Supplementary Figure S4. Characterization of additional cell lines. A. Immunofluorescence for DAPI (blue) and CGA (red) on murine NEPC cells (TC566 and TC411K). B. Immunofluorescence for DAPI (blue) and OPN (red) on HMC-1 cells. C. Western blot for OPN in HMC-1 cells treated or not with Brefeldin A (BFA; 5 g/ml). Vinculin was evaluated as housekeeping control. D. Gating strategy applied to distinguish human NEPC cells (CD49f+c-Kit-) from HMC-1 MCs (CD49f-c-Kit+) by flow cytometry. E. OPN evaluation by ELISA (left panel) or real time PCR (the Spp1 transcript; right panel) in WT and OPN−/− MCs, and in MC/9, T23, and ST4787 cells. F. Elisa for OPN in parental MC/9 cells and in MC/9-CTR, MC/9-OPNf, MC/9-iOPN cells. T23 cells were used as control.

Adverse events in CA-3 vs vehicle treated mice

Supplementary table 1 - list of antibodies used in the study

Inhibition of α2 reduces invasiveness of PNE and PAC cells

Supplementary Figure S1. Flow cytometry evaluation of murine MC purity and maturation in vitro. A. Gating strategy used to evaluate the purity and maturation of WT, OPN-/-, MyD88-/- and TNFa-/-MCs by flow cytometry. Mature MC population is identified as c-Kit+FceRI+.

Supplementary Figure S5. Expression of OPN in MC/9 cells and effect on T23 cell proliferation. A. Separate channels of immunofluorescence for DAPI (cyan), OPN (red) and WGA (blue) of MC/9, MC/9-CTR, MC/9-OPNf, and MC/9-iOPN cells showed in Fig. 2E. B. Quantification of A as percentage of OPN positive cells. C. Murine adenocarcinoma (T23) cells (50.000/well) were cultured either alone or with MC/9, MC/9-CTR, MC/9-OPNf or MC/9-iOPN cells (tumor cell:MC ratio 1:1). After 4 days the growth rate of cancer cells was evaluated through trypan blue count. Cancer cells and MC/9 could be distinguished thanks to their grown in adhesion or suspension, respectively. All histograms depict mean ± SD of biological replicates (represented by dots). One-way ANOVA followed by Tukey’s multiple comparison test was used: *, P < 0.05; **, P <0.01. Where P-value is not indicated, the comparison between groups is not statistically significant.

Supplementary table 3 - List of putative NEPC-specific surface TLR ligands

Divergence of the proteomes of ECMs obtained from TRAMP and WT prostates

Polarized light analysis of ECM linearization in human prostate cancer specimen

This file include the Supplementary Table S2

Supplementary Figure S2. OPN evaluation in bone marrow-derived MCs. A. Separate channels of immunofluorescence for DAPI (cyan), OPN (red) and WGA (blue) of WT and OPN-/- MCs showed in Fig. 1A B. Representative images of 3 different biological replicates of immunofluorescence for DAPI (cyan), OPN (red) and WGA (blue), in WT and OPN-/- MCs. These pictures were used for the quantification reported in Fig. 1B.

Supplementary Figure S8. Hematoxylin and eosin staining in TRAMP mice. A. Hematoxylin and eosin staining in serial slides of tumors reported in Fig. 5A, showing an untreated TRAMP mouse with adenocarcinoma (ADENO) and a TRAMP mouse subjected to surgical castration showing a focal t-NEPC area.

Differentially expressed matrisome genes discriminate human NEPC and CRPC-Adeno

Upregulation of integrin α2 is characteristic of PNE and not PAC cells

Gene signatures used for transcriptomic analysis

PNE cells translocate YAP into the nucleus when cocultured with PAC cells

Supplementary Material and Methods including a key resource Table

Supplementary table 4 - transcript levels of the 29 TLR ligands measured by high throughput customized Taqman assay

This file includes Supplementary Figures S1-S5

Abstract The emergence of the neuroendocrine phenotype in castration-resistant prostate cancer (CRPC) is associated with poor patient prognosis. Castration-induced death of fully differentiated, androgen-sensitive prostate cancer cells might foster interactions among rare androgen-independent, poorly differentiated cancer cells and the extracellular matrix (ECM) that promotes the development of neuroendocrine prostate cancer (NEPC). In this study, we investigated physical and molecular interactions between poorly differentiated prostate cancer cells with exocrine (PAC) or neuroendocrine features (PNE), which recapitulated preexisting human CRPC-like cells, and decellularized prostate ECM. Without androgens, PAC cells and prostate cancer–derived ECM promoted in vitro invasiveness of PNE cells by inducing integrin α2 upregulation and YAP activation, indicating a cell-to-cell and cell-to-matrix contact-driven process. Inhibition of RANK/RANKL and NF-κB prevented integrin α2 upregulation in PNE cells, and integrin α2β1 and YAP inhibition also reduced PNE invasiveness. Microenvironment-conditioned PNE cells showed YAP-dependent metastatic behavior in vivo, and YAP inhibition suppressed the development of NEPC and metastasis in castration-naïve mice and of neuroendocrine CRPC in transgenic mice with prostate cancer. Importantly, YAP inhibitors also restrained the growth of human CRPC organoids. These findings unveil mechanisms of NEPC development and implicate the integrin α2–YAP axis as a therapeutic target in patients with prostate cancer receiving androgen deprivation therapy. Significance: Targeting signaling pathways activated by interactions between poorly differentiated neuroendocrine and exocrine prostate cancer cells and the surrounding ECM suppresses NEPC development and metastasis.