Single-Cell Imaging Exposes Senescence Complexity and Rapamycin's Selective Power

Cellular senescence — the state of permanent cell cycle arrest associated with aging and disease — is notoriously difficult to measure because no single biomarker reliably identifies all senescent cells. This study tackled that challenge head-on by applying single-cell fluorescence imaging to quantify multiple senescence markers simultaneously, revealing a level of biological complexity that bulk measurements routinely miss.

The research team induced senescence in primary human dermal fibroblasts (HDFs) using two distinct methods: mitomycin C (MMC), a DNA-damaging chemotherapy agent, and D-galactose, which generates oxidative stress. Senescence was confirmed through EdU incorporation assays (showing proliferative arrest), SA-βgal enzymatic activity, immunofluorescence for p21 and IL-6, and morphological measurements (nuclear area, cell area) using high-content imaging platforms including the Opera Phenix Plus and IN-Cell Analyser 2200.

The most striking finding was profound single-cell heterogeneity in SA-βgal activity. Even within the same senescent culture, cells displayed a wide spread of SA-βgal fluorescence intensities, forming distinct subpopulations rather than a uniform senescent state. Nuclear and cell area enlargement — classic morphological hallmarks — were more consistently elevated across the senescent population and showed similar variability patterns to each other, making them robust complementary indicators. Importantly, specific nuclear area subpopulations correlated strongly with IL-6 expression levels, suggesting that the morphological state of a cell is linked to its inflammatory secretory activity (SASP), a connection with significant implications for understanding how senescent cells harm surrounding tissue.

To better capture this heterogeneity in practice, the authors introduced an 'induction threshold' method. Rather than comparing population averages, this approach defines a threshold based on the distribution of biomarker values in control (non-senescent) cells and quantifies what fraction of treated cells exceed that threshold. This enables more accurate and reproducible identification of genuinely senescent cells across experiments.

The study's most therapeutically significant finding concerned rapamycin, an mTOR inhibitor widely studied as a senomorphic (senescence-suppressing) agent. When senescent HDFs were treated with 500 nM rapamycin, it did not uniformly reduce senescence markers across the entire population. Instead, it selectively reduced marker expression in specific subpopulations — particularly those with high SA-βgal activity or large nuclear areas — while leaving other subpopulations largely unaffected. This subpopulation selectivity was observed in both the MMC and D-galactose models. By contrast, the senolytic drug ABT-263, which kills senescent cells via apoptosis, effectively eliminated senescent cells across all subpopulations regardless of biomarker heterogeneity. These contrasting profiles suggest that rapamycin and senolytics act through fundamentally different mechanisms and may be complementary rather than interchangeable therapeutic strategies.

These findings carry important implications for how senomorphic drugs are evaluated in preclinical and clinical settings. If rapamycin only targets certain senescent subpopulations, studies relying on population-average readouts may underestimate or mischaracterize its effects. Single-cell resolution analysis, as demonstrated here, provides a more nuanced and accurate picture of drug action.