CAMBRIDGE, Massachusetts — Researchers at MIT and McMaster Universities have used artificial intelligence algorithms to identify a new antibiotic that can kill certain types of bacteria that cause many drug-resistant infections. .
If developed for patients, the drug could help fight infections Acinetobacter baumanniiA type of bacteria commonly found in hospitals that can cause pneumonia, meningitis, and other serious infections. This organism is also a major cause of infections among wounded soldiers in Iraq and Afghanistan.
“Acinetobacter They can survive for long periods on hospital doorknobs and equipment, and may pick up antibiotic resistance genes from the environment.I see it really often now A. Baumani There are isolates that are resistant to almost every antibiotic,” says Jonathan Stokes, a former MIT postdoctoral fellow and now an assistant professor of biochemistry and biomedicine at McMaster University.
Researchers identified new drugs from a library of nearly 7,000 potential drug compounds using machine learning models trained to assess whether compounds inhibit bacterial growth. A. Baumani.
“This finding further supports the premise that AI can significantly accelerate and scale up the search for novel antibiotics,” said Thermeer, professor of biomedical engineering sciences at the Institute of Biomedical Engineering Sciences (IMES) and bioengineering at MIT. One James Collins said: “We are excited that this research has shown that AI can be used to combat problematic pathogens such as: A. Baumani“
Collins and Stokes are senior authors of the new study, announced today natural chemical biology. The paper’s lead authors are McMaster University graduate students Gary Liu and Dennis Kataktan, and McMaster University freshman Khushi Rasod.
drug discovery
Over the past decades, many pathogenic bacteria have developed resistance to existing antibiotics, while few new antibiotics have been developed.
A few years ago, Collins, Stokes, and Massachusetts Institute of Technology professor Regina Bergeley, who also authored a new study, used machine learning, a form of artificial intelligence that can learn to recognize vast numbers of patterns. and set out to address this growing problem. amount of data. Collins and Vergilai, co-directors of MIT’s Abdul Latif Jameel Clinic for Health and Machine Learning, hoped that this approach could be used to identify new antibiotics with chemical structures that differ from existing drugs.
In an initial demonstration, researchers trained a machine learning algorithm to identify chemical structures that could inhibit microbial growth. Escherichia coli. After screening over 100 million compounds, the algorithm produced a molecule the researchers dubbed Halicin, after the fictional artificial intelligence system in 2001: A Space Odyssey. This molecule they showed not only kills people, Escherichia coli However, there are also several other bacterial species that are resistant to treatment.
“After that paper, when I showed that these machine learning approaches worked well for complex antibiotic discovery tasks, I realized we were the biggest enemy against multidrug-resistant bacterial infections. I turned my attention to things. Acinetobactersays Stokes.
To obtain training data for the computational model, the researchers first A. Baumani About 7,500 different compounds were grown in experimental dishes to see which compounds could inhibit microbial growth. The structure of each molecule was then entered into the model. They also told the model whether each structure could inhibit bacterial growth. This allowed the algorithm to learn the chemical signatures associated with growth inhibition.
Once the model was trained, the researchers used it to analyze a set of 6,680 never-before-seen compounds from the Broad Institute’s drug repurposing hub. This analysis took him less than two hours and yielded hundreds of top hits. Of these, the researchers focused on compounds with structures that differed from existing antibiotics and training data molecules, and selected 240 for laboratory experiments.
These tests detected 9 different antibiotics, including some very potent antibiotics. The compound was originally studied as a potential antidiabetic drug, but it turned out to be highly lethal. A. Baumani However, it had no effect on other species of bacteria, including: Pseudomonas aeruginosa, Staphylococcus aureuscarbapenem resistance Enterobacteriaceae.
This ‘narrow-spectrum’ killing ability is a desirable feature for antibiotics because it minimizes the risk of bacteria rapidly spreading resistance to the drug. Another advantage is that the drug will likely help control opportunistic infections, such as: Clostridium difficile.
“Antibiotics often have to be administered systemically, and the last thing you want to do is cause a serious gut dysbiosis and expose an already sick patient to a secondary infection,” Stokes said. .
novel mechanism
In a mouse study, the researchers showed that the drug, which they named Abaucin, could treat wound infections caused by bacteria. A. Baumani. They have also shown in laboratory tests that it is effective against a variety of drug-resistant bacteria. A. Baumani A strain isolated from a human patient.
Further experiments revealed that the drug kills cells by interfering with a process known as lipoprotein transport, which cells use to transport proteins from the cell interior to the cell envelope. Specifically, the drug appears to inhibit her LolE, a protein involved in this process.
The researchers were surprised to find that abaucin is highly selective for its targets, since all Gram-negative bacteria express the enzyme. A. Baumani. They hypothesize that there are slight differences in how they do it. A. Baumani Performing this task may explain the drug’s selectivity.
“Acquisition of experimental data has not yet been conclusively completed, but the following reasons are possible.” A. Baumani Lipoprotein transport differs slightly from other Gram-negative bacteria. We believe that’s why this narrow spectrum of activity is occurring,” says Stokes.
Stokes’ lab is currently collaborating with other researchers at McMaster to optimize the compound’s medicinal properties in hopes of developing it for eventual use in patients.
The researchers also plan to use modeling approaches to identify potential antibiotics against other types of drug-resistant infections, including causative infections. Staphylococcus aureus and Pseudomonas aeruginosa.
###
The study was supported by the David Braley Center for Antibiotic Discovery, Weston Family Foundation, Audacious Project, C3.ai Digital Transformation Institute, Abdul Latif Jameel Clinic for Machine Learning in Health, 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 School of Health Sciences, Boris Family, Marshall Scholarship, Department of Energy Bioenvironmental Research Program.
journal
natural chemical biology
research method
Computational simulation/modeling
article title
Discovery of antibiotics targeting Acinetobacter baumannii based on deep learning
Article publication date
May 25, 2023
