Genome-Wide Profiling of Histone H3 Lysine 4 and Lysine 27 Trimethylation Reveals an Epigenetic Signature in Prostate Carcinogenesis Xi-Song Ke1*, Yi Qu1, Kari Rostad1, Wen-Cheng Li1,2, Biaoyang Lin3,4, Ole Johan Halvorsen5, Svein A. Haukaas6,7, Inge Jonassen8,9, Kjell Petersen8,9, Naomi Goldfinger10, Varda Rotter10, Lars A. Akslen1,5, Anne M. Oyan1,11, Karl-Henning Kalland1,11* 1 The Gade Institute, University of Bergen, Bergen, Norway, 2 Department of Urology, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, People���s Republic of China, 3 Swedish Medical Center, Seattle, Washington, United States of America, 4 Zhejiang-California International Nanosystems Institute (ZCNI), Zhejiang University, Hangzhou, People���s Republic of China, 5 Department of Pathology, Haukeland University Hospital, Bergen, Norway, 6 Department of Surgical Sciences, University of Bergen, Bergen, Norway, 7 Department of Surgery, Haukeland University Hospital, Bergen, Norway, 8 Computational Biology Unit, Bergen Center for Computational Science, University of Bergen, Bergen, Norway, 9 Department of Informatics, University of Bergen, Bergen, Norway, 10 Department of Molecular Cell Biology, Weizmann Institute of Science, Rehovot, Israel, 11 Department of Microbiology and Immunology, Haukeland University Hospital, Bergen, Norway Abstract Background: Increasing evidence implicates the critical roles of epigenetic regulation in cancer. Very recent reports indicate that global gene silencing in cancer is associated with specific epigenetic modifications. However, the relationship between epigenetic switches and more dynamic patterns of gene activation and repression has remained largely unknown. Methodology/Principal Findings: Genome-wide profiling of the trimethylation of histone H3 lysine 4 (H3K4me3) and lysine 27 (H3K27me3) was performed using chromatin immunoprecipitation coupled with whole genome promoter microarray (ChIP-chip) techniques. Comparison of the ChIP-chip data and microarray gene expression data revealed that loss and/or gain of H3K4me3 and/or H3K27me3 were strongly associated with differential gene expression, including microRNA expression, between prostate cancer and primary cells. The most common switches were gain or loss of H3K27me3 coupled with low effect on gene expression. The least prevalent switches were between H3K4me3 and H3K27me3 coupled with much higher fractions of activated and silenced genes. Promoter patterns of H3K4me3 and H3K27me3 corresponded strongly with coordinated expression changes of regulatory gene modules, such as HOX and microRNA genes, and structural gene modules, such as desmosome and gap junction genes. A number of epigenetically switched oncogenes and tumor suppressor genes were found overexpressed and underexpressed accordingly in prostate cancer cells. Conclusions/Significance: This work offers a dynamic picture of epigenetic switches in carcinogenesis and contributes to an overall understanding of coordinated regulation of gene expression in cancer. Our data indicate an H3K4me3/H3K27me3 epigenetic signature of prostate carcinogenesis. Citation: Ke X-S, Qu Y, Rostad K, Li W-C, Lin B, et al. (2009) Genome-Wide Profiling of Histone H3 Lysine 4 and Lysine 27 Trimethylation Reveals an Epigenetic Signature in Prostate Carcinogenesis. PLoS ONE 4(3): e4687. doi:10.1371/journal.pone.0004687 Editor: Mikhail V. Blagosklonny, Ordway Research Institute, United States of America Received December 18, 2008 Accepted December 25, 2008 Published March 5, 2009 Copyright: �� 2009 Ke et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This work was supported by Helse Vest (grants no. 911005, 911227 and 911401), the Research Council of Norway (grants. no. 154942/310, 163920/V50 and 185676/V40) and The Norwegian Cancer Society (grants no. HS01-2006-0446, HS01-2006-0468, HS02-2008-0188), the European Commission 6th Framework Program Contract 504743 and the UroBergen Research Fund. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: Xisong.Ke@gades.uib.no (X-SK) Kalland@gades.uib.no (K-HK) Introduction Epigenetics refers to heritable, but potentially reversible, alternated phenotypic states without difference in genotype. The proteins that mediate epigenetic changes are involved in dynamic transcriptional control of gene expression and are encoded by more than 100 genes including DNA methyltransferases (DNMTs), histone acetyltransferases (HATs), histone deacetylases (HDACs), histone methyltransferases (HMTs), histone demethylases (HDMTs) and chromatin remodelling enzymes [1]. EZH2 (Enhancer of Zeste homolog 2), a known core component of the polycomb repressive complex 2 (PRC2), is one of the best characterized HMTs, and can trimethylate the histone H3 lysine 27 (H3K27) and thereby repress gene transcription [2]. Previous work has shown that EZH2 was significantly upregulated and associated with high proliferation rate and aggressive tumor subgroups in prostate cancer [3]. However, the targets and output of epigenetic regulation in prostate cancer are still not completely understood. Genome-wide profiling of the H3K27me3 modification in prostate cancer has been carried out by a few groups. Yu et al. have analyzed H3K27me3 location and suggested a polycomb PLoS ONE | www.plosone.org 1 March 2009 | Volume 4 | Issue 3 | e4687

repression signature in metastatic prostate cancer [4]. It would be informative if H3K27me3 locations were also mapped in benign tissues to show the cancer specificity of the signature. Very recently, H3K27me3 modifications were mapped in both prostate cancer and normal cell lines and a set of genes silenced by EZH2- mediated H3K27 trimethylation specifically in prostate cancer was identified [5]. Both works shed light on the silencing function of EZH2 in prostate cancer, but little is known about epigenetic gene activation in prostate carcinogenesis. To systematically examine the role of epigenetic regulation in prostate cancer, we have screened dysregulated genes in prostate cancer tissues and cell lines using microarray techniques. We found that the most significantly changed epigenetic regulators in both prostate cancer tissues and cell lines were EZH2, SMYD3 and DNMT3A, which function as H3K27 trimethyltransferase, H3K4 di/tri-methyltransferase [6] and DNA methyltransferase [7], respectively. Trimethylation of H3K27 (H3K27me3) and tri- methylation of H3K4 (H3K4me3) are associated with repression and activation of gene transcription, respectively [2,6]. To examine the hypothesis that dysregulated genes in prostate cancer contain a distinct pattern of H3K4me3 and H3K27me3, ChIP- chip analysis was performed for genome-wide profiling of H3K4me3 and H3K27me3 modification patterns in both prostate primary cells and cancer cells. Comparison of the epigenetic switches and gene expression switches between normal primary and cancer cells indicated an H3K4me3/H3K27me3 epigenetic signature in prostate carcinogenesis. Results Dysregulated Epigenetic Genes in Prostate Cancer To analyze the dysregulated epigenetic genes in prostate cancer, we screened the microarray gene expression data of prostate cancer and benign tissues as previously published [8,9] and created a subset of the summarized epigenetic genes, and ranked them according to fold change between prostate benign and cancer tissues. The top three (fold change.1.5, p-value,3E-10) changed epigenetic enzymes were EZH2, SMYD3 and DNMT3A (Figure 1A, Table S1 in the Supplemental Data available online). All of them were overexpressed in prostate cancer. EZH2 is a known H3K27 trimethyltransferase, catalyzing H3K27me3, a transcriptional repressive mark [2]. SMYD3 is an HMTase specific for both di- and trimethylation of H3K4, catalyzing H3K4me2/me3, a transcriptional activating mark [6]. DNMT3A is a de novo DNA methyltransferase repressing gene transcription [7]. To our knowledge, this is the first report of overexpression of both SMYD3 and DNMT3A in prostate cancer tissues. Very interestingly, five of the top eight changed epigenetic genes were involved in histone H3 lysine methylation (Figure 1B). We also did genome-wide profiling of the expression patterns of prostate cell lines and screened the changed epigenetic genes between them using the same strategy as in prostate tissues. PC3 cells and EP156T cells were used as prostate cancer and primary cells, respectively. The PC3 cell line was initiated from a bone metastasis of a prostatic adenocarcinoma. The EP156T cell line was established from benign prostate tissue and was characterized as an immortalized primary prostate epithelial cell line [10]. As shown in Figure 1C, the top eight differentially expressed epigenetic genes in prostate cancer have the same expression patterns as in PC3 cells compared with EP156T cells. The high consistency of the most significantly changed epigenetic genes between prostate tissues and cell lines suggested that main epigenetic regulations in prostate carcinogenesis could be mediated by histone H3 lysine methylations. Genome-wide Analysis of H3K4me3 and H3K27me3 Modifications in Prostate Cells It was very interesting to identify EZH2 and SMYD3 as the most significantly changed and overexpressed histone modifiers in prostate cancer, since the two modifications are antagonistic and Figure 1. Microarray-based Screening of Changed Epigenetic Genes in Prostate Cancer. A. The top significantly changed epigenetic genes in prostate cancer were overexpressed EZH2, SMYD3 and DNMT3A. B. The most significantly changed epigenetic genes in prostate cancer were involved in histone 3 lysine methylation. C. Changed epigenetic genes in prostate cancer tissues have the same changed patterns as the prostate cancer cell line PC3 compared with primary prostate EP156T cells. Asterisks indicate gene expressions with fold change values.1.5 and p-values,3E- 10 in prostate cancer tissues compared with benign tissues. doi:10.1371/journal.pone.0004687.g001 Prostate Carcinogenesis PLoS ONE | www.plosone.org 2 March 2009 | Volume 4 | Issue 3 | e4687