Oncogene (2014) 33, 5546–5558 & 2014 Macmillan Publishers Limited All rights reserved 0950-9232/14 www.nature.com/onc

ORIGINAL ARTICLE Nuclear loss of arginine N-methyltransferase 2 in breast carcinoma is associated with tumor grade and overexpression of protein

J Zhong1,3, R-X Cao1,2,3, J-H Liu1,2, Y-B Liu1, J Wang1, L-P Liu1, Y-J Chen1, J Yang1,2, Q-H Zhang1,YWu1, W-J Ding1, T Hong1,2, X-H Xiao1,2, X-Y Zu1 and G-B Wen1

Human protein arginine N-methyltransferase 2 (PRMT2, HRMT1L1) is a protein that belongs to the arginine methyltransferase family, and it has diverse roles in transcriptional regulation through different mechanisms depending on its binding partners. In this study, we provide evidences for the negative effect of PRMT2 on cell proliferation in vitro and in vivo. Morever, cyclin D1, one of the key modulators of cell cycle, was found to be downregulated by PRMT2, and PRMT2 was further shown to suppress the estrogen receptor a-binding affinity to the activator protein-1 (AP-1) site in cyclin D1 promoter through indirect binding with AP-1 site, resulting in the inhibition of cyclin D1 promoter activity in MCF-7 cells. Furthermore, a positive correlation between the expression of PRMT2 and cyclin D1 was confirmed in the breast cancer tissues by using tissue microarray assay. In addition, PRMT2 was found to show a high absent percentage in breast caner cell nuclei and the nuclear loss ratio of PRMT2 was demonstrated to positively correlate with cyclin D1 expression and the increasing tumor grade of invasive ductal carcinoma. Those results offer an essential insight into the effect of PRMT2 on breast carcinogenesis, and PRMT2 nuclear loss might be an important biological marker for the diagnosis of breast cancer.

Oncogene (2014) 33, 5546–5558; doi:10.1038/onc.2013.500; published online 2 December 2013 Keywords: nuclear loss; PRMT2; cyclin D1; breast cancer

INTRODUCTION fetal brain, and to enhance androgen receptor-mediated Estrogen receptor a (ERa) is a transcription factor and engaged in a transactivation dependent on the cellular background.16 Recently, broad range of biological processes, including cell proliferation, PRMT2 has been demonstrated to have weak methyltransferase differentiation, morphogenesis and apoptosis,1–3 as well as the activity on a histone H4 substrate, but its optimal substrates have not development and progression of breast cancer.4 ERa-mediated yet been identified.17,18 A series of reports show that PRMT2 is clearly transcription is through binding directly to specific estrogen involved in a variety of cellular processes, including lung function, response elements (EREs) in the promoters of responsive genes5 or the inflammatory response, apoptosis promotion, Wnt signaling and via protein–protein interaction with other promoter-bound tran- leptin signaling regulation,19–23 suggesting that PRMT2 has diverse 6 7 scription factors, such as specificity protein 1 (Sp1), AP-1 or roles in transcriptional regulation through different mechanisms 8 nuclear factor-kB. In either case, coactivators or are depending on its binding partners. further recruited to form a functional receptor complex that In previous study, we identified that PRMT2 is capable of 9 specifies transcriptional activities of downstream targets. Cyclin binding to ERa both in vitro and in vivo, and we showed that D1 (CCND1) is a D-type cyclin that regulates G1-S cell cycle PRMT2 expression was higher in human breast tumors when progression during cell proliferation. CCND1 is also E2/ERa compared with adjacent normal tissue, suggesting a role for responsive and is thought to have major roles in breast PRMT2 in breast tumorigenesis.24 To further understand the 10,11 cancer. However, there is no classical ERE in the CCND1 physiological function of PRMT2, as well as its potential role in the promoter, and the contribution of E2/ERa to CCND1 action in development and progression of breast cancer, MCF7 breast 12 breast cells remains unclear. The roles of CCND1 transcription cancer cells with high level of PRMT2 expression was used for a regulation in breast tumorigenesis need to be further defined. loss-of-function cell model by artificial microRNA (miRNA) based Human protein arginine N-methyltransferase 2 (PRMT2, HRMT1L1) on the murine miR-155 sequence.25 This study provides evidences is a protein that belongs to the arginine methyltransferase family and for the negative effect of PRMT2 on breast cancer cell contains a highly conserved catalytic Ado-Met-binding domain and proliferation, uncovers the molecular mechanism of PRMT2 unique Src homology 3 domain that binds with proline-rich regulating the expression of CCND1, which accounts for PRMT2- motifs.13,14 Initially, PRMT2 enhances ERa-mediated transactivation in induced proliferation suppression in breast cancer cells, and CV-1 cells,15 and subsequent research showed that PRMT2 was found demonstrates the correlation between PRMT2 nuclear loss and to be aberrantly expressed in skeletal muscle, ovary, prostate and grade of invasive ductal breast carcinoma.

1Institute of Clinical Medicine, First Affiliated Hospital of University of South China, Hengyang, PR China and 2Department of Metabolism and Endocrinology, First Affiliated Hospital of University of South China, Hengyang, PR China. Correspondence: Dr X-Y Zu or G-B Wen, Institute of Clinical Medicine, First Affiliated Hospital of University of South China, 69 chuanshan Road, Hengyang 421001, China. E-mail: [email protected] or [email protected] 3The first two authors contributed equally to this work. Received 30 April 2013; revised 17 September 2013; accepted 18 October 2013; published online 2 December 2013 PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5547 RESULTS to evaluate knockdown of PRMT2 mRNA, real-time PCR was miRNA induced downregulation of PRMT2 expression in breast performed, and as shown in Figure 1d, PRMT2 mRNA levels in cancer cells MCF7 cells with PRMT2-miRNAs were reduced by almost 80% and Initially, experiments were performed to select the best miRNA to 70% respectively, as compared with those in control cells. knockdown PRMT2 expression. A pcDNA6.2-GW/EmGFP-miR- based miRNA expression plasmid with a pre-miRNA sequence was constructed and transfected to MCF7 cells. Two kinds of Suppression of PRMT2 expression promotes the cell proliferation miRNA were designed to test their inhibitory effect on PRMT2 and colony formation of MCF7 cells expression in MCF7 cells. PRMT2-miRNA1 was targeted to 4 We next examined the effect of miRNA-induced PRMT2 down- of PRMT2, and PRMT2-miRNA2 was targeted to exon 9 of PRMT2 regulation on the proliferation ability of MCF7 cells. Without (Figure 1a). Stable cell lines with the PRMT2-miRNAs were treatment of 17b-Estradiol (E2), MCF7 cells carrying PRMT2- obtained after the transfected MCF7 cells were selected by miRNA1, PRMT2-miRNA2 or LacZ-miRNA showed no significant blasticidin for 2 weeks (Supplementary Figure S1). Immuno- difference in their growth rates at the indicated time points fluorescence staining and western blotting were performed to (Figure 2a). Whereas, MCF7 cells with either PRMT2-miRNA1 or verify the expression of PRMT2 in the transfected MCF7 cells. The PRMT2-miRNA2 showed markedly increased proliferate ability cells with the pre-miRNA sequence against PRMT2 showed compared with that of cells with LacZ-miRNA, when treated with significantly decreased expression of PRMT2 compared with 10 nM E2 for 5 days (Figure 2b). We speculated that suppression of control cells with LacZ-miRNA (Figures 1b and c). Furthermore, endogenous PRMT2 expression could increase responsiveness of

Antisense target sequence miRNA loop Sense target sequence Overhang (Mature miRNA Sequence) derived from miR-155 (nucleotides 1-8 and 11-21

PRMT2-miR1 5’-TGCTGTGGAAGTGGACGCTAAACCAGGTTTTGGCCACTGACTGACCTGGTTTAGTCCACTTCCA-3’ 3’- CACCTTCACCTGCGATTTGGTCCAAAACCGGTGACTGACTGGACCAAATCAGGTGAAGGTGTCC-5’

PRMT2-miR2 5’-TGCTGTCTTCATCCTGCCACGTGTCCGTTTTGGCCACTGACTGACGGACACGTCAGGATGAAGA-3’ 3’-CAGAAGTAGGACGGTGCACAGGCAAAACCGGTGACTGACTGCCTGTGCAGTCCTACTTCTGTCC-5’

GFP PRMT2 DAPI Merge

LacZ-miRNA

PRMT2-miR1

PRMT2-miR2

1000 800 600

LacZ-miRNAPRMT2-miR1PRMT2-miR2 400 PRMT2 200 0 -actin to PRMT2-miRNA cells PRMT2 expressive folds of LacZ-miRNA cells compared

LacZ-miRNAPRMT2-miR1PRMT2-miR2 Figure 1. Downregulation of PRMT2 in MCF7 cells with PRMT2-miRNA. (a) Pre-miRNA double-strand oligo inserted into miRNA expression vector-pcDNA 6.2-GW/EmGFP-miR. (b) MCF7 cells with either LacZ-miRNA or PRMT2-miRNA were fixed in paraformaldehyde, permeabilized with triton X-100, and were incubated with cy3-conjugated PRMT2 antibody. The cells were then stained with 4,6-diamidino-2-phenylindole (DAPI) and viewed under a Zeiss LSM 510 confocal microscope. (c) Western blot analysis of MCF7 cells transfected with PRMT2-miRNA compared with MCF7 cells with LacZ-miRNA. (d) Summary of real-time reverse transcription–PCR results of MCF7 cells transfected with PRMT2-miRNA compared with LacZ-miRNA cells. GFP, green fluorescent protein.

& 2014 Macmillan Publishers Limited Oncogene (2014) 5546 – 5558 PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5548

) 3.0 ) LacZ-miRNA 4 LacZ-miRNA 4 8 * * PRMT2-miR1 PRMT2-miR1 PRMT2-miR2 7 PRMT2-miR2 * 2.5 6 5 4 2.0 3 2 Cell numbers (x 10 Cell numbers 1.5 (x 10 Cell numbers 1 1234567 1 234567 (days) (days)

0.85 LacZ-miRNA 1.0 LacZ-miRNA PRMT2-miR1 PRMT2-miR1 0.80 PRMT2-miR2 PRMT2-miR2 * * 0.8 * 0.75 0.6 OD/well OD/well 0.70 0.4 0.65 -11 -10 -9 -8 -7 -6 1234567 E2 (M/L) 010 10 10 10 10 10 (days)

LacZ-miRNA PRMT2-miR1 PRMT2-miR2 - E2 + E 160 * 2 140 * 120 -E2 100 80 60 40

Colony numbers Colony 20

+E2 0

LacZ-miRNA PRMT2-miR1PRMT2-miR2 Figure 2. Suppression of PRMT2 expression promotes the cell proliferation and colony formation of MCF7 cells. (a and b) Logarithmic population of the MCF7 cells expressing LacZ-miRNA, PRMT2-miR1 or PRMT2-miR2 without (a) or with 10 nM E2 (b). MCF7 cells with different PRMT2 levels were cultured for 3 days in phenol red-free Dulbecco’s modified Eagle’s medium supplemented with steroid-stripped fetal bovine serum (5%) and then treated with E2 at different concentrations (c) or 100 nM 4-hydroxytamoxifen (d) for 24 h. Cell proliferation rate was quantified using CellTiter 96 AQueous assay. Each data point represents the mean±s.e.m. number of cells counted in triplicate dishes (*Po0.05; statistical analyses with LacZ-miRNA cells). This experiment was repeated thrice. (e) MCF7 cells stably expressing corresponding plasmids were maintained in steroid-deprived medium (under the presence of 10 mg/ml blasticidin), and treated with 10 nM E2 or ethanol vehicle for 14 days. Subsequently stained with crystal violet. The number of colonies in each condition was counted and expressed as mean±s.d. from triplicate experiments; bars, s.d. *Po0.05.

MCF7 cells to estrogen. To test this hypothesis, the stable MCF7 cell we transplanted three types of breast tumors cells developed from lines with different PRMT2 expression levels were used to MCF7 cells (PRMT2-miRNA1, PRMT2-miRNA2 and LacZ-miRNA) determine whether the suppression of PRMT2 has any effect on into female nude mice. Growth of the implanted tumors was the cell sensitivity to E2. As shown in Figure 2c, E2 could stimulate measured in mice (n ¼ 5 for each group) over a period of 8 weeks. proliferation of MCF7 cells carrying either PRMT2-miRNAs or LacZ- Tumors collected from mice engrafted with MCF7-PRMT2-miRNA miRNA in a dose-dependent manner, but the MCF7 with PRMT2- cells showed a reduction of PRMT2 expression compared to that miRNAs present a more sensitive response to E2 compared that with MCF7-LacZ-miRNA both at mRNA and protein levels with LacZ-miRNA. Meanwhile, the treatment of 4-hydroxytamoxifen (Figure 3a). Knocking-down PRMT2 dramatically increased the could reverse the stimulate effect of E2 on growth rates for MCF7 size of tumors either engrafted with MCF7-PRMT2-miRNA1 cells or cells both with PRMT2-miRNAs and with LacZ-miRNA (Figure 2d). with MCF7-PRMT2-miRNA2 cells (Po0.05, Figure 3b), but the Colony formation assay showed that the number of formed difference between the MCF7-PRMT2-miRNA1 cells and MCF7- colonies from PRMT2-miRNA1- or PRMT2-miRNA2-transfected PRMT2-miRNA2 cells was not significant. CCND1, one of the MCF7 cells was markedly increased compared with the cells with targets of ERa was found to be upregulated in both MCF7-PRMT2- LacZ-miRNA when treated with 10 nM E2, whereas, without the miRNA1 cells and MCF7-PRMT2-miRNA2 cells (Figure 3c), indicat- treatment of E2, the number of formed colonies of MCF7 cells ing that CCND1 might be involved in the cell proliferation carrying either PRMT2-miRNA1 or PRMT2-miRNA2 was increased enhancement induced by PRMT2 knockdown. Those data suggest slightly compared with the cells with LacZ-miRNA (Figure 2e). that knocking-down PRMT2 enhances proliferation of breast Those results indicated that the effect of PRMT2 downregulation cancer cells in vivo. on cell proliferation is probably dependent on E2 and E2 could enhance the effect of PRMT2 on colony formation in MCF7 cells. Identification of PRMT2-targeting Knocking-down of PRMT2 increases tumor growth in a MCF7 To better understand the molecular mechanisms involving in the xenograft mouse model effects of PRMT2 on cell proliferation and tumorigenesis, the To further examine the effects of PRMT2 on cell proliferation whole genome-wide transcriptome profile of LacZ-miRNA and in vivo and to investigate its possible role in breast carcinogenesis, PRMT2-miRNA1 cells were analyzed by Agilent oligo microarray

Oncogene (2014) 5546 – 5558 & 2014 Macmillan Publishers Limited PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5549

8 P =0.035 P =0.012 6 LacZ-miRNAPRMT2-miR1PRMT2-miR2 PRMT2 4 CCND1 2 PRMT2 mRNA ERα (% of GAPDH mRNA) 0 LacZ- PRMT2- PRMT2- β-actin miRNA miR1 miR2

LacZ-miRNA PRMT2-miR1 PRMT2-miR2

1200 LacZ-miRNA * ) PRMT2-miR1 3 LacZ-miRNA 1000 PRMT2-miR2 * 800 * PRMT2-miR1 600

400 PRMT2-miR2 Tumor volume (mm 200 12345678 Weeks

LacZ-miRNA PRMT2-miR1 PRMT2-miR2 Figure 3. Knockingdown PRMT2 increased tumor growth in xenografted mice. (a) PRMT2 mRNA and protein levels were decreased in tumors from knocking-down PRMT2 mice. PRMT2 mRNA in tumors was quantified using real-time PCR; PRMT2 protein was detected by western blot analysis and immunohistochemical assay. (b) Increased tumor volume in knocking-down PRMT2 mice. MCF7 cells with LacZ-miRNA, PRMT2- miRNA1 or PRMT2-miRNA2 were transplanted into ovariectomized athymic mice. Left, tumor appearances in different treated groups. Right, tumors were measured weekly using a vernier calliper and the volume was calculated according to the formula: p/6  length  width2. Each point represents the mean±s.d. for different animal measurements (n ¼ 5; *Po0.05). (c) Detection of CCND1 protein expression by immunohistochemical assay of tumor tissues in nude mice in different treated groups (streptavidin biotin complex,  400).

(41 000 þ ). According to fold-change (X2.0) screening between proliferation, was found to be one of the downstream targets of LacZ-miRNA and PRMT2-miRNA1 cells, we found that a total of PRMT2. 2173 genes had altered expression, of which 795 and 1378 were upregulated and downregulated, respectively. Most of these differentially expressed genes are involved in transcription PRMT2 suppresses CCND1 expression in breast cancer cells regulation, cell adhesion, immune responses, cell growth, Previous study showed that PRMT2 function as co-regulator of ERa proliferation and carcinogenesis. Forty-seven cancer-associated in the regulation of ERa-targeting expression,15 so the role of genes and forty-six proliferation-related genes were selected E2/ERa signaling in PRMT2-induced gene alterations in breast for cluster mapping on the MeV microarray analysis platform cancer cells is of worth to be investigated. To address the effect of (Figures 4a and b). PRMT2 on estrogen-induced ERa target genes, CCND1 and c-myc To validate the microarray data, nine upregulated genes and were chosen for estrogen response detection. MCF7 cells were nine downregulated genes (Table 1) were chosen for real-time transfected with pcDNA3.1-PRMT2 or empty vector and treated PCR confirmation. Primers for selected genes analyzed by real- with 10 nM E2 or ethanol vehicle for 4 h. As shown in Figures 5a time PCR are showed in Supplementary Table S1, and as shown in and b, an average twofold increase of E2-induced CCND1 mRNA Figures 4c and d, five of nine upregulated genes and eight of nine and protein was observed in empty vector-transfected cells and downregulated genes showed consistent alteration as microarray increased PRMT2 expression obviously diminished E2-induced analysis. Interestingly, CCND1, a key modulator in cell cycle and CCND1 mRNA and protein expression. We also found that

& 2014 Macmillan Publishers Limited Oncogene (2014) 5546 – 5558 PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5550 cancer-associated genes proliferation-related genes

–3.0 0.0 3.0 –3.0 0.0 3.0 1 2 3 1 2 3

BMP2 EPCAM WNT10B CCNG2 PIAS2 DUSP5 PGF ICAM5 EGLN3 MAPT BIRC7 CCNG2 GAB1 ELK1 CEBPA CDK8 ELK1 PRKACB APPL1 CCNF STK4 STK4 RALBP1 MAPK12 SMAD2 CCND1 RAC2 RAC2 PTEN MAP2K1 LAMA4 PRKX CASP9 MAP2K5 CDH1 FGFR4 RPS6KA5 NF1 FZD3 RPS6KA5 MAP2K1 CCNL2 EGLN1 PPP5C PTEN DUSP2 VEGFA DUSP16 CCND1 VEGFA TNFSF8 MYC VCAN EFEMP1 MYADM PDCD1 CXCL1 RERG VEGFB ICAM2 PDCD1 ABI3 RERG ITGAM S100A16 MAG DIRAS3 BCAS3 CTGF ITGB2 MDK ITGB8 PPDPF CD2 HDGFL1 LCN2 GHRH HLA–B S100A13 HLA–C GHSR CNTNAP1 IGFBP2 CLDN1 GAS7 MADCAM1 FGF6 ICOSLG FGF13 CLDN6 FGF1 CD8B S100A8

60 LacZ-miRNA 30 LacZ-miRNA

) PRMT2-miRNA ) PRMT2-miRNA -4 -4 40 20

*** 20 10 * * ** 8 4 ** 6 ** 3 Relative mRNA Relative mRNA *** ** 4 * 2 *** expression of GAPDH (10 expression of GAPDH (10 2 1 ** * *** 0 0

ABI3 Slug CDH1 RAC2 BMP2 c-myc RERG VCAN LCN2 CCND1 DUSP1 C-FOS PDCD1 ICAM2 BCAS3 SMAD2 MAP2K1 MADCAM1 Figure 4. Target gene identification by cDNA microarray and real-time PCR analysis. (a and b) Clustering map of differentially expressed genes overlapped with either cancer-associated genes or proliferation-related genes set in the Molecular Signatures Database. Row represents gene, column represents the ratio of experimental cells (PRMT2-miR1 vs LacZ-miRNA) from triplicate experiments. Upregulated genes are shown in red and downregulated genes in green. (c and d) Verification of PRMT2 target genes by real-time PCR. This experiment was repeated thrice. Each data point represents the mean±s.e.m (*Po0.05, **Po0.01, ***Po0.001; statistical analyses with LacZ-miRNA group). GAPDH, glyceraldehyde 3-phosphate dehydrogenase.

overexpression of PRMT2 enhanced E2-induced increases in c-Myc PRMT2 downregulates CCND1 expression via the suppression of expression levels (Figures 5a and b). These data also suggest that Akt/GSK-3b signaling in breast cancer cells increased PRMT2 expression may have various effects on ERa It has been reported that CCND1 expression is mediated by Akt/ target genes. GSK-3b pathway,26–29 and we therefore exploited whether PRMT2 Having observed that increased PRMT2 expression decreased was involved in Akt/GSK-3b/CCND1 axis in MCF7 cells. First, we E2-induced CCND1 expression, it could be reasoned that PRMT2 examined the involvement of Akt/GSK-3b signaling pathway in silencing would result in an increase in the E2 induction of CCND1. E2-induced CCND1 expression by using both LacZ-miRNA and As shown in Figure 5c, PRMT2 knockdown leaded PRMT2-miRNA MCF-7 cells. As shown in Figure 5d, pre-treatment to a statistically significant (Po0.05) enhancement of E2-induced either with wortmannin or LY294002 abrogated E2-induced CCND1 protein expression levels, consistent with the results from phosphorylation of Akt and GSK-3b in MCF7 cells, resulting in the microarray and real-time PCR analysis. Whereas the knock- obvious downregulation of CCND1 in both PRMT2-miRNA and down of PRMT2 caused to a decrease of E2-induced c-myc LacZ-miRNA cells. In addition, LiCl and SB216763, two inhibitors of expression (Figure 5c), and the suppression of PRMT2 showed no GSK-3b, were applied in both LacZ-miRNA and PRMT2-miRNA obvious effect on the ERa expression. cells. As shown in Figure 5e, pre-treatment either with LiCl or

Oncogene (2014) 5546 – 5558 & 2014 Macmillan Publishers Limited PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5551 Table 1. Verified genes and their description by real-time PCR

Symbol Accession number Gene description

Upregulated DUSP1 NM_004417 Homo sapiens dual specificity phosphatase 1 MAP2K1 NM_002755 Homo sapiens mitogen-activated protein kinase kinase 1 RAC2 NM_002872 Homo sapiens ras-related C3 botulinum toxin substrate 2, rho family, small GTP-binding protein Rac2 CCND1 NM_053056 Homo sapiens cyclin D1 (CCND1) c-myc NM_002467 Homo sapiens v-myc myelocytomatosis viral oncogene homolog (avian) (MYC) c-fos NM_005252 Homo sapiens FBJ murine osteosarcoma viral oncogene homolog (FOS) CDH1 NM_004360 Homo sapiens cadherin 1, type 1, E-cadherin (epithelial) (CDH1) BMP2 NM_001200 Homo sapiens bone morphogenetic protein 2 SMAD2 NM_001003652 Homo sapiens SMAD family member 2 (SMAD2), transcript variant 2

Downregulated PDCD1 NM_005018 Homo sapiens programmed cell death 1 ICAM2 NM_000873 Homo sapiens intercellular adhesion molecule 2 ABI3 NM_016428 Homo sapiens ABI family, member 3 (ABI3), transcript variant 1, RERG NM_032918 Homo sapiens RAS-like, estrogen-regulated, growth inhibitor (RERG), transcript variant 1 MADCAM1 NM_130760 Homo sapiens mucosal vascular addressin cell adhesion molecule 1 (MADCAM1), transcript variant 1 VCAN NM_004385 Homo sapiens versican (VCAN), transcript variant 1 BCAS3 NM_017679 Homo sapiens breast carcinoma amplified sequence 3 (BCAS3), transcript variant 2 Slug NM_003068 Homo sapiens snail homolog 2 (Drosophila) (SNAI2) LCN2 NM_005564 Homo sapiens lipocalin 2 (LCN2)

SB216763 also abrogated E2-induced phosphorylation of GSK-3b, assay from chromatin precipitated by either anti-PRMT2 or ERa leading to obvious downregulation of CCND1 in both PRMT2- antibody (Ab; Figure 6c, upper panel), and the knockdown PRMT2 miRNA and LacZ-miRNA cells. Those results suggested that Akt/ could increase the binding affinity of ERa to CCND1 promoter GSK-3b is implicated in E2-induced CCND1 expression in MCF-7 (Figure 6c, lower panel). We therefore reasoned that PRMT2 could cells. As shown in Figures 5d and e, miRNA-mediated PRMT2 suppress CCND1 through directly or indirectly binding to the AP-1 knockdown markedly increased phosphorylation of Akt and GSK- site of CCND1 promoter. To further validate this proposal, we 3b and enhanced the expression of CCND1 with or without E2 tested the binding affinity of PRMT2 to the AP-1 sequence by treatment, suggesting that the knockdown of PRMT2 could using recombinant human full-length PRMT2 protein, and enhance Akt/GSK-3b/CCND1 signaling regardless of the presence electrophoretic mobility shift assays (EMSAs) showed that of E2. The above results support the notion that PRMT2 could recombinant PRMT2 could not directly bind to the AP-1 probe suppress the CCND1 expression in MCF-7 cells, at least partially via in vitro (Supplementary Figure S2). However, a strong specific the suppression of Akt/GSK-3b signaling. band shift with anti-PRMT2 Ab was observed in the EMSA assay with extracts from the MCF-7 cells and AP-1 probe (Figure 7). The EMSA results demonstrated that PRMT2 bind to the AP-1 site of PRMT2 downregulates CCND1 expression via indirectly binding to CCND1 promoter in an indirect manner in MCF-7 cells. Together, it AP-1 site of the CCND1 promoter in breast cancer cells could be concluded that PRMT2 suppresses the ERa-binding We further explored the mechanism by which PRMT2 regulates affinity to the AP-1 site in CCND1 promoter through indirect the expression of CCND1. The promoter region containing binding with AP-1 site, resulting in the inhibition of CCND1 1980 bp upstream of the transcription start site of CCND1 was promoter activity in MCF-7 cells. cloned and used to drive the expression of luciferase reporter in MCF7 and T47D breast cancer cells. Progressive deletions of the 50 end of the CCND1 promoter revealed that the À 1182-bp PRMT2 correlates with CCND1 expression in breast carcinoma fragment possesses full promoter activity (Figure 6a). Promoter To further identify the relationship between PRMT2 and CCND1 motif analysis identified that an AP-1 site (TGAGTCA) locating at expression in breast cancer, tissue microarray (BR1921a, À 895 to À 901 bp and an ERE-like site (GGTCAAG) locating at US Biomax), consisting of 160 breast cancer cases and 32 À 779 to À 785 bp are in this obtaining promoter sequences, and normal cases, was used. The specificity of PRMT2 Ab had been those sites had been shown to be potential binding sites of ERa. tested with ectopic PRMT2 expression before the immuno- Therefore, this promoter fragment was used to investigate the histochemistry staining, and as showed in Supplementary regulatory role of PRMT2 in its activity. As shown in Figure 6b, Figure S3. The tissue microarray analysis showed that CCND1 cotransfection of pcDNA3.1/PRMT2 greatly decreased the activity was stained in 39.5% of breast tumors, predominantly in nuclear of pluc-1182 in both MCF7 and T47D cells, whereas PRMT2- with faint cytoplasmic staining, and was almost undetectable in miRNA1 significantly increased its basal activity, suggesting matched normal breast tissue. As shown in Table 2, statistical PRMT2 as a negative regulator of CCND1 promoter. Furthermore, analysis confirmed that the PRMT2 and CCND1 expression were replacing TGAGTCA with GAGACAG in the AP-1 site of pluc-1182 positively correlated (r ¼ 0.346, P ¼ 0.000), when analyzed regard- (luciferase plasmid at À 901 bp) abrogated the promoter’s less of breast cancer types. Furthermore, the correlation between response to PRMT2 in both MCF7 and T47D cells, whereas PRMT2 and CCND1 expression was also found in invasion ductal replacing GGTCAAG with AAAAGGA in the ERE-like site of pluc- cancer (r ¼ 0.327, P ¼ 0.003) and invasion lobular cancer (r ¼ 0.309, 1182 (luciferase plasmid at À 785 bp) has no influence on the P ¼ 0.006), respectively. Interestingly, the level of PRMT2 in cell promoter’s response to PRMT2 in both MCF7 and T47D cells nuclear of breast cancer tissue was found to show obvious decrease (Figure 6b). In addition, the interactions between either PRMT2 or compared with those in normal breast tissue, although the total ERa and CCND1 promoter were further confirmed by chromatin amount of PRMT2 presented an increase in breast cancer tissue, immunoprecipitation (ChIP) assay. As shown in Figure 6c, CCND1 indicating that the elevated expression of PRMT2 in breast cancer promoter fragment containing AP-1 site was amplified by PCR tissue might locate mainly in cytoplasm. The above results

& 2014 Macmillan Publishers Limited Oncogene (2014) 5546 – 5558 PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5552 demonstrated that PRMT2 could negatively regulate the expression Loss of nuclear PRMT2 expression correlates with increasing tumor of CCND1 in breast cancer cells, so the loss of nuclear PRMT2 in grade of invasive ductal carcinoma breast cancer tissue could partly explain the enhanced level Because of the suppressive effect of PRMT2 on CCND1 expression, of CCND1. it could be inferred that the loss of nuclear PRMT2 expression in

pcDNA3.1 + - + - PRMT2 - + - + CCND1 * - E2 E2 - - + + 2.5 + E2 CCND1 2.0

1.5 c-myc 1.0 ER 0.5

GAPDH Relative mRNA expression 0.0 Empty vector PRMT2

pcDNA3.1 + - + - PRMT2 - + - + CCND1

E2 - - + + 2.5 * - E2 + E2 PRMT2 2.0

CCND1 1.5

c-myc 1.0

ER 0.5

Relative protein expression 0.0 -actin Empty vector PRMT2

LacZ PRMT2LacZ PRMT2 miRNA miR1 miR2 miRNA miR1 miR2 CCND1

E2 ---+++ * - E2 * + E2 PRMT2 4

CCND1 3

2 c-myc

1 ER

Relative protein expression 0 -actin LacZ- PRMT2- PRMT2- miRNA miR1 miR2

LacZ-miRNA PRMT2-miRNA

E2: - + + + - + + + Wortmannin: - - + - - - + - LY294 002: - - - + - - - + p-AKT

AKT

p-GSK-3

GSK-3

CCND1

-actin 1234 5678

LacZ-miRNA PRMT2-miRNA

E2: - + + + - + + + LiCl: - - + - - - + - SB216 763: - - - + - - - + p-GSK-3

GSK-3

CCND1

-actin

1234 5678

Oncogene (2014) 5546 – 5558 & 2014 Macmillan Publishers Limited PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5553 Schemes of wild type and mutant CCND1 promoters

Luc pGL4.1-basic +1 -747 Luc pLuc-747 +1 -1182 Luc pLuc-1182 +1 -1182 Luc pLuc-901m +1 -1182 Luc pLuc-785m +1 0 1000 2000 3000 4000 5000 6000 7000 8000 Realtive luciferase activity (u)

Effect of PRMT2 on CCND1 promoter activity

MCF7 cell T47D cell 12000 14000 - E2 - E2 + E * + E2 12000 2 10000 * 10000 8000 8000 * 6000 6000 * 4000 4000 2000 2000 0 0

Relative Luciferase Activity (u) Relative Luciferase Activity (u) iR1

pLuc-1182 pLuc-901m pLuc-785m pLuc-1182 pLuc-901m pLuc-785m pGL4.1-basic pGL4.1-basic

pLuc-1182+PRMT2 pLuc-1182+PRMT2

pLuc-1182+PRMT2-miR1pLuc-901m+PRMT2-miR1pLuc-785m+PRMT2-miR1 pLuc-1182+PRMT2-miR1pLuc-901m+PRMT2-miR1pLuc-785m+PRMT2-m

CHIP assay MCF7 cell T47D cell

Input PRMT2 IgG Input PRMT2 IgG

PRMT2-miRNA :- - + +- PRMT2-miRNA : - + +

E2:E- ++- 2: - ++-

ER ER

IgG IgG

Input Input

Figure 6. Recruitment of ERa and PRMT2 to the CCND1 promoter in MCF7 and T47D cells. (a) Schematic representation of CCND1 promoter luciferase reporter plasmids, including wild-type CCND1 promoter and 50 progressive deletions, as well as site-directed mutants at the ERE-like site, pLuc-785m (at À 785 bp) and the AP-1 site pLuc-901m (at À 901 bp). Refer to Materials and methods for the plasmid construction. CCND1 promoter activity was analyzed in MCF7 cells by luciferase assay. (b) Regulation of CCND1 promoter activity by PRMT2 in MCF7 and T47D breast cancer cells. Wild-type or mutant promoter was cotransfected with PRMT2 gene or PRMT2-miRNA, and luciferase activity was measured with pRL-TK as an internal control. Data represent the mean of three independent experiments±s.d. *Po0.05. (c) MCF7 cells carrying either LacZ-miRNA or PRMT2-miRNA were cultured in steroid-deprived medium for at least 3 days, and subsequently treated with 10 nM E2 or ethanol vehicle for 4 h. The precleared chromatin was immunoprecipitated with PRMT2 or ERa antibodies. CCND1 promoter sequence was detected by PCR with specific primers, as detailed in Materials and methods. To determine input DNA, the CCND1 promoter fragment was amplified from 5 ml purified soluble chromatin before immunoprecipitation. PCR products obtained at 35 cycles. ChIP with non-immune IgG was used as negative control.

Figure 5. PRMT2 suppresses CCND1 expression in breast cancer cells. (a) MCF7 cells were first cultured in steroid-deprived medium for at least 3 days, and subsequently MCF7 cells were transfected with pcDNA3.1-PRMT2 or empty vector and treated with 10 nM E2 or ethanol vehicle for 4 h. RNA was extracted and subjected to reverse transcription–PCR (RT-PCR). RT–PCR results quantitated for changes in CCND1 mRNA using glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as a control. Columns, mean of three independent replicates; bars, s.e.m, *Po0.05. (b) Representative western blot analysis using indicated antibodies for immunoblotting. Changes in CCND1 protein levels were quantitated and normalized to b-actin. Columns, mean of three independent replicates; bars, s.e.m., *Po0.05. (c) MCF7 cells carrying either PRMT2-miRNA or LacZ-miRNA were first cultured in steroid-deprived medium for at least 3 days, and subsequently treated with 10 nM E2 or ethanol vehicle for 4 h. Representative western blot analysis indicating miRNA-mediated silencing of PRMT2 in MCF7 cells. E2-induced changes in CCND1 between LacZ and PRMT2 miRNA stable cells were quantitated and normalized to b-actin. Columns, mean of three independent replicates; bars, s.e.m., *Po0.05. (d) MCF7 cells carrying either PRMT2-miRNA or LacZ-miRNA were cultured as in c, and with the addition of wortmannin (200 nM) or LY294002 (20 mM). Total protein extracts were analyzed by western blot analysis for total and phosphorylated Akt and GSK-3b. (e) MCF7 cells carrying either PRMT2-miRNA or LacZ-miRNA were cultured as in c, and with the addition of LiCl (20 mM) or SB216763 (10 mM). Total protein extracts were analyzed by western blot analysis for total and phosphorylated GSK-3b.

& 2014 Macmillan Publishers Limited Oncogene (2014) 5546 – 5558 PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5554 Nuclear extract + + + + - Table 3. Correlation of PRMT2 nuclear expression with AP-1 probe + + + + + clinicopathological parameters Cold probe - - - + - PRMT2 antibody - + - - - n Nuclear expression of PRMT2 P-value Normal IgG - - + - - Negative Positive

Supershift Age (years) 0.316 450 132 79 53 DNA/protein p50 184 123 61 complex Tumor size (cm) 0.133 42 287 179 108 p229236 Node metastasis 0.572 Positive 132 82 50 Negative 184 120 64 Tumor types 0.000 Invasion ductal 159 84 75 Free probe Invasion lobular 157 118 39 ER 0.556 Figure 7. The interaction between PRMT2 and AP-1 probe in MCF-7 Positive 112 74 38 cells. A strong supershift band with anti-PRMT2 antibody (Ab) could Negative 204 128 76 be detected in the EMSA assay with extracts from the MCF-7 cells PR 0.919 and AP-1 probe. Free probes migrate at the bottom of the gels. IgG Positive 87 56 31 was used as a negative control. Cold probe, unlabeled probe. Negative 229 146 83 HER2 0.751 Positive 64 42 22 Negative 252 160 92 Table 2. Correlation of PRMT2 and CCND1 expression Abbreviations: PRMT2, protein arginine N-methyltransferase 2; ER, estrogen CCND1 expression n PRMT2 expression P-value Spearman receptor; PR, progesterone receptor; HER2, human epidermal growth factor (r) receptor 2. Àþþþ

Breast cancer 156 0.000 0.346 carcinoma did not show a statistically significant correlation with À 64939 nodal status, metastasis, ER, progesterone receptor (PR) and þ 01535 human epidermal growth factor receptor 2 (HER2) status of the þþ 0111 tumor (Table 3, P40.05). Those tissue microarray analyses strongly Invasion ductal 79 0.003 0.327 argued that PRMT2 protein showed a high absent percentage in cancer tumor cell nuclei and this absent percentage of cell nuclei À 31621 þ 0823 correlates with increasing tumor grade of invasive ductal þþ 00 8 carcinoma. Invasion lobular 77 0.006 0.309 Furthermore, a positive correlation between nuclear PRMT2 loss cancer ratio and CCND1 expression (r ¼ 0.394, P ¼ 0.000; Table 4) was À 33318 verified in invasion ductal cancer, which is consistent with the þ 0712 finds that PRMT2 absent percentage in cell nuclei correlates with þþ 01 3 increasing tumor grade of invasive ductal carcinoma. In a tumor Abbreviations: CCND1, cyclin D1; PRMT2, protein arginine N-methyltrans- tissue overexpressing CCND1, nuclear expression of PRMT2 was ferase 2. not detected, whereas in the matched normal glands, the PRMT2 expression was in cell nucleic, and CCND1 protein expression was low (Figure 9a). In another tumor tissue where PRMT2 expression was in cell nucleus, and CCND1 expression was almost breast tumors would lead to an enhancement of CCND1 undetectable (Figure 9b). Those tissue microarray data further expression in breast cancer cells. To further confirm this confirm that CCND1 is at least in part responsible for the role of hypothesis, the correlations between nuclear PRMT2 loss and PRMT2 in proliferation regulation of breast cancer cells. CCND1 expression were analyzed by using tissue microarrays (BR1921 and BR1921a, US Biomax), and the clinicopathologic data were available in Table 3. The immunohistochemistry analysis showed that PRMT2 protein was readily detectable in the nuclei of DISCUSSION adjacent normal breast tissue but was absent in 64% of the tumor Protein arginine N-methyltransferases participates in a number of cell nuclei. PRMT2 immunostaining of tumor and adjacent normal cellular processes, including cell growth,30 nuclear/cytoplasmic tissue and the corresponding hematoxylin and eosin stains of a protein shuttling,31 differentiation and embryogenesis,32,33 RNA representative case of breast invasive ductal carcinoma are shown splicing and transport,34,35 and posttranscriptional regulation.36 in Figure 8. The percentage of tumors showing nuclear loss of Recently, studies showed that overexpression of these PRMT2 protein is shown to be associated with the pathological are often associated with carcinogenesis and metastasis, especially types (P ¼ 0.000; Table 3). In invasive ductal carcinoma, the in breast cancer.37–40 It has been reported that PRMT2 percentage of tumors showing nuclear loss of PRMT2 protein interacts with a number of nuclear receptors in a ligand- significantly increased with the progression of the tumor grade: independent manner, and enhances nuclear receptor-mediated 17% of grade 1, 62% of grade 2 and 82% of grade 3 (r ¼ 0.431, transactivation.15,16 PRMT2 directly represses E2F activity in an RB- P ¼ 0.000; Table 4). Loss of nuclear PRMT2 expression in breast dependent manner and depletion of PRMT2 increases

Oncogene (2014) 5546 – 5558 & 2014 Macmillan Publishers Limited PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5555 HE stain PRMT2 PRMT2

Normal

Tumor

Figure 8. Nuclear expression of PRMT2 is absent in breast carcinoma (images in original magnification, Â 400). Tumor cells lack nuclear PRMT2 protein, whereas the adjacent normal glands express PRMT2 in the nucleus. H&E, hematoxylin and eosin.

on the effect of E , and other regulation mechanisms might also Table 4. Correlation of PRMT2 and CCND1 expression in invasion 2 be involved in this process. ductal cancer The CCND1 gene is a well-recognized oncogene that is 44 n Nuclear expression P-value Spearman (r) overexpressed in over 50% of human mammary carcinomas. of PRMT2 Induction of CCND1 gene transcription by ERa has an important role in estrogen-mediated proliferation.45 Although CCND1 gene transcription is directly induced by estrogen, there is no classical Negative Positive ERE in the CCND1 promoter.12 Instead, the CCND1 promoter Grade 0.000 0.431 contains multiple potential estrogen-responsive sites, including an I41734 atypical cyclic AMP response element that bind c-Jun/ATF-2 II 101 63 38 proteins,46,47 an AP-1 sites that bind Jun/Fos48,49 and Sp1 sites.50,51 III 17 14 3 In this study, using luciferase reporter assay and ChIP PCR, we CCND1 0.000 0.394 have demonstrated that downregulation of PRMT2 expression led Positive 71 52 19 to an increased recruitment of ERa to promoter region of AP-1 site Negative 88 32 56 in CCND1 promoter, and increased the promoter activity of Abbreviations: CCDN1, cyclin D1; PRMT2, protein arginine N-methyltrans- CCND1, indicating that PRMT2 could downregulate the expression ferase 2. of CCND1 via suppressing the recruitment of ERa to the AP-1 site within the CCND1 promoter. Further EMSA assays showed that PRMT2 bind to AP-1 site in an indirect manner, and it could be endogenous E2F activity and causes cells to enter S phase inferred that the indirect binding of PRMT2 to AP-1 site could have earlier.41 In addition, PRMT2 can block nuclear factor of kappa a negative effect on the recruitment of ERa to the AP-1 site of light polypeptide gene enhancer in B-cells inhibitor alpha (IkBa) CCND1 promoter, leading to the downregulation of CCND1 in nuclear export, inhibit nuclear factor-kB transcription and promote MCF-7 cells. Those findings reveal a novel mechanism by which cell apoptosis.23 These findings suggest that PRMT2 has diverse PRMT2 is involved in CCND1 expression regulation in breast roles in transcriptional regulation through different mechanisms. cancer cells. The molecules involved in the binding of PRMT2 with Estrogens elicit proliferative responses in breast cancer cells AP-1 site of CCND1 promoter is of importance to unravel the through ERa-mediated transcriptional mechanisms.42 PRMT2 has accurate mechanism of PRMT2-induced CCND1 downregulation been implicated in the positive regulation of ERa-mediated gene and deserves further investigation. activation in response to estrogen signaling by binding to ERa in It is well known that CCND1 overexpression is a common event CV-1 cells.15 Thus, PRMT2 may be involved in the control of cell in cancer and is usually a result of defective regulation at the post- proliferation through the effect on cell cycle genes. This study translational level.52,53 Therefore, regulation of the CCND1 protein provides the first evidence for the role of PRMT2 downregulation level is one of the critical aspects in cell proliferation and tumor in proliferation regulation of breast cancer cells, strongly development. Previous studies have reported that CCND1 suggesting that PRMT2 function as negative modulator for the expression level is mediated by Akt/GSK-3b pathway.27–29 It was proliferation of breast cancer cells. Using the whole genome-wide also reported that GSK-3b could be phosphorylated and transcriptome profile and real-time PCR analysis, we have shown inactivated by Akt leading to an increase of CCND1 that in breast cancer cells, miRNA-mediated RNA interference of stabilization.54 In this study, we demonstrated that miRNA- PRMT2 affect a large body of genes expression, some of which are mediated PRMT2 downregulation promoted CCND1 expression involved in cancer-associated genes and proliferation-related via activation of Akt/GSK-3b/CCND1 signaling in breast cancer genes. CCND1, playing a critical role in G1-S phase cell cycle cells. Thus, it could be inferred that, except for its effect on CCND1 progression,11,43 was found to be one of the PRMT2 affecting promoter activity, PRMT2 also serve as a negative modulator of downstream targets. Furthermore, we have shown that increased Akt/GSK-3b/CCND1 aix leading to the proliferation suppression of PRMT2 expression in MCF7 cells led to a decrease in CCND1 breast cancer cells. expression, whereas downregulation of PRMT2 expression led to The expression of PRMT2 and CCND1 was found to be positively an enhancement of CCND1 expression. It was observed that on correlated in breast cancer tissues through the analysis of tissue the condition of E2, the effect of PRMT2 on CCND1 expression microarray, which seem to be inconsistent to the negative became more obvious. Thus, it could be inferred that the negative regulatory effect of PRMT2 on CCND1 expression. Further analysis effect of PRMT2 on cell proliferation would be partially depended to the tissue microarray data unraveled a fact that the level of

& 2014 Macmillan Publishers Limited Oncogene (2014) 5546 – 5558 PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5556 HE stain PRMT2 CCND1

Normal

Tumor

HE stain PRMT2 CCND1

Normal

Tumor

Figure 9. PRMT2 and CCND1 protein expression show an inverse correlation in breast carcinoma (images in original magnification, Â 400). (a) A representative case of breast carcinoma shows loss of nuclear PRMT2 expression, whereas strong nuclear PRMT2 expression is seen in the adjacent normal glands. The tumor overexpresses the CCND1 protein with adjacent normal glands showing weak CCND1 staining. (b) Another representative case of breast carcinoma in which PRMT2 expression is nuclear in the tumor cells, which are negative for CCND1. Nuclear PRMT2 expression and weak nuclear staining with the CCND1 antibody are observed in the adjacent, uninvolved glands. H&E, hematoxylin and eosin.

PRMT2 in cell nucleus of breast cancer tissue showed obvious found to be involved in the regulation of CCND1 expression. decrease compared with those in normal breast tissue, although Furthermore, it was found that the absent percentage of PRMT2 in the total amount of PRMT2 presented an increase in breast cancer cell nuclei correlates with increasing tumor grade of invasive tissue, indicating that the elevated expression of PRMT2 in breast ductal carcinoma. Those results provide an essential insight into cancer tissue might locate mainly in cytoplasm. Furthermore, a the effect of PRMT2 on carcinogenesis of breast tumor, rending positive correlation between nuclear PRMT2 loss ratio and CCND1 PRMT2 to be important biological marker for the diagnosis of expression in invasion ductal cancer was verified, which is breast cancer. consistent with the results from luciferase and ChIP assays in cells models, suggesting that loss of PRMT2 may be of importance in deregulating CCND1 expression in human breast cancer. MATERIALS AND METHODS Further analysis identify that the PRMT2 absent percentage of Cell culture cell nucleus correlates with increasing tumor grade of invasive MCF7 and T47D were cultured as described previously.24 ductal carcinoma, and this phenomenon was also observed in the investigation on the role of BTG2 in breast cancer progression, in Design of artificial miRNAs and transfection which the loss of nuclear BTG2 expression in ERa-positive breast We designed the PRMT2-targeting sequence, PRMT2-miRNA1 was targeted tumors was showed to be correlated significantly with increased 55 to exon 4 and PRMT2-miRNA2 was targeted to exon 9 (Figure 1a). These histologic grade and tumor size. The mechanism by which miRNA gene double strands were ligated and cloned into the PRMT2 nuclear transportation is blocked during breast pcDNA6.2GW/EmGFP vector (Invitrogen, Carlsbad, CA, USA) according to carcinogenesis remains elusive, and the clarification of the manufacturer’s instructions. Then, the vectors were transfected into the associated molecular events would be helpful for understanding MCF7 cells using Lipofectamine2000 (Invitrogen) according to the the role of PRMT2 cytoplasm accumulation in the progress of manufacturer’s instruction. breast cancer. To our knowledge, this is the first report for the role of PRMT2 in Western blot analysis proliferation regulation of breast cancer cells. The present work Total cell or tissue lysates were lysed on ice for 30 min. Soluble proteins provides evidences for the underling mechanism by which PRMT2 (30 mg) were probed with anti-PRMT2, anti-CCND1, anti-c-myc and anti-ERa negatively regultes the expression of CCND1 in breast cancer cells, antibodies (1:500; Abcam, Cambridge, MA, USA) and anti-p-AKT, anti-AKT, and both genomic effect and non-genomic effect of PRMT2 were anti-p-GSK-3b, anti-GSK-3b (1:800; Cell Signaling Technology, Danvers,

Oncogene (2014) 5546 – 5558 & 2014 Macmillan Publishers Limited PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5557 MA, USA). Loading variations were normalized against b-actin, which ACKNOWLEDGEMENTS was identified by anti-b-actin monoclonal Ab (1:1000; Abcam). This work is supported by projects from the National Natural Science Foundation of P.R.China (Grant No. 31200573, 81272906, 81272355 and 81172542), Hunan Provincial Natural Science Foundation of China (12JJ3116) and The Education Plasmid construction Department of Hunan Province Youth Fund (12B108). PRMT2-expressing plasmid was generated by inserting the encoding region into pcDNA3.1 expression vector (Invitrogen, Carlsbad, CA, USA) at the sites of the BamHI and XhoI restriction . PRMT2 primers were 50- REFERENCES 0 0 GACGGATCCATGGCAACATCAGGTGAC-3 and 5 -GCCCTCGAGTCATCTCCAG 1 Sayeed A, Konduri SD, Liu W, Bansal S, Li F, Das GM. 0 ATGGGGAA-3 . CCND1 promoter was subcloned into pGL4.10 basic inhibits p53-mediated transcriptional repression: implications for the regulation plasmid (Promega, Madison, WI, USA) at the KpnI and HindIII restriction of apoptosis. Cancer Res 2007; 67: 7746–7755. enzyme sites to drive luciferase expression. The amplification primers were 2 Feng Y, Manka D, Wagner KU, Khan SA. Estrogen receptor-alpha expression in the 0 0 0 5 -GACGGTACCGTTTGCGATAATATTGGG-3 and 5 -GCCAAGCTTACGTTACTG mammary epithelium is required for ductal and alveolar morphogenesis in mice. 0 0 TTGTTAAGCAAAG-3 for a 1182-bp fragment, 5 -GACGGTACCCTAGAAGG Proc Natl Acad Sci USA 2007; 104: 14718–14723. 0 0 0 ACAAGATGAAGG-3 and 5 -GCCAAGCTTACGTTACTGTTGTTAAGCAAAG-3 3 Le Romancer M, Poulard C, Cohen P, Sentis S, Renoir JM, Corbo L. Cracking the for a 747-bp fragment. estrogen receptor’s posttranslational code in breast tumors. Endocr Rev 2011; 32: 597–622. 4 Ferguson AT, Davidson NE. Regulation of estrogen receptor alpha function in Transient transfection and luciferase activity assay breast cancer. Crit Rev Oncog 1997; 8: 29–46. Transient gene delivery was carried out to assess the effect of PRMT2 on 5 Metivier R, Penot G, Hubner MR, Reid G, Brand H, Kos M et al. Estrogen receptor- 24 CCND1 promoter activity in MCF7 and T47D cells as described previously. alpha directs ordered, cyclical, and combinatorial recruitment of cofactors on a A luciferase assay kit (Promega) was used to measure the reporter activity natural target promoter. Cell 2003; 115: 751–763. according to the manufacturer’s instructions. Luciferase activity was 6 Saville B, Wormke M, Wang F, Nguyen T, Enmark E, Kuiper G et al. Ligand-, cell-, normalized by using a Renilla luciferase internal control. and estrogen receptor subtype (alpha/beta)-dependent activation at GC-rich (Sp1) promoter elements. J Biol Chem 2000; 275: 5379–5387. 7 DeNardo DG, Kim HT, Hilsenbeck S, Cuba V, Tsimelzon A, Brown PH. Global gene Site-directed mutagenesis expression analysis of estrogen receptor transcription factor cross talk in breast Mutants (pLuc-901m and pLuc-785m) of CCND1 promoter were generated cancer: identification of estrogen-induced/activator protein-1-dependent genes. using a QuikChange Lightning Site-Directed Mutagenesis Kit (Stratagene, Mol Endocrinol 2005; 19: 362–378. La Jolla, CA, USA). The TGAGTCA sequence in the binding site of AP-1 was 8 Stein B, Yang MX. Repression of the interleukin-6 promoter by estrogen receptor replaced by GAGACAG, the GGTCAAG sequence in the ERE-like site was is mediated by NF-kappa B and C/EBP beta. Mol Cell Biol 1995; 15: 4971–4979. replaced by AAAAGGA, and confirmed by sequencing. 9 Cheng AS, Jin VX, Fan M, Smith LT, Liyanarachchi S, Yan PS et al. Combinatorial analysis of transcription factor partners reveals recruitment of c-MYC to estrogen receptor-alpha responsive promoters. Mol Cell 2006; 21: 393–404. ChIP assay 10 Bocchinfuso WP, Korach KS. Mammary gland development and tumorigenesis in MCF7 cells were fixed by the addition of 1% formaldehyde to the medium estrogen receptor knockout mice. J Mammary Gland Biol Neoplasia 1997; 2: for 10 min. Formaldehyde was quenched by the addition of 1 Â for 323–334. 5 min at room temperature. A ChIP assay was performed as described 11 Eeckhoute J, Carroll JS, Geistlinger TR, Torres-Arzayus MI, Brown M. previously56 with 5 ml of immunoprecipitated DNA and primers (forward 50- A cell-type-specific transcriptional network required for estrogen regulation AGGTGAAGGGACGTCTAC-30 and reverse 50-TAGCCTGGAGACTCTTCG-30) of cyclin D1 and cell cycle progression in breast cancer. Genes Dev 2006; 20: located at À 1019 to À 1002 bp and À 764 to À 747 bp of the CCND1 2513–2526. promoter. 12 Klein EA, Assoian RK. Transcriptional regulation of the cyclin D1 gene at a glance. J Cell Sci 2008; 121: 3853–3857. 13 Katsanis N, Yaspo ML, Fisher EM. Identification and mapping of a novel human Tissue microarray and immunohistochemical analysis gene, HRMT1L1, homologous to the rat protein arginine N-methyltransferase 1 Two tissue microarrays (BR1921 and BR1921a, US Biomax, Rockville, MD, (PRMT1) gene. Mamm Genome 1997; 8: 526–529. USA) were used, and each of them consists of 160 breast cancer cases and 14 Scott HS, Antonarakis SE, Lalioti MD, Rossier C, Silver PA, Henry MF. Identification 32 normal cases, respectively, and were histologically interpretable and and characterization of two putative human arginine methyltransferases analyzed for the correlation with clinicopathological parameters. Immu- (HRMT1L1 and HRMT1L2). Genomics 1998; 48: 330–340. nohistochemistry staining was performed as detailed in our previous 15 Qi C, Chang J, Zhu Y, Yeldandi AV, Rao SM, Zhu Y-J. Identification of protein studies.24 The rabbit polyclonal PRMT2 (1:50; Aviva Systems Biology, arginine methyltransferase 2 as a for estrogen receptor alpha. J Biol Beijing, China) and CCND1 Ab (1:10; Cell Signaling Technology) were used. Chem 2002; 277: 28624–28630. 16 Meyer R, Wolf SS, Obendorf M. PRMT2, a member of the protein arginine methyltransferase family, is a coactivator of the androgen receptor. J Steroid Additional assays and methods Biochem Mol Biol 2007; 107: 1–14. 17 Herrmann F, Pably P, Eckerich C, Bedford MT, Fackelmayer FO. Human protein Cell growth and colony formation assay, global cDNA microarray analysis, arginine methyltransferases in vivo—distinct properties of eight canonical real-time PCR, mouse tumour xenografts experiments and EMSAs are members of the PRMT family. J Cell Sci 2009; 122: 667–677. described in Supplementary Materials and methods. 18 Lakowski TM, Frankel A. Kinetic analysis of human protein arginine N-methyl- transferase 2: formation of monomethyl- and asymmetric dimethyl-arginine residues on histone H4. Biochem J 2009; 421: 253–261. Statistical analysis 19 Blythe SA, Cha SW, Tadjuidje E, Heasman J, Klein PS. Beta-catenin primes orga- All experiments were performed with three replicates and the results were nizer gene expression by recruiting a histone H3 arginine 8 methyltransferase, expressed as the mean±s.e.m. or mean±s.d. Statistical analysis was done Prmt2. Dev Cell 2010; 19: 220–231. using SPSS, version 13.0 (SPSS, Inc., Chicago, IL, USA). A statistical 20 Iwasaki H, Kovacic JC, Olive M, Beers JK, Yoshimoto T, Crook MF et al. Disruption association between clinicopathological and molecular parameters was of protein arginine N-methyltransferase 2 regulates leptin signaling and tested, using non-parametrically two-tailed Mann–Whitney U-test. P values produces leanness in vivo through loss of STAT3 methylation. Circ Res 2010; 107: o0.05 were considered significant. Spearman’s rank-correlation coeffi- 992–1001. cients were used to assess the relationship between PRMT2 and CCND1 21 Yildirim AO, Bulau P, Zakrzewicz D, Kitowska KE, Weissmann N, Grimminger F et al. expression. Increased protein arginine methylation in chronic hypoxia: role of protein argi- nine methyltransferases. Am J Respir Cell Mol Biol 2006; 35: 436–443. 22 Besson V, Brault V, Duchon A, Togbe D, Bizot J-C, Quesniaux VFJ et al. Modeling the monosomy for the telomeric part of human 21 reveals CONFLICT OF INTEREST haploinsufficient genes modulating the inflammatory and airway responses. Hum The authors declare no conflict of interest. Mol Genet 2007; 16: 2040–2052.

& 2014 Macmillan Publishers Limited Oncogene (2014) 5546 – 5558 PRMT2 nuclear loss correlates with cyclin D1 overexpression J Zhong et al 5558 23 Ganesh L, Yoshimoto T, Moorthy NC, Akahata W, Boehm M, Nabel EG et al. Protein 40 Thomassen M, Tan Q, Kruse TA. Gene expression meta-analysis identifies chro- Methyltransferase 2 inhibits NF-{kappa}B function and promotes apoptosis. Mol mosomal regions and candidate genes involved in breast cancer metastasis. Cell Biol 2006; 26: 3864–3874. Breast Cancer Res Treat 2009; 113: 239–249. 24 Zhong J, Cao RX, Zu XY, Hong T, Yang J, Liu L et al. Identification and char- 41 Yoshimoto T, Boehm M, Olive M, Crook MF, San H, Langenickel T et al. The acterization of novel spliced variants of PRMT2 in breast carcinoma. FEBS J 2012; arginine methyltransferase PRMT2 binds RB and regulates E2F function. Exp Cell 279: 316–335. Res 2006; 312: 2040–2053. 25 Chung K-H, Hart CC, Al-Bassam S, Avery A, Taylor J, Patel PD et al. Polycistronic 42 Katzenellenbogen BS, Montano MM, Ediger TR, Sun J, Ekena K, Lazennec G et al. RNA polymerase II expression vectors for RNA interference based on BIC/miR-155. Estrogen receptors: selective ligands, partners, and distinctive pharmacology. Nucleic Acids Res 2006; 34: e53. Recent Prog Horm Res 2000; 55: 163–193, discussion 194-165. 26 Fu M, Wang C, Li Z, Sakamaki T, Pestell RG. Minireview: CYCLIN D1: normal and 43 Baldin V, Lukas J, Marcote MJ, Pagano M, Draetta G. Cyclin D1 is a nuclear protein abnormal functions. Endocrinology 2004; 145: 5439–5447. required for cell cycle progression in G1. Genes Dev 1993; 7: 812–821. 27 Ong CS, Zhou J, Ong CN, Shen HM. Luteolin induces G1 arrest in human naso- 44 Arnold A, Papanikolaou A. Cyclin D1 in breast cancer pathogenesis. J Clin Oncol pharyngeal carcinoma cells via the Akt-GSK-3beta-Cyclin D1 pathway. Cancer Lett 2005; 23: 4215–4224. 2010; 298: 167–175. 45 Steeg PS, Zhou Q. Cyclins and breast cancer. Breast Cancer Res Treat 1998; 52: 28 Takahashi-Yanaga F, Sasaguri T. GSK-3beta regulates cyclin D1 expression: a new 17–28. target for chemotherapy. Cell Signal 2008; 20: 581–589. 46 Sabbah M, Courilleau D, Mester J, Redeuilh G. Estrogen induction of the cyclin D1 29 Ye X, Guo Y, Zhang Q, Chen W, Hua X, Liu W et al. BetaKlotho suppresses tumor promoter: involvement of a cAMP response-like element. Proc Natl Acad Sci USA growth in hepatocellular carcinoma by regulating Akt/GSK-3beta/cyclin D1 sig- 1999; 96: 11217–11222. naling pathway. PLoS One 2013; 8: e55615. 47 Castro-Rivera E, Samudio I, Safe S. Estrogen regulation of cyclin D1 gene 30 Lin W-J, Gary JD, Yang MC, Clarke S, Herschman HR. The mammalian immediate- expression in ZR-75 breast cancer cells involves multiple enhancer elements. J Biol early TIS21 protein and the leukemia-associated BTG1 protein interact with a Chem 2001; 276: 30853–30861. protein-arginine N-methyltransferase. J Biol Chem 1996; 271: 15034–15044. 48 Albanese C, Johnson J, Watanabe G, Eklund N, Vu D, Arnold A et al. Transforming 31 McBride AE, Silver PA. State of the arg: protein methylation at arginine comes of p21ras mutants and c-Ets-2 activate the cyclin D1 promoter through distin- age. Cell 2001; 106: 5–8. guishable regions. J Biol Chem 1995; 270: 23589–23597. 32 Chen SL, Loffler KA, Chen D, Stallcup MR, Muscat GEO. The coactivator-associated 49 Shen Q, Uray IP, Li Y, Krisko TI, Strecker TE, Kim HT et al. The AP-1 transcription arginine methyltransferase is necessary for muscle differentiation. CARM1 coac- factor regulates breast cancer cell growth via cyclins and E2F factors. Oncogene tivates myocyte enhancer factor-2. J Biol Chem 2002; 277: 4324–4333. 2008; 27: 366–377. 33 Torres-Padilla ME, Parfitt DE, Kouzarides T, Zernicka-Goetz M. Histone arginine 50 Marampon F, Casimiro MC, Fu M, Powell MJ, Popov VM, Lindsay J et al. Nerve methylation regulates pluripotency in the early mouse embryo. Nature 2007; 445: growth factor regulation of cyclin D1 in PC12 cells through a p21RAS extracellular 214–218. signal-regulated kinase pathway requires cooperative interactions between Sp1 34 Lukong KE, Richard S. Arginine methylation signals mRNA export. Nat Struct Mol and nuclear factor-kappaB. Mol Biol Cell 2008; 19: 2566–2578. Biol 2004; 11: 914–915. 51 Bartusel T, Schubert S, Klempnauer KH. Regulation of the cyclin D1 and cyclin A1 35 Meister G, Fischer U. Assisted RNP assembly: SMN and PRMT5 complexes coop- promoters by B-Myb is mediated by Sp1 binding sites. Gene 2005; 351: 171–180. erate in the formation of spliceosomal UsnRNPs. EMBO J 2002; 21: 5853–5863. 52 Tashiro E, Tsuchiya A, Imoto M. Functions of cyclin D1 as an oncogene and 36 Li H, Park S, Kilburn B, Jelinek MA, Henschen-Edman A, Aswad DW et al. Lipo- regulation of cyclin D1 expression. Cancer Sci 2007; 98: 629–635. polysaccharide-induced methylation of HuR, an mRNA-stabilizing protein, by 53 Kim JK, Diehl JA. Nuclear cyclin D1: an oncogenic driver in human cancer. JCell CARM1. J Biol Chem 2002; 277: 44623–44630. Physiol 2009; 220: 292–296. 37 Zhong J, Cao RX, Hong T, Yang J, Zu XY, Xiao XH et al. Identification and 54 Ye Y, Xiao Y, Wang W, Yearsley K, Gao JX, Shetuni B et al. ERalpha signaling expression analysis of a novel transcript of the human PRMT2 gene resulted from through slug regulates E-cadherin and EMT. Oncogene 2010; 29: 1451–1462. alternative polyadenylation in breast cancer. Gene 2011; 487: 1–9. 55 Kawakubo H, Brachtel E, Hayashida T, Yeo G, Kish J, Muzikansky A et al. Loss of 38 Goulet I, Gauvin G, Boisvenue S, Cote J. Alternative splicing yields protein arginine B-cell translocation gene-2 in estrogen receptor-positive breast carcinoma is methyltransferase 1 isoforms with distinct activity, substrate specificity, and associated with tumor grade and overexpression of cyclin d1 protein. Cancer Res subcellular localization. J Biol Chem 2007; 282: 33009–33021. 2006; 66: 7075–7082. 39 El Messaoudi S, Fabbrizio E, Rodriguez C, Chuchana P, Fauquier L, Cheng D et al. 56 Zu X, Ma J, Liu H, Liu F, Tan C, Yu L et al. Pro-oncogene Pokemon promotes breast Coactivator-associated arginine methyltransferase 1 (CARM1) is a positive reg- cancer progression by upregulating survivin expression. Breast Cancer Res 2011; ulator of the Cyclin E1 gene. Proc Natl Acad Sci USA 2006; 103: 13351–13356. 13: R26.

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

Oncogene (2014) 5546 – 5558 & 2014 Macmillan Publishers Limited