An unbiased chemical screen of actin organisation enables drug discovery and mechanistic analyses of the cytoskeleton
School of Medical Sciences, UNSW Australia, Sydney, NSW, Australia.
The actin cytoskeleton is a highly organised and dynamic system that processes both external and internal signalling cues into changes in the structure of cellular architecture. Dysregulation of actin organisation is seen in many tumour types and disease processes, however actin itself is not a clinical drug target, due to cardiotoxic side-effects. Actin binding proteins are a diverse group of proteins that have distinct functions and tissue and cellular expression profiles, and as such present as more attractive therapeutic targets. This project aims to identify compounds that impact on the organisation of the actin cytoskeleton either through signalling pathways or direct actin binding proteins. We screened 114,400 diverse compounds in an image-based screen using F-actin organisation as a readout, quantifying 75 features per condition and defining >2600 unknown compounds impacting actin organisation. Unbiased clustering of responses delineated 27 distinct actin phenotypes. This suggests that there are a discrete number of ways that polymeric actin can be reorganised in a mammalian cell. We hypothesized that similar actin phenotypes may have similar drug targets, establishing 3 strategies to accelerate mechanism discovery: 1) Comparison with known actin compounds used as positive controls in the screen (>260 compounds identified); 2) Comparison with known genetic changes (3 talin-binding compounds defined), and; 3) Retrospective integration (RetInt) of additional positive controls into the original screen data (2 ROCK inhibitors defined). The RetInt strategy may enhance accelerated drug discovery by providing near-unlimited capacity to extend mechanistic analyses in this and other phenotypic screens.