Identification of Protein Arginine Methyltransferase 2 as a Coactivator for Estrogen Receptor α*

In an attempt to isolate cofactors capable of influencing estrogen receptor α (ERα) transcriptional activity, we used yeast two-hybrid screening and identified protein arginine methyltransferase 2 (PRMT2) as a new ERα-binding protein. PRMT2 interacted directly with three ERα regions including AF-1, DNA binding domain, and hormone binding domain in a ligand-independent fashion. The ERα-interacting region on PRMT2 has been mapped to a region encompassing amino acids 133–275. PRMT2 also binds to ERβ, PR, TRβ, RARα, PPARγ, and RXRα in a ligand-independent manner. PRMT2 enhanced both ERα AF-1 and AF-2 transcriptional activity, and the potential methyltransferase activity of PRMT2 appeared pivotal for its coactivator function. In addition, PRMT2 enhanced PR, PPARγ, and RARα-mediated transactivation. Although PRMT2 was found to interact with two other coactivators, the steroid receptor coactivator-1 (SRC-1) and the peroxisome proliferator-activated receptor-interacting protein (PRIP), no synergistic enhancement of ERα transcriptional activity was observed when PRMT2 was coexpressed with either PRIP or SRC-1. In this respect PRMT2 differs from coactivators PRMT1 and CARM1 (coactivator-associated arginine methyltransferase). These results suggest that PRMT2 is a novel ERα coactivator.

The estrogen receptor (ER) 1 is a transcription factor that belongs to the nuclear receptor superfamily (1,2). Upon estrogen binding, ER regulates the transcription of specific target genes by binding to specific DNA response elements referred to as estrogen response elements (EREs) in their promoters or by interacting with other transcription factors such as Jun and Fos (1,2). In addition to hormone-mediated activation, ER is also activated by growth factors including epidermal growth factor and insulin-like growth factor-1 probably through phosphorylation (3,4). ER contains two transcriptional activation function (AF) domains: AF-1 located in the N terminus and the ligand-dependent AF-2 located in the ligand binding domain (5). The ability of AF-1 and AF-2 to activate transcription varies according to the promoter context and the cell type (6). There are two isoforms of estrogen receptors, namely ER␣ and ER␤ (7). ER␣ and ER␤ recognize identical DNA elements and have similar affinity for a certain estrogen, but exhibit distinct tissue distribution (7). Evidence provided by gene knock experiments indicates that ER␣ is the receptor responsible for the estrogen-induced growth of mammary gland and the reproductive tract (8).
The precise mechanism by which ER modulates cell-and gene-specific transcription is not fully understood. Recent evidence suggests that ER activates transcription by recruiting coactivators that appear to act by modifying chromatin structure or facilitating the formation of transcriptional initiation complexes (9,10). Among a growing list of cofactors that regulate nuclear receptors, including ER, are the well studied coactivators of the SRC-1 family (9), CREB-binding protein (CBP/ p300) (11,12), and PBP (13). PBP is a component of the thyroid hormone receptor-associated protein (TRAP)/vitamin D 3 receptor-interacting protein (DRIP) complexes (14 -16). Both SRC-1 and CBP/p300 have intrinsic histone acetyltransferase activity and recruit other acetyltransferases (17)(18)(19)(20). The acetylation of histone results in the modification of chromatin and increases the access of the DNA to other components of transcription apparatus. The multiprotein TRAP/DRIP complexes exhibit no intrinsic histone acetyltransferase activity and appear to function through the direct interaction with general transcriptional machinery (15,16). The observation that certain coactivators such as SRC-3 (AIB1; ACTR, p/CIP, RAC3) (21-24), AIB3 (PRIP, ASC2, RAP250, NRC, TRBP) (25)(26)(27)(28)(29), and PBP (30) are amplified and overexpressed in some breast cancers underscores the importance of nuclear receptor coactivators in transcriptional activation and also points to their possible role in neoplastic conversion.
Post-translational modification of proteins by arginine methylation has recently been implicated in a variety of cellular processes including nuclear receptor transcriptional regulation (31). Among the five members of protein arginine methyltransferases (PRMTs) identified thus far based on protein sequences, PRMT1 is the first identified and the predominant PMRT in mammalian cells (32). PRMT1 has been shown to interact with SRC-2 (GRIP1) and enhance the nuclear receptor transactivation function (33). Coactivator-associated arginine methyltransferase 1 (CARM1)/PRMT4 was identified by its interaction with nuclear receptor coactivator SRC-2 (GRIP1) (34). PRMT1 and CARM1 are able to methylate the histones H4 and H3, respectively, suggesting their role in modulating the chromatin structure (34,35). In addition, CARM1 also methylates the CBP/p300, which disables the interaction between CREB and CBP/p300 and blocks the CREB activation (36). PRMT2 was isolated based on its sequence similarity with PRMT1 (37). So far no methyltransferase activity has been revealed for PRMT2 (37). Here we report the identification of PRMT2 as a new ER␣-binding protein through yeast two-hybrid screening. We now demonstrate that PRMT2 binds to ER␣ directly and also enhanced both its AF-1 and AF-2 transcriptional activity. We also demonstrate that the potential methyltransferase activity was pivotal for PRMT2 coactivator function. These results suggest that PRMT2 is a new ER␣ coactivator.
Yeast Two-hybrid Screening-Yeast two-hybrid screening was performed using the matchmaker two-hybrid system kit (CLONTECH). Briefly, the yeast strain HF7C was cotransformed with a human matchmaker mammary gland cDNA expression library and pGBKT7-ER␣. The positive clones were selected by their growth in medium lacking histidine and the expression of ␤-galactosidase in the presence of 1 ϫ 10 Ϫ7 M of 17␤-estradiol.
Quantitative ␤-Galactosidase Assays-Appropriate plasmids were cotransformed into yeast strain HF7C, plated on selective media plates in the presence or absence of 10 Ϫ7 M 17␤-estradiol, and then incubated for 4 days at 30°C. Ten colonies from each plate were suspended in 150 l of buffer Z (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 1 mM MgSO 4 , 35 mM 2-mercaptoethanol). An equal number of cells in suspension was collected by centrifugation, and ␤-galactosidase activity was determined (Galacto-light kit, Tropix, Bedford, MA). Three independent assays were performed.
GST Pull-down Assays-The GST alone and GST fusion proteins were produced in Escherichia coli BL21 and bound to glutathione-Sepharose beads according to the manufacturer's instructions (Amersham Biosciences). In vitro translation was performed using rabbit reticulocyte lysate (Promega) and labeled with [ 35 S]methionine. In GST pull-down assays, a 25-l aliquot of GST fusion protein loaded on glutathione-Sepharose beads was incubated with 5 l of [ 35   Cell Culture and Transfection-CV-1 cells (1 ϫ 10 5 ) were plated in 6-well plates and cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum for 24 h before transfection. Cells were transfected for 5 h with 1.25 g of luciferase reporter DNA, 20 ng of plasmid expressing the receptor, and 1.25 g of appropriate expression plasmid DNA or as indicated in the figure legends using Lipo-fectAMINE 2000 (Invitrogen); 0.1 g of ␤-galactosidase expression vector pCMV␤ (CLONTECH) DNA was always included as an internal control. Cell extracts were prepared 24 h after transfection and assayed for luciferase and ␤-galactosidase activities (Tropix). Three independent transfections were performed for each assay.

Isolation of PRMT2 as an ER␣-binding Protein by Two-
hybrid Screening-Using full-length ER␣ as a bait in yeast two-hybrid system, we isolated from human mammary gland cDNA library a partial cDNA encoding PRMT2 (amino acids 10 -433). To examine the influence of estrogen on the interaction, pACT2-PRMT2, which was isolated by yeast two-hybrid screening and expressed as fusion protein between GAL4 activation domain and PRMT2 (amino acods 10 -433), or pACT2 was cotransformed with PGBKT7-ER␣ expressing fusion protein between GAL4 DNA binding domain and ER␣ or PG-BTKT7 into yeast HF7C. The ␤-galactosidase activity was measured as an indication of the relative strength of interaction in the presence or absence of ligand. In the absence of ligand, we observed an interaction between ER␣ and PRMT2 that resulted in a ϳ40-fold increase in the ␤-galactosidase activity (Fig. 1). The presence of the ligand estrogen did not significantly affect the interaction between PRMT2 and ER␣ (Fig. 1).
Interaction of PRMT2 with ER␣ and Other Nuclear Receptors in Vitro-The direct interaction between PRMT2 and ER␣ was further tested by in vitro GST binding assay. The immobilized GST-PRMT2, but not GST alone, retained [ 35 S]methioninelabeled ER␣ both in the presence and absence of estrogen (Fig.  2). Moreover, PRMT2 also showed the ligand-independent in- teraction with ER␤, PR, TR␤, RAR␣, PPAR␥, and RXR␣ (Fig. 2).
To determine which region of ER␣ binds to PRMT2, a GST pull-down assay was performed using fusion proteins between GST and different regions of ER␣. As shown in Fig. 3A, PRMT2 bound to the AF1 region, DNA binding domain, and hormone binding domain but not to the hinge region. The binding to ER␣ hormone binding domain was ligand-independent. The interaction between PRMT2 and ER␣ AF-1 region or DNA binding domain is stronger than that between PRMT2 and ER␣ hormone binding domain.
PRMT2 contains a Src homology 3 (SH3) domain that binds to proteins with a proline-rich motif and plays a pivotal role in a wide variety of biological processes (38). A GST pull-down assay revealed that PRMT2 with a SH3 domain deletion was still able to bind to the GST-ER␣ fusion protein but not GST alone. Therefore, this domain is not considered necessary for PRMT2 and ER␣ interaction (Fig. 3B). The region of PRMT2 that interacts with ER␣ was further defined by GST pull-down assay using different truncated PRMT2 fragments. A fragment from amino acid 133 to 275 was found to interact with ER␣ (Fig. 3C).
PRMT2 Interacts with ER␣ in Vivo-The potential interaction between PRMT2 and ER␣ in the intact cell was examined by coexpressing ER␣ and FLAG-tagged PRMT2 in COS-7 cells followed by immunoprecipitation and Western blot analysis. As shown in Fig. 4, PRMT2 interacts with ER␣ both in the presence and absence of estrogen.
Interaction of PRMT2 with PRIP, SRC-1, and with PRMT2 Itself-The potential interaction between PRMT2 and other known nuclear receptor coactivators was investigated using the GST pull-down assay. We detected the interaction of PRMT2 with PRIP and SRC-1 (Fig. 5). No interaction was observed between PRMT2 and PBP or PRMT2 and CBP (Fig. 5).
The methyltransferase PRMT1 is able to form homodimer or homooligomers (39,40). A GST pull-down assay was performed to see if PRMT2 exhibits this property. GST and PRMT2 fusion protein but not GST alone retained [S 35 ]methionine-labeled PRMT2 suggesting that PRMT2 is capable of forming homodimer or homooligomers (Fig. 6) PRMT2 Binds S-Adenosylmethionine-PRMT2 was initially isolated by its protein sequence similarity to other PRMTs and so far no methyltransferase activity has been revealed. Using bacterially expressed GST-PRMT2 fusion protein, we did not demonstrate that PRMT2 was capable of methylating histone and ER␣ (data not shown). We then tested the ability of PRMT2 to bind the methyl donor S-adenosylmethionine by a filter binding assay. Just like PRMT1, PRMT2 was found to be able to bind S-adenosylmethionine, whereas PRMT2 with point mutation in the S-adenosylmethionine binding motif (41) failed to bind S-adenosylmethionine (Fig. 7).
PRMT2 Potentiates ER␣ Transcriptional Activity and Its Potential Methyltransferase Activity Is Pivotal While Its SH3 Domain Is Dispensable for This Function-Having established that PRMT2 is an ER␣-binding protein, we investigated the effect of increased levels of PRMT2 upon ER␣ transcriptional activity in CV-1 cells. The luciferase activity expressed from ERE-TK-LUC that contains one copy of ERE serves as the indicator of the ER␣ transcriptional activity. Expression of PRMT2 increased the estrogen-dependent transcription of luciferase gene by about 8-fold with minimal effort on basal transcription, which provided evidence that PRMT2 acts as a coactivator for ER␣ (Fig. 8). However, the mutated PRMT2 that was incapable of binding S-adenosylmethionine enhanced the ER␣ transcriptional activity by about 2.5-fold, which is much less than the 8-fold obtained with wild-type PRMT2 indicating the importance of the potential methyltransferase activity for the role of PRMT2 as a coactivator (Fig. 8). On the other hand, PRMT2 with deletion of the SH3 domain increased ER␣ activity to the same extent as that for wild-type PRMT2, indicating that the SH3 domain is dispensable for its coactivator function (Fig. 8).
PRMT2 Increases Both AF-1 and AF-2 Transcriptional Activity of ER␣-As ER␣ contains the autonomous activation domain AF-1 and ligand-dependent activation domain AF-2, we further examined the effect of increased expression of PRMT2 on their individual activities. The AF-1 (1-184) and AF-2 (251-595) were fused to GAL4 DNA binding domain, respectively, and then cotransfected with GAL4 responsive element-directed luciferase as the reporter gene. In comparison with the GAL4 DNA binding domain alone, AF-1 increased the luciferase activity by about 3-fold. The expression of PRMT2 further increased AF-1-mediated luciferase expression by about 4.5-fold (Fig. 9A). Therefore, PRMT2 is able to enhance the ER␣ AF-1 activity. Just as other nuclear receptors show transcription repression in the absence of their corresponding ligands, the hormone-dependent AF-2 slightly decreased the luciferase activity without estrogen over the control. The addition of estrogen increased the AF-2-mediated luciferase expression by about 3-fold, which is further enhanced by coexpression of PRMT2 by about 5-fold, demonstrating that PRMT2 also potentiates the AF-2 activity (Fig. 9B).
No Synergistic Enhancement of ER␣ Activity by Coexpression of PRMT2 and SRC-1 or PRIP-Given that PRMT2 binds to SRC-1 and PRIP, we sought to determine whether there was synergistic enhancement of ER␣ activity by PRMT2 and SRC-1 or PRIP. In transient transfection assay with ER␣ and its reporter gene, PRMT2, SRC-1, and PRIP all enhanced the expression of reporter gene to different levels (Fig. 10). When PRMT2 was cotransfected with either SRC-1 or PRIP, the expression of the reporter gene luciferase was modestly de-creased in comparison with PRMT2 alone (Fig. 10). Therefore, there appears no synergistic activation when PRMT2 are coexpressed with either PRIP or SRC-1.
PRMT2 Also Enhances PR, PPAR␥, and RAR␣-mediated Transactivation-To investigate whether PRMT2 acts as a coactivator for other nuclear receptors, transient transfection assays were performed with PR, PPAR␥, and RAR␣ responsive element-directed luciferase gene. PRMT2 enhanced the PR ligand-dependent transcriptional activity by about 8-fold. In comparison with ER␣ and PR, PRMT2, which also increased the ligand-dependent PPAR␥ and RAR␣ transactivation by about 5-and 4.5-fold, respectively, showed less effect on PPAR␥ and RAR␣ transactivation (Fig. 11).
PRMT2 Contains No Intrinsic Transcriptional Activity-In an effort to define the mechanism by which PRMT2 acts as a coactivator, we tested if PRMT2 contains intrinsic transcriptional activity similar to that reported with other coactivators such as SRC-1 (42). PRMT2 was linked to the GAL4 DNA binding domain and transfected into CV-1 cells along with GAL4 responsive element-directed reporter gene luciferase. In comparison with GAL4 DNA binding domain alone, GAL4-PRMT2 fusion protein produced no additional activity whereas GAL4-SRC-1 fusion protein increased the luciferase activity by about 7-fold and served as a positive control (Fig. 12). Therefore, PRMT2 does not have intrinsic transcriptional activity. DISCUSSION Using a yeast two-hybrid system with ER␣ as bait to screen a human mammary gland cDNA library, we isolated PRMT2 as a new ER␣-interacting protein. The interaction between PRMT2 and ER␣ was confirmed by in vitro binding and in vivo immunoprecipitation assay. A transient transfection assay demonstrated that PRMT2 increased the ER␣ transcriptional activity. In agreement with the finding that PRMT2 bound to both ER␣ AF-1 domain and the hormone binding domain, PRMT2 enhanced both ER␣ AF-1 and AF-2-mediated transactivation. These results established that PRMT2 is a coactivator of ER␣. However, unlike other coactivators such as SRC-1 family and CBP that show ligand-dependent binding to the nuclear receptors, PRMT2 binds to ER␣ both in the presence and absence of estrogen but enhances the ER␣ activity only with the estrogen. It appears that the interaction between ER␣ and PRMT2 is not enough for ER␣ transcriptional activation, which occurs only after the binding of estrogen resulting in most probably the recruitment of other ligand-dependent coactivators.
Two types of PRMT activities have been identified in mammalian cells (31). Type 1 PRMT enzymes including PRMT1, PRMT3, and CARM1 catalyze the formation of monomethylarginine and asymmetric dimethylarginine. Type 2 PRMT enzymes catalyze the formation of monomethylarginine and symmetric dimethylarginine. PRMT5/JBP1 is the only type II enzyme identified so far (43). Based on the protein sequence, PRMT2 was identified as a methyltransferase most probably belonging to type I enzyme, but so far its methyltranferase activity has not been identified (37). Although we demonstrated PRMT2 is capable of binding S-adenosylmethionine, we failed to detect any methyltransferase activity using bacterially expressed GST-PRMT2 fusion protein with substrates including histone and ER␣ (data not shown). A systematic approach to identify the substrates for PRMT2 will be required, and it is also possible that some modification such as phospho-rylation or some cofactor may be required for its activity. Nevertheless, the mutation in the conserved PRMT2 binding site for S-adenosylmethionine, which would abolish the potential methyltransferase activity, substantially diminished the PRMT2 coactivator function. The finding that PRMT2 does not have any intrinsic transcriptional activity favors the hypothesis that PRMT2 acts by modifying chromatin structure or the transcriptional apparatus through methylation. The elucidation of the substrates will be crucial for the understanding of PRMT2 coactivator function.
PRMT1 and CARM1 are two arginine methyltransferases that have been found to participate in the nuclear receptor transcriptional activation. Both PRMT1 and CARM1 interact with the carboxyl-terminal activation domain of coactivator GRIP1 and are able to methylate histones H4 and H3, respectively. PRMT1 or CARM1 enhance the nuclear receptor activity mildly by itself but substantially when coexpressed with GRIP1, suggesting that PRMT1 and CARM1 act as the secondary coactivators that are recruited by the first coactivator to modify the chromatin structure. Instead, PRMT2 strongly enhances the ER␣ transcriptional activity by direct interaction with ER␣. Although PRMT2 was found to interact with other coactivators SRC-1 and PRIP, no synergistic activation was found with coexpression of PRMT2 and PRIP or SRC-1. The coexpression of SRC-1 or PRIP even modestly decreased PRMT2 coactivation. Therefore, although PRMT2 is a protein arginine methyltransferase highly homologous to PRMT1 and CARM1, PRMT2 may have a very different mechanism by which it acts as a coactivator, possibly because it has different substrates involved in transcriptional activation.