|Home | About | Journals | Submit | Contact Us | Français|
Vaccination is one of the most effective preventive measures to combat influenza. Prospectively, cell culture-based influenza vaccines play an important role for robust vaccine production in both normal settings and urgent situations, such as during the 2009 pandemic. African green monkey Vero cells are recommended by the World Health Organization as a safe substrate for influenza vaccine production for human use. However, the growth of influenza vaccine seed viruses is occasionally suboptimal in Vero cells, which places limitations on their usefulness for enhanced vaccine production. Here, we present a strategy for the development of vaccine seed viruses with enhanced growth in Vero cells by changing an amino acid residue in the stem region of the HA2 subunit of the hemagglutinin (HA) molecule. This mutation optimized the pH for HA-mediated membrane fusion in Vero cells and enhanced virus growth 100 to 1,000 times in the cell line, providing a promising strategy for cell culture-based influenza vaccines.
Although several antivirals against influenza viruses, including neuraminidase (NA) inhibitors, have been developed and used worldwide, vaccination is still considered one of the most effective preventive measures to combat influenza (12, 23). Currently, most conventional influenza vaccines are produced from viruses grown in embryonated chicken eggs. However, the limited capacity of the egg-dependent vaccine supply could be problematic in terms of securing enough doses when facing a pandemic situation, such as occurred in 2009, or in the event of a pandemic originating from a highly pathogenic avian virus, such as an H5N1 virus. In these situations, cell culture-based systems could play an important role for robust vaccine production (4).
Presently, cell culture-based inactivated influenza vaccines are in clinical trials or have been approved for use in some countries (1, 7, 8, 13, 19). This approach has considerable advantages over egg-based vaccines because (i) it can lead to more rapid and larger-scale vaccine production (10); (ii) it may avoid the potential for selecting variants adapted for chicken eggs, which alters virus antigenicity (18); (iii) selection of high-yield vaccine seed viruses is needed for egg-based production; and (iv) it does not contain allergic components of eggs (16). Due to these advantages, the World Health Organization (WHO) has recommended the establishment of mammalian cell culture-based vaccines (41).
Several cell lines are currently approved for cell culture-based influenza vaccine production. One of them, the African green monkey Vero cell line, has a good track record for the production of other viral vaccines for human use (e.g., polio and rabies vaccines) (26). In their long history, Vero cells have proven safe for vaccine production, so the WHO now recommends this cell line as an alternative substrate for influenza vaccine production (2). However, since seed viruses for seasonal inactivated vaccines occasionally grow suboptimally in Vero cells, seed viruses that grow well in Vero cells must be carefully selected for robust vaccine production (37). Here, we present a strategy for the development of vaccine seed viruses with enhanced growth in Vero cells by changing an amino acid residue in the hemagglutinin (HA) stem region. This approach could help overcome shortages in the influenza vaccine supply in emergency pandemic situations.
African green monkey Vero WCB cells, approved for use in human vaccine production (38), were maintained in Dulbecco's modified Eagle's minimal essential medium (DMEM) with 10% fetal calf serum and antibiotics. Madin-Darby canine kidney (MDCK) cells were grown in Eagle's minimal essential medium (MEM) with 5% newborn calf serum and antibiotics. The cells were maintained at 37°C in 5% CO2.
The A/Puerto Rico/8/34 [PR8(UW)] strain (27, 31) was generated by using reverse genetics (29) and propagated in 10-day-old embryonated chicken eggs for 2 days at 37°C, after which the allantoic fluids containing viruses were harvested and stored at −80°C. PR8 virus was inoculated into Vero cells in bovine serum albumin (BSA) (0.3%)-containing MEM with tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-trypsin (1 μg/ml). Three to 4 days after infection, virus-containing supernatants were collected and inoculated into fresh Vero cells at 1:100 or 1:1,000 dilution. After 11 passages, virus-containing supernatant was collected and stored at −80°C. Stock virus titers were determined by using a plaque assay in MDCK cells.
Viral RNAs were extracted from supernatants by using a commercial kit (QiaAmp viral RNA isolation kit; Qiagen) and were converted to cDNAs by using reverse transcriptase (SuperScript III; Invitrogen) and primers based on the consensus sequences of the 3-prime ends of the RNA segments for the H1N1 viruses. The full-length cDNAs were then PCR amplified with PfuUltra DNA polymerase (Stratagene) and PR8-specific primer pairs for each segment. The amplified cDNAs were cloned by using the Zero Blunt TOPO PCR Cloning Kit (Invitrogen). For each segment, four clones were sequenced by using PR8-specific primers. Primer sequences are available upon request.
We used our previously produced series of PolI constructs, derived from PR8(UW), for reverse genetics (15, 27) and PolI plasmids containing the HA and NA genes derived from A/Kawasaki/173/2001 (H1N1; Kawasaki173; GenBank accession numbers AB671296 and AB671297 for HA and NA, respectively), A/Kawasaki/UTK-4/09 (H1N1; UTK-4; GenBank accession numbers AB671291 and AB671292 for HA and NA, respectively), A/California/04/09 (H1N1; CA04; GenBank accession numbers FJ966082.1 and FJ966084.1 for HA and NA, respectively), and A/Yokohama/2013/03 (H3N2; Yok2013; GenBank accession numbers AB671293 and AB671294 for HA and NA, respectively) (20, 30). To generate HA, NA, and PB2 mutants, PolI plasmids expressing the HA, NA, and PB2 genes of PR8, Kawasaki173, UTK-4, CA04, or Yok2013 were used as templates for site-directed mutagenesis by the inverse PCR method. PR8 mutants and PR8 backbone 6:2 reassortants containing the six internal segments of PR8 and the HA and NA segments of seasonal or pandemic viruses were generated by using reverse genetics (29).
Virus was inoculated into Vero or MDCK cell monolayers at a multiplicity of infection (MOI) of 0.01 PFU/cell with MEM containing BSA and 1.0 μg/ml TPCK-trypsin and incubated at 37°C (for PR8, PR8/Kawasaki173 6:2 reassortant, PR8/UTK-4 6:2 reassortant, PR8/CA04 6:2 reassortant, and PR8/Yok2013 6:2 reassortant viruses) or at 33°C (for PR8/CA04 6:2 reassortant viruses). Viruses in the culture supernatants were collected at a given number of hours postinfection (p.i.) and then titrated by use of an MDCK plaque assay to determine the virus titers.
The cell fusion assay was performed as previously described (33) with some modifications. Briefly, Vero cells were transfected with PR8 HA or mutant HA (N117DHA2) expression plasmids (pCAGGS-PR8HA or pCAGGS-PR8HA2N117D, respectively), as well as with a green fluorescent protein (GFP) expression plasmid (pCAGGS-GFP) to conveniently visualize fused cells under a fluorescence microscope. After transfection, the cells were incubated at 37°C for 24 h. The cells were then washed several times with Mg2+- and Ca2+-containing phosphate-buffered saline (PBS+) and treated with 5 μg/ml of TPCK-trypsin for 5 min at 37°C. The trypsin was then inactivated by washing with PBS+ containing FCS. To initiate cell fusion, the cells were treated with acidic PBS (adjusted with citric acid) for 1 min and then incubated in FCS-containing medium at 37°C for 30 min. Fused cells were observed under a fluorescence microscope (Biozero; Keyence).
Comparisons of endosomal pHs between Vero and MDCK cells were performed as previously described (25, 35), with some modifications. Briefly, Vero and MDCK cells were incubated with Alexa Fluor 647 (30 μg/ml; Invitrogen)- and Oregon green 488 (250 μg/ml; Invitrogen)-conjugated dextran and incubated for 15 min at 37°C. After incubation, the cells were immediately placed on ice and washed 5 times with ice-cold PBS+, and the intensities of the Alexa Fluor 647 and Oregon green 488 were measured by using confocal microscopy (LSM 510; Carl Zeiss) in five microscopic fields for each sample. The Oregon green 488/Alexa Fluor 647 intensity ratio was then calculated.
All comparisons of the infectivity titers of each virus and the intensity ratio for Oregon green 488/Alexa Fluor 647 relied on Student's t test with two-tailed analysis to determine significant differences.
To obtain a virus that grows to a high titer in Vero cells, we performed serial passages of the PR8 virus in the cell line. Initially, wild-type (WT) PR8 virus-infected Vero cells showed an ambiguous cytopathic effect. After eight passages, however, we observed a clear cytopathic effect in Vero cells. After the 11th passage, we collected the virus (referred to as the PR8-Vero virus). We then compared wild-type and PR8-Vero virus titers in the supernatants of infected Vero cells. Wild-type virus grew to 2.0 × 104 PFU/ml, whereas PR8-Vero virus grew to 1.9 × 109 PFU/ml. These data suggest that PR8-Vero virus possesses mutations that enhance its replication in Vero cells.
To identify the mutation(s) responsible for PR8 adaptation to Vero cells, the virus genome was sequenced. The cDNAs of the PR8-Vero virus were cloned into plasmids, and the sequences of four clones for each segment were read. As shown in Table 1, PR8-Vero virus contained mutations in the HA (4/4 clones), NA (4/4 clones), and PB2 (2/4 clones) genes, and all of these mutations caused amino acid changes. No mutations were identified in any of the other gene segments. Next, we introduced the mutation(s) into wild-type PR8 by using reverse genetics and examined the growth kinetics of the mutants (Fig. 1). The D740N PB2 PR8 mutant virus grew similarly to wild-type PR8, whereas the N255Y NA mutation augmented viral growth to 8.5 × 105 PFU/ml. Interestingly, the N117D HA2 mutant virus grew to the highest titer (6.3 × 108 PFU/ml). Both the N255Y NA and N117D HA2 mutations caused virus to grow to a level comparable to that of PR8-Vero virus. These data demonstrate that the N117D HA2 mutation is primarily responsible for Vero cell adaptation.
Based on a National Center for Biotechnology Information (NCBI) database search, more than 99% of viruses possess asparagine at position 117 (H1 subtype) or 116 (H3 subtype) of HA2. Asparagine at this position is located in the α-helix of the stalk region of the HA molecule, which is close to the viral membrane surface (Fig. 2A). We therefore sought to determine whether the HA2 N117D (or N116D) amino acid substitution could enhance the growth of other viruses in Vero cells. To this end, we introduced the HA2 N117D or N116D mutation into the HAs of the H1 seasonal strains A/Kawasaki/173/01 (H1N1) and A/Kawasaki/UTK-4/09 (H1N1), the pandemic strain A/California/4/2009 (H1N1), and the H3 seasonal strain A/Yokohama/2013/2003 (H3N2). We then generated PR8/H1N1 or PR8/H3N2 6:2 reassortants with these mutated HAs by using reverse genetics. We also produced 6:2 reassortants with wild-type HA from each virus for comparison. Each virus possessing mutated HA grew to a 100 to 1,000 times higher titer than its wild-type HA-bearing counterpart in Vero cells (Fig. 2C). In contrast, we did not observe any marked differences in virus titers between wild-type and mutant reassortants in MDCK cells (data not shown). These data suggest that the HA2 N117(116)D mutation would universally enhance the growth of a vaccine seed virus in Vero cells.
Since the HA2 region mediates virus membrane fusion and the HA2 N117 residue is located close to the fusion peptide (Fig. 2B), we examined whether the mutation affects the optimal pH for viral membrane fusion. We constructed plasmids expressing PR8 wild-type HA or HA2 N117D mutant HA and transfected them into Vero cells. A GFP expression plasmid was cotransfected into the cells for a fusion assay. At 24 h posttransfection, we treated the cells with several low-pH buffers and observed fused cells (Fig. 3). At neutral pH (pH 7.4), we did not observe any fused cells with either wild-type HA or N117D mutant HA-transfected Vero cells. At pH 5.0 and 5.2, numerous fused cells were observed in both wild-type HA and mutant HA-transfected cells. Interestingly, at pH 5.4, few fused cells were observed in wild-type HA-transfected cells, whereas many fused cells were observed in mutant N117D HA-transfected cells. At pH 5.6, both wild-type HA and mutant HA-transfected cells produced limited cell-cell fusion. In addition to the change in optimal pH for membrane fusion in mutant HA-expressed fused cells, the number of cells involved in cell-cell fusion was greater in these cells than in wild-type HA-transfected cells, even at pH 5.0 and 5.2. We also performed cell fusion assays in MDCK cells with wild-type and mutant HAs, and the results were very similar to those obtained in Vero cells (data not shown). These data indicate that the HA2 N117D mutation both augments membrane fusion and widens the optimal pH range for virus membrane fusion in Vero cells.
Although wild-type PR8 virus growth is suboptimal in Vero cells, wild-type PR8 grew to a level comparable to that of HA2 N117D mutants in MDCK cells (Fig. 4A). The optimal pH for viral membrane fusion was higher for the HA2 N117D mutant than for the wild type (Fig. 3), allowing us to hypothesize that the endosomal pH of Vero cells is higher than that of MDCK cells. To test this hypothesis, we compared the endosomal pH between Vero and MDCK cells by introducing dextran-conjugated fluorescent dye as a marker and measuring the intracellular intensity (Fig. 4A). This assay is based on the principle that the fluorescence intensity of Oregon green 488 is sensitive to low pH whereas the intensity of Alexa Fluor 647 is not pH sensitive (25). Thus, pHs can be compared by measuring the intensity of each fluorescent dye and calculating the intensity ratio between Alexa Fluor 647 and Oregon green 488. Vero and MDCK cells were incubated with Alexa Fluor 647- and Oregon green 488-conjugated dextran for 15 min at 37°C. After incubation, the cells were washed and the intensities of Alexa Fluor 647 and Oregon green 488 were measured by using confocal microscopy. The Oregon green 488/Alexa Fluor 647 intensity ratio was appreciably higher in Vero cells than in MDCK cells (Fig. 4B), suggesting that the early endosomal pH value is higher in Vero cells than in MDCK cells.
Here, we described an approach to enhance the growth of influenza vaccine seed viruses in Vero cells, which are approved for use in human vaccine production. Influenza vaccine seed viruses that provide robust growth in cell culture are needed to ensure an adequate supply of influenza vaccines as either a supplement to or an alternative method for egg-based vaccine production. Our approach involved introducing a single amino acid mutation (N117D) into the HA2 subunit of HA, which was found in a Vero cell-adapted PR8 virus. The seasonal influenza vaccine seed-like viruses (6:2 reassortants with a PR8 backbone) tested in this study grew poorly in Vero cells. However, the introduction of this HA2 single mutation into these viruses produced mutants that grew to 100 to 1,000 times higher titers in Vero cells than wild-type viruses. This strategy for virus growth enhancement based on the HA2 mutation could thus be feasible for the production of growth-enhanced seasonal or pandemic vaccine seed viruses.
The amino acid at position 117 of HA2 is located on the stem region of HA. One concern is the possibility that changing the amino acid residue at this position may affect the antigenicity of inactivated vaccines, although the major antigenic sites of H1 HA (Ca, Cb, Sa, and Sb), against which most neutralization antibodies elicited by inactivated vaccines are raised, are located on the globular head region that surrounds the receptor binding site (3). Recent studies revealed that antibodies against the HA stem region confer universal protection from influenza virus infection (5, 9, 17, 39, 40). However, such neutralizing antibodies are rarely induced by conventional vaccinations (28). These findings suggest that the HA2 N117D mutation would not affect the antigenicity or efficacy of the vaccines.
Here, we suggest that the growth enhancement was most likely due to broadening of the optimal pH range for virus membrane fusion mediated by HA2. The mutation site is located in the HA stalk region, close to the fusion peptide in the HA trimer (Fig. 2B). To accomplish virus membrane fusion, cleaved HA needs to be exposed to low pH (14). Previous reports demonstrate that the substitution of the neutral amino acid residue(s) in the HA stalk region for charged amino acids affects the optimal pH for virus membrane fusion (24, 32, 33, 43). Since the N117D substitution introduces a negative charge, it may change the electrostatic balance between the residue at position 117 and the fusion peptide, possibly resulting in fewer proton-dependent conformational changes in the HA molecule.
We revealed that the endosomal pH was higher in Vero cells than in MDCK cells 15 min after dextran intake. Based on this observation, we assume that this difference in endosomal pH affects virus growth in these cell lines. Indeed, cell-type-dependent endosomal pH kinetics are important for influenza virus infection (21, 22). In MDCK cells, influenza virus reaches the late endosome (pH 5.0) 10 min after endocytosis (21). On the other hand, in HeLa cells, which is a nonpermissive cell line for influenza virus infection (6), it takes 40 min for influenza virus to colocalize with a late endosome marker following endocytosis (36). However, highly pathogenic H5N1 virus, which requires a higher pH for optimal membrane fusion (pH 5.9) (32), grows in MDCK, Vero, and HeLa cells with comparable titers (44). Moreover, vesicular stomatitis virus (VSV) infectivity is much lower in MDCK cells than in Vero cells (11), possibly due to the higher optimal pH (pH 6.0) for membrane fusion of the VSV G protein (34). These facts imply that Vero cells may have a higher pH in the early endosome than MDCK cells. A precise assessment of endosomal pH changes is needed to better understand the mechanism of enhanced virus growth in Vero cells.
We used PR8 virus as a reassortant backbone virus in this study because it is attenuated in humans and is approved by the WHO for use as a genetic backbone for vaccine seed viruses (42). However, introducing the HA2 mutation identified here into any wild-type virus, including seasonal influenza viruses, could also enhance virus replication in Vero cells, providing alternative vaccine seed viruses for the production of inactivated influenza vaccines.
In conclusion, we propose a mutant virus possessing aspartic acid at HA2 position 117 as a seed virus for Vero cell-based influenza vaccine production. A virus with this single amino acid mutation can be produced easily by using reverse genetics. A cell culture-based vaccine strategy with this seed virus would allow the production of more doses of inactivated influenza vaccines in a timely, cost-effective manner, not only for seasonal, but also for pandemic vaccines.
We thank Y. Kino (Chemo-Sero-Therapeutic Research Institute, Japan) for Vero WCB cells and S. Watson for scientific editing.
This work was supported in part by Grants-in-Aid for Specially Promoted Research and for Scientific Research (B); by a contract research fund from the Ministry of Education, Culture, Sports, Science and Technology of Japan; by the Program of Founding Research Centers for Emerging and Reemerging Infectious Diseases; by Grants-in-Aid for Specially Promoted Research and for Scientific Research; by ERATO (Japan Science and Technology Agency); and by National Institute of Allergy and Infectious Diseases Public Health Service research grants.
Published ahead of print 16 November 2011