Cytoskeletal organization and nuclear morphology are reliable biomarkers of TGF- <b> <i>β</i> </b> induced fibroblast activation

The cytoskeleton is a hierarchical network of biopolymers essential for key cellular physiological and pathological functions, as well as for adaptation to different stimuli. Distinct cytoskeletal phenotypes are associated with numerous pathological states and can change upon drug treatments, underscoring their potential as biomarkers. Here, we present a quantitative approach based on cytoskeletal and nuclear morphologies to classify cell populations at the single-cell level. We validate this strategy by identifying fibroblast subpopulations in response to transforming growth factor β1 (TGF-β1). Using immunofluorescence images of sparsely plated fibroblasts labeled for phalloidin, α-smooth muscle actin (α-SMA), and DAPI, we analyzed cells with our custom-built CSKmorphometrics pipeline to extract multiple descriptors of cytoskeletal and nuclear morphology. From these, we identified nine descriptors that predict fibroblast activation with 92% accuracy and 82% kappa metrics. Predictions were confirmed through α-SMA expression analysis. Importantly, our approach revealed two distinct fibroblast subpopulations: one strongly responsive to TGF-β1, displaying significant increases in activation levels, and another mildly responsive. We also identified specific descriptors of cytoskeletal organization that anticipate the degree of individual cell responsiveness to stimulation. Overall, this study highlights the potential of cytoskeleton quantification to generate robust biomarkers that enhance diagnosis, prognosis, and disease stratification. Moreover, our framework provides high-content, single-cell data suitable for drug discovery and for understanding cellular heterogeneity across diverse pathophysiological states.