Objective: It is well established that cancer results from the complex interplay of mutated or otherwise phenotypically reprogrammed cells with the surrounding tissue and immune microenvironments. However, understanding this process has often been inferred by collecting data at snapshots in time, limiting our understanding of tissue and molecular-level dynamics at the root of disease pathogenesis and heterogeneous drug responses. Our objective was to address this limitation by combining ex vivo tissue culture, fluorescent biosensor, and imaging techniques to develop a novel live-cell dynamic model of cancer metastasis in the lung. We applied this model to determine how local physical and signaling microenvironments of the lung modulate osteosarcoma ERK and AKT signaling dynamics and gene expression heterogeneity, to single cell precision.
Methods: Osteosarcoma cells co-expressing ERK and AKT biosensors were disseminated into human or mouse lung tissues. Precision cut lung slices were generated, and live-cell imaging performed using spinning disk confocal microscopy under environmental control for CO2, heat, and humidity to capture biosensor dynamics over minute time scales for periods of several days. Fixed immunofluorescent imaging was performed to correlate ERK and/or AKT target gene expression with dynamic live-cell signaling data. Data was analyzed using custom MATLAB and Python software.
Results: We found that osteosarcoma signaling dynamics were highly sensitive to spatiotemporal variation in the physical microenvironment. Cells persisting in vascular niches often exhibited distinct ERK activity dynamics comprised of low frequency, high amplitude pulses and modest cell migration, whereas AKT signaling remained mostly active and stable. By contrast, cells that successfully escaped into alveolar spaces often exhibited high frequency, high amplitude ERK pulses and were often highly motile. These features were also linked to the density of tumor cells, whereby clusters of cells exhibited behavioral stability and reduced signaling heterogeneity compared to individual cells, providing evidence of collective tumor cell behaviors. These factors converged upon gene expression pathways to produce significant cell-to-cell heterogeneity in ERK and AKT targets, such as the pioneer factor, Fra-1, and pro-survival and drug resistance factor, MCL-1
Conclusion: Our results demonstrate the utility of a novel dynamic live-cell tissue model to measure and understand tumor signaling dynamics and behaviors within the context of the lung metastatic niche. Data obtained from this model provided new insights into how spatial constrains and local tissue microenvironments influence signaling dynamics and gene expression to create intratumoral heterogeneity, with significant implications for single cell drug response variation.
