The potential of influenza A viruses to generate new human pathogenic strains from a vast natural reservoir in aquatic birds means that eradication of influenza is not feasible. Correspondingly, disease control requires the monitoring of virus reservoirs and the development of improved antiviral therapies and vaccines. The most widely used human influenza vaccines are those made from subunits of inactivated viruses propagated in embryonated chickens' eggs. These influenza vaccines are essentially genetic modifications of those generated in the mid-1970s, when production was improved by reassorting the circulating strain with A/Puerto Rico/8/34 (PR8), an H1N1 virus adapted for high growth in eggs (
49). Since the mid-1970s, the influenza vaccine has been a so-called 6 + 2 reassortant containing the surface hemagglutinin (HA) and neuraminidase (NA) genes from the vaccine target strain and the remaining genes from PR8. Such reassortants are made by coinfecting eggs with both viruses and screening progeny for the desired 6 + 2 configuration.
Although routinely used to prepare human influenza virus vaccines and diagnostic reagents, embryonated chickens' eggs have potentially serious limitations as a host system, not least of which is that the cultivation of influenza viruses in eggs can lead to the selection of variants characterized by antigenic and structural changes in HA (
19,
36,
40). Other problems include the lack of reliable year-round supplies of high-quality eggs, the possible presence of adventitious pathogens, and the low susceptibility of summer eggs to infection with influenza virus (
27). The current cycle of interpandemic influenza vaccine production requires detailed planning up to 6 months before vaccine manufacture to ensure an adequate supply of embryonated eggs (
10). Because a pandemic event cannot be predicted and a 6-month delay in vaccine production is unacceptable, there is an urgent need to develop improved cell culture systems for vaccine production. Such an improvement is a priority for the World Health Organization (WHO). As part of their Global Agenda on Influenza, the WHO has urged the development of novel vaccines and production strategies or technologies (
44).
An additional need for improved cell-based protocols for the production of influenza vaccines has emerged with the development of reverse genetics, which enables the production of influenza vaccines from cloned viral cDNA (
7,
16,
30). The ability to custom make influenza viruses by using this technology may dramatically improve the speed with which we can respond to pandemic emergencies. Advantages include the ability to attenuate pathogenic strains (
45) and the elimination of the need to screen reassortant viruses for the 6 + 2 configuration, a procedure that can be time-consuming. The potential of reverse genetics to generate vaccine candidates has already been described (
15,
39). The main drawback of this methodology is the need to use vaccine-approved cell lines; those commonly used to obtain influenza viruses from cDNA are 293T and Madin-Darby canine kidney (MDCK) cells. The 293T cell line is a transformed cell line and is therefore unlikely to be used for human vaccine production, and there are lingering concerns over the tumorigenic potential of MDCK cells (
12). In addition, the use of host-specific RNA polymerase I promoters in reverse genetics systems limits usable cell lines to those of primate origin.
As an alternative cell-based system, the well-characterized African green monkey kidney (Vero) cell line has potential. Vero cells are suitable for the production of several human virus vaccines, including those against poliomyelitis and rabies (
26). Despite earlier findings that influenza viruses do not replicate well in Vero cells (
24,
28), the repeated addition of trypsin to the culture medium improves virus yields (
20), and preliminary studies with a limited number of strains have indicated that Vero cells support the primary isolation and replication of influenza A viruses (
11). Influenza virus vaccines derived from Vero cells have been produced and evaluated for immunogenicity, and their production has been scaled up to commercial levels (
2,
23). These Vero-derived vaccines elicit humoral responses comparable to those elicited by the egg-grown vaccines but are more potent stimulators of the cellular response (
2). These Vero-based vaccines have, however, consisted of unreassorted viruses rather than the accepted 6 + 2 PR8 reassortants.
The aim of the present study was to develop a reproducible, high-yielding Vero-based reverse genetics system to produce influenza vaccine. Govorkova and colleagues (
11) previously derived a high-yielding influenza virus by passaging an H1N1 reassortant isolate, A/England/1/53 [HG], several times in Vero cells. This reassortant virus was reported to contain the surface glycoprotein genes from A/England/1/53 and the remaining genes from PR8, the standard vaccine master strain. Our aims were to identify the molecular changes responsible for this high-yielding phenotype and to produce an altered PR8 vaccine master strain adapted for optimal efficiency of viral rescue from cDNA in the eight-plasmid reverse genetics system and for growth in Vero cells.
DISCUSSION
Recent discoveries in influenza pathogenicity and reverse genetics have the potential to revolutionize the way we prepare and manufacture pandemic and interpandemic influenza vaccines. Much of this technology has, however, been confined to experimental protocols; the realization of this potential awaits refinements of the methods. The results we report here go a long way toward helping to fulfill the potential of these technologies. By incorporating the NS gene segment of Eng53/v-a into the standard PR8 vaccine master strain, we have shown a reproducible improvement in vaccine virus rescue and growth in Vero cells with no drop in virus titers in eggs. Although we were able to rescue these viruses with the standard PR8 vaccine strain, the improved efficiency of rescue with our system may be crucial with other HA and NA combinations that are poorly infective in Vero cells.
With each of the HA and NA subtypes we tested, the PR8/Eng-NS viruses reached peak titers significantly faster than did the corresponding PR8 virus (Fig.
3). The peak titers of the PR8/Eng-NS variants containing the surface glycoproteins of the two contemporary vaccine strains (A/New Caledonia/20/99 and A/Panama/2007/99) and of A/teal/Hong Kong/W312/97, a virus implicated in the genesis of the 1997 H5N1 human viruses (
18), were also significantly higher than those of the PR8 viruses (Table
1). In contrast, there was no difference in the peak titers of the PR8/Eng-NS and PR8 viruses carrying the surface glycoproteins of A/quail/Hong Kong/G1/97 or PR8 itself. However, by the time these peak titers were reached with the PR8 variants, the cells infected with the PR8/Eng-NS variants had been completely destroyed by cytopathic effects, and a manufacturing process in which cells and fresh media can be continually added or replenished is likely to produce higher yields for those viruses on the PR8/Eng-NS backbone.
Our data also provide support for the continued use of 6 + 2 high-growth reassortants for vaccine production in Vero cells. For example, the PR8/Eng-NS variant of the H6N1 virus had growth characteristics significantly superior to those of the wild-type H6N1 virus; half times to peak titers were three times longer in the wild-type virus. The use of 6 + 2 reassortants also reduces the risks of growing adventitious agents with influenza viruses isolated directly from clinical samples.
One of the benefits of cell-based production of vaccines is that it appears to allow the amino acid sequences of influenza virus HA molecules to remain unchanged; these molecules are altered during the adaptation of viruses to eggs (
19,
36,
40). It should, however, be noted that changes in the biologic activity of a virus can occur in the absence of sequence changes. It has been shown that differences in the abilities of the same human H3N2 strain to agglutinate human and chicken red blood cells were associated with the type of cells from which the strain was isolated (MDCK or Vero) (
13). Romanova and colleagues have recently shown that the inability of the Vero-grown variant to grow in eggs and agglutinate chicken red blood cells is related to the higher proportion of oligosaccharides of high mannose type in this variant than in the corresponding MDCK-derived isolates. This observation was made in the absence of any amino acid differences between the variants (
37). We found that the HA genes of the PR8/Eng-NS-derived viruses were the same before and after rescue and propagation from Vero cells. This stability is an advantage in a vaccine that derives much of its protective qualities from the production of neutralizing antibody directed against the HA molecule. However, as discussed by Kemble and Greenberg (
21), such benefits will be lost unless candidate vaccine viruses are first isolated in approved cell lines rather than in the widely used MDCK cells or eggs. The retention of these benefits of cell-based vaccine production requires a commitment from many agencies, including the WHO influenza network.
Host range in influenza viruses is a polygenic trait for which many influenza virus proteins have been implicated (
14,
41,
43,
46). Likewise, many genes, including the NS genes, have been implicated in the attenuation of influenza viruses in different hosts (
4,
25,
43). However, in our system, the transfer of the NS gene segment from Eng53/v-a to PR8 was alone sufficient to confer the high-growth phenotype (Fig.
5). Although many factors have been reported to determine the efficiency of growth of influenza viruses, the NS1 protein (one of the two proteins expressed by the NS gene segment) is thought to play a significant role in translation and replication. For example, NS1 and its interaction with host proteins have been reported to play central roles in inhibiting the nuclear export and splicing of host mRNA (
8,
29,
34,
35,
47,
48). Furthermore, NS1 protein plays an important role in regulating interferon activity (
42). It is unlikely, however, that the interferon pathway is involved in viral growth in Vero cells, because Vero cells do not produce interferon (
6). Another study failed to identify major changes in the shutoff of host protein synthesis or in viral protein expression in Vero cells after infection with NS1 deletion mutant viruses (
38). Garcia-Sastre et al. showed, however, that a mutant virus that did not express NS1 protein (delNS1) grew approximately 10 times slower on Vero cells than did the wild-type strain (
9). The second protein encoded by the NS gene segment is the NS2 protein, or nuclear export protein (NEP). Although little is known about the function of NEP, this polypeptide interacts with nucleoprotein and contributes to the nuclear export of the viral ribonucleoproteins (
32). Additionally, amino acid residues of the NEP have been shown to be crucial for viral replication (
3). Studies are ongoing to identify the NS gene products responsible for increased viral replication in Vero cells and the mechanisms involved.
The applicability of reverse genetics to the production of influenza vaccines depends on the use of a suitable cell culture system. Technical constraints and the limited number of cells licensed for vaccine production severely limit the options available. It is therefore likely that timely improvements in reverse genetics-derived vaccines will arise from optimization of current protocols rather than identification of alternative systems. The data presented in this paper show that use of the Vero cell system with an improved master strain virus is a viable option for the rapid manufacture of influenza vaccines in pandemic emergencies and for the production of vaccines during annual epidemics. Although the technologies are now available, use of the vaccines created by them requires approval by regulatory agencies, which must be prepared and equipped to rapidly give such approval.