Open Full Page

Published OnlineFirst June 10, 2014; DOI: 10.1158/1541-7786.MCR-14-0034

Molecular Cancer Chromatin, Gene, and RNA Regulation Research

EZH2 Represses Target Genes through H3K27-Dependent and H3K27-Independent Mechanisms in Hepatocellular Carcinoma

Shu-Bin Gao1,2, Qi-Fan Zheng1,2, Bin Xu1,2, Chang-Bao Pan1, Kang-Li Li1, Yue Zhao1, Qi-Lin Zheng1, Xiao Lin1,2, Li-Xiang Xue3, and Guang-Hui Jin1,2

Abstract Alterations of polycomb group (PcG) genes directly modulate the trimethylation of histone H3 lysine 27 (H3K27me3) and may thus affect the epigenome of hepatocellular carcinoma (HCC), which is crucial for controlling the HCC cell phenotype. However, the extent of downstream regulation by PcGs in HCC is not well defined. Using cDNA microarray analysis, we found that the target gene network of PcGs contains well-established genes, such as cyclin-dependent kinase inhibitors (CDKN2A), and genes that were previously undescribed for their regulation by PcG, including E2F1, NOTCH2,andTP53. Using chromatin immunoprecipitation assays, we demonstrated that EZH2 occupancy coincides with H3K27me3 at E2F1 and NOTCH2 promoters. Interestingly, PcG repress the expression of the typical tumor suppressor TP53 in human HCC cells, and an increased level of PcG was correlated with the downregulation of TP53 in certain HCC specimens. Unexpectedly, we did not find obvious H3K27me3 modification or an EZH2 binding signal at the TP53 promoters, suggesting that PcG regulates TP53 expression in an H3K27me3-independent manner. Finally, the reduced expression of PcGs effectively blocked the aggressive signature of liver cancer cells in vitro and in vivo. Implications: Taken together, our results establish the functional and mechanistic significance of certain gene regulatory networks that are regulated by PcGs in HCC. Visual Overview: http://mcr.aacrjournals.org/content/12/10/1388/F1.large.jpg. Mol Cancer Res; 12(10); 1388–97. 2014 AACR.

Introduction nance complex, includes B-lymphoma moloney murine leukemia virus insertion region-1 (BMI1), chromobox Polycomb group (PcG) proteins are highly conserved – epigenetic gene silencers that play important roles in the homolog 8 (CBX8), and others (1 3). The EZH2 SET (suppressor of variegation, enhancer of zeste, and trithorax) maintenance of cell fates, developmental programs, and fi tumorigenesis (1). PcG proteins bind to specific regions of domain speci cally catalyzes the trimethylation of Histone fi 3 lysine 27 (H3K27me3). Methylated H3K27 is recognized gene promoters and direct posttranslational modi cations at fi certain histone sites, thereby silencing gene expression (2–5). by other speci c binding proteins, such as PRC1 compo- nent, which leads to the compression of chromatin structure In mammals, PcG is made up of at least two multiprotein – complexes: PRC1 (polycomb repressive complex 1) and and the repression of gene transcription (1 3). In addition, PRC2. These complexes work together to carry out their some PRC1 complexes can regulate gene expression repressive effect (1). PRC2 is the core of PcG in humans, by monoubiquitylation at Lys 119 of histone H2A (H2AK119ub) via the ubiquitin ligases RING1A (ring which consists of the proteins enhancer of zeste homolog 2 fi (EZH2), suppressor of zeste 12 (SUZ12), and embryonic nger protein) and RING1B (3). ectoderm development (EED; ref. 1). PRC1, the mainte- Hepatocellular carcinoma (HCC) is the most common type of liver cancer and the third most common cause of cancer mortality worldwide (6). EZH2 is a useful biomarker 1Department of Basic Medical Sciences, Medical College, Xiamen Univer- of the aggressive stages of HCC, and its overexpression is sity, Xiamen, P.R. China. 2Fujian Provincial Key Laboratory of Chronic Liver associated with the malignant progression of HCC (7). The Disease and Hepatocellular Carcinoma, Xiamen University, Xiamen, P.R. China. 3Department of Biochemistry and Molecular Biology, Health Sci- same study reported a high level of global H3K27me3 in ence Center, Peking University, Beijing, P.R. China. HCC, which is closely correlated with the vascular invasion Note: Supplementary data for this article are available at Molecular Cancer of HCC (8). LncRNA-HEIH is an oncogenic long-non- Research Online (http://mcr.aacrjournals.org/). coding RNA that promotes HCC progression by upregulat- Corresponding Author: Guang-Hui Jin, Xiamen University, Xiang'an ing the expression of EZH2 in humans (9). Several micro- South Road, Xiamen, Fujian, P.R. China 361102. Phone: þ86-592-218- RNAs, such as the putative tumor suppressor microRNA- 6173; Fax: þ86-592-218-1535; E-mail: [email protected] 124, repress the expression of EZH2. Dysregulation of these doi: 10.1158/1541-7786.MCR-14-0034 microRNAs could contribute to EZH2 overexpression in 2014 American Association for Cancer Research. cancer (10). Inversely, EZH2 exerts its oncogenic function

1388 Mol Cancer Res; 12(10) October 2014

Downstream Regulation by PcG in HCC

by way of epigenetic silencing of well-characterized tumor Western blotting and immunofluorescence suppressor miRNAs, such as miR139-5p, miR125b, Western blotting and immunofluorescence (IF) staining miR101, let-7c, and miR200b (11). These findings provide were performed as previously described (16). Antibodies are significant insights into the molecular mechanisms of PcG listed in the Supplementary Table S1. involvement in hepatocellular carcinogenesis. Tumor protein p53 (TP53), a typical tumor suppressor, is Real-time qRT-PCR a major genetic player in the development of various cancers, qRT-PCR was performed as previously described (16) including HCC (12). The aberrant expression of EZH2 has using an ABI Step One detection system with the primers been associated with P53 alteration and contributed to listed in Supplementary Table S2. qRT-PCR reactions were progression and pool prognosis in human squamous cell performed using SYBR Green (Roche Applied Science) carcinoma of the esophagus (13). Activation of TP53 sup- technology and were repeated at least three times. presses EZH2 expression through the repression of the EZH2 gene promoter (14). Inversely, the reduction of Cell proliferation assays EZH2 expression by RNA interference increases the expres- Cell growth and viability were evaluated by using either sion of TP53 (15). These features point to an interesting Bromodeoxyuridine (5-bromo-20-deoxyuridine, BrdUrd) or negative feedback loop between oncogenic EZH2 and 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2-H-tetrazolium tumor suppressor TP53. Despite the importance of EZH2 bromide (MTT) cell proliferation assays. The cells were in controlling the aggressive nature of HCC, the regulation plated at 2 105 cell/mL in 100 mL/well of DMEM media, profile of tumor suppressors by PcG in HCC is not well and the BrdUrd incorporation and detection were per- defined. formed using a cell proliferation ELISA BrdU kit (Millipore) In this study, we demonstrate potential tumor-promoting according to the provided protocol. For the MTT assay, the actions of PcG in HCC by the H3K27me3-dependent and cells were seeded at a density of 1 103 cells/well into 96- H3K27me3-independent regulation of downstream gene well plates, and measured as according to the manufacturer's expression. protocol (Merck). For BrdUrd or MTT detection, the optical density at 450- or 570-nm absorbance was measured Materials and Methods by using an ELISA plate reader (Bio-Tek Instruments), Cell culture respectively. For both assays, the mean absorbance values The human HCC cell line HepG2 (wild-type P53) and of at least three independent experiments were used for Hep3B (P53 null) were obtained from the Type Culture statistical analysis. Collection of the Chinese Academy of Sciences (Shanghai, China). PLC/PRF5 (human HCC with P53 mutant), Colony-forming assays SMMC-7721 (human HCC with wild-type P53), and The colony-forming assays were performed as previously HO-8910 (human ovarian cancer cell) were obtained from described (16). In brief, 1,000 cells were seeded in 6-well the China Center for Type Culture Collection (Wuhan, plates for 10 days. Colonies were stained with 0.05% crystal China). Wilm's (human nephroblastoma), MCF-7 (human violet, photographed, and counted in triplicate using micro- mammary gland adenocarcinoma), and A549 (human lung scope (Nikon Corporation). adenocarcinoma) were obtained from the Cell Culture Center of the School of Basic Medicine, Peking Union Cell migration assay Medical College (Beijing, China). The cell cultures were Migration assays were performed, as previously described performed as previously described (16–19). Briefly, the cells (19), in 24-well transwell plates (Millipore) using uncoated were cultured in DMEM (Gibco), RPMI-1640 medium polycarbonate membranes with 8-mm pores. Serum-starved (Gibco), or F12 (Gibco) supplemented with 10% FBS cells were harvested and resuspended at a concentration of 4 (Hyclone) and 1% penicillin–streptomycin. 1 10 cells per 0.2 mL in serum-free medium containing 10% BSA, and added to the upper chamber. At 24 hour after Chromatin immunoprecipitation assays incubation, nonmigrated cells were scraped off the upper Chromatin immunoprecipitation (ChIP) assays were per- side of the filter, and filters were stained with 0.1% crystal formed as previously described (16). In brief, 1 106 violet. The number of migrated cells on the reverse side of the HepG2 cells were treated with 1% formaldehyde for 10 membrane was quantified by average cell counts from six minutes at 37C;, followed by pulsed ultrasonication to random fields in each well, under a microscope at 100 shear cellular DNA. ChIP assays were then carried out with magnifications. Each condition was assayed in triplicate the indicated antibodies according to the protocol of the wells, and the resulting data were subjected to statistical ChIP Assay Kit (Millipore). After overnight incubation with analysis using the Student t test. the antibodies, protein A- or protein G-agarose beads were added. The cross-links between nuclear proteins and geno- Terminal deoxynucleotidyl transferased UTP nick end mic DNA were reversed, and DNA pulled down by the labeling assay antibody was purified by phenol/chloroform extraction. The Terminal deoxynucleotidyl transferased UTP nick end primer pair sequences and antibodies for the ChIP assays are labeling (TUNEL) assay (Roche Applied Science) was used shown in Supplementary Tables S1 and S2. to detect apoptotic cells. Apoptotic cells were determined by

www.aacrjournals.org Mol Cancer Res; 12(10) October 2014 1389

Gao et al.

counting positive-stained cells in five randomly selected that specifically target EZH2, SUZ12, BMI1,orCBX8. fields in each tumor. At least three tumors in each mouse Western blotting results indicate that correlated with the were randomly selected. reduction level of EZH2, SUZ12, BMI1, and CBX8, the global level of H3K27me3 was remarkably reduced in both Statistical analysis PcG KD HepG2 cells (Supplementary Fig. S1A). The All the statistical analyses were performed using SPSS CBX8–shRNA3 failed to reduce the expression of CBX8 software version 13.0. Data from cell growth, clonal assays, and was unable to reduce the global H3K27me3 levels and cell migration experiments were analyzed by two-tailed (Supplementary Fig. S1A). IF staining results indicate that Student t tests. The results for parametric variables are EZH2 was colocalized with SUZ12 and H3K27me3 at the expressed as the mean SD or the mean SEM. In all nucleus (Supplementary Fig. S1B). In addition, the stably cases, P < 0.05 was considered statistically significant. ectopic expression of EZH2 substantially increased the levels of SUZ12 and H3K27me3 in HepG2 (Supplementary Fig. S1B). These data indicate that the components of PcG Results effectively catalyze the global trimethylation of H3K27 in A role for EZH2 in hepatic cancer cell growth in vitro and liver cancer cells. Next, the shRNA-mediated KD of EZH2, in vivo SUZ12, CBX8, or BMI1 dramatically suppressed the A recent report indicates that knockdown (KD) of EZH2 proliferation of HepG2 cells in vitro (Fig. 1A). Control – suppresses HCC tumorigenicity and extrahepatic metastasis and EZH2 shRNA cells were synchronized in the G0 G1 in vivo (11); nonetheless, few studies have explored whether phase by serum starvation and reentered cell cycle with the other PcG components are involved in the control of the restoration of serum culture. Cyclin B1 (CCNB1) protein aggressive phenotype of HCC cells. To explore a potential expression was analyzed by Western blotting, which cor- – functional implication of PcG components in HCC, we responded to the G2 M checkpoint. We found that both stably transfected HepG2 cells, which are often used to study control and EZH2 shRNA HepG2 cells cultured in a the behavior of HCC cells (16), with either vectors or serum-free medium had a relatively low level of CCNB1 constructs expressing one of the 2 to 3 distinct shRNAs expression. Following the addition of serum into the

Figure 1. A role for EZH2 in liver cancer cell proliferation and migration. A, the proliferation rates of HepG2 cells that were stably transfected with shEZH2, shSUZ12, shBMI1, or shCBX8 were estimated using an MTT assay. B, the expression of CCNB1 and pCCNB1 was analyzed by Western blotting in shRNA- mediated KD of EZH2 in HepG2 cells at the indicated time points. C and D, the proliferation of HepG2 cells transiently transfected with Luc, EZH2, or SUZ12 siRNA was estimated by BrdUrd (OD450) or MTT (OD570) assay. E, HepG2 cells transiently transfected with Luc or EZH2 siRNA were added to the upper-ﬁlter, and 1 104 cell migration was determined by transwell chamber assay. F, colony-forming assays were performed using HepG2 cells transfected with Luc or EZH2 siRNA.

1390 Mol Cancer Res; 12(10) October 2014 Molecular Cancer Research

Downstream Regulation by PcG in HCC

medium, the protein expression of phospho-CCNB1 dra- tumor growth and weight in subcutaneous xenografts (Fig. matically increased at 4 to 24 hours after serum stimulation 2A and B). In addition, the incidence of apoptotic tumor in control HepG2 cells (Fig. 1B, lanes 3–7). However, in cells, as assessed by the TUNEL assay, was significantly shEZH2 HepG2 cells, the protein expression levels started increased in EZH2, SUZ12, and CBX8 shRNA HepG2 to increase at 6 hours and returned to a level less than xenografts, with no obvious effect observed for BMI1 baseline after 12 hours (Fig. 1B, lanes 11 and 12). These shRNA (Fig. 2C and D). Collectively, these results strongly data indicate that EZH2 promotes HepG2 cell prolifera- argue that both components of PcG play a crucial role in – in vitro in vivo tion at least partly by accelerating G2 Mprogress.Con- promoting liver cancer and . sistent with this hypothesis, the siRNA-mediated KD of either EZH2 or SUZ12 dramatically suppressed the pro- Identification of downstream pathway of PcG in HCC liferation of HepG2 cells at the indicated time point, as Although recent evidence suggests that the abnormal determined by BrdUrd or MTT incorporation, respectively expression of EZH2 and a high level of global H3K27me3 (Supplementary Fig. S1C; Fig. 1C and D). Because cancer can be attributed to the aggressive nature of HCC (7, 8), the metastasis usually involves enhanced cell migration, we extent of downstream regulation by PcGs in HCC is not well sought to determine whether one of the functions of EZH2 defined. To obtain a broader understanding of the molecular is involved in controlling hepatic cancer cell migration. To network of PcG in HCC, cDNA microarray analysis was this end, we performed a modified transwell chamber performed on shRNA KDs of EZH2, SUZ12, BMI1, or assay to evaluate the impact of altered EZH2 expression CBX8 HepG2 cells. Interestingly, the results indicated that on HCC cell migration. The results indicate that EZH2 PRC1 (CBX8 and BMI1) and PRC2 (EZH2 and SUZ12) siRNA KD significantly reduced the extent of migration display very different characteristics of gene expression (Fig. 2E, P < 0.001) and the colony-forming ability of regulation in HCC. We found genes that are targeted by HepG2 cells (Fig. 2F, P < 0.001). PRC2 without PRC1 and vice versa (Fig. 3A and Supple- We next set out to explore whether PcGs promote the mentary Table S3, GSE 54108). The microarray results growth of HepG2 cell–derived tumors in nude mice. The indicated that EZH2, SUZ12, BMI1, and CBX8 shRNA size of the solid tumor was measured after various periods revealed a significant coregulation of only 61 or 159 genes in following transplantation. The shRNA-mediated KD of HepG2 cells up or down, respectively (Fig. 3B). Figure 3C EZH2, SUZ12, CBX8, or BMI1 significantly suppressed highlights the major molecular functions (MF) from the list

Figure 2. The KD of PcGs represses tumor growth in HepG2 xenografts. A, the control, shEZH2, shSUZ12, shBMI1, or shCBX8 HepG2 cells were injected subcutaneously into nude mice. Tumor formation was examined on day 6 after transplantation. The tumor volume of recipient mice (N ¼ 10 per group) is shown. B, recipient mice were sacrificed on day 30 after transplantation; the tumor weights were measured (N ¼ 10). C and D, the recipient mice were sacrificed 30 days after transplantation, and the apoptotic cells were detected by TUNEL assay (green) with Propidium Iodide (PI) staining (red) in these tumors on the same day. The number of apoptotic cells was quantified.

www.aacrjournals.org Mol Cancer Res; 12(10) October 2014 1391

Gao et al.

ABCATPase regulator activity shLucshEZH2shSUZ12shCBX8shBMI1 shLucshEZH2shSUZ12shCBX8shBMI1 tRNA-specific ribonuclease activity Structural constituent of ribosome Downstream of PRC2 WDR75 shSUZ12 shCBX8 RWDD1 RNA helicase activity shEZH2 shBMI1 BNIP2 251 802 Ribonuclease activity CYB5D1 81 Hydrogen ion transmembrane transporter activity MAL2 3076 196 LYRM4 rRNA binding 354 10 ATP6V1C1 562 739 RNA-dependent ATPase activity TYW1 61 NADH dehydrogenase (quinone) activity PPP2R2B 72 20 GTPBP4 12 139 NADH dehydrogenase (ubiquinone) activity UBL5 22 NADH dehydrogenase activity ACAT1 IPO5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 AKAP1 Upregulated genes GART MAP kinase tyrosine/serine/threonine RPL26L1 Phosphatase activity RDH14 Oxygen transporter activity

SRSF10 MAP kinase phosphatase activity Downstream of PRC1 BLOC1S3 shCBX8 shSUZ12 Aromatase activity VAPB shEZH2 shBMI1 TJP2 238 2157 Oxygen binding HP1BP3 Oxidoreductase activity 1861 148 2048 NGFRAP1 Coreceptor activity MKI67IP 742 39 Rab GTPase binding C13orf27 305 810 Protein tyrosine/serine/threonine phosphatase activity MRPL45 159 BAMBI 113 24 Acting on the aldehyde or oxo group of donors 53 70 FNBP4 Phospholipid transporter activity RAB5B 26 Insulin receptor binding UBE2K Top 200 Downregulated genes 0246810 12 14 058 upregulated genes Fold enrichment ((Count/Pop.Hits)/(List.Total/Pop.Total))

Figure 3. Screen of target network of PcG in HepG2. A, expression microarray experiments were performed with control, shEZH2, shSUZ12, shCBX8, and shBMI1 HepG2 cells, and the Venn diagram analysis of target genes regulated by EZH2, SUZ12, BMI1, and CBX8 shRNA in the indicated categories. Hierarchical clustering analysis was performed for the top 200 most affected target genes in microarray assays. Each row, a single gene. The microarray data have been deposited in NCBI-GEO under accession numbers GSE 54108 B, Venn diagram analysis of target genes regulated by EZH2, SUZ12, CBX8, and BMI1 in the indicated categories. C, GO enrichment analysis of the expression microarray assays of PRC2 or PRC1 in HepG2 cells, categorized by MF.

of limbal genes based on the significance of the increase in relationship between PcG and TP53 using HepG2 cells. either CBX8 or EZH2 shRNA KD (P < 0.05). ATPase The reduction of EZH2 levels by shRNA increased the regulator activity, ribonuclease activity, and NADH dehy- expression of TP53 and its downstream effector P21 levels in drogenase activity–associated genes have the highest signif- HepG2 cells, as determined by qRT-PCR and Western icance in both the EZH2 and SUZ12–shRNA microarray blotting (Fig. 4A). Similarly, the KD of EZH2 by three Gene Ontology (GO) term analysis. MAP kinase phospha- distinct siRNA clearly increased TP53 and P21 mRNA tase activity, oxygen transporter activity, and oxidoreductase expression (Fig. 4B). Next, the shRNA-mediated KD of activity were involved in both CBX8 and BMI1-shRNA GO CBX8 increased the mRNA expression of TP53 and P21 in term analysis. HepG2 cells (Fig. 4C). The CBX8–shRNA3 failed to reduce CBX8 expression and was unable to increase the level of PcG indirectly regulates TP53 expression in HCC TP53 or P21 mRNA (Fig. 4C). In time-course detection, the Interestingly, the TP53 gene, a tumor suppressor, was mRNA expression of TP53 was elevated by the substantial strikingly regulated by both PRC1 and PRC2. Although KD of EZH2 or CBX8 siRNA at days 2 and 4 in both extensive studies have demonstrated that the TP53 mutation HepG2 and SMMC-7721 cells (Supplementary Fig. S2A is a major genetic event for HCC, the transcriptional and S2B). Furthermore, the KD of SUZ12 also increased the regulation of TP53 by critical effectors in HCC has not protein levels of TP53 and P21 in HepG2 cells (Fig. 4D). been well defined. It has been reported that the expression of EZH2 siRNA KD moderately increased the mRNA level of EZH2 is negatively correlated with the p53 pathway in TP53 in another type of human cancer cell line, lung human cancers, suggesting that EZH2 regulates p53 activity adenocarcinoma A549 (Fig. 4E). However, we did not or expression (15). We thus turned our attention to the observe an obvious effect of EZH2 on repressing TP53 impact that PcG has on the regulation of TP53 expression. mRNA level in other type tumor cell lines, HO-8910, MCF- To rule out the effect of the TP53 mutation on its own 7, and Wilms (Supplementary Fig. S2C). This result sug- expression, we first performed a mutation screen of the entire gests that EZH2 regulates TP53 expression in a tissue- TP53 locus in HepG2 cells. We focused our analysis on specific manner. Furthermore, inversely correlated with the novel changes that were predicted to affect the protein- reduction of PcGs, the mRNA levels of TP53 and CDKN2A coding sequence. We discovered only one variation, P72R, were increased in PcG KD HepG2 xenografts (Fig. 4F). which has been described in a previous report. P72R affects Interestingly, we further found that EZH2 did not regulate exon 4 of the TP53 gene in HepG2 and is considered a the TP53 mRNA expression in Hep3B with p53-null neutral SNP (20). Thus, we investigated the regulatory background and PLC/PRF/5 with p53-mutant background

1392 Mol Cancer Res; 12(10) October 2014 Molecular Cancer Research

Downstream Regulation by PcG in HCC

Figure 4. PcG indirectly regulates TP53 expression in HCC. A, the mRNA and protein levels of EZH2, TP53, and P21 were determined in shLuc and shEZH2 HepG2 cells. B, control (siLuc) and each of the three distinct siRNAs that speciﬁcally target EZH2 were transiently transfected into HepG2, and TP53 and P21 mRNA expression was determined by qRT-PCR at 4 days after transfection. C, the mRNA expression of CBX8, TP53, and P21 was detected by qRT-PCR in control and shCBX8 HepG2 cells. D, the expression of TP53 and P21 proteins was detected by Western blotting assays in control and shSUZ12 HepG2 cells. E, the mRNA levels of EZH2 and TP53 were determined by qRT-PCR in A549 cells transfected with control or EZH2 siRNA at 2 days after transfection. F, qRT-PCR detection of PcGs, TP53, and CDKN2A mRNA levels in PcG KD xenograft (N ¼ 9 or 10).

(Supplementary Fig. S2D). This result suggests that EZH2 negatively correlated with the upregulation of EZH2, represses TP53 expression depending on wild-type TP53 CBX8, and BMI1, but not SUZ12 (Fig. 5). This observation genomic status. further suggests a potential role for PcGs in the regulation of To gain insight into the role of the PcG in regulating TP53 TP53 in HCC. expression, we performed qRT-PCR analysis on total RNA We were interested in whether PcG regulated TP53 extracted from HCC liver samples and their matched non- transcription through an epigenetic mechanism. Thus, we neoplastic counterparts. First, the whole-exome sequencing designed ChIP primers for the regions spanning 25-kb of TP53 in the primary HCC tissues of 48 patients excluded upstream and downstream of the transcriptional start site the typical TP53 mutation, and we further analyzed the of the TP53 promoter (Supplementary Fig. S3A) and per- mRNA expression of EZH2, SUZ12, CBX8, BMI1, and formed ChIP assays using control and shEZH2 HepG2 cells. TP53. These HCCs included cases with wild-type TP53 (19 IgG and H3-ChIP were used as negative and positive cases), TP53 with synonymous mutations (three cases), controls, respectively (Supplementary Fig. S3B). Unexpect- TP53 with the neutral SNP variation (P72R, 21 cases), and edly, we did not find any obvious H3K27me3 modification TP53 with four previously unreported SNP variations (a or an EZH2-binding signal at the promoter loci using total of five cases; data not shown). Using a large cohort of H3K27me3 and EZH2 antibody ChIP assays, respectively HCC specimens (n ¼ 48), we confirm here that the (Supplementary Fig. S3C and S3D), suggesting that EZH2 expression of EZH2, CBX8, BMI1, and SUZ12 were sig- regulates TP53 expression through an H3K27me3-inde- nificantly overexpressed in HCC samples compared with pendent mechanism. their nontumoral counterparts (Fig. 5). In addition, we found that the TP53 mRNA expression levels are decreased PcG directly controls target gene transcription through in HCC specimens compared with adjacent tissues (Fig. 5). H3K27me3 Spearman analysis performed using qRT-PCR expression In our previous ChIP-on-chip assay (GSE52300), we data indicates that the mRNA expression of TP53 was screened out the H3K27me3-related target gene network

www.aacrjournals.org Mol Cancer Res; 12(10) October 2014 1393

Gao et al.

Figure 5. A negative correlation between PcGs and TP53 expression in HCC specimens. The mRNA levels of TP53, EZH2, SUZ12, CBX8, and BMI1 were determined by qRT-PCR in 48 cases of HCC specimens and adjacent tissues, and the correlation between PcGs and TP53 expression is shown (linear regression analysis, Pearson correlation). The mRNA levels were normalized to those of b-actin.

of PcG, which contains well-established genes, such as that EZH2 was present at the E2F1 promoter in HepG2 cells CDKN2A, and other previously unstudied genes in liver (Fig. 6B). As expected, the KD of EZH2 by shRNA cancer cells, including E2F1 and NOTCH2 (21). The decreased the binding of EZH2 to the E2F1 loci (Fig. functional importance of these genes in controlling liver 6B). EZH2–shRNA also led to a significant decrease in function has been established (22, 23). This finding suggests the H3K27me3 level at the promoter of the E2F1 locus a preferential epigenetic action of EZH2 on the regulation of (Fig. 6C). In addition, we demonstrated that EZH2 and the transcription of E2F1 and NOTCH2 in HCC. To test H3K27me3 were present at the promoter of NOTCH2 and this hypothesis, we performed ChIP assays using three that the KD of EZH2 by shRNA led to a severe decrease in distinct pairs of ChIP primers at E2F1 promoter loci (Fig. the binding of EZH2 and the H3K27me3 levels at the 6A). Using an EZH2-specific antibody, we demonstrated promoters in HCC (Fig. 6D–F). Therefore, we conclude

Figure 6. PcG directly regulates E2F1 and NOTCH2 expression through H3K27me3 in HCC. A and D, schematic representation of human E2F1 and NOTCH2 primer pairs (PP) used for ChIP assays. B and E, ChIP assays using antibody against EZH2 were performed using the indicated primer pairs of the E2F1 or NOTCH2 promoters in control and shEZH2 HepG2 cells. Rabbit-IgG was used as a negative control. C and F, H3K27me3 ChIP assays were performed using the indicated primer pairs of the E2F1 or NOTCH2 promoter in control and shEZH2 HepG2 cells. Rabbit- IgG was used as a negative control.

1394 Mol Cancer Res; 12(10) October 2014 Molecular Cancer Research

Downstream Regulation by PcG in HCC

that the transcription of E2F1 and NOTCH2 were directly binding factor (CTCF) that binds at the P53 promoter repressed by EZH2-mediated H3K27me3 in HCC cells. and protects the P53 gene promoter against repressive histone marks (30). Interesting findings have shown that EZH2 and SUZ12 were directly bound at CTCF pro- Discussion moter loci with H3K27me3 modification and that the EZH2 promotes HCC cell motility and metastasis by increased accumulation of H3K27me3 at the CTCF repressing tumor suppressor miRNAs (11), but the specific promoter is associated with the reduction of CTCF in gene network and epigenetic characteristics are not well HCC specimens (21). These results suggest that PcG established. The CDKN2A locus encodes two distinct pro- regulates P53 expression in HCC at least partly through INK4A ARF teins, p16 and p14 , both of which have been CTCF in HCC. Functionally and mechanistically signif- implicated in a variety of cancers. In acute myeloid leukemia icant work remains to be done to identify the competitive cells, H3K27me3 and EZH2 seem to be distributed pri- functions between PcG and CTCF in HCC. TP53 was marily at the promoters of CDKN2A (24). It has also been frequently mutated in HCC and is a major genetic event demonstrated that H3K27me3 is an early-epigenetic event for the development of HCC. Our results emphasize the INK4A of p16 silencing in hepatoma cells (25). Our previous interesting fact that, independent of genetic mutation, the results indicated that EZH2 and H3K27me3 cooccupi- transcriptional silencing of TP53 by PcG contributes to ed the promoter of CDKN2A (21), indicating that the aggressive nature of HCC. Recently, Lu and colleagues H3K27me3-dependent repression is a common feature at (31) reported that p53 is not the target of EZH2 in INK4-ARF in tumors, including HCC. E2F1, a tumor nasopharyngeal carcinoma. Similar to this, we did not suppressor, binds to the promoter of EZH2 and EED and observe an obvious regulation effect of EZH2 on repres- functionally controls their expression in prostate cancer (26). sing TP53 mRNA level in other type tumor cell lines, An interesting negative-feedback loop was found in the including HO-8910, MCF-7, and Wilm. This result present study, in which EZH2 repressed E2F1 expression suggests an interesting tissue-specificregulationprofile of by binding to the promoter of E2F1 in HCC. Our findings EZH2 on TP53 expression. Conversely, Tang and col- highlight H3K27me3 remodeling as an interesting new leagues (14) reported that P53 repress EZH2 transcription direction in the elucidation of the E2F1 transcriptional through binding to its promoter locus in primary human regulation mechanism. fibroblasts. These findings suggest a potential feedback In agreement with a previous report (27), we identified loop of tumor suppressor P53 and tumor promoter gene genes that are targeted by PRC2 without PRC1 and vice EZH2, an important interplay governing tumor activation versa. Although EZH2 and SUZ12 occupy the promoter and suppression in the development of HCC. This process regions of thousands of human genes, only 340 genes is similar to our previous reports that the interplay share H3K27me3 in HepG2 (21). This result suggests between oncogene K-Ras and tumor suppressor menin that PcG regulates gene expression in both an plays an important role in regulating the development of H3K27me3-dependent and an H3K27me3-independent lung cancer (17). However, it is unclear whether the manner. The mutation-induced inactivation of TP53, activated P53 suppresses the expression of EZH2 in HCC. which is one of the important tumor suppressors, has The link between p53 and EZH2 in controlling aggressive frequently been associated with a variety of human can- phenotype of HCC needs further investigation. cers, including HCC. Nevertheless, the transcriptional Alteration of PcG components was observed in various repression of TP53 by critical effectors in HCC has not cancers such as melanoma, lymphoma, and breast and been well defined. In the present study, we found that the prostate cancer, leading to the hypothesis that PcG has expression of TP53 was dramatically repressed by both tumor-promoting functions (32). The overexpression of PRC1 and PRC2, and a significant negative correlation EZH2 promotes the growth and cell invasion of immor- between the expression of PcGs and TP53 was found in talized human mammary epithelial cells (33), and the primary HCC specimens. However, we failed to observe reduction of EZH2 expression inhibits prostate cancer cell an obvious impact of H3K27me3 on TP53 expression, proliferation in vitro (34). Further studies demonstrate suggesting that PcG regulates TP53 expression in an that the oncogenic effect of EZH2 is linked to the H3K27me3-independent manner in HCC. This activity H3K27me3-dependent repression of the tumor suppres- is presumably dependent on the direct regulation of sor genes, including E-cadherin, RUNX3, INK4/ARF (also CDKN2A or E2F1, which are known upstream regulators known as CDKN2A; refs. 24, 35, 36), and others. Here, of the TP53 pathway (28, 29). The ARF product func- we found that the alterations of PcG proteins directly tions as a stabilizer of the p53 protein, which was respon- modulate H3K27me3 and may thus affect the HCC sible for the degradation of p53 (28). CDKN2A product epigenome, which is crucial for HCC cell growth both interacts with and inhibits the oncogenic action of in vitro and in vivo. We also observed that the reduction of – MDM2 by blocking MDM2-induced degradation of EZH2 expression induces G2 M cell-cycle arrest in HCC p53 and enhancing p53-dependent transactivation and cells. This finding is consistent with a previous report that CDKN2A apoptosis (28). Confusingly, we found that both protein induces G2 arrest in a p53-independent manner and mRNA levels of TP53 were substantially repressed by by preventing the activation of cyclin B1/CDC2 com- PcG in HCC cells. A previous report found a CCCTC- plexes (37). Our findings demonstrate a potential tumor-

www.aacrjournals.org Mol Cancer Res; 12(10) October 2014 1395

Gao et al.

promoting action of EZH2 by directly or indirectly Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Q.-F. Zheng, C.-B. Pan, Q.-L. Zheng, X. Lin, regulating the expression of tumor suppressors with crit- L.-X. Xue, G.-H. Jin ical functions in HCC, at least partly through the Writing, review, and/or revision of the manuscript: S.B. Gao, Q.-F. Zheng, – – fi L.-X. Xue, G.-H. Jin CDKN2A TP53 P21 pathway. These ndings also Administrative, technical, or material support (i.e., reporting or organizing data, reveal the EZH2 pathway as a potential target for the constructing databases): S.B. Gao, K.-L. Li treatment of EZH2-positive human HCC with EZH2 inhibitors, such as GSK126, which is a specificsmall Acknowledgments molecular inhibitor of the catalytic SET domain of EZH2 We appreciate the valuable comments from other members of our laboratories. (38). Significant work remains to identify the major target of PcGs in HCC. Grant Support This work was supported by the grants from the Natural Science Foundation of Disclosure of Potential Conflicts of Interest China (81101763, to S.-B. Gao; 91229111 and 81272719, to G.-H. Jin; 81301754, to B. Xu) and the Natural Science Foundation of Fujian Province (2011J06016 to No potential conflicts of interest were disclosed. G.-H. Jin). The costs of publication of this article were defrayed in part by the payment of page Authors' Contributions charges. This article must therefore be hereby marked advertisement in accordance Conception and design: S.B. Gao with 18 U.S.C. Section 1734 solely to indicate this fact. Development of methodology: B. Xu, Q.-L. Zheng Acquisition of data (provided animals, acquired and managed patients, provided Received January 20, 2014; revised April 30, 2014; accepted May 19, 2014; facilities, etc.): S.B. Gao, Q.-L. Zheng, G.-H. Jin published OnlineFirst June 10, 2014.

References 1. Sparmann A, van Lohuizen M. Polycomb silencers control cell fate, 14. Tang X, Milyavsky M, Shats I, Erez N, Goldfinger N, Rotter V. Activated development and cancer. Nat Rev Cancer 2006;6:846–56. p53 suppresses the histone methyltransferase EZH2 gene. Oncogene 2. Cao R, Wang L, Wang H, Xia L, Erdjument-Bromage H, Tempst P, et al. 2004;23:5759–69. Role of histone H3 lysine 27 methylation in polycomb-group silencing. 15. Choi JH, Song YS, Yoon JS, Song KW, Lee YY. Enhancer of Science 2002;298:1039–43. zeste homolog 2 expression is associated with tumor cell pro- 3. Cao R, Tsukada Y, Zhang Y. Role of Bmi-1 and Ring1A in H2A liferation and metastasis in gastric cancer. APMIS 2010;118: ubiquitylation and Hox gene silencing. Mol Cell 2005;20:845–54. 196–202. 4. Milne TA, Briggs SD, Brock HW, Martin ME, Gibbs D, Allis CD, et al. 16. Xu B, Li SH, Zheng R, Gao SB, Ding LH, Yin ZY, et al. Menin promotes MLL targets SET domain methyltransferase activity to Hox gene hepatocellular carcinogenesis and epigenetically up-regulates Yap1 promoters. Mol Cell 2002;10:1107–17. transcription. Proc Natl Acad Sci U S A 2013;110:17480–5. 5. Yokoyama A, Somervaille TC, Smith KS, Rozenblatt-Rosen O, Meyer- 17. Wu Y, Feng ZJ, Gao SB, Matkar S, Xu B, Duan HB, et al. Interplay son M, Cleary ML. The menin tumor suppressor protein is an essential between menin and K-Ras in regulating lung adenocarcinoma. J Biol oncogenic cofactor for MLL-associated leukemogenesis. Cell 2005; Chem 2012;287:40003–11. 123:207–18. 18. Xu B, Zeng DQ, Wu Y, Zheng R, Gu L, Lin X, et al. Tumor suppressor 6. Forner A, Llovet JM, Bruix J. Hepatocellular carcinoma. Lancet 2012; menin represses paired box gene 2 expression via Wilms tumor 379:1245–55. suppressor protein-polycomb group complex. J Biol Chem 2011; 7. Cai MY, Tong ZT, Zheng F, Liao YJ, Wang Y, Rao HL, et al. EZH2 286:13937–44. protein: a promising immunomarker for the detection of hepatocellular 19. Feng ZJ, Gao SB, Wu Y, Xu XF, Hua X, Jin GH. Lung cancer cell carcinomas in liver needle biopsies. Gut 2011;60:967–76. migration is regulated via repressing growth factor PTN/RPTP beta/ 8. Cai MY, Hou JH, Rao HL, Luo RZ, Li M, Pei XQ, et al. High expression of zeta signaling by menin. Oncogene 2010;29:5416–26. H3K27me3 in human hepatocellular carcinomas correlates closely 20. Murata M, Tagawa M, Kimura M, Kimura H, Watanabe S, Saisho H. with vascular invasion and predicts worse prognosis in patients. Mol Analysis of a germ line polymorphism of the p53 gene in lung cancer Med 2011;17:12–20. patients; discrete results with smoking history. Carcinogenesis 1996; 9. Yang F, Zhang L, Huo XS, Yuan JH, Xu D, Yuan SX, et al. Long 17:261–4. noncoding RNA high expression in hepatocellular carcinoma facil- 21. Gao SB, Xu B, Ding LH, Zheng Q, Zhang L, Zheng QF, et al. The itates tumor growth through enhancer of zeste homolog 2 in humans. functional and mechanistic relatedness of the EZH2 and menin in Hepatology 2011;54:1679–89. hepatocellular carcinoma. J Hepatol 2014 May 15. [Epub ahead of 10. Zheng F, Liao YJ, Cai MY, Liu YH, Liu TH, Chen SP, et al. The putative print]. tumour suppressor microRNA-124 modulates hepatocellular carcino- 22. Dill MT, Tornillo L, Fritzius T, Terracciano L, Semela D, Bettler B, et al. ma cell aggressiveness by repressing ROCK2 and EZH2. Gut 2012; Constitutive Notch2 signaling induces hepatic tumors in mice. Hepa- 61:278–89. tology 2013;57:1607–19. 11. Au SL, Wong CC, Lee JM, Fan DN, Tsang FH, Ng IO, et al. Enhancer of 23. Ladu S, Calvisi DF, Conner EA, Farina M, Factor VM, Thorgeirsson SS. zeste homolog 2 epigenetically silences multiple tumor suppressor E2F1 inhibits c-Myc-driven apoptosis via PIK3CA/Akt/mTOR and microRNAs to promote liver cancer metastasis. Hepatology 2012; COX-2 in a mouse model of human liver cancer. Gastroenterology 56:622–31. 2008;135:1322–32. 12. Fujimoto A, Totoki Y, Abe T, Boroevich KA, Hosoda F, Nguyen HH, 24. Bracken AP, Kleine-Kohlbrecher D, Dietrich N, Pasini D, Gargiulo G, et al. Whole-genome sequencing of liver cancers identifies etiological Beekman C, et al. The polycomb group proteins bind throughout the influences on mutation patterns and recurrent mutations in chromatin INK4A-ARF locus and are disassociated in senescent cells. Genes Dev regulators. Nat Genet 2012;44:760–4. 2007;21:525–30. 13. Yamada A, Fujii S, Daiko H, Nishimura M, Chiba T, Ochiai A. Aberrant 25. Yao JY, Zhang L, Zhang X, He ZY, Ma Y, Hui LJ, et al. H3K27 expression of EZH2 is associated with a poor outcome and P53 trimethylation is an early epigenetic event of p16INK4a silencing for alteration in squamous cell carcinoma of the esophagus. Int J Oncol regaining tumorigenesis in fusion reprogrammed hepatoma cells. 2011;38:345–53. J Biol Chem 2010;285:18828–37.

1396 Mol Cancer Res; 12(10) October 2014 Molecular Cancer Research

Downstream Regulation by PcG in HCC

26. Bracken AP, Pasini D, Capra M, Prosperini E, Colli E, Helin K. EZH2 is and cancers of the endometrium, prostate, and breast. J Clin Oncol downstream of the pRB-E2F pathway, essential for proliferation and 2006;24:268–73. ampliﬁed in cancer. EMBO J 2003;22:5323–35. 33. Kleer CG, Cao Q, Varambally S, Shen R, Ota I, Tomlins SA, et al. EZH2 27. Tavares L, Dimitrova E, Oxley D, Webster J, Poot R, Demmers J, et al. is a marker of aggressive breast cancer and promotes neoplastic RYBP-PRC1 complexes mediate H2A ubiquitylation at polycomb transformation of breast epithelial cells. Proc Natl Acad Sci U S A target sites independently of PRC2 and H3K27me3. Cell 2012;148: 2003;100:11606–11. 664–78. 34. Varambally S, Dhanasekaran SM, Zhou M, Barrette TR, Kumar-Sinha 28. Zhang Y, Xiong Y, Yarbrough WG. ARF promotes MDM2 degra- C, Sanda MG, et al. The polycomb group protein EZH2 is involved in dation and stabilizes p53: ARF-INK4a locus deletion impairs progression of prostate cancer. Nature 2002;419:624–9. both the Rb and p53 tumor suppression pathways. Cell 1998;92: 35. Cao Q, Yu J, Dhanasekaran SM, Kim JH, Mani RS, Tomlins SA, et al. 725–34. Repression of E-cadherin by the polycomb group protein EZH2 in 29. Sherr CJ, McCormick F. The RB and p53 pathways in cancer. Cancer cancer. Oncogene 2008;27:7274–84. Cell 2002;2:103–12. 36. Fujii S, Ito K, Ito Y, Ochiai A. Enhancer of zeste homologue 2 (EZH2) 30. Soto-Reyes E, Recillas-Targa F. Epigenetic regulation of the human down-regulates RUNX3 by increasing histone H3 methylation. J Biol p53 gene promoter by the CTCF transcription factor in transformed cell Chem 2008;283:17324–32. lines. Oncogene 2010;29:2217–27. 37. Normand G, Hemmati PG, Verdoodt B, von Haefen C, Wendt J, Guner 31. Lu J, He ML, Wang L, Chen Y, Liu X, Dong Q, et al. MiR-26a inhibits cell D, et al. p14ARF induces G2 cell cycle arrest in p53- and p21-deﬁcient growth and tumorigenesis of nasopharyngeal carcinoma through cells by down-regulating p34cdc2 kinase activity. J Biol Chem repression of EZH2. Cancer Res 2011;71:225–33. 2005;280:7118–30. 32. Bachmann IM, Halvorsen OJ, Collett K, Stefansson IM, Straume O, 38. McCabe MT, Ott HM, Ganji G, Korenchuk S, Thompson C, Van Aller Haukaas SA, et al. EZH2 expression is associated with high prolifer- GS, et al. EZH2 inhibition as a therapeutic strategy for lymphoma with ation rate and aggressive tumor subgroups in cutaneous melanoma EZH2-activating mutations. Nature 2012;492:108–12.

www.aacrjournals.org Mol Cancer Res; 12(10) October 2014 1397

EZH2 Represses Target Genes through H3K27-Dependent and H3K27-Independent Mechanisms in Hepatocellular Carcinoma

Shu-Bin Gao, Qi-Fan Zheng, Bin Xu, et al.

Mol Cancer Res 2014;12:1388-1397. Published OnlineFirst June 10, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-14-0034

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2014/06/11/1541-7786.MCR-14-0034.DC1

Visual A diagrammatic summary of the major findings and biological implications: Overview http://mcr.aacrjournals.org/content/12/10/1388/F1.large.jpg

Cited articles This article cites 37 articles, 14 of which you can access for free at: http://mcr.aacrjournals.org/content/12/10/1388.full#ref-list-1

Citing articles This article has been cited by 7 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/12/10/1388.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/12/10/1388. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.