Single identical cell cytotoxicity assay on lithographically generated microstructures

Abstract

Existing cytotoxicity assays based on cell ensembles suffer from cell heterogeneity, where each cell has a unique character and responds differently to therapeutic stimuli. Among many others, two major techniques resolve the heterogeneity issue by tracking surface markers of each cell with fluorescent labeling or by profiling protein markers with mass spectroscopy. However, these single cell techniques cannot identify the responses of the same (identical) cell over time or after multiple exposures to stimuli.This dissertation proposes a novel concept to track the response of each cell by spatially locating each cell onto two-dimensionally coordinated micropatterns, where each residing site of the cell is uniquely and exclusively identified based on its Cartesian coordinate. Two identical cell platforms, namely cell patch array (passive attachment) and microwell array (natural attachment) have been developed to locate and identify over 100,000 cells. Cells are maintained in living condition for over 48 hours and are stable against hydrodynamic flow while rinsing and operating. Cell responses over time have been tracked continuously with a fluorescence microscope. The single cell images are processed with a MATLAB algorithm and then enhanced with artificial intelligence. The identical cell arrays and image processing methods are explored in the following three aspects: (1) Cytotoxicity upon exposures to ionizing radiation and chemotherapeutic conditions. The oxidative damages caused by radiation are examined for each identical cell after staining with CellROX dye. The apoptosis induced by chemicals is examined for each identical cell after staining with calcein AM, annexin V and TOTO-3 dyes. The fluorescence intensities from each individual cell derived from MATLAB are used to determine the cytotoxicity damage and lifetime of each cell. (2) T cell-mediated cytotoxicity of immunotherapy. Fluorescent labeled ovalbumin transfected melanoma cell (B16-OVA) is co-cultured with OVA-specific CD8 T cells (OT-1) in each microwell. The cytokines released from T cells during cell-cell contacts are quantified within the antibody-modified microwell. The cytolysis of the B16-OVA cell is determined by the decrease of fluorescence intensity, which has been used to assess the immunotherapeutic efficacy of T cells. (3) Artificial intelligence-assisted cytotoxicity prediction and classification. The single identical cell array allows pre-segmentation of cells, which can enhance prediction accuracy and possibly reduce data processing time. The fluorescence images acquired from the cell patch and microwell array are used to train a convolutional neural network to recognize cell damage. The level of cellular damages (major, moderate, and minor), as well as the live/dead status of each identical cell, are classified rapidly by the trained model without further image processing. The model can differentiate subtle damage levels with a high prediction accuracy, which significantly accelerates the single identical cell cytotoxicity assay.--Author's abstract