Ligand type-specific Interactions of Peroxisome Proliferator-activated Receptor γ with Transcriptional Coactivators*

The nuclear peroxisome proliferator-activated receptor γ (PPARγ) is a member of the nuclear receptor superfamily and acts as a ligand-dependent transcription factor mediating adipocyte differentiation, cell proliferation and inflammatory processes, and modulation of insulin sensitivity. Members of the 160-kDa protein (SRC-1/TIF2/AIB-1) family of coactivators, CBP/p300 and TRAP220/DRIP205, are shown to interact directly with PPARγ and potentiate nuclear receptor transactivation function in a ligand-dependent fashion. Because PPARγ ligands exert partially overlapping but distinct subsets of biological action through PPARγ binding, we wished to examine whether interactions between PPARγ and known coactivators were induced to the same extent by different classes of PPARγ ligand. The natural ligand 15-deoxy-Δ12,14-prostaglandin J2 induced PPARγ interactions with all coactivators tested (SRC-1, TIF2, AIB-1, p300, TRAP220/DRIP205) in yeast and mammalian two-hybrid assays, as well as in a glutathione S-transferase pull-down assay. However, under the same conditions troglitazone, a synthetic PPARγ ligand that acts as an antidiabetic agent, did not induce PPARγ interactions with any of the coactivators. Our findings suggest that ligand binding may alter PPARγ structure in a ligand type-specific way, resulting in distinct PPARγ-coactivator interactions.

member of the nuclear hormone receptor superfamily, acts as a ligand-inducible transcription factor (1,2). PPAR␥ forms a heterodimer complex with one of the three retinoid X receptor (RXR) proteins, which then binds to PPAR-responsive elements (PPRE) within the promoters of PPAR␥ target genes (3,4). It is thought that the ligand binding domain (LBD) mediates the ligand-dependent transactivation function of PPAR␥, although two transactivation domains, at the N-terminal (AF-1) and C-terminal ends (AF-2), are present in most nuclear receptors. Ligand-induced transactivation is achieved by the nuclear receptor recruiting one of several types of nuclear receptor coactivator complex. One class of coactivator complex includes three SRC-1 family members (5), CBP/p300 (6), and SRA (7), as well as other proteins (8,9). The SRC-1 family members (SRC-1 (p160/NCoA-1) (10), TIF2 (GRIP-2) (11), and AIB1 (p/CIP/ ACTR) (12)) interact with the AF-2 nuclear receptors. This interaction is highly ligand-dependent through direct binding to the minimal activation domain of AF-2 (AF-2 AD), mapped to the C-terminal ␣-helix 12 (H12) in the LBD (13). CBP/p300 serves as an essential coactivator not only for nuclear receptors but also for other classes of transcription regulatory factor (14) and, like the SRC-1 family members, possesses histone acetyltransferase activity (15). Another coactivator complex, TRAP/ DRIP, contains at least 12 components, one of which exhibits direct and ligand-dependent interaction with H12 in the LBD (TRAP220/DRIP205/PBP) (16,17).
Reflecting on the role of the PPAR␥ function in many biological events, a variety of endogenous and synthetic ligands have been reported to activate the transactivation function of PPAR␥. Of the natural ligands for PPAR␥, prostaglandin derivative 15-deoxy-⌬12,14 prostaglandin J 2 (15d-PGJ2) and 9or 13-hydoxyoctadienoic acid (9-HODE or 13-HODE) are known to mediate potent adipogenesis and anti-inflammatory effects. Several synthetic thiazolidinedione derivatives, such as troglitazone, BRL49653, and pioglitazone, have been developed for anti-hyperglycemic activity in vivo (18 -21). In this study, we hypothesized that the ligand type-specific effects mediated by PPAR␥ are exerted through ligand type-specific structures on PPAR␥, allowing different coactivator associations. This conjecture of ligand type-specific structures on PPAR␥ is supported by ER␣ LBD crystallographic findings showing that ligand binding altered LBD structure with distinct and ligand type-specific shifts in the H12 angle (22). To examine our hypothesis, we studied the interactions between PPAR␥ bound to distinct ligands with known nuclear receptor coactivators. Although both ligands used were equally potent in the transactivation function of PPAR␥, direct interactions of PPAR␥ with SRC-1, TIF2, AIB-1, p300, and TRAP220 were observed when 15d-PGJ2 was bound but not when troglitazone was bound. Consistent with these coactivator-specific interactions, the transactivation function of troglitazone-bound PPAR␥ was not potentiated by coactivator overexpression. Thus, the present findings suggest that PPAR␥ structure is altered in a * This work was supported in part by a grant-in-aid for priority areas from the Ministry of Education, Science, Sports and Culture of Japan (to S. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Mammalian Two-hybrid Assay-COS-1 cells were maintained in Dulbecco's modified Eagle's medium without phenol red, supplemented with 5% fetal calf serum stripped with dextran-coated charcoal. Cells were transfected by calcium phosphate coprecipitation as described previously (24). Reporter plasmid (1 g) containing GAL4-UAS (17mer ϫ2 (␤-globin promoter) CAT) was cotransfected with 0.2 g of pVP-PPAR␥2(DEF) plus 0.2 g of either pM-SRC-1, pM-TIF2, pM-AIB-1, pM-p300 or pM-TRAP220. As a reference plasmid for normalization, 2 g of pCH110 plasmid was used (Amersham Pharmacia Biotech). Bluescribe M13 ϩ (Stratagene) was used as the carrier to adjust the total amount of DNA to 5 g. 15d-PGJ2 or troglitazone (0.1-100 M) was added to the medium 12 h after transfection and every 8 h thereafter at each exchange of medium. After 48 h, CAT activity was assayed, and transfection efficiency was normalized to ␤-galactosidase activity as described previously (26).
Bound proteins were extracted into loading buffer, separated by 7.5% SDS-PAGE, and visualized by autoradiography. Polyacrylamide gels were lightly stained with Coomassie Brilliant Blue to verify loading of equal quantities of fusion proteins prior to drying and autoradiography.
Electrophoretic Mobility Shift Assay-The interaction of TIF2 with ligand-bound PPAR␥2-RXR was determined by electrophoretic mobility shift assays with DNA probes as described (24). GST-PPAR␥2, GST-RXR␤, and GST-TIF2 were expressed in E. coli as GST fusion proteins and purified by digestion with thrombin following affinity column chromatography. Digested samples were applied to Sephadex G-100 to further purify the PPAR␥2, RXR␤, and TIF2 proteins with protein purity and quantity monitored by SDS-PAGE. In a typical assay, 10 ng of recombinant PPAR␥2 and/or RXR␤ protein with or without 10 ng of TIF2 protein in the presence or absence of 10 Ϫ7 M 15d-PGJ2 or troglitazone were incubated for 30 min on ice in a binding buffer (5 mM Tris, pH 8.0, 40 mM KCl, 6% glycerol, 1 mM dithiothreitol, 0.05% Nonidet P-40), 2 g of poly(dI⅐dC), 0.1 g of denatured salmon sperm DNA, and 10 g of bovine serum albumin in a final volume of 20 l. Doublestranded consensus mouse acyl-CoA oxidase-PPRE (PPRE; 5Ј-GTCG-ACAGGGGACCAGGACAAAGGTCACGTTCGGGAGT-3Ј) DNA fragments (3) were end-labeled using [␥-32 P]ATP and T4 polynucleotide kinase and used as probe. PPRE DNA fragments were added to the binding mixtures, and the mixtures was incubated for 20 min at room temperature. The entire reaction mixture (20 l) was loaded onto 4.5% polyacrylamide gels in 0.5ϫ Tris-acetate-EDTA buffer and electrophoresed at 4°C. The gels were dried on filter paper and exposed to x-ray film.
Transactivation Assays-COS-1 cells were maintained as described above for the mammalian two-hybrid system. The following plasmids were used for transfection: respective reporter plasmid (1 g) containing the pGL-GAL4-UAS (17-mer ϫ 2-␤-globin promoter-luciferase) cotransfected with 0.1 g of pM(GAL4-DBD)-PPAR␥ (DEF) or pM-PPAR␥ (DEF-⌬AF-2) with or without 1 g of SRC-1, TIF2, or TRAP220 expression vector. As a reference plasmid for normalization, 10 ng of pRL-CMV plasmid (Promega) was used. Bluescribe M13 ϩ (Stratagene) was used as the carrier to adjust the total amount of DNA to 3 g. 1 M 15d-PGJ2 or troglitazone was added to the medium 12 h after transfection and every 8 h thereafter at each exchange of medium. After 48 h, firefly luciferase activity (from GAL4-UAS) was used to measure transfection efficiency by Renilla luciferase activity (from pRL-CMV) as described previously (27).

Ligand Type-specific Interactions of PPAR␥ with Coactivators in the Mammalian Two-hybrid System and GST Pull-down
Assay-We first tested for ligand-induced and dose-dependent interactions of PPAR␥ using two distinct coactivator classes in a mammalian two-hybrid system. For this assay, the LBD of PPAR␥ containing AF-2 was fused to the VP16 domain in the pVP vector (pVP-PPAR␥ (DEF)), and several coactivators (SRC-1, TIF2, AIB-1, p300, and TRAP220/DRIP205) (24) fused to the GAL4 activation domain in the pM vector. The natural ligand 15d-PGJ2 (0.1 M) induced PPAR␥ interactions with SRC-1, TIF2, AIB-1, p300, and TRAP220 (Fig. 1A); however, no ligand-dependent interaction with coactivators was detected after 9-HODE binding. Troglitazone (1 M), a synthetic PPAR␥ ligand, induced no PPAR␥ interactions with any of the coactivators tested, and showed only weak interaction between PPAR␥ and SRC-1 or p300 at 10 M. It is notable that both 15d-PGJ2 (0.1 M) and troglitazone were equally potent in ligand-induced transactivation by PPAR␥ (Fig. 3). This ligand type-specific interaction between PPAR␥ and coactivators was also observed using a yeast two-hybrid system (data not shown).
We then determined whether ligand-dependent interaction of PPAR␥ with coactivators was ligand type-specific in vitro using a GST pull-down assay. [ 35 S]Methionine-labeled SRC-1, TIF2, AIB-1, p300, and TRAP220 were applied to glutathionebeads with GST-fused PPAR␥ protein in the presence and absence of ligand. Consistent with the results from the mammalian and yeast two-hybrid systems, the 15d-PGJ2-bound PPAR␥ physically interacted with all coactivators (Fig. 1B), whereas troglitazone failed to induce PPAR␥-coactivator interaction. A GST did not bind to all of the coactivators in the presence of ligands (data not shown).
Troglitazone Binding Is Unable to Recruit TIF2 to the PPAR␥-RXR Heterodimer upon PPRE Binding-We next tested whether troglitazone binding induced coactivator interactions with a DNA-bound PPAR␥/RXR␤ heterodimer. An electrophoretic mobility shift assay with a well characterized consensus PPRE from the acyl-coenzyme A oxidase gene (acyl-CoA) promoter (3) was used. As shown in Fig. 2, despite the absence of 15d-PGJ2 or the RXR-specific ligand (LG268), PPAR␥/RXR␤ heterodimer DNA binding was observed, whereas binding of the single receptors was not observed (lane 4). TIF2 recruitment induced the formation of a larger complex, observed as a slow migrating band produced by the binding of 15d-PGJ or LG268 to PPAR␥/RXR␤ (lanes 9 and 11). However, TIF2 recruitment was not induced by troglitazone binding (lane 10).
Ligand-specific Potentiation of PPAR␥ Transactivation Function by Coactivators-The observations that PPAR␥ interactions with SRC-1 family members, p300, and TRAP220 proteins were ligand type-specific, suggesting that the transactivation function of ligand-bound PPAR␥ was differentially potentiated by these coactivators. A transient expression assay was performed in COS-1 cells using pM (GAL4-DBD)-PPAR␥ PPAR␥ Interactions with Transcriptional Coactivators 33202 (DEF), pM-PPAR␥(DEF-⌬AF-2), and a reporter plasmid (GAL4-UAS-luciferase) containing the luciferase gene along with consensus GAL4 upstream activating sequence. As shown in Fig. 3, the transactivation function of PPAR␥ induced by troglitazone (lane 4) was comparable with that induced by 15d-PGJ2 (lane 2). 9-HODE was unable to induce significant transactivation even with wild-type PPAR␥ (lane 3). Moreover, the transactivation function of PPAR␥ induced by the two ligands was disrupted in the same way when the AF-2 AD core (H12) sequence was deleted (PPAR␥(DEF-⌬AF-2)) (lanes 6 and 8). SRC-1 and TIF2 significantly enhanced the transactivation function of PPAR␥ induced by 15d -PGJ2 (lanes 10, 13), and a potentiation by TRAP220 was also seen but was not statistically significant (lane16). However, the troglitazone-induced transactivation function of PPAR␥ was not potentiated by these coactivators (lanes 11, 14, and 17). Thus, although 15d-PGJ2 appears to alter PPAR␥ LBD structure to allow coacti-vator recruitment, troglitazone-bound LBD may be modulated in some other way.
PPAR␥-mediated signaling is involved in a variety of biological events, such as adipocyte differentiation, cell proliferation, and inflammatory processes (2). To modulate particular PPAR␥-mediated events, synthetic PPAR␥ ligands have been developed in addition to the identification of endogenous ligands 15d-PGJ2 and 9-HODE. Interestingly, the biological actions mediated by PPAR␥ were reported to differ according to the ligand used (28). These observations led us to examine the molecular mechanism underlying the ligand-specific actions of the PPAR␥ ligands. Structural analyses by crystallography revealed that the LBD structures of many nuclear receptors were altered in a ligand type-specific way, particularly at helix 12 (29,30). As the alteration of the helix 12 angle upon ligand binding to the nuclear receptor LBD is now considered essential for coactivator recruitment, we decided to examine the interactions between coactivators and PPAR␥ bound to distinct

PPAR␥ Interactions with Transcriptional Coactivators 33203
classes of PPAR␥ ligands. Both 15d-PGJ2 and troglitazone at the same concentration (1 M) were equally potent in the induction of PPAR␥ transactivation function, whereas the other endogenous ligand tested, 9-HODE, was unable to activate PPAR␥ transactivation. Ligand-dependent interactions of PPAR␥ with the tested coactivators were observed using 15d-PGJ2 by both in vivo and in vitro assays. However, troglitazone binding to PPAR␥ failed to induce coactivator interactions in these assays, indicating that the mode of coactivator interaction with PPAR␥ was ligand type-specific. These findings imply that troglitazone-bound PPAR␥ may recruit components other than TRAP220/DRIP205 in the DRIP/TRAP coactivator complex, or proteins other than the 160-kDa family proteins and CBP/p300 in the SRC-1 family-type coactivator complex to form transcription initiation complexes. An alternative possibility is that an unknown coactivator complex may be recruited to troglitazone-bound PPAR␥. In this respect, it would be interesting to examine whether troglitazone could induce PPAR␥ interaction with PGC-1 and PGC-2, which are also reported to act as PPAR␥ coactivators (31,32). Nevertheless, as ligand-induced coactivator interactions with PPAR␥ appear to be distinct between 15d-PGJ2 and troglitazone, the overall structure of PPAR␥ and coactivator complexes may be different according to the ligands involved, resulting in the activation of a particular set of target gene promoters that exert different biological actions.