Inhibition of Class I Histone Deacetylases 1 and 2 Promotes Urothelial Carcinoma Cell Death by Various Mechanisms Maria Pinkerneil1, Michele� J

Published OnlineFirst January 15, 2016; DOI: 10.1158/1535-7163.MCT-15-0618

Cancer Biology and Signal Transduction Molecular Cancer Therapeutics Inhibition of Class I Histone Deacetylases 1 and 2 Promotes Urothelial Carcinoma Cell Death by Various Mechanisms Maria Pinkerneil1, Michele J. Hoffmann1, Rene Deenen2,KarlKohrer€ 2, Tanja Arent3, Wolfgang A. Schulz1, and Gunter€ Niegisch1

Abstract

Class I histone deacetylases HDAC1 and HDAC2 contribute to apoptosis-like cell death in all UCCs. Clonogenic growth cell proliferation and are commonly upregulated in urothelial was cell line– and HDAC-dependently reduced, with double carcinoma. To evaluate whether specific inhibition of these knockdown of HDAC1 and HDAC2 being usually most effi- enzymes might serve as an appropriate therapy for urothelial cient. Class I HDAC-specific inhibitors, especially the more carcinoma, siRNA-mediated knockdown and specific pharmaco- specific HDAC1/2 inhibitors romidepsin and givinostat, signif- logic inhibition of HDAC1 and HDAC2 were applied in urothelial icantly reduced proliferation of all UCCs (IC50,3.36nmol/L– carcinoma cell lines (UCC) with distinct HDAC1 and HDAC2 4.59 mmol/L). Romidepsin and givinostat also significantly expression profiles. HDACs and response marker proteins were inhibited clonogenic growth of UCCs, with minor effects followed by Western blotting and qRT-PCR. Effects of class I on nontumorigenic controls. Intriguingly, these compounds HDAC suppression on UCCs were analyzed by viability, colony induced primarily S-phase disturbances and nonapoptotic forming, and caspase-3/7 assays; flow cytometry, senescence and cell death in UCCs. Thus, although both ways of inhibiting lactate dehydrogenase cytotoxicity assays; and immunofluores- HDAC1/2 share mechanisms and efficaciously inhibit cell cence staining. Whereas single knockdowns of HDAC1 or proliferation, their modes of action differ substantially. Regard- HDAC2 were impeded by compensatory upregulation of the less, combined inhibition of HDAC1/2 appears to represent other isoenzyme, efficient double knockdown of HDAC1 and a promising strategy for urothelial carcinoma therapy. HDAC2 reduced proliferation by up to 80% and induced Mol Cancer Ther; 15(2); 299–312. 2016 AACR.

Introduction Pharmacologic inhibition of HDACs is considered as a promising cancer therapy and may be particularly appropriate for Bladder cancer is the fifth most common cancer in the devel- cancers with altered chromatin regulation (9–11). The human oped world with about 4,00,000 new diagnosed cases per year and HDAC family includes 18 genes, which are classified into the 1,50,000 deaths worldwide (1). In industrial countries, about þ classical Zn2 -dependent HDACs (HDAC1-11; classes I, II, and 90% of bladder cancers are urothelial carcinomas, which can be IV) and sirtuins (class III; refs. 8, 12). Class I HDACs (HDAC1, further classified into muscle-invasive cancers, papillary cancers, HDAC2, HDAC3, and HDAC8) act primarily as transcriptional and carcinoma in situ with different clinical and molecular char- corepressors, but especially HDAC1 and HDAC2 are also associ- acteristics (2, 3). Integrative analysis of urothelial carcinomas has ated with active genes (12). Furthermore, class I HDACs affect revealed frequent mutations in chromatin regulator genes, includ- processes beyond transcription, in particular DNA replication, ing several affecting histone acetylation (3–5). Although The histone deposition, mitosis, DNA repair, and splicing (12–14). Cancer Genome Project has identified genetic alterations of Many HDAC inhibitors developed for clinical application histone deacetylase (HDAC) genes in other cancer entities, these are pan-inhibitors acting on a broad range of classical HDAC are rare in urothelial carcinoma (6–8). However, upregulation of isoenzymes, albeit to different extents (9–11). Because of the class I HDACs is common (6–8). diverse functions exerted by HDACs, inhibition of specific enzymes may be more suitable, especially selective targeting of HDAC1 and/or HDAC2. These predominantly nuclear proteins 1Department of Urology, Medical Faculty, Heinrich Heine University, Dusseldorf, Germany. 2Biological and Medical Research Center are highly related with partly but not completely overlapping (BMFZ), Medical Faculty, Heinrich Heine University, Dusseldorf, Ger- functions (15–17). Simultaneous ablation of HDAC1 and 3 many. Institute of Forensic Medicine, Medical Faculty, Heinrich Heine HDAC2 impairs cell-cycle progression and cell viability in all University, Dusseldorf, Germany. proliferating cells, including primary fibroblasts, B cells, and Note: Supplementary data for this article are available at Molecular Cancer hematopoietic cells (12, 15, 16, 18). Therapeutics Online (http://mct.aacrjournals.org/). In urothelial carcinoma, some potential for therapeutic efficacy Corresponding Author: Gunter€ Niegisch, Heinrich Heine University, Medical has been demonstrated for the pan-HDAC inhibitor SAHA and Faculty, Moorenstrasse 5, Dusseldorf 40225, Germany. Phone: 4921-1810-8776; the general class I inhibitor valproic acid (6, 19). In contrast, Fax: 4921-1810-4640; E-mail: [email protected] neither selective inhibition of HDAC8 nor HDAC6 was efficacious doi: 10.1158/1535-7163.MCT-15-0618 (20, 21). We therefore sought to evaluate the efficacy of specific 2016 American Association for Cancer Research. HDAC1 and HDAC2 inhibition in urothelial carcinoma cell lines

www.aacrjournals.org 299

Pinkerneil et al.

(UCC) by either siRNA-mediated knockdown or specific phar- (MTT, Thiazolyl blue formazan, M2128-G, Sigma Aldrich). For macologic inhibition with the class I HDAC inhibitors romidep- IC50 determination, defined concentration ranges of the inhibi- sin (FK228; ref. 22), givinostat (ITF2357; ref. 23), entinostat (MS- tors were used in three independent experiments, and viability 275; ref. 24), and mocetinostat (MGCD0103; ref. 25). We then was measured after 72 hours. Cell viability after siRNA-mediated compared the modes of action of siRNA-mediated knockdown HDAC knockdown was measured 72 hours after transfection via and pharmacologic inhibition with romidepsin and givinostat. total cellular ATP using the CellTiter-Glo Luminescent Cell Via- bility Assay (Promega). Caspase activity was quantified by the Methods caspase-Glo 3/7 assay and normalized to cell viability (Promega). Lactate dehydrogenase (LDH) release from damaged cells was Cell culture, siRNA transfection, and exposure to drugs measured via the Pierce LDH Cytotoxicity Assay Kit (#88954, For this study, different human UCCs reflecting the heteroge- Thermo Fisher Scientific). neity of urothelial carcinoma were used, namely in decreasing order of differentiation, RT-112, VM-CUB1, SW-1710, 639-V, and UM-UC-3. Control cell lines comprised normal urothelial control Colony forming assay and Giemsa staining cells (HBLAK, spontaneously immortalized normal human blad- For colony forming assays, cells were plated in 6-cm plates at a der cell line and TERT-NHUC, TERT-immortalized normal density of 500 to 1,500 cells per plate, 72 hours after siRNA human urothelial cells) and nonurothelial noncancerous cells transfection or 24 and 48 hours after HDAC inhibitor treatment. fi (HEK-293, immortalized human embryonic kidney cells and Colonies formed after 10 to 15 days were xed in methanol and HFF, human foreskin fibroblasts). UCCs, HFF, and HEK-293 were stained with Giemsa (Merck). cultured in DMEM GlutaMAX-I (Gibco) supplemented with 10% FCS (Biochrom). The cell line TERT-NHUC was cultured Flow cytometry in keratinocyte serum-free medium (Gibco) supplemented with Cell-cycle analyses of treated cells were performed after 72 0.25 ng/mL EGF, 12.5 mg/mL bovine pituitary extract and 1:100 hours (siRNA-mediated transfection) or 24 and 48 hours (phar- insulin-transferrin-selenium (Gibco), 0.35 mg/mL N-epinephrine, macologic inhibition) by staining the attached and supernatant and 0.33 mg/mL hydrocortisone (Sigma Aldrich; ref. 26). Cell cells with PI buffer containing 50 mg/mL propidium iodide, 0.1% lines were provided by Dr. M.A. Knowles (Leeds Institute of sodium citrate, and 0.1% Triton X-100 (27). For assessing apo- Cancer and Pathology, Leeds, United Kingdom), Dr. J. Fogh ptotic cell death and necrosis, cells were incubated with Annexin (Memorial Sloan-Kettering Cancer Center, New York, NY), V-FITC (#31490013 2, Immunotools), Annexin V binding Dr. B. Grossman (MD Anderson Cancer Center, Houston, TX), buffer, and PI (2 mg/mL in PBS). Flow cytometry analyses were and by the DSMZ (Braunschweig, Germany). For all cell lines, performed using a Miltenyi MACSQuant Analyzer (Miltenyi short tandem repeat (STR) profiling was recently performed by Biotec). standard DNA fingerprint analysis (Supplementary Table S1). The human bladder cell line HBLAK, obtained from CELLnTEC, was Senescence assay cultured in CnT-Prime Epithelial Culture Medium (CELLnTEC). Inhibitor-treated cells were washed twice in PBS and fixed for 5 All cells were cultured at 37 C and 5% CO2. minutes in 2% formaldehyde and 0.2% glutaraldehyde. Subse- For siRNA-mediated knockdown, UCCs were transfected with quently, cells were washed twice in PBS before being stained 10 nmol/L HDAC1 and/or HDAC2 Silencer Select validated overnight at 37C with fresh senescence associated b-gal (SA- siRNA (#4390824, HDAC1: s73; HDAC2: s6493) or a Silencer b-gal) staining solution [1 mg/mL X-gal (5-bromo-4-chloro-3- Select Negative Control #2 validated siRNA (#4390846) using indolyl-beta-D-galacto-pyranoside; Merck), 150 mmol/L NaCl, Lipofectamine RNAi MAX (Life Technologies) according to the 2 mmol/L MgCl2, 5 mmol/L K3Fe(CN)6, and 5 mmol/L K4Fe manufacturer's protocol and analyzed after a further 72-hour (CN)6]. Then, cells were washed in PBS, and images were taken cultivation. using the NIS-Elements software with a Nikon Eclipse TE2000-S Inhibitors were added 24 hours after seeding as a single dose of microscope (Nikon). the selective class I HDAC inhibitors romidepsin (S3020), givinostat (S2170), mocetinostat (S1122), or entinostat (S1053, all RNA isolation, cDNA synthesis, and qRT-PCR Selleck Chemicals) or the pan-HDAC inhibitor SAHA (suberoy- Total mRNA was isolated using the Qiagen RNeasy Mini Kit lanilide hydroxamic acid; LKT-V5734, Cayman Chemicals). As a (Qiagen) according to the manufacturer's protocol. cDNA syn- control in cell death measurements, cells were treated with bor- thesis was performed using the QuantiTect Reverse Transcription tezomib (50 nmol/L, S1013, Selleck Chemicals), actinomycin D Kit (Qiagen) with an extended incubation time of 30 minutes at (4 mg/mL, A1410, Sigma Aldrich), etoposide (25 mmol/L, S1225, 42C. qRT-PCR was performed using the QuantiTect SYBR Green Selleck Chemicals), or H2O2 (30 mmol/L, 107210, Merck). Inhi- RT-PCR Kit (Qiagen) with QuantiTect Primer assays (Qiagen) and bitors were dissolved in DMSO as a stock of 10 or 50 mmol/L. self-designed primers for the housekeeping gene TBP (TATA-box Control cells were treated with DMSO only. Q-VD-OPh (30 binding protein) as a reference (Supplementary Table S2) on the mmol/L, SML0063, Sigma Aldrich) and Necrox-2 (20 mmol/L, ABI 7500 Fast PCR instrument (Life Technologies). sc-391057, Santa Cruz Biotechnology) were used as pan-caspase and necrosis inhibitors. Western blot analysis Total protein was extracted by cell lysis for 30 minutes on ice in Determination of IC50, viability, apoptosis induction, RIPA buffer containing 150 mmol/L NaCl, 1% Triton X-100, 0.5% and LDH assay desoxycholate, 1% Nonidet P-40, 0.1% SDS, 1 mmol/L EDTA, Proliferation and IC50 were measured by 3-(4,5-dimethylthia- 50 mmol/L TRIS (pH 7,6), and a protease inhibitor cocktail zol-2-yl)-2,5-diphenyltetrazolium bromide dye reduction assay (10 ml/ml, #P-8340, Sigma Aldrich). Protein concentrations were

300 Mol Cancer Ther; 15(2) February 2016 Molecular Cancer Therapeutics

Inhibition of Class I HDACs in Urothelial Carcinoma

determined by bicinchoninic acid protein assay (Thermo Fisher bition) or P (corr) < 0.01 (Benjamini–Hochberg FDR adjusted; Scientific). After separation in SDS-PAGE gels and transfer to HDAC1/2 knockdown) were subjected to GO analysis in different polyvinylidene difluoride (PVDF) membranes (Merck Millipore), combinations using DAVID analysis software (DAVID Bioinfor- membranes were blocked with 5% nonfat milk or BSA in TBST matics Resources 6.7; ref. 30). GO biologic process terms signif- (150 mmol/L NaCl; 10 mmol/L TRIS, pH 7.4; and 0.1% Tween- icantly enriched at adjusted P < 0.05 were considered. 20), washed, and then incubated with primary antibodies at room temperature for 1 hour or at 4 C overnight. After washing, the Statistical analysis – membranes were incubated with horseradish peroxidase conju- Significance between groups was assessed by Student t test. P gated secondary antibody at room temperature for 1 hour. Bands values of <0.05 were considered as significant and P <0.01 as were visualized by Super Signal West Femto (Thermo Fisher highly significant. IC values and dose–response curves were fi 50 Scienti c) or WesternBright Quantum kit (Biozym). a-Tubulin approximated by nonlinear regression analysis using Origin 8.0 or GAPDH was detected as a loading control. Antibodies used are (Origin Lab). listed in Supplementary Table S3.

Extraction and analysis of histones Results Histones were acid-extracted according to a published protocol siRNA-mediated double knockdown of HDAC1 and HDAC2 (28). One microgram of each sample was used for Western blot significantly impairs urothelial carcinoma cell proliferation analysis with 15% SDS-PAGE gels and PVDF membranes (Merck and clonogenic growth Millipore) as described above using antibodies listed in Supple- First, endogenous HDAC1 and HDAC2 expression was down- mentary Table S3. regulated in UCCs by transiently transfecting siRNA or irrelevant control siRNA into RT-112, VM-CUB1, and SW-1710 (epithelial fl Immuno uorescence staining phenotype) or 639-V and UM-UC-3 (mesenchymal phenotype). fl Immuno uorescence analysis was performed on UCCs after These cell lines were selected from a larger panel reflecting the treatment with the HDAC inhibitors or DMSO for 24 and 48 heterogeneity of urothelial carcinoma (6). All cell lines expressed fi hours. Subsequent to xation with 4% formaldehyde, cells were relatively high levels of HDAC1 and HDAC2 compared with permeabilized with 0.3% Triton X-100 in PBS for 10 minutes at noncancerous cells (Fig. 1A; Supplementary Fig. S1). In all UCCs, room temeprature. Blocking was performed in 10% goat serum efficient knockdown of both mRNAs and proteins was achieved (DAKO), 0.3 mol/L glycine, and 0.1% Triton X-100 in PBS for 1 72 hours after transfection as shown by qRT-PCR and Western blot hour at room temeprature. The fixed cells were incubated over- fi analysis (Fig. 1B and C). Of note, ef cient knockdown of night at 4 C with the primary antibodies pH2A.X (1:50, #2577, HDAC1 consistently led to upregulation of HDAC2 mRNA Cell Signaling Technology) and 53-BP1 (1:100, clone BP18, 05- and protein in all cell lines. HDAC2 knockdown upregulated 725, Merck Millipore) followed by 1:500 diluted Alexa Fluor 488 both HDAC1 mRNA and protein, although not as distinctively. Goat Anti-Rabbit IgG antibody and 1:250 diluted TRITC-Goat No changes in mRNA or protein of HDAC3 were discernible þ Anti-Mouse IgG (H L) Conjugate (Life Technologies) for 1 hour following downregulation of HDAC1, HDAC2, or both (Fig. 1B at room temeprature. Cells were counterstained with 1 mg/mL 0 and C). At 72 hours after siRNA transfection, the number of DAPI (4 ,6-diamidino-2-phenylindole) for 3 minutes before viable cells was significantly diminished by knockdown of fl mounting with uorescence mounting medium (DAKO). Micro- HDAC1 only in VM-CUB1 (30%) and by knockdown of fi fi laments and nuclei of xed cells were stained with 14 nmol/L either HDAC1 (30%) or HDAC2 (50%) in 639-V cells. In Rhodamine Phalloidin and 1 mg/mL DAPI for 3 minutes. Images contrast, double knockdown of HDAC1 and HDAC2 resulted were taken with a Nikon Eclipse 400 microscope. in significant inhibition of cell growth (30 to 80%) in all fi Gene expression analyses cell lines, most ef ciently in VM-CUB1, 639-V and UM-UC-3 For microarray analyses, we used independently replicated (Fig. 1D). Double knockdown of HDAC1 and HDAC2 inhib- RNA preparations from each DMSO, romidepsin, givinostat, and ited clonogenic growth in UCCs almost completely, whereas SAHA-treated (24 hours) VM-CUB1 and UM-UC-3 cells, or cells HDAC1 or HDAC2 siRNAs alone inhibited clonogenic growth transfected with HDAC1 plus HDAC2-siRNA or negative control only weakly (Fig. 1E). siRNA as described above. Total RNA preparations were checked for RNA integrity by Agilent 2100 Bioanalyzer quality control. All HDAC1/2 double depletion induces apoptotic cell death in samples in this study showed high quality RNA Integrity Numbers urothelial cancer cell lines (RIN; 10). RNA was further quantified by fluorometric Qubit RNA To further characterize the impact of HDAC1 and/or HDAC2 assays (Life Technologies). Synthesis of cDNA and subsequent knockdown, detailed analyses of cell-cycle distribution and apo- biotin labeling of cRNA was performed according to the manu- ptosis were performed in VM-CUB1 and UM-UC-3, which differ facturer's protocol (30 IVT Plus Kit; Affymetrix, Inc.). For details see in HDAC1/2 expression but responded strongly to siRNA-medi- Heubach and colleagues 2015 (29). The significance threshold ated double knockdown. In addition, these cell lines were chosen was set to P ¼ 0.05 (Bonferroni FWER adjusted; pharmacologic due to their different properties (VM-CUB1, epithelial-like; UM- HDAC inhibitors) and P ¼ 0.01 (Benjamini–Hochberg FDR UC-3, mesenchymal-like) to obtain results that apply to a large adjusted; HDAC siRNA experiments), respectively. See GEO range of urothelial carcinoma cell lines/tumors. Cell-cycle anal- fl fi accession number GSE74478 for data and further description. yses by ow cytometry revealed a signi cant increase of the subG1 -fraction following HDAC1/2 double knockdown in both cell Functional classification of differentially expressed genes lines, but not in single HDAC1 or HDAC2 knockdowns, com- Differentially expressed genes of VM-CUB1 and UM-UC-3 with pared with irrelevant siRNA controls (Fig. 2A). Accordingly, P (corr) < 0.05 (Bonferroni FWER adjusted; pharmacologic inhi- significantly increased caspase-3/7 activity in VM-CUB1 and more

www.aacrjournals.org Mol Cancer Ther; 15(2) February 2016 301

Pinkerneil et al.

Figure 1. Expression and siRNA-mediated modulation of HDAC1/HDAC2 in UCCs and effects on proliferation and clonogenicity. A, relative mRNA expression of HDAC1 and HDAC2 in 15 UCCs compared with the urothelial control cells (UCon) NHUC-TERT and HBLAK and the nonurothelial control cells (nUCon) HEK-293 and HFF measured by qRT-PCR. B and C, effects of individual siRNA-mediated HDAC1, HDAC2, or HDAC1/2 knockdown in the UCCs RT-112, VM-CUB1, SW-1710, 639-V, and UM-UC-3 on mRNA (B) and protein expression levels of HDAC1 and HDAC2 and the concomitant HDAC3 (C) were compared with irrelevant controls (72 hours). The mRNA expression values were adjusted to TBP as a reference gene and are displayed on the y-axis. As a loading control, GAPDH was stained on each blot. D, relative cell viability in several UCCs after siRNA-mediated knockdown of HDAC1, HDAC2, or HDAC1/2 compared with irrelevant control (72 hours). The percentage of viable cells was measured by ATP assay and is displayed on the y-axis. The calculated significances of the treated value refer to the irrelevant control. E, Giemsa staining of colonies from RT-112, VM-CUB1, SW-1710, 639-V, and UM-UC-3 cells transfected with irrelevant siRNA and HDAC1, HDAC2, or HDAC1/2 siRNA (72 hours). P < 0.05 was regarded as significant and marked as , whereas P < 0.01 and P < 0.001 were defined as highly significant and marked as and .

302 Mol Cancer Ther; 15(2) February 2016 Molecular Cancer Therapeutics

Inhibition of Class I HDACs in Urothelial Carcinoma

strongly in UM-UC-3 was observed only after HDAC1/2 double profiling was performed in VM-CUB1 and UM-UC-3 cells. At a knockdown (Fig. 2B), as was cleavage of PARP and caspase-3. cut-off P (corr) value < 0.01 (Benjamini–Hochberg FDR adjust- Cleavage of caspase-8 was only observed in UM-UC-3 (Fig. 2C), ed), we identified 3449 (VM-CUB1) and 7036 (UM-UC-3) indicating a potential involvement of the extrinsic apoptotic differentially expressed transcripts (Supplementary Fig. S2E). pathway in this cell line. Apart from the increased sub-G1 The results are further described below under "HDAC1/2-spe- fraction, no significant changes in cell-cycle distribution were cific inhibition induces cell death". apparent in VM-CUB1, whereas the G1 fraction decreased slightly in UM-UC-3 (Fig. 2A). Effects on cyclin expression HDAC inhibitors selectively targeting HDAC1 and HDAC2 varied between the treatments and the two cell lines (Fig. show strong antineoplastic effects in UCCs 2D). Cyclin B1, cyclin D1, and cyclin E expression increased We next investigated the sensitivity of the five UCC lines to the in VM-CUB1, and cyclin A and cyclin B1 decreased in UM-UC-3 HDAC1/2 inhibitors romidepsin, givinostat, entinostat, and cells following HDAC1/2 knockdown. Cyclin D1 was elevated mocetinostat. Normal urothelial (HBLAK and TERT-NHUC) and following single HDAC2 knockdown in UM-UC-3 cells, where- nonurothelial (HEK-293 and HFF) cell lines with lower HDAC1/2 as cyclin E was diminished in both UCCs. Increased p21CIP1 expression (Fig. 1A, Supplementary Fig. S1) served as controls. expression was observed in both cell lines (Fig. 2D). No global Treatment with either compound inhibited proliferation of all changes in histone H3 and histone H4 acetylation were dis- UCCs, with romidepsin and givinostat being most efficient at cernible in VM-CUB1 or UM-UC-3 24 or 48 hours after knock- nanomolar (IC50, 3.36–6.47 nmol/L) or low micromolar (IC50, down of HDAC1 and/or HDAC2 (Fig. 2E). To get a detailed 0.39–0.63 mmol/L) concentrations, respectively (Fig. 3A, Supple- characterization of the processes and the molecular response mentary Table S4). Intriguingly, IC50 values for romidepsin and induced by HDAC1/2 double knockdown, mRNA expression givinostat were lower in normal cells than in UCCs; IC50 of

Figure 2. Effects of HDAC1, HDAC2, or HDAC1/2 knockdown on cell-cycle distribution and apoptosis in VM-CUB1 and UM-UC-3 cells. A, changes in cell-cycle distribution and amount of apoptotic cells (as sub-G1 fraction) after siRNA-mediated knockdown of HDAC1, HDAC2, or HDAC1/2 (72 hours) were measured by cell-cycle analysis using ﬂow cytometry. Irrelevant siRNA served as a control. The relative distribution of the fractions is displayed on the y-axis. Caspase-3/7 activity (B) and protein expression levels of cleaved PARP (cl. PARP), cleaved caspase-3 (cl. capase-3), and expression and cleavage of caspase-8 (C; full length 57 kDa; cleaved intermediate 41/43 kDa; active fragment 18 kDa) were measured after siRNA-mediated knockdown of HDAC1, HDAC2, or HDAC1/2 (72 hours) in VM-CUB1 and UM-UC-3 cells in comparison with an irrelevant control (co, set as 1). D, cyclin A, cyclin B1, cyclin D1, cyclin E, and p21 protein expression levels subsequent to HDAC1, HDAC2, and HDAC1/2 knockdown were determined by Western blot analysis in comparison with an irrelevant control in the UCCs VM-CUB1 and UM-UC-3 (72 hours). E, levels of acetylated histone H3 and H4 were determined in VM-CUB1 and UM-UC-3 cells 24 and 48 hours after siRNA transfection. HDAC inhibitor–treated cells served as positive control. After 72 hours, no histones could be extracted. As a loading control, a-tubulin, GAPDH, or histone H3 was stained on each blot.

www.aacrjournals.org Mol Cancer Ther; 15(2) February 2016 303

Pinkerneil et al.

Figure 3. Effects of class I HDAC-specific inhibitors on viability of VM-CUB1 and UM-UC-3 cells with activity of class I HDACs after pharmacologic inhibition with romidepsin and givinostat. A, several UCCs and control cells were treated with different concentrations of the class I HDAC-specific inhibitors (see Supplementary Table S4). Here, the dose–response curves of the UCCs VM-CUB1 and UM-UC-3 are shown for romidepsin, givinostat, mocetinostat, and entinostat after 72-hour inhibitor treatment as measured by MTT assay. The percentage of viable cells is displayed on the y-axis. B, levels of acetylated a-tubulin and histone H3 and H4 were determined in VM-CUB1 and UM-UC-3 cells after 24- and 48-hour treatment with romidepsin, givinostat, and SAHA. For inhibitor treatment, DMSO served as a solvent control. C and D, effects of romidepsin, givinostat, and SAHA treatment (24 and 48 hours) on HDAC1, HDAC2, and HDAC3 mRNA (C) and protein expression (D). P < 0.05 was regarded as significant and marked as , whereas P < 0.01 and P < 0.001 were defined as highly significant and marked as and .The calculated significances refer to the DMSO solvent control.

romidepsin was 0.57 and 0.89 nmol/L in TERT–NHUC and level of acetylated histones H3 and H4 in UCCs (Fig. 3B). HBLAK, respectively, and IC50 of givinostat was 0.15 and 0.12 Acetylation of a-tubulin, which depends mostly on HDAC6, was mmol/L in TERT-NHUC and HFF, respectively. The action of the not affected by romidepsin but strongly induced by givinostat, most potent inhibitors romidepsin and givinostat was then fur- suggesting that this compound inhibits HDAC6 activity too ther characterized compared with the pan-HDAC inhibitor SAHA (Fig. 3B). Pharmacologic inhibition of HDAC1 and HDAC2 was in VM-CUB1 and UM-UC-3 with HBLAK and HEK-293 cells as not counteracted by upregulation of class I HDAC protein (Fig. 3C controls. In the following experiments, all cells were treated with 3 and D). nmol/L romidepsin, 0.5 mmol/L givinostat, and 2.5 mmol/L SAHA Romidepsin and givinostat treatment induced an increase in (approximate IC50s in UCCs). Our group has characterized the cell size and cell ﬂattening in VM-CUB1 and UM-UC-3, whereas impact of SAHA on UCCs previously (6, 20, 21). All HDAC SAHA elicited only minor morphologic changes (Supplementary inhibitors, most strongly romidepsin, signiﬁcantly enhanced the Fig. S3). HBLAK control cells treated with romidepsin became

304 Mol Cancer Ther; 15(2) February 2016 Molecular Cancer Therapeutics

Inhibition of Class I HDACs in Urothelial Carcinoma

elongated with fibroblastoid morphology and numerous filo- of nontransformed HEK-293 cells only weakly, whereas SAHA podia and lamellipodia. In contrast, givinostat- and SAHA- impaired clonogenic growth significantly (Fig. 4A). Clonogeni- treated HBLAK cells assumed a senescence-like phenotype city of the normal UCC HBLAK was diminished by romidepsin becoming larger and flatter. HEK-293 cells showed slight mor- (24 and 48 hours) and SAHA (48 hours) but not by givinostat phologic changes in cell attachment and shape (Supplementary treatment. Fig. S3). Distinctive differences were observed in the ability As a further senescence assay, we performed SA-b-gal staining of romidepsin and givinostat versus SAHA to inhibit clono- for VM-CUB1 and UM-UC-3 and the control cell lines HBLAK and genic growth (Fig. 4A). After treatment with romidepsin or HEK-293. HDAC inhibitor treatment triggered cellular senescence givinostat for 24 hours or 48 hours, clonogenicity was strongly only weakly in VM-CUB1 and HEK-293 cells but strongly in impaired in UCCs, whereas SAHA exerted only minor effects. HBLAK control cells after treatment with romidepsin, givinostat, Conversely, romidepsin or givinostat diminished clonogenicity or SAHA (Fig. 4B).

Figure 4. Effects of HDAC1/2-speciﬁc inhibition via romidepsin and givinostat on clonogenic growth and senescence of UCCs. Giemsa staining of grown colonies (A) and senescence- associated b-gal staining (B; 24/48 hours) from inhibitor-treated VM- CUB1 and UM-UC-3 cells in comparison with the UCC HBLAK and nonurothelial control HEK-293 cells are shown. Cells were treated with DMSO, romidepsin, givinostat, and SAHA. Phase-contrast images are shown with a 20 magniﬁcation.

www.aacrjournals.org Mol Cancer Ther; 15(2) February 2016 305

Pinkerneil et al.

Concomitant pharmacologic targeting of HDAC1 and HDAC2 SAHA, showed a significant increase in proapoptotic nuclei. A abrogates growth of UCCs by disturbing cell-cycle progression significantly increased number of micronuclei was evident in To characterize the mechanisms underlying growth im- many cases after 24- and 48-hour treatment. A parsimonious pairment by pharmacologic HDAC1 and HDAC2 inhibition, explanation of these findings is that the HDAC inhibitors disturb cell-cycle analyses were performed in VM-CUB1 and UM-UC-3 and, in some cells, prolong mitosis in the first cell cycle, account- and the controls HBLAK and HEK-293 after treatment with 3 ing for the increase in mitoses and the G2–M fraction, subse- nmol/L romidepsin and 0.5 mmol/L givinostat (Fig. 5A). After quently leading to replication stress and S-phase disturbances in 24 hours, the G2–M-phase fraction increased strongly in VM- the second cell cycle (see discussion). CUB1 and UM-UC-3. After 48 hours, the cell-cycle histograms often became highly irregular with a serrated profile, indicating HDAC1/2-specific inhibition induces cell death strong disturbances of cell-cycle progression and prohibiting To further characterize the molecular response of UCCs reliable quantification of the cell-cycle distribution. These to treatment with HDAC inhibitors or HDAC1/2 double knock- effects were less pronounced under SAHA treatment. In HBLAK, down, mRNA expression profiling was performed in VM- the G2–M fraction decreased with all three HDAC inhibitors, CUB1 and UM-UC-3 cells. At a cut-off P (corr) value < 0.05 whereas in HEK-293 cells, the G2–Mfractionincreasedmost (Bonferroni FWER adjusted), we identified 6020 (romidepsin), strongly with SAHA (Fig. 5A). 4995 (givinostat), and 2233 (SAHA), respectively, differentially Despite the apparent increased G2–M-phase fraction, the G2 expressed transcripts in VM-CUB1 cells and 3780 (romidepsin), –M-phase cyclins A and B1 were diminished upon treatment with 3383 (givinostat), and 1642 (SAHA), respectively, differentially romidepsin or givinostat, especially after 48 hours. Instead, the S- expressed transcripts in UM-UC-3 cells. Of these, 1660 (VM- phase related cyclin E increased. Following givinostat treatment, CUB1) and 856 (UM-UC3) transcripts were common to all three the G1-phase related cyclin D1, too, accumulated, especially after inhibitors and 4211 (VM-CUB1) and 1962 (UM-UC-3) common 48 hours but disappeared after romidepsin treatment. The to romidepsin and givinostat. Two thousand five hundred and response to SAHA resembled that with givinostat (Fig. 5B). Thus, fifty one transcripts in VM-CUB1, and 1106 transcripts in UM-UC- the expression of cyclins was evidently perturbed, in keeping with 3 were differentially expressed in cells treated with romidepsin the flow cytometry data, suggesting profound disturbances in cell- and givinostat but not with SAHA (Supplementary Fig. S2A and cycle progression. All HDAC inhibitors induced p21CIP1, albeit to Supplementary Table S5). At a cut-off P (corr) value < 0.01 different extents and with different kinetics. In particular, induc- (Benjamini–Hochberg FDR adjusted), we identified 3449 (VM- tion of p21CIP1 appeared transient in givinostat-treated VM-CUB1 CUB1) and 7036 (UM-UC-3) differentially expressed transcripts and UM-UC-3 cells and SAHA-treated UM-UC-3 cells but was after siRNA mediated HDAC1/2 double knockdown (Supple- maintained in romidepsin-treated cells (Fig. 5B). Effects observed mentary Fig. S2E). Of these, 1417 transcripts were common to in HBLAK and HEK-293 were less distinctive. All HDAC inhibitors both cell lines. GO database analysis using DAVID software induced p21CIP1 in HBLAK cells, most strongly romidepsin revealed that genes significantly affected by romidepsin and (Fig. 5B); in HEK-293, this protein was undetectable with the givinostat in VM-CUB1 cells are involved in cell-cycle regulation, used antibody. especially mitotic processes, DNA replication and chromosome To characterize the mechanisms leading to these cell-cycle organization, DNA damage response, and metabolic processes. In disturbances, we determined whether pharmacologic HDAC1/2 UM-UC-3 cells, differentially expressed genes are significantly inhibition induces gH2A.X (phosphorylated histone H2A.X on associated with cell-cycle regulation (especially regulation of Ser139), an indicator of DNA damage, especially of double-strand cyclin–dependent kinase activity, mitotic cell cycle), regulation breaks (DSB; ref. 31, 32). The gH2A.X signal was increased after of cell proliferation, DNA damage checkpoints, and metabolic 24- and 48-hour treatment with romidepsin or givinostat (Sup- processes (Supplementary Fig. S2B). Accordingly, the 1,660 tran- plementary Fig. S4A). In a subpopulation of both cell lines, scripts affected commonly by romidepsin, givinostat, and SAHA gH2A.X was focally induced, typical of cells with an activated in VM-CUB1 are significantly associated with GO groups related DNA damage response (31). Another subpopulation displayed to cell cycle, mitotic processes, DNA replication, DNA damage high levels of uniform pan-nuclear gH2A.X staining across the response, cell proliferation, and different metabolic processes. nucleus (Supplementary Fig. S4A). As gH2A.X can also be induced Somewhat differently, the 856 affected transcripts in UM-UC-3 by other forms of DNA damage (31), we investigated whether the common to romidepsin, givinostat, and SAHA treatment were gH2A.X foci induced by HDAC inhibitors correspond to DSBs by predominantly assigned to processes involved in endogenous double staining for gH2A.X and 53-BP1, another DSB marker. 53- hormone or chemical stimulus, different metabolic processes, BP1 colocalized with defined gH2A.X foci in a subpopulation of and cell proliferation (Supplementary Fig. S2C). Notably, pro- VM-CUB1 and UM-UC-3 cells in UM-UC-3 cells more strongly cesses affected by romidepsin and givinostat treatment but not by after pan-HDAC inhibition by SAHA than by specific HDAC1/2 SAHA again mainly relate to cell-cycle regulation, proliferation, inhibition. Notably, cells with uniform pan-nuclear gH2A.X DNA replication, and mitosis in addition to metabolic processes staining lacked 53-BP1 foci (Supplementary Fig. S4B). (Supplementary Fig. S2D). These findings concord with the We next quantified the number of mitotic cells, proapoptotic pronounced disturbances of cell-cycle regulation seen in the nuclei and micronuclei in DAPI-stained cells (Supplementary Fig. previous analyses. S5). In VM-CUB1 cells, the number of mitoses decreased signif- In comparison, GO database analysis of genes significantly icantly after 48 hours upon romidepsin and givinostat treatment. affected by siRNA-mediated HDAC1/2 knockdown indicated in In UM-UC-3 cells, the number of mitoses increased upon treat- VM-CUB1 predominantly involvement in negative and positive ment with SAHA and romidepsin but not upon givinostat treat- regulation of cell death (apoptosis, antiapoptosis, and pro- ment. VM-CUB1 (48 hours) and especially UM-UC-3 (24 and 48 grammed cell death) but also in angiogenesis, development, and hours) cells treated with givinostat, but not with romidepsin or cell motion and migration. In UM-UC-3 cells, genes associated

306 Mol Cancer Ther; 15(2) February 2016 Molecular Cancer Therapeutics

Inhibition of Class I HDACs in Urothelial Carcinoma

Figure 5. Effects of HDAC1/2-specific inhibition via romidepsin and givinostat on cell-cycle distribution. A, changes in cell-cycle distribution after inhibitor treatment (24/48 hours) were measured by cell-cycle analysis using flow cytometry. DMSO served as a solvent control. The relative distribution of the fractions is displayed on the y-axis of the cell-cycle profiles. B, cyclin A, cyclin B1, cyclin D1, cyclin E, and p21 (, not detectable with used antibody in HEK-293) protein expression levels subsequent to HDAC1/2 inhibitor treatment were determined by Western blot analysis in comparison with DMSO control in the UCCs VM-CUB1 and UM-UC-3 and control cells HBLAK and HEK-293 (24/48 hours). www.aacrjournals.org Mol Cancer Ther; 15(2) February 2016 307

Pinkerneil et al.

with controlling of different metabolic processes were especially 2 (48 hours) or HDAC1/2 knockdown (72 hours). As shown in affected. Further significantly associated groups relate to DNA Supplementary Fig. S6B, Q-VD-OPh significantly rescued both damage response, cell death (apoptosis and programmed cell cell lines from cell death after HDAC1/2 knockdown in com- death), and regulation of cell cycle (Supplementary Fig. S2E). The parison with the irrelevant control. In contrast, the parallel 1,417 transcripts affected commonly in VM-CUB1 and UM-UC-3 treatment with Necrox-2 had no significant positive effect on cells after HDAC1/2 double knockdown are significantly associ- the viability of VM-CUB1 and UM-UC-3 after HDAC1/2 knock- ated with GO groups related to cell cycle, regulation of cell down. Both cell death inhibitors protected against treatment proliferation, angiogenesis, regulation of cell death, and meta- with etoposide in UM-UC-3 cells, and only Q-VD-OPh was bolic processes (Supplementary Fig. S2E). protective in VM-CUB1 cells. However, following pharmaco- The most striking difference between the two ways of HDAC1/2 logic inhibition of HDACs positive effects on the viability of suppression seems therefore that in inhibitor treated cells, more VM-CUB1 (weak) and UM-UC-3 (strong) were only observed cell-cycle related processes and in cells with HDAC1/2 knock- in the combined treatment with givinostat and Necrox-2 (Sup- down, more processes controlling cell death are affected. To plementary Fig. S6B). highlight the similarities between the two ways, we performed DAVID analysis of 167 transcripts affected commonly in both cell lines after romidepsin/givinostat treatment and HDAC1/2 Discussion double knockdown (Supplementary Fig. S2F). This gene set is In this study, we present a systematic analysis of HDAC1 and significantly associated with GO groups related to cell cycle (cell- HDAC2 as potential therapeutic targets in urothelial cancer. cycle arrest, mitotic cell cycle, and cell-cycle phase), cell prolifer- We applied siRNA-mediated knockdown and pharmacologic ation, immune response, and regulation of cell death. Among inhibition in a representative set of bladder cancer cell lines these 167 transcripts, established marker genes for HDAC inhi- covering a broad range of HDAC1 and HDAC2 expression as bition as well as novel genes in UCCs can be defined (Supple- well as to urothelial and nonurothelial benign control cells. mentary Table S6). The most striking consistent response was the Combined siRNA-mediated targeting of HDAC1 and HDAC2 induction of CDKN1A/p21, but genes like CTGF and SOX4 were diminished short-term UCC proliferation by up to 80%, also strongly and consistently induced. depending on the cell line, and suppressed clonogenic growth We then investigated which type of cell death was induced of all UCC lines. For both endpoints, combined targeting of following pharmacologic HDAC inhibition in VM-CUB1 and HDAC1/2 was far more efficacious than knockdown of HDAC1 UM-UC-3 and the controls HBLAK and HEK-293 (Supplemen- or HDAC2 individually. Our results concurred with observa- tary Fig. S6A, Fig. 6). Activity of caspase-3/7 increased slightly tions in many tissue and cell types whereby targeted single in UM-UC-3 following treatment with romidepsin, givinostat, deletion of these enzymes did not significantly affect prolifer- or SAHA but in VM-CUB1 cells only upon SAHA and givinostat ation, differentiation, and cell survival (12, 15) but combined treatment after 24 hours. In the control cells, a slight significant targeting yielded strong effects (15, 16, 18). This finding is increase occurred only in HBLAK cells after 48-hour SAHA plausible as the functions of the two proteins are largely treatment (Supplementary Fig. S6A). Cleaved PARP was detect- overlapping, and downregulation of one enzyme leads to able especially after 48-hour treatment with romidepsin and, compensatory increases in the other (17, 33, 34), as also less prominently, with givinostat or SAHA in VM-CUB1 and observed in the UCCs. UM-UC-3 cells (Fig. 6A). In urothelial HBLAK cells, cleavage of UCCs were highly sensitive to all four reportedly HDAC1/2- PARP occurred after romidepsin treatment only. Notably, specific inhibitors, with low nanomolar (romidepsin) or low cleavage of caspase-3 (Fig. 6A) and caspase-8 (Supplementary micromolar (givinostat < mocetinostat < entinostat) IC50 values. Fig. S6C) was undetectable in any cell line after pharmacologic The substantially lower IC50 value of romidepsin is in keeping HDAC inhibition. Significant release of LDH, indicating with observations in other cancer types. Romidepsin itself has a necrotic cell death, was detected only in SAHA-treated UM- high affinity for class I HDACs and is further activated by reduc- UC-3 cells (48 hours) but not in romidepsin- and givinostat- tion after uptake into the cells (22). In comparison with givino- treated cells (Fig. 6B and C). Annexin V/PI staining suggested a stat, entinostat and mocetinostat, the activated form of romidep- mixture of apoptotic and necrotic cell death, with a predom- sin shows the highest cell-free assay–based activity against inant increase in early apoptotic and late apoptotic/necrotic HDAC1 and HDAC2 likely accounting for its most efficient effect cells especially after 48-hour treatment with romidepsin and on proliferation of UCCs (22, 23, 25, 35). Notably, the normal givinostat in VM-CUB1 and UM-UC-3 cells (Fig. 6C). Notably, urothelial cell lines HBLAK and TERT-NHUC were more sen- in UM-UC-3 cells, the necrotic fraction was also enriched sitive to the inhibitors in short-term assays, but their long-term significantly after 24-hour (givinostat) and 48-hour (romidep- growth was much less affected. In these cells, the inhibitors sin, givinostat, and SAHA) treatment. In HDAC inhibitor appear to act in a reversible cytostatic manner, but induced a treated VM-CUB1 cells, the late apoptotic/necrotic fraction senescence-like phenotype in many cells, according to mor- was enriched up to 8.9% (romidepsin, 48 hours) and 8.7% phology and SA-b-Gal staining. This difference between normal (givinostat, 48 hours); higher increases were seen in UM-UC-3 and transformed cells has been observed in other cell types too cells with 9.9% late apoptotic/necrotic cells upon romidepsin andisascribedtotheabilityofnormalcellstoreactmore and 11.4% late apoptotic/necrotic cells upon givinostat treat- appropriately to damage caused by the HDAC inhibitors by ment after 48 hours. cell-cycle arrest and DNA repair (36). To further elucidate the mechanisms of cell death, we mea- Due to the sensitivity of UCCs to these inhibitors, we decided to sured viability of VM-CUB1 and UM-UC-3 after applying the evaluate romidepsin and givinostat more closely. Moreover, there pan-caspase inhibitor Q-VD-OPh and necrosis inhibitor is little published preclinical data on these compounds in solid Necrox-2 together with pharmacologic inhibition of HDAC1/ tumors (22, 37–39).

308 Mol Cancer Ther; 15(2) February 2016 Molecular Cancer Therapeutics

Inhibition of Class I HDACs in Urothelial Carcinoma

Figure 6. Analysis of cell death mechanisms after specific pharmacologic HDAC1/2 inhibition via romidepsin and givinostat. A, protein expression levels of cleaved caspase-3 (cl. caspase-3) and cleaved PARP (cl. PARP) were measured after treatment with romidepsin, givinostat, and SAHA for 24 and 48 hours in VM-CUB1, UM-UC-3, HBLAK, and HEK-293 cells. For detection of cleaved caspase-3, Caspase-3 Control Cell Extract served as a positive control (#9663, Cell Signaling Technology, Inc.). B, LDH assay indicating necrotic cell death was performed after HDAC inhibition (24 and 48 hours) in VM-CUB1 and UM-UC-3 cells. The percentage of LDH release (normalized on lysis control) is displayed on the y-axis. Bortezomib and actinomycin D treatments served as additional controls. C, induction of necrosis and apoptosis was additionally analyzed in VM-CUB1 and UM-UC-3 cells by combined Annexin V and PI staining with subsequent flow cytometry after treatment with romidepsin, givinostat, and SAHA for 24 and 48 hours. The percentages in the figure indicate viable cells (bottom left), necrotic cells (top left), early apoptotic cells (bottom right), and late apoptotic/necrotic cells (top right). P < 0.05 was regarded as significant and marked as , whereas P < 0.01 and P < 0.001 were defined as highly significant and marked as and . The calculated significances refer to the DMSO solvent control.

www.aacrjournals.org Mol Cancer Ther; 15(2) February 2016 309

Pinkerneil et al.

Both siRNA-mediated knockdown and pharmacologic inhibi- (13, 14), and cause DNA damage (14, 36). All three effects might tion of HDAC1 and HDAC2 suppressed short-term proliferation be relevant in our context. However, although we found an and clonogenicity of UCCs and activated cell death. Thus, target- increased number of gH2A.X foci and gH2A.X positive nuclei, few ing HDAC1/HDAC2 seems promising for urothelial cancer ther- foci were also positive for p53-BP1, unlike true DSB foci. The most apy. A global comparison of gene expression revealed the induc- parsimonious explanation for our observations is therefore that tion of CDKN1A/p21, which is the best established marker for HDAC inhibitor treatment in UCCs induced disturbances espe- HDAC targeting across various cancer entities (40, 41) as the most cially of mitosis, leading initially to a prolonged G2–M-phase. striking consistent response between knockdown and pharma- Although not fully elucidated, HDAC1 and HDAC2 appear to exert cologic inhibition but revealed several novel genes that deserve important functions in mitosis, which would be abrogated by investigation as potential markers and mediators of HDAC inhi- pharmacologic inhibition (47). As tumor cells with defective bition. Nevertheless, substantial differences were evident between checkpoints eventually exit from mitosis regardless of correct the effects of the two approaches towards targeting HDAC1/2. completion (48) and UCCs are defective at the G1–S-checkpoint, Prominently, global histone acetylation increased substantially UCCs would then inappropriately reenter S-phase, resulting in following pharmacologic inhibition but not after siRNA-mediat- replication stress and cell death. In this interpretation, the pan- ed knockdown. There are three obvious explanations for these gH2A.X–positive nuclei may reflect cells unable to complete S- differences. First, knockdown depletes the proteins and thereby phase. Direct effects of the HDAC inhibitors on DNA replication, disturbs assembly of repressor complexes (Sin3, NuRD, and reported for SAHA (14), may contribute, but we have no direct CoREST) containing HDAC1 and HDAC2, whereas inhibition evidence for these. The HDAC inhibitor trichostatin A may delay G2 of enzymatic activity should leave these complexes intact. Second, –M transition and mitotic progression leading to aberrant chro- pharmacologic inhibition could be more stringent as even highly mosome segregation, failed cytokinesis, multi-nucleation (49), efficient siRNA-mediated knockdown may incompletely abrogate and in some cases death by mitotic catastrophe (50). The latter HDAC1/2 expression. Third, the drugs could have additional off- was not obvious in UCC treated with givinostat or romidepsin. target effects. For instance, romidepsin has been reported to In summary, HDAC1 and HDAC2 combined appear to be inhibit PI3K signaling (42). In fact, some differences between reasonable targets for urothelial cancer treatment. While the effects of romidepsin and givinostat on UCCs could be siRNA-mediated knockdown caused preferentially apoptotic cell explained by off-target activities. Especially a-tubulin acetylation, death in urothelial cancer cell lines, pharmacologic inhibition by which is catalyzed by HDAC6 (43), was induced by givinostat and romidepsin and givinostat severely disrupted cell-cycle progres- SAHA but not by romidepsin. Although labeled as "specific class I sion with irreversible arrest and eventual cell death. As observed HDAC inhibitors", givinostat and romidepsin both may target for other HDACi in other cell types, normal cell lines were non–class I HDACs to some degree (44). similarly sensitive to these inhibitors in the short run but arrested Differences between siRNA-mediated knockdown and phar- appropriately and maintained their ability for long-term prolif- macologic inhibition are particularly evident in the cell death eration. This difference might provide a window for therapeutic mechanisms and cell-cycle response. The combined knockdown application. of HDAC1/2 led to a significantly increased sub-G1 fraction and enhanced caspase-3/7 activity indicating apoptotic cell death, as Disclosure of Potential Conflicts of Interest in other cell types (15, 16, 33). In addition, cell death was partially G. Niegisch reports receiving a commercial research grant from 4SC. No fl rescued by concurrent treatment with the pan-caspase inhibitor potential con icts of interest were disclosed by the other authors. Q-VD-OPh. These findings concord with the prominence of genes involved in cell death regulation in the expression array analysis of Authors' Contributions Conception and design: M. Pinkerneil, M.J. Hoffmann, K. Kohrer,€ W.A. Schulz, the HDAC1/2 double knockdown. No G1 arrest was observed, which may be difficult to achieve in UCC due to their impaired G. Niegisch Development of methodology: M.J. Hoffmann, G. Niegisch RB1 and p53 functions (2). The degree of apoptosis achieved by Acquisition of data (provided animals, acquired and managed patients, romidepsin or givinostat treatment, as indicated by subG1 frac- provided facilities, etc.): M. Pinkerneil, M.J. Hoffmann, R. Deenen, K. Kohrer,€ tion, caspase activity, Annexin-V staining, and PARP cleavage, was G. Niegisch more circumscribed. We have previously reported similar results Analysis and interpretation of data (e.g., statistical analysis, biostatistics, for SAHA, suggesting that HDAC inhibitors in UCCs do not act computational analysis): M. Pinkerneil, R. Deenen, W.A. Schulz, G. Niegisch primarily by inducing apoptosis (6, 45). Accordingly, caspase Writing, review, and/or revision of the manuscript: M. Pinkerneil, M.J. Hoffmann, W.A. Schulz, G. Niegisch inhibitors had little effect on cell viability following HDAC Administrative, technical, or material support (i.e., reporting or organizing inhibitor treatment, whereas a necrosis inhibitor was partly pro- data, constructing databases): M. Pinkerneil, R. Deenen, K. Kohrer,€ G. Niegisch tective. Therefore, a necrotic mechanism may rather be involved. Study supervision: G. Niegisch We suspect that cell death is secondary to the profound cell-cycle Other (STR analysis): T. Arent disturbance by romidepsin and givinostat, which was milder following HDAC1/2 knockdown. This disturbance was evident Grant Support in the Western blot analysis for cyclins and dominated the This work was supported by a grant from the Deutsche Forschungsge- microarray gene expression data but was most striking in the meinschaft to G. Niegisch (NI 1398/1-1). PhD position of M. Pinkerneil was funded by grant of G. Niegisch (NI 1398/1-1). flow cytometry analyses. After 24 hours, the G2–M fraction The costs of publication of this article were defrayed in part by the payment of increased strongly and after 48 hours (i.e., after two cell-cycle advertisement fi page charges. This article must therefore be hereby marked in lengths), the pro les became highly irregular. The number of accordance with 18 U.S.C. Section 1734 solely to indicate this fact. mitoses first increased and then decreased, while micronuclei appeared. In addition to their epigenetic effects on gene expres- Received July 23, 2015; revised November 24, 2015; accepted December 14, sion, HDAC inhibitors disturb mitosis (46), DNA replication 2015; published OnlineFirst January 15, 2016.

310 Mol Cancer Ther; 15(2) February 2016 Molecular Cancer Therapeutics

Inhibition of Class I HDACs in Urothelial Carcinoma

References 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer cytokines in vitro and systemic inflammation in vivo. Mol Med 2005; statistics. CA Cancer J Clin 2011;61:69–90. 11:1–15. 2. Knowles MA, Hurst CD. Molecular biology of bladder cancer: new 24. Saito A, Yamashita T, Mariko Y, Nosaka Y, Tsuchiya K, Ando T, et al. A insights into pathogenesis and clinical diversity. Nat Rev Cancer 2015; synthetic inhibitor of histone deacetylase, MS-27-275, with marked in vivo 15:25–41. antitumor activity against human tumors. Proc Natl Acad Sci U S A 3. Cancer Genome Atlas Research N. Comprehensive molecular char- 1999;96:4592–7. acterization of urothelial bladder carcinoma. Nature 2014;507: 25. Fournel M, Bonfils C, Hou Y, Yan PT, Trachy-Bourget MC, Kalita A, et al. 315–22. MGCD0103, a novel isotype-selective histone deacetylase inhibitor, has 4. Guo G, Sun X, Chen C, Wu S, Huang P, Li Z, et al. Whole-genome and broad spectrum antitumor activity in vitro and in vivo. Mol Cancer Ther whole-exome sequencing of bladder cancer identifies frequent alterations 2008;7:759–68. in genes involved in sister chromatid cohesion and segregation. Nat Genet 26. Swiatkowski S, Seifert HH, Steinhoff C, Prior A, Thievessen I, 2013;45:1459–63. Schliess F, et al. Activities of MAP-kinase pathways in normal 5. Gui Y, Guo G, Huang Y, Hu X, Tang A, Gao S, et al. Frequent mutations of uroepithelial cells and urothelial carcinoma cell lines. Exp Cell Res chromatin remodeling genes in transitional cell carcinoma of the bladder. 2003;282:48–57. Nat Genet 2011;43:875–8. 27.NicolettiI,MiglioratiG,PagliacciMC,GrignaniF,RiccardiC.Arapid 6.NiegischG,KnievelJ,KochA,HaderC,FischerU,AlbersP,etal. and simple method for measuring thymocyte apoptosis by propidium Changes in histone deacetylase (HDAC) expression patterns and iodide staining and flow cytometry. J Immunol Methods 1991;139: activity of HDAC inhibitors in urothelial cancers. Urol Oncol 2013; 271–9. 31:1770–9. 28. Shechter D, Dormann HL, Allis CD, Hake SB. Extraction, purification and 7. Poyet C, Jentsch B, Hermanns T, Schweckendiek D, Seifert HH, Schmidtp- analysis of histones. Nat Protoc 2007;2:1445–57. eter M, et al. Expression of histone deacetylases 1, 2 and 3 in urothelial 29. Heubach J, Monsior J, Deenen R, Niegisch G, Szarvas T, Niedworok C, et al. bladder cancer. BMC Clin Pathol 2014;14:10. The long noncoding RNA HOTAIR has tissue and cell type-dependent 8. Witt O, Deubzer HE, Milde T, Oehme I. HDAC family: what are the cancer effects on HOX gene expression and phenotype of urothelial cancer cells. relevant targets? Cancer Lett 2009;277:8–21. Mol Cancer 2015;14:108. 9. MarksP,RifkindRA,RichonVM,Breslow R, Miller T, Kelly WK. Histone 30. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis deacetylases and cancer: causes and therapies. Nat Rev Cancer 2001; of large gene lists using DAVID bioinformatics resources. Nat Protoc 1:194–202. 2009;4:44–57. 10. Montezuma D, Henrique RM, Jeronimo C. Altered expression of histone 31. Bonner WM, Redon CE, Dickey JS, Nakamura AJ, Sedelnikova OA, Solier S, deacetylases in cancer. Crit Rev Oncog 2015;20:19–34. et al. GammaH2AX and cancer. Nat Rev Cancer 2008;8:957–67. 11. Olzscha H, Sheikh S, La Thangue NB. Deacetylation of chromatin and gene 32. Rogakou EP, Pilch DR, Orr AH, Ivanova VS, Bonner WM. DNA double- expression regulation: a new target for epigenetic therapy. Crit Rev Oncog stranded breaks induce histone H2AX phosphorylation on serine 139. J 2015;20:1–17. Biol Chem 1998;273:5858–68. 12. Moser MA, Hagelkruys A, Seiser C. Transcription and beyond: the 33. Dovey OM, Foster CT, Conte N, Edwards SA, Edwards JM, Singh role of mammalian class I lysine deacetylases. Chromosoma 2014; R, et al. Histone deacetylase 1 and 2 are essential for normal T- 123:67–78. cell development and genomic stability in mice. Blood 2013;121: 13. Stengel KR, Hiebert SW. Class I HDACs affect DNA replication, repair, and 1335–44. chromatin structure: implications for cancer therapy. Antioxid Redox 34. Kelly RD, Cowley SM. The physiological roles of histone deacetylase Signal 2015;23:51–65. (HDAC) 1 and 2: complex co-stars with multiple leading parts. Biochem 14. Conti C, Leo E, Eichler GS, Sordet O, Martin MM, Fan A, et al. Inhibition Soc Trans 2013;41:741–9. of histone deacetylase in cancer cells slows down replication forks, 35. Tatamiya T, Saito A, Sugawara T, Nakanishi O. Isozyme-selective activates dormant origins, and induces DNA damage. Cancer Res 2010; activity of the HDAC inhibitor MS-275. Cancer Res 2004;64:567 70:4470–80. (abstr.). 15. Yamaguchi T, Cubizolles F, Zhang Y, Reichert N, Kohler H, Seiser C, et al. 36. Lee JH, Choy ML, Ngo L, Foster SS, Marks PA. Histone deacetylase inhibitor Histone deacetylases 1 and 2 act in concert to promote the G1-to-S induces DNA damage, which normal but not transformed cells can repair. progression. Genes Dev 2010;24:455–69. Proc Natl Acad Sci U S A 2010;107:14639–44. 16. Wilting RH, Yanover E, Heideman MR, Jacobs H, Horner J, van der Torre J, 37. Sasakawa Y, Naoe Y, Inoue T, Sasakawa T, Matsuo M, Manda T, et al. et al. Overlapping functions of Hdac1 and Hdac2 in cell cycle regulation Effects of FK228, a novel histone deacetylase inhibitor, on human and haematopoiesis. EMBO J 2010;29:2586–97. lymphoma U-937 cells in vitro and in vivo.BiochemPharmacol 17. Jurkin J, Zupkovitz G, Lagger S, Grausenburger R, Hagelkruys A, Kenner 2002;64:1079–90. L, et al. Distinct and redundant functions of histone deacetylases 38. Slingerland M, Guchelaar HJ, Gelderblom H. Histone deacetylase inhibi- HDAC1 and HDAC2 in proliferation and tumorigenesis. Cell Cycle tors: an overview of the clinical studies in solid tumors. Anticancer Drugs 2011;10:406–12. 2014;25:140–9. 18. Haberland M, Johnson A, Mokalled MH, Montgomery RL, Olson EN. 39. Golay J, Cuppini L, Leoni F, Mico C, Barbui V, Domenghini M, et al. The Genetic dissection of histone deacetylase requirement in tumor cells. Proc histone deacetylase inhibitor ITF2357 has anti-leukemic activity in vitro Natl Acad Sci U S A 2009;106:7751–5. and in vivo and inhibits IL-6 and VEGF production by stromal cells. 19. Vallo S, Xi W, Hudak L, Juengel E, Tsaur I, Wiesner C, et al. HDAC inhibition Leukemia 2007;21:1892–900. delays cell cycle progression of human bladder cancer cells in vitro. 40. Ocker M, Schneider-Stock R. Histone deacetylase inhibitors: signalling Anticancer Drugs 2011;22:1002–9. towards p21cip1/waf1. Int J Biochem Cell Biol 2007;39:1367–74. 20. Lehmann M, Hoffmann MJ, Koch A, Ulrich SM, Schulz WA, 41. Bose P, Dai Y, Grant S. Histone deacetylase inhibitor (HDACI) mechanisms Niegisch G. Histone deacetylase 8 is deregulated in urothelial of action: emerging insights. Pharmacol Ther 2014;143:323–36. cancerbutnotatargetforefficient treatment. J Exp Clin Cancer 42. Saijo K, Katoh T, Shimodaira H, Oda A, Takahashi O, Ishioka C. Romi- Res 2014;33:59. depsin (FK228) and its analogs directly inhibit phosphatidylinositol 3- 21. Rosik L, Niegisch G, Fischer U, Jung M, Schulz WA, Hoffmann MJ. Limited kinase activity and potently induce apoptosis as histone deacetylase/ efficacy of specific HDAC6 inhibition in urothelial cancer cells. Cancer Biol phosphatidylinositol 3-kinase dual inhibitors. Cancer Sci 2012;103: Ther 2014;15:742–57. 1994–2001. 22. Furumai R, Matsuyama A, Kobashi N, Lee KH, Nishiyama M, Nakajima H, 43. Li Y, Shin D, Kwon SH. Histone deacetylase 6 plays a role as a distinct et al. FK228 (depsipeptide) as a natural prodrug that inhibits class I histone regulator of diverse cellular processes. FEBS J 2013;280:775–93. deacetylases. Cancer Res 2002;62:4916–21. 44. Khan N, Jeffers M, Kumar S, Hackett C, Boldog F, Khramtsov N, et al. 23. Leoni F, Fossati G, Lewis EC, Lee JK, Porro G, Pagani P, et al. The histone Determination of the class and isoform selectivity of small-molecule deacetylase inhibitor ITF2357 reduces production of pro-inflammatory histone deacetylase inhibitors. Biochem J 2008;409:581–9.

www.aacrjournals.org Mol Cancer Ther; 15(2) February 2016 311

Pinkerneil et al.

45. Knievel J, Schulz WA, Greife A, Hader C, Lubke T, Schmitz I, et al. Multiple 48. Stevens FE, Beamish H, Warrener R, Gabrielli B. Histone deace- mechanisms mediate resistance to sorafenib in urothelial cancer. Int J Mol tylase inhibitors induce mitotic slippage. Oncogene 2008;27: Sci 2014;15:20500–17. 1345–54. 46. Gabrielli B, Brown M. Histone deacetylase inhibitors disrupt the mitotic 49. Noh EJ, Lim DS, Jeong G, Lee JS. An HDAC inhibitor, trichostatin A, induces spindle assembly checkpoint by targeting histone and nonhistone pro- a delay at G2/M transition, slippage of spindle checkpoint, and cell death in teins. Adv Cancer Res 2012;116:1–37. a transcription-dependent manner. Biochem Biophys Res Commun 47. Khan DH, He S, Yu J, Winter S, Cao W, Seiser C, et al. Protein kinase CK2 2009;378:326–31. regulates the dimerization of histone deacetylase 1 (HDAC1) and HDAC2 50. Vakifahmetoglu H, Olsson M, Zhivotovsky B. Death through a tragedy: during mitosis. J Biol Chem 2013;288:16518–28. mitotic catastrophe. Cell Death Differ 2008;15:1153–62.

312 Mol Cancer Ther; 15(2) February 2016 Molecular Cancer Therapeutics

Inhibition of Class I Histone Deacetylases 1 and 2 Promotes Urothelial Carcinoma Cell Death by Various Mechanisms

Maria Pinkerneil, Michèle J. Hoffmann, René Deenen, et al.

Mol Cancer Ther 2016;15:299-312. Published OnlineFirst January 15, 2016.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-15-0618

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2016/01/15/1535-7163.MCT-15-0618.DC1

Cited articles This article cites 50 articles, 13 of which you can access for free at: http://mct.aacrjournals.org/content/15/2/299.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/15/2/299.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/15/2/299. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.