AI-guided mutations help viruses infect bacteria up to 1 million times more effectively

Biochemists at the University of Wisconsin-Madison are using artificial intelligence to revolutionize phage therapy as a means to combat antibiotic-resistant bacterial infections. Professor Vatsan Raman and his team built an AI model using data from their laboratory, which can identify mutations in bacteriophages (phages) to enhance their ability to target and kill specific bacteria.

This breakthrough could potentially accelerate the development of new treatments for bacterial illnesses, as antibiotics have become less effective due to the quick evolution of bacteria. Phage therapy employs naturally occurring or engineered viruses, known as bacteriophages or phages, to infect and destroy specific bacteria. While phages have evolved to be effective against bacteria, they are not always lethal enough to decimate the entire bacterial population, which is crucial for the survival and reproduction of the phages.

Bacteria, on the other hand, have co-evolved with phages and have developed protective mechanisms to evade phage infections. The AI model developed by the researchers learns the rules by which phages evolve to be successful, enabling the creation of phages that are highly effective against pathogens. After testing the model with real-world obstacles, the team engineered phages carrying AI-suggested mutations and evaluated their performance in the laboratory.

The results showed that the phages, when carrying the AI-suggested mutations, were capable of infecting their bacterial hosts up to six orders of magnitude more effectively than their natural counterparts. The AI model also helped identify phage mutations that target specific bacteria while sparing others, which is crucial for creating treatments that could combat infections without disrupting beneficial microbial communities in the gut.

The AI model builds upon previous research in the Raman Lab, where high-throughput experimental approaches were used to explore the effects of mutations on key phage proteins that control the phage's ability to recognize and infect bacterial hosts. The model's output of novel phage mutations is expected to lead to more efficient and viable options for phage therapy, including targets not yet achievable through previous methods.

Ultimately, the researchers aim to engineer phages that can address various bacterial pathogens, such as those responsible for urinary tract infections, microbiome-linked diseases, and agricultural pathogens like Salmonella in poultry.