Stable High-Producer Cell Clone Expressing Virus-Like Particles of the Japanese Encephalitis Virus E Protein for a Second-Generation Subunit Vaccine

Abstract

We produced and characterized a cell clone (J12#26 cells) that stably expresses Japanese encephalitis virus (JEV) cDNA, J12, which encodes the viral signal peptide, premembrane (prM), and envelope (E) proteins (amino acid positions 105 to 794). Rabbit kidney-derived RK13 cells were transfected with a J12 expression plasmid, selected by resistance to marker antibiotics, and cloned by two cycles of a limiting-dilution method in the presence of antibiotics, a procedure that prevents the successful generation of E-producing cell clones. J12#26 cells secreted virus-like particles containing the authentic E antigen (E-VLP) into the culture medium in a huge enzyme-linked immunosorbent assay-equivalent amount (2.5 μg per 10⁴ cells) to the internationally licensed JE vaccine JE-VAX. E-VLP production was stable after multiple cell passages and persisted over 1 year with 100% expressing cells without detectable cell fusion, apoptosis, or cell death, but was suspended when the cells grew to 100% confluency and contact inhibition occurred. Mice immunized with the purified J12#26 E-antigen without adjuvant developed high titers of neutralizing antibodies for at least 7 months and 100% protection against intraperitoneal challenge with 5 × 10⁶ PFU of JEV when examined according to the JE vaccine standardization protocol. These results suggest that the recombinant E-VLP antigen produced by the J12#26 cell clone is an effective, safe, and low-cost second-generation subunit JE vaccine.

Japanese encephalitis virus (JEV), a member of the flavivirus family, is the causative agent of Japanese encephalitis (JE), which is a pandemic infectious disease of major public health importance in Asia (9, 48). Vaccination is the only effective way to prevent flavivirus infection in humans and domestic animals. Inactivated JEV and tick-borne encephalitis virus vaccines and attenuated yellow fever virus vaccine are in widespread production and use, whereas other flavivirus vaccines are under development or in human trials (6, 26, 37). The only licensed JE vaccine, JE-VAX, which is distributed commercially and available internationally, is formalin-inactivated JEV prepared from a number of JEV-infected mouse brains. The brain-derived whole virion vaccine is costly to manufacture and carries potential risks of allergic reactions to brain basic proteins or contamination by mouse prion proteins, and there are biosafety issues of manufacturing an infectious pathogen. Thus, the development of second-generation JE vaccines that are not derived from the brain, do not involve infectious JEV, and are of low cost is a top priority.

The 53-kDa envelope (E) glycoprotein of JEV has an important role in virus adhesion and entry into target cells through receptor binding (20, 34) and, therefore, in inducing neutralizing antibodies that protect hosts against JEV infection (8, 24, 32, 36). The protective epitopes on the E antigen are suggested to be formed in highly conformational structures of JEV virions (20, 34) based on antigenic analyses with panels of monoclonal antibodies (MAbs) (4, 12, 19, 25, 36) and studies on protective immunity in animals, JE patients, and JE-VAX recipients (7, 29-32, 50). In addition, molecular biological studies on the JEV genome indicate that expression of the premembrane (prM) and E genes in mammalian cells leads to the production of small, capsidless, noninfectious virus-like particles (VLP) that possess the E antigen (E-VLP), and its conformation-dependent protective epitopes are almost equivalent to those of the authentic E antigen in JEV virions (29-31, 44, 50). Thus, some attempts to develop second-generation JE vaccines have focused on the efficient production of the E-VLP antigen.

A recombinant vaccinia virus expressing cDNA encoding the prM and E proteins was a promising JE vaccine candidate; it produced extracellular E-VLP in cell cultures and induced neutralizing antibodies and protective immunity against JEV in vaccinated mice and rabbits (6, 26, 31, 50). Phase I human trials tested with NYVAC-JEV, a recombinant vaccinia virus constructed from an attenuated vaccinia virus strain, or with ALVAC-JEV, based on a canarypox virus vector, however, revealed their low immunogenicity, in particular, lower humoral immune responses in vaccinia-preimmune recipients (26, 28, 37). Furthermore, recombinant vaccinia virus vaccines do not yet have general international acceptance due to regulatory issues. On the other hand, plasmid DNA vaccines expressing the same cDNA region might provide an alternative to recombinant vaccinia virus. The DNA vaccines, however, also have low immunogenicity (1, 29); multiple injections and injecting the DNA into the skin with special gold-particle guns are required for the induction of neutralizing antibodies and JEV protection in animals. Otherwise, a high dose (100 μg) of DNA is required for a single intramuscular immunization (7). In addition, similar to the live virus-vectored vaccines, DNA vaccines are not accessible to humans in a large population due to potential risks.

Another alternative to the live virus-vectored or plasmid DNA-based JE vaccines, and probably an ideal one, is recombinant immunogens composed of the E-VLP antigen biosynthesized in vitro. To accomplish this, it is essential to establish stable cell lines that continuously express cDNA for prM-E and produce a large amount of the E antigen in a VLP form. This biotechnology could certainly overcome the safety and cost issues of JE vaccine production because infectious JEV and live virus vectors are not used, and therefore the physical containments for manufacturing vaccines are unnecessary. A recent paper (27), however, has cast a question as to the practicality of this system. The paper reported that the E-VLP antigen is toxic to the expressing host cells through cell-fusion activity, which is a cytopathic effect of the flavivirus E protein (14, 20, 34, 45). The E-mediated fusion or polykaryocyte-forming activity is implicated to be highly associated with the virion assembly, morphogenesis, and maturation processes by a recent model of the flavivirus replication cycle (20, 34). During virion formation, the prM-E protein in the endoplasmic reticulum (ER) membrane is processed, and prM is cleaved to the membrane (M) domain by a cellular protease in the trans-Golgi network to form prM/M and E oligomers, resulting in toxic fusion activity. Consequently, expression of E-VLP decreases stability and viability of the expressing cells, and, conversely, prevents the generation of high-VLP-producer cell lines, leading to the generation of only low-producer lines.

This contradictory notion appears to be the case for a JE-4B cell line (22) established by transfection of COS-1 cells with a prM-E expression plasmid used as a DNA vaccine by the same authors (7); JE-4B cells produced E-VLP at a maximum enzyme-linked immunosorbent assay (ELISA) titer of 1:16 to 1:32. Similarly, our previous attempts with a cytomegalovirus (CMV) promoter/enhancer-based plasmid and Vero cells were unsuccessful. Only a few neomycin-resistant cell lines expressed low levels of the E antigen in low frequencies (unpublished data). To overcome the contradiction, Konishi et al. (27) introduced a mutation at the prM/M cleavage site to prevent authentic processing and to reduce the toxic fusion activity of the expressed VLP and thus established an F-cell line after multiple efforts. The amount of F-cell VLP was 1 μg per 10⁷ cells, and Freund's adjuvant was required to induce neutralizing antibodies and protective immunity in mice.

The present study describes the establishment of an extremely high-producing cell clone, J12#26, that expresses a cDNA clone, designated J12, encoding prM and E, accompanied by an authentic signal peptide located in the carboxyl-terminus of the capsid (C) protein to facilitate the extracellular secretion of E-VLP (7, 31, 34, 47). J12 cDNA was driven by a β-actin promoter with higher efficiency than that of CMV. Rabbit kidney-derived RK13 cells were used as host cells because kidney-derived cells develop an exocytic network (11, 41), and RK13 cells were markedly resistant to infection and cytopathic effects of JEV (unpublished data). The results demonstrate that J12#26 cells produced approximately 2.5 μg of E-VLP per 10⁴ cells (theoretically 2.5 mg per 10⁷ cells) stably and continuously with 100% positivity over 1 year after the establishment and that the purified J12#26 antigen induced high titers of neutralizing antibodies and 100% protection in mice without adjuvant, similar to JE-VAX when examined basically according to the JE vaccine standardization protocol.

MATERIALS AND METHODS

Construction of plasmids carrying prM and E genes of JEV.

A JEV cDNA clone, J12, encoding the viral signal peptide of the carboxyl terminus of C, prM, and E proteins (amino acid positions 105 to 794) of the Beijing-1 strain was recovered from a vaccinia virus transfer plasmid, pS13J12Gal (unpublished data), by digestion with SalI and XhoI and then inserted into the XhoI site of a pCAGGS expression vector (40) to generate pCAGJ12. The fragment containing the CMV enhancer, the β-actin promoter and intron, and J12 cDNA was recovered from the pCAGJ12 plasmid and replaced with the regulatory region of a pEFBOSbsr plasmid (a kind gift from M. Tatsumi) to confer resistance to blasticidin S (BS) (Calbiochem, Inc., Darmstadt, Germany). The resulting plasmid, pCAGJ12bsr (Fig. 1), was used to establish stable cell lines.

FIG. 1.

Schematic presentation of pCAGJ12bsr. The expression vector pCAGJ12bsr contains the human CMV immediate-early (CMV-IE) enhancer, the chicken β-actin promoter (P_β) and intron derived from pCAGGS, the simian virus 40 (SV40) polyadenylation site (polyA) and t intron, and the BS resistance (bsr) gene derived from pEFBOSbsr. A JEV cDNA fragment, J12, encoding the viral signal peptide, prM, and E genes was inserted between the β-actin intron and the SV40 poly(A) site. A part of the JEV genome is represented schematically at the top of the figure with amino acid numbers, where 1 is the C protein initiation codon.

Establishment of stable cell lines constitutively producing the E antigen.

Subconfluent RK13 cells (CCL-37; American Type Culture Collection) grown in Eagle's minimal essential medium (MEM) supplemented with 10% fetal bovine serum (FBS) were transfected with pCAGJ12bsr DNA with FuGene6 (Roche Diagnostic Co., Tokyo, Japan) and cultured for 2 days. Cells grown just to confluence were split at l:6 and selected for BS resistance by growth in medium containing BS (10 μg/ml). BS-resistant colonies were picked 10 days later and transferred to a 24-well plate. The E-antigen amounts in culture supernatants from wells with the growing resistant cells were examined by ELISA. A cell line, 20, expressing over 700 ng of the E antigen per ml (Fig. 2) was selected and cloned by two cycles of the limiting dilution method (0.2 cells/well) in BS-containing medium. One clone, J12#20.14.26, here designated J12#26, was selected and maintained in medium supplemented with BS (5 μg/ml).

FIG. 2.

Establishment of stable cell lines continuously producing JEV E antigen. (A) BS-resistant colonies were transferred to a 24-well plate, and culture medium from each well was screened for the amount of the E antigen by ELISA. (B and C) Cells were expanded from the 24-well plate to 6-cm dishes, and (B) cell lysates and (C) precipitates of the culture supernatants by ultracentrifugation were analyzed by Western blotting with rabbit anti-JEV antiserum. RK represents the parental RK13 cells. The asterisk indicates that the E antigen amount was over the detection limit of the ELISA. Arrowheads on the right of panels B and C indicate the 53-kDa E protein of JEV.

Antigen capture ELISA for quantification of E antigen.

The amounts of the E antigen were quantified by a sandwich ELISA with a JEV-specific MAb, 503, which is one of the strongest neutralizing antibodies against JEV (24). Briefly, MAb 503 (1 μg/ml) was coated onto 96-well ELISA plates overnight at 4°C. The coated plates were incubated with various dilutions of samples for 1 h at 37°C to capture the JEV E antigen. The amounts of the E antigen captured to the solid phase were detected with horseradish peroxidase-conjugated MAb 503 by incubation for l h at 37°C and subsequently with Dako TMB+ substrate (Dako Corp., Carpinteria, Calif.). The enzyme reaction was stopped with 2 M H₂SO₄, and the optical density was measured at 450 and 690 nm. A stock of purified formalin-inactivated JEV (JE-VAX; 9.1 μg of protein/ml) was serially twofold diluted starting from 455 ng of protein/ml and used as a standard. The E-antigen amounts of each sample were determined from a standard curve of the diluted JE-VAX and expressed as ELISA equivalents to JE-VAX.

Indirect immunofluorescence.

J12#26 cells were cultured to subconfluency on eight-spot glass slides. The cells were washed with phosphate-buffered saline (PBS), air-dried, fixed with acetone for 10 min, and stored at −20°C until examination. The fixed cells were treated with 0.1% Triton X-100 for 5 min and then incubated with neutralizing MAbs, JEV-specific 503, group-specific N.04 (24, 25), or rabbit anti-JEV antiserum, followed by fluorescein isothiocyanate-conjugated goat anti-mouse (Fab)₂ or anti-rabbit (Fab)₂, respectively. RK13bsr cells that were rendered BS resistant by transfection with the vector plasmid DNA were used as a negative control.

Electron microscopy.

J12#26 and RK13bsr cells were treated with trypsin and collected by low-speed centrifugation. The pelleted cells were fixed with 2.5% glutaraldehyde for 2 h and postfixed with 1% OsO₄ for 1 h. The fixed specimens were dehydrated and embedded in Epon 812 (Poly/Bed). Ultrathin sections were cut, stained with uranyl acetate and lead citrate, and examined with a JEOL 1200Ex transmission electron microscope (Hitachi Co., Tokyo, Japan).

Western blotting.

Culture supernatants of BS-resistant cells were ultracentrifuged at 150,000 × g for 2 h. The precipitates were washed with PBS and lysed with Laemmli's sample buffer. Monolayers of the cells were washed with PBS and lysed with 1% Triton X-100. Cell lysates were clarified by centrifugation at 3,000 rpm for 10 min and mixed with an equal volume of 2× Laemmli's buffer. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis, electrotransfer, and Western blotting were performed as described previously (21). Briefly, proteins separated in 10% acrylamide gels and transferred onto polyvinylidene fluoride membranes (Millipore, Bedford, Mass.) were exposed to rabbit anti-JEV serum followed by alkaline phosphatase-conjugated anti-rabbit immunoglobulin G (BioSource, Nivelles, Belgium). Protein bands were visualized with a 5-bromo-4-chloro-3-indolylphosphate-nitroblue tetrazolium (BCIP-NBT) substrate kit (Kirkegaard & Perry Labs Inc., Gaithersburg, Md.).

Rate zonal centrifugation.

Culture medium collected at 4 days after subculture of J12#26 cells or JEV infection of Vero cells was clarified by centrifugation at 3,000 rpm for 10 min and filtration through a 0.45-nm-pore-size membrane. A 200-μl aliquot of the clarified supernatants was layered onto 5 to 45% (wt/wt) linear sucrose gradients and subjected to rate zonal centrifugation at 150,000 × g for 2 h. Fractions were collected dropwise starting from the bottom of the tubes. Each fraction of the J12#26 antigen and Vero cell-derived JEV was diluted 1:40 and 1:10 with PBS, respectively, and examined for determination of E-antigen amounts by ELISA. When the E antigen was purified from J12#26 cell cultures, the polyethylene glycol-pelleted fractions were subjected to rate zonal centrifugation in 5 to 25% (wt/wt) linear sucrose gradients.

Purification of E antigen.

The culture medium of J12#26 cells was clarified by centrifugation at 3,000 rpm for 10 min and then at 10,000 rpm for 15 min. The E antigen was precipitated from the clarified supernatants with 12.5% polyethylene glycol 6000 (Wako Chemicals, Osaka, Japan) at 4°C for 2 h. The precipitates were collected by centrifugation at 10,000 × g for 30 min and dissolved in 50 mM Tris-HCl (pH 7.5)-150 mM NaCl. Insoluble materials were removed by low-speed centrifugation, and the supernatants were subjected to rate zonal centrifugation as described above. E-antigen content was determined for each fraction collected from the sucrose gradients by sandwich ELISA, and peak fractions were pooled. Sucrose and low-molecular-weight materials were removed from the E-antigen pool by Sephadex G-25 chromatography with NAP-25 columns (Amersham Pharmacia Biotechnology, Uppsala, Sweden) equilibrated and eluted with PBS. The purified E antigen in PBS was stored at 4°C until immunization of mice.

Immunization of mice and test for neutralizing antibodies in serum samples.

Groups of 6-week-old female BALB/c mice (SLC, Shizuoka, Japan) were immunized intraperitoneally twice at a 7-day interval with either 1/16 of the adult human dose of JE vaccine (JE-VAX; Biken, Osaka, Japan), the equivalent ELISA dose of the purified E antigen from J12#26 cell cultures, or PBS as a negative control, basically according to the protocol of the minimum requirements for JEV vaccine in Japan. Mice were bled at various times up to 210 days after the second immunization. Pooled serum samples from each group were tested for virus neutralizing antibodies as described previously (46). Briefly, serial twofold dilutions of heat-inactivated serum samples were incubated with 500 PFU of the JEV Beijing-1 strain in MEM-2% FBS for 1.5 h at 37°C. A 200-μl aliquot was examined in triplicate for residual virus infectivity by plaque assay on Vero cell monolayers in multiwell plates. The percent plaque reduction was calculated relative to virus controls incubated without mouse serum. Neutralizing antibody titers were expressed as the reciprocal of the serum dilution yielding a 50% reduction in the mean plaque number versus control wells.

Challenge of mice with JEV.

For protection experiments, 4-week-old female BALB/c mice were used. Groups of mice were vaccinated without adjuvant as described above for the immunogenicity study and challenged at 6 weeks of age with intraperitoneal injections of 5 × 10⁶ PFU of the Beijing-1 strain of JEV. Simultaneous with the JEV challenge, mice were injected intracerebrally with 20 μl of PBS to disrupt the blood-brain barrier as described previously (46). The challenged mice were observed daily for 3 weeks.

RESULTS

Establishment of stable cell clone continuously expressing E antigen.

RK13 cells, which are resistant to infection and cytopathic effects of JEV, were transfected with pCAGJ12bsr DNA (Fig. l) to establish stable cells expressing E-VLP of JEV. The transfected cells were selected by growth in BS-containing medium, and 23 BS-resistant cell colonies were picked and tested for expression of the E antigen. The amounts of E antigen determined by ELISA in culture supernatants collected at the growing stage in a 24-well plate are shown in Fig. 2A. The result indicates that 17 of 23 colonies expressed the E antigen and that colony 20 produced over 700 ng of the E antigen per ml before generating a monolayer. Colonies 1 to 23 were expanded from the 24-well plate to 6-cm dishes, and the E antigen in cell lysates (Fig. 2B) and in precipitated fractions of culture supernatants by ultracentrifugation (Fig. 2C) was examined by Western blotting. The 53-kDa protein band corresponding to the JEV E antigen was detected in both cell lysates and culture supernatants. The intensity of the 53-kDa band for colonies 1 to 23 closely correlated with the E-antigen amounts determined by ELISA (Fig. 2).

As colony 20 provided the best results, the cells were cloned with a limiting dilution method in the presence of BS. After two cloning cycles, a total of 51 progeny clones in 96-well plates all produced at least 5 μg of the E antigen per ml in culture supernatants. A clone, J12#20.l4.26, designated J12#26, was selected, expanded, and maintained in BS-containing medium. The J12#26 cell clone had undistinguishable morphology from parental RK13 cells and induced no polykaryocyte formation (Fig. 3A and B). When examined by indirect immunofluorescence, 100% of J12#26 cells were stained with rabbit anti-JEV antiserum (Fig. 3D) and with neutralizing MAbs JEV-specific 503 (Fig. 3E) and group-specific N.04 (Fig. 3F) (24, 25).

FIG. 3.

Expression of JEV E antigen in a stable cell clone, J12#26. RK13 (A and C) and J12#26 (B, D, E, and F) cells were cultured on chamber glasses. Micrographs of growing RK13 (A) and J12#26 (B) cells are shown at the top of the figure. Cells were fixed with acetone and stained with rabbit anti-JEV serum (C and D) and with neutralizing MAbs 503 (E) and N.04 (F).

The morphology of J12#26 cells was further examined by transmission electron microscopy. Figure 4A confirms that the appearance of J12#26 cells was a normal epithelial type, similar to that of parental RKl3 cells (not shown). Differences between J12#26 cells and RKl3bsr cells that were rendered BS resistant by transfection with vector plasmid DNA were clearly observed around the Golgi apparatus located in the juxtanuclear area. Many small particle-like and electron-dense structures were present in the cisternae of the ER and within the Golgi apparatus of a J12#26 cell (Fig. 4B), whereas such structures were not observed in an RKl3bsr cell (Fig. 4C). These results suggest that J12#26 is established with l00% clonality, without morphological changes such as elastic transformation and polykaryocyte formation, and expresses and releases E antigen possessing authentic JEV-neutralizing epitopes.

FIG. 4.

Electron micrographs of a J12#26 and a BS-resistant RK13bsr cell. (A) Low magnification of a J12#26 cell. Higher magnification of (B) the juxtanuclear area in a J12#26 cell and (C) an RK13bsr cell transfected with vector plasmid DNA. Note the small spherical electron-dense structures (arrows) within the cisternae of the ER and Golgi apparatus in a J12#26 cell (B).

High production of E antigen and passage stability of Jl2#26 cells.

The amounts of the E antigen produced by J12#26 cells were examined within five passages in 25-cm² flasks after the original cells were expanded and cryopreserved. The cells were subcultured with relatively high seeding densities of 1 × 10⁵ or 2 × 10⁵ cells/ml to a 24-well plate. The culture medium was removed and exchanged with fresh medium every 24 h, and E-antigen amounts in the medium were determined by ELISA. As shown in Fig. 5A, the rates of E-antigen production increased as cells grew but decreased after 48 h of cultivation when cells reached l00% confluency. In other experiments, lower densities (1 × 10⁴ to 10 × 10⁴ cells/ml) of cells were seeded and cultured for 72 h without medium changes. A l00-μl aliquot of culture medium was collected at 24, 48, and 72 h, and the amounts of accumulating E-antigen were determined. Figure 5B shows a representative result. Cultures plated at lower cell doses (1 × 10⁴ to 5 × 10⁴ cells) released and accumulated the E antigen in the medium in parallel with cell growth, and 2 × 10⁴ cells produced approximately 5 μg of E antigen per ml during 72 h of cultivation. When cells were inoculated at a high dose (10⁵ cells), they reached l00% confluency 48 h later, and further accumulation of the E antigen was slight from 48 to 72 h. These results suggest that J12#26 cells generate and release a huge amount of E antigen during growing phases but suspend production due to contact inhibition.

FIG. 5.

Production of E antigen by J12#26 cells. (A) Cells (10⁵ [open columns] or 2 × 10⁵ [shaded columns] cells) were seeded in a volume of 1 ml in 24-well plates. Culture medium was collected and replaced every 24 h with fresh medium. Amounts of the E antigen produced each 24 h were determined by ELISA. Bars show the mean of triplicate cultures ± standard deviation. (B) Various numbers (10⁴ [∗], 2 × 10⁴ [▴], 5 × 10⁴ [▪], and 10⁵ [•]) of cells were seeded in a volume of 1 ml in 24-well plates. A 100-μl aliquot of culture medium was collected 24, 48, and 72 h later. The amount of E antigen in the culture medium was determined by ELISA. Results are expressed as the mean of duplicate cultures.

To evaluate the stability of the high production rate of the J12#26 clone, the ability of the cells to produce E antigen was examined for 30 passages. J12#26 cells were subcultured every 3 or 4 days, that is, twice a week, in 25-cm² flasks after the original cells (Ori) were cryopreserved. E-antigen contents in culture medium collected at the 2nd, 5th, 10th, 20th, and 30th passages were determined by ELISA (Fig. 6). Although the amount of E antigen produced by J12#26 cells differed little between passages, probably due to cell confluency at the passage time, more than l0 μg of E antigen per ml was detectable throughout the monitoring period. Even on the 30th passage, the frequency of E-expressing cells was l00%, as judged by indirect immunofluorescence. In addition, similar levels of the E antigen were produced after more than a year. These results indicate that the J12#26 clone is extremely stable and preserves its high-E antigen-producing nature.

FIG. 6.

Effect of long-term passage on production of E antigen by J12#26 cells. J12#26 cells were passaged every 3 or 4 days in 25-cm² flasks. Culture supernatants collected at the time of passages were stored at 4°C and examined for E antigen contents by ELISA. Ori represents the first J12#26 cells expanded to a 75-cm² flask after the second limiting-dilution cloning.

Rate zonal centrifugation of J12#26 E antigen.

The E antigen in the culture supernatants of J12#26 cells was characterized by comparison with that of JEV-infected Vero cells. A portion of the culture fluids was resolved by rate zonal centrifugation on 5 to 45% linear sucrose gradients. Figure 7 shows ELISA profiles of the gradient fractions diluted 1:40 and 1:10 for the J12#26 antigen and Vero cell-derived JEV, respectively. The J12#26 antigen sedimented slowly and formed a single peak at the same sucrose density position as JEV-derived slowly sedimenting particles, termed SHA. The SHA peak of JEV was almost undetectable by ELISA when each fraction was diluted 1:40, but was detectable as a small peak at a 1:10 dilution. On the other hand, JEV virions sedimented at a higher density position and formed a large peak. In addition, the J12#26 E-antigen peak became scarcely detectable by ELISA when each fraction was diluted for the assay with buffer containing EDTA (l mM) or Triton X-100 (0.5%) instead of PBS. These results suggest that J12#26 cells produce E antigen as a small VLP form, which possesses the 503 neutralizing epitope that is disrupted in conditions that disrupt the physical particulate structure.

FIG. 7.

Rate zonal centrifugation of J12#26 E antigen. Culture supernatants harvested at 4 days after subculture of J12#26 cells (open circles) or JEV infection of Vero cells (solid circles) were clarified by centrifugation and filtration. A 200-μl portion of the clarified supernatants was fractionated in 5 to 45% sucrose gradients. Each fraction collected from the bottom was diluted 1:40 and 1:10 for the J12#26 antigen and JEV, respectively, and examined for E antigen amounts by ELISA.

Yield of J12#26 E antigen from culture supernatants.

The yield and purification of E antigen from 100 ml of medium of J12#26 cells cultured in 15-cm plates was examined by ELISA (Table l) and Western blotting (Fig. 8). The culture supernatant was clarified by centrifugation and used as starting material. Approximately 90% of the E antigen was concentrated and recovered by polyethylene glycol precipitation. The E-antigen concentrate was then purified by rate zonal centrifugation on linear sucrose gradients. Peak fractions of the E antigen in the gradients were pooled and applied to Sephadex G-25 columns. Loss of the E antigen during these purification processes was less than 50%, and therefore more than half of the E antigen in the starting culture fluids was recovered, as determined by ELISA (Table 1). The samples of preparations in each purification step were then analyzed by Western blotting (Fig. 8). JEV-specific E antigen was observed as a protein band with a 53-kDa molecular mass in formalin-inactivated JEV virions. Similarly, preparations at the polyethylene glycol precipitation, sucrose gradient purification, and Sephadex G-25 purification steps all contained the 53-kDa E antigen as a major protein band (Fig. 8).

FIG. 8.

Western blot analysis of J12#26 E antigen preparations. J12#26 E antigen in culture supernatants was purified according to the procedures in Table 1. The E antigen preparations were normalized by ELISA. Each 150 ng of E antigen in a 10-μl volume was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and analyzed by Western blotting with rabbit anti-JEV serum. Purified formalin-inactivated JEV (Beijing-1 strain) was used as a standard for identification of the JEV E antigen. Prestained molecular size markers (broad range; Bio-Rad Laboratories) are shown in the right lane. Numbers on the right of the figure indicate the molecular mass of the markers (in kilodaltons).

TABLE 1.

Purification and yield of J12#26 E antigen from culture supernatant^a

Purification step	Volume (ml)	E antigen ELISA-equivalent to JE-VAX		Yield (%)
		Concn (μg/ml)	Total amt (μg)
Clarified medium (10,000 rpm, 15 min)	100	5.9	590	100
Polyethylene glycol precipitation	3.56	152.0	541	92
Sucrose gradient	11.4	37.4	426	72
Sephadex G-25 chromatography	11.4	27.0	308	52

^aCulture medium of J12#26 cells was harvested 24 h after subculture. An aliquot of samples from each purification step was examined for JEV E-antigen content by ELISA with purified and inactivated JEV virions (JE-VAX) as a standard.

Immunogenicity and protective efficacy of purified J12#26 E antigen.

Induction of neutralizing antibodies was examined for the purified J12#26 E antigen in the absence of adjuvant. Groups of BALB/c mice were intraperitoneally immunized twice at a 7-day interval with either 1/16 of human dose of JE vaccine (approximately 300 ng of JE-VAX), the purified J12#26 antigen in an ELISA equivalent to the licensed vaccine, or PBS as a negative control. Mice were bled at various times after the second immunization and tested for neutralizing antibodies with a plaque reduction assay. Table 2 shows neutralizing antibody titers to JEV in pooled serum samples collected from five mice at various days after the second immunization. Mice immunized with the purified J12#26 E antigen developed high levels of neutralizing antibodies, as did the mice immunized with the licensed JE-VAX. The neutralizing antibody levels on day 14 were as high as those on day 7 (experiment 1). When neutralizing antibody responses were examined for up to 7 months, comparable neutralizing titers were maintained for the period of time examined without reduction or loss of the titers in both the J12#26 antigen and JE-VAX groups (experiment 2).

TABLE 2.

Induction of neutralizing antibodies in mice immunized with J12#26 E antigen^a

Expt	Day	Neutralizing antibody titer
		PBS	JE-VAX	J12#26 E antigen
1	7	>1/20	1/640	1/640
	14	ND	1/640	1/640
2	10	>1/20	1/1,280	1/640
	30	ND	1/640	1/320
	120	ND	1/640	1/640
	210	>1/20	1/640	1/1,280

^aGroups of 6-week-old BALB/c mice were intraperitoneally immunized twice at a 7-day interval with PBS, 1/16 of the human dose of JE-VAX, or the ELISA-equivalent dose of J12#26 E antigen without adjuvant. JEV-neutralizing titers in serum collected at the indicated days after the second immunization were expressed as the reciprocal of the serum dilution yielding a 50% reduction in plaque number. ND, not determined.

The protective efficacy of the J12#26 E antigen was examined by challenge with infectious JEV. Groups of 4-week-old BALB/c mice were immunized without adjuvant as in the immunogenicity study described above and intraperitoneally challenged at 6 weeks of age with 5 × 10⁶ PFU of the Beijing-1 strain (Table 3). As the challenge dose of JEV was more than 50 times the 50% lethal dose in 6-week-old BALB/c mice, all the animals that received PBS died within 6 days after the challenge. In contrast, all mice vaccinated with the J12#26 E antigen developed no detectable clinical signs and were completely protected against challenge with JEV. This was also the case for mice vaccinated with the licensed JE-VAX. These results indicate that the J12#26 E antigen is as immunogenic and protective as the licensed JE vaccine in 503 ELISA-equivalent doses.

TABLE 3.

Protection of mice against challenge with JEV by vaccination with J12#26 E antigen^a

Vaccination	No. of mice surviving/ no. tested	Survival rate (%)
PBS	0/6	0
JE-VAX	5/5	100
J12#26 E antigen	8/8	100

^aGroups of 4-week-old BALB/c mice were vaccinated twice with PBS, JE-VAX, or purified J12#26 E antigen as in Table 2. The mice were intraperitoneally challenged at 6 weeks of age with 5 × 10⁶ PFU of the Beijing-1 strain and intracerebrally injected with 20 μl of PBS.

DISCUSSION

Japanese encephalitis is still endemic and epidemic in Asia and has recently expanded its geographic range to Indonesia and Australia, although the internationally licensed, formalin-inactivated, whole-virion JE vaccine (JE-VAX) has greatly reduced the incidence of the disease in certain Asian countries, including Japan (9, 16, 23, 35, 48). As the inactivated vaccine is prepared from virus-infected mouse brain and therefore costly, improvement of the JE vaccine and development of new-generation vaccines have been promulgated by the World Health Organization (6). There are inactivated cell culture vaccines (46), live attenuated vaccines (33), poxvirus-vectored vaccines (31, 50), chimeric yellow fever virus-JEV vaccines (5, 15, 38), and naked DNA vaccines (7, 29; reviewed in references 6, 26, and 37). Here we describe another alternative, a recombinant subunit vaccine.

We established an extremely high-producing cell clone, J12#26, which released more than 2.5 μg of E-VLP ELISA equivalents to JE-VAX per ml in culture supernatants of 10⁴ cells. This value was determined in experimental conditions with 24-well plates without medium changes and will be further increased in optimal culture conditions. In addition, J12#26 cells were highly stable in serum-free medium and continuously produced E-VLP in more than a JEV-equivalent amount of 10⁹ PFU/ml obtained from infected Vero cells for cell culture vaccine production (unpublished data). J12#26 cells grew well in cultures and there was little, if any, polykaryotic, apoptotic, and lytic cell morphology. The J12#26 cells had the typical features of normal epithelial cells and contact inhibition that was indistinguishable from those of parental cells. This normal cell nature appears to be advantageous as a producer of immunogens with minimal release of undesirable cellular DNA and proteins into culture supernatants and hence makes it easier to purify the immunogens than a Vero cell vaccine system because JEV is a lytic virus and the infected Vero cells leak significant amounts of the undesirable substances due to cell death.

A major concern of our subunit vaccine system may be the suitability of an RK13-derived cell line, J12#26, for the production of E antigen. Continuous cell lines are now considered suitable substrates for the production of biological medical substances and possess distinct advantages over primary cell substrates (13). Therefore, a World Health Organization Study Group recommended the establishment of well-characterized cell lines (49). However, Bolin et al. reported that many American Type Culture Collection cell lines, including RK13 (CCL-37) cells, contained bovine viral diarrhea virus (BVDV) (3).

Our RK13 cells were supplied more than 20 years ago from the Department of Viral Disease and Vaccine Control, National Institutes of Health, and probably originate from CCL-37. Thus, the J12#26 and parental RK13 cell lines thawed from frozen stocks and passaged twice were tested for the BVDV RNA genome by pestivirus-specific nested reverse transcription-PCR (17). The preliminary results demonstrated that BVDV RNA was positive and negative in the culture supernatants and cell lysates of the cell lines, respectively, and that the FBS used for the cell cultures was BVDV RNA positive (unpublished data). Although further experiments are in progress to test for BVDV in available FBS lots and several cell lines related to this study by nested reverse transcription-PCR, immunofluorescence, and infectivity analyses with BVDV-sensitive cells, these results suggest that the BVDV genome detected is from the contaminated FBS.

As reported previously, many lots of human live viral vaccines manufactured with primary chicken embryo, rabbit kidney, or monkey kidney cell cultures are contaminated with BVDV RNA (17, 43). As BVDV RNA or infectious BVDV in FBS is the most likely source of the contamination of human biologicals, FBS control is certainly important (10). Contaminated cell lines may become free of BVDV after multiple passages with BVDV-negative FBS-containing or serum-free medium. Formalin inactivation may be included in the purification processes of antigens similar to inactivated vaccines. However, it may be noted that primary cell cultures have been in worldwide use for the production of live and inactivated viral vaccines for more than 40 years, and experience has indicated that these products are safe (49). Although freedom from BVDV RNA is desirable, vaccines should be considered in view of the balance among their benefit, cost, and potential risks.

As to immunogenicity, the purified E-VLP antigen from the culture medium of J12#26 cells was as immunogenic as JE-VAX at ELISA-equivalent doses of 0.3 μg without adjuvant. In mice, the E-VLP antigen induced similar levels of persistent neutralizing antibody titers and 100% protection against JEV compared to JE-VAX, possibly affording significantly higher and prolonged neutralizing antibody titers than JE-VAX if mice are vaccinated at protein-equivalent doses. This efficacious property of the J12#26 antigen appears to depend on its particulate structure. A well-known characteristic of flavivirus is that expression of the prM and E genes without the C gene leads to the formation and secretion of capsidless VLP, which is smaller in size than virions but exhibits a native antigen structure of the oligomerized E protein on its surface very similar to that on the virion surface (20, 34). In this regard, the J12#26 antigen was expressed from prM-E cDNA of JEV.

Further support for a particulate form of the antigen is a finding that it sedimented slowly in a linear sucrose gradient and formed a single peak at the same density as JEV-derived SHA particles. This antigen peak was detected with neutralizing MAb 503 (24), which possibly recognized the conformational structure presumably arranged by amino acids at positions 52, 126, 136, and 275 of the E protein (39). In addition, our antigen-capture ELISA utilized MAb 503 as not only the capture antibody but also the detecting antibody. This ELISA system is likely to detect only the E antigen with the multiple 503 epitopes on steric conformation, because E-protein molecules with a single 503 epitope must be saturated with the capture MAb 503 before reacting with the detecting MAb 503. Another finding favorable to the particulate nature of the J12#26 antigen was electron microscopic observations of the cells; small spherical structures (approximate diameter of 20 nm) were abundant in the ER and Golgi apparatus, although the electron-dense particles were not directly proved by immunoelectron microscopy to be immature secreting forms of the E antigen. Detailed characteristics of the J12#26 E antigen purified from serum-free cultures, such as its negatively stained particle appearance, the existence of the M protein in the VLP form, and high titers of hemagglutinating activity will be described elsewhere.

It is important to understand the possible reasons why the J12#26 clone is such a stable high producer and resistant to the cytotoxic fusion activity of the E-VLP antigen. Our initial attempts to establish a high producer from Vero cells were unsuccessful with use of the same prM-E region and CMV promoter-based pCDNA vector. During the selection step with G418, a number of colonies changed cell morphology to a round shape and ceased cell growth. Only a few cell lines expressed the E antigen in levels detectable by Western blotting but not in realistic levels to function as a producer of the immunogen. The high-producing nature of the J12#26 clone might be due to the high efficiency of the β-actin promoter in mammalian cells (40) compared to that of CMV. In turn, however, the higher expression of the E-VLP antigen should be more toxic to the expressing cells, resulting in cell fusion, apoptosis, and death. Therefore, the high resistance of J12#26 cells might be explained by the shorter retention time of the toxic antigen.

In this regard, kidney cells develop the exocytic pathway relatively well and have been conveniently used since Golgi in the late 18th century for analysis of cellular transport or secretion mechanisms through the ER/Golgi apparatus (2). Vero and RK13 cells, however, originate from monkey and rabbit kidney, respectively, and it is unclear whether RK13 cells secrete more efficiently than Vero cells. Alternatively, the secretion signal sequences attached upstream of prM might provide an important contribution to the differences in releasing efficiency between our and other's cell lines. As presented in Table 4, the length of the authentic signal peptides of JEV used in each expression vector of prM-E cDNA was 23 amino acids for our pCAGJ12bsr (J12#26 cells), 19 for pcJEEP (F cells) (27, 29), and 29 for pCDJE2-7 (JE-4B cells) (7, 22), which seem to be high, medium, and low producers, respectively. The difference in length was only 4 amino acids shorter or 6 amino acids longer than our signal region and is unlikely to reflect the efficiency of E-VLP release.

TABLE 4.

Amino acid sequences used as a secretion signal in continuous cell lines expressing the prM-E gene of JEV^a

Cell line	Host cell	Recombinant plasmid	Plasmid vector	Promoter	Signal sequence upstream of prM	JEV strain of origin of prM-E	Refer- ence(s)
J12#26	RK13	pCAGJ12bsr	pCAGGS	β-Actin	MIRGGNEGSIMWLASLAVVIACAGA//	Beijing-1	This study
JE-4B	COS-1	pCDJE2-7	pCDNA3	CMV	MGRKQNKRGGNEGSIMWLASLAVVIACAGA//	SA14	7, 22
F	CHO-K1	pcJEEP/(pcJEME)	pCDNA3	CMV	MEGSIMWLASLAVVIACAGA//	Nakayama	27, 29

^aAmino acids are expressed in the single-letter code. The first M is the artificially introduced initiation codon. The junction between the C and prM proteins is indicated by //.

Another difference among those cell lines is the donor JEV strains of the prM-E gene (Table 4). J12#26, F, and JE-4B cells carried the gene from the Beijing-1, Nakayama, and SA14 strains of a Chinese isolate, virulent strain, and parental strain of attenuated live vaccine, respectively. A search of the GenBank database revealed that only 7 and 5 amino acid substitutions occurred in the E gene of Beijing-1 compared with that of Nakayama and SA14, respectively. If the toxicity of the expressed E-VLP is correlated with the virulence mapped to the E protein of donor strains (18, 20), its order might be medium, high, and low, and the production stability should have resulted in the reverse order. This was, however, not the case.

Accordingly, any bearing might be attributed to host cell difference (Table 4). Certain cell lines such as Vero, COS, CHO, BHK, and C6/36 cells have been so far used for JEV studies. Vero cells are one of the most susceptible cells to JEV infection and give rise to high virus titers (over 10⁹ PFU/ml). Therefore, this cell line is used as a JEV producer for the development of inactivated vaccines in cell culture systems (6, 26, 37). This JEV-susceptible nature of Vero cells appears to make them sensitive to the cytotoxic effects of the E-VLP antigen and hence inadequate as parental cells of stable producer lines, which we should have realized earlier. In marked contrast, RK13 cells are less susceptible to JEV infection. Our preliminary experiments demonstrated the following points: (i) JEV did not induce plaque formation nor cytopathic effects on RK13 cell monolayers even when inoculated at a high multiplicity of infection of 5; (ii) only low virus titers were detected in culture supernatants of RK13 cells inoculated with JEV in conditions where Vero cells were able to generate high titers; and (iii) JEV proteins were undetectable in infected RK13 cells, as determined by Western blotting and indirect immunofluorescence analyses, whereas they were easily detectable after transfection with expression plasmids such as pCAGJ12bsr (unpublished data).

Taken together, these preliminary observations suggest poor JEV receptor densities on the surface of RK13 cells, which do not allow induction of cytopathic effects but allow expression, processing, assembly, and release of viral structural proteins once JEV enters into the cells. This hypothesis likely explains the underlying mechanism for RK13 cells to resist cytotoxic fusion effects of E-VLP expressed within them, and the reason for the easy establishment of the stable high-producer cell clone J12#26.

Here we have described an alternative vaccine candidate composed of the recombinant nonreplicable E-VLP antigen. Our subunit JE vaccine system based on a high-producer clone likely overcomes the safety or cost issue of JE vaccine candidates. Moreover, this E-VLP antigen system is conceivably applicable to the development of other flavivirus vaccines, because of high similarities among the flaviviruses in their replication cycle, the E protein as the major protective antigen, and high antigenicity of E-VLP, and also applicable to diagnostic virus antigens. Further studies are in progress to address the usefulness of our E-VLP in JE diagnostic systems.

Finally, three JE cases occurred this September in a city in Japan. It is unclear at present whether the cases signify JE reemergence in Japan. It is important to keep in mind, however, that flavivirus diseases are not restricted to tropical or subtropical areas but are spreading globally due to the development of transportation and air-conditioning systems and the warming climate of the Earth, which enable infected mosquitoes to move across the sea and tide over the winter, as exemplified by the recent outbreak of JE in Australia (16, 35) and that of West Nile disease in North America (42). This study might provide an effective strategy to prevent emerging and reemerging flavivirus infections.