Machine learning algorithms have identified a compound that kills Acinetobacter baumannii, a bacterium that lurks in many hospital environments.
Researchers from the Massachusetts Institute of Technology (MIT) and McMaster University used artificial intelligence algorithms to discover a new antibiotic that can kill bacteria responsible for many drug-resistant infections.
The drug can help patients fight against Acinetobacter baumannii, a bacterium commonly found in hospitals that can cause pneumonia, meningitis, and other severe infections. This microorganism is also a major cause of infection in injured soldiers in Iraq and Afghanistan.
"Acinetobacter can survive for a long time on hospital doorknobs and equipment, and can acquire antibiotic resistance genes from the environment. Now, strains of Acinetobacter baumannii have been found to be resistant to almost all antibiotics," said Jonathan Stokes, former MIT postdoctoral fellow and current assistant professor of biochemistry and biomedical sciences at McMaster University.
The researchers used a machine learning model to identify this new drug from a library of nearly 7,000 potential drug compounds. They trained the model to evaluate whether a compound would inhibit the growth of Acinetobacter baumannii.
"This finding further supports the inference that artificial intelligence can significantly accelerate and expand our exploration of new antibiotics," said James Collins, professor of medical engineering and science at MIT's Institute for Medical Engineering and Science (IMES) and Department of Biological Engineering. "I am pleased that this work demonstrates that we can use artificial intelligence to combat problematic pathogens such as Acinetobacter baumannii."
Drug Discovery
In recent decades, many pathogens have become increasingly resistant to existing antibiotics, while few new antibiotics have been developed.
A few years ago, Collins, Stokes, and MIT professor Regina Barzilay (also an author of this new study) embarked on using machine learning to address this growing problem. Machine learning is a form of artificial intelligence that can learn to identify patterns in large amounts of data. Collins and Barzilay are co-directors of the Abdul Latif Jameel Clinic for Machine Learning in Health at MIT, and they hoped this approach could be used to identify new antibiotics with chemical structures different from any existing drugs.
In the initial demonstration, the researchers trained a machine learning algorithm to identify chemical structures that could potentially inhibit the growth of Escherichia coli. In a screening of over 100 million compounds, the algorithm produced a molecule called "halicin," named after the fictional artificial intelligence system in "2001: A Space Odyssey." They found that this molecule not only killed E. coli but also several other drug-resistant bacteria.
"After that paper, when we demonstrated that these machine learning methods could be used for complex antibiotic discovery tasks, we turned our attention to what I consider to be the public enemy number one in multi-drug-resistant bacterial infections, Acinetobacter baumannii," Stokes said.
To obtain training data for the computational model, the researchers first exposed Acinetobacter baumannii grown in laboratory petri dishes to around 7,500 different compounds to see which ones could inhibit the microorganism's growth. They then input the structure of each molecule into the model and told it whether each structure could inhibit bacterial growth. This allowed the algorithm to learn the chemical features associated with growth inhibition.
Once the model was trained, the researchers used it to analyze a set of 6,680 compounds that had never been seen before, obtained from the Broad Institute's Drug Repurposing Hub. This analysis took less than two hours and produced several hundred top hits. The researchers selected 240 compounds for experimental testing in the lab, focusing on compounds with structures different from existing antibiotics or the molecules in the training data.
These tests yielded nine antibiotics, one of which was highly effective. This compound was initially thought to be a potential diabetes drug, but it turned out to be highly effective in killing Acinetobacter baumannii while having no effect on other types of bacteria, including Pseudomonas aeruginosa, Staphylococcus aureus, and carbapenem-resistant Enterobacteriaceae.
This "narrow-spectrum" killing ability is an ideal characteristic of antibiotics as it minimizes the risk of bacteria rapidly developing resistance to the drug. Another advantage is that the drug may not harm beneficial bacteria living in the human gut and may help suppress opportunistic infections such as Clostridium difficile.
"Antibiotics typically have to be administered systemically, and the last thing you want to see is a severe ecological disruption that leads to secondary infections in these already sick patients," Stokes said.
A Novel Mechanism
In mouse studies, the researchers found that the drug they named "abaucin" could treat wound infections caused by Acinetobacter baumannii. Laboratory tests also showed its effectiveness against various drug-resistant strains of Acinetobacter baumannii isolated from human patients.
Further experiments revealed that the drug kills cells by interfering with the process of lipoprotein transport, which is the process by which cells transport proteins from the interior to the outer membrane. 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 abaucin had such strong selectivity for Acinetobacter baumannii. They speculate that the slightly different way in which Acinetobacter baumannii performs this task may be the reason for the drug's selectivity.
Stokes' lab is now collaborating with other researchers at McMaster University to optimize the drug's pharmaceutical properties with the hope of developing it for patients.
The researchers also plan to use their modeling approach to identify other potential antibiotics for treating other types of drug-resistant infections.
Source: https://news.mit.edu/2023/using-ai-scientists-combat-drug-resistant-infections-0525
Alphabet Q1 paid subscribers reached 350M, fueled by YouTube & Google One. YouTube ad revenue slightly missed at $9.88B amid Premium growth. Total Q1 revenue hit $109.9B, beating forecasts.
Anthropic releases Claude connectors for Adobe, Blender, and Ableton, enabling AI workflow automation and app control. It also sponsors the Blender Development Foundation.
Apple introduces an App Store model splitting 12-month plans into monthly installments. On iOS 26.4+, it lowers upfront costs without reducing total price, adding billing tracking.
Google launches Aadhaar digital IDs in Wallet for India, enabling secure QR/NFC verification. A phased partner rollout targets key sectors, while passport-based credentials expand globally.
YouTube tests "Ask YouTube," an AI tool blending text and video for complex queries. Initially for US Premium users, it supports interactive follow-ups, expanding Google's AI search.
Google grants US DoD classified AI access, joining OpenAI and xAI in defense deals amid Anthropic's dispute. Contractual safeguards against surveillance and lethal weapons remain unenforceable.
Mistral AI launches Workflows, a Temporal-powered orchestration platform for secure, Python-driven enterprise AI automation featuring human-in-the-loop approvals.