In 2003 there was an outbreak of highly pathogenic avian influenza (H7N7) in the Netherlands. Because of the zoonotic and pandemic potential of the viruses that caused the outbreak, the availability of an effective vaccine was highly desirable. The HA of a virus isolated from a mallard in 2000 had a high sequence identity to the HA gene of the viruses that caused the outbreak. This offered the opportunity to use this low-pathogenicity virus obtained through routine influenza surveillance in wild birds as the basis of a vaccine directed against highly pathogenic avian influenza H7N7 virus. Using reverse genetics, we generated a reassortant seed strain that was used for vaccine preparation. We compared a classical, nonadjuvanted influenza vaccine preparation containing HA and NA with an ISCOM-adjuvanted preparation.
Our experiments clearly showed that, although an antibody response was mounted, the classical influenza vaccine did not protect against lethal infection with influenza virus A/Netherlands/219/03, not even when two doses were administered. One dose of the ISCOM-adjuvanted vaccine preparation did not protect either. Two doses of the ISCOM-adjuvanted vaccine were required for clinical protection against influenza virus A/Netherlands/219/03. Although two doses of the ISCOM-adjuvanted preparation did not lead to sterile immunity against highly pathogenic avian influenza H7N7, virus titers in the lungs of vaccinated mice were reduced considerably. We speculate that this ~1,000-fold reduction of pulmonary virus titers was responsible for the survival of the animals receiving two doses of the ISCOM vaccine. In the event of a pandemic outbreak, a vaccine offering clinical protection could be sufficient to reduce the clinical impact of infection with the pandemic virus.
Our experiments also showed that it was possible to use a heterologous low-pathogenicity strain as the donor of the HA protein for vaccines directed against highly pathogenic avian influenza virus strains. It should be noted that the HA amino acid sequence of the low-pathogenicity mallard virus and the highly pathogenic avian influenza viruses differed very little; only the HA cleavage site and single-amino-acid substitutions in the signal peptide, HA1, and HA2 were different. Low-pathogenicity strains with such high amino acid sequence identity and such small antigenic differences from highly pathogenic avian influenza strains may not always be available.
An important question that remains is why infected animals were not protected from infection or death despite high HI antibody titers upon repeated vaccination with adjuvanted or nonadjuvanted vaccines, respectively. We performed virus neutralization assays with the sera from vaccinated mice and influenza viruses A/Mallard/Netherlands/12/00 (H7N3) and A/Netherlands/219/03 (H7N7). These experiments revealed that despite the high titers of HI antibodies in these sera (HI titers ranging up to 7,680), virus-neutralizing antibodies were not detectable (titers <40). In contrast, rabbits hyperimmunized with HA and NA of A/Seal/Massachusetts/1/80 (H7N7), which had HI antibody titers of 1,280 and 2,560 against A/Mallard/Netherlands/12/00 (H7N3) and A/Netherlands/219/03 (H7N7), respectively (Table ), had virus-neutralizing antibody titers against these viruses of 1,280 and 5,120, respectively.
Since the Dutch H7 viruses are thus not resistant to neutralization, the lack of virus neutralization with the mouse sera must be related to the poor induction of neutralizing antibodies upon vaccination. It is of interest to note that the postinfection ferret sera raised in this study also lacked detectable neutralizing antibody titers despite the presence of HI antibodies (Table and data not shown). Furthermore, the H7N7-infected humans in the Netherlands who developed HI antibodies during the outbreak in 2003 did not develop detectable virus-neutralizing antibodies (unpublished). Thus, it may be that the Dutch H7 viruses are poor inducers of neutralizing antibodies. Hence it is possible that cell-mediated immunity or antibodies with poor virus-neutralizing capacity were responsible for the clinical protection of mice vaccinated with the ISCOM-based vaccine.
Experiments like the one described here have been conducted with highly pathogenic strains of the H5N1 subtype, although it is difficult to compare these studies to our own because of the use of different virus subtypes, animal models, adjuvants, vaccine types, timing, and dosage. Using the low-pathogenicity A/Duck/Singapore/Q/F119-3/97 (H5N3) (12
), the low-pathogenicity A/Duck/Hokkaido/67/96 (H5N4), or an H5N1 reassortant thereof (24
), inactivated virus preparations were tested in mice for their possible use in a pandemic situation. In these studies, vaccinated mice survived challenge with the highly pathogenic avian influenza H5 strains A/Hong Kong/483/97 and A/Hong Kong/156/97, even when no adjuvant was used in the vaccine preparations. When alum was used as an adjuvant, mice were even protected from infection (12
However promising these studies may seem compared to ours, our results are more in agreement with vaccination studies in a chicken (20
) and monkey (19
) model and with experimental vaccination in humans (16
). When surface antigen of the same A/Duck/Singapore/Q/F119-3/97 strain used in the mouse studies was used as a vaccine in a phase I randomized trial in human volunteers, two doses of a preparation with the adjuvant MF59 were required to reach antibody levels that are considered protective (16
Although ISCOMs are not registered for use in humans, the immune response to an ISCOM-adjuvanted influenza vaccine has been tested in humans. When ISCOMs were tested in humans in a randomized, double-blind study, antibody responses to the ISCOM-adjuvanted preparation were improved compared to antibody responses to the conventional influenza vaccine preparation (21
). Although in this study individuals were vaccinated with an ISCOM preparation containing influenza A virus antigens against which there was preexisting immunity, the results in a range of animal models suggest that the ISCOM adjuvant could also work very well to induce immune responses in naive individuals. In addition, ISCOMs are used as an adjuvant in registered equine influenza vaccines (1
In case of a pandemic, a two-dose vaccine will be impractical. Therefore, there is an urgent need for new adjuvants or improved vaccine delivery approaches, yielding better immune responses and improved protection against lethal infection, preferably after administration of a single dose.
Although vaccination experiments using highly pathogenic H5 viruses with a deleted basic cleavage site have been successful in animals (9
), these vaccine strains can only be produced as soon as a pandemic threat arises. Furthermore, removal of the basic cleavage site requires an extra modification step and thus more time in the production of a reassortant seed virus. Since we show here that a heterologous vaccine could work well, prototypic envelope proteins of potentially pandemic viruses could be cloned in advance to speed up seed virus production further. Alternatively, vaccine seed viruses could be generated by classical reassortment, eliminating the need for reverse genetics technology. This would not only bypass the patent-related costs of influenza vaccines, but would also enable vaccine manufacturers to produce vaccines without the need for facilities equipped for working with genetically modified organisms.
An important conclusion that can be drawn from the experiments described above is that prototypic low-pathogenicity strains obtained through routine surveillance of wild birds could be used effectively to generate vaccines directed against highly pathogenic avian influenza viruses long before outbreaks in poultry or pandemic threats emerge. A repository of seed virus containing the HAs of viruses with known zoonotic or pandemic potential (H1, H2, H3, H5, H7, and H9) could be prepared in advance. High-growth strains could be generated in order to respond quickly to outbreaks and pandemic threats. Based on the antigenic properties of the virus strains, the most suitable vaccine candidate could then be selected from the repository using new methods for antigenic characterization (22