p300 Interacts with the N- and C-terminal Part of PPARγ2 in a Ligand-independent and -dependent Manner, Respectively*

The nuclear peroxisome proliferator-activated receptor γ (PPARγ) activates the transcription of multiple genes involved in intra- and extracellular lipid metabolism. Several cofactors are crucial for the stimulation or the silencing of nuclear receptor transcriptional activities. The two homologous cofactors p300 and CREB-binding protein (CBP) have been shown to co-activate the ligand-dependent transcriptional activities of several nuclear receptors as well as the ligand-independent transcriptional activity of the androgen receptor. We show here that the interaction between p300/CBP and PPARγ is complex and involves multiple domains in each protein. p300/CBP not only bind in a ligand-dependent manner to the DEF region of PPARγ but also bind directly in a ligand-independent manner to a region in the AB domain localized between residue 31 to 99. In transfection experiments, p300/CBP could thereby enhance the transcriptional activities of both the activating function (AF)-1 and AF-2 domains. p300/CBP displays itself at least two docking sites for PPARγ located in its N terminus (between residues 1 and 113 for CBP) and in the middle of the protein (between residues 1099 and 1460).

The three peroxisome proliferator-activated receptors (PPARs) 1 ␣, ␦ (or ␤), and ␥, each encoded by a separate gene and displaying different tissue distributions and distinct ligand selectivities, belong to the nuclear hormone receptor superfamily (1). PPAR␥ is an important transcription factor involved in adipocyte differentiation and glucose metabolism. The PPAR␥ gene gives rise to two different PPAR␥ proteins, i.e. PPAR␥1 and PPAR␥2. PPAR␥2 differs from PPAR␥1 by the presence at its N terminus of an additional 28-amino acid domain whose function is so far unknown (2). Expression of both PPAR␥ types is enriched in white adipose tissue (2), which is consistent with the major function this receptor plays in adipogenesis (3). To date, we have only a limited insight into the molecular basis by which PPARs control gene expression. Like other nuclear receptors, PPARs are suggested to have a modular structure consisting of six functional domains, A/B, C, D, and E/F (4). The A/B and E/F regions are each endowed with transcriptional activities: the activating functions (AF)-1 and -2, respectively. The E/F region also the ligand binding domain (LBD) and the AF-2 is ligand-dependent. Classically it is suggested that ligand binding facilitates the heterodimerization of PPAR with the retinoid X receptor (RXR) and the binding of the PPAR/ RXR heterodimers to peroxisome proliferator-responsive elements. Consecutively the ligand-activated heterodimer stimulates transcription of the target gene. In addition to this liganddependent regulation, it was recently demonstrated that the transcriptional activity of PPARs could be also altered by covalent modifications such as phosphorylation (5)(6)(7)(8)(9). Furthermore, the transcriptional activity of nuclear receptors can be influenced by cofactors, such as co-activators or co-repressors, which modulate signaling and interaction with the basal transcription machinery (10).
Among the cofactors shown to modulate nuclear receptor transcriptional activities, p300 and the CREB-binding protein (CBP), two homologous co-activators, have recently attracted interest because of the pivotal role they play in the cross-talk between different signal transduction pathways (11)(12)(13). Acting as factors capable of both influencing chromatin structure and establishing contacts between the nuclear receptors and the basal transcription machinery, p300 and CBP provide a model to explain how nuclear receptors exert their effect on gene expression (14 -20).
So far only a few studies addressed the interaction between PPARs and cofactors. Dowell et al. (21) have demonstrated that p300 could co-activate PPAR␣ ligand-dependent transcriptional activity and could interact with the PPAR␣ DEF domain in a ligand-dependent way. Aside from p300, the only cofactors described so far for PPAR␥ are members of the steroid receptor co-activator-1 (SRC-1) family (22)(23)(24)(25), the PPAR binding protein PBP (26), the PPAR gamma co-activator (PGC)-1 (27), and the receptor interacting protein (RIP)-140 (28). Although Mizukami and Taniguchi (29), using a yeast two-hybrid system, have shown an interaction between the ligand binding domain of PPAR␥ and CBP, they did not provide any evidence for a co-activation function or a physiological role for CBP in this interaction.
The aim of this work was to evaluate more precisely the role of p300 and CBP in PPAR␥-mediated gene expression. A detailed analysis of the interaction domains between PPAR␥ and p300/CBP revealed for the first time that PPAR␥ contacts p300/CBP not only through its DEF domain in a ligand-depend-ent manner but also through its AB domain in a ligand-independent manner. CBP itself contacts PPAR␥ through several domains located in its N terminus and in a region located in the middle of the protein. As a consequence, in transfection experiments, p300 was able to co-activate independently the AF-1and AF-2-mediated transcriptional activities of PPAR␥ when its ABC domain, on the one hand, and its DEF domain, on the other hand, were fused to the yeast Gal4 DNA-binding domain. The finding that the interaction between a cofactor such as p300/CBP and nuclear receptors involves numerous domains in both partners might help to understand how the N terminus region is able to regulate the whole activity of nuclear receptors.

Materials
BRL 49,653 was a kind gift of Dr. L. Hamann and R. Heyman (Ligand Pharmaceuticals, San Diego, CA). The CMV p300-CHA expression vector was a gift of Dr. R. Eckner. The different CBP-glutathione S-transferase (GST) constructs were a gift of Dr. R. Janknecht. The antibodies directed against the AB domain of PPAR␥ were produced in our laboratory and were a kind gift of Dr. J. Najib (2). The antibodies directed against the ligand binding domain (LBD) of PPAR␥ were a kind gift of Dr. J. Berger and Dr. M. Leibowitz (Merck Research Laboratories, Rahway, NJ). Anti-hemagglutinin antibodies (anti-HA.11) were purchased at BabCo (Richmond, CA). The protease inhibitor mixture was purchased at ICN (Orsay, France).

Cell Culture and Transient Transfection Assays
The HeLa cell line was maintained in Dulbecco's modified Eagle's minimal essential medium supplemented with 10% delipidated and charcoal-treated fetal calf serum, L-glutamine, and antibiotics.
Transfections with chloramphenicol acetyltransferase (CAT) reporter constructs were carried out exactly as described previously (30) in 6-well plates. The pGL3-(J wt ) 3 TKCAT reporter construct contains three tandem repeats of the J site of the apolipoprotein A-II promoter cloned upstream of the herpes simplex virus thymidine kinase (TK) promoter and the CAT reporter gene (30). The following expression vectors were used: CMV p300-CHA, a construct where the last 36 amino acids from the C terminus of p300 have been replaced by a hemagglutinin (HA) epitope (31); pSG5-hPPAR␥2, a construct containing the entire cDNA of the human PPAR␥2 (hPPAR␥2) (2); pcDNA3-BDGal4-hPPAR␥ ABC , a construct where the A, B, and C regions of PPAR␥2 (aa from 2 to 181) have been cloned downstream of the Gal4 DNA binding domain; pcDNA3-BDGal4-hPPAR␥ DEF , a construct where the D, E, and F regions of PPAR␥2 (aa from 181 to 507) have been cloned downstream of the Gal4 DNA binding domain; pGL3-(Gal4) 5 TKLuc, a reporter construct consisting of five tandem repeats of the Gal4 upstream activating sequence (UAS) cloned in front of the TK promoter and driving the expression of the luciferase reporter gene; and pCMV-␤Gal, a vector for the control of transfection efficiency.

Production of Proteins
The p300Nt-GST, CBP-GST, and SRC1 fusion proteins were generated by cloning the N-terminal part of the p300 protein (aa 2 to 516), or different domains of CBP, or the domain comprised between amino acids 568 and 780 of SRC-1 downstream of the glutathione S-transferase (GST) protein in the pGex-T1 vector (Amersham Pharmacia Biotech, Orsay, France). The p300Nt-GST and CBP-GST fusion proteins were then expressed in Escherichia coli and purified on a glutathione affinity matrix (Amersham Pharmacia Biotech). The PPAR␥2 AB1-146 (aa 1 to 146 of PPAR␥), the PPAR␥2 ABC1-181 (aa 1 to 181 of PPAR␥), and the PPAR␥2 DEF204 -507 (aa 204 to 507 of PPAR␥) proteins were produced following the same procedure, and the GST domain was removed by thrombin digestion.

Immunoprecipitation and Pull-down Experiments
Immunoprecipitations-Polyclonal antibodies (5 g) directed against the AB domain of PPAR␥ were added to nuclear extracts (150 g at 0.5 mg/ml) prepared as described previously (32). The samples were incubated for 1 h at 4°C in the presence or absence of 10 Ϫ6 M BRL 49,653. Hydrated protein A-Q-Sepharose beads (20 l, Sigma, St. Quentin Fallavier, France), which had been first blocked with 3% bovine serum albumin in lysis buffer (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 40 mM NaCl, 1% Nonidet P-40, protease inhibitor mixture), were then added, and the samples were incubated under constant agitation for 20 min at 21°C. The beads were then washed four times in lysis buffer. Complexes were recovered by boiling the beads in 2ϫ sample buffer (12.5 mM Tris-HCl, 20% glycerol, 0.002% bromphenol blue, 5% ␤-mercaptoethanol), separated by 8% acrylamide SDS-PAGE, and transferred to nitrocellulose membranes. Blots were then developed with anti-HA.11 antibodies.
Pull-down Experiments-The purified PPAR␥2 AB1-146 , PPAR-␥2 ABC1-181 , and PPAR␥2 DEF204 -507 proteins were incubated 1 h at 22°C in pull-down buffer (1ϫ phosphate-buffered saline, 10% glycerol, 0.5% Nonidet P-40) with either GST or the different GST fusion proteins, glutathione-Q-Sepharose beads, and different concentrations of BRL 49,653 when necessary. The beads were then washed four times in pull-down buffer and boiled in 2ϫ sample buffer. The samples were separated by 12% acrylamide SDS-PAGE and transferred to nitrocellulose membranes. Blots were developed with antibodies directed against PPAR␥2 AB or PPAR␥2 DEF .

Yeast Two-hybrid System
Different domains of hPPAR␥2 were cloned in the pBDGal4 vector for the construction of bait plasmids (Stratagene, La Jolla, CA); the different parts of the ABC region as well as PPAR␥2 DEF181-507 were cloned by polymerase chain reaction amplification on the pSG5-PPAR␥2 construct of the corresponding domains. The three PPAR␥2 DEF deletion constructs PPAR␥ DEF181-501 , PPAR␥ DEF181-281 , and PPAR␥ DEF181-224 were generated by removing the 3Ј-ends of the PPAR␥2 DEF181-507 insert located downstream of the SalI, EcoRI, and BglII sites, respectively. The N-terminal part of p300 (aa 2 to 516) was cloned in the pADGal4 vector (Stratagene). The pADGal4-SV40 construct was purchased from Stratagene. YRG-2 competent yeasts (Stratagene) were transformed with different combinations of expression vectors following the instructions of the manufacturer and grown at 30°C on synthetic medium agar plates in the presence of the appropriate amino acids for selection. When the Gal4-chimera proteins interact, induction of the HIS3 and the LacZ genes occur, and the yeasts can grow on histidine-deficient media. The ␤-galactosidase assay was performed as described before (30) but with yeast lysates from saturated yeast cultures lyzed with acid-washed beads (Sigma).

RESULTS
p300 Stimulates PPAR␥ Transcriptional Activity in HeLa Cells-As p300 has been shown to co-activate the transcriptional activity of several nuclear receptors, we first addressed the question of whether the co-activator p300 could also enhance PPAR␥2-mediated gene expression. HeLa cells were therefore co-transfected with the proliferator-responsive element-driven reporter construct pGL3-(J wt ) 3 TKCAT (30), together with an expression vector for human PPAR␥2 (pSG5-hPPAR␥2) (2) and increasing amounts of an expression vector for p300-CHA (CMV p300-CHA) (31), in the presence or absence of BRL 49,653, a synthetic PPAR␥ ligand (33) (Fig. 1). PPAR␥ transcriptional activity is stimulated in a dose-dependent way by co-transfection with the CMV p300-CHA expression vector. This effect is maximal in presence of 0.8 g of CMV p300-CHA.
To clarify the role of p300 toward each of the two PPAR␥2 transcriptional activities (AF-1 and AF-2), we performed transfections with expression vectors coding for chimeric proteins composed of either the A, B, and C or the D, E, and F domains of hPPAR␥2 fused to the binding domain of the Gal4 yeast transcription factor (BDGal4-hPPAR␥ ABC and BDGal4-hPPAR␥ DEF , respectively, Fig. 2A). These vectors were cotransfected in HeLa cells together with increasing amounts of CMV p300-CHA. Whereas we observed a significant stimulation of the transcriptional activity of the chimeric BDGal4-hPPAR␥ ABC protein by co-transfected p300, the stimulation of the DEF chimera was extremely weak (Fig. 2, B and C). These activities are maximally increased 2.4 times for BDGal4-hPPAR␥ ABC and 1.5 times for BDGal4-hPPAR␥ DEF in presence of 60 and 100 ng of co-transfected pCMV p300-CHA, respectively. The AF-1 activity was stimulated in the absence of ligand, whereas BRL 49,653 was required for the stimulation of p300/CBP and PPAR␥ Interact through Multiple Domains the AF-2 activity. Surprisingly, the BDGal4-hPPAR␥ DEF chimera, lacking the ABC region, not only was weakly co-activated by p300 but also needed higher amounts of BRL 49,653 than the full-length receptor to be fully activated (10 Ϫ6 versus 10 Ϫ7 M, respectively). p300 and PPAR␥ Interact in a Cellular Context-The enhancement of PPAR␥ transcriptional activity by p300 suggested that the two molecules are part of the same protein complex driving gene expression. To verify this, co-immunoprecipitation experiments were carried out. HeLa cells were therefore transfected with different combinations of the pSG5-hPPAR␥2 and CMV p300-CHA expression constructs and of the corresponding empty expression vectors. PPAR␥ was then immunoprecipitated from the cell nuclear extracts with antibodies directed against its AB domain, either in presence or in absence of BRL 49,653. The immunoprecipitates were analyzed by immunoblotting using anti-HA antibodies (Fig. 3). A clear band corresponding to the p300-CHA protein with an approximate molecular mass of 270 kDa was observed only for the immunoprecipitates from cells co-transfected with both PPAR␥ and p300-CHA. For the immunoprecipitates from cells which had been transfected either by PPAR␥ or p300-CHA alone, no clear band was visible in the immunoblot. This specific coimmunoprecipitation of p300-CHA with PPAR␥ suggests that PPAR␥ and p300 associate in the cell. A 2-fold increase in the amount of immunoprecipitated p300-CHA was observed when BRL 49,653 was added.
p300 Interacts in Vitro with the AB and the DEF Domains of PPAR␥2-The association of p300 with PPAR␥ in a cellular environment could be due either to a direct interaction between the two molecules or to the interaction of both of them with a third partner, either a cofactor such as SRC-1, or a nuclear receptor such as RXR. To test the hypothesis of a direct interaction, pull-down experiments with purified proteins were carried out. The domain by which p300 interacts with the LBDs of other nuclear receptors has been localized in the N-terminal part of the protein (11,12). To verify that the same domain is involved in the interaction of p300 with PPAR␥, the N-terminal part of p300 (amino acids from 2 to 516) was produced as a GST fusion protein in E. coli and purified. The PPAR␥2 AB1-146 (aa from 1 to 146 of PPAR␥), the PPAR␥2 ABC1-181 (aa from 1 to 181 of PPAR␥), and the PPAR␥2 DEF204 -507 (aa from 204 to 507 of PPAR␥) proteins were also produced and purified following the same procedure. The GST part of these proteins was then removed by thrombin cleavage. p300Nt-GST interacted with both the AB and the DEF domains of PPAR␥2 but following two different modes: in a ligand-independent way with the AB domain (Fig. 4A) and in a ligand-dependent way with the DEF domain of PPAR␥2 (Fig. 4B). The ligand-dependent interaction between the PPAR␥2 DEF and p300Nt-GST was enhanced by increasing amounts of BRL 49,653. Similar data were obtained when another synthetic PPAR␥ ligand, troglitazone, was used (data not shown). No interaction was detected with the GST protein alone.
As each sub-region of PPAR␥2 apparently displayed different FIG. 1. p300 enhances PPAR␥2 transcriptional activity in HeLa cells. HeLa cells were co-transfected with an expression vector for hPPAR␥2 (0.1 g/ well), increasing amounts of an expression vector for p300, and with either the pGL3-(J wt ) 3 TKCAT or the pGL3-TKCAT reporter constructs (1 g/well). They were then grown during 24 h in the presence or absence of 10 Ϫ7 M BRL 49,653. The numbers above the shaded bars indicate the -fold increase of the normalized CAT activity compared with control (ctrl). Each point was performed in triplicate, and this figure is representative of four independent experiments. DMSO, dimethylsulfoxide properties for the binding to p300Nt, we studied the overall mode of interaction between the full-length receptor and its co-activator. In a pull-down experiment, full-length PPAR␥2 produced with rabbit reticulocyte lysates or purified PPAR␥2 DEF were incubated with p300Nt-GST in presence or absence of BRL 49,653. Nonprogrammed reticulocyte lysate was added to the samples with purified PPAR␥2 DEF to rule out any artifact because of the potential presence of a PPAR␥ ligand in this crude lysate (used for the full-length PPAR␥). In presence of ligand, both full-length PPAR␥2 and PPAR␥2 DEF interacted with p300Nt (Fig. 4C), but in absence of ligand, only the interaction between full-length PPAR␥2 and p300Nt was substantial, indicating that the ABC domain was also involved in the interaction of the full-length nuclear receptor with its co-activator and giving a potential explanation for the important ligand-independent association of p300 and PPAR␥2 observed in the co-immunoprecipitation experiments.
CBP Interacts with the ABC Region of PPAR␥2 through Multiple Domains-Because the direct interaction in vitro of p300/CBP with the ABC domain of a nuclear receptor had never been studied so far, we investigated more precisely the regions in CBP susceptible to contact this part of the receptor. We performed pull-down experiments using the purified PPAR␥2 ABC1-181 or PPAR␥2 AB1-146 proteins and different subregions of CBP fused to the GST protein (Fig. 5, A and B). p300 and CBP contact the ABC domain of PPAR␥2 mainly through their N-terminal part, i.e. aa from 2 to 516 for p300 (Fig. 5A, lane 2), and aa from 1 to 113 for CBP (Fig. 5B, lane 4). Surprisingly, another domain located between amino acids 1099 and 1460 of CBP displayed a weaker though unambiguous interaction with the ABC domain of PPAR␥2 (Fig. 5A, lanes 9 and 10). It appears therefore that p300/CBP and PPAR␥2 can associate through multiple contact points. A constitutive interaction occurs in absence of any ligand because of the presence of the ABC domain. Upon ligand binding, the DEF domain also contacts the co-activator, thereby strengthening the association. It is noteworthy that the domain in the SRC-1 co-activator known to interact with the PPAR␥ ligand-binding domain (24) did not interact with the PPAR␥ N-terminal domain (Fig. 5B,  lane 6), suggesting that the interaction observed with p300/ CBP is specific.
p300 and PPAR␥ Interact in the Yeast Two-hybrid System-The yeast two-hybrid system provides a very sensitive and functional test to study interactions between p300 and PPAR␥. Therefore, the N-terminal part of p300 (aa from 2 to 516) was cloned downstream of the activating domain of the Gal4 transcription factor (pADGal4-p300Nt), whereas different parts of hPPAR␥ were cloned downstream of the DNA binding domain of the Gal4 protein (Figs. 6A and 7A. In yeast, the BDGal4-PPAR␥2 DEF181-507 and the ADGal4-p300Nt fusion proteins interact without addition of any PPAR␥ ligand. It is unclear whether this interaction is because of the presence of potential PPAR␥ ligands in the yeast cells or whether a constitutive interaction between the DEF domain of PPAR␥ and the Nterminal part of p300 can actually occur in absence of any ligand in vivo (Fig. 6A). This interaction is disrupted when the   FIG. 2. p300 co-activates the PPAR␥2 AF-1 and AF-2 transcriptional activity in RK13 cells. A, pcDNA3-BDGal4-hPPAR␥ ABC is a construct where the A, B, and C regions of PPAR␥2 (aa from 2 to 181) have been cloned downstream of the Gal4 DNA binding domain and pcDNA3-BDGal4-hPPAR␥ DEF is a construct where the D, E, and F regions of PPAR␥2 (aa from 181 to 507) have been cloned downstream of the Gal4 DNA binding domain. HeLa cells were co-transfected with expression vectors for the BDGal4, BDGal4-hPPAR␥ AB (B), or BDGal4-hPPAR␥ DEF (C) chimeric proteins (5 ng/well), increasing amounts of an expression vector for p300, and the pGL3-(Gal4) 5 TKLuc reporter construct (1.5 g/well). Cells were then grown 24 h in the presence or absence of 10 Ϫ6 M BRL 49,653. The histograms present the transcriptional activity of the BDGal4-hPPAR␥ AB or BDGal4-hPPAR␥ DEF chimeric proteins compared with the activity of the BDGal4 protein in the same conditions. The numbers above the shaded bars indicate the relative increase when p300 was added compared with control (ctrl). Each point was performed in triplicate, and the means and standard deviations where calculated with data from three independent experiments. Comparisons between groups were made by nonparametric Mann-Whitney tests. *, indicates a statistically significant difference (p Ͻ 0.05) with control. DMSO, dimethylsulfoxide.

FIG. 3. p300 and PPAR␥ interact in vivo.
HeLa cells were transfected either with an empty expression construct or with expression vectors coding for hPPAR␥2 or p300-CHA. hPPAR␥2 was immunoprecipitated from the nuclear extracts with an antibody directed against the AB domain of PPAR␥ after incubation of the samples in absence or presence 10 Ϫ6 M BRL 49,653. The immunoprecipitates were separated on an 8% SDS-PAGE gel and immunoblotted with anti-HA.11 antibodies.
AF-2 domain of PPAR␥ is deleted, pointing to an important role for this domain in the interaction between the two molecules.
Yeast co-transfected with bait vectors containing different regions of the ABC domain of hPPAR␥2 and the pADGal4-p300Nt vector can also grow on histidine-deficient plates (Fig.  7A), confirming that p300Nt interacts with the ABC domain of PPAR␥2. The different constructs used suggest that the interaction domain in PPAR␥2 is located between aa 31 and 99 and that the B exon of PPAR␥2 is not required for this interaction.
Beside the HIS3 reporter system, YRG-2 yeast cells also have a Gal4-dependent lacZ reporter system that can be quantified more easily. We used that quantitative system to further investigate the effect of the presence of a PPAR␥ ligand on the strength of the interaction between p300Nt and PPAR␥2 DEF181-507 or PPAR␥2 ABC1-182 (Figs. 6B and 7B). Similar to the pull-down experiments, two distinct mechanisms of FIG. 4. p300 and CBP interact directly with the ABC and the DEF domains of PPAR␥in vitro. A, the purified PPAR␥2 AB1-146 and PPAR␥2 ABC1-181 proteins were incubated with purified GST or p300Nt-GST protein and glutathione-Q-Sepharose beads in presence or absence of BRL 49,653. The beads were then washed and the samples separated on a 12% SDS-PAGE gel. Blots were developed with antibodies directed against the AB domain of PPAR␥2. B, the purified PPAR␥2 DEF protein was incubated with purified p300Nt-GST protein, glutathione-Q-Sepharose beads, and different concentrations of BRL 49,653. The beads were then washed and the samples separated on a 12% SDS-PAGE gel. Blots were developed with antibodies directed against the DEF domain of PPAR␥2. C, full-length PPAR␥2 synthetized with rabbit reticulocyte lysates or purified PPAR␥2 DEF protein mixed with nonprogrammed lysate were incubated with the purified p300Nt-GST or GST proteins and glutathione-Q-Sepharose beads in presence or absence of BRL 49,653. The beads were then washed and the samples separated on a 4 -20% SDS-PAGE gel. Blots were developed with antibodies directed against the DEF domain of PPAR␥2.

FIG. 5. PPAR␥2 ABC interacts with different domains of CBP.
A, the purified PPAR␥2 ABC1-181 protein was incubated with the purified p300Nt-GST protein or CBP-GST constructs and glutathione-Q-Sepharose beads. The beads were then washed and the samples separated on a 12% SDS-PAGE gel. Blots were developed with antibodies directed against the AB domain of PPAR␥2. B, the purified PPAR␥2 ABC1-146 protein was incubated with the purified p300Nt-GST, CBP1-113-GST, or SRC568 -780-GST proteins and glutathione-Q-Sepharose beads. The beads were then washed and the samples separated on a 12% SDS-PAGE gel. Blots were developed with antibodies directed against the AB domain of PPAR␥2.
interaction for the two domains of PPAR␥2 and p300 were observed. PPAR␥2 ABC and p300Nt interact in absence of any ligand and the addition of BRL 49,653 has no effect on this interaction. In contrast, although hPPAR␥ DEF and p300Nt can interact in absence of any ligand, a significant increase of the ␤-galactosidase activity is observed in presence of BRL 49,653, suggesting an enhancement of the interaction. DISCUSSION Among the cofactors shown to interact with several transcription factors, the homologous molecules p300 and CBP are two of the most studied co-activators (19,20). Concerning nuclear receptors, there is now evidence for a direct interaction between p300/CBP and the estrogen receptor (ER), the retinoic acid receptor (RAR), RXR, and the thyroid hormone receptor (TR) (11,12,15,16). It has been shown recently that p300 could interact with PPAR␥2 DEF (29) and that p300 could enhance PPAR␣ ligand-dependent transcriptional activity by binding to its DEF domain (21). The scope of this work was to evaluate the role of p300 toward both the ligand-dependent and ligandindependent transcriptional activities of PPAR␥.
In a series of transfection experiments, PPAR␥2-transcriptional activity was enhanced in the presence of p300. The observation that full-length PPAR␥ is more activated than PPAR␥2 ABC or PPAR␥2 DEF alone sustains the hypothesis that nuclear receptors AF-1 and AF-2 are co-activated in a cooperative manner and confirms previous observations made with the progesterone receptor and SRC-1 (34). This result shows also for the first time that p300 is a co-activator of PPAR␥ and raises the question of whether p300 is crucial in PPAR␥2 physiology, most notably in fat tissue and colon (2) where PPAR␥ is expressed to high levels. We can speculate that some mutations or altered levels of expression of p300 or PPAR␥ could result in the disruption or the enhancement of their interaction and have important consequences in these processes. p300 or CBP have been shown to contact the DEF domains of the ER, RAR, RXR, TR, and PPAR␣ in a ligand-dependent manner (11,12,15,16,21). These interactions involve different domains of p300, all located in its N-terminal region (21). In this study, different parts of PPAR␥ were tested for their interaction with the N terminus of p300. A ligand-dependent FIG. 6. The N-terminal part of p300 interacts with the DEF domain of PPAR␥ in the yeast two-hybrid system. A, yeasts were co-transformed with bait vectors containing different parts of the DEF domain of hPPAR␥2 (pBDGal4 constructs) and with the pADGal4-p300Nt or pADGal4-SV40 constructs. Growth on a histidine-deficient media is indicated by a "ϩ." B, yeasts were co-transformed with the pBDGal4-PPAR␥2 DEF181-507 bait vector and the pADGal4-p300Nt or pADGal4-SV40 constructs and grown in absence or presence of BRL 49,653 (10 Ϫ6 M). The ␤-galactosidase activity was then measured in each yeast culture lysate. Data are presented as means of triplicates Ϯ standard deviations. The mean activity for the lysates from yeasts transformed with the pADGal4 vector and grown without BRL 49,653 was set to be 1. Comparisons between groups were made by nonparametric Mann-Whitney tests. *, indicates a statistically significant difference (p Ͻ 0.05) with the points where the empty pAD-Gal4 vector was used. DMSO, dimethylsulfoxide FIG. 7. The N-terminal part of p300 interacts with the ABC domain of PPAR␥ in the yeast two-hybrid system. A, yeasts were co-transformed with bait vectors containing different parts of the ABC domain of hPPAR␥2 (pBDGal4 constructs) and with the pADGal4-p300Nt or pADGal4-SV40 constructs. Growth on a histidine-deficient media is indicated by a "ϩ." B, yeasts were co-transformed with the pBDGal4-PPAR␥2 ABC1-182 bait vector and the pADGal4-p300Nt or pADGal4-SV40 constructs and grown in absence or presence of BRL 49,653 (10 Ϫ6 M). The ␤-galactosidase activity was then measured in each yeast culture lysate. Data are presented as means of triplicates Ϯ S.D. The mean activity for the lysates from yeasts transformed with the pADGal4 vector and grown without BRL 49,653 was set to be 1. Comparisons between groups were made by nonparametric Mann-Whitney tests. *, indicates a statistically significant difference (p Ͻ 0.05) with the points where the empty pADGal4 vector was used. DMSO, dimethylsulfoxide.
interaction between the N-terminal part of p300 and the DEF domain of PPAR␥2 was demonstrated, and the importance of the AF-2 domain in this process has been highlighted (Fig. 7). Most interestingly, it was shown that the AB domain of PPAR␥2 also displays a docking site for p300 and that this interaction is ligand-independent. This result is not surprising because p300 is already known to interact with transcription factors that do not have any ligand such as CREB (35) or the AP-1 complex (12,36,37). The AB region of nuclear receptors, which includes the AF-1 domain, hence could be considered as a ligand-independent interaction surface. Such a ligand-independent interaction through the AB region might not be restricted to PPAR␥ and p300/CBP. Indeed, CBP as well as the F-SRC-1 and RIP140 co-activators have been shown to coactivate the androgen receptor (AR) AF-1 activity (38) even though the existence of a direct interaction between CBP and the N terminus of the AR has not been demonstrated. Along the same lines, it was recently shown that also the estrogen receptor AF-1 works by binding p160 co-activator proteins (39). Another argument for different docking domains in PPAR␥ for cofactors comes from a paper by Puigserver et al. (27) who recently cloned a new cofactor that interacts only in a ligandindependent way with part of the DNA-binding and hinge domains of PPAR␥.
The existence of a ligand-independent interaction between p300 and the AB domain of PPAR␥2 might also indicate that beside the PPAR␥ ligands the interaction between PPAR␥ and p300 is susceptible to modulation by other transduction pathways. Indeed, it has previously been demonstrated that p300 could undergo phosphorylation during cell differentiation (40) and that the state of phosphorylation of p300 could orientate the interactions with its different partners (41). PPAR␥ activity itself is also subject to regulation by phosphorylation (5)(6)(7)(8)(9). During adipocyte differentiation for example, p300 or PPAR␥ could each be differentially phosphorylated and/or could undergo other post-translational modifications that might drastically affect their ability to interact and control transcription.
Several studies demonstrated the crucial role played by the N-terminal domain of nuclear receptors for promoter-and cellspecificity determination, ligand-dependent transactivation, and recruitment of co-activator (34,38,(42)(43)(44). Our results further sustain the idea that cofactor recruitment by the LBD of nuclear receptors is influenced by their N-terminal part and show that this could be because of the presence in this domain of one or several docking sites for these cofactors. More generally, the finding that domains other than the LBD of PPAR␥ and the N terminus p300/CBP are involved in the interaction between the two molecules (Figs. 4 and 6) points out that their dimerization (and most likely the dimerization of p300/CBP with other nuclear receptors) is more complex than was previously thought to be and complement the growing body of evidence showing that associations between nuclear receptors, cofactors, and the basic transcription machinery involve multiple molecular interaction domains (or 'boxes') in each partner (45)(46)(47)(48). We recently obtained preliminary results showing a direct interaction in vitro between the N-and C-terminal regions of PPAR␥2, an interaction strengthened by the presence of p300. 2 It is tempting to speculate that, by contacting both the AB and the DEF regions of nuclear receptors, p300 could act as an adaptor and could favor the interaction between the AF-1 and the AF-2 regions, thereby allowing for full receptor activation. p300 could favor this interaction by simply bringing both domains next to another. It is also possible that p300 induces concomitantly a conformational change in either one or both domains thereby unmasking new interfaces for interaction.
In conclusion, this study has demonstrated that p300 is a bona fide PPAR␥ co-activator and constitutes a node at which the PPAR␥-mediated transduction pathway is susceptible to be integrated in a more complex cellular signaling system. This work points also to a new ligand-independent mode of interaction between the two molecules. If these interactions are to modify PPAR␥ conformation, it will be of interest to test whether it also interferes with ligand-or responsive element-specificity.