Using artificial intelligence, researchers say, they’ve found a new type of antibiotic that works against a particularly menacing drug-resistant bacteria.

When they tested the antibiotic on the skin of mice that were experimentally infected with the superbug, it controlled the growth of the bacteria, suggesting that the method could be used to create antibiotics tailored to fight other drug-resistant pathogens.

What’s more, the compound identified by AI worked in a way that stymied only the problem pathogen. It didn’t seem to kill the many other species of beneficial bacteria that live in the gut or on the skin, making it a rare narrowly targeted agent.

If more antibiotics worked this precisely, the researchers said, it could prevent bacteria from becoming resistant in the first place.

The study was published in the journal Nature Chemical Biology.

“It’s incredibly promising,” said Dr. Cesar de la Fuente, an assistant professor at the University of Pennsylvania’s Perlman School of Medicine who is also using AI to find new treatments but was not involved in the new research.

De la Fuente says this type of approach to finding new drugs is an emerging field that researchers have been testing since about 2018. It dramatically cuts the time it takes to sort through thousands of promising compounds.

“I think AI, as we’ve seen, can be applied successfully in many domains, and I think drug discovery is sort of the next frontier.”

For the study, researchers focused on the bacteria Actinetobacter baumanii. It hangs out in hospitals and other health care settings, clinging to surfaces like doorknobs and counters. Because it is able to grab bits of DNA from other organisms it comes into contact with, it can incorporate their best weapons: genes that help them resist agents doctors use to treat them.

“It’s what we call in the laboratory a professional pathogen,” said Jon Stokes, one of the researchers and an assistant professor of biochemistry and biomedical sciences at McMaster University in Hamilton, Ontario.

This species causes difficult-to-treat skin, blood or respiratory infections. The US Centers for Disease Control and Prevention said in 2019 that Acinetobacter baumanii infections were “of greatest need” for new types of antibiotics to treat them.

A recent study of hospital patients with Actinetobacter baumanii infections that were resistant even to powerful carbapenem antibiotics found that 1 in 4 had died within a month of their diagnosis.

For the new study, Stokes and lab teamed up with researchers from the Broad Institute at MIT and Harvard. First, they used a technique called high-throughput drug screening to grow Acinetobacter baumanii in lab dishes and spent weeks exposing these colonies to more than 7,500 agents: drugs and the active ingredients of drugs. They found 480 compounds that blocked the growth of the bacteria.

They fed that information into a computer and used it to train an artificial intelligence algorithm.

“Once we had our model trained, what we could do then is start showing that model brand-new pictures of chemicals that it had never seen, right? And based on what it had learned during training, it would predict for us whether those molecules were antibacterial or not,” Stokes said.

They then had the model screen more than 6,000 molecules, which Stokes said the AI was able to do over the course of a few hours.

They narrowed the search to 240 chemicals, which they tested in the lab. The lab testing helped them whittle the list to nine of the best inhibitors of the bacteria. From there, they took a closer look at the structure of each one, eliminating those they thought might be dangerous or related to known antibiotics.

They were left with one compound, called RS102895, which Stokes thinks had been originally developed as a potential treatment for diabetes.

He says it appears to work in a completely new way, by preventing components of the bacteria from traveling from inside the cell to its surface.

“It’s a rather interesting mechanism and one that is not observed amongst clinical antibiotics so far as I know,” he said.

What’s more, he said, RS102895 – which the researchers renamed abaucin – works only on Actinetobacter baumanii.

Stokes says most antibiotics are broad-spectrum agents, working against many species of bacteria. Broad-spectrum antibiotics put a lot of selection pressure on many types of bacteria, causing many to quickly evolve, and share, genes that help them resist the drug and survive.

“With this molecule, because it only works very potently against Actinetobacter, it doesn’t impose that universal selective pressure, so it’s not going to spread resistance quite as quick,” he said.