Researchers at the Perelman School of Medicine at the University of Pennsylvania have used artificial intelligence to discover potential antibiotic candidates hiding within prion proteins. These misfolded proteins are best known for causing rare and fatal degenerative brain diseases. The study suggests that prion and prion-like proteins contain short peptides, dubbed "prionins," that can kill bacteria.
This finding upends the conventional view of prions as purely harmful entities. Instead, it suggests these proteins may harbor molecular features linked to ancient immune defense mechanisms. The work opens a new frontier in the search for antibiotics amid a worsening crisis of drug-resistant infections.
The AI system identified the prionins by scanning protein structures for antimicrobial activity. While the research is early-stage, the candidates showed effectiveness against certain bacterial strains in lab tests. No specific potency data or bacterial targets were disclosed in the available report.
The next steps involve synthesizing these prionins and testing them in animal models. If successful, the approach could yield a new class of antibiotics derived from an unexpected biological source. The team emphasizes that the work is preclinical and years away from clinical use.
Experts caution that prion-derived therapies carry inherent risks, given the role of these proteins in neurodegeneration. The researchers note that further study is needed to ensure safety before any therapeutic applications. However, the approach demonstrates AI's growing power to repurpose biological components for novel medical uses.