Scalable quantification of dynamic subcellular spatial organization in single cells across tissues

The polarized organization of cellular constituents is vital for cell migration, fate decisions, and tissue organization and often altered in disease. However, its quantification remains challenging because cells vary in size, shape, marker expression, and change over time. As existing approaches lack throughput, discard spatial information, and cannot measure polarity dynamics, we introduce CellPolariS, a novel image analysis framework to quantify polarity in diverse cell types from different tissues and in living cells. Unlike other methods, CellPolariS uses spatial subcellular organization to quantify the number, direction, magnitude, and concentration of cellular structures. We identify that differences in cell morphology and marker expression between cells are critical confounding factors that impair polarity quantifications. Through extensive validation and quantification of diverse cell types, including T-cells, natural killer cells, zygotes, neurons, yeast, and bacteria, we demonstrate that CellPolariS corrects for these effects and provides precise measurements of the polarization of Tubulin, Actin, and other structures. In addition, we analyze hundreds of thousands of primary stem and progenitor cells, map how Tubulin, CDC42, and Actin polarity change throughout hematopoietic differentiation, and discover that polarity increases during erythroid differentiation. Using long-term imaging of living cells, we further show that polarized subcellular structures are highly dynamic and can quickly change from polar to non-polar states, suggesting that cell polarity is more dynamic than previously appreciated.