CUL4B promotes prostate cancer progression by forming positive feedback loop with SOX4

Abstract

How to distinguish indolent from aggressive disease remains a great challenge in prostate cancer (PCa) management. Cullin 4B (CUL4B) is a scaffold protein and exhibits oncogenic activity in a variety of human malignancies. In this study, we utilized PCa tissue specimens, cell lines and xenograft models to determine whether CUL4B contributes to PCa progression and metastasis. Here, we show that CUL4B expression highly correlates with the aggressiveness of PCa. CUL4B expression promotes proliferation, epithelial-mesenchymal transition, and metastatic potential of PCa cells, whereas CUL4B knockdown inhibits. Mechanically, CUL4B positively regulates SOX4, a key regulator in PCa, through epigenetic silencing of miR-204. In turn, SOX4 upregulates CUL4B expression through transcriptional activation, thereby fulfilling a positive feedback loop. Clinically, CUL4B+/SOX4+ defines a subset of PCa patients with poor prognosis. Bioinformatics analysis further reveals that Wnt/ß-catenin activation signature is enriched in CUL4B+/SOX4+ patient subgroup. Intriguingly, Wnt inhibitors significantly attenuates oncogenic capacities of CUL4B in vitro and in vivo. Together, our study identifies CUL4B as a key modulator of aggressive PCa by a positive feedback loop that interacts with SOX4. This regulatory circuit may have a crucial role in PCa progression.

Fig. 1. Overexpression of CUL4B predicts poor prognosis in prostate cancer (PCa).

a, b The expression of CUL4B in PCa tissues compared with normal prostate samples in GSE6956 (a) and GSE68545 (b) dataset. c The expression of CUL4B in primary and metastatic PCa tissues compared with the matched normal prostate samples in GSE35988 dataset. d–g Analysis of CUL4B expression in TCGA dataset. d CUL4B levels in PCa subgroups with different TNM stages; e CUL4B levels in PCa subgroups with different Gleason scores; f the correlation between CUL4B expression and biochemical recurrence-free survival (P = 0.015, Log-rank test) and g clinical recurrence survival (P = 0.01, Log-rank test) in the TCGA cohort. h Kaplan–Meier survival analysis of microarray data from 94 PCa cases stratified by high and low CUL4B expression levels (P = 0.02, Log-rank test). i–k Analysis of CUL4B expression in Qilu cohort by IHC assay. i Representative IHC images of CUL4B expression of benign prostate tissue and PCa tissue with different Gleason score in Qilu cohort. j Kaplan–Meier survival analysis of PCa cases from Qilu cohort according to high and low CUL4B expression levels (P = 0.17, Log-rank test). k Kaplan–Meier survival analysis of PCa cases with Pre PSA level 4–10 ng/ml in Qilu cohort stratified by CUL4B levels (P = 0.01, Log-rank test). *P < 0.05, **P < 0.01, ***P < 0.001. PCa prostate cancer, TCGA The Cancer Genome Atlas, IHC immunohistochemistry

Fig. 2. CUL4B promotes proliferation and invasion of PCa cells.

a Cell viability as assessed by MTS assay at different time points, ranging from 0 to 72 h in VCaP and DU145 cells. si4B 1#/NC: CUL4B was knockdown in VCaP cells by transfection of 1# siRNA targeting CUL4B (si4B 1#) or a negative control siRNA (NC). Flag4B/Flagα: ectopic expression of CUL4B in DU145 cells by transfection of plasmid including Flag-tag and CUL4B gene sequence (Flag4B) or empty plasmid with Flag-tag only (Flagα). b Colony formation assay of indicated PCa cells with CUL4B ablation or expression results after knocking down or overexpressing CUL4B in VCaP and DU145 cells. Quantitative analysis of colony numbers is shown in the right panel. For colony formation assays, colonies containing more than 50 cells were counted and plotted. Left panel: VCaP cells stably transfected 1# shRNA targeting CUL4B (shCUL4B 1#) or negative control (shSCR) by virus. Right panel: DU145 cells stably transfected plasmid containing CUL4B gene sequence (CUL4B) or empty plasmid (Vector) by virus. VCaP and DU145 cells stably expressing corresponding shRNA or plasmids were maintained in culture media for 14 days prior to being stained with crystal violet. c Representative photographs of EdU assays of VCaP cells transfected with si4B 1#/NC or DU145 cells transfected with Flag4B/Flagα as indicated. d, e The effect of CUL4B on cell migration/invasion evaluated by transwell migration and matrigel invasion assays. d Transwell assay of VCaP cells transiently transfected with si4B 1#/NC. e Transwell assay of DU145 cells transiently transfected with Flag4B/Flagα. Left panel: Representative images of cell migration and invasion. Right panel: Quantitative results of migration and invasion assays from triplicate experiments. In each experiment, cells were counted in five random fields of each filter under a microscope using a ×40 magnification. f–h Effect of CUL4B on turmorigenesis in vivo evaluated with xenografts model. VCaP cells with stable expression of shSCR/shCUL4B 1# subcutaneously injected into nude mice. Tumors were measured every 3 days using a vernier calliper, and the volume was calculated according to the formula: 1/6 × length × width². Representative images of xenograft tumors (f), growth curves of tumors (g), and the average weight of tumor mass in each group (h) were shown. i Representative H&E images of xenograft tumor well encapsulated in the Vector control group vs. xenograft with capsule invasion in the CUL4B group. DU145 cells with stable expression of Vector/CUL4B subcutaneously injected into nude mice. HE stain was performed to each tumor at harvest time. Each bar represents the mean ± SD of three independent experiments. *P < 0.05, **P < 0.01. PCa prostate cancer

Fig. 3. CUL4B promotes EMT in PCa through SOX4.

a Gene set enrichment analysis (GSEA) of CUL4B coexpressed signature (TCGA) in mRNA microarray of metastasis PCa (GSE32269). Enrichment scores (ES) were shown on the y-axis. x-axis bars represent individual genes of the indicated gene sets. ES = 0.43, P < 0.001, FDR q < 0.001. b EMT gene signatures in GSE35988 grouped by CUL4B expression with GSEA. ES = 0.29, P = 0.01, FDR q = 0.14. c Representative morphologic changes of VCaP cells transfected with Vector/CUL4B (phase-contrast illustration). d Western blot analysis of E-cadherin, N-cadherin, Vimentin, and CUL4B expression. VCaP cells were transiently transfected with siRNA targeting CUL4B (si4B 1#) or negative control (NC) and DU145 with plasmid expressing CUL4B (Flag4B) or empty plasmid (Flagα). GAPDH was used as a loading control. e Immunofluorescence analysis of localization and expression levels of EMT markers (N-cadherin and Vimentin) in indicated PCa cells. f Circos plot displaying the interconnectivity among genes related to PCa EMT. The thickness and color of the ribbons correlate to the correlation of genes expression in TCGA dataset (Supplementary Table 2). g GSEA of CUL4B coexpressed signature (TCGA) in mRNA microarray of PCa cells ectopic expressing SOX4 (GSE11914). ES = 0.43, P < 0.001, FDR q < 0.001. h Western blot analysis of CUL4B and SOX4 expression in PCa cells with CUL4B knockdown or overexpression. i Representative IHC images of CUL4B and SOX4 expression in xenograft tumors derived from VCaP cells. j Immunofluorescence staining analysis of CUL4B and SOX4 expression in PCa cells. k–m Effect of SOX4 in oncogenic function of CUL4B. DU145 cell were transiently transfected with empty plasmid (Flagα), CUL4B expression plasmid (Flag4B), control siRNA (NC), or SOX4 siRNA (siSOX4) as indicated. Cells were subjected to MTS assay (k), transwell migration (l) and western blot assay (m). *P < 0.05, **P < 0.01. EMT epithelial−mesenchymal transition, PCa prostate cancer, TCGA The Cancer Genome Atlas

Fig. 4. CUL4B induces SOX4 expression by modulating miR-204 level.

a Venn diagram showing candidate miRNAs targeting SOX4. SOX4 3′-UTR region was used for analysis of miRNA-binding sites by using four different bioinformatics tools including TargetScan, miRDB, PicTar and RNA22. b Relative expression of precursor miR-204 (pre-miR-204) and mature miR-204 (miR-204) in PCa cells. VCaP or LNCaP cells transiently transfected with control siRNA (NC), or CUL4B siRNA (si4B 1#) were subjected to real-time PCR. c GSEA of miR-204 signature in TCGA dataset grouped by CUL4B expression. Enrichment scores (ES) were shown on the y-axis. x-axis bars represent individual genes of the indicated gene sets. ES = −0.52, P < 0.001, FDR q < 0.001. d, e Luciferase reporter assay of HEK293T (d) and VCaP cells (e). 293T and VCaP cells were transiently transfected with wild-type construct of pmirGLO-SOX4 3′-UTR (Wild), mutant construct of pmirGLO-SOX4 3′-UTR (Mutant), control RNA (NC) or miR-204 mimics (204 mimics) as indicated. f Western blot analysis of SOX4 expression in VCaP and DU145 cells transfected with miR-204 mimics or inhibitor and relative control RNA. g Western blot analysis of SOX4 and CUL4B expression in DU145 cells. DU145 cells were transient transfected with empty plasmid (Flagα), CUL4B expression plasmid (Flag4B), control RNA (NC), or miR-204 mimics (204 mimics) as indicated. prior to western blot assay. h Real-time PCR of indicated transcripts of TRPM3 in VCaP cells. VCaP cells with stable expression of shSCR/shCUL4B were used to exam the expression of indicated transcripts of TRPM3 with specific primes. i ChIP-qPCR analysis of recruitment of CUL4B and H2AK119ub1 at promoters of miR-204 in VCaP cells. Purified rabbit IgG was used as a negative control for background enrichment signal. ChIP enrichments were presented as fold over background signal. Error bars represent mean ± SD of three independent experiments. j, k ChIP-qPCR analysis of recruitment of CUL4B (j) and H2AK119ub1 (k) at promoters of miR-204 in VCaP after transfection with control shRNA(shSCR) or shRNAs targeting CUL4B (shCUL4B 1#). Error bars represent mean ± SD of three independent experiments. l Expression of miR-204 was determined by real-time PCR from VCaP cells. CUL4B-knockdown and control VCaP cells transiently transfected with control siRNA (NC), or EZH2 siRNA (siEZH2) as indicated before real-time PCR assay. m–o ChIP-qPCR analysis of recruitment of EZH2 (m), H3K27me3 (n), and H3K4me3 (o) at promoters of miR-204 in CUL4B-knockdown and control VCaP cells. Error bars represent mean ± SD of three independent experiments. *P < 0.05, **P < 0.01. PCa prostate cancer, TCGA The Cancer Genome Atlas

Fig. 5. SOX4 reinforces CUL4B by forming feedback loop.

a, b CUL4B mRNA (a) and protein (b) expression determined after SOX4 transient knockdown or overexpression. VCaP cells transiently transfected with control siRNA (NC), or SOX4 siRNA (siSOX4), and DU145 transiently transfected with empty plasmid (P-Enter), SOX4 expression plasmid (SOX4) were subjected to real-time PCR and western blot assay. c Schematic representation of SOX4 binding sites, HMG box, in the CUL4B promoter. The loci and sequence of HMG box among different species were shown in the middle. d ChIP-qPCR analysis of recruitment of SOX4 at promoter region of CUL4B in VCaP cells. Purified rabbit IgG was used as a negative control. ChIP enrichments were presented as fold over negative control. Error bars represent mean ± SD of three independent experiments. e Luciferase reporter assay was used to determine whether CUL4B was a direct target of SOX4 in HEK293T cells. 293T cells were transiently transfected with wild-type construct of pGL3-CUL4B HMG box (Wild), mutant construct of pGL3-CUL4B HMG box (Mutant), empty plasmid (P-Enter), SOX4 expression plasmid (SOX4) as indicated. *P < 0.05, **P < 0.01, ***P < 0.001. f Representative images of HE and IHC staining of CUL4B and SOX4 in Qilu cohort with different Gleason score. GS Gleason Score, IHC immunohistochemistry

Fig. 6. CUL4B+/SOX4+ subgroup with unfavorable outcome and aberrant activation of Wnt/β-catenin signaling pathway.

a Unsupervised clustering analysis of GSE35988 (left) and GSE70769 (right) datasets based on differentially expressed genes (DEGs) of indolent and aggressive PCa. Patients statue were shown in the annotation column. Patients were annotated according to CUL4B and SOX4 expression or PCa risk assessment. Red: patients with concurrent high CUL4B and SOX4 expression (CUL4B+/SOX4+); Blue: other patients (other); Green: patients with indolent PCa (indolent); Purple: Patients with aggressive PCa (aggressive). b Principle component analysis (PCA) of unique differentially expressed genes between CUL4B+/SOX4+ and CUL4B−/SOX4− patients in TCGA and GSE70769 datasets. Each plot represents a patient. c, d The correlation between CUL4B+/SOX4+ subgroup and overall survival in the Qilu PCa cohort (c) (n = 200, P = 0.01, Kaplan−Meier survival analysis, Log-rank test) and GSE70769 (d) (n = 94, P = 0.03, Kaplan−Meier survival analysis, Log-rank test). e GSEA of statistically significant overlapping gene signatures in TCGA (left panel) and GSE70769 (right panel) datasets. Green for E2F activated signature, blue for C-MYC activated signature, and red for signature containing genes downregulated upon β-catenin activation (P < 0.05 and FDR < 0.2). f Immunofluorescence analysis of the localization and expression levels of SOX4 and β-catenin in DU145 cells. DU145 cells stable transfected with empty plasmid (Vector) or CUL4B expressed plasmid (CUL4B) were subjected to immunofluorescence analysis. Scale bar, 25 μm. g, h Cell migratory (g) and proliferation (h) capacities of DU145 cells. CUL4B-overexpressing and control DU145 cells were treated with LGK974 (10uM) or control before transwell migration (g) and MTS assay (h). i, j DU145 cells with stable expression of Vector/CUL4B subcutaneously injected into nude mice. The experiments were separated into four groups: Vector + PBS, CUL4B + PBS, Vector + LGK974, CUL4B + LGK974, n = 5/group). i The weight of tumors formed at harvest time. j Representative images of pulmonary metastatic loci. k Schematic model of positive feedback loop between CUL4B and SOX4 in regulating PCa progression. *P < 0.05, **P < 0.01, ***P < 0.001. PCa prostate cancer, TCGA The Cancer Genome Atlas