Fig 2: Androgen-deprivation therapy promotes CXCL5/CXCR2-driven neuroendocrine differentiation (NED) of prostate cancer (PCa)(A) Relative protein levels of CXCL5 and CXCR2 in LNCaP, C4-2, 22Rv1, PC3, and LASCPC-01 cells.(B) Relative protein levels of CXCL5, CXCR2, ENO2, CHGA, KLK3, and NKX3-1 in LNCaP cells cultured in CSS-containing medium for 0–96 h.(C) Relative mean mRNA expression levels of KLK3, NKX3-1, CXCL5, and CXCR2 in LNCaP cells from 3 weeks to 11 months of ADT in the GDS3358 database. ∗ vs. Control, as determined by a one-way ANOVA.(D) Relative mRNA levels of CXCL5, CXCR2, neuroendocrine (CHGA, SYP, and ENO2), and androgen-responsive (KLK3 and NKX3-1) markers in LNCaP cells cultured in CSS-containing medium for 1 or 5 days, followed by treatment with 10 nM dihydrotestosterone (DHT) for 1 day ∗ vs. FBS; # vs. CSS 5days, as determined by a two-way ANOVA.(E) Relative mRNA levels of CXCL5, CXCR2, neuroendocrine, and androgen-responsive markers in LNCaP cells stably transfected with an empty vector (EV) or a CXCL5-expressing vector. ∗ vs. EV, as determined by a one-way ANOVA.(F) Relative protein levels of CXCL5, CXCR2, ENO2, CHGA, KLK3, and NKX3-1 in LNCaP cells stably expressing the EV or CXCL5-expressing vector.(G and H) Relative mRNA levels of CXCL5, CXCR2, neuroendocrine, and androgen-responsive markers in LNCaP (G) and C4-2 (H) cells treated with PBS or 20 ng/mL of the CXCL5 recombinant protein, followed by treatment with DMSO or 5 μM navarixin for 48 h ∗ vs. PBS+DMSO; # vs. CXCL5+DMSO, as determined by a one-way ANOVA.(I) Relative protein levels of CXCL5, CXCR2, ENO2, CHGA, KLK3, and NKX3-1 in LNCaP and C4-2 cells treated with PBS or 20 ng/mL of the CXCL5 recombinant protein, followed by treatment with DMSO or 5 μM navarixin for 48 h.(J) GSEA of the TCGA PCa dataset revealed a positive correlation between high CXCL5 expression in prostate tissues and gene signatures representing NEPC-response signaling (Li).(K) GSEA of TCGA PCa dataset revealed an inverse correlation between high CXCL5 expression in prostate tissues and gene signatures representing androgen response signaling (PID, Wang, Nelson, Hallmark, and GO). NES, normalized enrichment score; FDR, false discovery rate. Quantification of relative mRNA levels is presented as the mean ± SEM from three biological replicates. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.005.

Fig 3: Knockdown of WPMY-1 suppresses LIF-driven neuroendocrine differentiation (NED) and malignant progression in prostate cancer (PCa) cells(A) Relative protein levels of LIF, LIFR, WISP1, SOCS3, and PDL1 were measured in AR-positive PCa cell lines (LNCaP, C4-2, and 22Rv1), an AR-negative PCa cell line (PC3), and an NEPC cell line (LASCPC-01).(B) Relative protein levels of WISP1, phosphorylated (p)-STAT3, STAT3, PDL1, and SOCS3 were measured in LNCaP cells treated with PBS or 100 ng/mL of the LIF recombinant protein for a duration ranging 0 to 96 h.(C) Relative mRNA levels of WISP1, neuroendocrine (CHGA, SYP, and ENO2), stem cell (SOX2 and NANOG), and anti-inflammatory (SOCS3 and PDL1) markers in LNCaP cells treated with PBS or 100 ng/mL of the LIF recombinant protein or combined treatment with DMSO or 35 nM EC330 for 48 h ∗ vs. PBS+DMSO; # vs. LIF+DMSO, as determined by a one-way ANOVA.(D) Relative protein levels of WISP1, p-STAT3, STAT3, PDL1, and SOCS3 in LNCaP and C4-2 cells cultured in PBS or 100 ng/mL of the LIF recombinant protein, followed by DMSO or 35 nM EC330 treatment for 48 h.(E) Relative mRNA expression levels of WISP1, neuroendocrine markers, stem cell markers, and anti-inflammatory markers (SOCS3 and PDL1) were measured in LNCaP cells expressing the non-target control (NC) or WISP1 siRNA, followed by treatment with PBS or 100 ng/mL of the LIF recombinant protein for 48 h ∗ vs. PBS+NC; # vs. LIF+NC, as determined by a one-way ANOVA.(F and G) Relative cell proliferation (F) and sphere formation (G) were measured in LNCaP cells expressing either NC or WISP1 siRNA, followed by treatment with either PBS or 100 ng/mL of LIF recombinant protein for 5 days (F) or 1 week (G). Scale bars represent 100 μm (G). ∗ vs. PBS+NC; # vs. LIF+NC, as determined by a one-way ANOVA. (H and I) Relative cell migration (H) and invasion through Matrigel (I) were measured in LNCaP cells expressing the NC or WISP1 siRNA, followed by treatment with PBS or 100 ng/mL of LIF recombinant protein for 12 h. Scale bars representing 20 μm are shown. ∗ vs. PBS+NC; # vs. LIF+NC, as determined by a one-way ANOVA. Quantification of relative mRNA levels, cell proliferation, sphere formation, and cell invasion through Matrigel is presented as the mean ± SEM from three biological replicates. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.005.(J) A GSEA of the TCGA PCa dataset revealed significant associations between high WISP1 expression in prostate tissues and a gene signature representing NEPC-responsive signaling. NES, normalized enrichment score; FDR, false discovery rate.

Fig 4: WISP1 abundance in serum relative to prostate cancer (PCa) progression(A) Relative mRNA levels of CXCL5, CXCR2, LIF, WISP1, neuroendocrine (CHGA, SYP, and ENO2), and stem cell (SOX2 and NANOG) markers in C4-2 cells expressing the empty vector (EV) or CXCL5-expressing vector, followed by treatment with DMSO or 35 nM EC330 for 48 h ∗ vs. EV + DMSO; # vs. CXCL5+DMSO, as determined by a one-way ANOVA.(B and C) Tumor growth analysis was conducted by subcutaneously inoculating male nude mice with C4-2 cells expressing either the EV or a CXCL5-expressing vector. The mice were then treated bi-daily with either DMSO or 2.5 mg/kg EC330 via intraperitoneal injection and allowed to grow for 8 weeks. Tumor sizes were measured weekly (B). Tumor weights were measured upon tumor collection (C). n = 5 per group. ∗ vs. EV + DMSO; # vs. CXCL5+DMSO, as determined by a one-way ANOVA and t-test.(D and E) Immunohistochemical (IHC) staining and intensity analyses were performed to assess protein levels of CXCL5, CXCR2, LIF, WISP1, and ENO2 in subcutaneous tumors derived from (B). ∗ vs. EV + DMSO; # vs. CXCL5+DMSO, as determined by a two-tailed Student’s t-test. Scale bars, 100 μm.(F) WISP1 concentrations were measured in patient sera derived from samples of benign prostatic hyperplasia (BPH; n = 10), hormone-sensitive PCa (HSPC, n = 10), and metastatic castration-resistant PCa (mCRPC; n = 8). ∗ vs. BPH; # vs. HSPC, analyzed by a one-way ANOVA.(G) GSEAs of the TCGA PCa dataset revealed significant associations between high LIF and WISP1 expressions in prostate tissues with gene signatures representing metastatic prostate stroma-responsive signaling. NES, normalized enrichment score; FDR, false discovery rate.(H) A schematic summary of this study is presented. Crosstalk between PCa cells and prostate stromal cells within the tumor microenvironment (TME) may enhance the interaction between the CXCL5/CXCR2 and LIF/LIFR pathways. This interaction can potentially upregulate the secretion of LIF and WISP1 proteins, facilitated by the critical roles of STAT3 signaling-driven activation of the WISP1 and LIF genes.

Fig 5: ESRRG expression in the cellular landscape of human and mouse PCa. (A) Feature plot (top) and violin plot (bottom) illustrating ESRRG cell type-specific expression in the HuPSA data set (Cheng et al. 2024). (B) Feature plot (top) and violin plot (bottom) illustrating Esrrg cell type-specific expression in the MoPSA data aset (Cheng et al. 2024). (C) ESRRG in AdPC patient tumors harboring high versus low SYP/CHGA NEPC marker expression (NE high, Z-score >0; NE low, Z-score <0) (Taylor et al. 2010). Significance was determined by Mann–Whitney test (unpaired, two-tailed). (D) ESRRG in metastatic CRPC patient tumors of adenocarcinoma (adeno) or neuroendocrine lineage (NEPC) (Labrecque et al. 2019). Significance was determined by Mann–Whitney test (unpaired, two-tailed). (E) ESRRG in PDXs representing CRPC adenocarcinoma (adeno) or neuroendocrine lineage (NEPC) (Labrecque et al. 2019). Significance was determined by Mann–Whitney test (unpaired, two-tailed). (F) ESRRG in PDXs representing CRPC adenocarcinoma (adeno) or neuroendocrine lineage (NEPC) (Anselmino et al. 2024). Significance was determined by Mann–Whitney test (unpaired, two-tailed). (G) Gene expression levels of mouse Esrrg in normal (WT) prostate tissue and in PCa tumors arising from Pten loss alone (AdPC) or in combination with the loss of Rb1 and/or Trp53 (NEPC-like) (Ku et al. 2017). Significance was determined by one-way ANOVA multiple comparisons test (Tukey). (H) ESRRG expression levels across a panel of six commonly used human PCa cell line models based on data from the CTPC portal (https://pcatools.shinyapps.io/CTPC_V2). The cell lines were classified as either castration-sensitive PCa (CSPC) or castration-resistant PCa (CRPC). CRPC cell line models were subclassified into AR-dependent PCa (CRPC-AR), double-negative PCa (DNPC), or neuroendocrine PCa (NEPC). (I) Heat map of mRNA expression (Z-score) of ESRRG and the indicated markers across a human PCa cell line panel based on data from the CTPC portal (https://pcatools.shinyapps.io/CTPC_V2). Genes were clustered using one minus Pearson distance measure and complete linkage. (J) Normalized ERRγ protein abundance levels across a panel of PDXs of primary PCa origin or from distinct subtypes of CRPC (Zhang et al. 2025). Levels of ERRγ protein are also shown as a function of disease progression, with each PDX model annotated with a pseudotime progression score. Linear correlations were evaluated for samples in all four molecular groups or for CRPC subtypes only. (r) Pearson, (P) two-tailed. (*) P < 0.05, (**) P < 0.01, (***) P < 0.001.