Fig 2: BMX is required for DHEA metabolism by 3βHSD1.(A) LNCaP cells were treated with ibrutinib or zanubrutinib for 1 hour and subsequently treated with [3H]-DHEA for 5 hours, followed by steroid extraction from media and steroid separation and quantitation with HPLC. The experiment was done in triplicate and repeated in independent experiments. Mean ± SEM represents 3 replicates in 1 experiment. 3 independent experiments were performed. **P < 0.01, ***P < 0.001, ****P < 0.0001 (1-way ANOVA test with Dunnett’s multiple comparisons test). (B) Cells stably expressing shNT or 2 shRNA sequences against BMX were treated with [3H]-DHEA for 6 hours and analyzed as in (A). Mean ± SEM represents combined data from 3 biological independent replicates performed in technical triplicate. ***P < 0.001, ****P < 0.0001 (1-way ANOVA test with Dunnett’s multiple comparisons test). (C) 293T cells were transiently cotransfected with HA-BMX, EGFR, SRC, or YES and GST-3βHSD1, followed by GST pull-down and Western blot (left). 293T cells were transiently cotransfected with HA-BMX and GST-3βHSD1, followed by HA immunoprecipitation and Western blot (right). (D) LNCaP and C4-2 cells were cultured, protein collected, and immunoprecipitation and Western blot performed for endogenously expressed proteins. WCL blots run in parallel, contemporaneously, using identical samples are shown. (E) LNCaP cells were transiently cotransfected with HA-BMX, and GST-3βHSD1cells were starved with medium containing 10% charcoal-stripped FBS for 24 hours, then treated with steroids for 2 hours, followed by GST-pull down and Western blot to detect interaction of HA-BMX and GST-3βHSD1. (F) LNCaP cells were starved as in (E), then transfected with HA-BMX and treated with steroids for 2 hours; p-BMX was detected by Western blot. Blots run in parallel, contemporaneously, using identical samples are shown. (G) Stable C4-2 cell lines with HSD3B1 gRNA or control gRNA were transfected with HA-BMX, starved as in (E) and treated with DHEA for 2 hours; p-BMX was detected by Western blot.

Fig 3: BMX blockade and inhibition of 3βHSD1 phosphorylation inhibit expression of androgen-regulated genes and prostate cancer proliferation.(A) C4-2 cells with stable shRNA-mediated knockdown of 3βHSD1 were stably infected with lentivirus expressing either WT 3βHSD1 or 3βHSD1-Y344F and subsequently treated with [3H]-DHEA for 5 hours, followed by steroid extraction from media and steroid separation and quantitation with HPLC. (B) As in A, but cells were deprived of serum overnight, followed by treatment with DHEA for the indicated days; cell proliferation was assessed with the WST-1 assay and growth for each cell line was normalized to WT control for each designated day. (C) As in A, but cells were deprived of serum overnight and treated with DHEA for 48 hours, followed by RNA extraction and qPCR. Expression is normalized to control and RPLP0 expression. (D) Cells stably expressing shNT or 1 of 2 shRNA sequences against BMX were deprived of serum overnight, followed by treatment with DHEA and cell proliferation assessment as in B. (E) As in C, except that RNA extraction and qPCR was done on shBMX or control cells treated with DHEA for 48 hours. (F) LNCaP or C4-2 cells were deprived of serum overnight, treated with zanubrutinib or DHEA for the indicated times, and cell proliferation assessed as in B. (G) LNCaP or C4-2 cells were deprived of serum overnight and treated with zanubrutinib or DHEA for 48 hours, followed by RNA extraction and qPCR. Expression is normalized to control and RPLP0 expression. For A and C, mean ± SEM represents combined data from 3 biological independent replicates performed in technical triplicate (unpaired 2-tailed t test). For B, D, and F, mean ± SEM represents 3 replicates in 1 experiment. Three independent experiments were performed (2-way ANOVA with Bonferroni’s multiple comparisons test). For E, mean ± SEM represents combined data from 3 biological independent replicates performed in technical triplicates (1-way ANOVA with Bonferroni’s multiple comparisons test). For G, mean ± SEM represents 4 replicates in 1 experiment. Three independent experiments were performed (1-way ANOVA with Bonferroni’s multiple comparisons test). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig 4: Targeting BMX inhibits phosphorylation and enzymatic activity of 3βHSD1 in prostate tissue of prostate cancer patients.(A) Fresh prostate tissues from 3 representative patients with prostate cancer from a total of 7 patients in whom DHEA metabolism was detectable; DHEA metabolism was analyzed in a total of 42 patients (MT, homozygous HSD3B1(1245C); HZ, heterozygous). Tissues were obtained and aliquoted in 2 equal portions. 1 portion was treated with zanubrutinib and the other with DMSO. Both portions were maintained in 3 ml DMEM containing 10% FBS, incubated for 12 hours, and then [3H]-DHEA was added to each portion. Cell culture medium was collected at the indicated times, and HPLC was performed. (B) DHEA metabolism was analyzed on day 7 and day 15. Mean ± SEM represents DHEA metabolism from 4 patients. *P < 0.05, **P < 0.01 (unpaired 2-tailed t test). (C) Protein was extracted from about 20 mg patient tissue, followed by 3βHSD1 immunoprecipitation and Western blot. (D) The remaining tissue was used for Western blot: tissue cores were minced and aliquoted in 2 equal parts and treated as in A. After 12 hours of culture, DHEA (10 nM) was added to each portion. 7 days later, protein was collected, and immunoprecipitation and Western blot were performed. (E) Proposed model for 3βHSD1 phosphorylation. BMX phosphorylates 3βHSD1 Y344 upon activation by DHEA. Y344 phosphorylation enhances 3βHSD1 activity by increasing its dimerization, which subsequently promotes androgen production and prostate cancer proliferation. When BMX is inhibited, 3βHSD1 phosphorylation–stimulated dimerization is lost, reducing cellular enzyme activity, potent androgen production, and prostate cancer proliferation. 3βHSD1 inhibition also blocks estrogen synthesis. For all panels, error bars represent the SEM; P values were calculated using paired 2-tailed t tests.

Fig 5: BMX pharmacologic blockade and 3βHSD1-Y344F inhibit androgen biosynthesis and CRPC growth in vivo.(A) C4-2 cells with stable shRNA-mediated knockdown of 3βHSD1 were stably infected with a lentivirus expressing WT 3βHSD1 or 3βHSD1-Y344F. Mice were subcutaneously injected with 10 million cells, and castration plus DHEA pellet implantation were performed after tumors reached 200 mm3. Tumor growth is shown as fold change from the time of treatment initiation for each tumor. The numbers of mice in the WT-3βHSD1/castration and 344F-3βHSD1/castration groups were 13 and 12, respectively. (B) Progression-free survival was assessed as time to 3-fold increase in tumor volume from treatment initiation. (C) The testosterone concentration in xenograft tumors was detected by mass spectrometry. (D) RNA was extracted from xenograft tumors, and expression of AR-responsive genes (PSA, FKBP5, and TMPRSS2) and HSD3B1 was determined by qPCR. Expression was normalized to control and RPLP0 expression. (E) 6 million C4-2 cells were injected subcutaneously in mice, and castration, DHEA-pellet implantation, and treatment with vehicle or zanubrutinib at a dose of 15 mg/kg by oral gavage twice daily was performed after tumors reached 150 mm3. Tumor growth was assessed as fold change from time of treatment initiation. The numbers of mice in the castration/vehicle and castration/zanubrutinib groups were 13 and 12, respectively. (F) Progression-free survival was assessed as time to 3-fold increase in tumor volume from treatment initiation. (G) Tumor testosterone in xenograft tumors was detected by mass spectrometry. (H) Expression of AR-regulated genes was assessed by qPCR and expression was normalized to control and RPLP0 as in D. (I) RNA-Seq and GSEA was performed, showing AR inhibition as the top upstream regulator predicted to be inhibited. (J) Ten million VCaP cells were injected subcutaneously. Castration, DHEA-pellet implantation, and vehicle or zanubrutinib at a dose of 15 mg/kg by oral gavage twice daily were performed after tumors reached 200 mm3, and fold change in tumor volume from the time of treatment initiation was assessed. The numbers of mice in the castration/vehicle and castration/zanubrutinib groups were 12 and 11, respectively. (K) Progression-free survival was assessed as in B. (L) Xenograft testosterone detection by mass spectrometry. (M) AR-regulated genes were assessed as in D. For A, E, and J, P values were calculated by 2-way ANOVA with Bonferroni’s multiple comparisons test. For B, F, and K, P values were calculated with a log-rank test. For C, D, G, H, L, and M, P values were calculated using an unpaired 2-tailed t test. *P < 0.05, **P < 0.01, ***P < 0.001.