Identification and characterization of novel spliced variants of PRMT2 in breast carcinoma

Authors

  • Jing Zhong,

    1.  Clinical Medical Research Institute of the First Affiliated Hospital, University of South China, Hengyang, China
    2.  Department of Pathophysiology, University of South China, Hengyang, China
    Search for more papers by this author
    • These authors contributed equally to this work

  • Ren-Xian Cao,

    1.  Clinical Medical Research Institute of the First Affiliated Hospital, University of South China, Hengyang, China
    2.  Department of Pathophysiology, University of South China, Hengyang, China
    Search for more papers by this author
    • These authors contributed equally to this work

  • Xu-Yu Zu,

    1.  Clinical Medical Research Institute of the First Affiliated Hospital, University of South China, Hengyang, China
    Search for more papers by this author
  • Tao Hong,

    1.  Clinical Medical Research Institute of the First Affiliated Hospital, University of South China, Hengyang, China
    2.  Department of Pathophysiology, University of South China, Hengyang, China
    Search for more papers by this author
  • Jing Yang,

    1.  Clinical Medical Research Institute of the First Affiliated Hospital, University of South China, Hengyang, China
    2.  Department of Pathophysiology, University of South China, Hengyang, China
    Search for more papers by this author
  • Ling Liu,

    1.  Clinical Medical Research Institute of the First Affiliated Hospital, University of South China, Hengyang, China
    Search for more papers by this author
  • Xin-Hua Xiao,

    1.  Clinical Medical Research Institute of the First Affiliated Hospital, University of South China, Hengyang, China
    Search for more papers by this author
  • Wen-Jun Ding,

    1.  Clinical Medical Research Institute of the First Affiliated Hospital, University of South China, Hengyang, China
    Search for more papers by this author
  • Qiang Zhao,

    1.  Department of Pathology of the First Affiliated Hospital, University of South China, Hengyang, China
    Search for more papers by this author
  • Jiang-Hua Liu,

    1.  Clinical Medical Research Institute of the First Affiliated Hospital, University of South China, Hengyang, China
    Search for more papers by this author
  • Ge-Bo Wen

    1.  Clinical Medical Research Institute of the First Affiliated Hospital, University of South China, Hengyang, China
    2.  Department of Pathophysiology, University of South China, Hengyang, China
    Search for more papers by this author

Ge-Bo Wen, Clinical Medical Research Institute of the First Affiliated Hospital, University of South China, Hengyang 421001, China
Fax: +86 734 8279009
Tel: +86 734 8279392
E-mail: gb_wen@yahoo.com.cn

Abstract

Protein N-arginine methyltransferases (PRMTs) participate in a number of cellular processes, including cell growth, nuclear/cytoplasmic protein shuttling, differentiation, RNA splicing and post-transcriptional regulation. PRMT2 (also known as HRMT1L1) is clearly involved in lung function, the inflammatory response, apoptosis promotion, Wnt signaling and leptin signaling regulation through different mechanisms. In this study, we report the molecular and cell biological characterization of three novel PRMT2 splice variants isolated from breast cancer cells and referred to as PRMT2α, PRMT2β and PRMT2γ. Compared with the wild-type PRMT2, these variants lack different motifs and therefore generate distinct C-terminal domains. Confocal microscopy scanning revealed a distinct intracellular localization of PRMT2 variants, suggesting that the alternatively spliced C-terminus of PRMT2 can directly influence its subcellular localization. Our findings reveal that these variants are capable of binding to estrogen receptor alpha (ERα) both in vitro and in vivo, and the N-terminal regions of these variants contribute to ERα–PRMT2 interactions. Furthermore, these variants were proved to be able to enhance ERα-mediated transactivation activity. Luciferase reporter assays showed that PRMT2s could modulate promoter activities of the ERα-targeted genes of Snail and E-cadherin. In addition, PRMT2 silencing could enhance 17β-estradiol-induced proliferation by regulating E2F1 expression and E2F1-responsive genes in ERα-positive breast cancer cells. Real-time PCR and immunohistochemistry showed that overall PRMT2 expression was upregulated in breast cancer tissues and significantly associated with ERα positivity status both in breast cancer cell lines and breast cancer tissues. We speculate that PRMT2 and its splice variants may directly modulate ERα signaling and play a role in the progression of breast cancer.

Structured digital abstract

Abbreviations
AR

androgen receptor

E2

17β-estradiol

ERE-LUC

estrogen-responsive-element-containing luciferase reporter

ERα

estrogen receptor α

FBS

fetal bovine serum

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

GFP

green fluorescent protein

GST

glutathione S-transferase

PR

progesterone receptor

PRMT2

arginine N-methyltransferase 2

Introduction

Protein arginine N-methyltransferases (PRMTs) are eukaryotic enzymes that catalyze the transfer of methyl groups from S-ade nosylmethionine to arginine residues of numerous PRMT substrates [1,2]. Their activities influence a wide range of cellular processes, including cell growth [3], nuclear/cytoplasmic protein shuttling [4], differentiation and embryogenesis [5,6], RNA splicing and transport [7,8] and post-transcriptional regulation [9]. PRMT2 (also known as HRMT1L1) is a type I enzyme that belongs to the arginine methyltransferase family. Although not initially described to have methyltransferase activity [10], subsequent research indicated that PRMT2 acts as an estrogen receptor α (ERα) coactivator and enhances estrogen-related transcription [11]. Recently, further studies reported that PRMT2 exerts multiple biological activities [12–14] and shows weak enzymatic activity [15,16].

ERα is a transcription factor involved in a broad range of biological processes, including cell proliferation, differentiation, morphogenesis and apoptosis [17–20], as well as the development and progression of breast cancer [21]. Regulation of gene expression by ERα requires the coordinated activity of ligand binding, phosphorylation and cofactor interactions; the latter may modulate tissue specificity of ERα responses [22]. A growing list of cofactors that regulate ERα includes the coactivators and the corepressors [23–25].

In eukaryotic cells, a single gene may give rise to various mRNA isoforms using alternative initiation, splicing, and termination and polyadenylation mechanisms [26]. The mRNA isoforms could code for different products, thereby increasing protein diversity. Tumorigenesis and tumor progression have been related to the expression of alternatively spliced mRNAs encoding altered forms of proteins [27–29]. Previous studies have shown PRMT2 to be tissue-specific mRNA expressed, but no alternative variants had been found except the wild type [10,13,30]. Recently, we identified a novel transcript of the human PRMT2 gene resulting from alternative polyadenylation in breast cancer, and the results of northern blot show that multiple alternative variants of PRMT2 may be present in humans [31]. To understand the physiological function of PRMT2, as well as its potential role in the development and progression of cancer, in this study we report the identification and characterization of these novel splice variants of PRMT2 which we have named PRMT2α, PRMT2β and PRMT2γ. Luciferase reporter assays showed that PRMT2s could modulate promoter activities of the ERα-targeted genes of Snail and E-cadherin. Moreover, PRMT2 silencing could enhance 17β-estradiol (E2) induced proliferation by regulating E2F1 expression and E2F1-responsive genes in ERα-positive breast cancer cells. Our data of real-time PCR and immunohistochemistry indicate that PRMT2 and its novel variants are upregulated in breast cancer tissues, and significantly associated with ERα positivity status both in breast cancer cell lines and breast cancer tissues. These results suggest that PRMT2 and its splice variants may play a role in the formation and development of breast cancer by modulating the estrogen–ERα signaling pathway.

Results

Identification and characterization of novel alternative splicing variants of PRMT2

To identify PRMT2 variants in breast cancer, the encoding region was amplified by RT-PCR using specific primers designed from its mRNA (NM_206962). Results showed that multiple bands were detected in several breast cancer cell lines such as MCF7 and T47D (Fig. 1A). Sequencing analysis revealed that these products represented new splicing variants of the PRMT2 gene, named PRMT2α, PRMT2β and PRMT2γ (GenBank accession numbers FJ436410, FJ436411 and FJ436412). Figure 1B shows the genomic organization of wild-type PRMT2 gene and these new alternative splicing variants. The difference between PRMT2α and PRMT2β PCR products was too small to be separated (Fig. 1A).

Figure 1.

 Identification of novel PRMT2 variants. (A) One-step RT-PCR was performed using total RNAs from the indicated breast cancer cells and the resulting products were separated on 1.0% agarose gel. Three distinct DNA bands represent different PRMT2 variants. Marker, 1 kb DNA molecular weight ladder. (B) Splicing events for PRMT2 variants, compared with wild-type PRMT2. Exons are indicated by boxes with numbers. Stop codons are designated by underlining and asterisks. (C) Deduced amino acid sequences and domains of PRMT2 variants, compared with wild-type PRMT2. PRMT2 domains containing the SH3, I, post I, II, III, post III and THW loop were aligned with comparison to the variants. PRMT2α and PRMT2β unique amino acid sequences are shaded. (D) Distributions of endogenous and exogenous PRMT2 variants in breast cancer cells were examined by western blotting using PRMT2 antibody.

Wild-type PRMT2 gene consists of 11 exons, stretching about 30 kp (NT_011515.11), and the protein contains 433 amino acids (∼ 48.5 kDa), constituting motifs SH3, I, post I, II, III, post III and THW loop. The alternative splicing occurs at the 3′ end of the PRMT2 gene, resulting in partial or complete loss of exons 7–10 and downstream frame-shifting (Fig. 1B). In particular, PRMT2β possesses 83 new amino acids and PRMT2α possesses 12 new amino acids at the C-terminus. The size of these variants is 289, 301 and 228 amino acids for PRMT2α, PRMT2β and PRMT2γ, respectively (Fig. 1C).

To examine and confirm the protein expressions of endogenous PRMT2 variants in cells, MCF7 cells were transfected with pcDNA3.0–PRMT2α, pcDNA3.0–PRMT2β and pcDNA3.0–PRMT2γ, respectively. These whole-cell extracts and untransfected 293T, SK-BR-3 and MCF7 cell extracts were immunoblotted with PRMT2 antibody. The antibody against the N-terminus of PRMT2 theoretically recognizes the wild-type PRMT2 and all variants. As shown in Fig. 1D, three major bands were detected with molecular weights of 48.5 kDa (PRMT2), 34 kDa/32.6 kDa (PRMT2β/PRMT2α) and 25.8 kDa (PRMT2γ), consistent with the predicted size. According to the band density, wild-type PRMT2 is the predominant form expressed in almost all cell fractions, and was also significantly higher in ERα-positive breast cancer MCF7 cells than in ERα-negative breast cancer SK-BR-3 cells and 293T cells. PRMT2α and PRMT2β syncretized into one band and nearly vanished in 293T cells. In addition, three exogenous PRMT2 variants were detected in corresponding cell extracts of transfected MCF7 cells.

Subcellular localization of PRMT2 isoforms

The previously described PRMTs are essentially located either in the cytoplasm (PRMT3) or in the nucleus (PRMTs 1, 4 and 6) [32–34] or in both (PRMTs 5 and 9) [15,35,36]. We wanted to compare the subcellular localization of PRMT2 and its variants. At present, reliable antibodies recognizing the different PRMT2 variants are not available. We therefore utilized green fluorescent protein (GFP) fusion proteins to study the expression of the PRMT2 variant proteins. Plasmid constructs expressing the different PRMT2 isoforms fused to an N-terminal GFP tag were generated and used to transiently transfect MCF7 cells grown on glass coverslips. Intracellular distribution of each PRMT2–GFP isoform was assessed 48 h post-transfection using confocal microscopy (Fig. 2A). Equal expression of each PRMT2–GFP fusion was confirmed by anti-GFP immunoblotting (Fig. 2B). As shown in Fig. 2A, cells expressing wild-type PRMT2 appear to be largely localized to the nucleus excluding the nucleoli and a weak fluorescence was detected in the cytosol as described previously [15,32,37], whereas cells expressing GFP–PRMT2β showed an even distribution of the GFP fusion proteins between the nucleus including the nucleoli and the cytoplasm. Strikingly, the GFP–PRMT2α and GFP–PRMT2γ fusion protein expressions were predominantly localized to the nuclear compartment excluding the nucleoli, as well as the distribution of wild-type PRMT2. In addition, the fluorescence signals related to ERα and PRMT2 variants overlap, indicating colocalization of the PRMT2 variants and ERα inside MCF7 cells (Fig. 2, Merge). To examine whether the subcellular localization of PRMT2 and its variants is MCF7 cell specific, this study was conducted in HepG2 cells, and similar results were observed (Fig. 3). Furthermore, the fluorescence signals related to androgen receptor (AR) and PRMT2 variants also overlap, indicating colocalization of PRMT2 variants and AR inside HepG2 cells (Fig. 3, Merge). Taken together, these results show that the N-terminus of PRMT2 contributes to its nuclear localization, and the alternatively spliced C-terminus of PRMT2 can directly influence its intracellular localization.

Figure 2.

 Subcellular distribution of GFP-tagged PRMT2 isoforms in MCF7 cells. (A) MCF7 cells were transiently transfected with N-terminal GFP-tagged PRMT2 variants for 48 h and cells were viewed under a Zeiss LSM 510 confocal microscope. Green represents the pixel intensity distribution of the GFP signal, blue depicts the profile of the exclusively nuclear DAPI staining, and localization of endogens ERα (in red) was determined by Cy3-conjugated antibody. (B) Total cell lysates of MCF7 cells transiently expressing the GFP-tagged PRMT2 variants were resolved by SDS/PAGE, transferred onto a polyvinylidene difluoride membrane and immunoblotted with GFP antibodies. A pcDNA3.1/NT–GFP expression plasmid was used as a control.

Figure 3.

 Subcellular distribution of GFP-tagged PRMT2 isoforms in HepG2 cells. HepG2 cells were transiently transfected with N-terminal GFP-tagged PRMT2 variants for 48 h; cells were viewed under a Zeiss LSM 510 confocal microscope. Green represents the pixel intensity distribution of the GFP signal, blue depicts the profile of the exclusively nuclear DAPI staining, and localization of endogens AR (in red) was determined by Cy3-conjugated antibody.

Overexpression of PRMT2 variants enhance ERα-mediated transactivation activity

In this study we evaluated the regulatory activity of the PRMT2 variants using human breast cancer MCF7 cells. Using this cell system, we tested the effect of PRMT2 variants on estrogen-responsive-element-containing luciferase reporter (ERE-LUC) activity. As shown in Fig. 4A, in the presence of estrogen the PRMT2 variants stimulated ERE-LUC activity in a dose-dependent manner. In particular, PRMT2α (0.5 μg) and PRMT2γ (0.5 μg) enhanced the transcriptional activity of ERα by 2.5- and 2.6-fold, respectively, similar to the action of wild-type PRMT2 (2.8-fold at 0.5 μg) (P < 0.01). Differentially, PRMT2β demonstrated weaker stimulatory activity to ERE-LUC with a 1.5-fold increase at 0.5 μg (P < 0.05). This transactivation activity was estrogen dependent. In the absence of estrogen, PRMT2 and the variants had no activity to ERE-LUC. The estrogen dependence of PRMT2 and its variants was further confirmed by anti-estrogen chemicals 4-hydroxytamoxifen (4-OHT) and ICI 182,780. The presence of either 4-OHT or ICI 182,780 at 100 nm nearly completely inhibited the regulatory activity (Fig. 4B).

Figure 4.

 PRMT2 and its variants enhance ERα-mediated transactivation. (A) MCF7 cells were cotransfected with 0.2 μg of ERE-LUC and increasing amounts of either PRMT2 or PRMT2 variants as indicated. Cells were then treated with control (0.1% ethanol) or 10 nm E2 for 24 h before luciferase assay. The luciferase activity obtained on transfection of ERE-LUC without exogenous PRMT2 and PRMT2 variants in the absence of E2 was set as 1. Results are expressed as mean ± SE for three independent experiments. *P < 0.05, **P < 0.01, versus empty-vector-transfected cells treated with E2. (B) MCF7 cells were cotransfected as in (A) and then treated with control (0.1% ethanol), 10 nm E2, 100 nm 4-OHT or 100 nm ICI 182,780 for 24 h before luciferase assay. Cells were analyzed as in (A). *P < 0.05, **P < 0.01, versus empty-vector-transfected cells treated with E2. (C) 293T cells were cotransfected with 0.2 μg of ERE-LUC and 0.5 μg of PRMT2 or PRMT2 variants in the absence or presence of ERα. Cells were analyzed as in (A). *P < 0.05, **P < 0.01, versus ERα-transfected cells treated with E2.

To examine whether the transactivation activity of PRMT2 and its variants is MCF7 cell specific, this cotransfection study was conducted in ERα-negative 293T cells and SK-BR-3 breast cancer cells, and similar results were observed (Table 1). The regulatory activity of PRMT2 variants on ERE-LUC was also ERα dependent. Without cotransfection of ERα in 293T cells, the coactivation activity of PRMT2 and its variants was negligible, even in the presence of estrogen (Fig. 4C). Taken together, our results suggest that PRMT2 and its variants could function as a co-regulator to enhance ERα-mediated transactivation activity in a ligand-dependent manner.

Table 1.   PRMT2 and its variants enhance ERα-mediated transcriptional activity. Cells were transfected, treated and analyzed as described in the legend to Fig. 4A.
Expressed proteinActivation of transcription (fold activation)
293T cellsSK-BR-3 cells
−E2+E2−E2+E2
ERα12.32 ± 0.7212.14 ± 0.41
ERα, PRMT21.30 ± 0.256.68 ± 0.821.21 ± 0.225.88 ± 0.62
ERα, PRMT2α1.29 ± 0.385.92 ± 0.661.18 ± 0.184.85 ± 0.56
ERα, PRMT2β1.25 ± 0.313.65 ± 0.581.20 ± 0.213.15 ± 0.76
ERα, PRMT2γ1.24 ± 0.286.11 ± 0.721.08 ± 0.234.81 ± 0.85

PRMT2 variants enhance promoter activities of Snail and repress promoter activities of E-cadherin

It is reported that Snail downregulates E-cadherin through ERα in the presence of E2 [38]. To further address the effect of PRMT2 variants on ERα target genes, we performed luciferase promoter assays and examined the ability of PRMT2 variants to transactivate luciferase reporter plasmids containing the human Snail or E-cadherin promoter. As shown in Fig. 5, in the presence of estrogen (10 nm), PRMT2 variants could strongly enhance the promoter activities of Snail gene. In particular, PRMT2α (0.5 μg) and PRMT2γ (0.5 μg) enhanced the promoter activity of Snail by 2.9- and 3.0-fold, respectively, consistent with the action of wild-type PRMT2 (2.9-fold at 0.5 μg). Differentially, PRMT2β showed weaker stimulatory activity to ERE-LUC with a 2.4-fold increase at 0.5 μg. This transactivation activity of PRMT2s was estrogen dependent. In the absence of estrogen, PRMT2 and the variants showed no obvious effect on the promoter activity of Snail. As transcription of E-cadherin is known to be regulated by the Snail repressors [39], we were interested to address the effect of PRMT2s on the promoter activities of E-cadherin gene. As shown in Fig. 5, in the absence of estrogen PRMT2α (0.5 μg) and PRMT2γ (0.5 μg) could reduce the promoter activity of E-cadherin (P < 0.05), whereas wild-type PRMT2 and PRMT2β showed no obvious effect. Moreover, with the same amount of PRMT2 variants, PRMT2s could significantly reduce the promoter activities of E-cadherin in the presence of estrogen (10 nm) (P < 0.01).

Figure 5.

 PRMT2 and its variants regulate gene promoter activities of Snail and E-cadherin. MCF7 cells were co-transfected with 0.2 μg of reporter construct of either Snail or E-cadherin promoter and 0.5 μg of either PRMT2 or PRMT2 variants. All cells were cotransfected with 25 ng Renilla as an internal control for transcription efficiency. Cells were then treated with control (0.1% ethanol) or 10 nm E2 for 24 h before luciferase assay. The luciferase activities were calculated relative to the promoterless vector (pGL4-Basic) and expressed as fold change relative to vehicle control. The data are shown as mean ± SE of three repeated experiments. *< 0.05 versus empty-vector-transfected cells treated with control (0.1% ethanol). **P < 0.01 versus empty-vector-transfected cells treated with 10 nm E2.

PRMT2 and its variants directly associate with ERα

To understand the molecular mechanisms of the ERα dependence of PRMT2 and its variants, we performed a protein–protein interaction study in vitro and inside the cells. The wild-type PRMT2 could bind to ERα both in the presence and absence of estrogen but could enhance the ERα activity only with estrogen as reported [11]. The following interaction studies were performed with treatment of 10 nm estrogen as described in the Experimental procedures section. As shown in Fig. 6A, PRMT2, PRMT2α and PRMT2γ demonstrated similar binding affinity to ERα, but PRMT2β was weaker. This binding activity of PRMT2 and its variants to ERα was confirmed inside the MCF7 cells. As shown in Fig. 6B, A-V5 tagged PRMT2 or its variants was co-immunoprecipitated by V5 antibody, and like the in vitro study the affinity of PRMT2β was weaker. To examine the binding activity of PRMT2 and its variants to AR, this study was conducted in HepG2 cells. After treatment with 10 nm AR agonist dihydrotestosterone (DHT) for 10 h, as shown in Fig. 6C, A-V5 tagged PRMT2 and its variants were co-immunoprecipitated by V5 antibody, and their affinities were similar. These results show that PRMT2 variants directly associate with ERα or AR in the presence of 10 nm estrogen or DHT.

Figure 6.

 PRMT2 and PRMT2 variants bind to ERα and AR in vitro and in vivo. (A) Interaction of PRMT2 and PRMT2 variants with ERαin vitro. Full-length GST–PRMT2 and GST–PRMT2 variant fusion proteins immobilized on beads were mixed with recombinant human ERα proteins (Invitrogen, P2187) in the presence of E2 (10 nm). Bound proteins were subjected to SDS/PAGE separation, followed by immunoblotting. (B) Interaction between PRMT2 or PRMT2 variants and ERαin vivo. MCF7 cells were cotransfected with the expression vectors for V5-tagged PRMT2 or PRMT2 variants as indicated. Lysates from the transfected cells were immunoprecipitated (IP) using V5 antibody and the immunoprecipitates were probed with an ERα antibody. (C) Interaction between PRMT2 or PRMT2 variants and AR in vivo. HepG2 cells were cotransfected with the expression vectors for V5-tagged PRMT2 or PRMT2 variants as indicated. Lysates from the transfected cells were IP using V5 antibody and the immunoprecipitates were probed with an AR antibody.

N-terminus of PRMT2 and its variants is responsible for the binding to ERα protein and transactivation activity

Except for PRMT2β, the PRMT2 variants showed similar binding and activation action to ERE-LUC (Fig. 4). To elucidate the functional domain(s) of PRMT2 and its variants in the transactivation regulation of ERE-responsive genes, we constructed a series of truncations to test their regulatory and binding activity to ERα as described in the Experimental procedures section. As shown in Fig. 7A, PRMT2 (1–218) and PRMT2 (1–277) possessed full transactivation activity compared with wild-type PRMT2, but PRMT2 (278–433) showed negligible regulatory activity to ERE-LUC, and PRMT2β (△1–218) relatively suppressed the transcription of ERE-LUC. Therefore, the N-terminus of PRMT2 and its variants may be responsible for their regulatory activity, but their C-terminus also partially affects the regulatory activity.

Figure 7.

 N-termini of PRMT2 and its variants are responsible for binding to ERα protein and transactivation activity. (A) MCF7 cells were cotransfected with 0.2 μg of ERE-LUC and 0.5 μg of V5-tagged PRMT2β (△1–218), PRMT2 (1–218), PRMT2 (278–433), PRMT2 (1–277) or V5-tagged PRMT2 as indicated. Cells were then treated and analyzed as in Fig. 4A. **P < 0.01 versus wild-type PRMT2-transfected cells treated with E2. (B) Different truncations, immobilized on beads, was mixed with the MCF7 cell lysates and subjected to immunoblotting with ERα antibody.

A protein–protein interaction assay confirmed this functional result. As shown in Fig. 7B, mutants PRMT2 (1–218) and PRMT2 (1–277) displayed binding activity to ERα, but PRMT2β (△1–218) and PRMT2 (278–433) did not. These data are consistent with the binding assays of PRMT2 and its variants, suggesting that the N-terminus of PRMT2 and its variants is responsible for their binding and regulatory activity.

PRMT2 silencing enhances E2-induced proliferation in ERα-positive breast cancer cells

To explore the role of PRMT2 in ERa-positive breast cancer cell proliferation, we examined the consequences of PRMT2 suppression on the growth stimulatory effect of E2 by using MCF7 cells as a model. Two kinds of small interfering RNA (siRNA) were designed to test the effect of endogenous PRMT2 on the growth of MCF7 cells. siPRMT2-N was targeted to exon 5 matching all PRMT2 variants, whereas siPRMT2-C was targeted to exon 9 specifically directing against the C-terminal domain of the wild PRMT2 (Fig. 8A). Expression of PRMT2 was efficiently and specifically silenced in MCF7 cells (Fig. 8B) with either of two different PRMT2 siRNA oligonucleotide duplexes. Upon silencing of PRMT2, we observed that the E2-mediated stimulation of MCF7 cell cycle progression was strongly enhanced (Fig. 8C). Furthermore, we observed an increase in the number of MCF7 cells following growth stimulation with E2 when PRMT2 was silenced compared with a control siRNA (Fig. 8D). We found a similar effect of PRMT2 silencing on E2-induced proliferation of T47D cells, another model of ERα-positive breast cancer (Fig. S1). These results indicate a critical role for PRMT2 in the proliferative response to estrogen in ERα-positive breast cancer cells.

Figure 8.

 Effect of PRMT2 silencing in MCF7 cells on E2-induced cell cycle progression. (A) Scheme representing the wild PRMT2 cDNA structure and the location of the siRNA: siPRMT2-N was located in exon 5 and siPRMT2-C was located in exon 9. (B) siRNA-mediated silencing of PRMT2 in MCF7 cells. Two different PRMT2-specific siRNAs or a control siRNA were transfected into hormone-depleted MCF7 cells. Extracts were collected and western blotting was done with PRMT2 antibodies. Anti-calnexin was used as a loading control. (C) Cell cycle analysis of PRMT2-depleted MCF7 cells. MCF7 cells were transfected as described in (A). Forty-eight hours later, cells were treated with 10 nm E2 or ethanol for 24 h and harvested to analyze DNA content by propidium iodide staining and flow cytometry. *P < 0.05 versus siControl transfected cells treated with E2. (D) Cell viability assays were done as described in the Experimental procedures section using MCF7 cells transfected with the indicated siRNAs. Cell proliferation is expressed as the percentage of the cells compared with the siControl transfected wells treated with E2. Columns, mean from one representative experiment done in triplicate; bars, SE. **< 0.01 versus similarly treated siControl transfected cells.

PRMT2 controls breast cancer proliferation by regulating E2F1 expression and E2F1-responsive genes

To address whether PRMT2 regulates the cell cycle by modulating transcription of specific genes that are involved in this process, we examined the effect of PRMT2 depletion on the expression of specific cell cycle genes. It was reported that the key cell cycle regulator E2F1 and, subsequently, its downstream target genes are critical for hormone regulation of the proliferative program of ERα-positive breast cancer cells [40]. We therefore investigated the effect of PRMT2 silencing on the expression of E2F1. As a result, estrogen stimulation induced a 3- and 5-fold increase in E2F1 mRNA after 3 and 12 h, respectively, and this effect was significantly increased by PRMT2 silencing (Fig. 9A). Accordingly, we also observed that E2F1 protein expression levels were strongly increased by PRMT2 silencing (Fig. 9B). E2F1 expression was also enhanced when PRMT2 was silenced in another breast cancer cell model, the T47D cell line (Fig. S2). Having shown that PRMT2 is important for estrogen regulation of E2F1, we examined whether PRMT2 silencing resulted in increased expression of the E2F1-responsive genes including CCNE1 and CCNA1. As predicted in E2-stimulated MCF7 cells in which PRMT2 was silenced, we observed a significant increase in the levels of CCNE1 and CCNA1 mRNA, as well as E2F1 mRNA, compared with the control (Fig. 9C,D). Hence, these results support an important role for PRMT2 in the regulation of E2F1 and E2F1-responsive genes.

Figure 9.

 PRMT2 silencing affects E2-induced expression of E2F1 and E2F1-responsive genes. (A) PRMT2-specific or control siRNAs were transfected into hormone-depleted MCF7 cells. Forty-eight hours after transfection, cells were treated with vehicle or 10 nm E2 for the indicated time period, and RNA was isolated to measure the expression of E2F1 gene by real-time RT-PCR. Columns, mean of three independent replicates; bars, SE. Shown are results obtained from siPRMT2-N transfections; similar results were obtained from siPRMT2-C transfections. *P < 0.05 versus siControl transfected cells. (B) After 48 h of transfection with siPRMT2 or control siRNA, MCF7 cells were stimulated with 10 nm E2 for 0, 24 and 48 h and whole-cell extracts from MCF7 cells were analyzed by western blot for E2F1. (C), (D) siRNA-mediated silencing of PRMT2 results in an increased level of E2F1-responsive mRNAs. PRMT2 was silenced by siRNA transfection and the relative levels of CCNE1 and CCNA1 were determined by real-time RT-PCR. *P < 0.05 versus siControl transfected cells.

Overexpression of PRMT2 variant mRNA in breast cancer is associated with ERα positivity

To assess whether the expression of PRMT2 C-terminal variant is distinct in breast cancer cells, we examined the expression of the PRMT2 gene in breast cancer cell lines using real-time PCR. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA was used as the control. The primer design allowed us to distinguish between the alternative splice forms (Fig. 10A, Table 2). PRMT2 variant expression in ERα-positive breast cancer cells is higher than that in ERα-negative cells. Specifically, the expression of the variant PRMT2γ is relatively higher in MCF7 cells than that of wild-type PRMT2 (Fig. 10B).

Figure 10.

 Differential expression of the human PRMT2 variants in breast cancer cells and tissues. (A) Schematic representation of gene-specific real-time PCR primers (arrows) and TaqMan probes (thick lines) used to amplify and detect each cDNA; boxes represent exons. (B) Relative expression of the PRMT2 variant mRNAs in ERα-positive and ERα-negative human breast cancer cells determined by quantitative real-time PCR. The experiments were performed in triplicate and the results are expressed as mean ± SE. (C) RNA samples representing adjacent normal breast tissues and 61 breast cancer cases were analyzed for transcripts encoding PRMT2 and its variants. GAPDH mRNA expression was used to normalize their expression. The SEs of three parallel analyses were <5% of the means.

Table 2.   Sequence of primers and TaqMan® Eclipse probes used in the quantitative real-time RT-PCR. FAM, 6-carboxyfluorescein.
Gene namePrimers and probes sequencesSize (bp)
PRMT2Forward: 5′-GTCCACTTCCAGAGCCTGCA-3′96
Reverse: 5′-CATGAACAGCGTCTGCTTCCA-3′
Probe: 5′-(FAM) AGCCGCCGCAGGTGCTCAGC (Eclipse)-3′
PRMT2αForward: 5′-CAGCGCTCTGAAGTTGGAGA-3′148
Reverse: 5′-ACCGACTTTCGCAGGAGTGT-3′
Probe: 5′-(FAM) CAAGCAAACCAAGTTGCACCTGGC (Eclipse)-3′
PRMT2βForward: 5′-GAGAAGGTGGACGTGCTGGT-3′79
Reverse: 5′-CATGAACAGCGTCTGCTTCCA-3′
Probe: 5′-(FAM) TGCAGCAGGCAGGTCCCCATCCA (Eclipse)-3′
PRMT2γForward: 5′-ATGTGGTGCTGCCCGAGAAG-3′98
Reverse: 5′-CTGTCATCTCCAGATGGGGAAGA-3′
Probe: 5′-(FAM) TCTCCAACCAGCAGGCAGGTCCCC (Eclipse)-3′
GAPDHForward: 5′-GGACCTGACCTGCCGTCTAG-3′99
Reverse: 5′-TAGCCCAGGATGCCCTTGAG-3′
Probe: 5′-(FAM) CCTCCGACGCCTGCTTCACCT (Eclipse)-3′
E2F1Forward: 5′-CCATCCAGGAAAAGGTGTGA-3′94
Reverse: 5′-GCTCAGCAGCTCCAGGAA-3′
CCNA1Forward: 5′-GGGCTCCCAGATTTCGTCTT-3′103
Reverse: 5′-GACCTCGGGCCACTGTAGC-3′
CCNE1Forward: 5′-TACCCAAACTCAACGTGCAA-3′104
Reverse: 5′-CATGATTTTCCAGACTTCCTCTC-3′

To obtain data regarding the relative expression of PRMT2 variants in breast carcinoma, 61 clinical breast cancer samples were analyzed by real-time PCR. As shown in Fig. 10C, in absolute terms, breast cancer tissues expressed PRMT2 variant mRNA levels similar to, but less than, that of the wild type. In addition, all PRMT2 variant expression in breast cancer tissues is higher than that in normal tissues (P < 0.01), similar to that of wild-type PRMT2 (P < 0.001). Except for PRMT2β, PRMT2α and PRMT2γ mRNA expression was associated with tumor ERα expression in these specimens (P < 0.05), similar to that of wild-type PRMT2 (P < 0.001). These results show that their mRNA levels are consistent with the presence of ERα in the majority of the breast carcinoma specimens and in the breast carcinoma cell lines.

PRMT2 protein is overexpressed in human breast tumor specimens

To determine the expression of PRMT2 variants in breast cancer patients, immunohistochemistry was performed in 198 breast cancer specimens with the antibody against the N-terminus of PRMT2 that theoretically recognizes the wild-type PRMT2 and all variants. As shown in Fig. 11A, five major bands were detected using this antibody in the protein extracts from breast cancer tissues and adjacent normal breast tissues. Based on its predicted molecular mass, the upper band should correspond to the wild-type PRMT2 (48.5 kDa). The third band should correspond to PRMT2β (34.0 kDa) and PRMT2α (32.6 kDa), and the smallest band should correspond to PRMT2γ (25.8 kDa), consistent with the predicted size. The fourth band, slightly smaller than the variant PRMT2α, should be the variant PRMT2L2 (31.3 kDa), which results from alternative polyadenylation in our report [31]. The possibility that other unknown variants such as the second band may exist cannot be ruled out. As shown in Fig. 11B, in normal breast tissues PRMT2 was stained weakly or unstained, but stained strongly in ERα-positive breast carcinomas. As shown in Table 3, overall PRMT2 protein immunocytochemical positivity was found in four of 20 (20.0%) normal specimens, but 158 of 198 (79.8%) breast carcinoma specimens were positive for PRMT2 protein, significantly different from normal specimens (P = 0.000, chi-squared test). Furthermore, 126 of 142 (88.7%) ERα-positive specimens were positively stained, whereas low level staining for PRMT2 was observed in ERα-negative breast tumor tissues; the cases of positive staining were 32 of 56 (57.1%), significantly different from the ERα-positive carcinomas (P = 0.003, chi-squared test).

Figure 11.

 Abnormally enhanced PRMT2 expression in human breast cancer specimens. (A) Western blot. Distributions of endogenous PRMT2 variants in the total protein extracts were examined by western blotting using PRMT2 antibody. T, breast cancer tissues; N, adjacent normal breast tissues. (B) Immunohistochemistry of PRMT2 expression in normal breast tissues and cancer was conducted using PRMT2 antibody. Representative data from immunohistochemical studies of 198 breast cancer specimens are shown. PRMT2 was expressed at low levels in adjacent normal breast tissue (a), but showed uniformly high expression in ductal carcinoma in situ (b), medullary carcinoma (c), invasive ductal cancer (d), lobular cancer (e) and mucous carcinoma (f). Original magnification 400 ×.

Table 3.   Identification of PRMT2 protein in breast specimens using immunocytochemistry. Positive immunocytochemistry is defined as > 10% of epithelial cells staining; negative immunocytochemistry is defined as <10% of epithelial cells staining.
Clinical specimensNo.PRMT2 expressionPositive (%)
PositiveNegative
  1. a P = 0.003, statistically significantly different from ERα positive (chi-squared test). b P = 0.000, statistically significantly different from normal specimens (chi-squared test).

Normal2041620.0
ERa-positive carcinomas1421261688.7
ERa-negative carcinomas5632a2457.1
Total198158b4079.8

Discussion

Alternative pre-mRNA splicing is a sophisticated and ubiquitous nuclear process which plays an essential role in the regulation of many biological processes [41]. In this report, we identified and characterized three novel PRMT2 splice variants that were isolated from human breast cancer cells. Compared with wild-type PRMT2, they contain distinct C-terminal domains. The PRMT2γ mRNA is generated by the removal of all the four exons 7, 8, 9 and 10 through alternative splicing, leading to an in-frame deletion of 205 amino acids in the C-terminus. In contrast, PRMT2α is generated by alternative splicing of exons 8, 9 and 10, resulting in a deletion of 156 amino acids in the C-terminus and downstream shifting. Therefore, 12 new amino acids are added at the C-terminus. Unlike PRMT2α and PRMT2γ, PRMT2β lacks exons 7, 8 and 9, resulting in a deletion of 215 amino acids and a gain of 83 amino acids at the C terminus due to downstream shifting (Fig. 1).

PRMT2 family members affect transcriptional regulation through their effects on transcription coactivators, such as PRMT1 and CARM1 [42–44]. Since PRMT2 has been shown to be one of the interaction proteins of ERα and it enhances ERα transcriptional activity upon ligand binding [11], we next sought to investigate the relationship between PRMT2 variants and ERα. Although there are some domain deletions in its variants, the interactions between these variants and ERα have been found by glutathione S-transferase (GST) pull-down assay and co-immunoprecipitation, suggesting that that the intact III, post III and THW loop domain are not necessary for PRMT2–ERα interaction. Meyer et al. [13] reported that PRMT2 is also a coactivator of the AR and the progesterone receptor (PR). The interactions between PRMT2 variants and AR have been confirmed by co-immunoprecipitation in vivo (Fig. 6C). We speculate that perhaps PRMT2 variants cooperate with numerous nuclear hormone receptors and are cell type specific under certain conditions. Transient cotransfection assays demonstrated that all the PRMT2 variants increased ERE-LUC transcriptional activity when estrogen was used at 10 nm, and anti-estrogen chemicals 4-OHT and ICI 182,780 reversely inhibited the ERE-LUC activity. We also demonstrated that PRMT2 and its variants could enhance promoter activities of Snail and repress promoter activities of E-cadherin through ERα in the presence of E2, indicating a functional connection between PRMT2 and ERα signaling pathways. It is noteworthy that the ERE-LUC transcriptional activation induced by PRMT2α and PRMT2γ is slightly lower compared with that induced by the wild type; however, variant PRMT2β showed a lower effect on ERE-LUC transcription activation compared to that of the wild type PRMT2 with treatment of 10 nm E2, and the affinity to ERα of PRMT2β was also weaker compared with other variants either in vitro or in mammalian cells. This surprising observation led us to further explore the role of the PRMT2β C-terminus in transcriptional activation. Using a series of truncations, we found that either truncation PRMT2 (1–277) or PRMT2 (1–218) was able to bind to ERα and activate transcription. Therefore, the amino acids between 1 and 218 are essential for the full transcriptional activity of human PRMT2. It was found that the SH3 domain of PRMT2 is not necessary for PRMT2–ERα interaction [11]; thus, it could be inferred that the ERα-interacting region on PRMT2 was mapped to a region encompassing the amino acids 133–218 of PRMT2. The truncation PRMT2 (278–433) was not able to bind with recombinant human ERα as reported previously [11], but slightly increased the ERE-LUC transcriptional activity indicating that truncation PRMT2 (278–433) may recruit other proteins that help to mediate PRMT2 coactivator function. It was reported that truncation PRMT2 (271–433) was sufficient for the association with the AR [13], indicating that the binding of truncation PRMT2 (271–433) with proteins depends on different cellular contexts. The novel part of PRMT2β protein cannot directly interact with ERα but relatively suppresses transcription, suggesting that it may recruit other ligand-dependent proteins that help to suppress PRMT2 coactivator function. The significance of this observation remains to be explored.

A series of reports show that PRMT2 is clearly involved in lung function, the inflammatory response, apoptosis promotion, Wnt signaling and leptin signaling regulation [14,45–48], suggesting that PRMT2 plays diverse roles in transcriptional regulation through different mechanisms depending on its binding partner. Meyer et al. [13] reported that under androgen-free conditions both AR and PRMT2 are confined to the cytoplasm of HepG2 cells, whereas in the presence of androgen both proteins colocalize and translocate into the nucleus. Our findings of confocal microscopy (Figs 2 and 3) showed that both PRMT2α and PRMT2γ were predominantly localized to the nucleus excluding the nucleoli and a weak fluorescence was detected in the cytosol of MCF7 and HepG2 cells in physiological conditions, as well as the distribution of wild-type PRMT2; PRMT2β was distributed in the nucleus and the cytoplasm, suggesting that the N-terminus of PRMT2 contributes to its nuclear localization, and the alternatively spliced C-terminus of PRMT2 can directly influence its subcellular localization. The colocalization of PRMT2 variants and ERα/AR inside mammalian cells, as indicated in the merged image (yellow) consistent with the results of interactions (Fig. 6), suggests that PRMT2 variants associate with various nuclear receptors and may play a role in cellular physiological procedures. The different subcellular localizations of these PRMT2 variants is indicative of potentially different functional properties, as is the case for PRMT1 in breast cancer [49], and the significance of this distinction remains to be explored.

Estrogens elicit proliferative responses in breast cancer cells through ERα-mediated transcriptional mechanisms [50]. PRMT2 has been implicated in the positive regulation of ERα-mediated gene activation in response to estrogen signaling [11]. Thus, it may be involved in the control of cell proliferation through the activation of key cell cycle genes. In this study, we describe a specific role for PRMT2 in regulating the cell cycle by inducing the expression of E2F1 in MCF7 cells following E2 treatment. We show that PRMT2 silencing could enhance E2-induced proliferation in ERα-positive breast cancer cells. PRMT2 silencing is required for the E2-mediated activation of E2F1 and for the induction of E2F1 target genes, including CCNE1 and CCNA1. These data are in agreement with a report showing that PRMT2 directly represses E2F activity in an RB-dependent manner and increases G1-S phase progression [12]. We speculated that since there are many ERα binding proteins which positively regulate ERα function, the loss of PRMT2 protein might be compensated by other coregulators such as E2F1. This pathway may be involved in the progression of ERα-positive breast cancer and may reflect the self-maintaining nature of ERα signaling in estrogen-sensitive cells.

PRMT2 has been found to be aberrantly expressed in skeletal muscle, ovary, prostate and fetal brain [10,13]. In this study, using several breast cancer cell lines, we provide evidence that PRMT2 is positively correlated with the expression of ERα. This link, initially derived from cell culture systems, is also supported by data from clinical breast carcinomas (Figs 10 and 11). Statistical analysis showed that the expressions of overall PRMT2 in breast malignant tissues were significantly higher than those in adjacent normal breast tissues, and their expressions were positively correlated to ERα status, indicating a potential role of PRMT2 in the ERα-dependent signaling pathway. Still, in parts of ERα-negative tumors their expressions were relatively higher, suggesting that their expressions in human tumors are possibly also associated with other nuclear receptors. Further work will be required to address the functional significance of PRMT2 overexpression in breast cancers. The results of immunohistochemistry show that the expression of PRMT2 was predominantly localized to the tumor cytoplasm of cells, with little distributed to the tumor nucleus (Fig. 11B), not as described above in mammalian cells (Figs 2 and 3). These discrepancies indicate that intricate regulation mechanisms are involved in the translocation of PRMT2 and might be related to disturbance of estrogen metabolism in breast cancer patients. In addition, novel variants possibly appear in human breast tumor specimens but are not observed in the cancer cell lines, suggesting the contribution from distinct regulatory mechanisms at the translational or post-translational level in human breast tumor tissues. The possibility remains that these variants may be expressed at higher levels in breast tumor tissues not examined in cancer cell lines or may be induced in response to specific stimuli or metabolic conditions [11,13]. The significance of this observation remains to be explored.

To date, increasing amounts of evidence strongly suggest that the alternative pre-mRNA splicing of some PRMTs, such as PRMT1, CARM1 and PRMT7, occurs frequently in human cancer cells [49,51,52]. They are considered coactivators of nuclear receptors and are overexpressed in hormone-dependent cancers, including breast cancer [1]. In this study, three novel PRMT2 variants were identified; they showed a tissue-specific expression pattern, and their levels changed in different cells and tissues. All these results indicate that human PRMT2 gene expression may be regulated at the post-transcriptional level by use of alternative splicing. However, further studies are still needed to explore the potential role of PRMT2 and its variants in the progression of breast cancer.

Experimental procedures

Identification and molecular cloning of PRMT2 splice variants

To identify novel splice variants of PRMT2 in breast cancer, we designed oligonucleotide primers based on the coding sequence of mRNA (NM_206962). The primers used were as follows: PRMT2-F, 5′-GCCTAAGAGAACTATGGCAACAT-3′; PRMT2-R, 5′-AAATAAAGCATCAACTGTCATCTCC-3′. PCR reactions were initiated with incubation at 94 °C for 2 min, followed by 30 cycles of 94 °C, 30 s; 55 °C, 30 s; and 72 °C, 1.5 min. Reactions were finished with a 72 °C, 7-min extension. Products were subcloned into pGEM-T Easy vector (Promega, Madison, WI, USA). Twenty clones were selected at random and subjected to DNA sequencing.

Cell culture

All cells were obtained from the American Type Culture Collection. MCF7, BT474, SK-BR-3, HepG2 and 293T were routinely cultured in DMEM with 10% fetal bovine serum (FBS). T47D and ZR-75-1 were cultured in RPMI 1640 with 10% FBS. All cells were maintained in 5% CO2 at 37 °C. MDA-MB-231 and MDA-MB-453 were cultured in Leibovitz’s L15 medium, 10% FBS and were maintained in a humidified atmosphere at 37 °C without additional CO2.

Plasmid construction

pcDNA3-ERα, pRL-TK, pGEX-5T and the ERE-LUC were kindly provided by P. Qinong Ye (Beijing Institute of Biotechnology, China). Full-length PRMT2 variants were cloned from T47D cells. Expression plasmids for GFP were constructed by inserting the cDNA into pcDNA3.1/NT–GFP-TOPO (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s protocol. Various truncations of PRMT2 (1–218) (N-terminus conserved in wild-type PRMT2 and all variants), PRMT2 (1–277) (N-terminus retained only in wild-type PRMT2 and PRMT2α), PRMT2 (278–433) (C-terminus of PRMT2) or PRMT2β (△1–218) (novel 83 amino acids at the C-terminus of PRMT2β) were generated by PCR amplification. 5′-BamHI site (GGATCC) and 3′-XhoI site (CTCGAG) were introduced into BamHI and XhoI sites of pGEX-5T vector. Expression plasmids for V5-tagged proteins were constructed by inserting the cDNA into the pcDNA3.2/V5/GW/D-TOPO vector (Invitrogen) following the manufacturer’s protocol. Snail and E-cadherin promoter were subcloned into PGL4.10-basic plasmid (Promega, Madison, WI, USA) at KpnI and HindIII sites to drive luciferase expression. The amplification primer pairs were as follows: Snail, 5′-GAAGGTACCTCAGGTGACCCGCCTCTTAACG-3′, 5′-GACAAGCTTCGCAGAAGAACCACTCGCTAGG-3′ for the 864-bp fragment; E-cadherin, 5′-GACGGTACCAAAGAACTCAGCCAAGTG-3′, 5′-GCTAAGCTTGCTGGAGTCTGAACTGAC-3′ for the 363 bp fragment, All constructs were sequenced to ensure proper in-frame ligation and Taq polymerase fidelity.

Western blot analysis

Total cell and tissue lysates were lysed on ice for 30 min. Soluble protein (30 μg) was separated on 12% SDS/PAGE gels, and western blot was conducted using PRMT2 antibodies (1 : 1000, ARP40196_T100; Aviva Systems Biology, Beijing, China). The same membrane was reprobed for GAPDH, which served as the loading control for the experiment.

Confocal microscopy

Transfections were performed using the Lipofectamine Plus transfection reagent (Invitrogen) according to the manufacturer’s instructions. Plasmid constructs expressing the different PRMT2 variants fused to an N-terminal GFP tag were generated and used to transiently transfect the MCF7 cells or HepG2 cells grown on glass coverslips. At 48 h post-transfection, the cells were fixed in paraformaldehyde, permeabilized with Triton X-100 and incubated with Cy3-conjugated ERα antibody or AR antibody. The cells were then stained with 4,6-diamidino-2-phenylindole (DAPI) and viewed under a Zeiss LSM 510 confocal microscope (Carl Zeiss, Oberkochem, Germany); images were acquired from typical cells using a × 63 oil-immersion lens.

Mammalian cell transfection and dual luciferase reporter assays

MCF7 cells were cultured as described above. For transfection, cells were seeded in 12-well plates containing phenol-red-free DMEM (Invitrogen) supplemented with 10% FBS (charcoal/dextran treated FBS, Hyclone). The cells were transfected using Lipofectamine 2000 (Invitrogen) with 0.2 μg of ERE-LUC reporter plasmid, Snail promoter, or E-cadherin promoter respectively, 25 ng of pRL-TK Renilla luciferase vector and 0.1–0.5 μg of PRMT2 or its variants and the respective empty vector was used to adjust the total amount of DNA. After treatment with 10 nm E2, 100 nm 4-OHT or 100 nm ICI 182,780 for 24 h, the cells were harvested and firefly luciferase and Renilla luciferase assays were performed using the Promega Dual-Luciferase Reporter Assay System. The activities of the luciferase constructs were normalized to the Renilla luciferase activity of the pRL-TK construct. All transfections were done in triplicate and repeated at least thrice.

Protein purification and GST pull-down assay

GST and GST fusion proteins were expressed in Escherichia coli BL-21 cells by induction with a final concentration of 0.8 mm isopropyl-β-d-thiogalactopyranoside. Cells were lysed by sonication in 10 mL of 1 × phosphate-buffered saline (NaCl/Pi) supplemented with complete protease inhibitor tablets (Roche Applied Science). GST fusion proteins with the PRMT2 variants or their truncations were purified using glutathione-agarose beads (Sigma). The recombinant human ERα proteins (Invitrogen, P2187) were mixed with 10 mg of GST derivatives bound to glutathione-agarose beads in 0.5 mL of binding buffer (50 mm Tris/HCl pH 7.5, 150 mm NaCl, 1 mm EDTA, 0.3 mm dithiothreitol, 0.1% NP-40 and protease inhibitor tablets from Roche) and supplemented with 10 nm E2. The binding reaction was performed at 4 °C for 3 h and the beads were subsequently washed four times with the washing buffer (the same as the binding buffer), 30 min each time. The beads were eluted by boiling in SDS sample buffer and analyzed by SDS/PAGE, and ERα was analyzed by immunoblotting.

Co-immunoprecipitation

MCF7 cells or HepG2 cells were transfected with the indicated plasmids. At 24 h, cells were treated with 10 nm E2 or DHT for an additional 10 h and then lysed (50 mm Tris at pH 8.0, 500 mm NaCl, 0.5% NP-40, 1 mm dithiothreitol and protease inhibitor tablets from Roche). Five hundred nanograms of lysate were precleared with 50 μL protein A-Sepharose beads (Sigma) for 2 h at 4 °C. V5 antibody (Invitrogen) was then added and incubated overnight at 4 °C. One hundred microliters of protein A agarose were then added to the antibody/lysate mixture for another 2 h at 4 °C, and the beads were pelleted and washed thrice with lysis buffer. Bound proteins were eluted in SDS sample buffer, subjected to SDS/PAGE and analyzed using an ERα antibody (HC-20; Santa Cruz Biotech, Santa Cruz, CA, USA) or AR antibody (H-280; Santa Cruz Biotech, Santa Cruz, CA, USA).

RNA interference

MCF7 cells were transfected with small interfering RNA oligonucleotide duplexes (siRNA) at a final concentration of 50 nm using Lipofectamine 2000 (Invitrogen). Forty-eight hours after transfection, cells were stimulated and harvested for analysis. Cells had been hormone deprived for at least 3 days at the moment of stimulation. To silence PRMT2 expression, we designed the PRMT2 targeting sequence with siRNA design soft (Invitrogen). siPRMT2-N was targeted in exon 5, and siPRMT2-C was targeted in exon 9: siPRMT2-N, 5′-CCCTGACGGATAAAGTCAT-3′; siPRMT2-C, 5′-CCTGGTTTAGCGTCCACTT-3′.

Cell cycle distribution and growth assays

MCF7 cell cycle distribution was analyzed 24 h after stimulation with 10 nm E2 or vehicle using propidium iodide staining and flow cytometry. For growth analysis, MCF7 cells transfected with the indicated siRNAs were replated 48 h after transfection (seeding density, 3 × 104 cells per well) in phenol-red-free DMEM containing 5% charcoal-stripped serum and 10 nm E2. Cells were supplemented with fresh medium containing hormones every 48 h throughout the course of the experiment. The total cell number was quantified with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT) viability assay after 3 days of E2 stimulation and is expressed as percentage change relative to control (siControl transfected cells treated with vehicle).

Quantitative real-time RT-PCR

Isolated RNA was reverse transcribed using SuperScript II reverse transcriptase (Invitrogen). TaqMan quantitative PCRs and SYBR Green quantitative PCRs were carried out in duplicate for the target genes (Table 2) on a Light Cycler Real Time PCR Sequence Detector (Roche Diagnostics). To test that the probes were specific, we used as template plasmids DNA containing PRMT2 variants or GAPDH cDNA inserts. The PCR reactions were initiated with incubation at 95 °C for 10 s, followed by 40 cycles of 95 °C, 5 s and 60 °C, 20 s. For each reaction, standard curves for the reference gene were constructed by using six tenfold serial dilutions of plasmids. All samples were run in triplicate and reported as PRMT2 variant expression levels relative to GAPDH in the cell lines.

Immunohistochemistry of breast cancer tissues

Formalin-fixed paraffin-embedded sections of cancerous and noncancerous breasts (obtained from First Affiliated Hospital of South China University, Hengyang) were used to determine PRMT2 expression. The sections were deparaffinized, rehydrated and pretreated with 3% H2O2 for 20 min to quench endogenous peroxidase activity. The antibody-binding epitopes of the antigens were retrieved by microwave treatment, and the sections were then pre-incubated with 10% goat serum to block nonspecific binding. The rabbit polyclonal PRMT2 (ARP40196_T100; Aviva Systems Biology) antibody which theoretically recognizes all variants was used. Images of immunostained breast cancer tissue slides were assessed by two independent researchers not involved in any step of their preparation and without the clinical information regarding these slides. The results were classified into two categories, depending on the percentage of cells stained and/or the intensity of staining: −, 0–10% positive tumor cells; +, > 10% positive tumor cells.

Statistical analysis

All experiments were conducted in triplicate and the results are expressed as the mean ± SEM. Statistical analysis was done using spss, version 13.0 (SPSS, Inc., Chicago, IL, USA). Student’s t-test, Fisher’s exact test and the chi-squared test were applied to assess the statistical significance. P values <0.05 were considered significant.

Acknowledgements

We are grateful to Dr Deliang Cao (Southern Illinois University, USA) for critically reading the manuscript. This work was supported by the National Natural Science Foundation of PR China (Grant nos 30670993 and 30840052).

Ancillary