The Human Senescence Atlas Reveals That Aging Cells Come in 15 Flavors
Aging is driven, in part, by cells that stop dividing but refuse to die. These senescent cells accumulate in tissues over a lifetime, secreting inflammatory signals that damage surrounding tissue and contribute to diseases from arthritis to Alzheimer’s to cancer. For years, the field has searched for a universal marker, a single molecular flag that identifies every senescent cell, regardless of where it sits in the body.
A landmark study published June 11 as the cover article of Cell suggests that goal may be fundamentally misguided. Senescence is not a single cellular state. It is a spectrum.
The paper, led by researchers at Yale University, Stanford University, and the NIH SenNet Consortium, presents the Human Senescence Atlas, a comprehensive map of senescent cells across 18 human tissues, integrating single-cell transcriptomics, spatial proteomics, and AI-driven classification. Rather than identifying a single senescence program, the atlas reveals 15 distinct “senotypes”, senescence subtypes with unique molecular signatures that vary by tissue, cell type, and physiological context.
“Senescence is not a single state,” said Dr. Rong Fan, the Harold Hodgkinson Professor of Biomedical Engineering at Yale and the study’s corresponding author. “We found that the molecular profile of a senescent cell depends on where it lives, what type of cell it is, and what triggered it to become senescent in the first place.”
The study is the flagship output of the NIH SenNet Consortium, a multi-institutional effort to map cellular senescence across the human lifespan. The research team, including first authors Vidyani Suryadevara (Stanford), Negin Farzad (Yale), and Mingyu Yang (Yale), assembled tissue samples from healthy donors spanning a wide age range, plus diseased tissue from patients with fibrotic liver disease and chronic skin wounds.
For each tissue, the team applied single-cell RNA sequencing, spatial transcriptomics (using technologies including Seq-Scope and Pixel-Seq developed within the consortium), and proteomic profiling. Machine learning algorithms classified senescent cells by their combined molecular features, not just a handful of established senescence markers, but the full transcriptomic and proteomic landscape.
The 15 Senotypes
The atlas identifies 15 distinct senotypes, each characterized by a specific combination of markers, secretory profiles, and microenvironmental interactions. Key findings include:
- No universal marker exists. Classical senescence markers like p16^INK4a, SA-β-gal, and the SASP (senescence-associated secretory phenotype) are present in some but not all senotypes. Relying on any single marker would miss the majority of senescent cells in most tissues.
- Tissue-specific programs dominate. Senescent cells in the lung look molecularly different from senescent cells in the liver. Senescent neurons in the prefrontal cortex share almost no markers with senescent fibroblasts in the skin.
- Immunosenescence hotspots. The atlas reveals localized clusters of dysfunctional B cells in aging lymph nodes, a previously uncharacterized form of immune system aging at the cellular level.
- Disease-specific signatures. Senescent cells in fibrotic liver tissue express a distinct molecular program not seen in healthy aging liver, suggesting that disease-associated senescence is a separate biological category, not just accelerated aging.
- Microenvironmental feedback. The spatial data reveal that senescent cells remodel their surrounding tissue, and the tissue in turn shapes the senescent state, a bidirectional relationship that has been difficult to study in culture dishes.
Why It Matters
The absence of a universal senescence marker has been a practical problem for the field. Drug companies developing senolytic therapies, drugs that selectively kill senescent cells, have struggled with patient selection: if you cannot reliably measure senescence in a patient’s tissue, you cannot tell whether your drug is working, or even whether the patient has the condition you are treating.
The atlas addresses this by providing machine-learning-derived molecular signatures that can recognize senotypes across tissues, datasets, and species. These signatures could serve as the basis for clinical biomarkers, blood tests or tissue biopsies that measure senescent cell burden and classify the specific type of senescence present.
“This sets the foundation for senolytic drug development targeting specific senotypes,” the authors note, rather than treating senescence as a monolithic target.
The Companion Study
A companion paper published simultaneously in Molecular Cell, the SenCat study, profiled 14 primary human cell types across more than 30 senescence induction paradigms, examining how different triggers (DNA damage, oncogene activation, replicative exhaustion) produce different senescent states in the same cell type. Together, the two papers provide the most comprehensive molecular portrait of human cellular senescence to date.
The atlas is being released as an open-access community resource, available for researchers to query by cell type, tissue, disease state, or molecular marker.
Source: Suryadevara, V., Farzad, N., Yang, M., Fan, R. et al. (2026). “Charting human cellular senescence in aging and disease.” Cell, 189(12). DOI: 10.1016/j.cell.2026.05.028.