AURORA: a longitudinal single-cell framework for resolving autophagy-associated remodeling under nutrient stress

Abstract

Nutrient stress induces dynamic and heterogeneous cellular responses that are often reduced to population averages or endpoint measurements. AURORA is a longitudinal single-cell phenotyping framework integrating high-content time-lapse imaging, trajectory-aware quality control, multi-compartment morphology, reference phenotypic states, and cross-cell-line aggregation of state-occupancy behaviors. AURORA is applied to MCF-7, MDA-MB-231, A-549, and HT-1080 cancer cells during 4 h exposure to Earle's balanced salt solution (EBSS). Population-level Total Autophagy shows sustained EBSS-associated elevation in MCF-7, A-549, and HT-1080, but a stronger early response followed by attenuation in MDA-MB-231. Longitudinal analysis shows that these averages arise from cell-line- and time-dependent changes in single-cell distributions rather than uniform displacement. Feature-level analysis further identifies distinct combinations of nuclear, whole-cell, and autophagosome-associated organization, compactness, and shape in each model. Projection of EBSS-treated cells into Complete-defined phenotypic landscapes reveals line-specific redistribution among reference states. Despite this specificity, dominant state-occupancy signatures converge into six recurrent higher-order programs, linking different cellular routes to partially shared remodeling architectures. Because DAPGreen is not paired with a flux inhibitor or ratiometric reporter, these findings describe an autophagy-associated imaging phenotype rather than autophagic flux. AURORA therefore distinguishes population-average change from heterogeneous single-cell adaptation during dynamic stress responses.