by Ruggiero Corcella
Using artificial intelligence, MIT researchers identify new class of antibiotic candidates These compounds can kill methicillin-resistant Staphylococcus aureus (MRSA), a bacterium that causes deadly infections
Thanks to deep learning, an artificial intelligence (AI) method that teaches computers to process data in a way that is inspired by the human brain, researchers at the Massachusetts Institute of Technology (MIT) have discovered a class of compounds capable of kill a drug-resistant bacterium that causes more than 10,000 deaths each year in the United States. In a study published in Nature researchers demonstrated that these compounds could kill methicillin-resistant Staphylococcus aureus (MRSA) cultured in a laboratory dish and in two mouse models of MRSA infection.
The compounds also show very low toxicity towards human cells, making them particularly suitable drug candidates. A key innovation of the new study is that the researchers were also able to understand what types of information the deep learning model used to make predictions about the potency of antibiotics. This knowledge could help researchers design additional drugs that might work even better than those identified by the model.
What is antibiotic resistance
Antibiotic resistance is a natural biological phenomenon of adaptation of some microorganisms, which acquire the ability to survive or grow in the presence of a concentration of an antibacterial agent. It has now become one of the problems that make the nights of researchers, doctors and health policy makers sleepless at a global level: in Italy alone, it is estimated that increasingly aggressive microbes and increasingly less effective drugs (also due to the reckless use of antibiotics ) are responsible for approximately 15 thousand deaths per year.
Second the latest data from the World Health Organization (WHO), the rate of infections due to MRSA worsened by 14%, from 21% in 2016 to 35% in 2020 globally. And, as indicated by the WHO itself, in 2050 antibiotic resistance will be the leading cause of death globally, with 10 million deaths.
The role of artificial intelligence
The development of new drugs is a complex, expensive and long-term process (10 years on average). In the last 5 years, artificial intelligence has entered the field in the discovery of new molecules to be used as drugs. With its computing capacity, AI can analyze huge data sets
to identify potential drug candidates and predict their efficacy.
Machine learning models can simulate molecular interactions and evaluate the safety and efficacy of new drugs, significantly accelerating the drug development process.
Discover 7 new antibiotics in the space of 7 years
The target of the project
led by James Collins, Termeer Professor of Medical Engineering and Science in the Institute for Medical Engineering and Science (IMES) and the Department of Biological Engineering at MIT, is discovering new classes of antibiotics against seven types of deadly bacteria, over seven years.
“The insight here was that we could see what was being learned by the models to make their predictions that certain molecules would make good antibiotics. Our work provides a time-, resource- and mechanically efficient structure, from a chemical structure perspective, in ways that we haven’t had before,” he says.
Felix Wong, research fellow at IMES and the Broad Institute of MIT and Harvard, and Erica Zheng, of the IDMP (Infectious Disease and Microbiome Program), Broad Institute of MIT and Harvard, Cambridge, are the lead authors of the study, which is part of the Antibiotics-AI project.
Explainable forecasts
MRSA, which infects more than 80,000 people in the United States each year, often causes skin or pneumonia. Severe cases can lead to sepsis. In recent years, Collins and his colleagues at MIT’s Abdul Latif Jameel Clinic for Machine Learning in Health (Jameel Clinic) have begun using deep learning to try to find new antibiotics.
Their work has produced potential anti-drugs l’Acinetobacter baumannii
a bacterium often found in hospitals, and many other bacteria drug resistant. These compounds were identified using deep learning models that can learn to identify chemical structures associated with antimicrobial activity. The models then analyze millions of other compounds, generating predictions about which ones might have strong antimicrobial activity.
This is a very fruitful type of research, but a limitation of this approach is that the models are “black boxes”, which means that there is no way of knowing on which characteristics the model has based its forecasts. If scientists knew how models make their predictions, it might be easier for them to identify or design additional antibiotics. “What we set out to do in this study was open the black box,” says Wong. “These models consist of a very large number of calculations that mimic neural connections, and no one really knows what happens under the hood.”
39 thousand compounds tested
First, the researchers trained a deep learning model using extensive datasets. They generated this training data by testing the antibiotic activity of approximately 39,000 compounds against MRSA, then fed this data, plus information about the compounds’ chemical structures, into the model. “You can represent virtually any molecule as a chemical structure and also tell the model whether that chemical structure is antibacterial or not,” Wong says. «The model is trained on many examples like this. If you then give it a new molecule, a new arrangement of atoms and bonds, it can tell you the expected probability of that compound being antibacterial.”
Powerful algorithms
To understand how the model was making its predictions, the researchers adapted an algorithm known as Monte Carlo tree search, which has been used to help make other deep learning models, such as AlphaGo, more explainable. This search algorithm allows the model to generate not only an estimate of the antimicrobial activity of each molecule, but also a prediction for which substructures of the molecule are likely to represent that activity. To further narrow the pool of drug candidates, the researchers trained three additional deep learning models to predict whether the compounds were toxic to three different types of human cells.
By combining this information with predictions of antimicrobial activity, researchers discovered compounds that could kill microbes while having minimal adverse effects on the human body. Using this collection of models, the researchers screened approximately 12 million compounds, all commercially available. From this collection, the models identified compounds from five different classes, based on the chemical substructures within the molecules, that were predicted to be active against MRSA.
Tests on 280 compounds
The researchers purchased around 280 compounds and tested them against MRSA grown in a laboratory dish, allowing them to identify two, from the same class, that appeared to be very promising antibiotic candidates. In tests on two mouse models, one of cutaneous MRSA infection and one of systemic MRSA infection, each of these compounds reduced the MRSA population by a factor of 10. The experiments revealed that the compounds appear to kill the bacteria by disrupting the their ability to maintain an electrochemical gradient across cell membranes. This gradient is necessary for many critical cellular functions, including the ability to produce ATP (adenosine triphosphate, molecule that cells use to store energy).
An antibiotic candidate discovered by Collins’ lab in 2020, halicin, appears to work by a similar mechanism but is specific to Gram-negative bacteria (bacteria with thin cell walls). MRSA is a Gram-positive bacterium, with thicker cell walls. “We have pretty strong evidence that this new structural class is active against Gram-positive pathogens by selectively dissipating the proton motive force in the bacteria,” says Wong. «The molecules selectively attack bacterial cell membranes, in a way that does not cause substantial damage to human cell membranes. Our enhanced deep learning approach allowed us to predict this new structural class of antibiotics and find that it is not toxic against human cells.”
Corriere della Sera is also on Whatsapp. It’s enough click here to subscribe to the channel and always be updated.
December 26, 2023 (modified December 26, 2023 | 10:09)
© ALL RIGHTS RESERVED