Avian flu has a disturbing capacity: it can travel thousands of kilometers inside a migratory bird and appear in places where it had not been detected until then. But the real fear of scientists is not only in its spread among animals, but in the virus acquiring the ability to be easily transmitted between mammals and, finally, between people. To anticipate this possibility, A team from the University of Cambridge has turned to artificial intelligence to design a broad-spectrum vaccine which, in animal tests, has achieved something that current formulations had not achieved: protect against different variants of the avian flu virus.
The study, published in Cell Reports Medicinedescribes a preparation developed by Cambridge scientists together with the spin-off company DIOSynVax. In tests, the vaccine completely protected mice and ferrets exposed to several genetically different variants of the virusall of them related to human infections and considered potential threats. In contrast, the reference vaccines used by the World Health Organization (WHO) in the comparison only allowed approximately half of the animals exposed to those same viruses to survive.
The difference is in the way the antigen is designed, the component of a vaccine that teaches the immune system to recognize a pathogen. Traditional vaccines usually start from a specific variant that already circulates and try to generate defenses against it. The problem is that RNA viruses, like bird flu, change quickly. By the time a specific vaccine is developed, manufactured and distributed, the virus may have accumulated mutations that reduce its effectiveness.
The Cambridge team, led by Jonathan Heeney, has attempted to reverse that process. To do this, he gathered genetic sequences of avian flu viruses recorded in surveillance programs around the world, with samples from wild animals, livestock and infected people. He then developed an artificial intelligence tool capable of analyzing this enormous amount of information and locating common characteristics between different variants. From them he designed several “superantigens”, structures that bring together elements shared by a large family of viruses. The candidates were tested in cells to select the most promising one before taking it to animal trials.
“Rather than providing immunity against a single variant of bird flu, our new vaccine protects against a huge variety of them – explains Heeney -. It is designed to anticipate the evolution of the virus, rather than react to it. In this scenario, Artificial intelligence considerably speeds up the vaccine design process and then we can test the candidates in the lab to make sure they are robust and safe.”
The importance of this strategy is understood by observing the behavior of avian flu. Birds can carry the virus during their migrations and disperse it through their droppings. The danger increases if the virus acquires adaptations that facilitate its entry and multiplication in the respiratory tracts of mammals.
“Is when bird flu adapts to live in the respiratory tracts of mammals when we start to really worry – adds Heeney -. Not only because of the impact on animal health, but because of the threat to people. “If we can make better vaccines for animals, we can also better protect human health.”
The promise is considerable, but there is still a decisive step: verifying that the protection observed in mice and ferrets also works in humans. Until then, it is neither a vaccine available nor a guarantee against a future pandemic. It does, however, represent a change of focus: use artificial intelligence to design defenses that do not only pursue the virus that we already know, but also try to anticipate the one that could appear tomorrow.