On the Role of Vaccine Dose and Antigenic Distance in the Transmission Dynamics of Highly Pathogenic Avian Influenza (HPAI) H5N1 Virus and its Selected Mutants in Vaccinated Animals

Influenza virus infections can cause high morbidity and mortality rates among animals and humans, and result in staggering direct and indirect financial losses amounting to billions of US dollars. Ever since it emerged in 1996 in Guangdong province, People’s Republic of China, one particular highly pathogenic avian influenza (HPAI) H5N1 virus has spread globally, and is responsible for massive losses of poultry, as well as human infections. For these reasons, HPAI H5N1 is considered as one of the viruses possible to cause a future influenza pandemic. One of the main reasons why influenza is a recurring problem is its ability to constantly evolve through the selection of mutants that are able to avoid immunity (be it natural or acquired). Due to the accumulation of mutations during genome replication, diverse/variant influenza genome sequences co-exist in a virus pool (quasispecies). These sequences can contain mutations that are able to confer selective advantages to the influenza virus given the opportunity. As a consequence, whenever a situation arises that places the virus under any type of pressure that the dominant virus sequence cannot cope with (i.e. immune pressure, selective receptor binding, etc.), the virus with the genome sequence that allows it to better adapt to that particular pressure becomes selected and takes over. Because of the influenza virus’s high rate of mutations, a global surveillance network is in place to monitor changes in circulating strains among humans that would warrant an update of the vaccines used. For human influenza strains, vaccines are updated frequently (every one or two years) and a similar situation holds true for racehorse vaccination. For avian influenza vaccination, however, the situation is different. In most countries, vaccination against avian influenza is not used, and in the countries where vaccines are used (either as routine or emergency measures), they are not updated as frequently as human vaccines are. In addition, in many instances vaccination against avian influenza viruses has met with some spectacular failures, since it failed to produce a level of immunity that would protect against circulating field strains. These vaccination failures have often been attributed to the fact that without constant vaccine updating (as is done for human influenza), the vaccines used are not able to keep up with continuously evolving antigenic variants selected in the field, and thus to protect poultry against them. In addition, since it is known that immune pressure resulting from vaccination can be a driving force in the evolution of influenza viruses and the selection of immune-escape mutants, there is a school of thought that posits that vaccination against avian influenza is not only a very expensive affair (especially if vaccines need to be frequently updated), but can also lead to selection of mutants that are able to avoid vaccination-induced immunity. The research reported in this thesis started with addressing the gaps in the knowledge regarding the role of vaccination-induced immunity in the selection of immune-escape mutants of HPAI H5N1, and if there is a way for vaccines to still be able to protect against antigenically-distant variants of the vaccine seed strain, without the need for frequent vaccine updates. Our first step in studying influenza virus evolution and selection of immune-escape mutants was to investigate how antigenic pressure may drive the selection of such mutants, and what the effect of the selected mutations on the pathogenicity and transmissibility of the mutants may be. Although there exist a variety of methods to select for influenza virus mutations (i.e. monoclonal antibodies, site-directed mutagenesis, reverse genetics, etc.), none of them is representative of selection as it happens in a vaccinated animal. In Chapter 2, we discuss in detail a laboratory-based system we have developed, in which immune-escape mutants are selected using homologous polyclonal chicken sera, similar to how they are selected in the field due to vaccination- induced immune pressure. We find that selection takes place early on, and additional mutations are selected when immune pressure is increased. Antigenic distances between the selected mutants and their parent strains are also increased throughout the selection process, but not in a linear fashion. Our selection system proved to be robust and replicable, and to be representative of selection in the field, since the mutations we selected for are also found in naturally-selected field isolates, and the antigenic distances between our selected mutants and their parent strains are similar to antigenic distances between vaccine strains and field isolates. We continued our research by addressing the roles played by vaccine dose (and resulting immunity) and antigenic distance between vaccine and challenge strains, in the transmission of HPAI H5N1 viruses, by employing transmission experiments using vaccinated chickens (Chapter 3). To our surprise, we found that the effect of antigenic distances between vaccine and challenge strains on transmission is very small compared to the effect of vaccine dose. We then quantified, for the first time, the minimum level of immunity and minimum percentage of the vaccinated population exhibiting said immunity, in order for vaccines to be able to protect against transmission even of strains that are antigenically distant to the vaccine seed strain. Transmission of such strains in well-vaccinated populations would allow for a scenario where vaccination- induced immunity may drive the selection of immune-escape mutants. Our results show that in order for vaccines to prevent transmission of antigenically distant strains (such as the ones resulting from selection due to immune pressure), the threshold level of immunity against these strains should be ≥23 haemagglutination inhibition units (HIU), in at least 86.5% of the vaccinated population. This level of immunity can be estimated by knowing the antigenic distance between the vaccine and challenge (field) strain, and the HI titre against the vaccine strain, which would then allow the approximate level of immunity against the field strain to be deduced. For example, assuming the HI titre against a vaccine strain is 210 HIU, and the distance with the challenge (field) strain is 24 HIU, according to our results the vaccine should be able to protect against the challenge strain, because the difference in HI titres should be around 26 HIU (i.e. above 23 HIU). These results, taken together with our previous work on selection of mutants, where we showed that the antigenic distances between our mutants and their parent strains are representative of distances found in the field, point to the fact that it is unlikely that vaccination-induced immunity can lead to selection of mutants able to escape it, given that a threshold level of immunity in a minimum percentage of the vaccinated population is achieved. As a consequence, we believe that constant vaccine updating may not be necessary for avian influenza viruses, as long as a threshold level of immunity is maintained. This makes vaccination a more attractive control measure, both from a health perspective and a financial one, than just applying biosecurity measures. To examine the effect the mutations in the haemagglutinin protein of our selected mutants may have in their transmission among chickens vaccinated with the parent strain, we used reverse genetics techniques to insert the HA gene of our most antigenically distant mutant into the parent strain backbone (Chapter 4). We vaccinated animals with a sub-optimal dose of vaccine, and we concluded that the mutations we selected for did not allow the mutant to avoid even low levels of immunity, such as the ones resulting from a sub-optimal vaccine dose (which resembles a poor field vaccination scenario). At the same time, the HA mutations we selected for did not appear to have a negative effect either on the pathogenicity of the mutant, or its ability to transmit to unvaccinated animals, since both parameters were comparable to the parent strain. Finally, we studied the role inter-animal variation in immunity – as measured by HI titres – has in the accuracy of antigenic cartography calculations (Chapter 5). We found that using sera from more than one animal significantly increased the accuracy of antigenic distance calculations, since it takes into account individual differences in immune responses to vaccination, an inevitable phenomenon documented in both humans and animals. In addition, we increased the accuracy of antigenic maps by avoiding the use of dimension-reducing algorithms as is currently done. By not reducing the dimensionality of virus positioning in space, our maps retain the original geometry between strains or sera, leading to more accurate positioning (Chapters 2 and 5). We hope that improving the accuracy of antigenic cartography can lead to a more precise surveillance of influenza evolution and better informed decisions regarding the need to update vaccines. Taken collectively, our results can improve field vaccination outcomes, since they provide guidelines on how to increase vaccination efficiency in stopping transmission of even antigenically-distant strains. In addition, our method for selecting for immune- escape mutants can be a valuable addition to research on influenza virus evolution. Moreover, policy making decisions regarding vaccination against any type of influenza can also benefit from our improvement on antigenic cartography accuracy, saving unnecessary costs in vaccine updating, and reducing morbidity and mortality of both animals and humans.