AI Longevity Daily

Today’s Digest

A Harvard team found that overexpressing one regulatory protein reduced frailty in aged mice and extended median lifespan.
An oral Parkinson's disease drug completed its first human dosing in a UK early-stage trial, aiming to validate biological signals rather than improve symptoms.
Whether mouse study results apply to humans remains unknown, and current human trials have not yet assessed actual symptom improvement.

🔥 Top 8 Highlights

1. Single Gene Overexpression Cuts Frailty by 44% and Extends Median Lifespan by 14% in Mice

SRSF1, an RNA splicing factor—a regulatory protein that handles “splicing” when cells read genetic information—has been identified by a Harvard Medical School research team as a potential rejuvenation target, published in Nature Aging. The researchers first trained a machine learning aging clock based on gene expression in human skin cells aged 1 to 96 years, used it to screen a batch of genes related to the human aging network, and found that SRSF1 overexpression rivals the effects of the most famous reprogramming cocktail (OSKM) while avoiding the loss of cellular identity that plagues the latter. In aged mice, viral vector–induced SRSF1 expression reduced frailty by 44% and extended median lifespan by 14% in groups of 9–10 female mice each. Currently, the viral vector targets only the liver, limiting systemic effects, and precise control of dosage and timing remains under investigation; the path from mice to human therapy still has a long way to go. Further details on trial design and molecular mechanisms can be found in Lifespan.io’s coverage .

2. First Parkinson’s Patient Receives MTX325 Dosing, Targeting Mitochondrial Clearance

Mission Therapeutics announced the first dosing in a Phase 1b clinical trial for Parkinson’s disease at Liverpool Royal University Hospital in the UK. MTX325 is an oral, blood-brain barrier–penetrating USP30 inhibitor; USP30 acts as a “brake” that prevents cells from clearing damaged mitochondria—the cell’s “power plants”—and inhibiting it promotes the removal of damaged mitochondria. The trial plans to enroll 36 mild-to-moderate patients across the UK, Europe, and Australia, aiming to verify whether the drug produces biological changes in humans consistent with animal experiments, not directly assessing symptom improvement; data readout is expected by the end of 2027. Early animal experiments and Phase 1a studies in healthy volunteers have suggested the drug can enter brain tissue, but there’s still a long way from safety signals to Parkinson’s symptom improvement. Details can be found in Longevity.Technology’s coverage .

3. AI Model Identifies 351 Disease-Associated Signals from Routine CT Scans

Percival, a vision-language model—a type of AI that simultaneously understands images and text—was developed by a University of Pennsylvania team and published in npj Digital Medicine. Trained on paired CT scans and radiology reports from over 52,000 participants totaling more than 400,000 scans, the model found statistical associations between digital features in scan images and 351 diagnoses, and its predictions for future onset of heart failure, cirrhosis, and acute kidney injury outperformed models using only age, sex, or organ measurements. However, the study also notes that correlation between CT image features and these diseases does not equal accurate measurement of blood markers, and some “predicted” future diagnoses may be diseases already present at imaging time but not yet documented; the study also did not directly compare with radiologists. More evaluation details can be found in Lifespan.io’s coverage .

4. Programmable Cell Deletion Circuit Boosts Vascular Retention in Organoids

SynBITE, a genetic circuit built by a University of Pittsburgh team and published in Cell Stem Cell, can clear unwanted immature cells in situ during organoid development—laboratory-grown miniature organ mimics—without needing to disassemble and reassemble the tissue. This addresses a silent cost: traditional purification methods destroy already-formed vascular networks, whereas in-situ deletion retains about 23% more vessel length and about 122% more vascular nodes. Combined with CRISPR activation, treated liver organoids reached about 60% of primary hepatocyte drug metabolism activity, compared to about 20% for unapoptosed controls. Researchers also note that similar subtraction logic could potentially be used to clear senescent cells in living tissue in the future, though this remains conceptual; the circuit’s core components are borro