Uncovering the mode of action of engineered T cells in patient cancer organoids

ABSTRACT: The rise of numerous promising immunotherapies poses a need for technology platforms to study their efficacy and mechanism of action. Organoids are cellular 3D structures that recapitulate important tumour characteristics and there is growing interest to use such technology to evaluate immunotherapy efficacy. Imaging provides a powerful approach to probe the cellular complexity modelled with organoids, but requires the development of adequate analytical methodology to comprehensively investigate the large and complex datasets retrieved. Here, we developed an organoid-based 3D live cell imaging and data mining platform and applied it to TEGs (T cells engineered to express a V9/V2 T cell receptor (TCR)) that sense metabolic changes via the recently identified ligand butyrophilin 2A1 (BTN2A1) bound to to BTN3A19. Because identification of specific antigens for solid tumours is challenging pan-tumour targeting through metabolic recognition is considered an important step forward and has indeed revealed exciting broad-targeting potential. We set-up a rapid imaging strategy using real-time fluorescent dyes to specifically label T cells, organoids and dead cells and their single acquisition through spectral imaging and performed high-resolution single-cell 3D time-lapse imaging. To investigate the behavioural landscape of T cells in co-culture with organoids, we developed a computational framework called BEHAV3D that allowed to grasp the complexity of our data by extracting multiple dynamic imaging parameters and performed a pooled multivariate time-series analysis. We pooled active (TEG) and mock T cells (LM1) with organoids that were either sensitive to the TEGs (13T) and that were resistant (100T). This reference map of behavioral patterns of TEGs showed that T cells could be separated into nine subpopulations with unique behavioral patterns. Patterns varied from inactive behaviors (dying, static and lazy) to active motility (slow scanner, medium scanner and super scanner) and organoid engagement (tickler, engager and super engager), thus demonstrating a high level of behavioral heterogeneity. Correlation between single organoid dying dynamics and TEG engagement over time revealed that organoids contacted by super engagers, as compared to other organoid-engaging clusters, had the highest chance of being killed. This indicates that effective killing by TEGs relies on prolonged organoid contact, a main feature of super engagers (48 ± 8 min / hr; mean ± s.d.). To link tumor targeting behavior to population phenotypes, we also included experiments where CD4+ and CD8+ TEGs were differentially labeled the co-cultures. This revealed that prolonged organoid contact and super engager behavior was a preferred feature of CD8+ TEGs, whereas CD4+ TEGs showed a higher proportion of lazy cells, slow scanners, medium scanners, super scanners, and ticklers characteristic of high movement and short organoid contact. Here, we provide an organoid-based 3D imaging-transcriptomic platform; BEHAV3D, for understanding the mode-of-action of cellular anti-cancer immunotherapies in a patient-specific manner and apply it to diverse solid tumor PDO models and multiple engineered T cell products.

ORGANISM(S): Homo sapiens (human)

SUBMITTER:  

PROVIDER: S-BIAD448 | bioimages |

REPOSITORIES: bioimages