(1056-A) Live Cell Painting: Drug Responses in Human Primary Patient Cells with a New Nontoxic Dye
Monday, February 5, 2024
12:00 PM – 1:00 PM EST
Location: Exhibit Halls AB
Abstract: Image-based profiling has proven to be useful in generating phenotypic profiles that help identify healthy and diseased cellular states, or even predict drug mechanism of action [1]. With improvements in bio-image analysis methods and in the acquisition speed of microscopy images, image-based profiling can achieve high throughputs that make it useful for drug discovery, especially when compared against other phenotypic profiling techniques such as gene or protein expression methods. The cell painting method has been popularized for generating such image profiles. It uses multi-channel fluorescence microscopy images acquired after adding multiple different stains on fixed cells [2]. However, cell painting requires multiple sample processing steps that can result in selective loss of cells, as well as other sample alterations associated with permeabilization and fixation. Moreover, as an end-point assay, important kinetic information could be missed, which may necessitate assaying independent samples at different time points. The uneven loss of cells and use of independent samples for different timepoints are particularly challenging for drug discovery applications, where scale and replicability matter, but also for primary patient cells, which are inherently heterogeneous samples that may have differing drug responses, and which are often limited in supply. Here, we present what is to our knowledge a first “live cell painting” compound screening study. A novel mix-and-read and non-toxic dye – which generates unique phenotypic fingerprints consistent with cellular phenotypes – was used to identify optimal treatment regime for a late-stage prostate cancer patient. These features were especially useful when performing a version of this screen in 3D cultures, where even staining of 3D samples was achieved, a feature that is not possible with alternative fluorescent dyes. Further image analysis using biologist-friendly AI tools was performed on the samples to successfully identify clusters of compounds with similar drug mechanisms of action.