Researchers at the University of Texas at Austin have made a breakthrough in developing safe and effective antibiotics to combat resistant bacteria, leveraging artificial intelligence to engineer a new drug that has shown promise in animal trials. The team used a large language model, similar to the one powering ChatGPT, to reengineer an existing antibiotic called Protegrin-1, which is great at killing bacteria but toxic to humans.
By identifying ways to modify the antibiotic without losing its activity, the researchers created a safer and more effective version, dubbed bacterially selective Protegrin-1.2 (bsPG-1.2). In mice infected with multidrug-resistant bacteria, treatment with bsPG-1.2 significantly reduced detectable bacteria in their organs. The study’s co-senior authors, Claus Wilke and Bryan Davies, believe that large language models will be widely used for developing therapeutics going forward. This project highlights the potential of artificial intelligence to meet societal needs, a key theme at UT, which has declared 2024 the Year of AI.
AI-Powered Antibiotic Discovery: A New Frontier in Combating Resistant Bacteria
The rise of antibiotic-resistant bacterial strains has become a significant public health concern, with the development of new treatment options stagnating in recent years. However, researchers at The University of Texas at Austin have made a promising breakthrough by leveraging artificial intelligence (AI) to develop a safer and more effective version of an existing antibiotic.
AI-Driven Protein Engineering: A Game-Changer for Antibiotic Development
The researchers utilized a large language model, similar to the one powering ChatGPT, to engineer a version of the bacteria-killing drug Protegrin-1 that was previously toxic in humans. This approach marks a significant shift in the field of protein and peptide engineering, where machine learning applications are becoming increasingly important. According to Claus Wilke, professor of integrative biology and statistics and data sciences, “Large language models are a major step forward for machine learning applications in protein and peptide engineering… Many use cases that weren’t feasible with prior approaches are now starting to work.”
From Text Sequences to Protein Sequences: The Power of Large Language Models
Large language models were originally designed to generate and explore sequences of text. However, scientists have found creative ways to apply these models to other domains, including protein engineering. Proteins are made up of sequences of amino acids, similar to how sentences are composed of sequences of words. By clustering proteins that share similar functions in an AI embedding space, researchers can identify patterns and relationships that would be difficult or impossible to discern through traditional laboratory approaches.
Reengineering Protegrin-1: A Safer and More Effective Antibiotic
The researchers employed AI to identify ways to reengineer Protegrin-1, a naturally produced antibiotic by pigs to combat infections. They created over 7,000 variations of the drug and used a protein LLM to evaluate millions of possible variations for three key features: selectively targeting bacterial membranes, potently killing bacteria, and not harming human red blood cells. The model guided the team to a safer and more effective version of Protegrin-1, dubbed bacterially selective Protegrin-1.2 (bsPG-1.2).
Promising Results in Mice Trials
Mice infected with multidrug-resistant bacteria and treated with bsPG-1.2 were significantly less likely to have detectable bacteria in their organs six hours after infection compared to untreated mice. If further testing yields similarly positive results, the researchers hope to eventually take a version of the AI-informed antibiotic drug into human trials.
The Future of Antibiotic Development: Machine Learning’s Impact
The study highlights the twofold impact of machine learning on antibiotic development. Firstly, it can point out new molecules with potential therapeutic benefits, and secondly, it can show researchers how to take existing antibiotic molecules and make them better, focusing their work to more quickly get those to clinical practice. As Bryan Davies, co-senior author of the study, noted, “Machine learning’s impact is twofold… It’s going to point out new molecules that could have potential to help people, and it’s going to show us how we can take those existing antibiotic molecules and make them better.”
External Link: Click Here For More
