Phenotypic plasticity, spatial niches and cancer-associated fibroblast/mural cell interactions define the human prostate cancer microenvironment

Abstract

Due to their established roles in tumor progression and therapy resistance, cancer-associated fibroblasts (CAFs) are considered key therapeutic targets. However, a lack of pre-clinical models has hampered the functional characterization of tumor-promoting versus tumor-suppressive CAF substates, a priori for the development of CAF-targeted therapeutics. We describe and cross-validate a biobank of primary human prostate inflammatory (iCAFs) and myofibroblastic (myCAFs) CAFs that closely recapitulate fibroblast heterogeneity in clinical prostate cancer (PCa) and retain key functional hallmarks pertinent to their CAF substate. iCAFs and myCAFs coexisted in clinical specimens but localized to distinct spatial niches and differentially expressed androgen receptor (AR). Pro-migratory AR-negative myCAFs retained high plasticity in a biomechanical-sensitive manner and were associated with hallmarks of poor outcome. Consistently, myCAFs represented the dominant CAF subpopulation in fibroblast-rich high-grade PCa and castration-resistant patient-derived xenografts. Conversely, AR-positive iCAFs were associated with a favorable prognosis. In view of the androgen-dependent but tumor-suppressive nature of parenchymal smooth muscle cells, these findings imply the potential inadvertent ablation of protective AR+ stromal cell types during androgen deprivation therapy may favor accumulation of aggressive AR-negative/castration-resistant myCAFs. The CAF biobank generated herein thus represents a valuable experimentally-tractable tool for the future design of CAF-targeted therapeutics to overcome therapy resistance.