Aging is characterized by progressive functional decline and is hypothesized to reflect accumulation of epigenetic noise and loss of epigenetic information. While DNA methylation clocks are commonly used to estimate biological age, they are often difficult to interpret in terms of gene regulatory networks. Chromatin accessibility measured by ATAC-seq provides a complementary epigenetic readout more directly linked to regulatory elements and transcriptional control. The authors reasoned that building aging clocks from chromatin accessibility at the level of individual cell types could capture cell type–specific aging signals that are obscured in bulk measurements.
The study used single-nucleus ATAC-seq (snATAC-seq) profiles from the prefrontal cortex (PFC) of human donors. The cohort included 357 donors ranging in chronological age from 15 to 100 years. The authors derived pseudobulk accessibility profiles for individual cell types from the single-nucleus data in order to train cell type–resolved aging models.
Using pseudobulk profiles generated from snATAC-seq, the authors trained a set of aging clocks at two granularities: cell type–specific clocks and an all-cell clock that aggregates across cell types. These models predict chronological age from chromatin accessibility features and are designed to be interpretable with respect to regulatory elements, gene loci, and transcription factor motifs associated with aging signals.
The PFC-derived clocks were validated for generalizability. According to the authors, models trained on PFC snATAC-seq generalized to accurately predict age across different brain regions and across species. This cross-context performance supports the view that the accessibility features learned by the clocks capture conserved aspects of epigenetic aging that extend beyond the original training tissue.
Beyond predicting chronological age, the PFC clocks were applied to perturbation experiments. The clocks detected the rejuvenating effect of SIRT6 overexpression in mouse liver, indicating sensitivity to interventions that alter biological aging at the chromatin accessibility level. The authors additionally linked the SIRT6-related signal to regulatory changes: specifically, repression of the NF-kB pathway was observed in SIRT6 transgenic mice.
The PFC clocks were used to assess epigenetic age acceleration in neurodegenerative disease. The models identified cell type–specific age acceleration in Alzheimer’s disease (AD) and Parkinson’s disease. Among major brain cell types, microglial age acceleration correlated most strongly with measures of neuropathology. Sex differences emerged in certain glial populations: female oligodendrocytes and oligodendrocyte precursor cells (OPCs) displayed the largest sex-specific increases in age acceleration.
A major strength of chromatin accessibility clocks is interpretability. The authors analyzed the features driving clock predictions to reveal candidate regulatory elements, genes, signaling pathways, and transcription factor motifs associated with aging and disease signals. In disease, severe AD was associated with upregulation of immune and inflammatory pathway signals at the accessibility level. Across regions and species, the clocks identified conserved age-predictive accessibility peaks linked to histone regulation, metabolic processes, and neuronal survival pathways.
The authors present PFC snATAC-seq–based aging clocks that are cell type–resolved and interpretable. These clocks accurately predict chronological age, generalize across brain regions and species, detect rejuvenation effects from genetic perturbation (SIRT6 overexpression), and reveal cell type–specific age acceleration in neurodegenerative disease, with notable microglial and glial sex-differential signals. Interpreting accessibility features connects clock outputs to regulatory mechanisms, including NF-kB repression with SIRT6 and immune/inflammatory pathway activation in severe AD. The work positions cell type–specific chromatin accessibility clocks as a tool to evaluate interventions, characterize disease-related epigenetic aging, and generate mechanistic hypotheses about the regulatory bases of aging and neurodegeneration.