N-ter) (1:250; Santa Cruz Biotechnology), rabbit anti-PA (C-ter) (1:2,000; Gentex Inc., Irvine, CA), mouse anti-HA (1:2,000; eEnzyme LLC, Gaithersburg, MD), and rabbit anti-EGFP antiserum (1:1,500; Life Technologies). Secondary antibodies included horseradish peroxidase (HRP)-conjugated donkey anti-goat IgG (1:2,500; Santa Cruz Biotechnology), HRP-conjugated goat anti-rabbit IgG (1:4,000; Southern Biotechnology), and HRP-conjugated goat anti-mouse IgG (1:5,000; Southern Biotechnology).

Transfection-based inoculation of mice.

Animal studies were approved by the Institutional Animal Care and Use Committee of the University of Maryland, College Park. Studies were performed in two different mouse strains, DBA/2J and BALB/c (Charles River Laboratories, Frederick, MD). Studies were initiated when mice were 5 weeks old. Two independent studies were performed with DBA/2J mice in which the first and second studies consisted of 5 and 15 mice/group, respectively. BALB/c studies were performed subsequently with 15 mice/group. Mice were anesthetized with isoflurane prior to intranasal inoculation. Each mouse received 5 × 10⁶ 293T cells previously transfected with either 25 μg of bcmd-RGFlu DNA or 40 μg of 8-plasmid RG DNA (5 μg of each plasmid) and TransIT-LT1 at a 1:2 ratio (wt/vol [μg DNA/μl transfection reagent]). At 6 hpt, transfected cells were resuspended in transfection tissue culture medium and spun down at 1,000 rpm for 5 min. Subsequently, the transfection medium was removed and cells were resuspended in 100 μl of Opti-MEM and inoculated intranasally into mice. Negative control mice received mock-transfected 293T cells (5 × 10⁶ cells in 100 μl per mouse) or cells transfected with either empty bacmid (bcmd-DH10; 25 μg) or a bacmid carrying the 6 RG competent internal gene segments of PR8 (bcmd-P6PR8; 25 μg). Positive control mice were inoculated intranasally with wild-type (wt) PR8 virus (10⁴ TCID₅₀/100 μl/mouse). Mice were monitored daily for 14 days postinoculation (dpi) for clinical signs of disease, including lack of grooming, presence of rough coat, respiratory distress or discharge, neurological signs, body weight loss, and survival. A scoring system was used and mice were euthanized if a moribund state was reached. For the first study in DBA/2J mice, lung and brain tissues were collected at the time of death. For the second study in DBA/2J mice and for the BALB/c mouse study, 5 mice/group were sacrificed at 5 dpi to determine virus titers from the whole lung. Surviving mice were euthanized by 14 dpi. The left lungs and half of the cerebrum were collected and stored at −80°C for determination of virus titers. For virus titrations, the tissues were weighed and homogenized with a tungsten carbide bead (200 mm) in 0.01 M PBS to produce a 10% (wt/vol) homogenate, which was oscillated 50 times at 1/s for 3 min in a TissueLyser LT apparatus (Qiagen). After centrifugation at 12,000 rpm for 10 min, 100-μl aliquots of the supernatants were collected and serially diluted into MDCK plates. Virus titers were subsequently measured in TCID₅₀ assays. For the DBA/2J mouse studies, the right lung and the other half of the cerebrum were fixed in 10% formalin and subsequently embedded in paraffin for immunohistochemistry (IHC) assays.

Protective efficacy of TBI vaccination in mice.

For TBI vaccination studies, mice were inoculated as described above, except that 1 × 10⁷ 293T cells/100 μl/mouse were used. Two independent studies were performed: one in DBA/2J mice and another in BALB/c mice. Groups consisted of either 5 (DBA/2J) or 15 (BALB/c) mice/group. TBI experimental groups were inoculated intranasally with 293T cells transfected with either the bcmd-P8Ty04ts (H1N1_PR8) or 8-plasmid RG system. Negative control groups consisted of 293T cells transfected with either bcmd-P6Ty04ts or bcmd-DH10 or were mock transfected. A killed virus vaccine control group was included, and these animals received a formalin-inactivated PR8 virus vaccine (PR8 KV). The PR8 KV vaccine was prepared from a PR8 virus stock at a concentration of 1 × 10⁹ EID₅₀/0.1 ml and then inactivated with 0.2% formalin (buffered neutral) for 48 h at 4°C. After verification of virus inactivation in 9-day-old SPF eggs, the PR8 KV vaccine was mixed with the adjuvant Montanide ISA50 V2 (Seppic Inc., Fairfield, NJ) following the manufacturer's directions. The PR8 KV mouse group received a vaccine dose of 0.1 ml/mouse intramuscularly. In BALB/c mouse studies, an additional vaccine group was included that received an attenuated virus vaccine (H1N1Ty04att) that was administered intranasally, as previously described (27). At 15 days postpriming, mice were boosted with a second dose of vaccine identical to the prime vaccine for each corresponding group. At day 21 postpriming, mice were challenged intranasally with 100× the 50% mouse lethal dose (MLD₅₀) of PR8 virus (5 × 10³ TCID₅₀/30 μl). After challenge, mice were monitored for clinical signs as explained above.

Hemagglutination inhibition assays and NP ELISA.

Mice were bled using the submandibular bleeding method at days 0, 7, 14, 21, 28, 35 and 42 after primary immunization in order to test antibody titers. For hemagglutination inhibition (HI) assays, the serum samples were collected, aliquoted, and then diluted 1/10 with receptor-destroying enzyme (RDE-II; Denka Seiken Co., Tokyo, Japan) and incubated at 37°C overnight to remove nonspecific binding activity. Subsequently, the HI assay was performed as described previously (16, 28). NP antibodies in mouse sera were identified in a blocking enzyme-linked immunosorbent assay (ELISA; Synbiotics Corporation, College Park, MD). The ratios of the optical density at 630 nm of a sample well (S) to that of a negative control well (N) were calculated, and S/N values of ≤0.5 were considered positive (29).

IHC assays.

Paraffin-embedded tissues were sectioned and stained with hematoxylin and eosin (H&E). For IHC, the tissue sections were stained with horseradish peroxidase-conjugated rabbit anti-H1N1 NP polyclonal antibody (Bioss Inc., Woburn, MA). To block nonspecific antibody binding activity, the sections were blocked with 10% goat serum for 30 min (Sigma), and endogenous peroxidase activity was quenched with 1% H₂O₂. Finally, the staining was performed using avidin-streptavidin-peroxidase and diaminobenzidine as the substrate as described previously (30).

Statistical analysis.

Data were graphed and statistical analyses were performed using the Prism 6 software (GraphPad, La Jolla, CA). Comparisons between two groups' means were carried out with a two-tailed Student t test, whereas multiple comparisons were carried out by an analysis of variance (two-way ANOVA method). The differences were considered statistically significant at P values of <0.05. Survival curves across groups were compared with a log-rank test.

RESULTS

Virus generation in tissue culture cells from bcmd-RGFlu DNA.

Studies were performed to determine whether endotoxin-free bcmd-RGFlu DNAs (Fig. 1 and Table 1) were able to generate influenza virus de novo and to estimate the minimum amount of DNA required for bacmid rescue compared to that with the 8-plasmid RG system. After insertion of the RG influenza virus clone, the bacmid vector was approximately 170 kb long (Table 1). Initially, bcmd-P8PR8 (H1N1_PR8) was produced that carried genes for an RG influenza virus infectious clone of the PR8 virus. Subsequently, another construct was produced, bcmd-P8LP (∂H5N1_VN1203), which carried genes for 6 RG units from the internal gene segments of PR8 and two surface gene segment RG units from a low-pathogenicity version (∂H5N1) of A/Viet Nam/1203/2004 (H5N1). PCRs from bcmd-P8PR8 (H1N1_PR8) and bcmd-P8LP (∂H5N1_VN1203) revealed amplicons consistent with the presence of 8 influenza virus gene segments (Fig. 2A and B). In order to test whether other cell types were amenable for transfection and de novo influenza virus synthesis from the recombinant bcmd-RGFlu DNAs, an additional construct, bcmd-P8LPE (∂H5N1_VN1203/EGFP), was created that carried EGFP under the control of a CMV RNA pol2 promoter (Table 1). It must be noted that the EGFP gene segment is not part of the RG system for influenza virus; it only serves as a marker of transfection of the cells in question. An optimization of transfection conditions using 293T cells was carried out to maximize influenza virus rescue efficiency. As little as 250 ng of bcmd-P8LPE (∂H5N1_VN1203/EGFP) was sufficient to rescue influenza virus, compared to the 500 ng of total plasmid RG DNA (62.5 ng/each plasmid) that was required when using the 8-plasmid system (data not shown). If an extrapolation were based on DNA molecule copy numbers, influenza virus rescue required about 1.25 × 10⁹ copies of bcmd-RGFlu DNA, compared to 9.05 × 10¹⁰ copies of plasmid DNA. When 20 μg of bacmid DNA was used, significant precipitation of DNA/transfection reagent was observed. Thus, for subsequent studies bacmid DNA in the range of 250 ng to 5 μg was used at a 1:2 ratio for DNA (in μg):transfection reagent (in μl). With bcmd-P8PR8 (H1N1_PR8), influenza PR8 virus titers reached 2.32 × 10⁷ TCID₅₀/ml at 48 hpt; this was about 150 times higher than the amount of virus obtained using the corresponding RG reverse genetics plasmids (1.58 × 10⁵ TCID₅₀/ml; P < 0.05) (Fig. 2C). In Vero cells, which are usually harder to transfect than 293T cells, the PR8 virus titer from transfected bcmd-P8PR8 (H1N1_PR8, 5 μg/10⁶ cells) was 1.58 × 10⁷ TCID₅₀/ml by 72 hpt, which was over 300 times higher than that obtained from cells transfected with the 8-plasmid RG system (5.00 × 10⁴ TCID₅₀/ml; P < 0.01) (Fig. 2D). Using bcmd-P8LP (∂H5N1), the amount of influenza ∂H5N1 virus obtained was higher in 293T cells (5.00 × 10⁷ TCID₅₀/ml, 48 hpt) and Vero cells (1.58 × 10⁷ TCID₅₀/ml, 72 hpt) than that obtained from the corresponding 8-plasmid PR8 RG system (∂H5N1 RG system, 1.08 × 10⁶ TCID₅₀/ml in 293T cells and 1.08 × 10⁴ TCID₅₀/ml in Vero cells; P < 0.05 and P < 0.01, respectively). Taken together, these results show that transfection of a bcmd-RGFlu DNA results in de novo influenza virus synthesis with efficiencies superior to those with the 8-plasmid system.

FIG 2.

De novo influenza virus synthesis from bcmd-RGFlu constructs. (A) PCR amplicons were produced by using bcmd-P8PR8 (H1N1_PR8) as the template with the following primer sets: Ba-PB2-1F/Ba-PB2–2341R (1), Bm-PB1-1/Bm-PB1–2341R (2), Bm-PA-1/Bm-PA-2233R (3), Bm-HA-1F/Bm-NS-890R (4), Bm-NP-1F/Bm-NP-1565R (5), Ba-NA-1F/Ba-NA-1413R (6), Bm-M-1F/Bm-M-1024R (7), and Bm-NS-1F/Bm-NS-890R (8). (B) PCR amplicons produced as for panel A except that bcmd-P8PR8 (∂H5N1_VN1203) was used as the template with the following primer sets: PB2–1643F/Ba-PB-2341R (1), PB1–1240F/Bm-PB1–2341R (2), PA-892F/Bm-PA-2233R (3), H5-clvF/Bm-NS-890R (4), NP-1116F/Bm-NP-1565R (5), NA-788F/Ba-NA-1403R (6), M-741F/Bm-M-1042R (7), and NS-474F/Bm-NS-890R (8). (C) Five micrograms of bcmd-P8PR8 (H1N1_PR8) or bcmd-P8LP (∂H5N1) (or 8 μg of the corresponding plasmid RG set) was transfected into cocultured 293T/MDCK cells. Virus in supernatants of transfected cells was collected every 24 h until 96 hpt, and titers were determined on MDCK cells. (D) Results of an experiment similar to that in panel C, except that transfection was into Vero cells. (E) Five micrograms of bcmd-P8LPE (∂H5N1_VN1203/EGFP) was transfected into the indicated cells and analyzed for EGFP expression and examined under an Axiophot photomicroscope (Carl Zeiss; λEx of 488/543 nm and λEm of 522/590 nm for 100 ms). Pictures are representative of transfection efficiency at 48 hpt. Magnification, ×20. The amount of ∂H5N1 virus produced at 72 hpt is reported in Table 2. (F) Immunofluorescence assay for HA expression from bcmd-P8LPE (∂H5N1_VN1203/EGFP) transfected into 293T cells using monoclonal antibody DPJY01 against H5 HA and FITC-conjugated goat anti-mouse Ig (H+L) (Southernwest Biotech Associates Inc., Birmingham, AL). Nuclei were stained with 4′,6-diamidino-2-phenylindole. Magnification, ×20. (G) Western blot analysis of 293T cells transfected with bcmd-P8LPE (∂H5N1_VN1203/EGFP) or bcmd-P6PR8 or mock transfected (293T). Antibodies specific for viral PB2, PB1, PA, NP, and NS1 proteins as well as antibodies against EGFP and the host glyceraldehyde-3-phosphate dehydrogenase (GADPH) were used (see Materials and Methods).

When 5 μg of bcmd-P8LPE (∂H5N1_VN1203/EGFP) was used to transfect 1 × 10⁶ 293T cells, approximately 99% of those cells showed positive GFP expression by 48 hpt (Fig. 2E). In human-origin lung carcinoma A549 cells, transfection with bcmd-P8LPE (∂H5N1_VN1203EGFP) resulted in expression of EGFP by 48 hpt (Fig. 2E and G) and led to 1.08 × 10⁶ TCID₅₀/ml of ∂H5N1 virus by 72 hpt (Table 2). In addition, HBEC generated influenza virus by 72 hpt, although its detection was dependent on blind passage in 10-day-old chicken eggs (1.08 × 10⁷ TCID₅₀/ml after blind passage), consistent with their low transfection efficiency (Table 2 and Fig. 2E). Likewise, transfected swine-origin PK-15 cells generated detectable ∂H5N1 virus by 72 hpt after blind passage in MDCK cells (1.08 × 10⁵ TCID₅₀/ml) (Table 2 and Fig. 2E). Differences in virus rescue efficiencies can be partly attributed to differences in transfection efficiencies based on EGFP expression, as shown in Fig. 2E. As expected, no virus was obtained with either bcmd-RGFlu or RG plasmids transfected into mouse cells (Table 2 and data not shown), consistent with the lack of activity of the human pol1 promoter in these cells. In contrast, EGFP expression was readily detected in mouse cells (Fig. 2E and Table 2). Additional evidence of proper expression of influenza virus proteins from bcmd-RGFlu in 293T cells was observed by IFA and Western blot analysis (Fig. 2F and G).

A transfection-based inoculation method leads to production of lethal PR8 virus in mice via bcmd-P8PR8 (H1N1_PR8).

Recently, a TBI method was developed in which 293T cells previously transfected with a full set of RG competent plasmids were inoculated into ferrets and resulted in effective influenza virus infection in vivo (31). Because the hpol1 promoter is not active in mouse cells, TBI was used to determine whether 293T cells which were previously transfected with bcmd-P8PR8 (H1N1_PR8) and subsequently inoculated intranasally in mice would generate influenza virus in vivo. At 6 hpt, 293T cells were lifted from the plate, resuspended in tissue culture medium, and inoculated intranasally into mice. Each mouse received 100 μl of cell suspension containing 5 × 10⁶ cells [5 μg of bcmd-P8PR8 (H1N1_PR8)/10⁶ cells]. Alternatively, another group of mice received cells previously transfected with the 8-plasmid PR8 RG system (8 μg of total DNA/10⁶ cells). DBA/2J mice (n = 5/group) that received 293T cells transfected with either the bcmd-P8PR8 or the 8-plasmid PR8 RG started to show signs of disease similar to that in mice challenged with 10⁴ TCID₅₀ of the wild-type PR8 virus (Fig. 3A). By 7 dpi, all mice in the bcmd-P8PR8 (H1N1_PR8) group as well as 4 out 5 mice in the 8-plasmid PR8 RG had succumbed to the infection. Compared to mice challenged with wt PR8 virus, there were 24- and 48-h delays in the times of death in the bcmd-P8PR8 (H1N1_PR8) and 8-plasmid PR8 RG groups, respectively. This observation was consistent with the kinetics of weight loss over time (Fig. 3B), and with the levels of virus detected in the lungs of mice at the time of death (Fig. 3C). One mouse in the 8-plasmid group appeared to not have been infected during the TBI procedure and showed no evidence of virus replication by 14 dpi. Interestingly, mice that received either the bcmd-P8PR8 (H1N1_PR8) or 8-plasmid PR8 RG showed no evidence of virus infection in the brain, unlike the mice that were challenged with the PR8 virus (Fig. 3C and D). This difference is an important feature from a vaccine perspective, since an in vivo RG vaccination strategy would require that the replication of the virus be limited to the respiratory tract. However, lack of detection of viral antigen in the brain does not preclude virus presence below the limit of detection. In order to determine whether the TBI in mice was reproducible and applicable to more than one mouse strain, an additional study was performed in DBA/2J mice (n = 10 plus 5/group) and BALB/c (n = 10 plus 5/group). In this new study, 10 mice/group were monitored for signs of disease and survival over 14 dpi after TBI or inoculation with wild-type PR8 virus. An additional 5 mice/group were sacrificed at 5 dpi to determine virus titers in mouse lungs. The patterns of weight loss and survival of DBA/2J mice in the second study were similar to those from the first study (Fig. 3E and F). By 14 dpi, 10 out of 10 DBA mice in the bcmd-P8PR8 (H1N1_PR8) group succumbed to the infection, whereas 9 out of 10 mice in the 8-plasmid PR8 RG group succumbed to infection (Fig. 3F). Mice in the bcmd-P8PR8 (H1N1_PR8) group showed a tendency to succumb to the infection earlier than mice in the 8-plasmid PR8 RG group, although this was not statistically significant based on a log-rank test (P = 0.0825). Significant differences were detected, however, in virus titers in mouse lungs at 5 dpi (Fig. 3G), and these were significantly higher in the bcmd-P8PR8 (H1N1_PR8) group than in the 8-plasmid PR8 RG group (n = 5/group; P < 0.05). As expected, mice that received either a bacmid carrying just the internal gene segments (bcm-P6PR8), empty bacmid DNA (bcmd-DH10), or 293T cells showed neither signs of disease nor death, consistent with absence of virus in the lung homogenates (Fig. 3G). Similar observations were made for BALB/c mice, as mice in the bcmd-P8PR8 (H1N1_PR8) group showed significant body weight loss and 8 out of 10 succumbed to the infection by 14 dpi (Fig. 4A and B). In contrast, significantly fewer animals (log-rank test, P = 0.0332) died in the 8-plasmid PR8 RG group, with only 5 out 10 succumbing to the infection (Fig. 4A and B). Virus titers in lung homogenates prepared by 5 dpi (n = 5/group) (Fig. 4C) showed a profile similar to the one obtained in DBA/2J mice. The mean lung virus titers in the bcmd-P8PR8 (H1N1_PR8) group were significantly higher than those in the 8-plasmid group (P < 0.05).

FIG 3.

Transfection-based inoculation of all-in-one bacmids leads to lethal PR8 virus replication in DBA mice. Five-week-old DBA/2J mice (n = 5 [A to D] or n = 15 [E to G]) were intranasally inoculated with 100 μl of inoculum containing 5 × 10⁶ 293T cells transfected 6 h earlier with bcmd-P8PR8 (H1N1_PR8) or the corresponding 8-plasmid RG DNAs. Negative control mice received 5 × 10⁶ mock-transfected 293T cells or cells transfected with either bcmd-P6PR8 without HA and NA cassettes or bcmd-DH10 empty DNA. Positive control mice were intranasally inoculated with wt PR8 virus (10⁴ TCID₅₀/100 μl). Mice were monitored for clinical signs, including body weight changes (A and E) and survival (B and F). At the time of death (C) or at 5 dpi (G), mice were necropsied or euthanized and lungs and brain were collected and analyzed for levels of virus in tissues by a TCID₅₀ assay in MDCK cells. *, P < 0.05. A section of the collected tissues from panel C were subjected to IHC for detection of viral antigen (NP) (see Materials and Methods), which is indicated by the presence of a brown precipitate in panel D (representative staining is marked with arrows).

FIG 4.

Transfection-based inoculation of all-in-one bacmids leads to lethal PR8 virus replication in BALB/c mice. Five-week-old BALB/c mice (n = 15) were intranasally inoculated with 100 μl of inoculum containing 5 × 10⁶ 293T cells, which had been previously transfected 6 h earlier with either 25 μg of bcmd-P8PR8 (H1N1_PR8) or 40 μg of the corresponding 8-plasmid RG set. Negative control mice received 5 × 10⁶ mock-transfected 293T cells, or transfected bcmd-P6P without HA and NA cassettes, or bcmd-DH10. Positive control mice were intranasally inoculated with wt PR8 virus (10⁴ TCID₅₀/100 μl). Mice were monitored for clinical signs, including body weight changes (A) and survival (B). (C) Five mice were necropsied at 5 dpi, and lungs were collected and analyzed for virus levels by titration on MDCK cells. *, P < 0.05.

Transfection-based inoculation of a bcmd-RGFlu carrying genes for a temperature-sensitive H1N1 virus protects mice against lethal challenge.

To determine whether the TBI/bcmd-RGFlu system protects against the virulent PR8 virus, bcmd-P8Ty04ts (H1N1_PR8) was created and carried the internal gene segment backbone from temperature-sensitive Ty04 and the surface gene segments from the PR8 strain (Table 1). Transfection of bcmd-P8Ty04ts (H1N1_PR8) or the corresponding 8-plasmid RG system into 293T cells was performed as described above at 33°C (data not shown), 35°C (Fig. 5A and B), and 37°C (data not shown). The Ty04ts (H1N1_PR8) virus was readily detected in 293T and Vero cells although, as expected, with lower efficiency than bcmd-P8PR8 (H1N1_PR8). At 120 hpt, 293T/MDCK cocultured cells (Fig. 5A) or Vero cells (Fig. 5B) transfected with bcmd-P8Ty04ts (H1N1_PR8) produced about 100-fold more Ty04ts (H1N1_PR8) virus than cells transfected with the 8-plasmid RG system (P < 0.001).

FIG 5.

bcmd-P8Ty04ts (H1N1_PR8) protects mice against lethal PR8 virus challenge. (A and B) Five micrograms of bcmd-P8Ty04ts (H1N1_PR8) or 8 μg of the corresponding 8-plasmid RG set was transfected into cocultured 293T/MDCK cells (A) or Vero cells (B). Virus in supernatants of transfected cells was collected every 24 h until 144 hpt, and titers were determined based on the TCID₅₀ for MDCK cells. (C to E) Five-week-old BALB/c mice (n = 15) were vaccinated intranasally twice, 2 weeks apart, with 100 μl of inoculum containing 1 × 10⁷ 293T cells that had been transfected 6 h earlier with either 25 μg of bcmd-P8Ty04ts (H1N1_PR8) or 40 μg of the corresponding 8-plasmid RG set. Negative control mice received 1 × 10⁷ mock-transfected 293T cells or cells transfected with bcmd-P6Ty04ts or bcmd-DH10. Positive control mice were vaccinated intramuscularly, 2 weeks apart, with killed vaccine (KV) PR8 virus or were vaccinated intranasally with the H1N1Ty04att virus vaccine. (C) Mice were monitored for seroconversion by HI assay prior to challenge. (D and E) Survival and body weight changes over a 14-day period following challenge with 100 MLD₅₀ of PR8 virus (5 × 10³ TCID₅₀/30 μl) administered intranasally 7 days after boost vaccination. ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

For vaccination/challenge studies, DBA/2J (n = 5/group) and BALB/c (n = 15/group) mice were subjected to TBI intranasally using a prime-boost strategy 16 days apart. Similar results were obtained in DBA/2J and BALB/c mice. Only results with the BALB/c mice are shown, because the number of animals used allowed for more statistically significant observations (Fig. 5). Mouse groups were set up to receive by TBI the bcmd-P8Ty04ts (H1N1_PR8) or the corresponding 8-plasmid RG system (6:2 plasmids). Negative control groups included mice that received 293T cells transfected with either bcmd-P6Ty04ts (bacmid carrying the six internal gene segments of Ty04ts [Table 1]) or empty bcmd-DH10 or were mock transfected. Positive control mice included two groups that received the same prime-boost regimen described above. The first positive control group included mice vaccinated intranasally with a live attenuated influenza virus vaccine based on the Ty/04att backbone (H1N1Ty04att) (16), and the second group of mice were vaccinated intramuscularly with killed PR8 virus vaccine (PR8 KV). Prechallenge HI titers indicated seroconversion in mice vaccinated with PR8 KV, with a mean HI titer of 497.8 and with every mouse in the group having an HI titer of ≥320. Compared to the PR8 KV group, significantly lower HI titers were observed in the H1N1Ty04att live virus vaccine group. Mean HI titers in the H1N1Ty04att-vaccinated group were 38.7, with 9 out of 15 mice having HI titers of ≥40. In contrast, mice in the bcmd-P8Ty04ts (H1N1_PR8) group or 8-plasmid RG group had negligible HI titers, with just 2 and 1 mouse, respectively, having HI titers of 20 (Fig. 5C). Analysis of antibodies against the viral NP protein by ELISA showed seroconversion in the PR8 KV and H1N1Ty04att groups but not in any of the TBI groups (data not shown).

At 7 days postboost, mice were challenged with 100 MLD₅₀ of the PR8 virus. Interestingly and despite negligible seroconversion, TBI vaccination with bcmd-P8Ty04ts (H1N1_PR8) resulted in survival of 13 out of 15 mice at 14 days postchallenge, which corresponded to 87% survival (Fig. 5D). In the corresponding 8-plasmid RG group, 10 mice survived out of 15 (67% survival). No statistically significant differences were observed between these two groups (P = 0.179), although a trend was observed that was consistent with the former having more survivors than the latter. In contrast, only 1 mouse survived in the bcmd-P6Ty04ts group (7% survival), which was statistically significantly lower than the survival seen in the bcmd-P8Ty04ts group (P < 0.001) or 8-plasmid group (P = 0.0063) based on a log-rank test. As expected, no survivors were observed in the mock-transfected 293T cell group, whereas 100% of mice survived in the KV vaccine and H1N1Ty04att vaccine groups. Body weight loss was observed in all TBI groups and was indicative of active PR8 virus replication after challenge, although mice in the bcmd-P8Ty04ts (H1N1_PR8) group showed the least variation in body weight and no discernible clinical signs (Fig. 5E). Taken together, these results suggest that TBI vaccination with a bcmd-RGFlu can result in significant protection of mice against aggressive influenza virus challenge (Fig. 5D and E). A trend to lower protection in the 8-plasmid RG group compared to the bcmd-P8Ty04ts (H1N1_PR8) was consistent with differences in virus rescue efficiency (Fig. 5A and B).

DISCUSSION

According to the Centers for Disease Control and Prevention, influenza vaccination is the most effective method to reduce the impact of influenza and its severe complications. The current KV and LAIV vaccines are trivalent or quadrivalent products aimed at providing protection against H1 and H3 influenza A virus subtypes and influenza type B viruses. Influenza virus reverse genetics has changed the way seed stocks of influenza vaccines can be prepared (32). Virus-like RNAs are produced from DNA copies, usually under the control of the RNA polymerase I promoter (pol1), and viral proteins are expressed by using an RNA polymerase II promoter (pol2). A robust implementation of the system consists of 8 plasmids, with each viral cDNA cloned under the control of promoters running in opposite directions (13, 33, 34). Although RG provides an easier method for the generation of vaccine seed stocks, the vaccine virus itself is ultimately propagated by using the traditional egg method or tissue culture cells (35, 36). The manufacture of influenza vaccines requires approximately 8 months. During a pandemic, such a lag period greatly affects the ability to curtail or minimize the effects of such an event. The 2009 pandemic H1N1 virus (pH1N1) highlighted such a fact, as traditional vaccine production systems were too slow to significantly ameliorate the pandemic's impact. Thus, there continues to be a major emphasis on finding alternatives to production of influenza vaccines in systems other than eggs or tissue culture cells and to provide a universal influenza vaccine that does not require annual reformulation. These efforts have been aimed at retargeting the host immune response and stimulating the production of broadly reactive antibodies against the HA and/or other viral proteins. Plasmid DNA (37–40), recombinant adenoviruses (41–44), alphavirus replicon particles (45–48), VLPs (49–55), and other methods have been used for influenza virus gene/protein delivery, alone or in combination. Among these methods, DNA vaccination offers many potential advantages—most notably, ease and low-cost production. However, despite its simplicity and safety, DNA vaccination has been hampered by low immunogenicity in humans. Interestingly, successful DNA vaccines for other infections are commercially available for veterinary use, including canine melanoma (56), West Nile viruses in horses (57), and fish hematopoietic necrosis virus (58). We hypothesize that the limitation in immunogenicity of DNA vaccines can be overcome by providing alternative machinery capable of amplifying the antigen of interest. In this report, influenza viruses were generated de novo by reverse genetics using recombinant bacmids as delivery vehicles and a modified transfection-based inoculation system (31). Consolidating influenza virus RG competent units into fewer plasmids or other vectors to improve virus rescue efficiency has been performed before, although not with an attempt to establish activity in vivo. Neumann et al. were able to produce, from a single RG plasmid, about 10⁵ to 10⁶ TCID₅₀/ml of the laboratory-adapted WSN strain from 5 × 10⁵ Vero cells by 72 hpt (59), although this needed the cotransfection of helper protein expression plasmids carrying genes for the viral ribonucleoprotein polymerase complex. More recently, a single plasmid system based on chicken pol1 was constructed on a low-copy-number vector, p15A (60). Very limited virus production of the WSN strain was observed in chicken embryo fibroblasts with this p15A-based plasmid (≤1,280 PFU/ml by 6 days posttransfection). Perhaps one of the limitations of the latter system was that chicken cells were sensitive to TPCK-trypsin and therefore 5 to 10% allantoic fluid had to be used in order to favor virus rescue.

In this report, bacmids were used in order to overcome some of the limitations of cloning a full influenza virus RG set into a single DNA unit constrained by stability or copy number. Two sets of high-copy-number plasmids were initially constructed, one to contain an RG set of the internal gene segments of the influenza virus strain in question (PR8 or Ty04ts) and another to contain the RG units for the HA and NA surface genes (Table 1). Both sets of these plasmids had additional molecular features to generate recombinant bacmids in which the HA and NA RG cassettes could be easily incorporated by recombination while maintaining the chosen influenza virus strain backbone. bcmd-RGFlu constructs had improved transfection efficiency (≥100-fold) compared to the 8-plasmid counterparts, particularly in hard-to-transfect Vero cells (Fig. 2). This observation represents an improvement over previous reports regarding transfection of Vero cells. Fodor reported the generation of 10 to 20 PFU of influenza virus from 10⁷ Vero cells by 4 dpi (61). Wood and Robertson generated an H5N1 reference vaccine strain in Vero cells by reverse genetics but they did not report the rescue efficiency (62), whereas PR8 or PR8-based viruses were generated in Vero cells with low efficiency (<10³ PFU/ml). Our studies showed that virus rescue is also possible in human primary bronchial epithelial cells, although the amount of virus produced was initially below the limit of detection and needed amplification through blind passage in eggs. A similar observation was noted when swine PK15 cells were used; in these cells the hpol1 gene product is expected to work less efficiently. These observations, however, are good indications for the potential of bcmd-RGFlu as a platform for vaccination against influenza virus in humans or pigs. Further studies, particularly rescue of influenza viruses under the context of a swine pol1 promoter, will likely provide further insights into this process.

Since the hpol1 promoter shows no activity in mice or in mouse cells, we resorted to a transfection-based inoculation system to determine whether the bcmd-RGFlu DNA would work in vivo. 293T cells transfected with either bacmid or plasmid DNA were inoculated into mice by 6 hpt, a time in which very few virus particles, if any, are predicted to be produced. In fact, 6-hpt supernatants were tested for the presence of virus, which was found to be below the limit of detection after blind passages into eggs. Thus, it is tempting to speculate that transfected cells remained alive in the respiratory tract of mice long enough to produce influenza virus particles capable of establishing an influenza virus infection (Fig. 3 and 4). This observation is also consistent with the protection seen against lethal influenza virus challenge in mice that received 293T cells transfected with a bacmid carrying genes for a temperature-sensitive influenza virus vaccine (Fig. 5). Although the TBI vaccination strategy did not prevent body weight loss after challenge, there was remarkable protection, with 13 out of 15 mice surviving in the group that received bcmd-P8Ty04ts (H1N1_PR8). Survival in the bcmd-P8Ty04ts (H1N1_PR8) or the 8-plasmid counterpart groups could not be correlated to HI titers, since in both groups there was either very limited or negligible seroconversion. Lack of detectable seroconversion responses was not unexpected. Previous studies have shown that vaccination with low amounts of live attenuated influenza virus vaccines often results in serological responses below the limit of detection (16). It is plausible that virus replication of an attenuated strain generated from cells transfected with either a bacmid or a plasmid RG set would be even lower than by direct inoculation of live attenuated virus. Analysis of mucosal immunity should provide a better understanding of the mechanism of protection imparted by the TBI method; such an analysis was beyond the scope of the present study.

Taken together, the results presented in this report are encouraging for exploration of alternative delivery strategies in order to induce better protection against influenza virus by using bcmd-RGFlu DNAs. It must be noted that use of the mouse polymerase I promoter, instead of hpol1, would be more suitable to further explore the potential for in vivo influenza virus reverse genetics. Unfortunately, a search of the literature revealed limited characterizations of mouse pol1 as a suitable promoter for rescue of influenza viruses (63, 64). To our knowledge, an entire reverse genetics system for influenza virus using the mouse pol1 promoter is currently lacking. It is also worth noting that the bcmd-RGFlu DNAs described here can be easily converted into recombinant baculoviruses with the potential for an alternative method of DNA delivery into cells. Furthermore, the bcmd-RGFlu DNAs have significantly higher cloning capacities than plasmids, which would allow inclusion of additional virus- or host-based strategies aimed at improving immune responses to influenza virus. In summary, we have described a new strategy with the potential to produce live influenza viruses de novo through in vivo reverse genetics. This strategy is, in principle, not limited to influenza virus but amenable for other pathogens and may overcome the current limitations of many vaccines.