C-Terminal Binding Protein-2 Regulates Response of Epithelial Ovarian Cancer Cells to Histone Deacetylase Inhibitors

ORIGINAL ARTICLE C-terminal binding protein-2 regulates response of epithelial ovarian cancer cells to histone deacetylase inhibitors

L Barroilhet1, J Yang1, K Hasselblatt1, RM Paranal2, S-K Ng3, JA Rauh-Hain1, WR Welch4, JE Bradner2, RS Berkowitz1 and S-W Ng1

Ovarian cancer survival rates have stagnated in the last 20 years despite the development of novel chemotherapeutic agents. Modulators of gene expression, such as histone deacetylase (HDAC) inhibitors, are among the new agents being used in clinical trials. Predictors of sensitivity to chemotherapy have remained elusive. In this study, we show that the expression of the transcriptional corepressor C-terminal binding protein-2 (CtBP2) is elevated in human ovarian tumors. Downregulation of CtBP2 expression in ovarian cancer cell lines using short-hairpin RNA strategy suppressed the growth rate and migration of the resultant cancer cells. The knockdown cell lines also showed upregulation of HDAC activity and increased sensitivity to selected HDAC inhibitors. Conversely, forced expression of wild-type CtBP2 in the knockdown cell lines reversed HDAC activity and partially rescued cellular sensitivity to the HDAC inhibitors. We propose that CtBP2 is an ovarian cancer oncogene that regulates gene expression program by modulating HDAC activity. CtBP2 expression may be a surrogate indicator of cellular sensitivity to HDAC inhibitors.

Oncogene (2013) 32, 3896–3903; doi:10.1038/onc.2012.380; published online 3 September 2012 Keywords: ovarian cancer; tumor antigens; HDAC inhibitors

INTRODUCTION histones can be counteracted by the HDAC inhibitor trichostatin A Ovarian cancer remains the deadliest gynecologic malignancy in (TSA).12 HDACs remove acetyl-groups from histones and alter the United States. It is estimated that B21 550 new cases of chromatin structure and gene transcription.13 HDAC inhibitors ovarian cancer were diagnosed in 2009 in the United States with have gained increasing amounts of attention as potential 14 600 deaths reported.1 Survival rates have stagnated in the last chemotherapeutic targets.14,15 Various HDAC inhibitors, 20 years despite the development of novel chemotherapeutic including suberoylanilide hydroxamic acid and TSA, have been agents. Predictors of sensitivity to chemotherapy have remained reported to exhibit antitumor activities against hematologic, elusive, further complicating treatment optimization in patients, breast and bladder malignancies.16 Suberoylanilide hydroxamic who commonly present with advanced disease. acid is currently being used in phase II clinical trials to treat Tumor antigens are produced in cancer cells and trigger an recurrent ovarian cancer. immune response in hosts. The identification of tumor antigens Even though CtBP1 and CtBP2 share very high sequence has been a subject of ongoing research, as they are useful in homology, there are some intrinsic differences between them. identifying tumor cells and are potential targets for novel CtBP1 is both cytoplasmic and nuclear, with its subcellular therapies. The C-terminal binding protein-2 (CtBP2) was identified localization regulated by sumoylation, phosphorylation and as a tumor antigen in epithelial ovarian cancer using a reverse binding to a PDZ protein.17 In contrast, CtBP2 is exclusively capture antibody array.2 We sought to further characterize this nuclear, and its amino acid sequence lacks a C-terminal PDZ- protein and determine its role in oncogenesis. binding motif. The CtBP2 sequence has a unique N-terminal Human CtBP2 protein3 is homologous to human CtBP1 protein, domain that confers a dominant nuclear localization and provides a 48-kDa cellular phosphoprotein that binds to the C-terminal Lys residues for acetylation by the nuclear acetylase p300.18 region of the human adenovirus E1A proteins.4 CtBP family Recent studies have also suggested the dual activator and proteins are conserved transcriptional corepressors5 that have repressor roles of CtBP2 in transcriptional regulation.6,19 While crucial roles during development and oncogenesis.6,7 They also Ctbp1 mutant mice are viable and fertile, Ctbp2-null mice show have specific functions in epithelial gene regulation and defects in axial patterning and die by E10.5 because of aberrant programmed cell death.8,9 As transcriptional co-repressors, CtBP extraembryonic development.20 proteins interact with and regulate the activity of more than 30 The goals of our study included the investigation of the different vertebrate transcriptional factors, many of which are expression pattern of CtBP2 in epithelial ovarian cancer and the involved in oncogenic cell signaling pathways.10,11 The potential role of CtBP2 in oncogenic properties and drug mechanism of transcription repression includes the recruitment responses of ovarian cancer cells. Intriguingly, our studies have of chromatin modifiers including histone deacetylases (HDACs) shown the specific function of CtBP2 in conferring the sensitivity and histone methyltransferases.6,10 CtBP-mediated modification of of ovarian cancer cells to HDAC inhibitors.

1Department of Obstetrics/Gynecology and Reproductive Biology, Brigham and Women’s Hospital, Boston, MA, USA; 2Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, USA; 3School of Medicine, Griffith Health Institute, Griffith University, Meadowbrook, Queensland, Australia and 4Department of Pathology, Brigham and Women’s Hospital, Boston, MA, USA. Correspondence: Dr S-W Ng, Department of Obstetrics/Gynecology and Reproductive Biology, Brigham and Women’s Hospital, 221 Longwood Avenue BLI-449A, Boston, MA 02115, USA. E-mail: [email protected] Received 1 November 2011; revised 4 May 2012; accepted 4 July 2012; published online 3 September 2012 C-terminal binding protein-2 expression in ovarian cancer L Barroilhet et al 3897 RESULTS Table 1. CtBP2 expression and its relationship with clinical CtBP2 expression in ovarian epithelial cancers and relationship characteristics of ovarian tumor samples (upper panel) and overall with survival survival (bottom panel) In our previous study, autoantibody that targets CtBP2 was identified in ovarian cancer patients.2 To investigate the Characteristics Number of Mean of scores P-value expression pattern of the tumor antigen CtBP2 in ovarian cases tissues, we have performed immunohistochemistry using a Diagnostic o0.001 monoclonal CtBP2-specific antibody on 103 paraffin-embedded Healthy 6 0.33 ovarian tissues samples. Strong immunoreactivity with a high Benign 20 1.15 mean weighted score of 4.77 (borderline tumors) and 5.49 Borderline 17 4.77 (invasive tumors) was found in 83% of ovarian epithelial tumors, Invasive 60 5.49 which showed a significant difference (P-value o0.001) when compared with normal and benign ovarian tissues that had a Histology mean score of 0.33 and 1.15, respectively (Table 1). There were Serous 22 5.48 0.879 Mucinous 35 4.77 no significant differences of staining among subtypes of ovarian Endometrioid 13 5.08 tumors (P-value ¼ 0.879). Examples of normal ovarian tissue, Clear cell 11 5.18 borderline tumors and invasive tumors stained for CtBP2 are shown in Figure 1a. The majority of the staining was in the nuclei Variable Hazard ratio P-value of tumor cells. No staining was observed in the stromal cells. (95% CI) When CtBP2 expression in the invasive ovarian tumors with survival data was analyzed, the overall survival of patients with CtBP2-positive 9.813 (1.50–64.37) 0.017 positive CtBP2 expression was found to be poorer (Figure 1b) versus CtBP2- negative compared with those patients with negative CtBP2 expression Log (CA125) 0.817 (0.65–1.03) 0.092 (P-value ¼ 0.05), based on the Kaplan–Meier survival estimation. Stages 3 and 9.231 (2.78–30.67) o0.0005 In the Cox proportional hazard regression analysis (Table 1), a 4 versus stages significant difference in overall survival between patients 1 and 2 with positive and negative CtBP2 expression was found (P-value ¼ 0.017). The estimated hazard ratio for overall Abbreviations: CI, confidence interval; CtBP2, C-terminal binding protein-2. survival of patients with positive CtBP2 expression as compared with the negative CtBP2 expression was 9.813 (95% confidence interval: 1.50 to 64.37), with the adjustment for CA125 value and CtBP2 regulates HDAC activity and affects cellular sensitivity to tumor stage. HDAC inhibitors Western blotting showed an increase in CtBP2 expression in The corepressor complex of CtBP2 mediates coordinated histone seventeen human ovarian cancer cell lines when compared with modification by deacetylation and methylation of histones 6,10 normal human ovarian surface epithelial cells (Figure 2a). through interactions of several factors including HDACs. As many of the pharmacological inhibitors of HDACs are effective antineoplastic agents against leukemias and solid tumors,16 we Functional analysis of ovarian cancer cell lines with CtBP2 asked the question whether the response of ovarian cancer cell knockdown lines to HDAC inhibitors changes with differential expression of To investigate the function of CtBP2 in ovarian cancer cells, CtBP2 CtBP2. In vitro cytotoxicity assays of MCAS control and knockdown expression was successfully knocked down in two ovarian cancer cell lines to a panel of HDAC inhibitors currently in clinical and cell lines, SKOV3 and MCAS, using two short-hairpin RNA (shRNA) preclinical trials showed enhanced sensitivity of the CtBP2 constructs (Figure 2b). Ovarian cancer cells harboring shRNA knockdown cell lines to most of the HDAC inhibitors compared constructs 2 and 3 demonstrated reduced expression of CtBP2 by with the control cell line (Figure 4). The cell line harboring 475%. The expression of CtBP1 was not affected by the presence knockdown construct 3 was in general more sensitive to all of the of shRNA constructs (data not shown). Both SKOV3 and MCAS inhibitors, which might be due to other side effects of the control and knockdown cell lines were employed for the analysis construct. It is also noted that in the treatment with MGCD0103 of CtBP2 function and derivatives of both cell lines showed similar and valproic acid, the differences between the control and results. knockdown cell lines are significant only in the very high doses. Cellular proliferation assays showed reduced proliferation rate Supplementary Figure S2 shows that the SKOV3 panel demon- of knockdown cell lines when compared with controls (Figure 3a). strated similar response to the same and other HDAC inhibitors. In Cell-extracellular matrix adhesion assays showed that the knock- comparison, the knockdown cell line harboring construct shRNA2 down cell lines have significantly reduced capacity to adhere to did not show big differences in the response to the antitumor extracellular matrix proteins collagen I and collagen IV, and agent carboplatin (Figure 4f), indicating that the sensitivity of modest reduction of adhesion to laminin (Figure 3b). In a transwell CtBP2 knockdown cells to HDAC inhibitors is not due to the migration assay, the knockdown cell lines also demonstrated triggering of a common survival or death pathway that responds reduced migration rate relative to control shRNA cell lines univerally to all kinds of antineoplastic agents. (Figure 3c). Supplementary Figure S1 shows the differences of To demonstrate that CtBP2 expression has a causative role in number of stained cells that migrated through the transwell the sensitivity of CtBP2 knockdown cell lines to HDAC inhibitors, membrane. we reintroduced a full-length wild-type CtBP2 complementry DNA As CtBP family proteins are involved in Wnt signaling,11,19 (cDNA) expression construct into the knockdown cells and we tested the b-catenin activity of the cell lines using a compared the sensitivity of the resultant cells in association with luciferase reporter gene with upstream binding elements for b- CtBP2 expression. As the SKOV3 knockdown cell lines harbored catenin/T-cell factor (TCF) complex. Similar to the findings by already a luciferase construct, which would hinder clonal selection Hamada et al.,21 the knockdown cell lines demonstrated after transfection, we have only conducted the reintroduction of significantly increased b-catenin activity (Figure 3d), suggesting CtBP2 into the MCAS knockdown cell line that harbored the that expression of CtBP2 can downregulate Wnt signaling shRNA construct 2. Western blot analysis of two CtBP2 cDNA target genes. transfectants and the control cell lines in Figure 5a confirmed

& 2013 Macmillan Publishers Limited Oncogene (2013) 3896 – 3903 C-terminal binding protein-2 expression in ovarian cancer L Barroilhet et al 3898

Figure 1. CtBP2 expression and its relationship with patient overall survival. (a) Representative immunohistochemical results for CtBP2 expression. Positive CtBP2 staining was indicated by the brown color. Scale bar represents 50 mm. (b) Kaplan–Meier curves to show the difference of overall survival of patients with positive and negative CtBP2 expression.

reappearance of CtBP2 expression in the CtBP2 cDNA transfectants. Cytotoxicity asssay with the cell lines using TSA showed significant reversal of the TSA sensitivity in the CtBP2 cDNA transfectants relative to the parental line (Figure 5b). For comparison, a vector transfected shRNA2 cell line did not show difference in the senstivity compared with the parental line. To account for the changes of the TSA sensitivity, we have determined the HDAC activity of the cell lines using a universal HDAC assay kit. As shown in Figure 5c, the shRNA construct 2-harboring MCAS cell line showed HDAC activity level that was 33% higher than that of the control shRNA-harboring cell line. In the transfectants of shRNA2-harboring MCAS cell lines, the vector control showed similar level of HDAC activity as the parental line, whereas two CtBP2 transfectants showed reduced levels of HDAC activities (19 and 13%, respectively) relative to the parental shRNA2 line (Figure 5c). Hence, HDAC activities of the cell lines showed inverse relationship with their responses to HDAC inhibitors. To further evaluate whether there are changes in Figure 2. Expression of CtBP2 in ovarian cancer cell lines. (a) CtBP2 is overexpressed in seventeen ovarian cancer cell lines when activity of specific HDAC classes, we employed two substrates to compared with normal human ovarian surface epithelial cells. b- perform HDAC assays. The substrate MAZ1600 is kinetically more actin was used as loading control. (b) Western blot analysis of CtBP2 specific for class I HDACs (HDAC1, HDAC2, HDAC3) and HDAC6, expression in ovarian cancer cells transduced with lentivirus- whereas the substrate MAZ1675 is kinetically favorable for the harboring control and CtBP2 knockdown shRNAs. evaluation of class IIa HDACs (HDAC4, HDAC5, HDAC7 and HDAC9)

Oncogene (2013) 3896 – 3903 & 2013 Macmillan Publishers Limited C-terminal binding protein-2 expression in ovarian cancer L Barroilhet et al 3899

Figure 3. Functional analysis of ovarian cancer cells harboring control and CtBP2 knockdown shRNAs. SKOV3 ovarian cancer cells with control and CtBP2 shRNAs were compared for (a) cell growth (*Pp0.007; **Pp0.001); (b) relative adhesion to extracellular matrix proteins (*Po0.015; **Po0.05); (c) relative cell migration performed using the transwell migration assay (*P ¼ 0.01; **Po0.03); and (d) b-catenin activity (*Po0.032).

and HDAC8. As shown in Figure 5d, the knockdown shRNA2 cell b-catenin levels and constitutive activation of b-catenin/TCF- line showed elevated activities to both the substrates (50% to regulated gene expression occur in many human cancers. b- MAZ1600 and 26% to MAZ1675), whereas both CtBP2 transfec- catenin is stabilized upon Wnt stimulation and moves into the tants showed reduced levels of activity to both the substrates. The nucleus, where it binds to members of the TCF transcription factor results suggest that the activity of both classes of HDACs might be family and the formed complexes modulate transcription.23 TCF affected by the cellular levels of CtBP2. mutations and alterations in the APC and b-catenin may contribute to oncogenesis.11,23,24 However, CtBP proteins have been reported to interact with APC protein and suppress the expression of Wnt signaling target genes.11,21 Consistent with DISCUSSION those reports, our b-catenin assay also showed that b-catenin activity was increased in the CtBP2 knockdown cell lines. Whether Oncogenesis is a multi-step process that involves both genetic suppression of Wnt signaling bears any significance to CtBP2 and epigenetic events. Transcription factor complexes are master function in ovarian cancer awaits further investigation. regulators of cell state during development and oncogenesis. CtBP2 complex is a chromatin remodeling complex that Many of these transcription factor complexes are formed by involves HDACs, a group of enzymes that have gained increasing sequence-specific DNA binding factors in association with amounts of attention as potential chemotherapeutic targets.14,15 chromatin-modifying enzymes for gene transcriptional activation Our CtBP2 knockdown cell lines showed increased sensitivity to all or repression.15 The corepressor complex of CtBP2 mediates HDAC inhibitors tested, with the exception of MGCD0103 and coordinated histone modification by deacetylation and valproic acid. In the cytotoxicity assay with carboplatin, the cancer methylation of histones through interactions of several factors cell line harboring shRNA2 did not show differences in response, including HDACs.6,10 Our immunohistochemical study has shown indicating that the CtBP2-dependent drug sensitivity is restricted the overexpression of CtBP2 in ovarian tumor cells and its to HDAC inhibitors. The cancer cell line harboring shRNA3 did expression is associated with poor survival (Figure 1 and Table 1). show some significantly increased sensitivity to carboplatin, CtBP2 is likely an oncogene that epigenetically alters gene which we believe was due to side effects of the construct, transcription and contributes to neoplasia. as this line showed significant drop in growth rate and was in Ovarian cancer cells with CtBP2 knockdown exhibit a multitude general hypersensitive to other stress treatments. Furthermore, of dysfunctions that include reduced cell proliferation, cell- reintroduction of wild-type CtBP2 partially rescued the cancer cells extracellular matrix adhesion and cell migration capacity. Hence, to cytotoxicity mediated by HDAC inhibitors (Figure 5), strongly the CtBP2 transcription program is responsible for cell growth and supporting the notion that cytotoxicity to HDAC inhibitors is in cell migration and invasion, properties that are pivotal to accordance with the level of CtBP2 expression. Our HDAC assays oncogenic transformation and metastasis.22 Dysregulation of have shown that cellular HDAC activity was elevated in the

& 2013 Macmillan Publishers Limited Oncogene (2013) 3896 – 3903 C-terminal binding protein-2 expression in ovarian cancer L Barroilhet et al 3900

Figure 4. Comparison of MCAS ovarian cancer cells harboring control and CtBP2 knockdown shRNAs for sensitivity to chemotherapeutic agents. For each cell line, the MTT reading of the untreated cells was assigned as 100%. HDAC inhibitors: (a) Trichostatin A; (b) Vorinostat; (c) Belinostat; (d) MGCD0103; (e) valproic acid; and (f) carboplatin, a non-HDAC inhibitor. The P-values for differences between the knockdown cell lines and the control are indicated. In (d) and (e), the doses with signiﬁcant differences are marked with an *, which represents Po0.05.

knockdown cell lines and reversed in the CtBP2 cDNA transfec- migration and b-catenin activity. We are the first to show that tants. Hence, cellular sensitivity to HDAC inhibitors is proportional CtBP2 knockdown cell lines have increased sensitivity specific to to the HDAC activity, which is inversely proportional to the CtBP2 HDAC inhibitors, suggesting that the expression of CtBP2 might expression level. Both class I and class II HDACs are believed to be be a marker for response to these chemotherapeutic agents in the recruited to CtBP either directly or via other complex proteins.6,25 clinical setting. It will be of great interest to investigate which of It is not known, however, whether HDAC activity can be the HDACs are active in ovarian tumors and how nonselective interferred by CtBP complex. As cellular response to HDAC inhibitors are more effective in treating ovarian cancer. As CtBP2 is inhibitors is of clinical relevance, further mechanistic studies on dominantly overexpressed in ovarian tumors, it is of profound potential modulation of HDAC activity by CtBP2 are warranted. importance to design adjuvant therapy that can overcome CtBP2 The results of class-specific HDAC assays have suggested that the expression in ovarian tumors. activity of both class I and class IIa enzymes were elevated in the CtBP2 knockdown cells. Valproic acid is unique in that it has less activity against class II HDACs.26 Similarly, MGCD0103 is also a 27 MATERIALS AND METHODS class I HDAC-specific inhibitor. The less drastic effects of both Ovarian clinical samples and ovarian cell lines agents on the knockdown cell lines suggest that isoform- All patient-derived biologic specimens were collected and archived under selectivity may account for the differences of efficacy of HDAC protocols approved by the Human Subjects Committee of the Brigham and inhibitors, and that the nonselective HDAC inhibitors may be more Women’s Hospital, Boston, Massachusetts. Archived formalin-fixed, paraf- desirable in treating human cancers. fin-embedded normal, benign and cancerous ovarian tissues were In conclusion, we have shown overexpression of CtBP2 protein collected from women undergoing surgery at the Brigham and Women’s in ovarian tumors and its function in regulating cell growth, cell Hospital for a diagnosis of primary ovarian cancer or from control subjects

Oncogene (2013) 3896 – 3903 & 2013 Macmillan Publishers Limited C-terminal binding protein-2 expression in ovarian cancer L Barroilhet et al 3901

Figure 5. Rescue of sensitivity of knockdown cells with reintroduction of CtBP2 expression. (a) Western blot analysis to show the CtBP2 expression of the investigated cell lines. b-actin was used as loading control. (b) Sensitivity of the cell lines to TSA using MTT assays. For each cell line, the MTT reading of the untreated cells was assigned as 100%. (c) Cellular HDAC activity measured by a universal HDAC assay kit. *, Po0.05. (d) Class-speciﬁc HDAC assays using MAZ1600 and MAZ1675 substrates. The results of a representative experiment are shown. (i) shRNA control cell line; (ii) shRNA2 knockdown parental line; (iii) shRNA2-CtBP2 transfectant 1; (iv) shRNA2-CtBP2 transfectant 2; (v) shRNA control cell line in the presence of 10 mM TSA.

who were undergoing the procedure of hysterectomy or oophorectomy (maximum immunoreactivity). The score was obtained by multiplying the for benign gynecologic indications. Samples were collected with written intensity of the stain (0 for no stain, 3 for intense stain) by the percentage informed consent from patients and confirmed histologically by pathol- of the epithelial tissue that stained positive (0 for no stain, 3 for 100% of ogists. Cases were staged according to the International Federation of the epithelium staining positive). Gynecology and Obstetrics system. The normal human ovarian surface epithelial cells and the ovarian cancer cell lines have been described previously.28 Normal human ovarian surface epithelial cells were collected Lentiviral Infection and DNA transfection by scraping the ovarian surface of the control subjects who were Mission lentiviral CtBP2-targeting and non-target control shRNA transduc- undergoing hysterectomy or oophorectomy for benign diseases. Long- tion particles were purchased from Sigma-Aldrich (St Louis, MO, USA). term human ovarian surface epithelial cells were immortalized by a HPV The RNAi Consortium numbers for the shRNAs 1–5 are N0000013743, E6/E7 gene introduction. All ovarian cell lines were maintained in a mixture N0000013744, N0000013745, N0000013746 and N0000013747, respec- of medium 199 and MCDB105 medium (1:1) (Sigma, St Louis, MO, USA) tively. For the effective knockdown shRNA2 and 3, they target the positions supplemented with 10% fetal calf serum (Invitrogen, Carlsbad, CA, USA). 1021–1042 and 1144–1165, respectively, of the NM_001329 CtBP2 transcript sequence. 1 Â 105 MCAS and SKOV3 ovarian cancer cells growing in complete medium were infected with 5 Â 105 transduction Immunohistochemistry particles in the presence of 8 mg/ml hexadimethrine bromide (Sigma- Paraffin-embedded ovarian tissue blocks were sectioned at a thickness of Aldrich) and incubated overnight at 37 1C and 5% CO2. On the third day, 7 mm and mounted on Superfrost Plus microscopic slides (Fisher Scientific, the medium was replaced with complete medium containing 2 mg/ml Pittsburgh, PA, USA), and dried at 50 1C for at least 3 h. Deparaffinization puromycin and changed every 3 days for 2 weeks. Knockdown of CtBP2 was performed using xylene and rehydration with a graded ethanol expression in the resultant cell lines was confirmed by western blot series. Antigen retrieval was performed in a pressure-cooker in antigen- analysis. To rescue the CtBP2 expression in the knockdown cell lines, full- unmasking solution (Vector Laboratories, Burlingame, CA, USA). Endogen- length CtBP2 cDNA expression construct and empty vector purchased ous peroxidases were blocked using 0.3% H202 in methanol for 20 min. from OriGene Technologies (Rockville, MD, USA) were transfected The sections were then blocked with normal horse serum for 20 min and separately into the CtBP2 shRNA clone-2 MCAS cell line using Lipofecta- were subsequently incubated overnight with anti-CtBP2 (monoclonal, 1:80 mine 2000 transfection reagent (Invitrogen Corp.). The transfected cells dilution; BD Biosciences, San Jose, CA, USA) as the primary antibody. After were selected using complete medium containing 500 mg/ml G418 incubation, the reaction was visualized using Vectastain Elite ABC Kit with (Invitrogen Corp.) and the CtBP2 expression in the transfected cells was diaminobenzidine chromogen as a substrate (Vector Laboratories). confirmed by Western blot analysis. It was noted that the reappearance of Sections were counterstained lightly with hematoxylin and mounted. CtBP2 expression was not stable. The selected positive clones reverted to Immunohistochemical results were evaluated by two trained investigators. loss of CtBP2 expression with time, especially after thawing from frozen CtBP2 staining was graded semi-quantitatively by considering the stocks. The experimental results described here were performed in the percentage and intensity of the staining. A histological score was obtained early stage of clone selection during which the CtBP2 expression was from each sample, which ranged from 0 (no immunoreaction) to 9 rescued. We have later transfected the CtBP2 cDNA expression construct

& 2013 Macmillan Publishers Limited Oncogene (2013) 3896 – 3903 C-terminal binding protein-2 expression in ovarian cancer L Barroilhet et al 3902 into the control shRNA-harboring cancer cell lines to establish CtBP2- solubilization solutions, 10% SDS in 0.01N HCl, and 0.1N HCl in 2-propanol overexpressing lines. Further experiments with these new lines showed (1:1), was added to each well. Absorbance at 562 nm was determined on similar results as provided by the transient rescued clones. an EL Â 800 absorbance microplate reader (Bio-Tek, Winooski, VT, USA). Each drug assay was performed in triplicates and repeated twice. Western blot Total cell lysates were prepared from growing cells using lysis buffer HDAC activity assays (50 mM Tris–HCl pH 8, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate Universal HDAC activity was measured using a colorimetric HDAC activity and 0.1% SDS) supplemented with PhosStop phosphatase inhibitor assay kit (BioVision). Fifty microgram of diluted nuclear extracts, prepared cocktail and complete protease inhibitor cocktail (Roche Applied Science, using nuclear and cytoplasmic extraction kit (Pierce Biotechnology, Indianapolis, IN, USA). Protein concentration was measured with a Rockford, IL, USA), were incubated with HDAC colorimetric substrate in MicroBCA protein assay kit (ThermoScientific, Rockford, IL, USA). Ten 1 Â HDAC assay buffer at 37 1C for 1 h. Reactions were stopped by adding microgram of total cell lysates of the cell lines were resolved by SDS– 10 ml of lysine developer and mixed for 30 min. Samples were read at polyacrylamide gel electrophoresis and transferred to a polyvinylidene 405 nm on an EL Â 800 absorbance microplate reader (Bio-Tek). Class- fluoride membrane using a SEMI-DRY transfer cell (Bio-Rad Laboratories, specific HDAC assays were performed as described.29 Briefly, nuclear Hercules, CA, USA). After blocking with 5% nonfat dry milk in 1 Â Tris extracts were incubated with a series of doses of fluorophore-conjugated Buffer Saline Tween20 buffer (10 mM Tris–HCl pH 7.5, 150 mM NaCl, 0.05% substrates in assay buffer (50 mM HEPES, 100 mM KCl, 0.001% (v/v) Tween-20) at room temperature for 1 h, the membrane was incubated with Tween-20, 0.05% (w/v) bovine serum albumin, 200 mM tris(2-carboxyethyl) the primary antibody (anti-CtBP2, BD Biosciences) at 4 1C overnight then phosphine, pH 7.4) and followed for fluorogenic release of 7-amino-4- washed at room temperature. The bound antibody was detected by the methylcoumarin from substrates upon deacetylase and trypsin enzymatic anti-mouse secondary antibodies conjugated with horseradish peroxidase activity. Fluorescence measurements were obtained every 5 min using a (Pierce Biotechnology, Rockford, IL, USA) and a supersignal west pico kit multilabel plate reader (Envision, Perkin-Elmer, Waltham, MA, USA). Data (Pierce Biotechnology). were analyzed for the mid-linear range of fluorescence over time and plotted versus substrate concentrations. In these assays, the substrate MAZ1600 was kinetically more specific for HDAC1, HDAC2, HDAC3 and Cell growth, adhesion and migration assays HDAC6, whereas the substrate MAZ1675 was kinetically favorable for Cell proliferation was determined by seeding 3 Â 104 cells to 35-mm HDAC4, HDAC5, HDAC7, HDAC8 and HDAC9. As a negative control, a culture dishes and allowed to grow to different time points. At each time reaction was performed in the presence of 10 mM of TSA. The assays were point, the number of cells in three culture dishes were counted and performed in duplicates and repeated twice. averaged for each cell line. Adhesion to components of the extracellular matrix was evaluated using CytoMatrix cell adhesion strips (Millipore, Billerica, MA, USA) and performed according to the instructions of the Statistical Analysis manufacturer. Cell migration was evaluated in transwell inserts with 8 mm All calculations were performed with MINITAB statistical software (Minitab, pore-sized polyethylene terephthalate membranes (BD Biosciences) coated State College, PA, USA) unless otherwise indicated. ANOVA was used to with 10 mg/ml collagen (Sigma-Aldrich) overnight. The following day, compare the mean immunohistochemistry scores between benign and 1 Â 105/ml cells in serum-free medium containing 0.05% bovine serum malignant paraffin sections and between different tumor histologies. albumin were added to the inserts. The inserts were then placed in the Significance of the test was considered at the 5% level (that is, P-value wells of a companion plate and incubated with complete medium p0.05). For the survival analysis, the length of overall survival was defined containing 10% fetal bovine serum in the lower chamber. After 24 h of from the date of operation to the date of the patient’s death (uncensored) incubation, the cells in the inserts were scraped with cotton swaps. The or the date of last visit (censored). We defined the low and high CtBP2 membranes were excised from the inserts and counterstained with Diff- expression groups using the cutoff score of 3.5. The choice of cutoff score Quik Stain Set (Dade Behring Inc., Newark, DE, USA). The cells that migrated was based on the ability to group cases with significant differences in through the membrane were counted in four random fields. Each overall survival. Overall survivals of patients with positive and negative experiment was repeated twice and average values±s.e.m are presented. CtBP2 expression were estimated using the Kaplan–Meier method, and compared with a log-rank test. The impact of CtBP2 expression on patient survival was further studied with the inclusion of potential clinical risk b-catenin activity assay factors, using the Cox proportional hazards regression model. The variables b-catenin activity was measured using a luciferase reporter assay. The cells applied for adjustment in the multivariate analysis included tumor grade, were transfected with a TOP-FLASH luciferase reporter plasmid (Addgene, stage of disease, histologic type, CA125 values, optimal or suboptimal Cambridge, MA, USA) that contains multimers of b-catenin/TCF-binding debulking and patient’s age. Patient’s age was treated as a continuous sequences upstream of the firefly luciferase reporter gene. A Renilla variable, while tumor grade, stage of disease and histologic type were luciferase-encoding plasmid pRL-CMV (Promega, Madison, WI, USA) was analyzed as categorical variables. Grade 1 was grouped as the reference included in the transfection as internal control. Another transfection with a group to identify the risk of death of grades 2 and 3 cases. Stage 1 and 2 FOP-FLASH (Addgene) plasmid that harbors mutant b-catenin/TCF-binding cases were grouped as the reference group to evaluate the risk of death of sites, together with pRL-CMV plasmid, was performed to serve as negative the combination of stage 3 and 4 cases. Histologic type was classified as control. Preparation of cell lysates and luciferase assay was performed either serous or nonserous. With the multivariate Cox proportional hazards using the Dual-Luciferase Reporter Assay System (Promega) according to analysis, the variables were entered into the model using a backward the manufacturer’s protocol. Luminescence was measured using a TD-20/ stepwise Wald-test process. The impact of CtBP2 expression on patient 20 luminometer (Turner Designs, Sunnyvale, CA, USA). The firefly luciferase survival was determined by examining the relative hazard ratios with activities of the cell lines were normalized by the respective Renilla respect to the positive/negative CtBP2 expression groups in the final luciferase activities. Each transfection was performed in triplicates and model. The significance of estimated hazard ratios was tested using the repeated twice. The results are reported as means±s.e.’s. Wald test. The Cox regression analysis was performed using the SPSS 17.0 statistical software. (SPSS, Inc., Chicago, IL, USA). Cytotoxicity assay Cellular sensitivity to HDAC inhibitors and carboplatin (Sigma-Aldrich) was CONFLICT OF INTEREST measured using 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assays. The following HDAC inhibitors were tested: belinostat, The authors declare no conflict of interest. vorinostat, MGCD0103 (Selleck, Pittsburgh, PA, USA), splitomicin, M344 (Biovision, Mountainview, CA, USA), TSA and valproic acid (Sigma-Aldrich). Cell suspensions (7 Â 104 cells/ml) were seeded in 100 ml of medium per ACKNOWLEDGEMENTS well in 96-well plates. After overnight incubation, 100 ml of medium with We thank critical reading of the manuscript by Professor YF Wong. We acknowledge serial dilutions of each HDAC inhibitor were added for 48 h. Twenty the provision of belinostat, vorinostat, MGCD0103 from Selleck Inc. LB was supported microliter of 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide by Ruth N White Gynecologic Oncology Research Fellowship. The Laboratory of solution (5 mg/ml in PBS, Sigma-Aldrich) was added to each well and Gynecologic Oncology was partly supported by Robert and Deborah First Fund, the incubated for 4 h. Medium was aspirated and 100 ml of combined Sperling Family Fund Foundation, Women’s Cancer Program and Gillette Center for

Oncogene (2013) 3896 – 3903 & 2013 Macmillan Publishers Limited C-terminal binding protein-2 expression in ovarian cancer L Barroilhet et al 3903 Women’s Cancer from Dana-Farber Cancer Institute, Ovarian Cancer Research 14 Minucci S, Pelicci PG. Histone deacetylase inhibitors and the promise of epige- Foundation, Adler Foundation, Inc., and Friends of Dana-Farber Cancer Institute. netic (and more) treatments for cancer. Nat Rev Cancer 2006; 6: 38–51. 15 Yoo CB, Jones PA. Epigenetic therapy of cancer: past, present and future. Nat Rev Drug Discov 2006; 5: 37–50. REFERENCES 16 Glaser KB. HDAC inhibitors: clinical update and mechanism-based potential. Bio- 1 Altekruse S, Kosary C, Krapcho M, Neyman N, Aminou R, Waldron W et al. SEER chem Pharmacol 2007; 74: 659–671. Cancer Statistics Review, 1975-2007. National Cancer Institue: Bethesda, MD, USA, 17 Lin X, Sun B, Liang M, Liang YY, Gast A, Hildebrand J et al. Opposed regulation of 2010. corepressor CtBP by SUMOylation and PDZ binding. Mol Cell 2003; 11: 1389–1396. 2 Tang L, Yang J, Ng SK, Rodriguez N, Choi PW, Vitonis A et al. Autoantibody 18 Zhao LJ, Subramanian T, Zhou Y, Chinnadurai G. Acetylation by p300 regulates profiling to identify biomarkers of key pathogenic pathways in mucinous ovarian nuclear localization and function of the transcriptional corepressor CtBP2. J Biol cancer. Eur J Cancer 2010; 46: 170–179. Chem 2006; 281: 4183–4189. 3 Katsanis N, Fisher EM. A novel C-terminal binding protein (CTBP2) is closely 19 Fang M, Li J, Blauwkamp T, Bhambhani C, Campbell N, Cadigan KM. C-terminal- related to CTBP1, an adenovirus E1A-binding protein, and maps to human binding protein directly activates and represses Wnt transcriptional targets in chromosome 21q21.3. Genomics 1998; 47: 294–299. Drosophila. EMBO J 2006; 25: 2735–2745. 4 Boyd JM, Subramanian T, Schaeper U, La Regina M, Bayley S, Chinnadurai G. A 20 Hildebrand JD, Soriano P. Overlapping and unique roles for C-terminal binding region in the C-terminus of adenovirus 2/5 E1a protein is required for association protein 1 (CtBP1) and CtBP2 during mouse development. Mol Cell Biol 2002; 22: with a cellular phosphoprotein and important for the negative modulation of T24- 5296–5307. ras mediated transformation, tumorigenesis and metastasis. EMBO J 1993; 12: 21 Hamada F, Bienz M. The APC tumor suppressor binds to C-terminal binding 469–478. protein to divert nuclear beta-catenin from TCF. Dev Cell 2004; 7: 677–685. 5 Turner J, Crossley M. Cloning and characterization of mCtBP2, a co-repressor that 22 Engers R, Gabbert HE. Mechanisms of tumor metastasis: cell biological aspects associates with basic Kruppel-like factor and other mammalian transcriptional and clinical implications. J Cancer Res Clin Oncol 2000; 126: 682–692. regulators. EMBO J 1998; 17: 5129–5140. 23 Moon RT, Kohn AD, De Ferrari GV, Kaykas A. WNT and beta-catenin signalling: 6 Chinnadurai G. Transcriptional regulation by C-terminal binding proteins. Int J diseases and therapies. Nat Rev Genet 2004; 5: 691–701. Biochem Cell Biol 2007; 39: 1593–1607. 24 Cuilliere-Dartigues P, El-Bchiri J, Krimi A, Buhard O, Fontanges P, Flejou JF et al. 7 Chinnadurai G. The transcriptional corepressor CtBP: a foe of multiple tumor TCF-4 isoforms absent in TCF-4 mutated MSI-H colorectal cancer cells colocalize suppressors. Cancer Res 2009; 69: 731–734. with nuclear CtBP and repress TCF-4-mediated transcription. Oncogene 2006; 25: 8 Grooteclaes M, Deveraux Q, Hildebrand J, Zhang Q, Goodman RH, Frisch SM. C- 4441–4448. terminal-binding protein corepresses epithelial and proapoptotic gene expression 25 Zhang CL, McKinsey TA, Lu JR, Olson EN. Association of COOH-terminal-binding programs. Proc Natl Acad Sci USA 2003; 100: 4568–4573. protein (CtBP) and MEF2-interacting transcription repressor (MITR) contributes to 9 Grooteclaes ML, Frisch SM. Evidence for a function of CtBP in epithelial gene transcriptional repression of the MEF2 transcription factor. J Biol Chem 2001; 276: regulation and anoikis. Oncogene 2000; 19: 3823–3828. 35–39. 10 Chinnadurai G. CtBP, an unconventional transcriptional corepressor in develop- 26 Khan N, Jeffers M, Kumar S, Hackett C, Boldog F, Khramtsov N et al. Determination ment and oncogenesis. Mol Cell 2002; 9: 213–224. of the class and isoform selectivity of small-molecule histone deacetylase inhi- 11 Sierra J, Yoshida T, Joazeiro CA, Jones KA. The APC tumor suppressor counteracts bitors. Biochem J 2008; 409: 581–589. beta-catenin activation and H3K4 methylation at Wnt target genes. Genes Dev 27 Fournel M, Bonfils C, Hou Y, Yan PT, Trachy-Bourget MC, Kalita A et al. MGCD0103, 2006; 20: 586–600. a novel isotype-selective histone deacetylase inhibitor, has broad spectrum 12 Shi Y, Sawada J, Sui G, Affar el B, Whetstine JR, Lan F et al. Coordinated histone antitumor activity in vitro and in vivo. Mol Cancer Ther 2008; 7: 759–768. modifications mediated by a CtBP co-repressor complex. Nature 2003; 422: 735–738. 28 Huang KC, Park DC, Ng SK, Lee JY, Ni X, Ng WC et al. Selenium binding protein 1 in 13 de Ruijter AJ, van Gennip AH, Caron HN, Kemp S, van Kuilenburg AB. Histone ovarian cancer. Int J Cancer 2006; 118: 2433–2440. deacetylases (HDACs): characterization of the classical HDAC family. Biochem J 29 Bradner JE, West N, Grachan ML, Greenberg EF, Haggarty SJ, Warnow T et al. 2003; 370: 737–749. Chemical phylogenetics of histone deacetylases. Nat Chem Biol 2010; 6: 238–243.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

& 2013 Macmillan Publishers Limited Oncogene (2013) 3896 – 3903