Linking Phenotypes and Modes of Action Through High-Content Screen Fingerprints

Abstract

High-content screening (HCS) is a powerful technique for monitoring phenotypic responses to treatments on a cellular and subcellular level. Cellular phenotypes can be characterized by multivariate image readouts such as shape, intensity, or texture. The corresponding feature vectors can thus be defined as HCS fingerprints that serve as a powerful biological compound descriptor. Therefore, clustering or classification of HCS fingerprints across compound treatments allows for the identification of similarities in protein targets or pathways. We developed an HCS-based profiling panel that serves as basis for characterizing the mode of action of compounds. This panel measures phenotypic effects in six different compartments of U-2OS cells, namely the nucleus, the cytoplasm, the endoplasmic reticulum, the Golgi apparatus, and the cytoskeleton. We profiled a set of 2,725 well-annotated compounds and clustered their corresponding HCS fingerprints to establish links between predominant cellular phenotypes and cellular processes and protein targets. We found various different clusters enriched for individual targets (e.g., HDAC, HSP90, TOP1, HMGCR, TUB), signaling pathways (e.g., PIK3/AKT/mTOR), or gene sets associated with diseases (e.g., psoriasis, leukemia). Based on this clustering we were able to identify novel compound-target associations for selected compounds such as a submicromolar inhibitory activity of Silmitasertib (a casein kinase inhibitor) on PI3K and mTOR.

Tales of 1,008 Small Molecules: Phenomic Profiling through Live-cell Imaging in a Panel of Reporter Cell Lines

“…Our panel of cell lines expressed 12 different fluorescently-tagged proteins from their endogenous chromosomal loci, enabling us to monitor the morphology of different organelles and the activity of various signalling pathways in three distinct cell types. We observed that 57% of reference compounds were phenotypically active in ≥1 reporter line, which is comparable to previous image-based profiling screens [26]. There was a clear trend for compounds co-annotated to the same MoA to be either all active or all inactive in any given reporter line.…”

Cell Painting, a high-content image-based assay for morphological profiling using multiplexed fluorescent dyes

“…Morphological profiling has successfully been used to characterize genes and compounds in a number of studies. For instance, morphological profiling of chemical compounds has been used to determine their mechanism of action 7,14–18 , identify their targets 19,20 , discover relationships with genes 20,21 , and characterize cellular heterogeneity 22 . Genes have been analyzed by creating profiles of cell populations where the gene is perturbed by RNA interference (RNAi), which in turn have been used to cluster genes 23,24 , identify genetic interactions 25–27 , or characterize cellular heterogeneity 28 .…”

Development of the Theta Comparative Cell Scoring Method to Quantify Diverse Phenotypic Responses Between Distinct Cell Types

“…18,20–25,27 However, the majority of these methods have been applied to single cell types amenable to high-content imaging or large-scale biochemical and proteomic analysis. 18,21–25,27 The TCCS method described in this article was developed to provide a practical method to enable comparative multiparametric phenotypic analysis across a panel of genetically distinct cell types, which provides rapid quantification and visualization of divergent compound-induced phenotypic response between cell types. An intuitive explanation of the TCCS method would be the cosine distance in degrees of vectors in the first two PCs; this is a variation on existing methods that largely rely on correlation or Euclidean distance between compound vectors.…”