Targeted Inactivation of HDAC2 Restores P16 Activity and Exerts

Published OnlineFirst November 21, 2012; DOI: 10.1158/1541-7786.MCR-12-0332

Molecular Cancer Oncogenes and Tumor Suppressors Research

Targeted Inactivation of HDAC2 Restores p16INK4a Activity and Exerts Antitumor Effects on Human Gastric Cancer

Jeong Kyu Kim1,2, Ji Heon Noh1,2, Jung Woo Eun1,2, Kwang Hwa Jung1,2, Hyun Jin Bae1,2, Qingyu Shen1,2, Min Gyu Kim1,2, Young Gyoon Chang1,2, Seung-Jin Kim1,2, Won Sang Park1,2, Jung Young Lee1,2, Jurgen€ Borlak3, and Suk Woo Nam1,2

Abstract Aberrant regulation of histone deacetylase 2 (HDAC2) was reported for gastric cancers. However, responsive cancer genes in disease onset and progression are less understood. HDAC2 expression was studied by quantitative RT-PCR and Western blotting. The functional consequences of HDAC2 knockdown on cell-cycle regulation, programmed cell death, and gene target identification was investigated by flow cytometry, Western blotting, electron microscopy, anchorage-independent colony formation, and cell migration assay and by whole-genome microarray. Therapeutic efficacy of HDAC2 knockdown was determined in nude mice with small hairpin INK4a expressing human gastric cancer cells. Epigenetic regulation of p16 was studied by methylation-specific PCR and chromatin-IP to evidence HDAC2 or acetylated-histone-H4 binding at gene specific promoter sequences. HDAC2 gene and protein expression was significantly upregulated in different histopathologic grades of human gastric cancers and cancer cell lines. HDAC2 inactivation significantly reduced cell motility, cell invasion, clonal – p16INK4a expansion, and tumor growth. HDAC2 knockdown-induced G1 S cell cycle arrest and restored activity of and the proapoptotic factors. This treatment caused PARP cleavage and hypophosphorylation of the Rb-protein, repressed cyclinD1, CDK4, and Bcl-2 expression and induced autophagic phenotype, that is, LC3B-II conversion. INK4a Some gastric tumors and cancer cells displayed p16 promoter hypermethylation but treatment with 5-aza- deoxycitidine restored activity. With others the methylation status was unchanged. Here, chromatin-IP evidenced INK4a HDAC2 binding. Nonetheless, expression of p16 was restored by HDAC2 knockdown with notable histone- INK4a H4-acetylation, as determined by chromatin-IP. Thus, p16 is regulated by HDAC2. HDAC2 is a bona fide target for novel molecular therapies in gastric cancers. Mol Cancer Res; 11(1); 62–73. 2012 AACR.

Introduction subject of intense research. Histone deacetylases (HDAC) fi Gastric cancer is one of the leading causes of cancer deaths are modi cation enzymes that regulate the expression and worldwide (1) and is particularly prevalent in Asia. Over the activity of numerous proteins involved in both cancer initiation and progression by removing the acetyl groups, last few years, several studies contributed toward an under- fi standing of the molecular causes of gastric cancer, thereby thus allowing compacted chromatin modi cation (4). The facilitating its early diagnosis and improved curability (2, 3). mammalian HDACs are diverse and divided into 4 different While the molecular mechanisms leading to undue cell classes of 18 isoforms based on their sequence homologies growth in gastric cancers remain uncertain the study of with their yeast counterparts (5) and available evidence enzymes that modify the histone code has become the indicates their involvement in cell proliferation, differentiation, and cell-cycle regulation (6). Importantly, deregulation of HDACs is frequently observed in cancer (7) and Authors' Affiliations: 1Lab of Oncogenomics, Department of Pathology, structurally diverse HDAC inhibitors are being developed as College of Medicine, and 2Functional RNomics Research Center, The Catholic University of Korea, Seoul, Republic of Korea; and 3Center of anticancer agents for the treatment of various solid and Pharmacology and Toxicology, Hannover Medical School, Hannover, hematologic malignancies. Unfortunately, most HDAC Germany inhibitors are not isoform specific (8) and the regulation Note: Supplementary data for this article are available at Molecular Cancer and contribution of individual HDACs in cancer develop- Research Online (http://mcr.aacrjournals.org/). ment remains unclear. Corresponding Authors: Suk Woo Nam, Department of Pathology, Col- In this regard, it was suggested that elevated HDAC2 lege of Medicine, The Catholic University of Korea, Seoul, Republic of expression is associated with poor prognosis in patients Korea. Phone: 82-2-2258-7314; Fax: 82-2-537-6586; E-mail: [email protected]; and Jurgen€ Borlak, Center of Pharmacology and suffering from oral, prostate, ovarian, endometrial, or Toxicology, Hannover Medical School, Carl-Neuberg Str.1, D-30625 Hann- gastric cancer (9) and research from our laboratory and over, Germany. E-mail: [email protected] by others evidenced HDAC2 overexpression in stomach doi: 10.1158/1541-7786.MCR-12-0332 cancer and in hepatocellular carcinoma to possible con- 2012 American Association for Cancer Research. tribute to cancer progression (10, 11). Therefore, aberrant

62 Mol Cancer Res; 11(1) January 2013

Oncogenic HDAC2 in Gastric Cancer

regulation of HDAC2 may be a key event in carcinogen- Gene Expression Omnibus (GEO) database (Accession esis, yet an identification of HDAC2-relevant cancer genes number; GSE31338). and the molecular consequences of silencing of hyperac- tive HDAC2 in gastric cancer development remain to be Western blot analysis elucidated. Protein lysates were separated by SDS-PAGE, transferred Overall, this study aims for a better understanding of to a nitrocellulose membrane, and immunoblotted with the oncogenic activity of HDAC2 in human gastrocarci- antibodies as indicated. Blots were developed with enhanced nogenesis and an identification of target genes relevant chemiluminescence Western blotting reagent. Densitomet- for the development of suitable therapeutic intervention ric analyses for protein quantification were done using Image strategies. J 1.44p software (http://rsbweb.nih.gov/ij/index.html). Each blot was repeated at least twice. Details are given in Materials and Methods Supplementary Data. Ethics approval This study was conducted with the approval of the Methylation-specific PCR Institutional Review Board (IRB) of the Songeui Campus, DNA was extracted, treated with bisulfate using the EZ College of Medicine, The Catholic University of Korea (IRB DNA methylation kit (Zymo Research), and purified approval number; CUMC09U118). using a QiaQuick DNA purification kit (Qiagen) according to the manufacturer's instruction. Primer sequences INK4a Cell culture and treatments and PCR conditions for the p16 gene were as The human stomach cancer cells (MKN-1, SNU-1, 216, previously described (Supplementary Data, Supplementa- 484, 620, 638, 668, and 719) were purchased from Korean ry Table S2; ref. 12). Primers were localized to regions in INK4a Cell Line Bank, whereas human stomach adenocarcinoma and around the transcription start site of the p16 cell line (AGS), NIH-3T3/Ras, and HT1080 cells were gene, that is, the region that is associated with loss of gene purchased from American Type Culture Collection (Man- expression. assas). To inhibit DNA methylation cells were treated with 2 mmol/L of 5-aza-deoxycytidine. To induce an autophagic Statistical analysis response, 10 mmol/L of ceramide (N-acetyl-d-erythro-sphin- For cell growth, cell proliferation, soft agar colony for- gosine; Calbiochem) dissolved in dimethyl sulfoxide was mation, and cell motility and invasion assays, statistical added. significances of differences between the means of sample sets were determined using the unpaired 2-tailed Student's t Histone deacetylase 2 gene silencing by small-interfering test in Prism 5.0 (GraphPad Software). Results are expressed RNA and stable cell lines as the means SDs of at least 3 independent experiments. HDAC2-specific small-interfering RNAs (siRNA) were Statistical significance was determined for p values of less designed and purchased through Silencer Predesigned siR- than 0.05. NAs (www.ambion.com) to transfect MKN-1 cells with a Detailed experimental procedures can be found in the siRNA targeted against the HDAC2 gene. The sequences of Supplementary Data. siRNA and generation of stable cell line are given in Sup- plementary Data. Results HDAC2 overexpression is associated with enhanced Reverse transcription PCR analysis and real-time PCR mitogenesis of gastric cancer cells analysis We previously reported aberrant expression of HDAC2 in Total RNA from each cell line was isolated using Trizol human gastric cancer (10). Additional report also suggested reagent (Invitrogen). One microgram of total RNA was HDAC2 overexpression to be associated with poor prognosis reverse transcribed into cDNA using a mixture of Super- by multivariate analyses (13). To generalize these findings, script II enzyme (Gibco-BRL) and 0.5 mg oligo (d)T we recapitulated HDAC2 gene expression from the large (Amersham Pharmacia Biotech), according to the manufac- cohorts of gastric cancer patients that are available from the turer's recommendations. Detailed information of the real- NCBI, GEO database, (accession numbers GSE24375 and time PCR, PCR reactions, and the primer used are given in GSE13196) and data are given as scatter plot. Consistently, Supplementary Data. HDAC2 gene expression was significantly upregulated in these 2 different gastric cancer cohorts (Fig. 1A). Increased Microarray analysis of whole genome expression expression of HDAC2 protein was confirmed by immuno- For each of the experimental conditions, total RNA was blotting of 12 randomly selected human gastric cancer extracted from 3 independent sets of the corresponding tissues. HDAC2 was overexpressed in 9 of the 12 selected cell lines. For each experimental condition, an RNA pool gastric cancer tissues compared with the corresponding was then obtained by mixing equal quantities of total noncancerous tissues (Fig. 1B; Supplementary Fig. S1A). RNA from each of the 3 independent RNA extractions. Human gastric cancer cell lines, with a few exceptions, also The primary microarray data are available from the exhibited considerable expression of HDAC2 in both National Center for Biotechnology Information (NCBI), mRNA and protein levels (Fig. 1C; Supplementary Fig.

www.aacrjournals.org Mol Cancer Res; 11(1) January 2013 63

Kim et al.

Figure 1. Aberrant regulation of HDAC2 and its association with malignant proliferation of human gastric cancer. A, HDAC2 expression in normal gastric mucosa, gastric adenoma, and gastric adenocarcinoma as recapitulated by gene expression data from NCBI GEO (GSE24375, GSE13195). , P < 0.005; , P < 0.01. B, Western blotting of HDAC2 in a subset of gastric cancer tissues. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as loading control. C, gene and protein expression of HDAC2 in various human gastric cancer cell lines as determined by qRT-PCR and Western blotting. D, mRNA expression of different HDAC isoforms in the human gastric cancer cell line MKN-1 cells determined by qRT-PCR. Data are determined by the 2(-Delta Ct) method and are expressed as the ratio of gene of interest relative to the expression of the housekeeping gene GAPDH. E, the effects of HDAC2 knockdown on cell proliferation for MKN-1, SNU-216, SNU-484, and SNU-719 cells as determined by the MTT assay. To ascertain the knockdown efﬁciency of HDAC2, protein expressions of HDAC2 were determined by immunoblotting (Top).

S1B). Importantly, in MKN-1 cells HDAC2 expression consequences of gene silencing of HDAC2 in gastric tumor- corresponded to histone deacetylase activity, as determined igenesis, knockdown of HDAC2 was attempted by means of by quantitative real-time PCR (qRT-PCR) for the various RNA interference and studied in the MTT cell proliferation HDACs (Fig. 1D). To better understand the molecular assays. As shown in Fig. 1E, HDAC2 knockdown resulted in

64 Mol Cancer Res; 11(1) January 2013 Molecular Cancer Research

Oncogenic HDAC2 in Gastric Cancer

INK4a a significant reduction in protein expression and in reduced bound to the p16 promoter region could be evidenced, proliferation rates of the MKN-1 and SNU-484 gastric whereas HDAC2 knockdown was associated with acetylated cancer cells, whereas the proliferation of the SNU-216 and histone H4 DNA binding (Fig. 2F). SNU-719 cells was unaffected even though efficient inhibition of the HDAC2 protein was achieved in these cell lines. HDAC2 overexpression enhanced the proliferation of gastric cancer cells by deregulating expressions of G –S INK4a 1 HADC2 knockdown recovers p16 tumor cell-cycle proteins suppressor activity In an attempt to identify molecular targets associated with Packing of DNA into chromatin requires dynamic oncogenic HDAC2 activity whole genome gene expression changes and involves nucleosome-remodeling complexes analysis was applied to mock-(empty vector, MKN1-mock) whereby gene silencing is achieved by DNA methylation and HDAC2 shRNA-expressing plasmid transfected MKN- and by the activity of the histone deacetylase complex. 1 cells (MKN1-shHDAC2). Such analysis revealed HDAC2 INK4a INK4a Specifically, a previous report suggested p16 to be knockdown to restore expression of p16 and to influ- hypermethylated in gastric carcinomas (14) but the rela- ence the expression of genes involved in cellular growth and tionship between nucleosomal histone deacetylation and death pathways (depicted as a heatmap in Fig. 3A). Indeed, DNA methylation of this tumor suppressor is unknown. the suppression of cyclin-dependent kinase 4 (CDK4) and – fi Initially, hypermethylation of the promoter region of the the G1 S speci c cyclin D1 observed for MKN1-shHDAC2 p16INK4a – gene in gastric cancer cell lines was assessed with a cells is suggestive for HDAC2 to disturb the G1 S phase by methylation-specific PCR (MSP) assay. For this MSP assay, deregulating cell-cycle regulatory proteins. To clarify the role 2 types of primers (p16-150 and p16-234) were designed of HDAC2 in cell-cycle progression, HDAC2 knockdown INK4a that enharbor the transcription start site of the p16 gene MKN-1 cells were treated with nocodazole. This treatment fi – and/or adjacent sequences. At rst, the MKN-1 and SNU- synchronizes the cells in the G2 M phase. After release from 484 cells that exhibited growth retardation after functional nocodazole block the proportions of cells in the G1 phase inactivation of HDAC2 were analyzed. Both cells lacked were determined by flow cytometry. HDAC2 knockdown p16INK4a – methylation sites in the chosen promoter region. In caused arrest in the G1 S phase and delayed cell-cycle contrast, the cell lines SNU-216 and SNU-719, where transition, suggesting that the proliferative defect and/or proliferation rates were unaffected by HDAC2 knockdown growth retardation of MKN-1 cells by HDAC2 knockdown (see also Fig. 1E) exhibited hypermethylation of the is caused, at least in part, by a G arrest of the cell cycle (Fig. INK4a 1 p16 promoter region (Fig. 2A). We next explored 3B). To generalize this finding, we assessed the effects of whether HDAC2 knockdown could restore p16INK4a pro- HDAC2 knockdown on cell cycle in the absence of noco- tein expression in gastric cancer cells. With MKN-1 and dazole. As we expected, HDAC2 deficient MKN-1 cells, SNU-484 cells HDAC2 knockdown recovered p16INK4a stable HDAC2 knockdown cell lines (HDAC2 KD1 and 2), – expression, albeit marginally, whereas treatment with 5-aza- exhibited increase of cells remaining in G1 S phase of cell dC alone or in combination had no effect. With the SNU- cycle and delayed cell-cycle transition compared with MKN- 719 and SNU-216 cell lines HDAC2 knockdown had no 1 or mock-transfected cells (Supplementary Fig. S2A and effect on p16INK4a expression (Fig. 2B). Notably, treatment S2B). We then observed that HDAC2 knockdown selec- of cell cultures with 5-Aza-dC resulted in significantly tively induced p16INK4a expression, and simultaneously reduced proliferation rates of the SNU-216 and SNU- suppressed the expression of CDK1 and cyclin D1 among 719 cells (Fig. 2C). These results suggest that both histone G –S cell-cycle regulators (Fig. 3C). In general, the activated INK4a 1 acetylation and promoter methylation control p16 gene CDK/cyclin complex can cause hyperphosphorylation of expression. Next, the p16INK4a protein expression in human pRb, which then loses its tumor suppressor activity and gastric cancers and its promoter methylation status was allows for increased E2F/DP1 transcriptional activity. studied. A total of 7 pairs of gastric tumor tissues and As we expected, HDAC2 knockdown led to hypophosphor- corresponding nontumor tissues were examined. Note, 3 ylation of pRb without change the total pRb in MKN-1 of these tissue pairs (cases #592, #805, and #1033) displayed gastric cancer cells (Fig. 3C, Supplementary Fig. S3). Fur- little difference between normal and tumor tissues in terms thermore, dual knockdown of HDAC2 and p16INK4a was of HDAC2 protein expression (Fig. 2D). However, all attempted and studied in the MTT cell proliferation assays INK4a selected gastric tumor tissues revealed loss of p16INK4a to verify the p16 as a key downstream target of onco- expression. These tissues were then analyzed for the meth- genic HDAC2 in gastric tumorigenesis. Consistent with our INK4a ylation status of the p16 gene promoter region. Normal results, HDAC2 knockdown resulted in the reduced pro- gastric mucosal tissues were found to contain both unmethy- liferation rates of the both MKN-1 and SNU-484 gastric INK4a INK4a lated and methylated p16 but for gastric tumors with cancer cells, whereas p16 knockdown did not affect on considerable HDAC2 overexpression, that is, cases #1094 the proliferation rates of both MKN-1 and SNU-484 cells. INK4a #1134, #1273, and #1152 the chosen promoter regions were However, dual knockdown of HDAC2 and p16 (si- basically unmethylated. In contrast, tumors #592, #805, and HDAC2þsi-p16) recovered reduced growth rate, suggesting #1033, with no changes in HDAC2 expression displayed that antigrowth effect on gastric cancer cells by HDAC2 significant methylation as determined by the MSP assay (Fig. inactivation could be explained, at least in part, by the 2E). Finally, by chromatin immunoprecipitation HDAC2 disruption of cell-cycle regulation via induction of p16INK4a

www.aacrjournals.org Mol Cancer Res; 11(1) January 2013 65

Kim et al.

INK4a INK4a Figure 2. Epigenetic regulation of p16 in human gastric cancer. A, MSP to identify changes in the methylation status at p16 promoter regions in human INK4a gastric cancer cell lines. Primers were designed to study DNA methylation patterns in CpG islands of the p16 promoter start site (genomic position: þ167, genomic position is the location of the 50 nucleotide of the sense primer in relation to the major transcriptional start site deﬁned in the Genbank accession numbers: X94154). In addition, the effect of treatment of human gastric cancer cell lines with 5-aza-dC was investigated. The presence of a PCR INK4a product in lanes labeled U indicates an unmethylated status for the p16 promoter region, whereas the presence of a PCR product in lanes labeled M indicates a methylated promoter status. B, p16INK4a levels are evaluated in HDAC2 knockdown or 5-aza-dC treated cells to ascertain the effects of HDAC2 knockdown or 5-aza-dC treatment on p16INK4a protein expression in gastric cancer cell lines. C, cell proliferation of MKN-1, SNU-216, SNU-484, and SNU-719 human gastric cancer cell lines treated with 5-aza-dC (2 mmol/L). Cells were grown in 24-well plates in medium supplemented with 5-aza-dC for up to 96 hours. Cell proliferation was determined by the MTT assay. Data are expressed as means SDs. , P < 0.01. All measurements were conducted in triplicate, and each experiment was repeated at least twice. D, Western blotting of HDAC2 and p16INK4a expressions in human normal gastric mucosa and cancer tissues. The blot shown is typical of at least 3 individual experiments. E, analysis of the methylation status of the p16INK4a promoter by MSP in paired-gastric cancer tissues of the same patient. N, normal gastric mucosa; T, gastric cancer tissue. F, chromatin-IP to evidence HDAC2 or acetylated histone H4 DNA binding. A marked increased in acetylated histone H4 DNA binding is noted in HDAC2 gene silenced MKN-1 human gastric cancer cells expressing short hairpin RNAs. This coincides with recovered gene expression of p16INK4a. IgG was used as a negative control, whereas total input DNA (Bottom) was used as a positive PCR control for the target promoter. Depicted is the proximal p16INK4a promoter regions, that is, nucleotide 177 to þ135, and the result shown is representative of 3 conducted experiments.

66 Mol Cancer Res; 11(1) January 2013 Molecular Cancer Research

Oncogenic HDAC2 in Gastric Cancer

Figure 3. The effects of HDAC2 knockdown on cell-cycle proteins. A, HDAC2 gene silencing was conducted by transfecting MKN-1 cells with HDAC2 shRNA expression plasmids for 48 hours, and then microarray analysis was done as described in materials and methods section. Heatmap of the gene expressions changes associated with cell cycle and apoptosis: green, reduced expression; red, induced expression when compared with the control (black). B, the effects of HDAC2 siRNA (si-HDAC2) or scramble sequence (s-Scr) on cell- cycle transition of MKN-1 cells. Data were obtained by ﬂow cytometry and cells were synchronized in their cell- cycle progression by treatment with 50 ng/mL nocodazole. N, nontransfected cell line (mother cells, MKN-1); R, transfection reagents- treated cells. C, Western blot analysis of cell-cycle regulatory proteins in MKN-1 cells treated with HDAC2 siRNA or scrambled RNA. b-actin served as a loading control. The blot is representative for at least 3 separate experiments. D, the effects of HDAC2 and p16INK4a knockdown on cell proliferation of MKN-1 and SNU-484 cells. Immublotting of HDAC2 and p16INK4a (Top). MTT assay (Bottom). , P < 0.05; , P < 0.01; N.S., not signiﬁcant.

expression, on HDAC2 targeting (Fig. 3D). These results stems from ﬂow cytometry measurements of annexin strongly suggested that HDAC2 hyperactivity modulates V-ﬂuorescein isothiocyanate (FITC) propidium iodine transcription of cell-cycle proteins to foster the transition stained MKN-1 cells transfected with HDAC2 siRNA. from G1 to S phase in gastric cancer cells. The proportion of apoptotic cells was greater for HDAC2 knockdown cells than for controls (Supplementary Fig. HDAC2 knockdown induced apoptotic and autophagic S4A). Taken collectively, HDAC2 oncogenic activity cell death of gastric cancer cells deregulates major components of the programmed cell On the basis of gene expression analysis deregulation of death, and thus confers resistance to apoptosis in gastric components of the programmed cell death can be inferred cancer. and included the apoptosis regulatory proteins Bcl-2 and In a previous study, it was reported that some histone Apaf-1. HDAC2 knockdown selectively induced the pro- deacetylase inhibitors induce apoptotic and autophagic tein expression of the proapoptotic factors Bax, apoptosis cell death (15). While the present results show that inducing factor (AIF), and Apaf-1, but suppressed the HDAC2 knockdown resulted in suppression of Bcl-2 the expression of the antiapoptotic Bcl-2. This suggests cross-talk between apoptosis and autophagic signaling has HDAC2 inactivation to restore programmed cell death not been investigated. As shown in Fig. 4B, HDAC2 in MKN-1 gastric cancer cells. Consequently, cleaved knockdown suppressed Bcl-2 and increased LC3B-II con- PARP levels were increased and the cleaved form of version, a hallmark of autophagic activation. Here, treat- caspase 3 was induced by HDAC2 knockdown (Fig. ment with ceramide, that is, an inducer of autophagic cell 4A). Further evidence for endorsed programmed cell death death served as a positive control. It was also suggested

www.aacrjournals.org Mol Cancer Res; 11(1) January 2013 67

Kim et al.

Figure 4. HDAC2 knockdown induces apoptosis and autophagic cell death in human gastric cancer cell line. A, left, Western blotting of apoptosis proteins in HDAC2 knockdown MKN-1 cells. Protein expression of AIF, Bcl-2, BAX, Apaf-1, cleaved-caspase3, PARP, and cleaved-PARP is depicted. N, nontransfected cell line (mother cells); R, transfection reagent-treated cells. Right, Western blot bands were quantified with the Image J (NIH) software and expression levels of indicated protein are given relative to scrambled siRNA. Each sample was normalized by related a-tubulin expression. , P < 0.05. B, left, activation of autophagic cell death induced by HDAC2 inactivation. Ceramide (20 mmol/L) was used as a positive control for turning on the autophagic signal in MKN-1 cells. Right, expression levels of indicated protein were quantified and expressed as described above. C, Western blot analysis of apoptosis related proteins in HDAC2 and/or Apaf-1 knockdown MKN-1 cells. b-actin served as a loading control. The blot is representative for at least 3 separate experiments. D, apoptosis assay of MKN-1 cells treated with HDAC2 and/or Apaf-1 siRNA. After 48 hours transfection of indicated siRNA, annexin-V-FITC/PI positive cells were determined by flow cytometry. Bar graph indicates the percent of Annexin V positive cells (apoptotic cells) in MKN-1 cells transfected with the indicated siRNA. All measurements were conducted in duplicate, and each experiment was repeated at least twice. , P < 0.01; , P < 0.005.

that targeted inhibition of HDAC2 suppresses Bcl-2 to trol siRNA-trasnfected cells, si-Scr) cells (Supplementary stimulate activity of Beclin 1 thereby facilitating the Fig. S3C). formationofanautophagosome.Toconﬁrm the forma- These data suggest that the cooperative suppression of tion of an autophagosome, we examined the ultrastructure caspase-dependent and independent cell death by HDAC2 of HDAC2 siRNA treated MKN-1 cells by transmission may exert a very potent mitotic stimulation causing uncon- electron microscopy (TEM). At 48 hours posttransfection trolled cell growth during gastric tumorigenesis. Thus, to approximately 40% to 45% of the HDAC2 siRNA-trans- clarify our hypothesis, selective knockdown of HDAC2 and/ fected cells developed autophagic vacuoles, some of which or Apaf-1 was conducted in MKN-1 gastric cancer cells. had possible merged to form larger vacuoles (Supplemen- Consistent with our results, HDAC2 knockdown selectively tary Fig. S4B). Further, immunoﬂuorescence staining for induced Bax, Apaf-1, and the cleavage of PARP in MKN-1 LC3B revealed that HDAC2 knockdown induced ring- gastric cancer cells, whereas the Apaf-1 knockdown did not shaped spots evenly distributed throughout cytoplasm, affect the aforementioned proteins. Moreover, dual knock- indicating an association between LC3B and autophago- down of HDAC2 and Apaf-1 induced Bax and cleavage of somal membranes compared with controls (negative con- PARP implying Apaf-1-independent modulation of

68 Mol Cancer Res; 11(1) January 2013 Molecular Cancer Research

Oncogenic HDAC2 in Gastric Cancer

Figure 5. Effects of HDAC2 inactivation on the invasive potential of human gastric cancer cells. A, after 48 hours transfection of HDAC2 shRNA, microarray analysis was done as described above. Heatmap of gene expression changes associated with the metastatic potential of cancer cells: green, repressed expression when compared with the control (black). B, motility assay. Assays were conducted by using the Boyden chambers. N, nontransfected cell line (mother cells); R, reagent-only treated cells. The graph shows the average number of cells that migrated based on 3 independent experiments; means SDs are given. , P < 0.01. C, the invasive capabilities of MKN-1 cells and of MKN-1 cells transfected with either scramble siRNA or HDAC2 siRNA were examined by using the Transwell invasion assay. Data are the means SDs of 3 independent experiments. , P < 0.01. The graph shows the average number of cells that migrated during 2 independent experiments. D, in vitro anchorage-independent growth assay. The effect of HDAC2 inhibition on the growths of cell colonies is depicted after 3 weeks of culture. Colony numbers are the means SDs of 3 independent experiments. , P < 0.01. E, tumor aggressiveness-related protein expression in HDAC2 knockdown MKN-1 cells measured by Western blot analysis. a-tubulin and GAPDH served as a loading control. The blot is representative for at least 3 separate experiments. apoptosis proteins by HDAC2 in gastric cancer cells (Fig. Overexpression of HDAC2 enhanced the invasive 4C). Note that p53 protein level was not affected by potential of gastric cancer cells HDAC2 knockdown, suggesting that proapoptosis effect Previous studies showed HDAC2 overexpression in tumors on HDAC2 targeting in gastric cancer cells appeared to be with nodal metastasis and in advanced gastric cancers (10, 13) p53-independent processing. This result was supported by to possible deﬁne a role for HDAC2 in metastatic spread. cytometry measurements of annexin V-FITC propidium Therefore, the gene expression data from HDAC2 knock- iodine stained MKN-1 cells transfected with either down cells was analyzed to identify some of the genetic events HDAC2and/orApaf-1siRNA(Fig. 4D). The proportion associated with the metastatic potential of gastric cancer of apoptotic cells was greater for HDAC2 knockdown cells cells. Figure 5A depicts gene regulations associated with than appropriate controls (N; nontreatment, si-Scr; neg- metastasis pathways as a heatmap. To support the gene ative control siRNA). The apoptotic cell proportions expression data, an in vitro tumorigenesis assay was conducted (upperrightofthedotplotgraphsinFig.4D)were using HDAC2 knockdown cells. HDAC2 knockdown atten- signiﬁcantly increased in MKN-1 cells transfected with uated tumor cell motility and invasion in a Boyden chamber HDAC2 siRNA or HDAC2 siRNA þ Apaf-1 siRNA assay (Fig. 5B and C). HDAC2 knockdown also reduced the compare with controls, suggestingthataberrantregulation anchorage-independent growth ability of tumor cells (Fig. of HDAC2 may modulate apoptosis proteins and promote 5D). In addition, to gain more insight into oncogenic malignant transformation of normal gastric cells. There- HDAC2 in tumor cell motility and invasion, we conducted fore, targeted-inactivation HDAC2 causes p53-indepen- Western blot analysis for proteins related to epithelial–mes- dent apoptotic and autophagic cell death in gastric cancer enchymal transition (EMT) in HDAC2 knockdown MKN-1 cells. cells, because EMT is a program of metastatic cancer cells

www.aacrjournals.org Mol Cancer Res; 11(1) January 2013 69

Kim et al.

characterized by loss of cell adhesion, repression of E-cadherin lines (Fig. 6A). These cell lines exhibited reduced growth expression, and increased cell mobility. As shown in Fig. 5E, rates, as compared with mock (empty vector) transfected or HDAC2 knockdown repressed the expression of SLUG, a key mother cell line, MKN-1 (Fig. 6B). On the basis of these regulator of EMT, whereas other molecules were not changed results, we next conducted in vitro tumorigenic assays, or nondetectable (data not shown). This result suggests that namely, anchorage-independent colony formation and cell the impairedmotilityand/orinvasivenessofgastric cancer cells migration and invasion assays. The results obtained clearly by HDAC2 inactivation, at least in part, due to suppression of indicate that HDAC2 deficient cells exhibited reduced cell SLUG expression suggesting a possible key regulator for motility, cell invasion, and clonal growth (Fig. 6C–E). In the enhanced motility and invasion potential by HDAC2 in transient transfection study described above, it was noted – gastric tumorigenesis. that HDAC2 caused G1 S cell-cycle arrest, of which cellular senescence is one of the biologic consequences. Therefore, The sustained downregulation of HDAC2 reduced the b-galactosidase (SA-b-Gal) staining was used to determine tumorigenic potentials of MKN-1 gastric cancer cells in whether HDAC2 suppression enhances the senescence of vitro and in vivo cell lines. As shown in Fig. 6F, HDAC2 deficient cell lines To investigate whether sustained suppression of HDAC2 (HDAC2 KD1 and HDAC2 KD2) were found to exhibit an lead to suppression of in vivo gastric tumorigenesis, we increase in cellular senescence as compared with MKN-1 or prepared HDAC2 deficient cells by establishing stable mock-transfected cells. Finally, to confirm that HDAC2 HDAC2 knockdown cell lines (HDAC2 KD1 and HDAC2 overexpression contributes to gastrogarcinogenesis in vivo, KD2), and confirmed the repression of HDAC2 by detect- we subcutaneously injected the HDAC2 KD1 and HDAC2 ing p16INK4a induction and cyclin D1 reduction in these cell KD2 cell lines into athymic nude mice. Overall tumor

Figure 6. Effect of sustained suppression of HDAC2 in MKN-1 cells on in vitro tumorigenicity. A, Western blot analysis of HDAC2, p16INK4a and cyclin D1 in total lysates of MKN-1 cells with stable expression of short hairpin RNA (HDAC2 KD1 and HDAC2 KD2) or mock (Empty vector, pSilencer 2.1). Note the considerable HDAC2 protein depletion in the cell lines. HDAC2 gene silencing restored p16INK4a expression but repressed cylin D1. a-tubulin expression served as loading control. The data are at least n ¼ 3 independent experiments. B, cell proliferation was determined by the MTT assay. Data are mean SD. , P < 0.01. C, cell migration assay. The assay was conducted in the Boyden chamber; here, NIH/3T3 cells were used as a positive control (¼ mock). The graph depicts the average number of cells migrating of n ¼ 3 independent experiment; data are means SD. , P < 0.01. D, the invasive abilities of the indicated cell lines were examined by the Transwell invasion assays. , P < 0.01 versus mock cells. E, anchorage-independent colony growth was visualized by staining with crystal violet; colonies were counted and analyzed. Data are means SD. , P < 0.01. F, left, speciﬁc senescence-associated marker, b-galactosidase (SA- b-gal), was determined by X-gal staining. Right, the bar graph shows the average number of senescent cells as a % of total cell that were positively stained by SA-b-gal. Results are means SD. , P < 0.01.

70 Mol Cancer Res; 11(1) January 2013 Molecular Cancer Research

Oncogenic HDAC2 in Gastric Cancer

Figure 7. In vivo tumorigenicity of MKN-1 cells with sustained suppression of HDAC2. A, MKN-1 parental, mock, HDAC2 KD1, and HDAC2 KD2 cells (5 106) were given subcutaneously into the right ﬂanks of nude mice (n ¼ 9 animals per cell line). Tumor sizes were measured every 3 days for 46 days, as described in "Supplementary Data and Supplementary Materials and Methods". , P < 0.01 versus mock cells. B, representative images of individual mice are shown (Top), alongside images of excised tumors (Bottom). C, HDAC2 and p16INK4a protein expression in xenograft tumors was evaluated by Western blot analysis. A typical result of 2 conducted experiments is shown.

growth rates were significantly lower for the HDAC2 defi- the molecular and genetic functions of HDAC2 was cient cell lines (HD2 KD1 and HD2 KD2) as compared studied in considerable detail. Evidence is presented for with MKN-1 or mock-transfected cells (Fig. 7A). Average HDAC2 to mediate the epigenetic silencing of the mul- tumor volumes at sacrifice were also significantly smaller in tiple tumor suppressor and cyclin-dependent kinase inhib- the HD2 KD1 and HD2 KD2 groups than in the mock or itor p16INK4a. Furthermore, the present study shows MKN-1 groups (Fig. 7A and B). Furthermore, the smaller HDAC2 knockdown to reactivate cellular apoptosis and – tumor volumes in the HD2 KD1 and HD2 KD2 groups autophagic cell death programs, to cause G1 S cell-cycle were associated with low HDAC2 but restored p16INK4a arrest in human MKN-1 gastric cancer cells, and to reduce expression in tumor tissues (Fig. 7C). tumor growth in an animal xenograft model. In terms of oncogenic enabling in gastric cancer, epigenetic instability via methylation of CpG islands may be a Discussion major contributor. Here, the more commonly methylated HDACs play an important role in chromatin remodeling tumor-related genes are APC, CDH1, hMLH1, CDKN2A INK4a and contribute to the silencing of the transcriptional (p16 ), CDKN2B, and RUNX (21) and the synergy machinery. There is also evidence for certain HDAC between methylation and histone deacetylase activity in the family members to be aberrantly expressed in various transcriptional silencing of genes is being increasingly under- human malignancies (6, 16) that inspired the development stood. We hypothesized that the expression of tumor sup- of a new class of anticancer agents. Nonetheless, a com- pressor genes during gastric tumorigenesis is epigenetically prehensive understanding of cancer relevant target genes as regulated by both methylation and histone deacetylase a result of HDAC oncogenic activity is still lacking as is the activity. In our efforts to identify relevant cancer genes, we INK4a contribution of individual HDACs in the regulation of found that the expression of p16 is regulated by gene transcription in the context of cancer initiation and HDAC2 deacetylation of histone H4 and by promoter progression (4, 5). In this regard, it was noted that methylation in gastric cancer cells (Fig. 2). Although there HDAC2 is aberrantly expressed in neoplastic tissues, and is evidence for HDAC2 inactivation to increase the it was consistently shown that targeted-inactivation of p21WAF/Cip1 activity (22), this is the first study to show in HDAC2 elicited growth arrest and apoptosis in certain gastric cancer cells, that HDAC2 binds to promoter regions INK4a human cancer cells (17–20). In gastric cancer, the expres- of p16 and suppresses the transcriptional activity of INK4a sions of class I HDACs were highly correlated with each p16 . Our data suggest that HDAC2 inhibits endoge- other, whereas HDAC2 expression was reported to be nous p16INK4a expression by establishing a repressive chro- INK4a elevated in tumors with nodal metastases and an advanced matin structure at p16 promoter to increase mitotic tumor stage (10, 13). However, no investigation has been activity and to prevent cell-cycle arrest or cellular senescence. conducted to determine the role of HDAC2 during gastric Consistent with this, we observed that HDAC2-deficient tumorigenesis and an identification of associated but piv- MKN-1 cells exhibited both in vitro and in vivo growth otal cancer related target genes. In the present investigation retardation and increased cellular senescence (Fig. 6 and 7).

www.aacrjournals.org Mol Cancer Res; 11(1) January 2013 71

Kim et al.

The specific function of HDAC2 with respect to the sion of HDAC2 may contribute to gastric malignant targeting of genes, proteins, and signaling cascades is the proliferation and transformation via resistance to cas- subject of intense research (9). In the present study, it was pase-dependent and independent cell death or accelerating – found that HAC2 inactivation caused a G1 S cell-cycle cellular growth rate by transcriptional activation of cell arrest via recovering p16INK4a activity and repressing the cycle components or enhancing tumor motility and expression of CDK4 and cyclin D1 in MKN-1 gastric invasiveness. INK4a cancer cells (Fig. 3). Evidence was obtained for HDAC2 to Taken collectively, this study shows that p16 is an silence proapoptotic proteins, such as AIF and Apaf-1, and important cancer relevant HDAC2 target gene that is to simultaneously induce the expression of Bcl-2. Reports inactivated by promoter hypermethylation and histone suggest HDAC2 to disrupt the apoptotic balance by deacetylation in gastric cancer. The knockdown of regulating the caspase 9-activating proteins Apaf-1 and HDAC2 induced growth arrest, apoptosis, and autophagic NOXA (23, 24). Here, we show for gastric cancer cells cell death in vitro and in vivo. In gastric cancer cells, that targeted inactivation of HDAC2 restored activity of HDAC2 overexpression was associated with disturb the proapoptotic factors Bax, AIF, and Apaf-1, but homeostasis by deregulating gene expressions of cell cycle, repressed the antiapoptotic Bcl-2 (Fig. 3 and 4). Note apoptosis, and autophagy components. Chromatin IP that both MKN-1 (Val143Ala, dominant negative muta- studies provided conclusive evidence for HDAC2 to be INK4a tion) and SNU-484 (Gly266Glu, DNA-binding region p16 promoter bound, whereas HDAC2 siRNA was mutation) cells have p53 gene mutation and express associated with acetylated histone H4 DNA binding. mutant p53. Thus, p53 protein level was not affected by Future studies will help to define the precise mechanisms HDAC2 knockdown, suggesting that proapoptosis effect of gene regulation by HDAC2 to influences cell prolifer- on HDAC2 targeting in gastric cancer cells appeared to be ation, apoptosis, and differentiation in cooperation with p53-independent processing. Therefore, HDAC2 inacti- others. Here, we propose that aberrant regulation of vation causes cell death processing by p53-independent HDAC2 results in the epigenetic regulation of genes caspase-activating apoptosis. coding for apoptosis and cell-cycle components and show The findings of the present study agree well with reports its important role in the development of gastric cancer. by other investigators that show that HDAC inhibitors Thus, selective HDAC2 inhibition provides a molecular induce both apoptosis and autophagy. Given the inherent rational for novel therapeutic intervention strategies. resistance to apoptosis that characterizes cancer, the targeting of an alternative pathway is an attractive therapeutic Disclosure of Potential Conflicts of Interest strategy. Our finding that HDAC2 knockdown suppresses No potential conflicts of interest were disclosed. Bcl-2 expression and increases annexin V staining led us to speculate whether HDAC2 inactivation could possible Authors' Contributions Conception and design: J.K. Kim, W.S. park, J. Borlak, S.W. Nam promote autophagic cell death as well. Indeed, many Development of methodology: J.K. Kim signaling pathways, including mTOR, AIF, reactive oxy- Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.K. Kim, J.H. Noh, K.H. Jung, S.W. Nam gen species (ROS), CDKs, and HDAC1/6, play important Analysis and interpretation of data (e.g., statistical analysis, biostatistics, compu- roles in the regulation of HDAC-induced autophagy (15). tational analysis): J.K. Kim, J.H. Noh, J.W. Eun, H.J. Bae, J. Borlak, S.W. Nam However, the role played by HDAC2 on the mechanism of Writing, review, and/or revision of the manuscript: J.K. Kim, W.S. park, J. Borlak, S.W. Nam autophagic cell death was not investigated as yet. After Administrative, technical, or material support (i.e., reporting or organizing data, HDAC2 knockdown, TEM analysis of the cell ultrastruc- constructing databases): J.K. Kim, W.S. park, J.Y. Lee, S.W. Nam ture showed that HDAC2 induced autophagic cell death Study supervision: J.K. Kim, W.S. park, J.Y. Lee, J. Borlak, S.W. Nam in gastric cancer cells. Note, Bcl-2 is known to inhibit Beclin1-dependent autophagy (25) and HDAC2 siRNA Grant Support This research was supported by the National Research Foundation of Korea (NRF) abrogated the interaction between Bcl-2 and Beclin 1 to grant funded by the Korean government (MEST; Grant No. 2011-0010705). influence the cross-talk between autophagy and apoptosis The costs of publication of this article were defrayed in part by the payment of page (Fig. 4). Our results show that the targeted inactivaton of charges. This article must therefore be hereby marked advertisement in accordance with HDAC2 can induce both mitochondria-mediated apo- 18 U.S.C. Section 1734 solely to indicate this fact. ptosis and caspase-independent autophagic cell death in Received May 31, 2012; revised September 20, 2012; accepted October 17, 2012; gastric cancer cells (Fig. 4A and B). Aberrant overexpres- published OnlineFirst November 21, 2012.

References 1. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA activation-induced cytidine deaminase in gastric epithelium. Nat Med Cancer J Clin 2005;55:74–108. 2007;13:470–6. 2. Polk DB, Peek RM Jr. Helicobacter pylori: gastric cancer and beyond. 4. Marks P, Rifkind RA, Richon VM, Breslow R, Miller T, Kelly WK. Histone Nat Rev Cancer 2010;10:403–14: Erratum in: Nat Rev Cancer. 2010 deacetylases and cancer: causes and therapies. Nat Rev Cancer Aug;(8):593. 2001;1:194–202. 3. Matsumoto Y, Marusawa H, Kinoshita K, Endo Y, Kou T, Morisawa T, 5. Glozak MA, Seto E. Histone deacetylases and cancer. Oncogene et al. Helicobacter pylori infection triggers aberrant expression of 2007;26:5420–32.

72 Mol Cancer Res; 11(1) January 2013 Molecular Cancer Research

Oncogenic HDAC2 in Gastric Cancer

6. Witt O, Deubzer HE, Milde T, Oehme I. HDAC family: what are the 17. Hrzenjak A, Moinfar F, Kremser ML, Strohmeier B, Staber PB, Zatloukal cancer relevant targets? Cancer Lett 2009;277:8–21. K, et al. Valproate inhibition of histone deacetylase 2 affects differen- 7. Yang XJ, Seto E. HATs and HDACs: from structure, function and tiation and decreases proliferation of endometrial stromal sarcoma regulation to novel strategies for therapy and prevention. Oncogene cells. Mol Cancer Ther 2006;5:2203–10. 2007;26:5310–8. 18. Huang BH, Laban M, Leung CH, Lee L, Lee CK, Salto-Tellez M, et al. 8. Batty N, Malouf GG, Issa JP. Histone deacetylase inhibitors as anti- Inhibition of histone deacetylase 2 increases apoptosis and p21Cip1/ neoplastic agents. Cancer Lett 2009;280:192–200. WAF1 expression, independent of histone deacetylase 1. Cell Death 9. Kramer OH. HDAC2: a critical factor in health and disease. Trends Differ 2005;12:395–404. Pharmacol Sci 2009;30:647–55. 19. Kaler P, Sasazuki T, Shirasawa S, Augenlicht L, Klampfer L. HDAC2 10. Song J, Noh JH, Lee JH, Eun JW, Ahn YM, Kim SY, et al. Increased deﬁciency sensitizes colon cancer cells to TNFalpha-induced apopto- expression of histone deacetylase 2 is found in human gastric cancer. sis through inhibition of NF-kappaB activity. Exp Cell Res 2008;314: APMIS 2005;113:264–8. 1507–18. 11. Noh JH, Eun JW, Ryu SY, Jung KH, Kim JK, Lee SH, et al. Increased 20. Zhu P, Martin E, Mengwasser J, Schlag P, Janssen KP, Gottlicher M. expression of histone deacetylase2 is found in human hepatocellular Induction of HDAC2 expression upon loss of APC in colorectal tumor- carconoma. Mol Cell Toxicol 2006;2:166–9. igenesis. Cancer Cell 2004;5:455–63. 12. Baur AS, Shaw P, Burri N, Delacretaz F, Bosman FT, Chaubert P. 21. Ushijima T, Sasako M. Focus on gastric cancer. Cancer Cell 2004; Frequent methylation silencing of p15(INK4b) (MTS2) and p16 5:121–5. (INK4a) (MTS1) in B-cell and T-cell lymphomas. Blood 1999;94: 22. Noh JH, Jung KH, Kim JK, Eun JW, Bae HJ, Xie HJ, et al. Aberrant 1773–81. regulation of HDAC2 mediates proliferation of hepatocellular carcino- 13. Weichert W, Roske€ A, Gekeler V, Beckers T, Ebert MP, Pross M, et al. ma cells by deregulating expression of G1/S cell cycle proteins. PLoS Association of patterns of class I histone deacetylase expression with One 2011;6:e28103. patient prognosis in gastric cancer: a retrospective analysis. Lancet 23. Fritsche P, Seidler B, Schuler€ S, Schnieke A, Gottlicher€ M, Schmid Oncol 2008;9:139–48. RM, et al. HDAC2 mediates therapeutic resistance of pancrea- 14. KimH,KimYH,KimSE,KimNG,NohSH.Concertedpromoter tic cancer cells via the BH3-only protein NOXA. Gut 2009;58: hypermethylation of hMLH1, p16INK4A, and E-cadherin in gastric 1399–409. carcinomas with microsatellite instability. J Pathol 2003;200: 24. Hanigan CL, Van Engeland M, De Bruine AP, Wouters KA, Weijenberg 23–31. MP, Eshleman JR, et al. An inactivating mutation in HDAC2 leads to 15. Shao Y, Gao Z, Marks PA, Jiang X. Apoptotic and autophagic cell death dysregulation of apoptosis mediated by APAF1. Gastroenterology induced by histone deacetylase inhibitors. Proc Natl Acad Sci U S A 2008;135:1654–64 e2. 2004;101:18030–5. 25. Pattingre S, Tassa A, Qu X, Garuti R, Liang XH, Mizushima N, et al. 16. Weichert W. HDAC expression and clinical prognosis in human malig- Bcl-2 antiapoptotic proteins inhibit Beclin 1-dependent autophagy. nancies. Cancer Lett 2009;280:168–76. Cell 2005;122:927–39.

www.aacrjournals.org Mol Cancer Res; 11(1) January 2013 73

Targeted Inactivation of HDAC2 Restores p16INK4a Activity and Exerts Antitumor Effects on Human Gastric Cancer

Jeong Kyu Kim, Ji Heon Noh, Jung Woo Eun, et al.

Mol Cancer Res 2013;11:62-73. Published OnlineFirst November 21, 2012.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-12-0332

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2012/11/21/1541-7786.MCR-12-0332.DC1

Cited articles This article cites 25 articles, 4 of which you can access for free at: http://mcr.aacrjournals.org/content/11/1/62.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/11/1/62.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://mcr.aacrjournals.org/content/11/1/62. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.