Gonorrhea, be gone! MIT researchers use AI to create 2 new antibiotics that kill superbugs

MIT researchers have developed two novel antibiotics using generative AI, including compounds that can kill drug-resistant gonorrhea and MRSA infections. The breakthrough demonstrates how AI can design entirely new molecules by exploring previously inaccessible chemical spaces, potentially addressing the growing crisis of antimicrobial resistance that causes nearly 5 million deaths annually.

The big picture: While only a few dozen new antibiotics have been approved over the past 45 years—most being variants of existing drugs—this AI-driven approach generated over 36 million theoretical compounds that are structurally distinct from any known antibiotics.

How it works: The MIT team, led by James Collins, employed two different generative AI strategies to design novel antimicrobial compounds.

• The first approach used chemical fragments with known antimicrobial activity as building blocks, starting with a library of 45 million fragments and narrowing down to promising candidates through machine learning models.
• The second method allowed AI algorithms to freely generate molecules without constraints, using only basic chemical rules for how atoms can combine.
• Both approaches used algorithms called CReM (chemically reasonable mutations) and VAE (variational autoencoder) to generate and evaluate millions of potential compounds.

In plain English: Think of it like two different approaches to designing new keys. The first method starts with pieces of keys that already work on some locks (chemical fragments with antimicrobial properties) and combines them in new ways. The second method creates entirely new key shapes from scratch, following only basic rules about what makes a functional key.

Key breakthroughs: Two lead compounds emerged from the research, each targeting different types of drug-resistant bacteria.

• NG1 proved effective against drug-resistant Neisseria gonorrhoeae (the bacteria that causes gonorrhea) by interfering with LptA protein, disrupting bacterial membrane synthesis in a novel way.
• DN1 successfully cleared methicillin-resistant Staphylococcus aureus (MRSA) skin infections in mouse models by broadly affecting bacterial cell membranes.
• Both compounds work through mechanisms that haven’t been seen in existing antibiotics, potentially making them harder for bacteria to develop resistance against.

What they’re saying: The research opens new possibilities for combating antimicrobial resistance through computational drug design.

• “We’re excited about the new possibilities that this project opens up for antibiotics development. Our work shows the power of AI from a drug design standpoint, and enables us to exploit much larger chemical spaces that were previously inaccessible,” says James Collins, the study’s senior author.
• “We wanted to get rid of anything that would look like an existing antibiotic, to help address the antimicrobial resistance crisis in a fundamentally different way,” explains lead author Aarti Krishnan.

What’s next: Phare Bio, a nonprofit partner in the Antibiotics-AI Project, is now working to modify NG1 and DN1 for additional testing and potential clinical development.

• The team plans to apply these AI platforms to other challenging bacterial pathogens, including Mycobacterium tuberculosis (which causes tuberculosis) and Pseudomonas aeruginosa.
• Further medicinal chemistry work will focus on creating analogs and advancing the best candidates through preclinical testing.