MOLECULAR AND CELLULAR BIOLOGY, June 2002, p. 3621–3632 0270-7306/02/$04.00⫹0 DOI: 10.1128/MCB.22.11.3621–3632.2002 Copyright © 2002, American Society for Microbiology. All Rights Reserved.
Vol. 22, No. 11
Synergy among Nuclear Receptor Coactivators: Selective Requirement for Protein Methyltransferase and Acetyltransferase Activities Young-Ho Lee,1 Stephen S. Koh,2 Xing Zhang,3 Xiaodong Cheng,3 and Michael R. Stallcup1,2* Departments of Pathology2 and Biochemistry and Molecular Biology,1 University of Southern California, Los Angeles, California 90089, and Department of Biochemistry, Emory University School of Medicine, Atlanta, Georgia 303223 Received 14 September 2001/Returned for modification 12 November 2001/Accepted 25 February 2002
Hormone-activated nuclear receptors (NR) bind to specific regulatory DNA elements associated with their target genes and recruit coactivator proteins to remodel chromatin structure, recruit RNA polymerase, and activate transcription. The p160 coactivators (e.g., SRC-1, GRIP1, and ACTR) bind directly to activated NR and can recruit a variety of secondary coactivators. We have established a transient-transfection assay system under which the activity of various NR is highly or completely dependent on synergistic cooperation among three classes of coactivators: a p160 coactivator, the protein methyltransferase CARM1, and any of the three protein acetyltransferases, p300, CBP, or p/CAF. The three-coactivator functional synergy was only observed when low levels of NR were expressed and was highly or completely dependent on the methyltransferase activity of CARM1 and the acetyltransferase activity of p/CAF, but not the acetyltransferase activity of p300. Other members of the protein arginine methyltransferase family, which methylate different protein substrates than CARM1, could not substitute for CARM1 to act synergistically with p300 or p/CAF. A ternary complex of GRIP1, CARM1, and p300 or CBP was demonstrated in cultured mammalian cells, supporting a physiological role for the observed synergy. The transfection assay described here is a valuable new tool for investigating the mechanism of coactivator function and demonstrates the importance of multiple coactivators, including CARM1 and its specific protein methyltransferase activity, in transcriptional activation. GRIP1/TIF2/SRC-2, and ACTR/RAC3/pCIP/AIB1/TRAM1/ SRC-3 (10). Once the p160 coactivators are bound to the hormone-activated NR, two activation domains, AD1 and AD2, are important for the transmission of the activating signal to the transcription machinery (23). AD1, located between amino acids 1040 and 1120 in the 1,462-amino-acid GRIP1 polypeptide, is a binding site for p300 and CREB binding protein (CBP), related proteins which serve as coactivators for many different classes of transcriptional activators (5, 40). The abilities of CBP and p300 to acetylate histones and other proteins in the transcription initiation complex and their abilities to form stable complexes with several basal transcription factors suggest multiple mechanisms by which they help to mediate the transcriptional activation process (5, 6, 18). Another histone acetyltransferase, p300/CBP-associated factor (p/ CAF), also functions as an NR coactivator; but since p/CAF exists in a separate large protein complex and can bind in vitro directly to NRs, p160 coactivators, and p300/CBP, the precise mechanism by which it participates in the coactivator complex and activity is not clear (10). AD2, located in the C-terminal region of p160 coactivators, also contains a protein acetyltransferase activity (5, 37), but substrates which are efficiently acetylated by this domain have not been identified, and its importance in the coactivator function of p160 proteins has not been determined. However, coactivator-associated arginine methyltransferase 1 (CARM1), a member of the arginine-specific protein methyltransferase family, was recently shown to bind the AD2 region of p160 coactivators and to enhance gene activation by NRs in collaboration with p160 coactivators (4). Since CARM1 can methylate specific arginine residues in the N-terminal tail of histone H3 in vitro (4, 34), CARM1 may contribute to transcriptional
Transcriptional activator proteins regulate transcription of specific genes by binding to specific enhancer elements associated with the promoters of target genes and recruiting coactivator proteins to remodel chromatin structure and to recruit and activate the transcription initiation complex containing RNA polymerase II (8, 10, 25). The identities and functions of the coactivator proteins, especially among the nuclear receptor (NR) superfamily of transcriptional activators, have recently been the subject of intense study (10, 25). The NR family includes many well-known hormonedependent transcriptional activators (2, 24). Binding of the appropriate hormone alters NR conformation in a way that prevents binding of corepressors and promotes binding of coactivators, thus leading to transcriptional activation (10). To date, several dozen potential coactivators for NR have been identified (10, 25). The most well characterized coactivators exist as multisubunit complexes. The SWI/SNF complex is one of several related complexes which bind to NR and other transcriptional activators and remodel chromatin structure through an ATP-dependent mechanism. The p160 coactivators exist in complexes containing one or more protein (including histone) acetyltransferases. The thyroid receptor-associated proteins/vitamin D receptor interacting proteins (TRAP/DRIP) complex is also involved in the recruitment and/or activation of RNA polymerase II. The present study focuses further on the p160 coactivator complexes and their mechanisms of transcriptional activation. The three members of the p160 coactivator family are SRC1, * Corresponding author. Mailing address: Department of Pathology, HMR 301, University of Southern California, 2011 Zonal Ave., Los Angeles, CA 90089-9092. Phone: (323) 442-1289. Fax: (323) 442-3049. E-mail: [email protected]
LEE ET AL.
activation through methylation of histones and/or other proteins in the transcription initiation complex. CARM1 and p300 have both been shown individually to cooperate with p160 coactivators to enhance NR function (3, 20), but each binds to a different domain of p160 coactivators and each has a different proposed mechanism of signaling to the transcription machinery, i.e., protein methylation for CARM1 versus protein acetylation and contact with basal transcription factors for p300. Different types of posttranslational protein modifications may function cooperatively to help promote transcription initiation. Based upon these considerations, we investigated the potential for synergistic coactivator function between the protein methyltransferase CARM1 and the protein acetyltransferases p300 and p/CAF. We previously showed that GRIP1, CARM1, and p300 together could enhance estrogen receptor (ER) function in a more-than-additive manner (3). In the present study, we demonstrate that use of very low levels of NR expression vectors with the three coactivator expression vectors in transient-transfection assays produces a dramatic coactivator synergy characterized by a stringent requirement for a p160 coactivator, CARM1, and a coactivator of the acetyltransferase class. To investigate the physiological significance of this synergy, we tested whether GRIP1, CARM1, and p300 can form a ternary complex in cells. Given the proposed importance of the enzymatic activities of these proteins for their coactivator function, we also determined the effects on coactivator function caused by mutations which selectively eliminate the enzymatic activities. In the case of CARM1, design of a novel and more selective mutation was aided by the recent determination of the threedimensional structure of the highly conserved methyltransferase domain of one member of the protein arginine methyltransferase family, to which CARM1 belongs (42). As a further test of the specificity of the observed synergy, we also investigated whether other members of the protein arginine methyltransferase family, which methylate different protein substrates than CARM1, can substitute for CARM1 in this NR functional assay where multiple coactivators are required. MATERIALS AND METHODS Plasmids. The mammalian cell expression vector pSG5.HA (4), which has simian virus 40 and T7 promoters, was used to express proteins with an N-terminal hemagglutinin (HA) tag; plasmids encoding the following proteins were previously constructed in pSG5.HA as indicated: GRIP1, CARM1, CARM1(VLD) (4); GRIP1⌬AD1 and GRIP1⌬AD2 (23); GRIP1⌬AD1 plus ⌬AD2 (3); and rat protein arginine methyltransferase 1 (PRMT1) (17). New pSG5.HA expression vectors encoding the following proteins were constructed by inserting the appropriate cDNA coding region (reviewed in reference 4) into the indicated restriction enzyme sites of the vector: human PRMT2, EcoRIBamHI; rat PRMT3, MfeI-XhoI fragment inserted into EcoRI and XhoI sites; and yeast arginine methyltransferase 1 (RMT1), EcoRI-BamHI. The mutation E267Q in CARM1 was generated with the Quickchange site-directed mutagenesis kit (Stratagene), using pSG5.HA-CARM1 as the template. The following mammalian expression vectors with cytomegalovirus promoters were used to express p300, CBP, and p/CAF: pCX-p/CAF, pCX-p/CAF⌬579-608, pCX-p/ CAF⌬609-624, pCMV-p300, pCMV-p300⌬1603-1653 (29), and pcDNA3-CBP, kindly provided by T.-P. Yao (Duke University). The p300 vectors included an N-terminal Flag epitope tag. Mammalian expression vectors encoding NRs included pHE0 for human ER␣, pSVAR0 for human androgen receptor (AR), and pCMX.hTR␤1 for human thyroid hormone receptor (TR) ␤1 (4). The following luciferase-expressing reporter genes were described previously (4, 12): for AR, MMTV-LUC with the native mouse mammary tumor virus (MMTV) promoter; for TR, MMTV(TRE)LUC with a single thyroid hormone response element substituted for the native
MOL. CELL. BIOL. glucocorticoid response elements; for ER, MMTV(ERE)-LUC with a single estrogen response element (ERE) substituted for the native glucocorticoid response elements, EREII-LUC(GL45) containing a basal herpes simplex virus thymidine kinase (TK) promoter and two EREs, and TK-LUC containing the basal TK promoter without EREs; and for Gal4 DNA binding domain (DBD) fusion proteins, GK1. The vector pGAL-CBP8 encoding Gal4 DBD fused to full-length CBP was described previously (9). The vector pVP16.CARM1, encoding the activation domain VP16 fused to CARM1, was constructed by inserting an EcoRI-BglII fragment encoding CARM1 into pVP16 (Clontech). Cell culture and transient transfections. CV-1 cells (11) were maintained in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum. Approximately 20 h before transfection, 105 cells were seeded into each well of six-well dishes. The cells in each well were transfected with SuperFect Transfection Reagent (Qiagen) or Targefect (Targeting Systems) according to the manufacturer’s protocol; total DNA was adjusted to 2.0 g by addition of the empty vector pSG5.HA (4). After transfection, the cells were grown in medium supplemented with 5% charcoal-stripped fetal bovine serum (Gemini Bioproducts) for 40 h before harvest; where indicated, the medium was supplemented with 20 nM dihydrotestosterone (DHT) for AR, 20 nM estradiol (E2) for ER, or 20 nM 3,5,5⬘-triiodo-L-thyronine (T3) for TR during the last 30 h of growth. Luciferase assays were performed with the Promega Luciferase Assay kit, and luciferase activities are shown as the mean and deviation from the mean of two transfected sets. The results shown are representative of at least three independent experiments. Because some coactivators enhance the activities of so-called constitutive promoters two- to threefold, internal controls by cotransfection of constitutive ␤-galactosidase expression vectors were not used to normalize luciferase data. However, internal controls were used strategically to show that variation in transfection efficiency was not a factor in key results (data not shown). Coimmunoprecipitation and immunoblot analyses. Cos-7 cells (11) were grown in 100-mm-diameter dishes seeded with 106 cells and transfected with combinations of the coactivator expression plasmids pCMV-p300, pSG5.HACARM1, and pSG.HA-GRIP1 as indicated. At 40 h after transfections, cell extracts were prepared by lysing the cells in 1.0 ml of RIPA buffer (50 mM Tris·Cl [pH 8.0], 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate [SDS]) and were clarified by centrifugation for 15 min at the maximum speed of a microcentrifuge. A portion of the supernatant (40 l) was removed for direct immunoblot analysis of CARM1, using rat monoclonal antibody 3F10 against the HA epitope (Boehringer Mannheim) at 100 ng/ml as the primary antibody and horseradish peroxidase-conjugated anti-rat immunoglobulin G (sc-2006; Santa Cruz Biotechnology) at 160 ng/ml (1:2,500 dilution) as the secondary antibody. The remaining supernatant (800 l) was incubated with anti-Flag antibody (no. F3165; Sigma) (1 g) for 16 h at 4°C; the immunoconjugates were precipitated by incubation with protein G agarose (no. P7700; Sigma) (50 l of a 50% suspension), followed by centrifugation. Immunoprecipitates were resolved by SDS-polyacrylamide gel electrophoresis, and immunoblotting was performed as described previously (23) with antibody against HA epitope. Immunoprecipitation of CARM1 and methyltransferase assays. Vectors encoding wild-type or mutant HA-tagged CARM1 (2.5 g) were transfected into 106 Cos-7 cells. After 40 h, the cells were lysed with 1.0 ml of RIPA buffer. After centrifugation, 40 l of supernatant was removed for direct immunoblot analysis of CARM1 (using antibody against HA tag), and 800 l was used for immunoprecipitation of CARM1 as described above, using 1 g of antibody against HA tag and 50 l of protein G agarose suspension. The pellet from the immunoprecipitation was resuspended in 50 l of methyltransferase reaction mixture (20 mM Tris-HCl [pH 8.0], 200 mM NaCl, 0.4 mM EDTA, 8 g of unfractionated core histones [no. H9250;Sigma], 500 M S-adenosylmethionine) and incubated at 30°C for 2 h. The reaction products were resolved by SDS-polyacrylamide gel electrophoresis (10% acrylamide), and methylated histone H3 was visualized by immunoblot analysis with antiserum raised against a peptide representing histone H3 amino acids 11 to 21 with asymmetric dimethylarginine at position 17 (Upstate Biotechnology) (22).
RESULTS Dependence of NR function on three different coactivators: GRIP1, CARM1, and p300. To test coactivator function, expression vectors for various NRs were transfected into CV-1 cells (which have little or none of most NRs) along with expression vectors for one or more coactivators and a luciferase reporter gene containing one or more enhancer elements spe-
VOL. 22, 2002
cific for the NR being used. The transfected cells were treated or not treated with the appropriate hormone, and cell lysates were subsequently tested for luciferase activity as a measure of NR and coactivator function. When the amounts of various expression vectors transfected were carefully titrated, it was observed that the amount of NR expression vector strongly influenced the degree of cooperation observed among multiple coactivators. For six-well petri dishes (3.3-cm-diameter wells), transfection of 100 ng of ER expression vector enhanced reporter gene expression in the presence of estradiol (Fig. 1A, bars 1 and 2) but not in the absence of the hormone (Fig. 1D). The luciferase reporter gene had an MMTV promoter with a single ERE substituted for the glucocorticoid response elements found in the native promoter. Coexpression of GRIP1 with ER enhanced reporter gene expression severalfold (bar 3), and addition of CARM1 with GRIP1 caused a further enhancement (bar 4). In this experiment, coexpression of p300 with GRIP1 caused no additional stimulation (bar 5); in multiple experiments, we have found the effect of p300 in the presence of GRIP1 to be variable within the range of no effect to twofold enhancement. Addition of p300 in the presence of GRIP1 and CARM1 caused a decrease in the activity observed with GRIP1 plus CARM1 (bars 4 and 6); in multiple experiments, coexpression of p300 in this situation caused either no effect or a decrease in activity of as much as 50%. In contrast, when lower levels of ER expression vector (⬍10 ng) were transfected, reporter gene activity was much more highly dependent on the coexpression of multiple coactivators. In the absence of any cotransfected coactivator vectors, 10 ng of ER vector produced no increase in luciferase activity above the background observed with reporter gene alone, even in the presence of estradiol (Fig. 1A, bars 7 and 8). Addition of GRIP1, GRIP1 plus CARM1, or GRIP1 plus p300 caused no enhancement or a very modest enhancement of reporter gene activity compared with the activity observed with ER alone (bars 9 to 11). However, when ER was supplemented with all three coactivators, a synergistic enhancement of reporter gene function was observed (bar 12). Thus, three-coactivator synergy and a high degree of dependence on three coactivators were observed at low ER concentrations but not at higher ER concentrations. The selectivity of the coactivator effect was demonstrated by the failure of GRIP1, CARM1, and p300 to stimulate the activity of the basal TK promoter (Fig. 1C) and an RSV–␤-galactosidase reporter plasmid (data not shown). Similar synergistic enhancement of NR function by these three coactivators was observed when low levels of AR (Fig. 1B) and TR (Fig. 2C) expression vectors were used. Systematic variation of the level of AR expression vector revealed the relationship between synergy and NR levels (Fig. 1B). At low levels of AR expression vector (10 to 50 ng), very little reporter gene activity was observed unless GRIP1, CARM1, and p300 were all coexpressed with AR (Fig. 1B). However, at 70 ng of AR vector, GRIP1 alone, GRIP1 plus CARM1, or GRIP1 plus p300 produced substantial activity; a lower degree of threecoactivator synergy was still observed at 70 ng of AR vector, but the requirement for three coactivators was not as stringent. At higher AR levels, the three-coactivator synergy was not observed (data not shown). The observed coactivator synergy depended on the presence of the entire hormone response system in addition to the three
NUCLEAR RECEPTOR COACTIVATOR SYNERGY
coactivators. Coexpression of ER (1 ng), GRIP1, CARM1, and p300 in the presence of estradiol caused activation of a reporter gene controlled by a basal herpes simplex virus TK promoter and two EREs (Fig. 1C, bar 16). Omission of the hormone response elements (bars 1 to 6), ER (bars 7 to 11), or GRIP1 (bar 17) resulted in a complete loss of reporter gene stimulation. Omission of CARM1 or p300 or both resulted in a dramatic but not complete loss of reporter gene stimulation (bars 13 to 15). If CBP was substituted for p300, a similar level of synergy was also observed (Fig. 1D, lanes 11 to 16). The synergistic effect of coactivators was completely dependent on the presence of estradiol (Fig. 1D, compare lanes 2 to 7 with lanes 9 to 14). Similar results were obtained when the various controls in Fig. 1C and D were performed with AR and TR instead of ER (data not shown). Thus, at low levels of NR, efficient hormone-dependent activation of an NR-dependent reporter gene was almost completely dependent on the coexpression of three coactivators, and the synergistic effect of the three coactivators depended entirely on the presence of a hormone-activated NR bound to its cognate enhancer element. Furthermore, the coactivator effects of CARM1 and p300 depended entirely on the presence of GRIP1. Titration of the various coactivator expression vectors demonstrated that the level of reporter gene activity increased as the level of each coactivator was increased (data not shown). The amount of reporter gene activity observed from day to day when GRIP1 alone or GRIP1 plus one other coactivator was coexpressed with low levels of NR varied from none to significant, as seen in the various figures in this report (e.g., compare Fig. 1A, right, with C; also see later figures); but the activity observed with three coactivators was always synergistic, i.e., much higher than the additive effects of individual or pairs of coactivators. Ternary coactivator complex formation by GRIP1, CARM1, and p300. CARM1 binds to the AD2 domain of GRIP1, while p300 and CBP bind to the GRIP1 AD1 domain (4, 23). Since CARM1 and p300 can bind to different domains of GRIP1 and can function synergistically as coactivators, we tested whether these three proteins can exist as a ternary complex in mammalian cells. CARM1 with an HA epitope tag, p300 with a Flag epitope tag, and GRIP1 were coexpressed in Cos-7 cells by transient transfection. HA-CARM1 was expressed well in all of the transfections (Fig. 2A, bottom). When an antibody against the Flag tag was used for immunoprecipitation, coprecipitating CARM1 was detected by immunoblotting with an antibody against the HA tag (Fig. 2A, top, lanes 3 and 7). However, the coprecipitated HA-CARM1 signal was substantially reduced if vectors for p300 or GRIP1 were omitted from the transfection (lanes 2, 5, and 6). The requirement for GRIP1 expression indicates that CARM1 and p300 do not interact directly but associate indirectly through their contacts with GRIP1; thus, the results indicate a ternary complex of the three coactivators. When GRIP1 mutants lacking the AD2 domain or the AD1 and AD2 domains were substituted for wild-type GRIP1, the coprecipitation of CARM1 was also substantially reduced (lanes 9 and 10). These data are consistent with previous reports that CARM1 associates with GRIP1 through contact with the AD2 domain. Surprisingly, when only the AD1 domain was deleted from GRIP1, there was no decrease in the coprecipitation of CARM1 (lane 8). Immunoblots demonstrated that wild-type and mutant GRIP1 proteins were ex-
LEE ET AL.
MOL. CELL. BIOL.
pressed at similar levels (23) (data not shown). Since p300 did not bind directly to CARM1 in this experiment (lanes 2 and 6), and since CBP did not bind directly to CARM1 in a modified mammalian two-hybrid system (Fig. 2B), the integrity of the ternary complex with the GRIP1 AD1 deletion mutant suggests that p300 and CBP bind directly or indirectly to other regions of GRIP1 in addition to the AD1 domain. A modified mammalian two-hybrid system was used to demonstrate that GRIP1 and CARM1 can also form a ternary complex with CBP. Coexpression of a Gal4 DBD-CBP hybrid protein with a VP16-CARM1 hybrid protein failed to activate a luciferase reporter gene controlled by Gal4 response elements (Fig. 2B, bar 6). Coexpression of GRIP1 with the CBP and CARM1 hybrid proteins strongly enhanced reporter gene expression, indicating that CARM1 and p300 interact indirectly through GRIP1 (bar 7). Without VP16-CARM1, GRIP1 enhanced Gal-CBP activity only two- to threefold (data not shown). As in the coimmunoprecipitation experiment (Fig. 2A), the ternary complex did not form when a GRIP1 mutant lacking AD2 was substituted for wild-type GRIP1 (Fig. 2B, bars 9 and 10). However, deletion of GRIP1 AD1 failed to prevent the formation of the complex (bar 8), supporting the conclusion from the immunoprecipitation experiments that CBP and p300 can interact with GRIP1 through the AD1 domain and also through another yet-undefined domain. The importance of the GRIP1 AD2 domain for the formation of the ternary coactivator complex and for the functional synergy of these three coactivators was also demonstrated by employing the low-NR coactivator assay system. GRIP1, CARM1, and p300 synergistically enhanced the activation of a reporter gene by low levels of TR, and the activity was dependent on the presence of all three coactivators (Fig. 2C, bars 1 to 6). When a GRIP1 mutant lacking AD1 was substituted for wild-type GRIP1, the synergistic activity among the three coactivators was reduced but not eliminated (bars 7 to 10). However, deletion of AD2 or of AD1 and AD2 from GRIP1 eliminated reporter gene activity (bars 11 to 18). Similar results were obtained when ER was used instead of TR (data not shown). Thus, the AD2 domain of GRIP1 was necessary for the formation and synergistic coactivator function of the ternary CARM1-GRIP1-p300 complex; the AD1 domain may contribute to the total coactivator activity but was not essential for complex formation or synergy.
FIG. 1. Requirement for three coactivators (GRIP1, CARM1, and p300) at low levels of NR. CV-1 cells in each well of six-well culture dishes were transiently transfected with plasmids as indicated below. The transfected cells were grown with 20 nM E2 for ER or DHT for AR. Cell extracts were prepared and assayed for luciferase activity. (A) MMTV(ERE)-LUC reporter plasmid, 250 ng; 10 ng (low ER) or 100 ng (high ER) of ER expression vector; 250 ng of pSG5.HAGRIP1, 500 ng of pSG5.HA-CARM1, and 500 ng of pCMV-p300. (B) MMTV-LUC, 250 ng; 10 to 70 ng of AR expression vector; coactivator vectors as for panel A. The graph on the right shows the activity observed with 10 ng of AR vector on an expanded scale. (C) EREIILUC(GL45) or TK-LUC reporter plasmid, 250 ng; 1 ng of ER expression vector; coactivator vectors as for panel A. (D) EREIILUC(GL45), 250 ng; ER expression vector, 1 ng; 500 ng of each coactivator expression vector (GRIP1, CARM1, p300, and CBP). ⫹, present.
VOL. 22, 2002
NUCLEAR RECEPTOR COACTIVATOR SYNERGY
Requirement for the acetyltransferase activity of p/CAF, but not of p300, for coactivator synergy with CARM1 and GRIP1. p/CAF, p300, and CBP are protein (including histone) acetyltransferases and have been shown to act as coactivators for NRs (10). We therefore tested whether p/CAF could substitute for p300 or CBP to act synergistically with GRIP1 and CARM1 as coactivators for low levels of NRs. The coactivator combination of GRIP1, CARM1, and p300 produced the highest reporter gene activity among the various combinations of three coactivators tested with ER (Fig. 3A, bar 7). p/CAF also acted synergistically with GRIP1 and CARM1 (bar 8), but the activity was generally lower than that obtained with p300 (although no effort was made to ensure equal expression of p300 and p/CAF). As shown for the coactivator trio GRIP1, CARM1, and p300 (Fig. 2C), the AD2 domain of GRIP1, but not the AD1 domain, was essential for coactivator synergy among GRIP1, CARM1, and p/CAF (data not shown). The combination of GRIP1 plus the two acetyltransferases (Fig. 3A, bar 9) was modestly synergistic but less effective than the combinations of GRIP1 with CARM1 and one acetyltransferase (bars 7 to 8). The combination of CARM1, p300, and p/CAF was ineffective in the absence of GRIP1 (bar 10), presumably because GRIP1 is necessary for formation of the coactivator complex. A similar pattern of relative activities for these coactivator combinations was also observed for AR and TR (data not shown). The fact that a strong dependence on three coactivators can be demonstrated suggests that a dependence on a larger number of coactivators may be established by careful titration of the levels of NR and coactivator expression vectors. In experiments using TR and a suitable reporter gene, we observed that the combination of GRIP1, CARM1, p300, and p/CAF produced a higher level of activity than any combinations of three of these coactivators (Fig. 3B). To test whether the acetyltransferase activities of p300 and p/CAF are required for their synergistic coactivator function
FIG. 2. Ternary coactivator complex formation among GRIP1, CARM1 and p300/CBP. (A) Coimmunoprecipitation. Cos-7 cells in 100-mm-diameter dishes were transfected with coactivator expression vectors as indicated: 2.5 g of pCMV-p300 (produces p300 with a Flag tag); 2.5 g of pSG5.HA-CARM1; 2.5 g of pSG5.HA-GRIP1 (wild type [WT]) or the equivalent vector encoding GRIP1⌬AD1 (⌬AD1), GRIP1⌬AD2 (⌬AD2), or GRIP1⌬AD1 plus ⌬AD2 (⌬AD1/2). Com-
plexes containing p300 were immunoprecipitated from transfected cell extracts with anti-Flag antibody, and coprecipitated CARM1 was detected by immunoblotting with antibodies against the HA tag (top). To check CARM1 expression before immunoprecipitation, 5% of the transfected cell extract was directly tested by immunoblotting with antibodies against HA tag (bottom). Lanes 1 to 3 and 4 to 10 represent two independent experiments. The diagram indicates the proposed interaction sites among the three coactivators. IP, antibody used for immunoprecipitation; WB, antibody used for Western immunoblotting; IgG-H, immunoglobulin G heavy chain from the immunoprecipitation, which is recognized by the secondary antibody used in the immunoblot. (B) Modified mammalian two-hybrid system. CV-1 cells in six-well dishes were transfected with 250 ng of GK1 reporter plasmid controlled by an E1b basal promoter (E1bTATA) and Gal4 response elements (Gal4RE) and, as indicated, 250 ng of pGAL-CBP8, encoding Gal4 DBD fused to full-length CBP; 250 ng of pVP16.CARM1, encoding VP16 activation domain (VP16 AD) fused to CARM1; and 500 ng of pSG5.HA-GRIP1 or the corresponding vector expressing the ⌬AD1, ⌬AD2, or ⌬AD1 plus ⌬AD2 mutant of GRIP1. The luciferase (Luc) activity of the cell extract is shown. (C) Coactivator assays. CV1 cells in six-well dishes were transfected with 250 ng of MMTV(TRE)LUC reporter plasmid, 1 ng of TR expression vector, 500 ng of CARM1 vector, 500 ng of p300 vector, and 250 ng of the indicated wild-type or mutant GRIP1 vector. The cells were grown with 20 nM T3, and the luciferase activity of the cell extract was determined. ⫹, present.
LEE ET AL.
MOL. CELL. BIOL.
FIG. 3. Synergy among various combinations of three or four coactivators at low NR levels. CV-1 cells were transiently transfected as for Fig. 1 and grown with 20 nM E2 for ER or 20 nM T3 for TR. The plasmids used were 250 ng of MMTV(ERE)-LUC, 1 ng of ER vector, 250 ng of GRIP1 vector, 500 ng of CARM1 vector, 500 ng of p300 vector, and 500 ng of p/CAF vector (A) and 250 ng of MMTV(TRE)LUC, 5 ng of TR vector, and coactivator vectors as for panel A (B). ⫹, present.
with GRIP1 and CARM1, we tested previously defined acetyltransferase-negative deletion mutants of p300 and p/CAF (29) in our coactivator synergy assay. The p300 mutant del7 lacks amino acids 1603 to 1653 in the acetyltransferase domain; the p/CAF mutants MT1 and MT2 lack amino acids 579 to 608 and 609 to 624, respectively. As shown above, the combination of GRIP1, CARM1, and p/CAF acted synergistically to enhance the ability of low levels of AR or TR to activate their respective reporter genes in the presence of the appropriate hormone (Fig. 4A, bars 7). Substitution of either p/CAF mutant for the wild-type p/CAF caused a substantial reduction in the ability of p/CAF to cooperate with GRIP1 and CARM1 as coactivators for TR and completely eliminated the ability to cooperate with GRIP1 and CARM1 as coactivators for AR (compare bars 3 and 7 to 9). Immunoblot studies indicated that the mutant and wild-type p/CAF proteins were expressed at similar levels in transfected cells (data not shown). When other NRs were tested in this system, the p/CAF mutations caused severe loss of synergistic coactivator function for glucocorticoid receptor (like AR) but caused only a partial loss of function for ER (like TR) (data not shown). The reason for the different effects of these mutations on different NRs is not clear at this time but may reflect different coactivator requirements to mediate effectively the activities of different NRs. In contrast to p/CAF, elimination of the acetyltransferase activity of p300 caused little or no loss of the ability of p300 to cooperate with GRIP1 and CARM1 as coactivators for AR and TR (Fig. 4B). Histone acetylation assays, performed with wild-type and mutant p300 immunoprecipitated (via the Flag tag) from transiently transfected cells, confirmed the previously defined (29) lack of acetyltransferase activity in the mutant p300 protein (data not shown). When the amounts of wild-type and mutant p300 expression vectors were varied, their activities were indistinguishable over a broad range of vector amounts (data not shown). Thus, the acetyltransferase activity played an important role in the synergistic coactivator function of p/CAF, but the ability of p300 to cooperate with GRIP1 and CARM1 was independent of the acetyltransferase
FIG. 4. Role of protein acetyltransferase activities of p/CAF and p300 in coactivator synergy. CV-1 cells in six-well dishes were transfected with 1 ng of AR or TR expression vector, 250 ng of MMTVLUC or MMTV(TRE)-LUC, 250 ng of GRIP1 vector, 500 ng of CARM1 vector, and 500 ng of a vector encoding the indicated wildtype or mutant form of p/CAF (A) or p300 (B). The cells were grown with 20 nM DHT or T3, and cell extracts were assayed for luciferase activity. WT, wild-type p/CAF or p300; MT1, p/CAF⌬579-608; MT2, p/CAF⌬609-624; MT, p300⌬1603-1653; ⫹, present.
activity. These results indicate that p300 and p/CAF contributed by different mechanisms to the synergistic enhancement of NR function. Requirement for the methyltransferase activity of CARM1 for coactivator synergy with GRIP1 and p300. We previously demonstrated that substitution of alanine for three highly conserved residues (valine-leucine-aspartate, or VLD; amino acids 189 to 191) in CARM1 caused loss of methyltransferase activity and loss of the ability of CARM1 to cooperate with GRIP1 as a coactivator for NRs (4). The VLD sequence is conserved among members of the protein arginine methyltransferase family, indicating that it plays an important role in the structure or function of the methyltransferase domain. A recent three-dimensional crystal structure of the related protein arginine methyltransferase PRMT3 (42) indicates that the VLD sequence is important for tertiary structure of the core methyltransferase domain and thus may affect the overall structure as well as the methyltransferase activity of the protein. Based upon the same crystal structure, we designed a mutation which is predicted to eliminate the methyltransferase activity without affecting the overall structure of the protein. The crystal structure allowed the prediction of a binding site for the arginine residue to be methylated and suggested that two glutamate residues, which are highly conserved among the protein arginine methyltransferase family, form salt bridges with the two terminal guanidino group nitrogen atoms of the substrate arginine residue. In CARM1, these conserved glutamate residues are located at amino acid positions 258 and 267. In a new
VOL. 22, 2002
CARM1 mutant, we converted glutamate 267 to glutamine (E267Q). Wild-type CARM1 and both CARM1 mutants (VLD-toAAA and E267Q) were expressed at equivalent levels in transfected cells by immunoblot analysis with an antibody against the HA tag on the CARM1 proteins (Fig. 5A, top). The antibody against the HA tag was used to immunoprecipitate the CARM1 proteins from the transfected cell extracts in order to test their histone methyltransferase activities. In this assay, wild-type CARM1 methylated histone H3, but neither mutant CARM1 protein had any detectable histone methyltransferase activity (Fig. 5A, bottom). Methylated histone H3 was detected by immunoblot analysis with antibodies which recognize CARM1-methylated histone H3 but not unmethylated H3. Both mutants retained the ability to bind GRIP1 (4) (data not shown). The CARM1 wild-type and mutant proteins were tested for the ability to cooperate with GRIP1 and p300 as coactivators for NRs. When various amounts of CARM1 expression vectors were cotransfected with a low level (1 ng) of TR vector (conducive for three-coactivator synergy) and vectors encoding GRIP1 and p300, wild-type CARM1 strongly enhanced TRmediated reporter gene expression in cooperation with GRIP1 and p300. The E267Q mutant of CARM1 had dramatically reduced but detectable activity as a coactivator, while the VLD mutant of CARM1 had no detectable coactivator activity (Fig. 5B). Similar results were obtained when ER was substituted for TR (Fig. 5C, left). We also investigated the ability of the mutant and wild-type CARM1 proteins to act as coactivators in the presence of low versus high levels of ER. With 1 ng of ER vector, wild-type CARM1 cooperated effectively with GRIP1 and p300 to enhance ER function (Fig. 5C, bar 7), but the CARM1 E267Q mutant was essentially inactive (bar 8). Similar results were observed when p/CAF was substituted for p300 (data not shown). In contrast, when 100 ng of ER vector was used (with GRIP1 vector but without p300 vector), the coactivator function of the E267Q mutant of CARM1 was almost as good as that of wild-type CARM1 (bars 12 to 13). Thus, the methyltransferase activity of CARM1 was required for synergistic coactivator action with GRIP1 and p300 or p/CAF. However, at higher NR levels, the methyltransferase activity of CARM1 was not required for CARM1 to cooperate with GRIP1 in the absence of p300, suggesting that the coactivator function of CARM1 may involve two different activities of CARM1, one methyltransferase dependent and one methyltransferase independent. In addition, the results shown in Fig. 5C and 1A indicate that the requirements for coactivator function are different at high and low levels of NR. CARM1, compared with other protein arginine methyltransferases, has a unique ability to cooperate with p300 and p/CAF as coactivators for NRs. In addition to CARM1, the current list of mammalian arginine-specific protein methyltransferases includes PRMT1, PRMT2, PRMT3, and JBP1 (42). All of the proteins except PRMT2 have been shown to methylate specific proteins in vitro, but each enzyme methylates different proteins (38). CARM1, as indicated above, can methylate histone H3 (4). PRMT1 can methylate a variety of RNA binding proteins, as well as histone H4 and STAT1 (27, 38). A yeast enzyme, RMT1/Pmt1/ODP1, has a protein substrate specificity similar to that of mammalian PRMT1. With
NUCLEAR RECEPTOR COACTIVATOR SYNERGY
the exception of PRMT1 and RMT1, which have been shown to methylate various RNA binding proteins (38), the in vivo targets for these methyltransferases have not been established. We have shown previously that, in the presence of GRIP1, PRMT1 can act as a coactivator for NRs and can function synergistically with CARM1 (17). We therefore tested whether PRMT1 and other members of the protein arginine methyltransferase family could substitute for CARM1 to function in synergy with GRIP1 and p300 as coactivators for low levels of TR. In the presence of GRIP1 and absence of p300, each methyltransferase enhanced the reporter gene activity above the level observed with TR plus GRIP1 (Fig. 6A, bars 3 and 5 to 9), with the degree of enhancement ranging from two- to ninefold in this experiment. Similar results were observed at higher NR levels in the absence of p300 (data not shown). Thus, in the absence of p300, all of the methyltransferases exhibited similar, although not quantitatively identical, coactivator functions with GRIP1. Addition of p300 as a third coactivator caused a synergistic increase in activity when CARM1 was present but not when any of the other methyltransferases was present (bars 10 to 14). Essentially all of the activity observed was eliminated when GRIP1 (bars 15 to 19) or TR (bars 20 to 24) was omitted. Like p300, p/CAF cooperated synergistically with CARM1 but not with any of the other methyltransferases (Fig. 6B). Thus, p300 and p/CAF have a specific and synergistic functional relationship with CARM1 but not with the other protein arginine methyltransferases. DISCUSSION Role of p160 coactivator complexes in transcriptional activation by NRs. Presumably, many of the steps in the complex process of transcriptional activation are not mediated by a single protein but by multiple proteins, possibly existing as preformed complexes, which work together to accomplish a single task or step in the process. Thus, studying the effect of an individual protein on gene activation is unlikely to produce a full understanding of that protein’s function in vivo and may in fact provide misleading clues about the protein’s true physiological role in the overall process. In the system reported here, the activity of low levels of NRs was almost completely dependent upon synergistic action of a p160 coactivator, a coactivator of the protein acetyltransferase family (p300, CBP, or p/CAF), and the protein methyltransferase CARM1. The absolute dependence of NR function and the functions of the other coactivators (CARM1, p300, and p/CAF) on the coexpression of GRIP1 further supports the model (3, 4, 10) that p160 coactivators play a central and direct role in bringing the coactivator complex into association with the NRs, whereas CARM1, p300, and p/CAF associate with the NRs through their association with GRIP1. The absolute requirement for the AD2 domain of GRIP1 (Fig. 2C) is also consistent with this model whereby CARM1 is recruited to the promoter through its association with the AD2 domain of GRIP1 (Fig. 7). The fact that the deletion of AD1 did not cause complete loss of coactivator synergy or disruption of the ternary complex of coactivators (Fig. 2) was surprising given the clear evidence that p300 and CBP associate with the AD1 domain (5, 40). Thus, our results indicate that p300 and CBP can bind to GRIP1 through the AD1 domain, through another as yet un-
LEE ET AL.
FIG. 5. Role of protein methyltransferase activity of CARM1 in coactivator synergy. (A) Methyltransferase activities of wild-type and mutant CARM1. Cos-7 cells (100-mm-diameter dishes) were transfected with 2.5 g of the indicated CARM1 expression vector: Con, control with pSG5.HA; WT, pSG5.HA-CARM1 wild type; E/Q, pSG5.HA-CARM1(E267Q) mutant; VLD, pSG5.HA-CARM1(VLD) mutant. Cell extracts were immunoprecipitated (IP) with antibody against HA tag (anti-HA), and immunoprecipitates were incubated with mixed histones and S-adenosylmethionine to allow methylation. Incubated reactions were analyzed by Western immunoblotting (WB), using antibodies specific for the CARM1-methylated form of histone H3 (anti-MeH3; bottom). Expression of CARM1 was assessed before immunoprecipitation by immunoblotting with antibodies against HA tag (top). (B) Coactivator synergy with different amounts of wild-type and mutant CARM1. CV-1 cells in six-well dishes were transfected with 1 ng of TR vector, 250 ng of MMTV(TRE)-LUC reporter plasmid, 250 ng of GRIP1 vector, 500 ng of p300 vector (line plots), or no p300 vector (bars) and the indicated amount of CARM1 wild-type or
MOL. CELL. BIOL.
defined domain of GRIP1, or through contact with another component of the complex. Most previous studies which documented the binding of the p160 AD1 domain to the C-terminal region of CBP and p300 used fragments of these two proteins in their binding studies and thus apparently missed the secondary interaction of CBP/p300 with another region of p160 proteins (5, 16, 40, 41). However, Torchia et al. (39) reported that the C-terminal region of CBP bound strongly to the AD1 region and more weakly to the N-terminal region of a p160 coactivator. This secondary site may be responsible for the AD1-independent interaction we have observed between GRIP1 and p300/CBP. The AD1-independent coactivator function of GRIP1 occurred under low-NR conditions in the presence of three coactivators (Fig. 2C) but not at higher-NR conditions when GRIP1 and p300 were used without CARM1 (3). Our results suggest that the ternary coactivator complex CARM1-GRIP1-p300 may stabilize an AD1-independent p160-p300 interaction which is otherwise too weak to support coactivator function. p/CAF is known to exist in a large cellular complex which does not include p160 coactivators, p300, or CBP (28), and yet p/CAF can bind directly to p160 coactivators, CBP, and p300, as well as some NRs (10). In our low-NR system, the coactivator function of p/CAF was absolutely dependent on the presence of GRIP1, suggesting that p/CAF may be recruited to the promoter through its association with GRIP1 or that p/CAF coactivator activity may depend indirectly on activities of p160 coactivators and/or CARM1. The role of protein acetyltransferase activity in p300 and p/CAF coactivator function. Histone acetylation by p300, CBP, and p/CAF plays an important role in transcriptional activation of native genes (6, 8, 30, 35). In addition, the ability of these coactivators to acetylate and make contact with other components of the transcription machinery may also help to transmit the activating signal from the transcriptional activator-coactivator complex to the transcription machinery (5, 6, 32). Thus, in different circumstances these multifunctional coactivators may use different portions of their signaling repertoire. In fact, various recent studies have reached different conclusions as to whether the protein acetyltransferase activities of p300 and CBP are required for their coactivator function (6, 18–20). In our system, NR function required the acetyltransferase activity of p/CAF but not that of p300 (Fig. 4). Korzus et al. (18) reached similar conclusions using a microinjection assay. The diverse results in various studies as to whether the acetyltransferase activity of p300 and CBP is required may not be contradictory but may rather reflect the use of different transcriptional activators, reporter genes (e.g., different promoters or transiently transfected versus chromosomally integrated reporter genes), combinations of coactivators, cell types, or assay
mutant vector. The cells were grown with 20 nM T3, and the luciferase activities of the cell extracts were determined. (C) Coactivator activities of mutant and wild-type CARM1 at low versus high NR levels. CV-1 cells in six-well dishes were transfected with 1 (Low NR) or 100 (High NR) ng of ER vector, 250 ng of MMTV(ERE)-LUC reporter plasmid, 250 ng of GRIP1 vector, 500 ng of p300 vector, and 500 ng of wild-type or mutant CARM1 vector. The cells were grown with 20 nM E2, and the luciferase activities of the cell extracts were determined. ⫹, present.
VOL. 22, 2002
NUCLEAR RECEPTOR COACTIVATOR SYNERGY
FIG. 7. Different mechanisms of transcriptional activation at low and high NR levels. (Top) At low NR levels, NRs shuttle on and off of the hormone response element (HRE), and occupancy of the HRE is relatively infrequent. Transcriptional activation by the bound NRs requires the assistance of multiple coactivators to remodel chromatin structure and recruit and activate RNA polymerase II (Pol II complex). (Bottom) High NR levels may force almost constant occupancy of the HREs by NRs. Perhaps such high occupancy allows NRs to recruit RNA polymerase through direct contact with TATA binding protein (TBP), TFIIB, or other components of the RNA polymerase II complex, independent of the action of some or many coactivators. FIG. 6. Selective synergy of protein arginine methyltransferases with p300 and p/CAF. CV-1 cells in six-well dishes were transfected with 1 ng of TR vector, 250 ng of MMTV(TRE)-LUC reporter plasmid, and coactivator vectors as indicated: 250 ng of GRIP1 vector, 500 ng of p300 (A) or p/CAF (B) vector, and 500 ng of the indicated methyltransferase vector encoding CARM1, PRMT1, PRMT2, PRMT3, or RMT1. The cells were grown with 20 nM T3, and the luciferase activities of the cell extracts were determined. ⫹, present.
methods. The different requirements for the acetyltransferase functions of p300 versus p/CAF (Fig. 4), as well as our demonstration that GRIP1, CARM1, p300, and p/CAF can all cooperate synergistically to enhance NR function (Fig. 3), suggest that p300 and p/CAF contribute to transcriptional activation by different mechanisms: p/CAF through its acetyltransferase function and p300 through protein-protein interactions. Role of protein methyltransferase activity of CARM1 in coactivator function and synergy. The results with the two CARM1 mutants suggest several conclusions relevant to the mechanism of CARM1 coactivator function. First, as predicted
from the crystal structure of the CARM1 homologue PRMT3 (42), the VLD mutation apparently affected more than simply the methyltransferase activity, since both mutants lacked methyltransferase activity but the VLD mutant had a more severe loss of coactivator function (Fig. 5). Second, the fact that the E267Q mutant was expressed at normal levels, binds GRIP1, and had wild-type coactivator activity when tested under high-NR conditions (Fig. 5 and data not shown) indicates that the overall protein structure is probably not adversely affected by the E267Q mutation and thus suggests that the mutation caused a selective loss of methyltransferase function. Thus, the severe loss of coactivator function by this mutant when tested under low-NR conditions reinforces the importance of the methyltransferase activity in the coactivator function of CARM1. Third, the fact that the methyltransferase-negative mutant E267Q retains substantial coactivator function when tested under high-NR conditions indicates that another domain of CARM1 (which is functionally separable from the methyltransferase activity) can contribute to downstream signaling by CARM1, at least under high-NR conditions. CARM1
LEE ET AL.
has an autonomous transcriptional activation activity (17), which is separable from its methyltransferase activity (our unpublished results) and is an obvious candidate for the additional downstream signaling domain. Another indication of the importance of the CARM1 methyltransferase domain is the fact that CARM1, alone among the various arginine-specific methyltransferases tested, was capable of participating in three-coactivator synergy with p300 and p/CAF (Fig. 6). Each of the protein arginine methyltransferase family members methylates different protein substrates in vitro (38). The protein(s) methylated by CARM1 in connection with its coactivator function remains to be determined. Since CARM1 can methylate histone H3 in vitro (4, 34) and cooperates synergistically as a coactivator with a histone acetyltransferase (Fig. 4 and 5), an attractive hypothesis is that methylation of arginine residues of histone H3 by CARM1 cooperates with acetylation of histone H3 (or another histone) to promote chromatin remodeling (Fig. 7). In fact, recent in vitro and in vivo evidence has provided strong support for the hypothesis that histone acetylation, phosphorylation, and methylation (on lysine) can regulate each other and/or cooperate in promoting chromatin remodeling (8, 15, 30). These histone modifications may alter chromatin structure by neutralizing the positive charge of the N-terminal histone tails (21) or by creating new binding sites to recruit proteins which contribute to chromatin remodeling (14, 15, 30). Our results suggest that arginine-specific methylation of histone H3 or another protein by CARM1 is an important part of the transcription activation process, at least for NRs. In fact, we have recently shown by chromatin immunoprecipitation that CARM1 is recruited to steroid hormone-responsive promoters in a steroid hormone-dependent manner. Furthermore, using antibodies that specifically recognize the CARM1-methylated form of histone H3, we also observed that methylated histone H3 is preferentially associated with glucocorticoid-activated versus transcriptionally inactive MMTV promoters which are stably integrated into chromatin (22). These findings further support the role of CARM1 and its histone methyltransferase activity in transcriptional regulation. We cannot rule out the possibility that other proteins, in addition to histones, are methylated by CARM1 in connection with its coactivator function. Low NR levels provide stringent conditions for studying coactivator synergy; different coactivator requirements indicate differences in the mechanism of transcriptional activation at low and high NR levels. It is important to note that the levels of NR vector required for three-coactivator synergy in our system are well below those normally used for transient-transfection studies with NRs. The requirements for and effects of various combinations of the three coactivators (GRIP1, CARM1, and p300 or GRIP1, CARM1, and p/CAF) were quite different when low versus high levels of NR vectors were employed in the transient-transfection assays. At low NR levels, (i) NR alone failed to activate the reporter gene, even when hormone was present; (ii) use of one or two of the coactivators with NR caused modest to no activation of reporter gene; (iii) expression of all three coactivators resulted in a strong synergistic enhancement of NR function and in fact was required for efficient NR function; (iv) the methyltransferase activity of CARM1 was required for the coactivator
MOL. CELL. BIOL.
synergy and thus for reporter gene activation; and (v) substitution of another member of the protein arginine methyltransferase family for CARM1 eliminated the synergistic coactivator activity and most or all of the observed reporter gene activity. In contrast, when high NR levels were used, (i) NR alone activated the reporter gene in a hormone-dependent manner; (ii) GRIP1 alone enhanced NR function, and addition of one of the secondary coactivators (CARM1, p300, or p/CAF) produced a further enhancement in most cases; (iii) the activity achieved by adding all three coactivators was no more effective, and was often less effective, than the activity observed with two coactivators; (iv) the methyltransferase activity of CARM1 was not required for its ability to cooperate with GRIP1 to enhance NR activity; and (v) all of the protein arginine methyltransferases, when coexpressed with GRIP1, were approximately equivalent in their abilities to enhance NR function. Thus, at high versus low NR levels, the degree to which coactivators were required, the number of coactivators and types of coactivators which could effectively contribute to NR function, and even the specific functional domains of individual coactivators which were required to enhance reporter gene activation by NRs were dramatically different. In other words, the mechanisms of transcriptional activation at high versus low NR levels are different. While the exact mechanistic differences remain to be determined, we speculate that overexpression of NRs, which almost certainly occurs in transient-transfection experiments, may allow NRs to accomplish directly tasks which normally require assistance from coactivators (Fig. 7). Recent fluorescence-quenching studies of fluorescently labeled steroid receptors bound to large tandem promoter arrays suggest that these receptors rapidly and repeatedly associate with and dissociate from their cognate enhancer elements (26). NR overexpression may lead to a higher level of enhancer element occupancy by the NR. NRs and many other transcriptional activators can bind to various basal transcription factors, such as TFIIB and TBP (1, 13, 31, 33). Perhaps at concentrations high enough to force almost constant NR occupancy of enhancer elements, the NRs can recruit and activate the transcription initiation complex without additional assistance from some coactivators (Fig. 7). More work is required to test such ideas. We have previously shown that coactivator synergy between CARM1 and a related protein methyltransferase, PRMT1, also requires low levels of NR (17). Other investigators have also reported that high levels of NRs reduce the abilities of coactivators to enhance reporter gene activity further (7). Which set of experimental conditions (i.e., high versus low NR levels) may produce results which more accurately reflect the physiological roles of the coactivators? Since transient transfections almost certainly cause substantial overexpression of the proteins encoded by the transfected plasmids, it seems reasonable to conclude that the lower NR concentrations are more likely to approach physiological conditions. While there are some differences in the mechanism of transcriptional activation for stably integrated versus transiently transfected reporter genes (36), it seems unlikely that the stringent and specific coactivator requirements observed in our transient-transfection system are entirely an artifact of the experimental system or are irrelevant to the physiological function of these coactivators. Thus, it seems likely
VOL. 22, 2002
NUCLEAR RECEPTOR COACTIVATOR SYNERGY
that the extremely high degree of synergy observed among the three coactivators and the stringent requirement for the methyltransferase activity of CARM1 (which cannot be replaced by any other methyltransferase protein) provide valid insights about the function of these coactivators in vivo. p160 coactivators, p300 and CBP (6, 35), and more recently CARM1 (22) have all been shown to be recruited in response to steroid hormones to the promoters of natural target genes in their native chromosomal locations; thus, these proteins play important roles in the transcriptional activation process on native genes in vivo. This provides a strong validation for the synergy we observed for these three coactivators in our transient-transfection assays. Thus, the low-NR conditions represent a powerful and convenient new system for studying the mechanism of coactivator function; such quantitative mechanistic studies are an essential complement for techniques such as chromatin immunoprecipitation, which examine physical recruitment of proteins to stably integrated reporter genes. Our demonstration that GRIP1, CARM1, and p300 form a ternary complex in vivo and that the methyltransferase activity of CARM1 is required for transcriptional activation provide important insights into the mechanism of coactivator function. ACKNOWLEDGMENTS We thank the following investigators for providing plasmids: Vittorio Sartorelli (NIH) for vectors encoding p/CAF, p300, and their mutants; Tso-Pang Yao (Duke University) for CBP expression vector; Adam Frankel, Steve Clarke, and Harvey Herschman (UCLA) for vectors encoding PRMT1, PRMT2, PRMT3, and RMT1; and Richard Goodman (University of Oregon Health Sciences Center) for the vector encoding the Gal4-CBP fusion protein. We thank Baruch Frenkel (University of Southern California) for critical comments on the manuscript. This work was supported by U.S. Public Health Service Grant DK55274 (to M.R.S.) and GM61355 (to X.Z. and X.C.) from the National Institutes of Health. S.S.K. was supported by a predoctoral training fellowship from Grant AG00093 from the National Institutes of Health. REFERENCES 1. Baniahmad, A., I. Ha, D. Reinberg, S. Tsai, M.-J. Tsai, and B. W. O’Malley. 1993. Interaction of human thyroid hormone receptor ␤ with transcription factor TFIIB may mediate target gene derepression and activation by thyroid hormone. Proc. Natl. Acad. Sci. USA 90:8832–8836. 2. Beato, M., P. Herrlich, and G. Schütz. 1995. Steroid hormone receptors: many actors in search of a plot. Cell 83:851–857. 3. Chen, D., S.-M. Huang, and M. R. Stallcup. 2000. Synergistic, p160 coactivator-dependent enhancement of estrogen receptor function by CARM1 and p300. J. Biol. Chem. 275:40810–40816. 4. Chen, D., H. Ma, H. Hong, S. S. Koh, S.-M. Huang, B. T. Schurter, D. W. Aswad, and M. R. Stallcup. 1999. Regulation of transcription by a protein methyltransferase. Science 284:2174–2177. 5. Chen, H., R. J. Lin, R. L. Schiltz, D. Chakravarti, A. Nash, L. Nagy, M. L. Privalsky, Y. Nakatani, and R. M. Evans. 1997. Nuclear receptor coactivator ACTR is a novel histone acetyltransferase and forms a multimeric activation complex with P/CAF and CBP/p300. Cell 90:569–580. 6. Chen, H., R. J. Lin, W. Xie, D. Wilpitz, and R. M. Evans. 1999. Regulation of hormone-induced histone hyperacetylation and gene activation via acetylation of an acetylase. Cell 98:675–686. 7. Chen, S., N. J. Sarlis, and S. S. Simons, Jr. 2000. Evidence for a common step in three different processes for modulating the kinetic properties of glucocorticoid receptor-induced gene transcription. J. Biol. Chem. 275: 30106–30117. 8. Cheung, P., C. D. Allis, and P. Sassone-Corsi. 2000. Signaling to chromatin through histone modifications. Cell 103:263–271. 9. Chrivia, J. C., R. P. Kwok, N. Lamb, M. Hagiwara, M. R. Montminy, and R. H. Goodman. 1993. Phosphorylated CREB binds specifically to the nuclear protein CBP. Nature 365:855–859. 10. Glass, C. K., and M. G. Rosenfeld. 2000. The coregulator exchange in transcriptional functions of nuclear receptors. Genes Dev. 14:121–141. 11. Gluzman, Y. 1981. SV40-transformed simian cells support the replication of early SV40 mutants. Cell 23:175–182.
12. Huang, S.-M., and M. R. Stallcup. 2000. Mouse Zac1, a transcriptional coactivator and repressor for nuclear receptors. Mol. Cell. Biol. 20:1855– 1867. 13. Ing, N. H., J. M. Beekman, S. Y. Tsai, M.-J. Tsai, and B. W. O’Malley. 1992. Members of the steroid hormone receptor superfamily interact directly with TFIIB (S300II) to mediate transcriptional induction. J. Biol. Chem. 267: 17617–17623. 14. Jacobson, R. H., A. G. Ladurner, D. S. King, and R. Tjian. 2000. Structure and function of a human TAFII250 double bromodomain module. Science 288:1422–1425. 15. Jenuwein, T. 2001. Re-SET-ting heterochromatin by histone methyltransferases. Trends Cell Biol. 11:266–273. 16. Kamei, Y., L. Xu, T. Heinzel, J. Torchia, R. Kurokawa, B. Gloss, S.-C. Lin, R. A. Heyman, D. W. Rose, C. K. Glass, and M. G. Rosenfeld. 1996. A CBP integrator complex mediates transcriptional activation and AP-1 inhibition by nuclear receptors. Cell 85:403–414. 17. Koh, S. S., D. Chen, Y.-H. Lee, and M. R. Stallcup. 2001. Synergistic enhancement of nuclear receptor function by p160 coactivators and two coactivators with protein methyltransferase activities. J. Biol. Chem. 276:1089– 1098. 18. Korzus, E., J. Torchia, D. W. Rose, L. Xu, R. Kurokawa, E. M. McInerney, T.-M. Mullen, C. K. Glass, and M. G. Rosenfeld. 1998. Transcription factorspecific requirements for coactivators and their acetyltransferase functions. Science 279:703–707. 19. Kraus, W. L., E. T. Manning, and J. T. Kadonaga. 1999. Biochemical analysis of distinct activation functions in p300 that enhance transcription initiation with chromatin templates. Mol. Cell. Biol. 19:8123–8135. 20. Li, J., B. W. O’Malley, and J. Wong. 2000. p300 requires its histone acetyltransferase activity and SRC-1 interaction domain to facilitate thyroid hormone receptor activation in chromatin. Mol. Cell. Biol. 20:2031–2042. 21. Luger, K., and T. J. Richmond. 1998. The histone tails of the nucleosome. Curr. Opin. Genet. Dev. 8:140–146. 22. Ma, H., C. T. Baumann, H. Li, B. D. Strahl, R. Rice, M. A. Jelinek, D. W. Aswad, C. D. Allis, G. L. Hager, and M. R. Stallcup. 2001. Hormonedependent, CARM1-directed, arginine-specific methylation of histone H3 on the mouse mammary tumor virus promoter. Curr. Biol. 11:1981–1985. 23. Ma, H., H. Hong, S.-M. Huang, R. A. Irvine, P. Webb, P. J. Kushner, G. A. Coetzee, and M. R. Stallcup. 1999. Multiple signal input and output domains of the 160-kDa nuclear receptor coactivator proteins. Mol. Cell. Biol. 19: 6164–6173. 24. Mangelsdorf, D. J., and R. M. Evans. 1995. The RXR heterodimers and orphan receptors. Cell 83:841–850. 25. McKenna, N. J., J. Xu, Z. Nawaz, S. Y. Tsai, M.-J. Tsai, and B. W. O’Malley. 1999. Nuclear receptor coactivators: multiple enzymes, multiple complexes, multiple functions. J. Steroid Biochem. Mol. Biol. 69:3–12. 26. McNally, J. G., W. G. Muller, D. Walker, R. Wolford, and G. L. Hager. 2000. The glucocorticoid receptor: rapid exchange with regulatory sites in living cells. Science 287:1262–1265. 27. Mowen, K. A., J. Tang, W. Zhu, B. T. Schurter, K. Shuai, H. R. Herschman, and M. David. 2001. Arginine methylation of STAT1 modulates IFN␣/␤induced transcription. Cell 104:731–741. 28. Ogryzko, V. V., T. Kotani, X. Zhang, R. L. Schlitz, T. Howard, X. J. Yang, B. H. Howard, J. Qin, and Y. Nakatani. 1998. Histone-like TAFs within the PCAF histone acetylase complex. Cell 94:35–44. 29. Puri, P. L., V. Sartorelli, X. J. Yang, Y. Hamamori, V. V. Ogryzko, B. H. Howard, L. Kedes, J. Y. Wang, A. Graessmann, Y. Nakatani, and M. Levrero. 1997. Differential roles of p300 and PCAF acetyltransferases in muscle differentiation. Mol. Cell 1:35–45. 30. Rice, J. C., and C. D. Allis. 2001. Histone methylation versus histone acetylation: new insights into epigenetic regulation. Curr. Opin. Cell Biol. 13: 263–273. 31. Sadovsky, Y., P. Webb, G. Lopez, J. D. Baxter, P. M. Fitzpatrick, E. GizangGinsberg, V. Cavaillès, M. G. Parker, and P. J. Kushner. 1995. Transcriptional activators differ in their responses to overexpression of TATA-boxbinding protein. Mol. Cell. Biol. 15:1554–1563. 32. Sartorelli, V., P. L. Puri, Y. Hamamori, V. Ogryzko, G. Chung, Y. Nakatani, J. Y. Wang, and L. Kedes. 1999. Acetylation of MyoD directed by PCAF is necessary for the execution of the muscle program. Mol. Cell 4:725–734. 33. Schulman, I. G., D. Chakravarti, H. Juguilon, A. Romo, and R. M. Evans. 1995. Interactions between the retinoid X receptor and a conserved region of the TATA-binding protein mediate hormone-dependent transactivation. Proc. Natl. Acad. Sci. USA 92:8288–8292. 34. Schurter, B. T., S. S. Koh, D. Chen, G. J. Bunick, J. M. Harp, L. Hanson, A. Henschen-Edman, D. R. Mackay, M. R. Stallcup, and D. W. Aswad. 2001. Methylation of histone H3 by coactivator-associated arginine methyltransferase 1. Biochemistry 40:5747–5756. 35. Shang, Y., X. Hu, J. DiRenzo, M. A. Lazar, and M. Brown. 2000. Cofactor dynamics and sufficiency in estrogen receptor-regulated transcription. Cell 103:843–852. 36. Smith, C. L., and G. L. Hager. 1997. Transcriptional regulation of mammalian genes in vivo. A tale of two templates. J. Biol. Chem. 272:27493–27496. 37. Spencer, T. E., G. Jenster, M. M. Burcin, C. D. Allis, J. Zhou, C. A. Mizzen, N. J.
LEE ET AL.
McKenna, S. A. Oñate, S. Y. Tsai, M.-J. Tsai, and B. W. O’Malley. 1997. Steroid receptor coactivator-1 is a histone acetyltransferase. Nature 389:194–198. 38. Stallcup, M. R. 2001. Role of protein methylation in chromatin remodeling and transcriptional regulation. Oncogene 20:3014–3020. 39. Torchia, J., D. W. Rose, J. Inostroza, Y. Kamei, S. Westin, C. K. Glass, and M. G. Rosenfeld. 1997. The transcriptional co-activator p/CIP binds CBP and mediates nuclear receptor function. Nature 387:677–684. 40. Voegel, J. J., M. J. S. Heine, M. Tini, V. Vivat, P. Chambon, and H. Grone-
MOL. CELL. BIOL. meyer. 1998. The coactivator TIF2 contains three nuclear receptor binding motifs and mediates transactivation through CBP binding-dependent and -independent pathways. EMBO J. 17:507–519. 41. Yao, T.-P., G. Ku, N. Zhou, R. Scully, and D. M. Livingston. 1996. The nuclear hormone receptor coactivator SRC-1 is a specific target of p300. Proc. Natl. Acad. Sci. USA 93:10626–10631. 42. Zhang, X., L. Zhou, and X. Cheng. 2000. Crystal structure of the conserved core of protein arginine methyltransferase PRMT3. EMBO J. 19:3509–3519.