AI platform discovers new powerful antibiotic

A new antibiotic compound has been discovered by a team of researchers in the US thanks to a machine-learning platform.

Using a computer model which can screen over a hundred million chemical compounds, researchers at the Massachusetts Institute of Technology (MIT) identified the new compound, which was able to kill many disease-causing bacteria, including some antibiotic-resistant strains. 

The machine-learning platform has been designed to identify potential antibiotics that kill bacteria using different mechanisms than existing drugs.

The platform was developed to “usher in a new age of antibiotic drug discovery,” according to James Collins, the Termeer Professor of Medical Engineering and Science in MIT’s Institute for Medical Engineering and Science (IMES) and Department of Biological Engineering.

“Our approach revealed this amazing molecule which is arguably one of the more powerful antibiotics that has been discovered.”

Currently, discovering and developing new antibiotics is both costly and time consuming, with many pharmaceutical companies pulling out of antibiotic pipelines due to a lack of reimbursement.

“We’re facing a growing crisis around antibiotic resistance, and this situation is being generated by both an increasing number of pathogens becoming resistant to existing antibiotics, and an anaemic pipeline in the biotech and pharmaceutical industries for new antibiotics,” Collins said.

The team’s model has been specifically designed to look for chemical features that make molecules effective at killing E. coli. The platform was trained on around 2,500 molecules and then tested against the Broad Institute’s Drug Repurposing Hub, which as a library of 6,000 compounds.

The platform discovered a drug which was previously investigated as a possible diabetes drug and which was shown to have a different chemical structure to any existing antibiotics. The drug - which the team named halicin - was then tested against a range of bacterial strains and was found to work against every species they tested it against, with the exception of one difficult-to-treat lung pathogen.

The researchers also used a halicin-containing ointment to test the drug’s effectiveness in living animals. The team used it to treat mice infected with A. baumannii – a bacterium with a strain that is resistant to all known antibiotics. Importantly, the application of halicin was found to completely clear the infection in 24 hours.

More so, the researchers found that E.Coli did not develop any resistance to halicin during a 30-day treatment period. In contrast, bacteria against the antibiotic ciprofloxacin developed resistance within one to three days, and after 30 days, were about 200 times more resistant to the drug.

The team believes halicin works by disrupting bacteria’s ability to maintain an electrochemical gradient across their cell membranes. This gradient allows bacteria to produce molecules that cells use to store energy, and when it breaks down, the cells die. The researchers state that this is a method which bacteria may find difficult to develop resistance to.

The researchers plan to use their model to design new antibiotics and to optimise existence molecules.

“This groundbreaking work signifies a paradigm shift in antibiotic discovery and indeed in drug discovery more generally,” says Roy Kishony, a professor of biology and computer science at Technion (the Israel Institute of Technology), who was not involved in the study. “Beyond in silica screens, this approach will allow using deep learning at all stages of antibiotic development, from discovery to improved efficacy and toxicity through drug modifications and medicinal chemistry.”