Construction and Characterization of Insect Cell-Derived Influenza VLP: Cell Binding, Fusion, and EGFP Incorporation

We have constructed virus-like particles (VLPs) harboring hemagglutinin (HA), neuraminidase (NA), matrix protein 1 (M1) ,and proton channel protein (M2) using baculovirus as a vector in the SF9 insect cell. The size of the expressed VLP was estimated to be ~100 nm by light scattering experiment and transmission electron microscopy. Recognition of HA on the VLP surface by the HA2-specific monoclonal antibody IIF4 at acidic pH, as probed by surface plasmon resonance, indicated the pH-induced structural rearrangement of HA. Uptake of the particle by A549 mediated by HA-sialylose receptor interaction was visualized by the fluorescent-labeled VLP. The HA-promoted cell-virus fusion activity was illustrated by fluorescence imaging on the Jurkat cells incubated with rhodamine-loaded VLP performed at fusogenic pH. Furthermore, the green fluorescence protein (GFP) was fused to NA to produce VLP with a pH-sensitive probe, expanding the use of VLP as an antigen carrier and a tool for viral tracking.


Introduction
Virus-like particle (VLP) has been extensively studied for its vaccine capacity in the last 20 years. More than a dozen of VLP-based vaccines have been tested in clinical trials. Two of them against hepatitis B and human papilloma viruses are commercially available as prophylactics [1]. The merits of VLP-based vaccine rest on its viral mimicry and devoid of genetic materials, hence alleviating the safety concerns [2]. The repetitive and ordered structures of VLPs are conducive to opsonization, resulting in enhanced phagocytosis [3,4], and are capable to activate B cells in a T cell-independent manner [5,6]. Owing to the virus-like dimension and particulate nature, VLPs are easily taken up by antigenpresenting cells (APCs) [7,8] and more efficient in crosspresentation of antigens to MHC I molecules compared to the soluble antigen and hence the possibility of activating antigen-specific cytotoxic T lymphocytes (CTLs) [9].
Although the VLP-based vaccines have the potential to induce CTL response, most of them favor the production of humoral immunity but poorly evoke CTL response if additional stimuli for innate immune system and, in particular, APC were neglected [1]. It is somewhat surprising that VLPs derived from baculovirus system incorporating viral structural proteins of influenza virus were found to induce both cell-mediated and humoral immunity to confer full protection from viral challenge [10][11][12]. This distinct feature drew our attention on the interaction between VLPs and cells since most of the VLP-based strategies for vaccine development do not involve antigens with functions of ligand binding and membrane fusion in contrast with the hemagglutinin of influenza VLP.
Influenza A virus is an enveloped RNA virus with three transmembrane proteins: hemagglutinin (HA), neuraminidase (NA), and ion channel protein (M2) along with matrix protein 1 (M1) underlying the viral envelope. The major tasks of these proteins include ligand binding and membrane fusion for HA [13,14], spreading of viral progeny and facilitating entry of the virus for NA [15][16][17], proton transfer for viral uncoating as well as assembly and budding  [18][19][20], recruitment of viral components to the assembly site, and a major driving force for viral budding for M1 [21,22]. Several laboratories have produced insect cell-derived VLPs encompassing the above-mentioned influenza proteins for vaccine development [23]. However, there has been no report on the characterization of the interaction between VLPs and cells. To afford a better understanding of these interactions, we used fluorescent images and dye dequenching in this study in the hope of shedding some light on the mechanism of antigen presentation, and virus-mediated membrane fusion and its inhibition.  Table 1). The HA PCR products were cloned into BamHI/NotI site under the control of the polyhedron promoter, the M1 PCR products into XhoI/KpnI site under p10 promoter of the pFastbac Dual baculovirus transfer vector (Invitrogen), the M2 PCR products into EcoRI /HindIII site under the polyhedron promoter, and the NA PCR products into XhoI/KpnI site under p10 promoter of the pFastbac Dual baculovirus transfer vector (Invitrogen), and the NA-EGFP PCR products into XhoI/KpnI site under p10 promoter of the pFastbac Dual baculovirus transfer vector (Invitrogen). All the inserted sequences were confirmed by DNA sequence analysis (Mission Biotech Inc., Taipei, Taiwan).

Generation of Recombinant Baculoviruses.
Production of recombinant baculoviruses followed the manufacturer's manual of Bac-to-Bac baculovirus system (Invitrogen). Briefly, the pFastbac Dual plasmids were transformed into E.coli strain DH10Bac (Invitrogen) and selected on LB plate containing kanamycin (Invitrogen), gentamicin (Invitrogen), tetracycline (Invitrogen), Bluo-gal (Invitrogen), and IPTG (BioRad). The recombinant bacmids were confirmed by sequencing and then transfected into Spodoptera frugiperda (Sf9) cells for recombinant baculovirus packaging via the aid of Cellfectin Reagent (Invitrogen). After 4 days, the recombinant baculovirus in the supernatant was collected as P1 viral stock and further amplified as P2 viral stock for VLP purification. The virus titer of viral stock was determined by the end-point dilution in Sf9 cells [24].

Production and
Purification of VLP. The Sf9 insect cells (cat. no. 11496-015, Invitrogen) were maintained as suspension cultures in sf900 II SFM (Invitrogen) at 27 • C. For the production of HA + VLP, 300 ml of Sf9 cells at 2 × 10 6 cells/ml were coinfected with Dual HA, M1 and Dual NA, M2 recombinant baculovirus at MOI of 3 and 1, respectively, whereas for the production of HA − VLP, the Dual HA, M1 recombinant baculovirus were replaced with M1 only recombinant baculovirus. To produce NA-EGFP VLP, the same infection procedure was carried out except for the replacement of Dual NA, M2 with Dual NA-EGFP, M2 recombinant baculovirus. At 72 h post-infection, the supernatant was collected by centrifugation at 12000 g for 20 min. The VLPs were precipitated at 33000 rpm and 4 • C for 2 h (RPS40ST rotor, Hitachi). The particles were resuspended in PBS buffer, loaded on a 0-60% (w/v) discontinuous sucrose gradient, and centrifuged for 4 h at 33000 rpm and 4 • C (RPS40ST rotor, Hitachi). After ultracentrifugation, fractions were collected from the top of the gradient and analyzed by Western blotting [25]. Fractions containing HA, NA, M1, and M2 proteins were collected, and the protein concentration was determined by BCA protein assay kit (cat. 23225, Pierce).

Western Blotting for Detection of the Expressed Viral
Proteins. Protein samples were resolved by 12.5% SDS polyacrylamide gel and then transferred to PVDF membranes. The membrane blots were blocked with 5% nonfat milk in Tris-buffered saline containing 0.1% Tween-20 and probed with primary antibodies against HA (abcam ab21297), NA (abcam ab70759), M1 (abcam ab25918), M2 (novus NB100-2073), separately. The presence of each protein was detected by HRP-conjugated secondary antibody and Enhanced Chemiluminescence (ECL) plus western blot detection system (Amersham Bioscience). The blot images were captured by the LAS-3000 imaging system (Fujifilm, Tokyo, Japan).

Fluorescent Dye Labeling and Fluorescence Microscopy.
The fluorescent dye labeling was performed as described previously with minor modification [26]. VLPs suspended in PBS were mixed with equal volume of R18-or DiOcontaining PBS to final concentration of 10 μM of R18 or DiO (Molecular probes, Invitrogen). The mixture was incubated in dark for 1 h at room temperature for R18 labeling while incubated for 20 min at 37 • C for DiO incorporation. The R18-labeled VLPs were washed three times with PBS, and the unincorporated dyes were filtered away by a Microcon devise with MW cutoff of 50 kDa. The fluorecentlabeled VLPs were mounted on a glass support and visualized by inverted fluorescence microscope (Aviovert, Carl Zeiss Inc.) with oil immersion objective (Plan-Apo 100X, N.A. 1.4). The images were processed by MetaMorph software (Carl Zeiss Meditec, Gottingen Germany).

Dynamic Light Scattering (DLS) and
Transmission Electron Microscope (TEM). The particle size measurement was executed on the miniDAWN (Wyatt Technology Corp., Santa Barbara, CA) equipped with a microCUVETTE and a 30 mW, 685 nm GaAs laser light source. One milliliter of PBS was filtrated through a 0.02 μm inorganic membrane filter (Anodisc, Whatman International Ltd., Maidstone, England) as the DLS background. For the hydrodynamic radius (Rh) measurement, one microliter of VLP solution was directly mixed with the PBS and transferred in the microCUVETTE. Data collection and analysis were performed on the ASTRA V software (Wyatt Technology Corp., Santa Barbara, CA).
The morphology of the influenza VLP was observed by transmission electron microscope JEM 2010 (Jeol Ltd., Tokyo Japan). 3 μl of diluted influenza VLPs were loaded onto formvar-carbon-coated 300-mesh copper grids (Polysciences Inc.) and allowed to absorb for 3 minutes. The grids were wicked dry with filter paper, rinsed once with PBS, and negatively stained with 1% ammonium phosphotungstate pH 7. Grids were examined by TEM at 200 kV accelerating voltage for the presence of influenza virus.

Spectroscopic
Measurements. 20 μl of NA-EGFP VLP (total protein amount 15.6 μg) was diluted in 50 μl of PBS with varying pH by titration with 25 mM of citric acid. The fluorescent signal was recorded on a fluorescence spectrophotometer F2500 (Hitachi). The excitation wavelength was 470 nm while emission spectra spanned 490 to 560 nm. Excitation and emission slit widths were set at 2.5 and 10 nm, respectively.

Endocytosis of VLPs by A549 Cells.
The fluorescentlabeled VLPs were diluted in ice-cold DMEM medium containing 50 mM NH 4 Cl and added to cover slips with a monolayer of grown A549 cells. The maximum amount of NH 4 Cl was tested for 24 h for minimal cellular morphological change. The cover slips were stood on ice for 1 h to prevent VLP engulfed by pinocytosis. To show that the endocytotic process is mediated by the binding between viral HA and cellular sialylated receptors, A549 cells were treated with neuraminidase from Vibrio cholerae (Sigma) at 4 × 10 −2 unit/ml and 37 • C for 2 h prior to the addition of fluorescent-labeled VLPs. After 1 h incubation, the unbound VLPs were washed away with ice-cold DMEM medium, and the cells were cultured in DMEM medium containing 50 mM NH 4 Cl at 37 • C for 2 h [27], which was replaced with a medium containing 5 μM of DiD dye for 15 min before the cells were washed extensively with PBS and fixed with 4% paraformaldehyde in PBS at room temperature for 10 min and mounted with 50% glycerol in PBS. The R18-and DiO-labeled VLPs taken up by A549 cells were observed by Zeiss LSM510 META laser scanning confocal microscope using oil immersion objective (Plan-Apo 100x, N. A. 1.4) in multitrack channel mode. Excitation wavelengths and emission filters were 488 nm/band-pass 500-530 nm for EGFP and DiO, 561 nm/band-pass 575-630 nm for R18, and 561 nm/band -pass 651-704 nm for DiD. The images were analyzed by LSM image browser (Zeiss).

Monoclonal Antibody
Binding to HA. Experiments of surface plasmon resonance (SPR) were conducted to evaluate the binding affinity of HA2-specific mAb IIF4 to HA + and HA − VLPs. In brief, the VLP was immobilized on an activated CM5 sensor chip, which was placed in the chamber of a BIACORE 3000 biosensor system [28] (BIAcore AB, Uppsala, Sweden). IIF4 used as the analyte was injected over the CM5 sensor chip at 10 μl/min flow rate at pH 7.4 or 5.0 in sodium phosphate buffer to examine the pH-dependent conformational change as observed for the native influenza virus. The sensor surface was regenerated with solution of 0.1 M NaCl and 0.02 N NaOH. The SPR kinetic data were fitted by the BIAevaluation 3.1 software.
2.11. Immunofluorescence Detection of VLP. 20 μl of VLP was diluted in 200 μl of PBS, and then added onto poly-Llysine-treated coverslip on ice for 2 h. After PBS washing, the pH 5 treated VLP was incubated with low pH buffer (10 mM HEPES, 10 mM MES in PBS, pH 5.0) for 2 min, rinsed with PBS, and fixed with 4% paraformaldehyde and subsequently blocked with 10% FBS in PBS at RT for 1 h. The VLP was probed with IIF4 mAB (kindly provided by Dr. Vareckova, 100x diluted) and subsequently hybridized with FITC conjugated antimouse IgG (500x diluted) at 37 • C for 1 h. The coverslips were mounted on a microscope slide and visualized with fluorescence microscope (Carl Zeiss, Germany). The images were captured and processed with Metamorph software (Molecular Devices, CA, USA ).   Figure 1 illustrates the construction of pFastbac Dual plasmids containing influenza M1/HA, NA/M2, −/HA, and NA-EGFP/M2 gene pairs for the production of recombinant baculoviruses. Using these recombinant baculoviruses, HA + VLPs, HA − VLPs, and NA-EGFP VLP were generated. Figure 2 displays the Western blot results (panel a) of the expressed proteins incorporated in HA + and HA − VLPs. The incorporated proteins were expressed at similar level except for HA protein devoid in HA − VLPs as expected. It is noted that the fraction of cleavage for hemagglutinin in the present expression system is estimated to be 33.7%. The morphology of the expressed VLP as visualized by transmission electron microscopy shown in Figure 2(b) is nearly spherical with a diameter of ∼100 nm, close to the virus in the native context [31,32]. The radius of VLP measured by light scattering was 63.3 ± 0.13 nm with a convergent size distribution (Figure 2(c)).

Endocytosis of VLP in A549 Cells.
Binding of hemagglutinins to cellular sialylated receptors initiates the host endocytotic process for influenza virus entry [33][34][35]. Equal amount of HA + VLP or HA − VLP, with regard to each protein concentration, was added to A549 cells. Compared to HA − VLP, HA-incorporated VLPs were significantly endocytosed by A549 cells. To confirm that the endocytic process is mediated by HA binding to the sialylated receptor, A549 cells were treated with neuraminidase from Vibrio cholerae. After 2 h of treatment, most of the sialic acids exposed on the cell surface were removed as confirmed by sialic acid-specific lectin staining (see Figure S1 in Supplementary Material available online at doi:10.11/2010/506363). As shown in Figure 3, removal of sialic acids starkly decreased HA + VLP engulfed by A549 cells. Also, attachment of VLP to another suspended cell line Jurkat cell could be hampered by fetuin (Figure 3(f)), which is a sialylated glycoprotein [36,37] Journal of Biomedicine and Biotechnology and used as a model inhibitor for influenza virus infection [38][39][40][41], in a dose-dependent manner. Taken together, the HA + VLP was endocytosed by A549 cells through the binding of viral HA and cellular sialylated receptor.

HA2 Exposure Recognized by Monoclonal Antibody IIF4
in Acidic Environment. IIF4 mAb has been reported to be specific to the epitope within HA2 aa 125-175 of the H3 subtype of influenza A virus and is well cross-reacting with H4 and H5 viruses [42][43][44]. Previous studies demonstrated that IIF4 epitopes on HA2 became accessible after acid treatment on HA [42,45]. The binding affinity of HA2specific mAb IIF4 to HA + VLPs at acidic pH was 0.18 ± 0.05 nM, 32-fold higher than binding affinity at neutral pH (Figures 4(a) and 4(c) and Table 2). This implied that the HA on VLP membrane could expose HA2 subunit leading to fusion of the viral and cellular membranes as a result of lowered pH.
To further support the SPR result that the HA + VLP has much higher affinity to IIF4 mAb at pH 5 than at pH 7.4, immunofluorescence staining was performed. As shown in Figure 4(d) acidic buffer-treated HA + VLP was recognizable by HA2 IIF4 mAb whereas at natural pH HA + VLP and HA − VLP were not, in agreement with the SPR measurement and indicating that the HA2 portion on the VLP is exposed at acidic pH. Graph in (f) depicts the inhibitory effect of fetuin on VLP binding. The values of relative VLP binding was determined by the ratio of R18 signals at each concentration of fetuin to the value obtained for the maximum binding of VLPs without the fetuin treatment.  [29,30]. Thus R18-labeled VLPs were mixed with Jurkat cells and collected by centrifugation of the cells. As shown in Figure 5(a), HA + VLP could adhere to the cells surface and form heterogeneous fluorescent dots while HA − VLP could be barely detected. After lowering the external pH of the cells, the cell-bound VLPs underwent fusion with the cells, as manifested by more homogenous dispersion of fluorescent signals around the cell. The fusion activity of HA + VLP was reaffirmed by the fluorescence dequenching experiment displayed in Figure 5(b) showing that only HA + VLP induced dramatic fluorescence increase in response to acidic pH.

Anchorage of NA-EGFP on VLP as a Model for Antigen
Incorporation. Since the NA activity has been inferred unnecessary for influenza VLP release from Sf9 insect cells [46][47][48][49], the extracellular domain of NA was replaced with the EGFP protein to demonstrate the possibility of accommodating other protein antigens on the VLP membrane to extend the application of influenza VLP as an antigen carrier. To construct the membrane-anchored chimera, the amino terminal 61 amino acids of NA protein were retained and fused with EGFP [50]. The NA-EGFP VLPs were stained with lipophilic dye R18 to characterize the virus-like To investigate the uptake of NA-EGFP-incorporated VLP by cells, the VLPs were labeled with R18 and then admixed with A549 cells at 4 • C to prevent pinocytosis. As shown in Figures 6(d)-6(f), NA-EGFP VLP can be taken up by A549 cells, and 20.6 ± 3.7% of the endocytotic vesicles contain the EGFP fluorescence.

Discussion
Using pFastbac Dual baculovirus transfer vector, we produced insect cell derived VLPs containing HA, M1, NA, and M2 from influenza virus as well as EGFP-incorporated VLPs. Despite the low NA content or deprived activity in the EGFP-fused NA, the production of VLPs was not impaired.
It was reported that the proteins expressed in Sf9 insect cells were devoid of sialylation as a result of the absence of detectable sialyltransferase activities [46,47] and CMP-sialic acids [48,49] and were N-glycosylated in high mannose type [51]. It implies that VLP production in insect cells is viable without the aid of NA activity for viral progeny release if HA is incorporated. Also, N-glycans in high mannose type could possibly enhance the HA binding to its sialylated receptors [52] and facilitate the uptake of VLPs by APCs.
From a previous report, HA expression in insect cells was retarded for posttranslational proteolysis [53] as noted in our Western blot result. However, the proteolytic process does not impair the binding capacity of HA but does for membrane fusion [54], as demonstrated in our fusion test ( Figure 5). It is noted that some VLPs did not fuse with the plasma membrane of the cells when pH was lowered to 5 and clumped on the cell surface. It raises an interesting question: since insect cell-derived VLP can induce cell-mediated immune response [10][11][12], how are the HA proteins anchored on the late endosomal membrane as a result of HA-induced membrane fusion or accumulated in the endosome lumen involved in cross presentation to MHC I molecules, leading to HA-specific CTL induction? To address this question, mammalian cell-derived influenza VLP with or without proteolytic site in HA protein could be tested for HA-specific CTL induction, protein processing as well as MHC I loading.
The size of expressed VLP is estimated to be ∼100 nm while the average size deduced from dynamic light scattering is ∼120 nm. The discrepancy between the two data can be at least partly accounted for by the fact that the latter reports the hydrodynamic diameter of the particle, namely, including the hydrated water molecules surrounding the VLP, thus the larger size [31].
Low pH-triggered conformational change can be detected by mAb IIF4 with the epitope within HA2 125-175, which was exposed upon lowering pH to 5.0 as deduced from 33-fold increase in binding affinity to IIF4 when switched to acidic pH. Again, this reaffirms that the constructed VLP can be used as a valid mimic to the native virus for the functional and structural study of hemagglutinin.
Further investigation of the biological activity for the synthesized VLP was performed on fusion to the Jurkat cell membrane induced by HA ( Figure 5) as probed by dequenching of the embedded R18 label due to its dilution arising from membrane merger. The result established the function exerted by HA as HA − VLP did not exhibit fusion activity. By measuring the fluorescence intensity of R18, IC50 of fetuin on binding inhibition of HA + VLP to Jurkat cell was estimated to be 147 μM (7.1 mg/ml), compared to a previous Journal of Biomedicine and Biotechnology result of 6.4 mg/ml for inhibition of viral infection by focusforming assays [40]. Based on viral binding and fusion both experiments can be used as screening platforms for inhibitors targeting viral entry. The fluorescent intensity of EGFP was shown to be influenced by the surrounding pH [55]. That could beget the low percentage of engulfed vesicles containing EGFP signal. However, it reinforced that the application NA chimera for antigen incorporation in VLP was feasible. Moreover, EGFP was reported as a noninvasive intracellular pH indicator [56]. The EGFP-VLP could be expanded for its utility as a reporter tag, particularly as a pH sensor inside the HA-targeted cells. Further study on the time variation of EGFP in the endosome can be conducted to monitor the time course of the fusion process and trafficking of VLP within the cell, which may provide insight into the antigen presentation of the virus and the dynamics and the route of virus invasion in the cell.

Conclusions
The extensive biophysical characterization of influenza VLP containing M1, HA, NA, and M2 viral structural proteins established a reliable protocol of synthesizing a large quantity of the virus mimetic for use in vaccine study, in particular for the HA as the major antigen. The incorporation of a fluorescent label demonstrates the feasible use of VLP as an antigen carrier as well as a possible tool for microscopic observation of the virus dynamics and interaction with the cellular components within the cell.