Researchers at the University of McMaster and the Massachusetts Institute of Technology (MIT) have made two scientific discoveries at the same time: they have not only discovered a brand new antibiotic aimed at inflammatory bowel diseases (IBD) and how to use a new type. From their own knowledge, this is a worldwide first for AI.
In detail on October 3, 2025 in the journal Nature Microbiology, Discovery presents a promising treatment option for millions of people affected by Crohn’s disease and other relevant conditions, also presenting important new applications for AI in Discovery Research research.
“This project shows that we are still awakening the surface until the discovery of medicines guided by AI,” says Jon Stokes, assistant professor in the Department of Biochemistry and Biomedical Sciences and a key researcher for the new study.
The development of our new drug, designed to target IBD, has been accelerated with the cooperation between humans and genetic AI. ”
Jon Stokes, Department of Biochemistry and Biomedical Sciences, McMaster University
An antibiotic for IBD
Most antibiotics used in clinics today are “broad spectrum” drugs, which means that they eliminate good bacteria except for those that cause disease – “they are nuclear,” Stokes explains.
This can create opportunities for medication resistant and bacteria, such as E. Coli, move and colonize the intestines, which can aggravate conditions such as Crohn’s.
But Enterololin, the new antibiotic discovered in McMaster, is a “narrow spectrum” drug, which means that it restores microbicide and attacks only a specific error group that causes a disease-in this case, a family of bacteria called Enterobacteriaceae, which occurs.
This means that not only does it kill E. Coli, but also reduces the opportunity for drug -resistant executives to colonize the gut in the first place.
“This new drug is a truly promising candidate for the treatment for millions of patients living with IBD,” says Stokes. “At the moment we have no cure for these conditions, so by developing something that could meaningfully relieve symptoms could help people experience a much higher quality of life.”
How do drugs work? Just ask AI
To date, AI has been used as a tool to predict which molecules may have a therapeutic potential, but this study has used it to describe what researchers call “MOA” (MOA) – or how drugs are added.
“AI has accelerated the rate at which we can explore the chemical space for new drug candidates, but so far it has done nothing to relieve a significant congestion in drug development, which understands what these new drug candidates are doing,” Stokes explains.
MOA studies, he says, are necessary for the development of drugs. They help scientists confirm security, optimize dosage, make modifications to improve effectiveness and sometimes even reveal completely new drug goals. They also help regulators to determine if a particular drug candidate is suitable for use in humans.
But they are also remarkably expensive – and slow.
Stokes says a MOA thorough study can last up to two years and cost about $ 2 million. However, using the AI, his team made Enterololin’s in just six months and for just $ 60,000.
Indeed, after the discovery of his new antibiotic of his laboratory, Stokes linked to his colleagues at MIT and the CSAIL workshop to determine if any of the emerging machine learning platforms could help MOA’s upcoming studies.
In just 100 seconds, he was given a prediction: his new medicine attacked a cluster of tiny proteins called Lolcde, which is essential for the survival of certain bacteria.
“A lot of AI use in the discovery of medicines was to search for a chemical space, identifying new molecules that may be active,” says Regina Barzilay, a professor at the MIT School of Engineering and Diffdock Developer, Model AI who made the prediction. “What we show here is that AI can also provide mechanistic explanations, which are critical for moving a molecule through the development pipeline.”
Barzilay was recently listed among the most important people in Time magazine at AI.
Stokes stresses that while the prediction was interesting, it was just that – a prediction. It should still have traditional MOA studies in the laboratory.
“At present, we can’t just assume that these AI models are completely right, but the idea that he could be right took the speculation from our next steps,” explains Stokes, a member of the Michael G. Degoote Institute for infectious disease research at McMaster.
And so his team, which led to a large extent by McMaster Denise Catacutan student, began researching Enterolin’s MOA, using MIT’s prediction as a starting point.
Within a few months, it became clear that AI was actually right.
“We did all the standard Moa work to validate the prediction – to see if the experiments would support the AI and they did it,” says Catacutan, a doctorate in the Stokes lab. “By doing this in this way shaved a year and a half of our normal timetable.”
With this, Stokes has now successfully used AI to discover sustainable drug candidates, quick efforts to discover drugs and determine how new drugs work. But, ask him, and he’ll tell you that – beneficial as it is – AI is just a means of an end.
“The resistance of the drugs and the lack of our new drugs are a tap leak,” he says. “You can leave it for a while. But you will eventually have a big problem.” AI is my key – it’s a tool to correct the leak before a flood is done and that’s really.
The trail to Stokes Spin-Out, Stoked Bio, has already granted Enterololin by McMaster and is currently optimizing it for human use.
The company also tests modified versions of the new antibiotic against other drug -resistant bacteria, such as Klebsiella, and the first results are very promising.
“Enterololin’s identification highlights the remarkable science that emerges in McMaster,” says Jeff Skinner, chief executive of Stoked Bio. “We are proud to work with the University to translate this revolution into real treatments for patients.”
If all goes well, Stokes says that the new drug will be ready for human tests within three years – a timetable whose research team is willing to meet.
“Working on something translation like this is surreal,” says the conqueror. “The fact that something we discovered in the lab can someday help patients is really amazing and really enhances the importance of our work.”
Source:
Magazine report:
Conqueror, db, et al. (2025) Discovery and artificial intelligence guided by the intelligence of an antibiotic narrow spectrum. Microbiology of nature. Doi.org/10.1038/S41564-025-02142-0.
