Topological defects induced by MSC-cancer interaction correlate with cancer cell apoptosis

Mesenchymal stem cells (MSC) can target and home into tumors by tracking the chemotactic factors secreted by cancer cells, thus making MSCs ideal candidates for active drug carriers in cancer therapy. Despite some success in preclinical models and clinical trials, little is known about how MSC homing occurs. To gain mechanistic insight into the interactions and response of MSC to cancer cell invasion, we employ a controlled co-culture assay using primary mouse MSC and a neuroblastoma cell line (Neuro2a) as our model cell types. The spindle-shaped MSC collectively develops orientational order and macroscopic patterns disrupted by MSC-Neuro2a interactions, forming topological defects that trigger apoptosis of Neuro2a cells and cell state transition of MSC, respectively. By modeling the MSC layer as an active nematic liquid crystal, we propose a computational framework combining deep learning and Bayesian Inference to explore MSC-cancer cell dynamics. We integrated cell segmentation with Variational Inference methods to predict intercellular aligning interactions similar to a classical XY-like model. To further probe the spatiotemporal correlation between apoptosis signaling and topological defects, we integrated imaging and single-cell transcriptomic profiling, we identified upregulated apoptotic signals in Neuro2a and MSC differentiation toward neural precursors induced by physical interactions. Gene network and ontology analyses revealed that the interactions triggered specific signaling pathways, particularly PTN, Collagen, and FN1 signaling pathways, which may indicate apoptosis in Neuro2a, enhanced adhesion between MSC/Neuro2a, and immunomodulation through MSC. Taken together, our study provides physical insights with molecular perspectives on the interactive motility of stem cells and cancer cells, offering a complementary approach to designing stem cell-based drug delivery strategies for cancer therapy.