Histone Demethylase KDM7A Regulates Androgen Receptor Activity, and Its Chemical Inhibitor TC-E 5002 Overcomes Cisplatin-Resistance in Bladder Cancer Cells

In order to investigate the possible role of KDM7A demethylase in the functioning of AR in bladder cancer cells, we first compared the AR expression levels in various bladder cancer cell lines, including 253J, RT4, T24, and J82. As expected, the levels of AR mRNA and protein in bladder cancer cells were quite low compared to that in LNCaP prostate cancer cells (Supplemental Figure S1A,B). Nonetheless, we were able to detect AR mRNA and protein in the bladder cancer cells that we tested, with the levels found to be comparable among them. Since a previous study has reported the development of cisplatin-resistant T24 cells [19], we selected this cell line for our experiments. We also included another bladder cancer cell line, J82, in order to demonstrate data reproducibility. To analyze the function of KDM7A histone demethylase in bladder cancer cells, we produced KDM7A knock-down bladder cell lines (T24 and J82), using a lenti-viral shRNA expression system. After antibiotics selection, efficient knock-down of the gene expression was confirmed by Western blotting and reverse transcriptase quantitative PCR (RT-qPCR) (Figure 1A). Since KDM7A is known to regulate AR activity in prostate cancer cells [32], we speculated that it may control AR activity as an epigenetic regulator in bladder cancer cells. Our studies showed that the levels of both AR mRNA and protein were regulated by KDM7A, before and after dihydrotestosterone (DHT) induction (Figure 1B, Supplemental Figure S2), even though we had expected changes only in the protein activity. A possible explanation could be the autoregulatory effect of AR on its own promoter [37]. In order to measure AR activity in KDM7A knock-down cells, the expression of previously reported downstream target genes of AR was screened using RT-qPCR [38,39,40,41,42,43]. The relative mRNA levels of these genes, compared with those of LNCaP, are listed in Supplemental Figure S1C. The PCR signals of six genes (KLK3, TMPRSS2, KLK4, IGF1R, VEGF, and MYC) were successfully amplified in bladder cancer cell lines. The mRNA levels of these AR target genes were elevated after DHT treatment, and this induction was reduced in KDM7A knock-down stable cells (Figure 1C).

Next, we analyzed the histone methylation status in KDM7A knock-down cells. The cell extracts from each of the cell lines were analyzed with specific antibodies for diverse histone methylation sites (Figure 2A). Among the sites tested, only H3K27 di-methylated lysine was elevated in KDM7A knock-down cells. The AR activity can be measured as the extent of binding of AR onto target promoters. We, therefore, analyzed the binding efficiency of AR to its target promoters, using the chromatin immunoprecipitation (ChIP) experiment. The T24 bladder cancer cells expressing control or KDM7A shRNA were treated with DHT, sonicated, and the chromatin was precipitated with AR antibody. The promoter binding by AR on the indicated genes was detected with IP PCR. The precise location of the PCR primers for each of the gene promoters is described in Supplemental Figure S3. Induction with DHT was shown to increase the binding of AR to these promoters, while this induction was abolished in the KDM7A knock-down cells (Figure 2B). We next attempted to investigate methylation statuses on AR responsive promoters in the KDM7A knock-down cells. Based on the changes in methylation (Figure 2A), we used H3K27 di-methyl-specific histone antibodies for immunoprecipitations, and H3K27 methylation was found to be increased in KDM7A knock-down bladder cells, before and after DHT induction (Figure 2C). Our data confirmed the molecular function of KDM7A on AR transcription factor activity in bladder cancer cells.

Since AR is known to be essential for cell growth in many cancers, we treated AR siRNA or enzalutamide, to confirm the growth inhibition effect on bladder cancer cell growth (Supplemental Figure S4). Next, we measured cell proliferation in both control and KDM7A knock-down cells. Although the morphology of KDM7A knock-down cells was not different from control cells (Supplemental Figure S5), the rate of cell proliferation in KDM7A knock-down bladder cancer cells was reduced compared to that of control shRNA-expressing cells (Figure 3A). When we seeded the same number of cells in cell culture dishes, knock-down cells showed a reduction in colony numbers and size compared to control cells (Figure 3B). We next observed the expression levels of cell cycle proteins in KDM7A knock-down cells (Figure 3C). Previous studies have reported that AR regulates cell cycle by controlling cyclin D1 [44], and cyclin B1 is a direct target of AR [45]. We observed a decrease in the cyclin B1 protein levels in KDM7A knock-down bladder cancer cells, which is consistent with the cell proliferation rate difference data in Figure 3A,B. It was observed that the decrease in cyclin D1 is not as obvious as that in cyclin B1 in our experimental conditions. Since AR is known to inhibit apoptosis induced by cytotoxic stimuli [46], we measured the levels of apoptotic proteins in KDM7A knock-down cells treated with the anti-cancer drug cisplatin (Figure 3D). We observed an increase in the PARP and caspase 3 cleavage products compared to control shRNA-expressing cells. These results strongly suggest that KDM7A regulates the rate of cell proliferation and drug-induced apoptosis in bladder cancer cells.

One of the major functions of AR in cancer progression is to facilitate cell migration and metastasis. It was observed that AR inhibition by enzalutamide affected the migration of bladder cancer cells (Supplemental Figure S6). To verify the effect of KDM7A in bladder cancer cell migration and metastasis, we measured migration and invasion in KDM7A knock-down T24 and J82 bladder cancer cells. One day after scratch, KDM7A knock-down cells showed decreased mobility when compared to control cells (Figure 4A). We demarcated the wound margin with a yellow line for better visualization, and added the original pictures and quantification of remaining scratched areas after the indicated times in Supplemental Figure S7. The cell invasion assay using Matrigel Transwell showed the decreased invasion of KDM7A knock-down cells compared to control cells (Figure 4B). To elucidate the molecular mechanism of this decrease in cell migration, we measured the expression of several epithelial-mesenchymal transition (EMT) markers in KDM7A knock-down cells. Although a decrease in the protein levels of N-cadherin and vimentin, the epithelial markers, was observed only in J82 cells (Figure 4C, Supplemental Figure S8), the mRNA expression levels of them were reduced in both T24 and J82 cells (Figure 4D). The protein and mRNA levels of mesenchymal marker E-cadherin were elevated in KDM7A knock-down cells in both T24 and J82 cells (Figure 4C,D, Supplemental Figure S8). The findings suggest that KDM7A is required for cell migration and EMT transition in bladder cancer cells.

The effect of AR induction on the cisplatin resistance process of bladder cancer has been previously reported [19]. Therefore, we wanted to explore the possibility of AR regulation as a target for overcoming the drug resistance of bladder cancer. Initially, we established a cisplatin-resistant T24 (CR-T24) bladder cancer cell line. After 2 months of exposure to increasing concentrations of cisplatin, we obtained T24 cells that survived in 2 μM cisplatin (Figure 5A). To compare the AR protein levels between original and CR-T24 cells, we compared their nuclear extracts, since the active form of AR protein occurs only in the nuclear fraction. We used the Lamin B1 antibody as the loading control of the nuclear extract, and for establishing the purity of the fraction, while GAPDH acted as the loading control of the cytosolic fraction. In addition to an increase in the total AR protein level, our study found that the level of active AR protein was higher in the nuclear fraction of CR-T24 when compared to the parental cells (Figure 5B). The protein level of KDM7A in CR-T24 cells decreased in the cytosolic fraction, but remained the same in nuclear fraction. The AR protein level in CR-T24 cells was found to be elevated before and after DHT treatment (Figure 5C), as reported previously [19]. The mRNA levels of several AR target genes also showed an increase, when compared to the parental T24 cells before and after DHT treatment (Figure 5D), suggesting that the elevated AR has functional activity. After confirming an increase in AR activity in CR-T24 cells, we utilized the KDM7A inhibitor TC-E 5002, in conjunction with AR antagonist enzalutamide, to evaluate the role of KDM7A on bladder cancer growth and drug resistance in terms of the AR pathway. Cell viability was tested in parental and CR-T24 cells in the presence of enzalutamide and/or TC-E 5002 (Figure 5E). In order to obtain a clearer picture of the changes in cell viability, we used a non-toxic dose and time of enzalutamide or TC-E 5002 when treated to parental T24 cells. CR-T24 cells were found to be more sensitive to a single treatment with enzalutamide or TC-E 5002 than parental cells. Moreover, when CR-T24 cells were treated with both enzalutamide and TC-E 5002, fewer cells survived compared to the single drug treatments. The anti-cancer effect of this co-treatment was significantly greater in CR-T24 compared to the parental T24 cells. Finally, we investigated the involvement of cellular signaling pathways involved in the anti-cancer effect of AR and/or KDM7A inhibitors on CR-T24 cells. Of the several pathways tested, Akt signaling pathway molecules were significantly decreased upon treatment with AR and/or KDM7A inhibitor. In our experiments, the total and phosphorylated protein levels of Akt and mTOR in T24 cells were decreased after enzalutamide and/or TC-E 5002 treatment. The changes were more significant in CR-T24 compared to the parental cells, and the effect was synergistically increased in co-treatment (Figure 5F; Supplemental Figure S9A). Upon cisplatin treatment, we observed the synergistic effect of two drugs on apoptosis signaling induction in the parental T24 cells, although the resistant cells did not show the apoptosis induction (Supplemental Figure S9B). For studying the changes in cell migration ability in CR-T24 cells, we performed a wound healing assay using both the cell lines treated with AR and/or KDM7A inhibitors (Supplemental Figure S10). The findings showed a decrease in the migration of cells upon treatment with AR and/or KDM7A inhibitors, in both original as well as CR-T24 cells.

To investigate the role of KDM7A in bladder tumor growth in vivo, we stably incorporated a luciferase-expressing vector into KDM7A shRNA-expressing bladder cancer cell lines and control cell line. After their inoculation into the bladders of NOD scid gamma (NSG) immune-deficient mice, the growth of the cancer cells was monitored using luciferase signal. The growth of bladder tumors was consistently higher in control cells compared to KMD7A knock-down cells (Figure 6A, Supplemental Figure S11). On the day of sacrifice, the tumors were extracted and the luminescence in control tumors was seen to be significantly higher than in KDM7A knock-down tumors (Figure 6B). Immunostaining of proliferation marker Ki-67 was used to evaluate the aggressiveness of the tumor (Figure 6C). The control tumor was positive for Ki-67, while the KDM7A knock-down tumor was found to be negative. The expression of vascular endothelial growth factor (VEGF) was detected in control tumor, but not in KDM7A knock-down tumors. These results implied that the growth and migration capability of bladder tumor is highly affected by the expression of KDM7A in vivo.

In the prostate study, we found that the KDM7A inhibitor TC-E 5002 effectively reduced tumor cell growth and migration in vitro [32]. In order to assess the in vivo effect of TC-E 5002 on bladder cancer development, we subcutaneously injected the T24 cells in the flanks of the NSG mice. After the tumor volume had reached 200 mm³, the mice were divided into 2 groups and injected with the vehicle or TC-E 5002 intraperitoneally daily for 8 days. After day 8, we observed a difference in the tumor sizes between the two groups (Figure 7A). The tumors excised from mice treated with TC-E 5002 weighed lesser than those excised from control mice (Figure 7B). Suppression of the AR activity by TC-E 5002 treatment was evident from the protein and mRNA expression levels of AR-dependent genes in the individual tumors (Figure 7 C,D). In tumors treated with TC-E 5002, the expression of the cell-cycle marker Ki-67 and VEGF protein were decreased, while the expression of the apoptotic DNA-fragmentation marker TUNEL was increased (Figure 7E). Our data exemplified the potential applications of TC-E 5002 in bladder cancer treatment in vivo.

In order to analyze KDM7A expression in bladder cancer patients, we performed immunohistochemistry (IHC) of KDM7A using tissue microarray. The results showed a higher level of KMD7A in tumor tissues compared to normal bladder tissue (Figure 8A). In the next step, patient tissue samples were collected, and the KDM7A protein and mRNA levels were analyzed. The protein (Figure 8B, Table 1) and mRNA (Figure 8C, Table 1) levels were significantly higher in tumor samples compared to normal bladder tissue from the same patients. Upon quantification of the signal from Western blot for KDM7A and its comparison with the tumor stages, the results were not statistically significant, which might be due to the small sample size (Supplemental Figure S12). Next, we evaluated the correlation between mRNA expression level of KDM7A and clinical outcome using a Kaplan–Meier plotter in public database (www.kmplot.com↗). We examined the prognostic value of KDM7A expression at each tumor stage, and in both men and women, in the bladder cancer database using KDM7A as the ‘key gene’ for data mining. High KDM7A mRNA expression was associated with significantly worse overall survival (OS) in men with stage 2 bladder cancer (Figure 8D). However, we were not able to identify a correlation in other stages of male cancer patients (Supplemental Figures S13 and S14), or in any stages of female patients.

RPMI-1640, DMEM, trypsin, anti-biotics, Trizol and Lipofectamine 2000 were purchased from Invitrogen (Carlsbad, CA, USA). Fetal bovine serum and culture media were obtained from HyClone Laboratories Inc. (South Logan, UT, USA). The detailed information of all primary antibodies is listed in Supplemental Table S1.

The T24, J82, and 293T cell lines were purchased from the American Type Culture Collection (Rockville, MD). T24 and J82 cells were cultured in RPMI-1640, and 293T cells for lentiviral package were cultured in DMEM medium at 37 °C in 5% CO₂, which was supplemented with 10% fetal bovine serum. For gene silencing, the control or KDM7A shRNA expressing lenti-virus packaging and stable cell line establishment were performed as described [32]. The oligo sequence used for KDM7A shRNA 01 cloning is 5′-CCGGTGGATTTGATGTCCCTATTATCTCGAGATAATAGGGACATCAAATCCAT TTTT-3′, and for shRNA 02 sequence is 5′-CCGGTTAGACCTGGACACCTTATTACTCGAGTAATAA GGTGTCCAGGTCTAATTTTT-3′. The oligo sequence for control shRNA cloning is 5′-CCGGCGTGA TCTTCACCGACAAGATCTCGAGATCTTGTCGGTGAAGATCACGTTTTT-3′. FUGW-luc vector (from the Molecular Imaging and Neurovascular Research Laboratory, Dongguk University Ilsan Hospital, Goyang, Korea) expressing cells were produced as described below. FUGW-luc vector was cut with XhoI enzyme and transfected into the T24 cell line, expressing either control vector or KMD7A shRNA vector. The cells with FUGW-luc incorporation were sorted with GFP channel using BD FACSAria II (BD Biosciences, Franklin Lakes, NJ, USA).

For the colony formation assay, 1000 cells were plated in 6-well plates. The cells were cultured for 14 days and stained with 0.1% crystal violet. The cell colonies were photographed, and the number of colonies comprising more than 50 individual cells was counted using SZX7 stereo microscope (Olympus, Tokyo, Japan). For the cell viability assay, cells (2000 to 3000 cells/well) were dispensed in 100 μL culture medium in a 96-well plate, and incubated for the indicated time. EZ-Cytox cell viability kit (Daeil-Lab, Seoul, Korea) solution (10 μL) was mixed with the culture medium in each well of the plate. Samples were incubated for 1 h at 37 °C, and the absorbance of each sample at 450 nm was measured using a microplate reader (PerkinElmer, Waltham, MA, USA).

The total cellular RNA was extracted using the Trizol reagent (Ambion, Austin, TX, USA), according to the manufacturer’s instructions. For the induction of AR activity, 5 nM 5α-dihydrotestosterone (DHT) was added after one day of serum deprivation. For each reverse-transcription reaction, 1 μg of total RNA was used for cDNA synthesis, using the MultiScribe Reverse Transcription Kit from Life Technologies (Carlsbad, CA, USA). RT-qPCR was performed using the EvaGreen qPCR Master Mix Kit from Applied Biological Materials Inc. (Richmond, BC, Canada) and a StepOne^TM Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). The quantity of 18S ribosomal RNA was measured as an internal control. The sequences of the primers used for RT-qPCR are listed in Supplemental Table S2.

The wound healing assay was performed on 100% confluent cells, plated into 6-well culture plates. Straight scratches were made by using a pipette tip. The cells were washed twice to remove debris, followed by the addition of fresh medium. The cells were incubated in a 5% CO₂ environment at 37 °C, and observed using a SZX7 stereo microscope at the indicated time. The scratched areas were measured using ImageJ program (ver. 1.43u; www.rsb.info.nih.gov/ij↗). For the invasion assay, cells (5 × 10⁴/well) were plated in the upper chambers of Transwells without serum, using Matrigel-coated polycarbonate membranes (Corning, Big Flats, NY, USA). The basal medium containing 10% fetal bovine serum was added into the lower chambers, as a chemoattractant for cell migration. After 48-h, non-migrated cells were removed from the upper chambers, while cells that migrated through chambers were fixed using 10% ethanol (Sigma-Aldrich). After cells were stained with the 0.01% crystal violet solution (Sigma-Aldrich), migrated cells were randomly counted in five different microscopic fields at 20× magnification.

The frozen tissues from bladder cancer patients were collected from Seoul Nation University Hospital Tissue Bank, with the approval of Institutional Review Board No. H-1004-037-315 (Date of approval: 06/11/2010). The demographic data of each patient are shown in Table 1. Tumor tissues and matching normal tissues from the same patient were identified from the pathology results. For Western blotting, 50–200 mg of tissues was ground in liquid nitrogen and lysed with RIPA buffer.

All animal experiments were performed in accordance with the Seoul National University Hospital institutional guidelines, under IACUC protocol No.16-0167-C2A0 (Date of approval: 07/20/2018). NOD scid gamma (NSG) mice were bred and maintained under specific pathogen-free (SPF) conditions. For generating orthotopic tumors, 1 × 10⁵ T24 human bladder cells expressing the indicated shRNA and Luciferase expression vector were injected into the bladder of six-week-old male NSG mice (n = 5 for each group). For injection, the cells were suspended with 100 μL of 50 % Matrigel (BD Biosciences) in complete media. The mice from each group were injected intraperitoneally with 150 mg/kg D-luciferin (Promega, Madison, WI, USA), 15 min before acquiring the image. After anesthetizing the mice using 1–3% isoflurane, the photons emitted from the tumor were detected with Xenogen IVIS imaging system 200 (Alameda, CA, USA) as described. The image acquisition period was 1 s. Living Image (Version 2.20, Xenogen) was used to quantify signals emitted from the regions of interest. The mice were sacrificed after 30 days of tumor implantation, and bladder tumor was fixed in 4% paraformaldehyde at 4 °C and embedded in paraffin. The specimens were subjected to IHC with the indicated antibodies. For subcutaneous xenografting, six-week-old male NSG mice were injected, in their lower flanks, with 1 × 10⁷ T24 cells in 100 μL of 50% (v/v) Matrigel (n = 5 for each group). When the average tumor volume reached 200 mm³, mice were randomly assigned into two groups and injected intraperitoneally everyday with vehicle (0.05 mL; 90% corn-oil and 10% DMSO (v/v)) or 10 mg/kg TC-E 5002 in the same vehicle, respectively. The tumors were measured every other day, and at the end of 8 days of treatment, the mice were sacrificed, and the tumors were excised, weighed, and either frozen or fixed in formalin for further analyses.

The cells (5 × 10⁶) and ground tissue (50–200 mg) were lysed in 1ml RIPA buffer (150 mM NaCl, 50 mM Tris-HCl [pH 7.2], 0.5% NP-40, 1% Triton X-100, and 1% sodium deoxycholate), containing a protease/phosphatase inhibitor cocktail (Sigma-Aldrich, St. Louis, MO, USA). For the induction of AR activity, 5 nM 5α-dihydrotestosterone (DHT) was added after one day of serum deprivation. The cell lysates were separated on sodium dodecyl sulfate-polyacrylamide gels and transferred to an Immobilon-P membrane (Millipore, Darmstadt, Germany). The membranes were blocked with 5% skim milk in 0.1% Tween-20 for 1 h, followed by overnight incubation at 4 °C with the indicated primary antibodies. The membranes were incubated with a horseradish peroxidase-conjugated secondary antibody (1:5000) for 1 h and developed using the ECL-Plus Kit (Thermo Scientific, Rockford, IL, USA).

Bladder cancer tissue microarrays purchased from SuperBioChips Laboratories (Seoul, Korea) were stained with an anti-KDM7A antibody. The slides were incubated with an anti-Rabbit IgG secondary antibody and hematoxylin and eosin (nuclear staining dye). The expression level of KDM7A from 25 different bladder tumors and 6 normal tissues was calculated and plotted. The expression of KDM7A positive cells was evaluated using the Cytoplasmic V2.0 algorithm in Aperio ImageScope software (Leica, Nussloch, Germany), and logistic regression analysis was used to compare the expression patterns between groups. Mouse tumor tissues were fixed in paraffin after formaldehyde fixation. Mouse tissue slides were deparaffinized and stained with indicated antibodies, and the slides were photographed under a Leica microscope (Wetzlar, Germany). Positive signals were counted from at least 3 different fields of the same area, and relative values were calculated.

The prognostic significance of the mRNA expression of KDM7A was evaluated using the Kaplan–Meier plotter (www.kmplot.com↗), an online database comprising gene expression data and clinical data. In order to assess the prognostic value of the KDM7A gene, the patient samples were divided into two cohorts according to the median expression level of the gene (high vs. low expression). We analyzed overall survival (OS) of bladder cancer patients by the Kaplan–Meier survival plot. KDM7A gene was uploaded into the database to obtain the Kaplan–Meier survival plot, in which the number-at-risk was shown below the main plot. Log rank p-value and hazard ratio (HR) with 95% confidence intervals were calculated and displayed on the webpage. We exported the plot data as a PowerPoint file.

All data were analyzed using Microsoft Excel 2010 software, unless otherwise stated. Continuous variables were analyzed using Student’s t-test if the data were normally distributed. All statistical tests were two-sided. Differences were considered significant in cases where p values were <0.05.