Bacteria are evolving resistance to antibiotics much faster than new drugs can be developed, potentially leading us to a dangerous future where infections are more likely to be deadly. Now, an artificial intelligence model has identified a powerful new antibiotic called halicin, which cleared infections of most superbugs in mouse tests.
Ever since antibiotics were invented in the early 20th century, we’ve been locked in an arms race with bacteria. Antibiotics work for a while, but eventually the bugs evolve resistance to those in wide use. Scientists develop new ones, so bacteria continue to evolve, and so on. The problem is, we’re starting to lose the battle as the bugs outpace us and fewer new drugs are in the pipeline.
Drug discovery is an arduous task, requiring huge amounts of data to be crunched – and that’s just the kind of job that AI excels at. For this study, researchers from MIT and Harvard started by training a machine learning model on around 2,500 molecules, including existing FDA-approved drugs and other natural products.
Once the system had a good grasp of what biological effects these molecules have, the team then set it loose on a library of about 6,000 drug compounds to search for those that would have strong antibacterial activity. And it found one.
“We wanted to develop a platform that would allow us to harness the power of artificial intelligence to usher in a new age of antibiotic drug discovery,” says James Collins, senior author of the study. “Our approach revealed this amazing molecule which is arguably one of the more powerful antibiotics that has been discovered.”
The molecule in question has been named halicin, after the AI system HAL from 2001: A Space Odyssey. No doubt it’s a tip of the hat to the method used to discover the new drug, but we can’t help but feel that it sounds oddly ominous.
Previously studied as a potential diabetes drug, halicin now has a new life as a strong antibiotic agent. In lab tests, the molecule killed almost every species of bacteria that it was tested against, including Clostridium difficile, Acinetobacter baumannii, and Mycobacterium tuberculosis, which are all resistant to other antibiotics. The only bacteria that halicin couldn’t take down was Pseudomonas aeruginosa, a tough customer that often infects the urinary or respiratory tracts.
Next, the team tested halicin in mice infected with a strain of A. baumannii that’s resistant to all known types of antibiotics. Applying an ointment that contained halicin, the infections were completely cleared in under 24 hours.
In other tests, the team found evidence that the drug works by disrupting the bacteria’s ability to maintain an electrochemical gradient on their outer membranes. This affects how they store energy, quickly killing them. This unique mechanism should make it difficult for bacteria to develop resistance to halicin – as shown in another experiment, where E. coli was found to not develop any resistance to it after 30 days of treatment.
“When you’re dealing with a molecule that likely associates with membrane components, a cell can’t necessarily acquire a single mutation or a couple of mutations to change the chemistry of the outer membrane,” says Jonathan Stokes, first author of the study. “Mutations like that tend to be far more complex to acquire evolutionarily.”
The team says that the focus of the next phase of work will be to develop halicin for eventual human use. The AI system also identified 23 other antibiotic candidates – eight of which showed antibacterial activity in lab tests, and two of which were especially strong. These will be tested further in future as well.
The research was published in the journal Cell.