Using an artificial intelligence algorithm, researchers at MIT and McMaster University have discovered a new antibiotic that can kill a type of bacteria responsible for many drug-resistant infections.
If created for use in patients, the drug can help fight Acinetobacter baumannii, a type of bacteria that is often found in hospitals and can lead to pneumonia, meningitis and other serious infections. The microbe is also a major cause of infection in wounded soldiers in Iraq and Afghanistan.
“Acinetobacter can survive on hospital doorknobs and equipment for long periods of time, and can pick up antibiotic resistance genes from its environment. Now it’s really common to find A. baumannii isolates that are resistant to almost all antibiotics,” says Jonathan Stokes, a former MIT postdoc who is now an assistant professor of biochemistry and biomedical sciences at McMaster University.
The researchers identified the new drug from a library of nearly 7,000 potential drug compounds using a machine learning model they trained to assess whether a chemical compound would inhibit growth. A. baumannii.
“This finding further supports the idea that AI can significantly accelerate and expand our search for new antibiotics,” said James Collins, director of medical engineering and science in MIT’s Institute for Medical Engineering and Science (IMES) and the Department of Biological Engineering. professor “I’m excited that this work shows that we can use AI to help fight problematic pathogens such as: A. baumannii“.
Collins and Stokes are senior authors of the new study, which appears today in the 2011 chemical biology of nature. The paper’s lead authors are McMaster graduate students Gary Liu and Denise Katakutan and recent McMaster graduate Khushi Rathod.
drug detection
Over the past few decades, many pathogenic bacteria have become more resistant to existing antibiotics, while very few new antibiotics have been developed.
A few years ago, Collins, Stokes, and MIT professor Regina Barzilay (who is also an author of the new study) set out to combat this growing problem by using machine learning, a type of artificial intelligence that can learn to recognize patterns in large numbers. : data quantities. Collins and Barzilai, who co-directs MIT’s Abdul Latif Jameel Clinic for Machine Learning in Health, hoped that this approach could be used to discover new antibiotics with chemical structures different from existing drugs.
In their initial demonstration, the researchers trained a machine learning algorithm to identify chemical structures that could inhibit growth. E. coli. In a screen of more than 100 million compounds, that algorithm identified a molecule that the researchers named hallis, “2001. “Space Odyssey” in honor of the artistic system of artificial intelligence. This molecule, they showed, could kill not only E. coli but several other types of bacteria are resistant to treatment.
“After that paper, when we showed that these machine learning approaches could work well for the complex task of antibiotic discovery, we turned our attention to what I consider public enemy no. 1 for multidrug-resistant bacterial infections that Acinetobactersays Stokes.
To obtain training data for their computational model, the researchers first identified A. baumannii about 7,500 different chemical compounds were grown in a lab plate to see which ones could inhibit the microbe’s growth. They then fed the structure of each molecule into the model. They also told the model whether or not each structure could inhibit bacterial growth. This allowed the algorithm to learn chemical properties associated with growth inhibition.
Once the model was trained, the researchers used it to analyze a set of 6,680 never-before-seen compounds obtained from the Broad Institute’s Drug Repurposing Hub. This analysis, which took less than two hours, yielded several hundred top hits. Of these, the researchers selected 240 to test in the lab, focusing on compounds with structures that differ from existing antibiotics or molecules from the training data.
Those tests yielded nine antibiotics, including one that was very potent. This compound, which was originally studied as a potential diabetes drug, was found to be extremely effective in killing A. baumannii but had no effect on other types of bacteria, incl Pseudomonas aeruginosa, Staphylococcus aureusand carbapenem resistant Enterobacteriaceae.
This “narrow-spectrum” killing ability is a desirable characteristic of antibiotics because it minimizes the risk of bacteria rapidly spreading resistance to the drug. Another advantage is that the drug is likely to preserve the beneficial bacteria living in the human gut and help suppress opportunistic infections such as: Clostridium difficile.
“Antibiotics often have to be given systemically, and the last thing you want to do is cause significant dysbiosis and open up these already sick patients to secondary infections,” says Stokes.
New mechanism
In studies on mice, the scientists showed that the drug, which they called abusin, could treat wound infections that occur; A. baumannii. They have also shown in laboratory tests that it works against resistance to various drugs A. baumannii strains isolated from human patients.
Further experiments showed that the drug kills cells by interfering with a process called lipoprotein trafficking, which cells use to move proteins from inside the cell to the cell envelope. Specifically, the drug appears to inhibit LolE, a protein involved in this process.
All gram-negative bacteria express this enzyme, so the researchers were surprised to find that abusin was so selective in its targeting. A. baumannii. They assume that slight differences are how A. baumannii performance on this task may explain drug selectivity.
“We haven’t finished collecting pilot data yet, but we think that’s why A. baumannii Lipoproteins do their trafficking a little differently than other Gram-negative species. We believe that’s why we’re getting this narrow-spectrum activity,” Stokes says.
Stokes’ lab is now working with other McMaster researchers to optimize the compound’s therapeutic properties, hoping to develop it for eventual use in patients.
The researchers also plan to use their modeling approach to identify potential antibiotics for other types of drug-resistant infections, including: Staphylococcus aureus and: Pseudomonas aeruginosa.
The research was funded by David Braille Center for Antibiotic Discovery, Weston Family Foundation, Audacious Project, C3.ai Digital Transformation Institute, Abdul Latif Jameel Health Machine Learning Clinic, DTRA Discovery of Medical Countermeasures Against New and Emerging Threats Program, DARPA Accelerated Molecular Discovery Program, Canadian Institutes of Health Research, Genome Canada, McMaster University Faculty of Health Sciences, Boris Family, Marshall Fellowship, and Department of Energy Biological and Environmental Research.