KDM4/JMJD2 Histone Demethylase Inhibitors Block Prostate Tumor Growth by Suppressing the Expression of AR and BMYB-Regulated Genes

Abstract

Histone lysine demethylase KDM4/JMJD2s are overexpressed in many human tumors including prostate cancer (PCa). KDM4s are co-activators of androgen receptor (AR) and are thus potential therapeutic targets. Yet to date few KDM4 inhibitors that have anti-prostate tumor activity in vivo have been developed. Here, we report the anti-tumor growth effect and molecular mechanisms of three novel KDM4 inhibitors (A1, I9, and B3). These inhibitors repressed the transcription of both AR and BMYB-regulated genes. Compound B3 is highly selective for a variety of cancer cell lines including PC3 cells that lack AR. B3 inhibited the in vivo growth of tumors derived from PC3 cells and ex vivo human PCa explants. We identified a novel mechanism by which KDM4B activates the transcription of Polo-like kinase 1 (PLK1). B3 blocked the binding of KDM4B to the PLK1 promoter. Our studies suggest a potential mechanism-based therapeutic strategy for PCa and tumors with elevated KDM4B/PLK1 expression.

Figure 1. Novel KDM4 inhibitors selectively inhibit prostate tumor growth

(A) Cell viability response of LNCaP cells to compound B3, A1, and I9. The chemical structures of B3, A1, and I9 are shown. (B) Dose-response curves of KDM4B, 4A, 4C, 4D, and KDM5A demethylase activities to various concentrations of B3, A1, and I9. (C) Expansions of ¹H-¹⁵N TROSY-HSQC spectra of KDM4B in the presence (red contours) and absence (black contours) of compounds B3 (left panel) or A1 (right panel). (D) IC₅₀ values reflecting the effects of B3 on the viability of various prostate immortalized epithelial and cancer cell lines, as well as cervical and breast cancer cell lines. Viable cells were quantified by MTT assays. The IC₅₀ values of the compounds were calculated by curve-fitting using Graphpad (N=6, mean ± SD). (E) LNCaP cells were cultured in the presence or absence of the androgen agonist R1881 and/or compound B3 as indicated. (F) PC3 xenograft tumor volume in SCID mice after treatment with vehicle or compound B3. N=9 (vehicle treated) and 10 (B3 treated), mean ± SD. *, p<0.05, compared over vehicle-treated tumors. (G) H&E of representative tumors treated with vehicle or B3.

Figure 2. KDM4 siRNA and inhibitors inhibited DNA replication of PCa cells

(A) Relative expression levels of KDM4 isoforms in normal and cancerous prostate tissue samples. Error bars are SEM. *, p<0.05, **, p<0.01. (B) Western blots of LNCaP cells transfected with control (ctl), KDM4A (4A), KDM4B (4B), or KDM4C (4C) specific siRNA and probed with antibodies against KDM4A, KDM4B, and GAPDH. (C) Growth curves of LNCaP (left panel) and PC3 cells (right panel) transfected with the indicated siRNA duplexes. (D) FACS analysis of LNCaP cells (left panel) and PC3 cells (right panel) transfected with control (ctl), 4A, or 4B specific siRNA. Two different siRNA duplexes were used. Representative ones are shown here. N=3, mean ± SD. *, p<0.05, compared to control-siRNA transfected cells. (E) FACS analysis of LNCaP and PC3 cells treated with vehicle DMSO, compound I9 (10 μM), A1 (5 μM), or B3 (1 μM). N=3, mean ± SD. *, p<0.05, compared to DMSO-treated cells. (E) Percentage of apoptotic cells treated with vehicle, A1, or B3 as assayed by Annexin V staining.

Figure 3. Inhibition of KDM4B by either siRNA or inhibitors down-regulated the expressions of cell cycle genes including PLK1

(A) Heat map of selected downregulated genes from LNCaP cells treated with compound I9, A1, or B3. Cell cycle genes that are also downregulated in KDM4B knockdown cells are highlighted in red. (B) qRT-PCR analysis of selected differentially expressed transcripts in compound-treated LNCaP and PC3 cells. mRNAs are expressed relative to vehicle (ctl) treated cells. (C) qRT-PCR analysis of selected genes downregulated in KDM4B siRNA-transfected LNCaP and PC3 cells. mRNAs were normalized against GAPDH and expressed relative to that in control siRNA-transfected cells. (D) Western blots of indicated proteins in LNCaP and PC3 cells treated without or with compound B3 (left panel), control, KDM4A, or KDM4B siRNA (right panel).

Figure 4. Cell cycle-dependent PLK1 expression requires KDM4 activity

(A) Western blot of indicated proteins from thymidine-released LNCaP cells. (B) Relative KDM4 demethylase activity at various time points of double thymidine released LNCaP cells. The activities were expressed relative to asynchronized (asy) cells. (C) Relative occupancy of H3K9me3, KDM4B, and BMYB on the PLK1 promoter at various time points of double thymidine-released LNCaP cells. The amounts of immunoprecipitated complex were normalized against DNA input and expressed relative to that from asynchronized cells. (D) Relative fold change of mRNA of the indicated genes from LNCaP cells described in (B) in the presence or absence of compound B3. mRNAs were expressed relative to the 0 h time point.

Figure 5. KDM4B interacts with BMYB and activates BMYB-targeted PLK1 transcription

(A) PLK1-luc reporter activities in cells transfected with full-length (1.5 kb), deletion (1.0, 0.5 kb) or point-mutant (x) of a PLK1-promoter construct and a vector that expresses KDM4B, KDM4B^H189G, or BMYB as indicated. (B) PLK1-luc activity in control, KDM4A (2A), or KDM4B (2B) knocked down cells that were transfected with various amount of BMYB. (C) LNCaP cell lysates were immunoprecipitated with control IgG or anti-KDM4B antibody, respectively. Immunoprecipitates were Western blotted with antibodies against BMYB and KDM4B. (D) 293T cells were transfected with Flag-BMYB and HA-KDM4B (left lower panel) or various deletion mutants of KDM4B (right panel). Whole cell lysates were immunoprecipitated with anti-Flag antibody and immunoprecipitates were Western blotted with anti-HA antibody. (E) ChIP-qPCR assays showing the relative occupancy of H3K9me3, KDM4B, BMYB, H3K36me3, and H3K4me3 at the PLK1 promoter in LNCaP cells treated with compound B3 (1 μM) or vehicle DMSO. N=3, mean ± SD. *, p<0.05.

Figure 6. Compound B3 inhibits AR and critical cell cycle genes in ex vivo solid human prostate tumors and growth of PC3 xenograft tumor in vivo

(A) Immunohistochemical staining of AR (upper panel) and H3K9me3 (middle panel), and H & E staining (lower panel) in sections of human prostate tumor in ex vivo cultures in the presence of vehicle DMSO or B3. (B) Relative fold change of mRNA of genes as indicated from tissues described in (A). mRNA levels were normalized against GAPDH and expressed relative to those in cultures in the presence of DMSO. N=5, mean ± SD.