(1062-C) Pooled image-based phenotypic compound screening in human cells enabled by Ghost Cytometry with DNA barcoding technology
Tuesday, February 6, 2024
12:00 PM – 1:00 PM EST
Location: Exhibit Halls AB
Abstract: Although target-based screening has been extensively used in the pharmaceutical companies, phenotypic screening is an emerging method to screen for first-in-class drugs. One possible reason is that phenotypic screening based on optical cell images is an unbiased, powerful approach in search of genes and compounds that induce physiologically relevant disease cellular phenotypes. The microscopy-based screening platform is well-established, common, and popular for High-content imaging. However, the current microscopy-based phenotypic screening approaches still can be inflexible. For example, cell staining requires multiple fluorescence markers with a combination of image processing and analysis. On the other hand, a pooled phenotypic screening platform is potentially high throughput, cost-effective, and smaller well-to-well batch effect. However, existing conventional flow cytometers can only detect total fluorescence intensity or cell size, thus it is difficult to obtain high content cell information. Here, we introduce a new approach for pooled phenotypic screening by combining ThinkCyte’s AI-enabled cell sorter (VisonSort TM) powered by Ghost Cytometry® (GC) technology with DNA barcoding technology for connecting the information of cell phenotypes and compounds. Herein, VisionSort has been recently developed based on our fluorescence and label-free machine learning-driven flow cytometry approach that analyzes cell image information without image production. For validation of DNA barcoding technology, we first performed a proof-of-concept experiment focusing on one fluorescent cellular phenotype; NFkB nuclear translocalization in THP1 cells. After incubation of each compound and corresponding DNA barcode in each 96-well, cells were pooled, and applied to VisionSort. The cells exhibiting inhibition of nuclear translocalization were collected and sequenced the corresponding barcodes. We confirmed that the sorted cells exhibiting inhibition of nuclear translocalization had specific DNA barcodes corresponding to positive control compounds. Next we conducted compound screening to identify novel small molecules against nonalcoholic steatohepatitis (NASH) using a selected compound library. NASH is a progressive and severe liver disease, characterized by lipid accumulation, inflammation, and downstream fibrosis. To evaluate our new technology, we performed two different types of screening assays; one is based on BODIPY fluorescence intensity to detect NASH phenotype, lipid accumulation in cells using microscopy and the other is based on label-free cell morphological information using VisionSort. After induction of NASH phenotype in HepG2 cells, the lipid accumulation was statistically detected by BODIPY intensity. We tested the capability of VisionSort to classify phenotypic changes of control and NASH phenotype using label-free high content imaging and found that the machine learning models exhibited high performances. Using the trained classifier, we conducted compound screening with DNA barcode at label-free. In this meeting, by comparing the conventional fluorescence-based screening results, we will discuss the novelty and efficacy of label-free screening and new potential target drugs obtained from our new screening platform.