Fig 2: TGFBI and ECM-1 have opposite effects on acinar morphogenesis and PC3 cell invasion. a Acinar morphogenesis assays for shDKK3 (sh6) RWPE-1 cells cultured in media only (control) or with ECM-1 (100 ng/ml) or TGFBI (1 μg/ml) for 7 days; error bars show SD, *p < 0.05, **p < 0.01, n = 4. b Acinar morphogenesis assays for shCTRL (NS11) RWPE-1 cells cultured as in a. c Invasion assays for PC3 cells plated in triplicate wells for 24 h in serum-free RPMI with PBS (Control) or with TGFBI (1 μg/ml), ECM-1 (100 ng/ml) or both TGFBI and ECM-1. Left, example photos of invaded cells, right, graph showing average number of invaded cells, normalized to control, error bars show SD, n = 3, **p < 0.01 by ANOVA; scale bars 100 μm. d, e Invasion assays for C4-2B cells (d) and enzalutamide-resistant C4-2B (MDVR) cells (e) treated as in c

Fig 3: Effects of DKK3 silencing on TGFBI and ECM-1 protein and mRNA levels in WPMY-1 and RWPE-1 cells. a Western blots of CM from equal numbers of shCTRL (PSM2) and shDKK3 (Wsh8) WPMY-1 cells cultured in serum-free medium for 48 h were probed for TGFBI; a Coomassie Blue (CB)-stained gel of samples run in parallel was used as a loading control. b Densitometry analysis of TGFBI in CM; graph shows average intensity normalized to PSM2, error bars show SD, **p < 0.01, n = 5. c q-RT-PCR analysis of TGFBI mRNA levels, relative to 36B4, in shCTRL (PSM2/PSM3) and shDKK3 (Wsh7/Wsh8) WPMY-1 cells; error bars show SD. d Western blots of CM from equal numbers of shCTRL (NS11) and shDKK3 (sh6) RWPE-1 cells cultured in serum-free medium for 48 h were probed for TGFBI; a CB-stained gel of samples run in parallel was used as a loading control. e Densitometry analysis of TGFBI in CM; graph shows average intensity normalized to NS11, error bars show SD, **p < 0.01, n = 3. f q-RT-PCR analysis of TGFBI mRNA levels, relative to 36B4, in shCTRL (NS11) and shDKK3 (sh6) RWPE-1 cells; error bars show SD, n = 3, *p < 0.05. g–l ECM-1 protein and ECM1 mRNA levels were analyzed as in a–f; h **p < 0.01, n = 6, k *p < 0.05, n = 3

Fig 4: Analysis of TGFBI and ECM-1 in prostate cancer patients. a Statistical analysis of TGFBI expression in benign prostate and prostate cancer; b Benign and tumor sections from a patient stained for TGFBI. c Tumor sections from two patients stained for TGFBI and Dkk-3; scale bars 65 µm. d Statistical analysis of ECM-1 expression in benign prostate and prostate cancer. e Example of a patient tumor with increased ECM-1 in cancer, compared to in benign epithelium. Benign and cancer sections from the same patient were stained for ECM-1, SMA, and pan-CK; scale bars 84 µm. Gl Gleason, PCaE prostate cancer epithelium, BS benign stroma, PCaS prostate cancer stroma, BE benign epithelium; Chi Sq. Chi square Yates correction, Fisher, Fisher’s Exact test, two-sided

Fig 5: Changes in serum CLU, VCAM1, and CRELD2 concentrations in response to SFN treatment differed significantly between the R group and NR group. Box‐and‐whisker plots of 3 proteins exhibiting a significant change. The rates of change in CLU, VCAM1, and CRELD2 concentration were significantly higher in the NR group than the R group 1 month after treatment initiation (R = 18, NR = 18). **P < 0.01, ***P < 0.001. Abbreviations: CLU, clusterin; VCAM1, vascular cell adhesion molecule‐1; CRELD2, cysteine‐rich with EGF‐like domain protein 2.