AI Predicts Mechanism of Investigational Antibiotic for Crohn's Disease

While AI has been used to identify potential drug candidates from libraries of promising molecules, investigators now claim a "global first" for applying AI to predict the mechanism of action of an investigational antibiotic. "A lot of AI use in drug discovery has been about searching chemical space, identifying new molecules that might be active," said Regina Barzilay, PhD, a study¹ co-author and developer of the AI predictive system, DiffDock, Computer Science & Artificial Intelligence Lab, Massachusetts Institute of Technology (MIT), Cambridge, MA, in a release² announcing the study publication.

"What we're showing here is that AI can also provide mechanistic explanations, which are critical for moving a molecule through the development pipeline," Barzilay explained.

Enterololin

The investigational agent, enterololin, targets pathogens associated with Crohn's disease and inflammatory bowel disease (IBD) including adherent-invasive Escherichia coli (AIEC). It exhibits a relatively narrow spectrum, which investigators anticipate will lessen the dysbiosis associated with other agents which can facilitate opportunistic organisms such as carbapenem-resistant Enterobacteriaceae (CRE).

Principle investigator, Jonathan Stokes, PhD, Biochemistry and Biomedical Sciences, McMaster University, Hamilton, Ontario, Canada, commented on the development in the news release.

"This new drug is a really promising treatment candidate for the millions of patients living with IBD. We currently have no cure for these conditions, so developing something that might meaningfully alleviate symptoms could help people experience a much higher quality of life," Stokes said.

Stokes and colleagues had screened over 10,000 bioactive small molecules before identifying the candidate agent that would be active against AIEC to >95% normalized growth inhibition.The identified agent, initially designated BAY-524 and renamed enterololin, is a Bub1 kinase inhibitor which the investigators coupled with an analog of polymyxin B to enhance disruption of the Gram-negative organism's outer membrane.

Check out this past interview with César de la Fuente, PhD, who provides insights on the promising work of his lab as they also accelerate the speed of finding new antimicrobial molecules through the use of AI.

Application of the agent with this additive demonstrated a minimum inhibitory concentration (MIC) of 16mcg/ml and 32mcg/ml against two selected AIEC strains.Importantly, they note, it showed minimal to no growth inhibition of most commensal isolates.

Stokes and colleagues then considered whether AI might be harnessed to reduce the time-consuming and costly challenge of ascertaining the agent's mechanism of action. "AI has expedited the rate at which we can explore chemical space for new drug candidates, but, until now, it has done little to alleviate a major bottleneck in drug development, which is understanding what these new drug candidates actually do," Stokes indicated.

They began with a literature search on mechanisms of antibacterial molecules with substructures common to enterololin. From this, they hypothesized activity against Gram-negative bacteria through inhibition of the lipoprotein transport system, specifically the LoICDE complex.

"To efficiently explore this mechanistic hypothesis, we leveragd DiffDock-L, a deep learning-based molecular docking algorithm, to inform on the potential interactions between enterololin and the E coli LoICDE complex," Stokes and colleagues wrote.

In contrast to the projected 6 months and estimated $2 million that would be

necessary to explore this hypothesis in the lab, Stokes reports that AI concurred with their prediction in 100 seconds, and they were then able to complete laboratory confirmation in 6 months at a cost of about $60,000.

"Currently, we can't just assume that these AI models are totally right, but the notion that it could be right took the guesswork out of our next steps," explained Stokes.

Stokes and colleagues anticipate that the investigational agent could be ready for clinical trials within 3 years.They also expect that this demonstrated application of AI will spur further advances in drug development.

"Collectively, this study shows the utility of machine learning to help elucidate the MOA (mechanism of action) of a novel antibacterial molecule and introduces a promising candidate for further translational optimization," Stokes and colleagues wrote.

References

1. Catacutan DB, Tu MM, Brown ED, et al. Discovery and artificial intelligence-guided mechanistic elucidation of a narrow-spectrum antibiotic. Nat Microbiol. 2025.https://doi.org/10.1038/s41564-025-02142-0

2. Dillion B. New antibiotic targets IBD—and AI predicted how it would work before scientists could prove it. News Release, McMaster University. Posted Oct 3, 2025.https://healthsci.mcmaster.ca/new-antibiotic-targets-ibd-and-ai-predicted-how-it-would-work-before-scientists-could-prove-it/