Subtype-specific collaborative transcription factor networks are promoted by OCT4 in the progression of prostate cancer

Abstract

Interactive networks of transcription factors (TFs) have critical roles in epigenetic and gene regulation for cancer progression. It is required to clarify underlying mechanisms for transcriptional activation through concerted efforts of TFs. Here, we show the essential role of disease phase-specific TF collaboration changes in advanced prostate cancer (PC). Investigation of the transcriptome in castration-resistant PC (CRPC) revealed OCT4 as a key TF in the disease pathology. OCT4 confers epigenetic changes by promoting complex formation with FOXA1 and androgen receptor (AR), the central signals for the progression to CRPC. Meanwhile, OCT4 facilitates a distinctive complex formation with nuclear respiratory factor 1 (NRF1) to gain chemo-resistance in the absence of AR. Mechanistically, we reveal that OCT4 increases large droplet formations with AR/FOXA1 as well as NRF1 in vitro. Disruption of TF collaborations using a nucleoside analogue, ribavirin, inhibited treatment-resistant PC tumor growth. Thus, our findings highlight the formation of TF collaborations as a potent therapeutic target in advanced cancer.

Fig. 1. Identification of distinct TF networks in advanced PC that contributes to the aggressiveness.

a Heatmap of TFs in benign prostate, primary PC, and CRPC tissues. Average RNA-seq results (RPKM) are summarized. Benign N = 6, Prostate cancer N = 8, CRPC N = 5. b Publicly microarray data (GSE35988) used for gene expression analysis. Pca: prostate cancer, mCRPC: metastatic CRPC. Two-sided Mann–Whitney test was performed between Pca and mCRPC. Benign N = 12, Pca N = 51, mCRPC N = 25). c Knockdown of FOXM1 and OCT4 expression by siRNAs. d Cell proliferation assay of multiple PC cells transfected with siControl or two siRNAs targeting indicated TFs. MTS assay was used to quantify the cell growth rate (N = 4, biological independent). N.S not significant, *P < 0.05, **P < 0.01, ***P < 0.001. two-sided t-test was performed. Data are presented as average ± S.D. Source data are provided as a Source Data file.

Fig. 2. OCT4 functions as a modulator of AR and FOXA1 TF complex in AR-positive PC cells.

a OCT4 ChIP-seq view of a representative AR-target gene, ACSL3. b (upper) OCT4-binding sites (OCT4BSs) by ChIP-seq in LNCaP and 22Rv1 cells. (lower) Venn diagram showing the overlap of AR with OCT4BSs in the presence of dihydrotestosterone (DHT) in 22Rv1 cells. c Composition plot of OCT4-binding signal in LNCaP cells. d Immunoblots of AR, AR-V7, and OCT4 after immunoprecipitation (IP) in LNCaP and 22Rv1 cells. e Immunofluorescence (IF) images of AR and OCT4 in 22Rv1 cells. Bar =  10 μm. f Heatmap of ChIP-seq signals showing the aligned peaks of active promoter (K4me3) and enhancer (K4me1) markers at OCT4 center. g Motif discovery analysis of OCT4 peak positions (±50 bp). HOMER, publicly available program, was used with the default settings to find the enriched motifs and P-value. h Distribution of FOXA1/AR/OCT motifs around OCT4-binding peaks. i Composition plots of FOXA1 and AR-binding signals around the peaks of OCT4-binding in 22Rv1 cells. j Heatmap of aligned ChIP-seq signals of FOXA1, FOXM1, and AcH3K27 at OCT4 center. Source data are provided as a Source Data file.

Fig. 3. The role of OCT4 in superenhancer (SE) establishments to facilitate AR transcription program.

a KEGG-pathway analysis of AR/OCT4 target genes. AR- and OCT4-binding genes were closest genes to binding sites. We selected androgen-induced genes (>1.5 fold) repressed by siOCT4 #1 (fold < 0.7) by microarray. b Heatmap showing representative genes regulated by OCT4 and AR. c Identification of Superenhancers (SEs) in advanced PC cell model. ROSE analysis of AcH3K27 ChIP-seq signals in 22Rv1 cells. Representative genes in the vicinity of SEs are indicated. Signal tracks of SE signal (AcH3K27), RNA pol II reads on four representative SE genes. d Repressed (Fold < 0.5) or induced (Fold > 1.5) genes by OCT4 silencing in microarray data. Significant enrichment of repressed genes among SE genes were determined by two-sided chi-square test. ***P < 0.0001. e ChIP assay of AcH3K27 was performed (N = 3, technical replicates). 22Rv1 cells were treated with siControl, siOCT4 #1, or #2 for 48 h. **P < 0.01. two-sided t-test was used to compare siOCT4 with siControl. Data are presented as average ± S.D. f (Upper) Luciferase vectors (wild type or mutation of FOXA1 motif #1, #2 or ARE) were constructed by inserting OCT4-binding sequence to pGL3-promoter vector. (Lower) LNCaP cells were transfected with luciferase vectors complexed with HA-OCT4 or control vector. Cells were then treated with DHT or vehicle for 24 h and luciferase assay was performed (N = 3, biological replicates). Two-sided t-test was performed to compare with WT. Data are presented as average ± S.D. g Venus-FOXA1 and Venus-AR were co-transfected with mCherry or mCherry-OCT4 into 293T cells. The number of nuclear foci is counted and quantified (N = 5, biological independent cells). Two-sided t-test was performed. Data are presented as average ± S.D. h FRAP kinetic plots and representative images from pre- and post-bleaching cells (N = 4, biological independent cells). 293T cells were transfected with (1) Venus-FOXA1, Venus-AR, and mCherry-OCT4 (AR + FOXA1 + OCT4), (2) Venus-AR and mCherry-OCT4 (AR + OCT4), (3) Venus-FOXA1, Venus-AR and mCherry (AR + FOXA1), or (4) Venus-AR and mCherry (AR). Representative images of (1) and (3) are shown. Cells were treated with DHT for 2 h. Bar = 10 μm. Two-sided t-test was performed. Data are presented as average ± S.D. Source data are provided as a Source Data file.

Fig. 4. The effect of 1,6-HD and a nucleoside analog ribavirin (Riba) on OCT4-occupied enhancer activity.

a Immunoblots showing that AR, FOXA1, and OCT4 are precipitated by biotinylated isoxazole (b-isox). TRIM25, a representative RNA-binding protein, was used as a positive control. GAPDH was used as a negative control. WCLs: whole cell lysates, Sup: supernatants. b (Left) IF images of AR and OCT4 in 22Rv1 cells. Cells were treated with 1,6-hexandiol (HD) for 5 min, recovered for 1 h after 1,6-HD for 5 min, or ribavirin for 24 h. Bar = 10 μm. (Right) Quantification of OCT4 puncta (N = 4, biological independent cells) Two-sided t-test was performed. N.S not significant. Data are presented as average ± S.D. c ChIP analysis was performed in 22Rv1 cells treated with DHT for 24 h (N = 3, technical replicates). Cells were treated with 2,5-HD, 1,6-HD, or vehicle for 5 min before fixation. *P < 0.05, **P < 0.01, ***P < 0.001 vs Control. Two-sided t-test was performed. Data are presented as average ± S.D. d Venn diagram showing the overlap of SE regions with OCT4BSs in the absence or presence of ribavirin. e Heatmap of ChIP-seq signals showing the aligned peaks of RNA pol II and OCT4-binding signals at AR binding sites (ARBSs). Cont: control. f Composition plot of RNA pol II and OCT4-binding signal around the OCT4-binding peaks. Cells were treated with siAR or ribavirin. Source data are provided as a Source Data file.

Fig. 5. Distinct TF complex formation of OCT4 with NRF1 in AR-negative PC cells.

a HOMER analysis of OCT4-binding peaks in DU145 cells. b Distribution of OCT4BSs in the genomic regions around the closest gene was analyzed. c Heatmap of ChIP-seq signals shows the aligned peaks of NRF1, AcH3K27 signals at OCT4-binding peaks of control cells. d Microarray analysis was performed to analyze gene expression. Significantly enriched gene ontologies (GOs) among OCT4 -regulated genes with OCT4/NRF1 bindings (Fold < 0.5 by siOCT4 #1) in DU145 cells are summarized. e ChIP-seq views of NRF1 and OCT4 reads on three representative OCT4-regulated genes. Significant OCT4 and NRF1-binding regions (P < 1.0E-4) are shown by boxes. f The effect of 1,6-HD (hexandiol) on chromatin recruitments of NRF1 and OCT4. ChIP analysis was performed in DU145 cells (N = 3, technical replicates). Cells were treated with 2,5-HD, 1,6-HD or vehicle for 5 minutes before fixation. N.C: negative control. **P < 0.01. Two-sided t-test was performed. Data are presented as average ± S.D. g DU145 cells were treated with siControl, siOCT4 #1, or #2 for 48 h. ChIP analysis was performed using anti-NRF1 and anti-OCT4 antibodies (N = 3, technical replicates). **P < 0.01. Two-sided t-test was performed. Data are presented as average ± S.D. h FRAP kinetic plots from pre- and post-bleaching cells (NRF1:N = 5, NRF1 + OCT4:N = 6, biological independent cells). 293T cells were transfected with Venus-NRF1 or mCherry-OCT4. Two-sided t-test was performed. Data are presented as average ± S.D. i IF images of NRF1 and OCT4 in DU145 cells overexpressing OCT4 or control cells. OCT4-overexpressing cells were treated with vehicle or ribavirin for 48 h. Bar = 10 μm. Source data are provided as a Source Data file.

Fig. 6. OCT4 cooperates with NRF1 to activate its target.

a Growth inhibition of chemotherapy resistant PC cells by siRNA-mediated knockdown of OCT4 and NRF1. Chemoresistant PC cells (PC3-CR and DU145-CR) and their parental cells were treated with cabazitaxel (Cbz) 5 nM or vehicle. Cells were transfected with siControl, two siRNAs targeting OCT4 (#1, #2) or two siRNAs targeting NRF1(#1, #2). After 72 h incubation, cell viability was measured by MTS assay. N = 4, biological independent samples. Two-sided t-test was performed. *P < 0.05, **P < 0.01, ***P < 0.001. Data are presented as average ± S.D. b Representative colony formation assays and quantification of cells following 72 h treatment with cabazitaxel 10 nM. DU145-CR, PC3-CR, and parental cells were transfected with siControl, siOCT4 #1, or siNRF1 #1 (10 nM). N = 4, biological independent samples. Two-sided t-test was performed. ***P < 0.001. Data are presented as average ± S.D. Bar = 100 μm. c Representative colony formation assays and quantification of cells following 72 h treatment with cabazitaxel 10 nM. DU145 cells stably expressing HA-OCT4 (#1 and #2) or empty vector (Vec) (#1 and #2) were used. N = 4, biological independent samples. Two-way ANOVA was performed. ***P < 0.001. Data are presented as average ± S.D. Bar = 100 μm. d Immunoblotting of OCT4, HA, and NRF1 was performed in DU145 cells stably expressing HA-OCT4 (#1 and #2) or empty vector (Vec) (#1 and #2). e qRT-PCR analysis to measure mRNA levels of OCT4 target genes, OCT4 and NRF1 in DU145 cells stably expressing HA-OCT4 (#1 and #2) or empty vector (Vec) (#1 and #2). N = 3, technical replicates. Two-way ANOVA was performed. ***P < 0.001. Data are presented as average ± S.D. f Interaction of OCT4 with NRF1 in chemoresistnat PC cells. Immunoblots after immunoprecipitation (IP) by anti-NRF1 and OCT4 antibodies in DU145-CR and PC3-CR cells. Source data are provided as a Source Data file.

Fig. 7. Enhanced liquid droplet formation by collaborating TF complexes in vitro.

a Representative images of droplet formation of mCherry, Venus, mCherry-OCT4, Venus-FOXA1, and Venus-AR (total 10 μM) as indicated in droplet formation buffer with 125/500 mM NaCl and 10% PEG-8000. Bar = 10 μm. b Quantification of droplet formations of mCherry-OCT4, Venus-FOXA1, and Venus-AR in the presence of DHT. Average numbers (AR, AR + FOXA1 + OCT4: N = 4, others: N = 5) or sizes (N = 20) of droplets. One-way ANOVA and post-doc Dunnett’s test were performed. Data are presented as average ± S.D. c Effects of ribavirin and ribavirin-phosphate (P) on the droplet formation. Representative images of droplet formation of mCherry-OCT4, Venus-FOXA1, and Venus-AR (total 10 μM), as indicated in droplet formation buffer with 125 mM NaCl and 10% PEG-8000. Bar = 10 μm. Source data are provided as a Source Data file.

Fig. 8. The OCT4-centered phase separation could be a promising therapeutic target of aggressive PC.

a Immunohistochemistry (IHC) of OCT4, FOXA1, AR, CD56, and NRF1 in CRPC/NEPC tissues (N = 16). Bar = 100 μm. Representative images of NEPC pathology is shown. b Representative images of OCT4-associated TFs (in specimens with high or low expression of AR). +: strongly positive, +/−: weakly positive and −: negative. Bar = 50 μm. c The expression level of OCT4 protein detected by IHC correlated with NRF1 in AR low PC tissues (N = 80). Regression analysis was performed. d NRF1 protein level is upregulated in CRPC/NEPC tissues (N = 16) compared with localized PC tissues (N = 143) or locally invasive PC (N = 16). One-way ANOVA and followed Dunnett’s tests were performed. Data are presented as average ± S.D. e NRF1 is a prognostic factor for PC (N = 159). PC patients were classified into two groups according to the median value of NRF1 H-score. Survival analyses were performed using the Kaplan–Meir method and curves were compared by the log rank test. f Nude mice were inoculated with 22Rv1 cells. Mice were castrated after tumor development. Growth of tumors in nude mice treated with vehicle or ribavirin are shown (N = 6, biological independent animals). Two-sided t-test was performed. Data are presented as average ± S.D. g Immunohistochemical (IHC) images and labeling indexes of Ki67 in 22Rv1 tumor specimens (N = 3, biological independent samples). Two-sided t-test was performed. Data are presented as average ± S.D. Bar = 50 μm. h Immunoblots of AR and OCT4 after immunoprecipitation (IP) in 22Rv1 tumors. i Nude mice were inoculated with DU145-CR cells. (left) Mice were treated with vehicle, Cabazitaxel (Cbz), Ribavirin, Cbz +  Ribavirin. (right) Growth of tumors in nude mice treated with vehicle or ribavirin (Riba) are shown (N = 4, biological independent animals). One-way ANOVA and followed two-sided t-tests were performed. Data are presented as average ± S.D. j Immunoblots of NRF1 and OCT4 after immunoprecipitation (IP). k Schematic diagram of OCT4 function in PC progression. Source data are provided as a Source Data file.