Image-based temporal profiling of autophagy-related phenotypes.

Abstract

Autophagy is a dynamic process critical in maintaining cellular homoeostasis. Dysregulation of autophagy is linked to many diseases and is emerging as a promising therapeutic target. High-throughput methods to characterise autophagy are essential for accelerating drug discovery and characterising mechanisms of action. In this study, we developed a scalable image-based temporal profiling approach to characterise ~900 morphological features at a single cell level with high temporal resolution. We differentiated drug treatments based on morphological profiles using a random forest classifier with ~90% accuracy and identified the key features that govern classification. Additionally, temporal morphological profiles accurately predicted biologically relevant changes in autophagy after perturbation, such as total cargo degraded. Therefore, this study acts as proof-of-principle for using image-based temporal profiling to differentiate autophagy perturbations in a high-throughput manner and has the potential identify biologically relevant autophagy phenotypes. Ultimately, approaches like image-based temporal profiling can accelerate drug discovery.