MIT researchers have successfully used generative AI to design novel antibiotics that can combat two hard-to-treat infections: drug-resistant Neisseria gonorrhoeae and multi-drug-resistant Staphylococcus aureus (MRSA). The AI-designed compounds, built atom-by-atom, successfully killed the bacteria in laboratory and animal tests.
The researchers employed two distinct AI-driven approaches, focusing on fragment-based design and unconstrained generation of compounds. They analyzed 36 million real and theoretical compounds, learning how molecular structures affect bacterial growth. The AI then generated novel antibiotic designs while filtering out any that resembled existing drugs, were likely toxic to humans, or unsuitable for use as medicines.
Two promising candidates emerged from the study: NG1 and DN1. NG1 effectively eradicated Neisseria gonorrhoeae in laboratory tests and animal models by targeting LptA, a protein essential for building the bacterial outer membrane. DN1, on the other hand, cleared MRSA skin infections in mice.
According to Prof. James Collins of MIT, this breakthrough could signal a "second golden age" in antibiotic discovery. The use of generative AI allows researchers to explore vast chemical spaces and identify novel compounds that could help combat antibiotic resistance.
While the compounds require further refinement and testing, experts believe this development has enormous potential. However, they also caution that safety and efficacy testing will be crucial before these antibiotics can be used in humans. Additionally, the economic challenge of producing new antibiotics that must be used sparingly to preserve effectiveness could impact commercial viability.
