EGFR-activated myofibroblasts promote metastasis of pancreatic cancer

Pancreatic ductal adenocarcinoma (PDAC) has a dismal prognosis. Cancer-associated fibroblasts (CAFs) are recognized potential therapeutic targets, but poor understanding of these heterogeneous cell populations has limited the development of effective treatment strategies. We previously identified transforming growth factor beta (TGF-β) as a main driver of myofibroblastic CAFs (myCAFs). Here, we show that epidermal growth factor receptor/Erb-B2 receptor (EGFR/ERBB2) signaling is induced by TGF-β in myCAFs through an autocrine process mediated by amphiregulin. Inhibition of this EGFR/ERBB2-signaling network in PDAC organoid-derived cultures and mouse models differentially impacts distinct CAF subtypes, providing insights into mechanisms underpinning their heterogeneity. Remarkably, EGFR-activated myCAFs promote PDAC metastasis in mice, unmasking functional significance in myCAF heterogeneity. Finally, analyses of other cancer datasets suggest that these processes might operate in other malignancies. These data provide functional relevance to myCAF heterogeneity and identify a candidate target for preventing tumor invasion in PDAC.

Escape from cell-cell and cell-matrix adhesion dependence underscores disease progression in gastric cancer organoid models

OBJECTIVE

Cell-cell (CC) and cell-matrix (CM) adhesions are essential for epithelial cell survival, yet dissociation-induced apoptosis is frequently circumvented in malignant cells.

DESIGN

We explored CC and CM dependence in 58 gastric cancer (GC) organoids by withdrawing either ROCK inhibitor, matrix or both to evaluate their tumorigenic potential in terms of apoptosis resistance, correlation with oncogenic driver mutations and clinical behaviour. We performed mechanistic studies to determine the role of diffuse-type GC drivers: ARHGAP fusions, RHOA and CDH1, in modulating CC (CCi) or CM (CMi) adhesion independence.

RESULTS

97% of the tumour organoids were CMi, 66% were CCi and 52% were resistant to double withdrawal (CCi/CMi), while normal organoids were neither CMi nor CCi. Clinically, the CCi/CMi phenotype was associated with an infiltrative tumour edge and advanced tumour stage. Moreover, the CCi/CMi transcriptome signature was associated with poor patient survival when applied to three public GC datasets. CCi/CMi and CCi phenotypes were enriched in diffuse-type GC organoids, especially in those with oncogenic driver perturbation of RHO signalling via RHOA mutation or ARHGAP fusions. Inducible knockout of ARHGAP fusions in CCi/CMi tumour organoids led to resensitisation to CC/CM dissociation-induced apoptosis, upregulation of focal adhesion and tight junction genes, partial reversion to a more normal cystic phenotype and inhibited xenograft formation. Normal gastric organoids engineered with CDH1 or RHOA mutations became CMi or CCi, respectively.

CONCLUSIONS

The CCi/CMi phenotype has a critical role in malignant transformation and tumour progression, offering new mechanistic information on RHO-ROCK pathway inhibition that contributes to GC pathogenicity.