An image based high throughput screen to identify regulators of Imp containing RNP granules

In vivo, RNAs and proteins are frequently packaged into diverse dynamic macromolecular structures named mRNP granules. These assemblies form upon phase separation of individual RNA and protein components, a process involving the establishment of multivalent weak interactions and their regulations via post-translational modifications. Defects in their properties have been associated with several human pathologies. However, our knowledge of these dynamic structures relies essentially on the study of P bodies and stress granules. We are interested in the highly conserved RNA binding protein Imp whose mammalian counterpart's overexpression correlates with poor prognosis in several cancers. In vivo, Imp is present in cytoplasmic RNP granules, distinct from P-bodies and visible both in neuronal cell bodies and axons. They are also detected in Drosophila S2R + cultured cells. Taking advantage of this cellular model, we have undertaken a genome-wide RNAi-based visual screen to identify factors that regulate the properties of Imp-containing granules. This implies combining high throughput microscopy with the development of a computational pipeline for automatic image analysis. This pipeline first segments and discriminates healthy from dead nuclei, storing this information in an interactive SQLite database that enables experimental quality control. Then, GFP-Imp granules are detected using the SPADE algorithm in the cytoplasm of healthy cells. Data from the pilot screen we have performed to validate the experimental design and develop our pipeline for data mining are presented.