Abstract: Phenotypic drug discovery (PDD) is a powerful approach for identifying new drug targets and elucidating disease mechanisms. Current approaches for PDD mainly rely on either microscopy-based or flow cytometry-based technologies for analyzing readouts. Microscopy-based screening platforms are well established, but are limited by their low throughput and resource-intensive infrastructure. On the other hand, conventional flow-based screening platforms allow for higher efficiencies and large scale screening, but only work for a limited number of simple phenotypic readouts with defined markers. For many diseases, including cancer, infectious diseases, neurodegenerative diseases, etc., complex phenotypes that involve morphological transformation or intracellular dynamics are important readouts for drug screening.
Here we introduce a new approach of flow-based high content phenotypic screening by combining ThinkCyte’s AI-enabled Ghost Cytometry® (GC) with PhoreMost’s PROTEINi® technology, targeting intracellular protein aggregation as the target readout. GC is a novel flow cytometry technology that enables analysis and sorting of cells based on both label-free morphological features and conventional fluorescence parameters with high intracellular spatial resolution. PROTEINi is capable of probing the entire proteome to systematically unmask new and unanticipated druggable sites, directly linking them to useful therapeutic functions. When used together, GC and PROTEINi have the potential to rapidly screen for a range of new, complex phenotypes and unmask truly novel drug targets for treating diseases.
In this study, we prepared an MCF10a cell line model that expressed Dox-inducible aggregation of GFP-tagged proteins. Cells were labeled with unique DNA barcodes based on their phenotypes, either intracellular proteins that are diffused or aggregated, and were mixed together for analysis via GC. To identify and classify the different phenotypes, we captured the intracellular spatial distribution of the GFP signals as an optical waveform and performed dimensionality reduction using UMAP to visualize and identify unique clusters represented by the phenotypes. 20 million cells were analyzed and sorted in 2 hours based on their UMAP clustering with maintenance of cell viability and integrity. Downstream deep sequencing analysis validated that the enriched cells indeed contained DNA barcodes for the aggregated phenotype. This study demonstrates the utility of GC for high content phenotypic screening in flow and can be broadly applied for innovations in drug discovery.