Invasive phenotypes of triple-negative breast cancer-associated fibroblasts are mechanosensitive, AhR-dependent and correlate with disease state.

Cancer associated fibroblasts (CAFs) play a critically important role in facilitating tumour cell invasion during metastasis. They also modulate local biophysical features of the tumour microenvironment through the formation of fibrotic foci, which have been correlated with breast cancer aggression. However, the impact of the evolving three-dimensional biophysical tumour microenvironment on CAF function remains undefined. Here, by isolating CAFs from primary human triple-negative breast cancer tissue at the time of surgery, we find that their ability to remodel the local microenvironment and invade into a three-dimensional matrix correlates with disease state. We then engineered culture models to systematically deconstruct and recreate mechanical tissue features of early breast cancer fibrotic foci; and demonstrate that invasion is mechanically-activated only in CAFs from patients with no detectable pre-existing metastases, but is independent of mechanical cues in CAFs isolated from patients with later-stage axillary lymph node metastases. By comparing the differential transcriptional response of these cells to microenvironmental tissue stiffness, we identify the aryl hydrocarbon receptor (AhR) as being significantly upregulated in invasive sub-populations of both mechanically-activated and mechanically-insensitive CAFs. Increasing AhR expression in CAFs induced invasion, while suppressing AhR significantly reduced invasion in both mechanically-activated and mechanically-insensitive CAF populations, even on stiffnesses that recapitulate late-stage disease. This work therefore uses mechanobiological analyses to identify AhR as a mediator of CAF invasion, providing a potential stratification marker to identify those patients who might respond to future mechanics-based prophylactic therapies, and provides a targetable mechanism to limit CAF-associated metastatic disease progression in triple-negative breast cancer patients. STATEMENT OF SIGNIFICANCE: By designing a mechanically-tunable tissue-engineered model of fibroblastic foci, and using this to culture patient-derived cancer-associated fibroblasts, we demonstrate that these cells are differentially mechanosensitive, depending on disease stage of the patient. While comparing transcriptomic profiles of patient-derived cells produces too many pathways to screen, identifying the pathways activated by local tissue mechanics that were common across each patient allowed us to identify a specific target to limit fibroblast invasion. This broad discovery strategy may be useful across a variety of biomaterials-based tissue engineered models; and these specific findings suggest (1) a strategy to identify patients who might respond to CAF- or matrix-targeting therapies, and (2) a specific actionable target to limit CAF-associated metastatic disease progression.