Scientists Built an 'Age Detector' for Cells — Then Used It to Hunt for a Rewind Button
How old is a cell, really? Not by the calendar — by how it behaves. A Harvard-led team has built a machine that reads exactly that, and its first big catch is a gene called SRSF1.
The team — led by geneticist George Church of Harvard Medical School and the Wyss Institute, with Antonio del Sol's group at CIC bioGUNE in Spain — first trained a machine-learning model to predict a skin cell's 'transcriptomic age' from its gene activity. Then they used that age-reader as a scoreboard: in aged human skin cells, they switched on candidate genes one at a time — 89 candidates in a 95-construct screen — to see which ones made old cells score young again.
The Winner: A Humble Editing Gene
SRSF1 topped the screen. It is an RNA splicing factor — part of the machinery that edits genetic messages before cells build proteins. Boosting it in old skin cells reduced senescence (cells retiring and causing trouble), protein-handling stress, and inflammatory signalling. In one donor's cells, it cut the predicted age by a median of 16.2 years. Crucially, the cells stayed skin cells — unlike the famous Yamanaka reprogramming factors, which can erase a cell's identity.
Delivered into aged mice with a virus that mostly targeted the liver, and switched on in a controlled way, SRSF1 slowed frailty progression by 44% and was linked to a 14% extension of median lifespan.
The Honest Part
This is early, lab-stage science — not a treatment. The mouse lifespan groups were tiny, just 9–10 animals each. Nothing here has been tested in people, and the details matter enormously: leaving the gene switched on constantly did not improve skin or muscle repair in mice. Timing and dose made the difference.
The Twikup Take
The real breakthrough may be the platform itself: a reliable way to measure cell age, then screen gene after gene for ones that reverse it. SRSF1 is the first headline catch. Expect more — but don't expect a fountain-of-youth pill anytime soon.
Image: AI-generated illustration.