Matrix biophysical cues drive transient or prolonged breast cancer dormancy programmes
Breast cancer metastasis can emerge after disease-free periods through the reawakening of dormant disseminated cancer cells. Despite evidence that the extracellular matrix (ECM) modulates this phenomenon, biophysical cues regulating dormancy fate and therapeutic responses remain poorly understood. Here, we engineered stiffness-matched gelatin methacryloyl (GelMA) and alginate-based 3D matrices to disentangle matrix stiffness from adhesion and degradability and uncover their contributions to dormancy. Deep learning-based 3D segmentation revealed that adhesive and degradable matrices supported rounded and proliferative tumour masses, whereas inert matrices favoured single cell dormancy. Within inert matrices, stiffness further determined dormancy fate, with soft matrices activating SOX9-associated transient stem-like dormancy, while stiff matrices reduced cell volume, increased intracellular macromolecular crowding of proteins and induced NR2F1-associated prolonged dormancy marked by autophagy and cellular stress. Transiently dormant cells retrieved from soft matrices retained dormancy memory and failed to form tumours in chicken chorioallantoic membranes, demonstrating persistence of matrix-induced states after removal of biophysical cues. Moreover, patient-derived cells in soft inert matrices recapitulated transient dormant states and exhibited therapy resistance. We demonstrate that tuneable biophysical cues shape distinct dormancy states, linking matrix mechanics to dormancy regulation and therapeutic response.