p300 Functions as a Coactivator for the Peroxisome Proliferator-activated Receptor α*

The integrator protein, p300, was demonstrated to interact with mouse peroxisome proliferator-activated receptor α in a ligand-enhanced manner. The PPARα-interacting domain of p300 was mapped to amino acids 39–117 which interacted strongly with PPARα but did not interact with retinoic acid receptor-γ or retinoid X receptor-α. Amino acids within the carboxyl terminus of PPARα as well as residues within the hinge region were required for ligand-dependent interaction with p300. p300 enhanced the transcriptional activation properties of PPARα and, therefore, can be considered a bona fide coactivator for this nuclear receptor. These observations extend the group of p300-interacting proteins to include mPPARα and further characterize the molecular mechanisms of PPARα-mediated transcriptional regulation.

Mammalian peroxisome proliferator-activated receptors (PPARs) 1 are ligand-inducible transcription factors comprised of Ϫ␣, Ϫ␦ (also known as FAAR and NUC1), and Ϫ␥ subtypes that regulate expression of genes involved in lipid homeostasis (reviewed in Refs. 1 and 2). PPARs, along with the steroid, thyroid, and retinoid receptors, all contain conserved protein regions that function in transcriptional activation, DNA binding, and homo-and/or heterodimerization (reviewed in Refs. 3 and 4). PPARs form a complex with the common heterodimeric partner, retinoid X receptor (RXR), thereby establishing a high affinity DNA binding complex that preferentially binds to a degenerate direct repeat of the canonical AGGTCA sequence separated by 1 base pair (DR1, Refs. [5][6][7][8][9]. The PPAR⅐RXR complex appears to be responsive to both PPAR and RXR ligands with synergistic activation profiles when ligands for both receptors are bound (5)(6)(7)(8)(9).
The molecular mechanism(s) of transcriptional activation mediated by nuclear receptors, including PPAR⅐RXR complexes, has not been fully elucidated. However, evidence has accumulated indicating that ligand may promote a receptor conformation that favors interaction with transcriptional coactivators while promoting dissociation from receptor-associated corepressors, yielding a transcriptionally active complex (see Refs. 23 and 24 and references therein). The functional significance of receptor interactions with coactivators or corepressors and the resulting transcriptional activation or repression, respectively, may involve chromatin remodeling (reviewed in Ref. 25). The coactivators CBP and p300 have been shown to interact with the histone acetyltransferase, P/CAF (26), in addition to possessing intrinsic histone acetyltransferase activity (27,28). Conversely, receptor-corepressor complexes have been shown to associate with histone deacetylase 1 (HDAC1 in Refs. 29 and 30). The resulting covalent modification of histones, either acetylation or deacetylation, may result in alterations within chromatin structure thereby facilitating or inhibiting, respectively, access of transcription factors and/or the core transcriptional machinery to DNA templates (reviewed in Ref. 31). It should also be noted that both receptors (32)(33)(34) and associated coactivators (35)(36)(37) have been shown to interact directly with components of the basal transcription machinery suggesting that recruitment of the transcriptional apparatus, in addition to chromatin remodeling, may contribute to the molecular mechanisms of transcriptional activation. It remains to be determined if additional mechanisms are operant and, if so, the relative importance of these additional pathways.
Two main families of transcriptional coactivators that interact with nuclear receptors have been identified as follows: 1) the integrator protein family including CBP and p300 (38 -41), and 2) the SRC-1 family of proteins that includes SRC-1 (40 -42), TIF2/GRIP1 (43,44), RAC3 (45), AIB1 (46), ACTR (47), and P/CIP (48). Proteins within the SRC-1 family appear to interact exclusively with transcription factors of the nuclear receptor superfamily (48). In contrast, CBP and p300, which appear to * This work was supported in part by Oregon Affiliate of the American Heart Association Grants OR-94-GS-16 and by NIEHS Grants ES00210 and ES00040 from the National Institutes of Health. 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.
¶ Supported by a predoctoral fellowship from the American Foundation for Pharmaceutical Education.
be functionally homologous proteins, have been shown to interact with a plethora of signaling molecules and, therefore, have been termed "integrator" proteins (Ref. 40 and reviewed in Ref. 49). In fact, several signaling pathways appear to converge at the level of CBP and/or p300 which may represent the molecular basis of reciprocal antagonism between AP-1 and 1) retinoid receptors (40), 2) signal transducers and activators of transcription (STATs, see Refs. 50 and 51), and 3) p53 (52,53). A third group of nuclear receptor coactivators that neither exhibit homology to each other nor to other classes of coactivators includes RIP140 (54), ARA70 (55), SUG-1/TRIP1 (56,57), and TIF1 (58). Although the diversity of nuclear receptor coactivators most likely contributes to the complexity of receptormediated signaling, the precise role of individual coactivators and possible synergies or antagonistic interactions between coactivators remains to be elucidated.
We have previously demonstrated that peroxisome proliferating agents such as WY-14,643 bind directly to and induce conformational change within the PPAR␣ ligand binding domain (LBD, see Ref. 59). We and others (59 -62) have hypothesized that this ligand-induced conformational change promotes interaction of the receptors with transcriptional coactivators. Toward the goal of identifying PPAR␣ coactivators, we have used a yeast two-hybrid system to isolate several proteins that interact with ligand-activated mouse PPAR␣ (mPPAR␣). One of the mPPAR␣-interacting proteins isolated in this screen was SRC-1, which has been demonstrated previously to interact with PPAR␥ (63,64). Another such protein isolated in our mPPAR␣ screen was p300 which serves as a transcriptional coactivator for many other transcription factors, including nuclear receptors (38,40,65). In this report, we focus on the interaction of p300 with mPPAR␣ in yeast, in mammalian cells, and in vitro, and we demonstrate that the manner by which p300 interacts with mPPAR␣ appears to be distinct from that of either retinoic acid receptor (RAR) or retinoid X receptor (RXR).

MATERIALS AND METHODS
Plasmids and Receptor Constructs-Plasmids encoding the receptors and reporters described below were used either directly or as templates for PCR to assemble all constructs described herein using standard techniques. All plasmids were kind gifts from the following individuals: mouse PPAR␣ (10) from Drs. S. Green  Vectors for the yeast two-hybrid screen (pBL1, pBTM116, and pASV, see Ref. 58), the yeast reporter strains PL3(␣) and L40 (Refs. 58 and 70, respectively), and the randomly primed mouse cDNA library inserted into the VP16 acidic activation domain (AAD)-encoded pASV3 vector (58) were all kind gifts from Drs R. Losson and P. Chambon (IGBMC, Illkirch, France). All yeast expression vectors encoding receptor baits were constructed by PCR amplification of the corresponding receptors (mPPAR␣, hRAR␥, or mRXR␣) with primers containing appropriate restriction sites for insertion into the XhoI and  (66) and the corresponding proteins were expressed by in vitro transcription/translation for use in electrophoretic mobility shift assays and GST pull-down experiments as described previously (59).
The construct encoding GST/mRXR␣ has been described (59). The construct encoding GST/hRAR␥ was prepared by insertion of a BamHI fragment encoding full-length hRAR␥ (67) into pGEX-2T (Pharmacia). Constructs encoding GST/p300 and GST/CBP fusion proteins were gen-erated by PCR amplification of the corresponding coding regions from pBSK/p300 (68) and pRc-RSV/CBP (69), respectively, with all forward primers containing a BglII site and reverse primers containing an EcoRI site. PCR products were then digested appropriately and inserted into BamHI/EcoRI-digested pGEX-2T.
The ER DBD fusion expression vectors used in HEK 293 transient transfection experiments (see below), ER DBD and ER DBD/mPPAR␣, were constructed by inserting the EcoRI fragment (including initiator methionine and favorable translation initiator sequence) of pBL1 and pBL1/mPPAR␣, respectively, into the EcoRI site of pTL1. A mammalian expression vector for full-length p300 was generated by inserting the NotI/HindIII fragment of pBSK/p300 (68) into the NotI/HindIII sites of pTL2. The region of p300 encoding amino acids 39 -221 was amplified by PCR with a 5Ј primer introducing a HindIII site, favorable translation initiator sequence, and initiator methionine and a 3Ј primer introducing a BglII site and stop codon. The PCR product was digested with HindIII and BglII and inserted into the HindIII/BglII site of pTL1 generating p300 dominant negative (dn). The ER-responsive CAT reporter, 17M-ERE-globin-CAT, has been described previously (71). The integrity of all constructs was verified by restriction digest and/or sequence analysis. Additional information concerning any of the constructs described herein can be obtained upon request.
Yeast Two-hybrid Screening-The Saccharomyces cerevisiae PL3(␣) reporter strain expressing the histidine-selectable pBL1/mPPAR␣ plasmid was co-transformed with a mouse embryo cDNA library fused to the VP16 AAD (on a leucine-selectable plasmid; see Ref. 58 and references therein for details on this two-hybrid system). Approximately 1 ϫ 10 6 yeast transformants (quantitated on His Ϫ Leu Ϫ plates) were screened for URA3 reporter activity on His Ϫ Leu Ϫ Ura Ϫ plates containing 30 g/ml 6-azauracil. Preliminary results indicated that the pBL1/mP-PAR␣ expressing PL3(␣) strain exhibited residual URA3 expression sufficient to allow growth on Ura Ϫ media (data not shown); therefore, inclusion of 6-azauracil (an inhibitor of the URA3 gene product) was necessary to restore uracil auxotrophy. Ligand-independent and -dependent screens were conducted in the absence and presence of 100 M WY-14,643, respectively, in all synthetic media. Positive clones from both screens were isolated and replated on Leu Ϫ plates to allow loss of the histidine-selectable pBL1/mPPAR␣ plasmid which was verified by replica plating on His Ϫ Leu Ϫ plates. Leu ϩ Ura Ϫ His Ϫ Trp Ϫ segregants were then mated to the S. cerevisiae L40 reporter strain expressing the tryptophan-selectable pBTM116/mPPAR␣ plasmid. Diploids from the mating were then selected on Leu Ϫ Trp Ϫ plates and screened for HIS3 reporter activity on Leu Ϫ Trp Ϫ His Ϫ plates containing 10 mM 3-azatriazole (an inhibitor of the HIS3 gene product) in the absence and presence of 100 M WY-14,643. Positive clones were individually grown as liquid cultures in Leu Ϫ media to allow loss of the tryptophan-selectable pBTM116/mPPAR␣ plasmid and verified by plating on Leu Ϫ Trp Ϫ plates. Plasmid DNA from positive clones was then isolated, and the resulting mouse embryo cDNAs were re-tested for interaction with the PPAR␣ baits, bait DBDs alone, and non-PPAR related baits. All cDNA clones that exhibited a specific interaction with the PPAR␣ baits were then sequenced using the standard dideoxynucleotide chain termination method.
␤-Galactosidase Assays-Assays were conducted essentially as described in Ref. 72 with minor modifications. Briefly, yeast cultures derived from single colonies were grown in selective media for 24 h and then diluted into media containing either vehicle or appropriate ligands and grown for an additional 3 h. Cells were permeabilized for 30 min at room temperature in buffer Z (100 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , pH 7.8, 10 mM MgSO 4 , 1 mM KCl, 35 mM ␤-mercaptoethanol) containing 0.1% (w/v) sarcosyl. ␤-Galactosidase assays were initiated by addition of o-nitrophenyl-␤-D-galactopyranoside, and reactions were carried out at 30°C until a visible color change was evident (0.5-6 h). ␤-Galactosidase activity was determined as indicated in Equation 1.
where A 420 is derived from the colorimetric ␤-galactosidase assay, t is the time in hours, and N corresponds to cell number. Ligand dependent assays were carried out by inclusion of 1 M 9-cis RA, 100 M WY-14,643 (final concentrations), or vehicle. GST Fusion Protein Production-All GST fusion proteins were produced and crude bacterial lysates prepared as described previously (59) except the bacteria were grown in 2 ϫ YT supplemented with glucose instead of LB media.
Protein-Protein Interaction Assays-GST fusion proteins were partially purified using glutathione-Sepharose 4B (Pharmacia Biotech Inc.) as per manufacturer's recommendations and used directly in protein-protein interaction studies. Assays were initiated by addition of 2 l of [ 35 S]methionine-labeled receptor to GST fusion protein-bound resin equilibrated in binding buffer (10 mM HEPES-NaOH, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, 100 mM NaCl, 10% glycerol, 0.1% Nonidet P-40), and incubated with rotation at 4°C for 2 h. Unbound proteins were removed by 3 sequential washes with binding buffer containing vehicle or the appropriate ligand. Bound proteins were eluted from the resin by addition of 30 l of SDS sample buffer and boiled. Samples were then analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography as described previously (59). Ligand-dependent assays were carried out by inclusion of 1 M 9-cis-RA, 10 M WY-14,643 (final concentrations), or vehicle in binding buffer. Representative gels were stained with Coomassie Blue before being subjected to autoradiography to ensure that equal amounts of GST fusion proteins were included in each reaction (data not shown). Independent experiments were conducted at least three times and representative results are shown.
Protein-protein interaction assays using P19 embryonal carcinoma cell nuclear extracts were conducted as described above except that GST or GST/p300  was incubated with P19 extracts (see below) instead of [ 35 S]methionine-labeled receptor preparations, and liganddependent assays were carried out in binding buffer containing 100 M WY-14,643 or vehicle. Bound proteins were analyzed by immunoblot analysis as described previously (62) using a polyclonal antibody that recognizes an epitope in the highly conserved carboxyl-terminal region of PPAR subtypes (Affinity Bioreagents, catalog PA3-820).
Electrophoretic Mobility Shift Assays-Electrophoretic mobility shift assays were conducted essentially as described previously (59) except receptor-DNA complexes were incubated with ϳ500 ng of purified GST or GST/p300-(39 -117) for an additional 10 min. The acyl-CoA oxidase peroxisome proliferator response element (PPRE) probe has been described previously (14). Ligand-dependent assays were carried out by inclusion of 1 M 9-cis-RA, 10 M WY-14,643, or vehicle.
Cell Culture and Nuclear Extract Preparation-Human embryonic kidney (HEK) 293 cells (ATCC CRL 1573) were cultured in minimum essential medium with Earle's salts (Sigma) supplemented with 10% horse serum and 4 mM glutamine (both from Life Technologies, Inc.). Cells were grown to 50 -60% confluence and transiently transfected using the calcium phosphate method. Each 100-mm plate was cotransfected with 0.5 g of expression vector encoding the ER DBD or ER DBD/PPAR␣ fusion protein, 2 g of the 17M-ERE-globin-CAT reporter construct, variable amounts of either wild-type p300 or p300dn, and 2 g of pCH110 (Pharmacia) encoding ␤-galactosidase which was used to normalize for transfection efficiency. 24 h after transfection, cells were incubated with WY-14,643 (10 M) or vehicle and grown for an additional 24 h. ␤-Galactosidase and CAT activities were quantified spectrophotometrically and by thin layer chromatography/scintillation counting, respectively, using standard techniques.
P19 embryonal carcinoma cells (73) were cultured on 100-mm plates in minimum essential medium supplemented with 10% fetal bovine serum (Hyclone). Nuclear extracts were prepared from cells grown to approximately 80% confluence essentially as described in Ref. 74. Briefly, cells were washed once with phosphate-buffered saline, scraped from plates, and resuspended in cold buffer A (10 mM HEPES-KOH, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride). Cells were allowed to swell on ice for 10 min, vortexed, and centrifuged to pellet nuclei. The pellet was resuspended in buffer C (20 mM HEPES-KOH, pH 7.9, 25% glycerol, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride) and incubated on ice for 20 min. Cellular debris were pelleted by centrifugation, and the supernatant containing the crude nuclear extract was frozen at Ϫ80°C.
Chemicals and Reagents-WY-14,643 was purchased from Chemsyn (Lenexa, KS), and all radioisotopes were purchased from NEN Life Science Products.

RESULTS
Using a previously described yeast two-hybrid system (58), two mPPAR␣-interacting clones were isolated, sequenced, and determined to encode fragments of the nuclear receptor coactivators, p300 (40) and SRC-1 (Ref. 42, Fig. 1, A and B). Both isolated fragments contain at least one copy of the recently identified LXXLL motif that has been shown to mediate interactions between coactivators and nuclear receptors (see Fig. 1, A and B, Refs. 48, 75, and 76). Although an interaction between PPAR␥ and SRC-1 has been reported previously (63,64), to our knowledge, this is the first report of a p300-PPAR␣ interaction, and for this reason, we chose to conduct further experiments with PPAR␣ and p300.
p300 Amino Acids 39 -221 Interact with PPAR␣ but Not RXR␣ or RAR␥-Interestingly, the isolated fragment of p300 (encoding amino acids 39 -221, Fig. 1A) that exhibited a strong interaction with the PPAR␣ bait did not interact with either RXR␣ or RAR␥ baits in the absence or presence of 9-cis-retinoic acid (9-cis-RA) in yeast ( Fig. 2A). However, the isolated fragment of SRC-1 (encoding amino acids 579 -932, Fig. 1B), which interacted with PPAR␣ in a ligand-enhanced manner, did exhibit a strong ligand-dependent interaction with both RXR␣ and RAR␥ demonstrating that all three baits are expressed in a functional form in this yeast strain ( Fig. 2A). Compared with the strong 9-cis-RA-dependent interaction observed between both the RXR␣ and RAR␥ baits and the SRC-1 fragment, stimulation of the interaction between the PPAR␣ bait and both coactivator fragments by the PPAR ligand, WY-14,643, was modest yet significant (p Ͻ 0.05, Fig. 2A). A strong interaction between PPAR␣ and both SRC-1 and p300 was observed in the absence of WY-14,643 ( Fig. 2A) consistent with observations previously reported for the interaction of PPAR␥ and SRC-1 (63). This may be due, at least in part, to the presence of high levels of endogenous PPAR activators in yeast, such as fatty acids and/or derivatives of fatty acids.
To confirm the observed interactions in vitro, p300-(39 -221) was expressed in bacteria as a GST fusion protein and immo- p300 Interacts with mPPAR␣ bilized on glutathione-Sepharose beads. The immobilized protein was then used as an affinity matrix to examine interactions with in vitro translated PPAR␣, RXR␣, and RAR␥. As observed in the yeast two-hybrid system, PPAR␣ exhibited a strong specific interaction with the GST/p300-(39 -221) fusion. However, the in vitro interaction was strictly dependent on the presence of WY-14,643 (Fig. 2B, lanes 2 and 3). This ligand-dependent interaction was in contrast to that of the ligand-independent interaction observed between GST/RXR␣ and PPAR␣ (see below, Fig. 4A, lanes 4 and 5). RAR␥ and RXR␣ did not interact with the GST/p300-(39 -221) fusion protein in the absence or presence of 9-cis-RA (Fig. 2B, lanes 5, 6, 8, and 9, respectively) in agreement with the yeast two-hybrid data ( Fig.  2A). The integrity of the in vitro translated RXR␣ and RAR␥ proteins was confirmed by interaction with GST/RAR␥ and GST/RXR␣ fusion proteins, respectively (see below, lanes 4 and 5 of Fig. 4, B and C, respectively), and DNA binding activity (data not shown). Therefore, amino acids 39 -221 of p300, which interacted strongly with PPAR␣, were not sufficient for interaction with the retinoid receptors, RXR␣ and RAR␥, as detected in our assays in yeast and in vitro. It should be noted, however, that other groups have reported CBP-RXR␣ (38), CBP-RAR␣, and p300-RAR␣ (40) interactions using yeast twohybrid systems. The region of p300 examined in our studies is partially overlapping, but not identical to, those (or homologous regions of CBP) used by Kamei and co-workers (40) and Chakravarti and collaborators (38).
Residues within the Carboxyl-terminal and Hinge Regions of PPAR␣ Are Required for Interaction with p300 -To determine which regions of PPAR␣ are required for interaction with p300-(39 -221) in yeast, we constructed three additional PPAR␣ baits as follows: 1) PPAR␣⌬448-(91-447) which is missing 21 carboxyl-terminal amino acids (in addition to the 90 aminoterminal residues which are also truncated in PPAR␣; see "Materials and Methods"), 2) PPAR␣ D/E-(166 -468) encompassing the hinge and ligand binding domain (LBD), and 3) PPAR␣ E-(281-468) encompassing the LBD alone as defined by Issemann and Green (10). When these baits were examined in yeast, we were unable to detect an interaction between p300-(39 -221) and either PPAR␣⌬448 or PPAR␣ E in the absence or presence of WY-14,643 (Fig. 3A). In contrast, a strong interaction was detected with both SRC-1 and p300 using the PPAR␣ D/E bait, and these interactions were modestly but significantly enhanced by WY-14,643 (Fig. 3A), as observed for PPAR␣ ( Fig. 2A).
Protein-protein interaction assays, as described above, were also conducted with in vitro translated PPAR␣⌬448, PPAR␣ D/E, and PPAR␣ E. In agreement with our yeast two-hybrid data, no interaction was detected between GST/p300-(39 -221) and either PPAR␣⌬448 or PPAR␣ E (Fig. 3B, lanes 2, 3, 8, and  9). In contrast to PPAR␣⌬448 and PPAR␣ E, a strong interaction was detected between GST/p300-(39 -221) and PPAR␣ D/E (Fig. 3B, lanes 5 and 6). As observed with PPAR␣, the in vitro interaction between p300-(39 -221) and PPAR␣ D/E was more strongly ligand-enhanced than that in yeast. Our yeast twohybrid and in vitro data suggest that the carboxyl-terminal 21 amino acids as well as residues within the hinge of PPAR␣ are required for interaction with p300- (39 -221).
Amino Acids 39 -117 of p300 Are Sufficient for Interaction with PPAR␣-To define further the region of p300 required for

FIG. 2. Interactions between coactivator fragments and PPAR␣, RAR␥, and RXR␣.
A, interactions between coactivator fragments and nuclear receptors in yeast. LexA DBD/receptor baits (PPAR␣, RAR␥, and RXR␣) were coexpressed with VP16AAD/coactivator fragment fusions (SRC-1 and p300) and examined for the ability to activate an integrated LexA-responsive lacZ reporter in the yeast strain L40. Ligand-dependent interactions were examined in the presence or absence of 100 M WY-14,643 (PPAR␣ bait), 1 M 9-cis-RA (RAR␥ and RXR␣ baits), or vehicle where indicated. At least three independent transformants were assayed for each experiment, and the results shown represent the mean Ϯ S.E. of three independent experiments. Statistical significance at the 95% (p Ͻ 0.05) and 99% (p Ͻ 0.01) confidence levels are indicated by * and ** symbols, respectively. No reporter activity was detected when VP16AAD/coactivator fragment fusions were coexpressed with either LexA alone or LexA fused to an unrelated protein (data not shown). B, in vitro protein-protein interactions between coactivator fragments and nuclear receptors. A GST/p300 fusion protein corresponding to the isolated p300 fragment (GST/p300-(39 -221)) was bound to glutathione-Sepharose and used as an affinity matrix to examine interactions with in vitro translated [ 35 S]methionine-labeled PPAR␣, RAR␥, and RXR␣ in the absence and presence of ligand (10 M WY-14,643 for PPAR␣ or 1 M 9-cis-RA for RAR␥ and RXR␣ as indicated). All receptors preparations expressed in yeast or translated in vitro were truncated in the amino-terminal (⌬AB) region (see "Materials and Methods").

FIG. 3. Interactions between coactivator fragments and truncation mutants of PPAR␣.
A, interactions between p300 coactivator fragment and mutant PPARs in yeast. The 21-amino acid carboxylterminal truncation mutant, PPAR␣⌬448, and the amino-terminal truncation mutants, PPAR␣ D/E and PPAR␣ E, encompassing the hinge/LBD and LBD of PPAR␣, respectively, were examined as LexA DBD/receptor baits as described in Fig. 2A. B, in vitro protein-protein interactions between p300-(39 -221) and mutant PPARs. Assays were conducted as described in Fig. 2B. p300 Interacts with mPPAR␣ interaction with PPAR␣, several fragments of p300 were expressed as GST fusion proteins and examined for interaction with PPAR␣ as well as RXR␣ and RAR␥ as described above. PPAR␣, RAR␥, and RXR␣ all exhibited specific interactions with GST/p300 amino acids 1-450 (GST/p300-(1-450)), but the interaction with RXR␣ in the presence of 9-cis-RA was relatively weak (Fig. 4C, lanes 6 and 7) and was more readily apparent upon overexposure of the dried gel (data not shown). In contrast, RXR␣ exhibited a strong ligand-independent interaction with GST/RAR␥ (Fig. 4C, lanes 4 and 5) demonstrating the integrity of this preparation of RXR␣. RAR␥ consistently exhibited a ligand-enhanced interaction with GST-p300-(1-450) (Fig. 4B, lanes 6 and 7) in addition to a strong ligandindependent interaction with GST/RXR␣ (Fig. 4B, lanes 4 and  5). Surprisingly, the GST/p300-(1-450)-PPAR␣ interaction was unaffected by addition of WY-14,643 (Fig. 4A, lanes 6 and 7). In contrast, PPAR␣ exhibited a ligand-dependent interaction with GST/p300-(1-117) and -(39 -117) (Fig. 4A, lanes 8-10 and 11) as observed for GST/p300-(39 -221) (see above, Fig. 2B). The molecular basis for this variable degree of ligand dependence when examining PPAR␣ interactions with different regions of p300 and vice versa (see below) is presently unknown. Nonetheless, a minimal PPAR␣ interaction domain encompassing p300 amino acids 39 -117 was defined. In addition, results from experiments conducted with GST/CBP-(1-115) demonstrate that PPAR␣, but neither RAR␥ nor RXR␣, interact with CBP amino acids 1-115 in a ligand-dependent manner in vitro (data not shown). lt should be noted that although Kamei and coworkers (40) demonstrated an interaction between RAR␣ and p300-(1-117) in a yeast two-hybrid assay, we did not observe an interaction between RAR␥ and p300-(1-117) in vitro (Fig. 4B,  lanes 8 and 9). However, we did observe ligand-enhanced, in vitro interactions between both RAR␥ and RXR␣ and GST/ p300- (1-450) (lanes 6 and 7 of Figs. 4, B and C, respectively) in agreement with the observations of Kamei and collaborators (40). Discrepancies between our results and those published previously may simply reflect the different subtypes of RAR used (␣ versus ␥) and/or differential sensitivities when comparing the two assays. In fact, other groups have also reported conflicting results when comparing in vitro interactions with those in yeast (56).
We next conducted additional protein-protein interaction experiments to define regions of both PPAR␣ and p300 that are required for efficient interaction. PPAR␣⌬448, PPAR␣ D/E, and PPAR␣ E all exhibited a ligand-independent interaction with GST/p300-(1-450) (lanes 6 and 7 of Fig. 5, A, B, and C, respectively). PPAR␣ D/E, containing both the hinge and ligand binding domain, exhibited an interaction profile similar to that of PPAR␣ (compare Figs. 4A with 5B) except that the interaction of PPAR␣ D/E with GST/RXR␣ was enhanced by WY-14,643 (Fig. 5B, lanes 4 and 5). Neither PPAR␣⌬448 nor PPAR␣ E was capable of interacting with smaller fragments of p300 fused to GST (GST/p300-(1-117) and -(39 -117)) in the absence or presence of WY-14,643 (lanes 8 -11 of Fig. 5, A and C, respectively). Although a strong ligand-independent interaction between PPAR␣⌬448 and GST/mRXR␣ was observed (Fig. 5A, lanes 4 and 5), only a weak interaction between PPAR␣ E and GST/mRXR␣ was observed which was more apparent upon overexposure of the dried gel (data not shown). As demonstrated previously, PPAR␣⌬448 responded to WY-14,643 in a differential proteolytic sensitivity assay demonstrating that this mutant receptor is capable of binding ligand (59). Our in vitro results further support the conclusion that the carboxyl-terminal 21 PPAR␣ residues are required for interaction with the minimal PPAR␣ interaction domain of p300- (39 -117). In addition, residues contained within the hinge of PPAR␣-(166 -281) are also required for the interaction of the receptor with p300- (39 -117). In contrast, it appears that only PPAR␣ amino acids 281-447 are required for the ligand-independent interaction observed with p300-(1-450) as all PPAR␣ mutants examined were capable of interacting with this fragment regardless of the presence of ligand.
p300 Interacts with a PPAR␣/RXR␣ Heterodimer Bound to DNA-We next conducted electrophoretic mobility shift assays to determine if PPAR␣/RXR␣ heterodimers could interact with a p300 fragment while bound to DNA. Addition of GST/p300-(39 -117), but not GST alone, to a RXR␣⅐PPAR␣⅐PPRE complex resulted in a pronounced mobility shift indicating the forma-  Fig. 2B. B, in vitro protein-protein interactions between coactivator fragments and RAR␥. Assays were carried out as described above (A). C, in vitro protein-protein interactions between coactivator fragments and RXR␣. Assays were carried out as described above (A) except the integrity of the in vitro translated RXR␣ was demonstrated by interaction with GST/hRAR␥.

FIG. 5. Interactions between regions of p300 and truncation mutants of PPAR␣.
A, in vitro protein-protein interactions between coactivator fragments and PPAR␣⌬448. Assays were conducted as described in Fig. 4A. B, in vitro protein-protein interactions between coactivator fragments and PPAR␣ D/E. Assays were carried out as described above (A). C, in vitro protein-protein interactions between coactivator fragments and PPAR␣ E. Assays were carried out as described above (A).
p300 Interacts with mPPAR␣ tion of an oligomeric RXR␣⅐PPAR␣⅐GST/p300-(39 -117)⅐PPRE complex (Fig. 6, compare lanes 1 and 2 with lanes 5 and 6). The presumed oligomeric complex formation was only modestly stimulated by ligand, in contrast to results obtained from in vitro protein-protein interaction assays with this particular fragment of p300 (Fig. 4A). No binding of RXR␣ homodimers to the PPRE was observed excluding the possibility that the GST/ p300 protein was interacting with a PPRE-bound RXR homodimer (data not shown). In addition, no interaction between a canonical DR1-bound RXR␣ homodimeric complex and GST/ p300-(39 -117) was observed in the absence or presence of 9-cis-RA (data not shown). RXR␣/RAR␥ heterodimers similarly did not exhibit interactions with the GST/p300-(39 -117) protein when bound to either DR1 or DR5 response elements (data not shown).
p300 Enhances the Transcriptional Activation Properties of PPAR␣-To examine the effect of p300 on the transcriptional activation properties of PPAR␣, we utilized a fusion protein encoding the DBD of ER fused to amino acids 91-468 of PPAR␣ (ER DBD/PPAR␣) in conjunction with an ER-responsive CAT reporter. Coexpression of p300 in HEK 293 cells significantly enhanced the ability of the ER DBD/PPAR␣, but not that of the empty ER DBD expression vector, to activate transcription of CAT both in the presence and absence of WY-14,643 (Fig. 7). ER DBD/PPAR␣ activated transcription strongly in the absence of exogenous activator, and this activity was slightly stimulated by 10 M WY-14,643 (Fig. 7, lanes 1 and 2). A dominant negative form of p300, p300dn (corresponding to amino acids 39 -221), was used to address the role of endogenous p300 or that of other transcriptional coactivators, in the constitutive activity of ER DBD/PPAR␣ observed in HEK 293 cells. As observed with assays in yeast, endogenous PPAR activators may be responsible for the ostensibly ligand-independent activity of ER-DBD/PPAR␣. Coexpression of p300dn with ER DBD/PPAR␣ resulted in a striking inhibition of both ligand-independent and -dependent transcriptional activation properties of the receptor (Fig. 7, lanes 7-10). When considered together, our results suggest that p300 functions as a bona fide PPAR␣ coactivator by enhancing the transcriptional activation capabilities of the receptor. Conversely, p300dn appears to function as a potent dominant negative inhibitor of PPAR␣mediated activation, suggesting that endogenous p300 or possibly other transcriptional coactivators mediate the constitutive activity of PPAR␣ in HEK 293 cells.
p300 Interacts with Endogenous PPAR from Mammalian Cells-Finally, experiments were carried out to determine if a fragment of p300 was capable of interacting with endogenous PPAR expressed in a mammalian cell line. GST alone, GST/ mRXR␣, or GST/p300-(1-450) was bound to glutathione-Sepharose and incubated with nuclear extracts from P19 mouse embryonal carcinoma cells in the absence or presence of WY-14,643. Bound proteins were then eluted, electrophoresed, transferred to nitrocellulose, and subjected to immunoblot analysis using antiserum that recognizes all PPAR subtypes. A ligand-independent interaction was detected between endogenous PPAR and both GST/mRXR␣ and GST/p300-(1-450) (Fig.  8, lanes [5][6][7][8], in agreement with data discussed above (see Fig. 4A), whereas no interaction was observed with GST alone (Fig. 8, lanes 3 and 4). The immunoreactive signals we observed were specific for PPAR as preimmune serum was ineffective (Fig. 8. lane 1 and data not shown). Two P19 cell nuclear proteins that migrated with predicted molecular mass consistent with that of PPAR subtypes (ϳ51-55 kDa) were specifically retained by GST/p300-(1-450) (see arrows in Fig. 8). These proteins could represent different subtypes of PPAR or degradation products thereof. Alternatively, these proteins may represent unphosphorylated and phosphorylated forms of a single or multiple PPARs as recent reports have shown that both PPAR-␣ and -␥ subtypes are phosphorylated (77)(78)(79). Additional studies will be required to address these questions. Nonetheless, our results suggest that native PPAR expressed in mammalian cells is capable of interacting with p300. DISCUSSION Results described herein extend the group of p300/CBP-interacting proteins to include the nuclear receptor PPAR␣. We have shown that PPAR␣ residues within the carboxyl-terminal region of the receptor (amino acids 448 -468) are required for ligand-dependent interactions with the minimal PPAR␣ interaction domain of p300 (amino acids 39 -117). Based on the crystal structures of RAR␥ (80) and RXR␣ (81,82) and the predicted structural similarities between PPAR, RAR, and ]chloramphenicol converted to acetylated forms for a given treatment. Statistical significance at the 95% (p Ͻ 0.05) and 99% (p Ͻ 0.01) confidence levels are indicated by * and ** symbols, respectively. Note that PPAR⌬AB is ER DBD/PPAR␣. RXR (80 and data not shown), PPAR␣ amino acids 448 -468 correspond to a portion of putative helix 11 and all of putative helix 12 (see Refs. 59 and 80). Helix 12, which contains a conserved transcriptional activation function (AF-2), appears to fold back upon the core receptor molecule in the ligandbound state and may contribute part of a larger ligand-dependent interaction surface for receptor-interacting proteins such as coactivators. Indeed, point mutations within the AF-2 of other nuclear receptors have been shown to abolish or inhibit interactions with coactivators (38,40,43,57,58). In this model, deletion of helix 12 or mutation of key residues within helix 12 would effectively destroy part or all of the ligand-dependent coactivator interaction surface. The PPAR␣ mutant, PPAR␣⌬448, which is missing putative helix 12 but still binds WY-14,643 (59), does not interact with the minimal PPAR␣ interaction domain of p300 (amino acids 39 -117) in contrast to PPAR␣ which interacts with this region of p300 in a strictly ligand-dependent manner. Our results are in agreement with the model proposed above, and recent studies have demonstrated that deletion of 11 amino acids from the carboxylterminal region of PPAR␥ abolishes transcriptional activation although the effects on coactivator interactions were not examined (83). This region is well conserved across PPAR subtypes, and it will be of interest in future studies to determine if loss of transcriptional activation correlates with loss of coactivator interactions. Studies with other nuclear receptors have shown that the AF-2, while being required, is not sufficient for receptor-coactivator interactions (76) suggesting that other regions within the receptor may also be involved. Although we did not directly examine if the 21 carboxyl-terminal amino acids of PPAR␣ alone were sufficient for interaction with p300, we were unable to demonstrate an interaction between a larger PPAR␣ fragment containing this region (PPAR␣ E) and all p300 fragments examined except 1-450, and the latter interaction was entirely ligand-independent. The ligand-dependent nature of the interaction with p300 fragments 39 -221, 1-117, and 39 -117 was restored when the hinge region was added to the LBD (PPAR␣ D/E) suggesting that the extreme carboxyl-terminal region of the receptor as well as amino acids that comprise the hinge region are required for ligand-dependent interaction with the smaller p300 fragments. The region of mPPAR␣ referred to herein as the hinge, amino acids 166 -282, is predicted to include ␣ helices 1-4 based on structural similarity with the retinoid receptors (Ref. 80 and data not shown). Hsu and coworkers (84) have demonstrated that point mutation of Glu 282 3 Gly attenuated the constitutive transcriptional activation properties of mPPAR␣ highlighting the importance of this region in mPPAR␣ signaling. However, mutation of Glu 282 likely affected the ligand binding properties of the mPPAR␣ rather than its ability to interact with coactivators because the maximal extent of inducible transcriptional activation was unchanged (84). Similarly, the observed differential ability of mPPAR␣ E and mPPAR␣ D/E to interact with p300 fragments may be related to differing affinities of these two receptor regions for WY-14,643. The results of these studies using mP-PAR␣ E and mPPAR␣ D/E did not allow discrimination between an inability to bind ligand and a defect in receptor/p300 interactions.
We have demonstrated that 79 amino acids of p300-(39 -117) and 115 amino acids of CBP-(1-115) are sufficient for interaction with PPAR␣ in vitro. One copy of the recently identified LXXLL nuclear receptor interaction motif (48,75,76) is contained within these regions (amino acids 81-85 of p300 and 69 -73 of CBP; see Fig. 1A and data not shown), and our studies demonstrate that this single motif is clearly sufficient for interaction with mPPAR␣, but not RAR␥ or mRXR␣, in vitro. A second LXXLL motif is present outside this region within amino acids 344 -348 of p300 and 357-361 of CBP (75), and it is conceivable that both motifs are required for interaction with the retinoid receptors examined in this study. However, Kamei and co-workers (40) previously demonstrated that p300-(1-117), which contains the identical LXXLL motif present in p300-(39 -221), interacts with RAR␣ when examined in the yeast two-hybrid system. The basis for this discrepancy is presently unknown but may involve a lack of sufficient sensitivity in our assays which would have permitted detection of weak interactions. Nonetheless, our observations clearly demonstrate that p300 amino acids 39 -221 interact robustly with PPAR␣, but not detectably with RAR␥ or RXR␣, both in yeast and in vitro. We hypothesize that these differential interactions are a result of different affinities of each receptor with this particular coactivator/integrator protein and are not indicative of an exclusive PPAR␣ interaction. Indeed, other coactivators appear to interact preferentially, although not exclusively, and/or augment the transcriptional activation properties of particular nuclear receptors (47). These observations may be particularly relevant to the functioning of receptor-coactivator complexes in vivo where the concentrations of receptors, coactivators, or both may be limiting and in competition with each other. In this manner, an additional layer of regulation is afforded by coactivators in establishing a finely tuned signaling system.
The nuclear proteins p300 and CBP are increasingly identified as common components of a variety of signaling pathways (reviewed in Ref. 49, also see Refs. 50 -53, 85, and 86). The central role of CBP in vivo is underscored by the identification of three human diseases that are attributable to mutations in the CBP gene: 1) Rubinstein-Taybi syndrome (87), 2) myelodysplastic syndrome (88), and 3) acute myeloid leukemia (89). Although translocations involving chromosome regions that map within the same region as the p300 locus have been linked to a leukemia similar to that of acute myeloid leukemia (89), it remains to be firmly established if dysfunctional forms of p300 also give rise to human maladies. It will be of critical importance to determine the precise role of p300 and/or CBP in PPAR signaling and if other signaling pathways cross-talk at the level of PPAR␣ coactivators such as p300/CBP and SRC-1.
Our results demonstrate that p300 is a bona fide coactivator for the nuclear receptor PPAR␣. Recently, two independent groups reported that p300 is also capable of interacting directly with and augmenting the transcriptional activation properties FIG. 8. In vivo interaction between p300 and endogenous PPAR. GST/p300-(1-450) was bound to glutathione-Sepharose and used as an affinity matrix to examine interactions with endogenous PPAR(s) from mammalian cell lysates. Experiments were conducted essentially as described in Fig. 4A except nuclear extracts from P19 cells were incubated with GST or GST/p300-(1-450) in the absence and presence of 100 M WY-14,643. Bound proteins were then eluted and subjected to immunoblot analysis with antiserum that recognizes an epitope common to all known PPAR subtypes (see "Materials and Methods"). The arrows indicate the positions of two proteins that are recognized by this antibody. Each protein is in the range of 51-55 kDa which is consistent with the calculated mass of PPARs. Immunoblot analysis of material retained on the GST/p300-(1-450) bound matrix using preimmune (PI) serum did not reveal the immune complexes indicated above by the filled arrows (data not shown). of the tumor suppressor protein, p53 (52,53). When these results are considered together, it is tempting to speculate that peroxisome proliferator-induced hepatocarcinogenesis in rodents may be the result of PPAR␣-mediated interference with p53 signaling through p300. A mechanistically similar phenomenon has been reported for the mutual antagonism between nuclear receptors and AP-1 (40). In this manner, sequestration of the common coactivator, p300, by chronically activated PPAR␣ may titrate cellular levels of p300 thereby compromising the tumor suppressor activity of p53. Although this scenario may seem plausible, the exact molecular mechanisms that may occur and the determinants of specificity of such a scenario remain to be discovered.