Single-cell multiomic modelling of early metastatic events promoted by the extracellular matrix

Cancer metastasis, primarily driven by epithelial-to-mesenchymal transition (EMT), is responsible for most cancer-related mortalities. Traditional pre-clinical models fail to fully capture mesenchymal characteristics due to the loss of human stroma. The extracellular matrix (ECM) plays a crucial role in EMT, yet conventional in vitro models often rely on defined ECM components, which may not adequately replicate the human physiological ECM niche.

To mimic the in situ dissemination of cancer cells, we employed a patient-derived extracellular matrix (pdECM). We transitioned the culture matrix for patient-derived colorectal cancer organoids from a basement membrane extract (BME) to a patient-derived ECM (pdECM). We performed single-cell multiomic analyses, integrating transcriptomic and epigenomic data, to investigate changes in organoid phenotypes and reconstruct the EMT trajectory.

Organoids cultured in the pdECM exhibited increased tumor cell dissemination and motility, resembling in situ lesions without exogenous ligand treatment. Single-cell multiomic analysis revealed TNF-α signaling as an early metastatic event in the EMT trajectory. Epigenomic changes led to increased accessibility of AP-1 complex target genes, particularly MMP7, which promoted an invasive phenotype. Our multimodal computational approach distinguished early and late EMT states, demonstrating that pdECM-induced EMT occurs independently of traditional EMT master regulators. Notably, pdECM organoids exhibited a partial EMT phenotype, characterized by hybrid epithelial-mesenchymal states.

This study presents an advanced in vitro model that closely recapitulates in situ tumorigenesis and provides novel insights into the metastatic cascade. The pdECM system enables the reconstruction of EMT dynamics, highlighting the critical role of ECM composition in metastasis and offering a physiologically relevant platform for the development of targeted therapies.