Single-cell analysis of sterol-induced Ca^2+^ signaling in human astrocytes by dynamic mode decomposition

Abstract

Ca^2+^ signaling in astrocytes is a central mechanism of intercellular communication in the brain and plays a key role in regulating neuronal excitability, synaptic plasticity, and energy metabolism. Disruption of astrocytic Ca^2+^ dynamics is a characteristic of neurodegenerative diseases, as are deviations in cholesterol trafficking and metabolism, which are essential for maintaining membrane structure and function. Although recent studies have begun to explore links between Ca^2+^ signaling and sterol homeostasis in astrocytes, unbiased analytical workflows and mechanistic insight into how cholesterol and related sterols regulate astrocytic Ca^2+^ dynamics remain limited. Here, we apply dynamic mode decomposition to dissect and classify Ca^2+^ signals obtained from time-lapse imaging of human astrocytes. Using both synthetic and experimental datasets, we show that delay-embedded dynamic mode decomposition combined with clustering separates heterogeneous Ca^2+^ activity into distinct dynamical states. This analysis reveals that increasing cholesterol levels shift astrocytes toward more active oscillatory states, whereas acute cholesterol depletion suppresses Ca^2+^ activity. In addition, pretreatment with the oxysterols 24-, 25-, and 27-hydroxycholesterol impaired cholesterolinduced Ca^2+^ oscillations. Together, this work presents a general computational framework for decomposing and analyzing complex spatiotemporal Ca^2+^ signals, with broad applicability to quantitative imaging in cell biology.