BEHAV3D: A 3D live imaging platform for comprehensive analysis of engineered T cell behavior and tumor response

The use of patient-derived material and immune cell co-cultures in modeling immune-oncology has gained significant interest for understanding and manipulating immune cell tumor targeting in a patient-specific context. However, current protocols have limitations in visualizing and analyzing the dynamic cellular features of these living culture systems. We recently developped a workflow names BEHAV3D,  that combines multi-color live 3D imaging and computational tools to analyze cell death dynamics, classify T cell behavior, and generate data-informed 3D images and videos. Here we provide some example pre-processed dataset of videos of two co-culture set ups: breast cancer Patient Derived Organoids with αβ T cells engineered to express a γδ TCR (TEGs) and Acute Lymphoblastic Leukemia cells with CD19 CART cells.

Methods
Our approach involves the development of a co-culture assay that allows for 3D live imaging to simultaneously monitor tumor cell response and T cell dynamic behavior. We tested and designed different combinations of live cell dyes to label viable, dead cells, and up to two types of immune cells. This method does not require genetic manipulation or the use of fluorescent markers, making it fast, widely applicable, and preserving cellular heterogeneity. We also determined the optimal proportions of tumor cells to therapeutic cells for solid and liquid tumors, and optimized the imaging medium to ensure compatibility with live imaging.

For image acquisition was performed on a Zeiss LSM880 confocal system. This allows for the acquisition of four spectrally resolved fluorophores in a single scan, enabling fast imaging with minimal photobleaching and phototoxicity. Image processing is performed with Imaris (Oxford Instruments) sofware and an image processing pipeline was developed to extract statistics on cellular and organoid tracks. Single cells and tumor cells are segmentated and tracked in these example datasets.