Fig. 4. Senescence biomarker intensities reflect cell-cycle histories and can resolve predicted-senescent cells with varying levels of accuracy.

a-e MCF10A cells expressing the CDK2 activity sensor were treated with 10 μM etoposide for 24 h, washed, and subjected to time-lapse microscopy from 6-10 d after release. The cells were fixed and stained for SA-β-Gal, LAMP1, succinimidyl ester, Hoechst, IL8, 53BP1, p21, and Lamin B1 after the last frame was taken. b, c Cells were split into fast cycling, slow cycling, or predicted senescent based on their duration spent CDK2^low during live cell imaging, and the intensity of each marker was plotted for each cellular behavior. d The duration cells spent in the CDK2^low state was plotted against the bottom, middle, and top 10% of signal for each marker. e ROC analysis using markers to identify predicted-senescent cells amidst all cells (fast-cycling, slow-cycling, and predicted-senescent cells), or amidst only slow-cycling and predicted-senescent cells. AUC indicates the area under the curve for each condition.