Penn AI Finds 1,179 Prionin Antibiotic Candidates, With 2 Cutting Bacteria in Mice
Updated
Updated · Newswise · Jun 19
Penn AI Finds 1,179 Prionin Antibiotic Candidates, With 2 Cutting Bacteria in Mice
3 articles · Updated · Newswise · Jun 19
Summary
Penn researchers used the APEX 1.1 deep-learning platform to scan 19.3 million peptide fragments from 2,897 prion and prion-like proteins, identifying 1,179 antimicrobial candidates they named prionins.
Lab validation narrowed that AI list to 75 peptides, of which 59 inhibited at least one of 11 bacterial pathogens and 42 showed strong activity at low concentrations.
Safety and animal tests strengthened the case: 16 active peptides showed no measurable harm to red blood cells or human cells, and two reduced Acinetobacter baumannii levels in a mouse skin-infection model with effects comparable to polymyxin B.
The Nature Microbiology study points to prion-related proteins—better known for links to fatal neurodegenerative disease—as an unexpected reservoir for antibiotic discovery amid rising drug resistance.
Researchers said the work does not show prionins are naturally released during infection or alter what is known about harmful misfolded prions, but it opens a new search space linking protein aggregation and innate immunity.
Are prion proteins a secret weapon in our immune system, waiting to be activated against invading microbes?
Could antibiotics from deadly prions accidentally trigger the brain diseases they are known to cause?
Why would evolution hide life-saving antibiotics inside proteins also capable of causing fatal brain diseases?
Prionins: AI-Driven Identification of Over 1,000 New Antimicrobial Peptides from Prion Proteins
Overview
A landmark study published in Nature Microbiology on June 19, 2026, announced the discovery of prionins, a new class of antimicrobial peptides (AMPs) derived from prion and prion-like proteins. This breakthrough redefines the understanding of these proteins, revealing them as a rich, previously untapped source of potential antibiotics. Using the advanced APEX 1.1 deep learning platform, researchers systematically explored nearly 2,900 prion-related proteins, generating over 19 million peptide fragments. This innovative approach led to the identification of prionins, offering new hope in the fight against drug-resistant pathogens and opening fresh avenues for antibiotic discovery.