Buffio.

A field journal of machine-made science

← All discoveries

Medicine · 25 May 2023

A narrow-spectrum antibiotic against a hospital pathogen

A neural network trained on a growth-inhibition screen discovered abaucin, a compound with narrow-spectrum activity against Acinetobacter baumannii. It acts through lipoprotein trafficking and controlled an infection in a mouse wound model.

Field
Medicine
Date
25 May 2023
AI system
Chemprop neural network
Institution
McMaster University and MIT
Published in
Nature Chemical Biology
Identifier
10.1038/s41589-023-01349-8

Acinetobacter baumannii is a hospital-associated Gram-negative pathogen that often carries multidrug resistance. The World Health Organization has listed it among the bacteria for which new antibiotics are most urgently needed. Conventional screening has struggled to find new drugs against it.

Researchers at McMaster University and MIT screened about 7,500 molecules for growth inhibition of A. baumannii, trained a neural network on that dataset, and used the model to predict structurally new active compounds. Through this approach they discovered abaucin, an antibacterial compound with narrow-spectrum activity against A. baumannii.

Further work showed that abaucin perturbs lipoprotein trafficking through a mechanism involving the protein LolE, and that it could control an A. baumannii infection in a mouse wound model. Narrow-spectrum activity is itself part of the finding: a drug that hits one pathogen, rather than many, may reduce collateral damage to the microbiome and the pressure that drives broad resistance.