Scientists at the University of Pennsylvania used an AI platform to scan millions of peptide fragments inside prion proteins — the same proteins linked to fatal brain diseases — and discovered nearly 1,200 potential antibiotic candidates that could one day help fight drug-resistant infections.
Scientists studying some of biology’s most feared proteins — the misfolded prions linked to rare and fatal brain diseases — have stumbled onto an unexpected trove of potential antibiotics. Researchers at the University of Pennsylvania Perelman School of Medicine deployed an artificial intelligence tool to systematically scan prion and prion-like proteins, uncovering nearly 1,200 short molecular fragments that may be capable of killing bacteria. The findings were published June 19 in the journal Nature Microbiology.
A Surprising Place to Look for New Drugs
The antibiotic pipeline has been running dangerously low for decades. Drug-resistant bacteria — sometimes called superbugs — are increasingly rendering standard treatments ineffective, creating an urgent global health challenge. While researchers have explored everything from ocean microbes to venom in search of new compounds, prion proteins were never on the shortlist. That changed when the Penn team asked a deceptively simple question: what if these proteins were hiding something useful?
Earlier research had suggested that fragments of certain proteins involved in neurodegeneration, including amyloid-beta and the cellular prion protein, might have some antimicrobial properties. But those observations were scattered and anecdotal — no one had conducted a large-scale, systematic search across the full landscape of prion and prion-like proteins. The Penn team used AI to do exactly that.
How AI Scanned Millions of Molecular Fragments
The researchers fed a deep-learning platform called APEX 1.1 a dataset of 19.3 million short peptide fragments drawn from 2,897 prion and prion-like proteins. The platform is designed to predict antimicrobial activity based on amino acid sequences alone. Out of those millions of candidates, it flagged 1,179 with potential antibiotic properties. The team named this newly identified class of molecules “prionins.”
Senior author César de la Fuente, a presidential associate professor and director of the Machine Biology Group at the University of Pennsylvania Perelman School of Medicine , described the significance of finding these candidates in such an unexpected place.
“This work changes where we think antibiotics might be hiding,” de la Fuente said in a news release. “Prions have long been seen almost entirely through the lens of disease, but AI let us ask a different question: whether these proteins also encode useful molecular fragments. The answer appears to be yes.”
From Screen to Lab Bench to Living Organisms
Computational predictions only go so far, which is why the team moved 75 of the most promising prionins into the laboratory. Tested against 11 bacterial pathogens — including several drug-resistant strains — 59 of those peptides inhibited at least one pathogen, and 42 demonstrated strong activity at low concentrations. Many appeared to work by rupturing bacterial membranes, a well-established mechanism used by natural antimicrobial peptides.
Safety profiles were encouraging as well. Sixteen of the active prionins showed no detectable harm to human cells or red blood cells even at the highest concentrations tested. Two lead candidates — one derived from a fungus, the other from a roundworm — then moved into animal trials. In a mouse model of skin infection caused by Acinetobacter baumannii, a particularly difficult-to-treat pathogen, both peptides reduced bacterial burden at a level comparable to polymyxin B, a last-resort antibiotic, with no signs of treatment-related weight loss.
Co-first author Marcelo D. T. Torres emphasized that the animal results were what elevated the work from promising theory to genuine discovery.
“This is where the story becomes more than a computer screen,” Torres said in the news release. “The AI search gave us a short list of candidates, but the important point is that many of those molecules worked in the lab, and two worked in an animal infection model. That is what makes this a discovery platform, not just a prediction exercise.”
A Broader Hunt for Hidden Biology
The prionin study is part of a larger research program at the de la Fuente Lab focused on what the team calls “encrypted peptides” — short sequences embedded within larger proteins that only reveal their biological activity when isolated. The group has previously mined human proteins, extinct organisms, archaea, microbiomes and venoms for similar hidden compounds. Prion proteins represent one of the most counterintuitive additions to that list yet.
Importantly, the findings do not suggest that prions naturally function as antibiotics in the body, nor do they revise the understanding of how misfolded prions cause neurodegeneration. What the work does suggest is that these proteins may be an entirely overlooked reservoir for future drug candidates.
“For a long time, drug discovery has been limited not only by what we can test, but by where we choose to look,” de la Fuente added. “AI is changing that. It gives us a way to search the hidden layers of biology and ask whether molecules associated with one story – in this case, disease – may also carry another story with therapeutic potential.”
Why It Matters for Students and Future Researchers
For students pursuing careers in medicine, pharmacology, biology or bioinformatics, this study illustrates just how dramatically AI is reshaping the discovery process. The ability to computationally screen tens of millions of molecular fragments in a fraction of the time it would take in a traditional lab is opening entirely new corridors of inquiry — and prion proteins are a vivid example of how longstanding assumptions about what is “useful” in biology may be worth revisiting.
Drug-resistant infections already claim millions of lives annually, and that toll is projected to grow. Research programs that expand the universe of places scientists look for antibiotics — including inside proteins once written off as purely destructive — are exactly the kind of work that could shape medicine for the next generation of practitioners and scientists.