CDK9 inhibition inhibits multiple oncogenic transcriptional and epigenetic pathways in prostate cancer

Abstract

Background: Cyclin-dependent kinase 9 (CDK9) stimulates oncogenic transcriptional pathways in cancer and CDK9 inhibitors have emerged as promising therapeutic candidates.

Methods: The activity of an orally bioavailable CDK9 inhibitor, CDKI-73, was evaluated in prostate cancer cell lines, a xenograft mouse model, and patient-derived tumor explants and organoids. Expression of CDK9 was evaluated in clinical specimens by mining public datasets and immunohistochemistry. Effects of CDKI-73 on prostate cancer cells were determined by cell-based assays, molecular profiling and transcriptomic/epigenomic approaches.

Results: CDKI-73 inhibited proliferation and enhanced cell death in diverse in vitro and in vivo models of androgen receptor (AR)-driven and AR-independent models. Mechanistically, CDKI-73-mediated inhibition of RNA polymerase II serine 2 phosphorylation resulted in reduced expression of BCL-2 anti-apoptotic factors and transcriptional defects. Transcriptomic and epigenomic approaches revealed that CDKI-73 suppressed signaling pathways regulated by AR, MYC, and BRD4, key drivers of dysregulated transcription in prostate cancer, and reprogrammed cancer-associated super-enhancers. These latter findings prompted the evaluation of CDKI-73 with the BRD4 inhibitor AZD5153, a combination that was synergistic in patient-derived organoids and in vivo.

Conclusion: Our work demonstrates that CDK9 inhibition disrupts multiple oncogenic pathways and positions CDKI-73 as a promising therapeutic agent for prostate cancer, particularly aggressive, therapy-resistant subtypes.

Fig. 1. CDK9 is commonly over-expressed and amplified in prostate cancer.

a Expression level of CDK9 in localized prostate cancer according to increasing Gleason score in the TCGA [26] cohort. Boxes extend from the 25th to 75th percentiles; the middle line is the median; top and bottom lines are minimum and maximum. Groups were compared using ANOVA and Tukey’s multiple comparison tests. b Kaplan–Meier curve showing estimated disease-free probability following radical prostatectomy in patients with high or low levels of CDK9 in the TCGA cohort. P value and hazard ratio were determined using log-rank tests. c Expression level of CDK9 in localized prostate cancer (TCGA) and adenocarcinoma (CRPC-Adeno) or neuroendocrine (NEPC) subtypes of CRPC [28]. CDK9 mRNA levels were normalized to ACTB mRNA levels. Groups were compared using ANOVA and Tukey’s multiple comparison tests. d CDK9 is commonly amplified in prostate cancer. Data is from The Cancer Genome Atlas (TCGA), Beltran [28] and Stand Up 2 Cancer (SU2C) [29] cohorts. e CDK9 gene amplification in the SU2C cohort is associated with increased mRNA expression. Groups were compared using ANOVA and Tukey’s multiple comparison tests. f CDK9 protein expression in clinical prostate samples (benign, primary tumors, or bone metastases) was measured by mass spectrometry [55]. Groups were compared using ANOVA and Tukey’s multiple comparison tests. g CDK9 protein expression is elevated in malignant compared to benign prostate tissues. Left, stacked bar graph showing the proportion of cells with low (1+), intermediate (2+), and high (3+) intensity CDK9 staining. Error bars are standard errors of the mean (s.e.m.). P value was determined using an unpaired t-test comparing the 3+ proportions. Middle, H scores of matched tumors and non-malignant samples (n = 8). P value was determined using a paired t test. Right, representative staining of non-malignant and Gleason grade 3 and 4 tumors (bar = 100 µm). In all panels, p values are: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Fig. 2. CDKI-73 causes cell death and suppresses anti-apoptotic pathways in prostate cancer cells.

a Chemical structure of CDKI-73. b Live-cell confluency analysis (Incucyte) demonstrates a dose-dependent reduction of LNCaP cell growth in response to CDKI-73. Error bars are ± s.e.m. of 6 biological replicates. Data is representative of 3 independent experiments. c Flow cytometry-based Annexin V/7-AAD apoptosis assay after 72 h of treatment with the indicated doses of CDKI-73. Data represents the mean of triplicate samples and are representative of 3 independent experiments. Error bars are s.e.m. Apoptotic cell proportions were compared to the vehicle using ANOVA and Tukey’s multiple comparison tests (*p < 0.05; ****p < 0.0001). d Representative Western blots showing decreased levels of RNAPII, pSer2-RNAPII, BCL-2, MCL-1, and MYC following treatment of LNCaP cells with the indicated doses of CDKI-73 or vehicle control (DMSO) for 10 or 24 h. GAPDH is shown as a loading control. e Levels of pSer2-RNAPII normalized to total RNAPII (both normalized to GAPDH) following treatment of LNCaP cells with the indicated doses of CDKI-73 or vehicle control (DMSO) for 10 or 24 h. This data is from an independent experiment to that shown in panel d. f Expression of genes encoding anti-apoptotic factors, as measured by qRT-PCR, following 4, 8, or 12 h of treatment with the indicated doses of CDKI-73 or vehicle control (DMSO). Gene expression was normalized to GAPDH, HPRT1, and ACTB; expression for DMSO (4 h) was set to 1. Error bars are ± s.e.m. of 3 biological replicates; P values (treatment compared to vehicle) were determined using ANOVA and Dunnett’s multiple comparisons tests. The data shown is representative of 3 independent experiments. In all panels: a or *p < 0.05; b or **p < 0.01; c or ***p < 0.001; d or ****p < 0.0001; NS, not significant.

Fig. 3. CDKI-73 has potent anti-tumor activity in diverse, clinically-relevant models of aggressive prostate cancer.

a Growth of LNCaP xenografts in mice treated with vehicle (n = 10) or CDKI-73 (n = 9). CDKI-73 (50 mg/kg) was administered orally on day 1 and then daily from day 3. For vehicle-treated mice, tumor size in 4 mice reached the ethical end-point at day 11; hence, this is the final time-point in this group. Graphs show the mean ± s.e.m. at each time-point. b Average body weight of mice harboring LNCaP xenografts and treated with vehicle or CDKI-73 (50 mg/kg). For vehicle-treated mice, tumor size in 4 mice reached the ethical end-point at day 11; hence, this is the final time-point in this group. c CDKI-73 inhibits proliferation and promotes cell death in prospectively collected human tumors grown as patient-derived explants (PDEs). PDEs (from n = 8 patients) were treated for 48 h. Ki67 and cleaved caspase-3 were evaluated by immunohistochemistry. Boxes extend from the 25th to 75th percentiles; the middle line is the median; top and bottom lines are minimum and maximum. Groups were compared using ANOVA and Tukey’s multiple comparison tests. d Representative immunohistochemistry staining of CDK9 in 3 independent PDXs, 27.1A, 201.1A and 305R. Source of tissue for the PDXs and AR status are shown in Supplementary Fig. 3. Met, metastasis. The bar is 50 µM. e CDKI-73 inhibits the growth of organoids grown from PDXs. Organoid viability was determined using Cell Titer Glo viability assays 3 days post-treatment. Data represents the mean ± s.e.m. of quadruplicate samples; treatment groups were compared to vehicles using ANOVA and Tukey’s multiple comparison tests. In all panels: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; NS, not significant.

Fig. 4. CDKI-73 reduces AR expression and activity.

a Volcano plot demonstrating genes differentially expressed (≥2-fold change and false-discovery rate (FDR) q value ≤ 0.05) in response to 4 h of treatment with 250 nM CDKI-73. b Hallmark pathways [75] suppressed (down) and activated (up) by CDKI-73, as determined using GSEA. NS, not significant. c Expression of AR mRNA and AR target genes (FKPB5 and TMPRSS2), as measured by qRT-PCR, following 4, 8 or 12 h of treatment with the indicated doses of CDKI-73 or vehicle control. Gene expression was normalized to GAPDH, HPRT1, and ACTB; expression for the vehicle (4 h) was set to 1. Error bars are ± s.e.m. of 3 biological replicates; P values (treatment compared to vehicle) were determined using ANOVA and Dunnett’s multiple comparisons tests. The data shown is representative of 3 independent experiments. d Representative Western blots showing levels of pSer81-AR and total AR following treatment of LNCaP cells with the indicated doses of CDKI-73 or vehicle control (DMSO) for 10 and 24 h. GAPDH is shown as a loading control. The values below the pSer81-AR blots are levels of pSer81-AR normalized to total AR (both normalized to GAPDH, with DMSO set to 1); the values below the AR blots are its levels normalized to GAPDH (with DMSO set to 1). In all panels: a or *p < 0.05; b or **p < 0.01; c or ***p < 0.001; d or ****p < 0.0001; NS, not significant.

Fig. 5. CDKI-73 induces epigenomic reprogramming and disrupts prostate cancer-associated enhancers.

a Volcano plot demonstrating genomic sites with differential H3K27ac ChIP-seq signal in response to 48 h of treatment with 150 nM CDKI-73. b Genomic distribution of H3K27ac lost sites, determined using ChIPseeker [45]. c GIGGLE plot showing overlap between H3K27ac lost sites and publicly available cistrome data for transcription factors and epigenetic regulators. Points indicate individual datasets, line indicates mean score. d Motifs enriched in the H3K27ac lost sites, were identified using HOMER [46]. P values represent enrichment over the genomic background. e Overlap between H3K27ac lost sites and LNCaP super-enhancers defined in a previous study [76]. Arrows indicate the number of sites in that particular dataset that overlap with the other. f CDKI-73 suppresses BRD4 oncogenic gene sets [68, 77], as demonstrated by GSEA. NES, normalized enrichment score. g Fold-change of mRNAs following CDKI-73 treatment for genes associated with or without super-enhancers (SE status + and -, respectively). LNCaP genes with SEs were obtained from the SEdb 2.0 database [78] using published H3K27ac ChIP-seq data (GSM1902615) [76].

Fig. 6. Synergism of CDKI-73 with BRD4 inhibition.

a Growth assays assessing the effect of CDKI-73, AZD5153, or the combination of both drugs in the indicated cell lines. Error bars are ± s.e.m. Treatment groups were compared using t-tests. b Representative Western blots showing levels of pSer2-RNAPII, MCL-1, and MYC in LNCaP cells following 24 h of treatment with CDKI-73, AZD5153, or the combination. GAPDH is shown as a loading control. c Expression of MCL1 and MYC mRNA, as measured by qRT-PCR, in LNCaP cells following 8 h of treatment with the indicated doses of CDKI-73, AZD5153 or the combination. Gene expression was normalized to GAPDH and ACTB; expression for vehicle (DMSO) was set to 1. Error bars are ± s.e.m. of 3 biological replicates; P values (treatment compared to vehicle) were determined using ANOVA and Tukey’s multiple comparisons tests. The data shown is representative of 2 independent experiments. d LNCaP tumor volume in mice treated with vehicle (n = 10), CDKI-73 (n = 9), AZD5153 (n = 10), or the combination (n = 9) following 1 or 11 days of engraftment. Single-agent treatments were administered orally on day 1 and then daily from day 3; for the combination treatment, drugs were administered orally on days 1, 3–7, and 10–11. Boxes are minimum to maximum; the line indicates the mean. Groups were compared using ANOVA and Tukey’s multiple comparison tests. e Synergistic activity of CDKI-73 and AZD5153 in 201.1A-Cx organoids. Organoid viability was determined using Cell Titer-Glo viability assays at 3 days post-treatment. Data represents the mean ± s.e.m. of 6 individual wells. f Synergy map (highest single agent (HSA) model) for the experiment shown in e, generated using SynergyFinder Plus [54]. g Summary of synergy between CDKI-73 and AZD5153 in 4 organoid models of aggressive prostate cancer. Mean synergy scores from 3 distinct synergy models (zero interaction potency (ZIP), HSA, and Bliss) are shown; the average of the 3 models is shown on the right. In all panels: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; NS, not significant.