Adenocarcinoma-neuroendocrine Transition Of Androgen Resistant PCa Depends On SPARC Down-regulation In Stromal Accessory Cells

Adenocarcinoma-neuroendocrine differentiation of castration resistant prostate cancer often occurs in a relevant subset of patients as a mechanism of resistance to androgen deprivation and androgen-receptor targeted drugs. Tumor cell plasticity toward neuroendocrine differentiation can be sustained by signals provided by the tumour microenvironment. The matricellular protein SPARC is a crucial modulator of cancer progression and phenotype, exerting different functions depending on its tumour or stromal expression. Here, we elucidated the different contribution of tumour- or stromal-derived SPARC in prostate cancer, using the TRAMP mouse model. Interestingly, crossing TRAMP mice with Sparc-/- mice, we observed the appearance of focal areas of neuroendocrine differentiation within adenocarcinoma. The same percentage of neuroendocrine differentiation observed in Sparc-/-TRAMP mice was also observed in castrated TRAMP mice suggesting that SPARC deficiency and castration converge to the same disease outcome. Supporting this hypothesis, we found that SPARC expression in stromal cells populating TRAMP prostate was dramatically reduced after castration. Furthermore, adenocarcinoma cells acquired neuroendocrine phenotype after injection in Sparc-/- hosts. These results suggested a pivotal role for stromal-derived SPARC in limiting neuroendocrine differentiation of prostate cancer. Accordingly, prostate cancer cell lines, independently by their own expression of SPARC, when co-cultured in presence of Sparc-deficient fibroblasts acquired neuroendocrine features. This transition occurs through the effect of IL-6, which is released by SPARC-deficient, but not proficient, fibroblasts. Indeed, molecular targeting of IL-6 receptor was able to reduce neuroendocrine differentiation both in vitro and in vivo. We further detailed a tumor-stroma crosstalk, triggered in TRAMP mice by castration and mimicked in vitro by co-culturing tumour cells and fibroblasts in presence of Enzalutamide, in which miR29b produced by tumour cells consequently induce Sparc down-regulation. We also identified the heat shock protein GRP78 as the primer of this tumor-stromal crosstalk leading to neuroendocrine differentiation. Accordingly, we found that the molecular targeting of GRP78 strongly reduces neuroendocrine differentiation both in vivo and in vitro. We were able to confirm the modulation of GRP78 and SPARC, and focal up regulation of NE markers, in paired cancer samples collected before and after androgen deprivation therapy in a small cohort of patients. Our results highlight the pivotal role of stromal components in driving neuroendocrine differentiation of prostate cancer also suggesting possible diagnostic and therapeutic targets.