Single Transcription Factor Switch Reverses Aging in Cells and Mouse Livers

Aging at the cellular level is accompanied by sweeping changes in gene expression, and reversing those changes — without triggering dedifferentiation or cancer risk — represents one of the most promising frontiers in longevity medicine. Partial reprogramming with Yamanaka factors has shown proof of concept, but the approach carries risks and only a handful of alternative transcription factor (TF) perturbations have ever been described. A UCSF team led by Deng, Villeda, and Li set out to change that systematically, developing a platform they call the Transcriptional Rejuvenation Discovery Platform (TRDP) and publishing results in PNAS in January 2026.

TRDP begins with bulk RNA sequencing comparing old and young cell states, then uses bioinformatic tools to identify which TFs most likely regulate the genes differentially expressed between those states. The platform prioritizes TFs whose perturbation would be predicted to shift gene expression back toward the youthful state. The team applied this to passaged human neonatal dermal fibroblasts — an established replicative aging model — defining early passage (0–20 population doublings, PD), middle passage (21–30 PD), and late passage (>31 PD) as proxies for young, middle-aged, and old cells. From the computational pipeline, 200 candidate TFs were selected for CRISPRa (activation) and CRISPRi (inhibition) Perturb-seq screening — 400 total perturbations — in late-passage fibroblasts, followed by single-cell RNA sequencing.

Rejuvenation was quantified by R_rej, defined as the correlation between the gene expression fold-change of late versus early passage cells and the fold-change induced by each TF perturbation. Perturbations with significantly negative R_rej values reversed the aging transcriptional signature. More than a dozen TFs met the threshold (R_rej ≤ −0.3), including DLX6 (−0.57), E2F3 (−0.53), FOXM1 (−0.47), EZH2 (−0.36) via CRISPRa, and EGR1 (−0.55), ZFX (−0.51), ATF4 (−0.48) via CRISPRi. Four candidates — E2F3 overexpression, EZH2 overexpression, STAT3 repression, and ZFX repression — were validated through extensive phenotyping. All four increased KI67+ proliferating cells and proteasome activity, decreased p21/CDKN1A and TIMP1/TIMP2 senescence markers, enhanced mitochondrial membrane potential (TMRE staining), and reduced lysosomal accumulation — a full reversal of the classic cellular aging hallmark panel. Importantly, these effects were distinct from Yamanaka factor overexpression, which caused abnormal phenotypes, and none of the four validated TF perturbations upregulated cancer-progression genes.

A key mechanistic insight emerged from SCENIC transcription factor module analysis: all four validated perturbations converged on a shared downstream transcriptional program, despite acting through different primary mechanisms. This program was conserved across species — it matched the gene expression signature seen in young versus old mouse tissues and in aged mice rejuvenated by heterochronic parabiosis, spanning multiple tissues and cell types from published datasets. This cross-species, cross-tissue conservation strongly implies a universal molecular logic underlying rejuvenation.

The most striking in vivo result came from EZH2 overexpression in aged mice via AAV liver delivery. EZH2-treated aged livers showed reversal of aging-associated gene expression profiles, significant reductions in hepatic steatosis (fat accumulation) and fibrosis, and improved glucose tolerance in metabolic challenge tests. This demonstrates that a single TF perturbation is sufficient to produce meaningful tissue-level rejuvenation in a living aged organism — without requiring a cocktail of factors. The authors note that EZH2 is a histone methyltransferase (PRC2 component) that broadly silences gene expression through H3K27 methylation, providing a potential epigenetic mechanism for the broad transcriptional reset observed. These results substantially expand the toolkit of candidate rejuvenating TFs available for future translational development.