Particle.news

Download on the App Store

AI-Designed Antibiotic Candidates Defeat Drug-Resistant Gonorrhea and MRSA in Mice

MIT alongside nonprofit Phare Bio is applying medicinal chemistry to refine these candidates into forms suitable for clinical testing.

Overview

  • Researchers used fragment-based and unconstrained AI pipelines incorporating CReM mutations and a fragment-VAE model to generate and computationally screen more than 36 million novel molecules.
  • Of the top computational hits, only two molecules could be synthesized and were designated NG1 and DN1 for experimental evaluation.
  • NG1 demonstrated potent activity against drug-resistant Neisseria gonorrhoeae in vitro and in mice and was shown to bind LptA, a novel bacterial membrane protein target.
  • DN1 cleared methicillin-resistant Staphylococcus aureus skin infections in a mouse model after showing strong antibacterial effects in laboratory cultures.
  • Next steps include medicinal chemistry, scale-up of synthesis and preclinical safety studies, with researchers acknowledging challenges in manufacturing, regulatory approval and demonstrating human efficacy.