Identification of Nuclear Receptor Corepressor as a Peroxisome Proliferator-activated Receptor α Interacting Protein*

Nuclear receptor corepressor (NCoR) was demonstrated to interact strongly with peroxisome proliferator-activated receptor α (PPARα), and PPARα ligands suppressed this interaction. In contrast to the interaction of PPARα with the coactivator protein, p300, association of the receptor with NCoR did not require any part of the PPARα ligand binding domain. NCoR was found to suppress PPARα-dependent transcriptional activation in the context of a PPARα·retinoid X receptor α (RXRα) heterodimeric complex bound to a peroxisome proliferator-responsive element in human embryonic kidney 293 cells. This repression was reversed agonists of either receptor demonstrating a functional interaction between NCoR and PPARα·RXRα heterodimeric complexes in mammalian cells. NCoR appears to influence PPARα signaling pathways and, therefore, may modulate tissue responsiveness to peroxisome proliferators.

Nuclear receptor corepressor (NCoR) was demonstrated to interact strongly with peroxisome proliferator-activated receptor ␣ (PPAR␣), and PPAR␣ ligands suppressed this interaction. In contrast to the interaction of PPAR␣ with the coactivator protein, p300, association of the receptor with NCoR did not require any part of the PPAR␣ ligand binding domain. NCoR was found to suppress PPAR␣-dependent transcriptional activation in the context of a PPAR␣⅐retinoid X receptor ␣ (RXR␣) heterodimeric complex bound to a peroxisome proliferator-responsive element in human embryonic kidney 293 cells. This repression was reversed agonists of either receptor demonstrating a functional interaction between NCoR and PPAR␣⅐RXR␣ heterodimeric complexes in mammalian cells. NCoR appears to influence PPAR␣ signaling pathways and, therefore, may modulate tissue responsiveness to peroxisome proliferators.
Members of the steroid/thyroid hormone receptor superfamily function by binding to specific DNA response elements within the regulatory regions of target genes and modulating expression of these genes at the transcriptional level (1)(2)(3). Regulation of target gene expression mediated by nuclear receptors may occur in response to activation of the receptors by ligands (4) or by phosphorylation (5) or by a combination of both events (5).
The mammalian PPAR 1 family is composed of at least three genetically and pharmacologically distinct subtypes, PPAR␣, -␥, and -␤/␦ (also referred to as NUCI; reviewed in Refs. 6 and 7). The primary physiological roles for the ␣ and ␥ subtypes of PPAR appear to be regulation of lipid metabolism and adipogenesis, respectively, and both subtypes have been implicated in modulating inflammatory responses (8 -12). A physiological role for PPAR␤/␦ has not been elucidated but this receptor subtype is expressed ubiquitously in the mouse, possibly suggesting a more general function (13).
PPAR-dependent transcriptional activation of many genes is well documented, and direct, ligand-enhanced interactions between PPARs and the coactivators, p300/CBP (39), SRC-1 (39 -41), PPAR-binding protein (PBP; Ref. 42), and PGC-1 (43) are thought to play a role in such activation. In contrast, PPARmediated transcriptional repression of target genes, as observed for RAR and TR (see above), is relatively unexplored or at best controversial. PPAR␥ has been shown to interact in solution with the corepressors, NCoR and SMRT, but weakly if at all when bound to DNA, possibly suggesting that neither of these corepressors mediate putative PPAR-dependent gene repression (44). However, Lavinsky and co-workers (45) demonstrated SMRT-dependent gene repression mediated by a phosphorylated form of PPAR␥. Such findings illustrate the need for a more complete mechanistic understanding of potential PPARdependent repression of gene expression. We report here the isolation of NCoR from a yeast two-hybrid screen using PPAR␣ as a bait. We describe results from studies in yeast, in mammalian cells, and in vitro that were conducted to characterize PPAR␣-NCoR interactions and to examine the influence of PPAR ligands upon such interactions.

MATERIALS AND METHODS
Plasmids and Receptor Constructs-Plasmids encoding the receptors described below were used either directly or as templates for polymerase chain reaction to assemble all constructs described herein using standard techniques. All plasmids were kind gifts from the following individuals: mouse PPAR␣ (21) from Drs. S. Green  , and identical fragments of all receptors and mutants thereof listed above were subcloned in pTL1 (46) and have been described previously (39,51). PPAR␣⌬288 (amino acids 91-287 of mPPAR␣), PPAR␣⌬227 (amino acids 91-226 of mPPAR␣), PPAR␣⌬202 (amino acids 91-201 of mPPAR␣), and PPAR␣⌬180 (amino acids 91-179 of mPPAR␣) were constructed by polymerase chain reaction amplification of the indicated receptor regions with primers containing appropriate restriction sites for insertion into pTL1. Receptor proteins and derivatives thereof were expressed by in vitro transcription/translation for use in GST pulldown experiments as described previously (39,51). Note that all PPAR␣ constructs encode receptors truncated in the amino-terminal (⌬AB).
A (PPRE) 3 -tk-CAT reporter construct was prepared by insertion of a multimerized acyl-CoA oxidase PPRE (8) into the XbaI site of pBL2CAT2 (52). Additional information concerning any of the above constructs described herein can be obtained upon request.
Yeast Two-hybrid Screening-The yeast two-hybrid screen was conducted as described (39) except that a mouse brain cDNA library inserted into the GAL4 activation domain-encoded yeast expression vector, pACT2 (CLONTECH), was used in place of the mouse embryo library and approximately 2 ϫ 10 6 yeast transformants were screened. Plasmid DNA from positive clones was then isolated, and the resulting cDNAs were re-tested for interaction with PPAR baits, bait DBDs alone, and non-PPAR-related baits. cDNA clones, which exhibited a specific interaction with the PPAR baits, were then sequenced using the standard dideoxynucleotide chain termination method.
␤-Galactosidase Assays and Data Analysis-␤-Galactosidase assays were conducted as described previously (39). All titration data were analyzed using an iterative, curve-fitting routine (GraphPad Prism) and the four-parameter logistic equation. Yeast ␤-galactosidase results were analyzed using a two-tailed Student's t test.
Protein-Protein Interaction Assay-GST pulldown experiments were conducted as described previously (39), and ligand-dependent assays were carried out by inclusion of 1 M 9-cis-RA, 1 mM clofibrate, 100 M troglitazone, 100 M LY-171883, 100 M ETYA, 100 M WY-14,643 or vehicle in binding buffers except where otherwise noted. GST and GST fusion proteins were produced as described previously (39).
Mammalian Cell Transfection Experiments-Human embryonic kidney 293 (HEK293) cells were maintained and transiently transfected as described previously (53). Dose-response curves were fit using Graph-Pad Prism software as described above.
Chemicals and Reagents-WY-14,643 and clofibrate were purchased from Chemsyn Science Laboratories (Lenexa, KS) and Sigma, respectively. LY-171883 and ETYA were obtained from Biomol (Plymouth Meeting, PA). Troglitazone was kindly supplied by Dr. S. Kliewer (Glaxo Wellcome). All radioisotopes were purchased from NEN Life Science Products.

RESULTS
A previously described yeast two-hybrid system (49) was used to isolate a carboxyl-terminal NCoR (29) fragment as a PPAR␣-interacting protein from an oligo(dT)-primed, mouse brain cDNA library (NCoR amino acids 2110 -2453, Fig. 1). Previous studies have identified two domains, ID I and ID II, that mediate interactions between NCoR and nuclear receptors  (55). For clarity we have referred to our clone throughout the text as NCoR amino acids 2110 -2453. (29,54). The NCoR fragment isolated in this screen encompasses the last 33 amino acids of ID II and the entirety of ID I (Fig. 1).
PPAR␣ and interaction domains of the coactivators, p300 and SRC-1, exhibit strong ligand-independent association when examined in a yeast two-hybrid system (39). When examined in vitro, however, these interactions are strictly liganddependent, suggesting that yeast may contain endogenous PPAR␣ activating ligands (39). The isolation of NCoR in a yeast two-hybrid screen was therefore unexpected as we (39) had hypothesized that PPAR␣ existed in a liganded state in yeast that may not facilitate receptor interactions with potential corepressor proteins as has been observed for other liganded nuclear receptors (28,29). Toward the goal of understanding the influence of NCoR on PPAR␣ signaling mechanisms, we chose to conduct a more thorough analysis of the interaction between these two proteins.
To confirm the observed interactions in yeast, in vitro protein-protein interaction assays were conducted. A GST/NCoR fusion protein containing amino acids 2110 -2453 of NCoR was examined for the ability to interact with radioactively labeled PPAR␣, RAR␥, and RXR␣ prepared by in vitro translation. PPAR␣ interacted strongly with GST/NCoR in vitro, but not GST alone, in the absence of ligand (Fig. 2B, lanes 2 and 10), and the presence of WY-14,643 did not significantly affect this interaction (Fig. 2B, lane 3). Similarly, RAR␥ interacted with GST/NCoR in vitro, and this interaction was not significantly affected by the presence of 9-cis-RA (Fig. 2B, lanes 5 and 6). As observed in yeast, RXR␣ interacted weakly with GST/NCoR and 9-cis-RA appeared to stimulate this interaction modestly (3.4-fold; Fig. 2B, lanes 8 and 9). A weak, ligand-enhanced interaction between RXR and the related corepressor, SMRT, has also been observed by Chen and co-workers (28). In addition, Seol and collaborators (55) observed an approximately 3-fold increased interaction between NCoR/RIP13 (see Fig. 1 legend) and RXR in the presence of 9-cis-RA.
To probe the ligand dependence of PPAR␣-NCoR interactions further, several additional PPAR ligands were examined using the yeast two-hybrid system as described above. Clofibrate, LY-171883, and ETYA, as well as the PPAR␥-specific ligand, troglitazone, did not significantly affect PPAR␣-NCoR interactions in yeast (Fig. 3A). The only ligand examined in yeast that efficiently promoted dissociation of PPAR␣ and NCoR was WY-14,643, and this effect was dose-dependent with an apparent IC 50 of 134 nM (Fig. 3B).
NCoR and p300 Require Distinct PPAR␣ Regions for Interaction-To determine which regions of PPAR␣ are required for association with NCoR, we examined GST/NCoR interactions with several carboxyl-and amino-terminal PPAR␣ truncation mutants in vitro using standard GST pulldown methodology (Fig. 4A). The results of these in vitro studies are depicted schematically in Fig. 4A. As shown above, PPAR␣ interacted strongly with both GST/NCoR and GST/p300 in vitro (Fig. 4B,  lanes 4 -7). The former interaction was only modestly inhibited by ligand, while the latter interaction was strictly ligand-dependent. In contrast, PPAR␣⌬448, which lacks 21 carboxylterminal amino acids, efficiently interacted with GST/NCoR but ligand-dependent interaction between this truncation mutant and GST/p300 was abolished (Fig. 4B, lanes 11-14). The hinge/LBD region of PPAR␣, PPAR␣ D/E, interacted in vitro with GST/NCoR and GST/p300 in a manner similar to that of PPAR␣ (Fig. 4B, lanes 18 -21), although the WY-14,643-induced dissociation of receptor⅐GST/NCoR complexes was more apparent (lanes 18 -19). In contrast, PPAR␣ E, which lacks residues contained within the hinge region of PPAR␣ (amino acids 166 -281), did not interact with either GST/NCoR or GST/p300 (Fig. 4B, lanes 25-28). These results demonstrate that both NCoR and p300 require the hinge region of the receptor for efficient interaction, while only the latter requires an intact PPAR␣ LBD. Therefore, the regions of PPAR␣ required for interaction with NCoR and p300 are partially overlapping but largely distinct.
The carboxyl-terminal PPAR␣ truncation mutants, PPAR␣⌬425, PPAR␣⌬288, PPAR␣⌬227, and PPAR␣⌬202, all interacted efficiently with GST/NCoR (Fig. 4C, lanes 3, 6, 9, and 12). PPAR␣⌬180, which lacks the entire LBD, also interacted with GST/NCoR (Fig. 4C, lane 15). PPAR␣ amino acids 166 -179 within the hinge region are common to all receptor proteins observed to interact with GST/NCoR, suggesting that these residues may play a role in PPAR␣ interactions with NCoR. However, this isolated region of the receptor did not interact with NCoR either in yeast or in vitro under a variety of conditions (data not shown), suggesting that this region of PPAR␣ is necessary but not sufficient to mediate interaction with NCoR.
NCoR ID II Is Not Necessary for PPAR␣-NCoR Interactions-Protein-protein interaction studies carried out with a GST/NCoR fusion protein lacking the entire ID II, GST/NCoR (⌬ID II), demonstrated that the ID II region of NCoR was not required for interaction with PPAR␣ (Fig. 5). As observed for interactions between PPAR␣ and GST/NCoR (amino acids 2110 -2453, see Fig. 2B, lanes 2 and 3), WY-14,643 did not significantly affect receptor-NCoR interactions in vitro (Fig. 5,  lanes 4 and 5).
PPAR␣ Ligands Promote Both PPAR␣-NCoR Dissociation and PPAR␣-p300 Association-The yeast two-hybrid system was used to compare the influence of several PPAR ligands on receptor interactions with the corepressor, NCoR (amino acids 2110 -2453), and the coactivator, p300 (amino acids 39 -221). PPAR␣ D/E was used as the receptor component in this series of experiments because we previously observed a readily detectable WY-14,643-enhanced interaction in yeast between the D/E region of PPAR␣ and p300 amino acids 39 -221 (39). As observed for yeast two-hybrid analyses using PPAR␣ (see Fig.  3), WY-14,643, but not troglitazone, clofibrate, LY-171883, or ETYA, significantly repressed the interaction between PPAR␣ D/E and NCoR (Fig. 6A). Similarly, troglitazone, clofibrate and LY-171883 had no significant influence on the strong ligandindependent PPAR␣ D/E interaction with p300, while both ETYA and WY-14,643 modestly, but significantly, enhanced this interaction (Fig. 6A). Thus, while both ETYA and WY-14,643 promoted p300-PPAR␣ interactions in yeast and in vitro, only WY-14,643 was observed to induce dissociation of NCoR-PPAR␣ complexes, and this was only significantly ap-  (29). Note that PPAR␣ refers to an amino-terminally truncated receptor lacking the A/B domain (see "Materials and Methods"). ND, not determined. B, in vitro protein-protein interactions between PPAR␣ truncation mutants and both GST/NCoR (2110 -2453) and GST/p300 (39 -221). These GST pulldown experiments were carried out as described in Fig. 2B and ligand concentrations were the same as those indicated in Fig. 3A. C, in vitro protein-protein interactions between PPAR␣ carboxyl-terminal truncation mutants and GST/NCoR (2110 -2453). GST pulldown assays were conducted as described in B. Representative experiments that were replicated three to five times are shown in B and C.

FIG. 5. ID II of NCoR is not required for interaction with PPAR␣.
In vitro protein-protein interactions between PPAR␣ and amino acids encompassing ID I, but not ID II, of NCoR (amino acids 2218 -2381, see Fig. 1). Assays were carried out as described in the legend to Fig. 2B with GST or GST/NCoR (⌬ID II). Shown is a representative experiment that was replicated three times. parent in yeast.
To confirm the observed interactions in yeast, in vitro protein-protein interaction assays were conducted as described above. GST/NCoR (amino acids 2110 -2453) and GST/p300 (amino acids 39 -221) were examined for interaction with both PPAR␣ D/E and PPAR␣ in the absence and presence of several, above-mentioned PPAR ligands. None of the ligands tested significantly affected the in vitro interaction of NCoR with either PPAR␣ D/E (Fig. 6B) or PPAR␣ (Fig. 6C). In contrast, in vitro association between GST/p300 and both PPAR␣ D/E (Fig.  6B) and PPAR␣ (Fig. 6C) was observed only in the presence of ETYA or WY-14,643 (lanes 18 and 19, respectively).
NCoR Represses PPAR␣/RXR␣-mediated Transcriptional Activation from a PPRE in HEK293 Cells-Cotransfection experiments in HEK293 cells were conducted to determine the physiological significance of the strong interaction between PPAR␣ and NCoR observed in yeast and in vitro. The reporter construct used for these studies, (PPRE) 3 -tk-CAT, exhibited low basal or ligand-stimulated activity in the absence of cotransfected receptor (Fig. 7, lanes 1-6). However, cotransfection of PPAR␣ and RXR␣ resulted in a strong, constitutive activity of this reporter that was only minimally stimulated by PPAR␣ (WY-14,643 and ETYA) or RXR␣ (9-cRA) agonists or both types of ligands together (Fig. 7, lanes 7-12). Cotransfection of full-length NCoR (2 g) dramatically reduced constitutive, PPAR␣⅐RXR␣-dependent transcriptional activation (compare lanes 7 and 13 of Fig. 7), and this repression was reversed by treating the cells with either PPAR␣ agonists alone or in combination with 9-cis-RA (lanes 13-18). Cotransfection of a SMRT expression vector (28) similarly repressed PPAR␣⅐RXR␣dependent transcriptional activation (data not shown). NCoRdependent repression of transcriptional activation mediated by PPAR␣⅐RXR␣ was also reversed by 9-cis-RA alone (Fig. 7, lane  15), suggesting that RXR␣ may be competent to bind ligand and/or activate transcription in the context of a PPAR␣⅐RXR␣ heterodimeric complex bound to the acyl-CoA oxidase PPRE (8). The potencies with which WY-14,643, 9-cis-RA, and ETYA activated PPAR␣⅐RXR␣ in the presence of cotransfected NCoR were determined in a series of concentration-response experiments using HEK293 cells. WY-14,643 (EC 50 ϭ 288 Ϯ 81 nM; n ϭ 5, Fig. 8A) and ETYA (EC 50 ϭ 189 Ϯ 108 nM, n ϭ 5; Fig. 8C) were roughly equipotent, while 9-cis-RA activated this reporter construct with an EC 50 ϭ 40 Ϯ 8 nM (n ϭ 3, Fig. 8B). These EC 50 values are in general agreement with previous estimates of the affinity of PPAR␣ for WY-14,643 (21,51) as well as that of RXR␣ for 9-cis-RA (56). These findings demonstrate that NCoR interacts with and represses transcriptional activation mediated by PPAR␣⅐RXR␣ heterodimeric complexes bound to a PPRE in mammalian cells and this repression is reversed by agonists of either receptor. From these results, we conclude that the cellular interaction between NCoR and PPAR␣ and/or RXR␣ is likely to be of physiological relevance and may influence tissue responsiveness to both peroxisome proliferators and retinoic acids. DISCUSSION We have demonstrated that PPAR␣ interacts strongly with NCoR, and the PPAR␣ ligand, WY-14,643, inhibits this interaction in yeast and in mammalian cells. The PPAR␣ ligand, ETYA, also inhibits PPAR␣-NCoR interaction in mammalian FIG. 6. Influence of ligands on NCoR and p300 interactions with PPAR␣. A, interactions between PPAR␣ D/E and both NCoR (amino acids 2110 -2453) and p300 (amino acids 39 -21) in yeast. Assays were conducted as described in Fig. 3A. Each determination represents 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, for ligand versus vehicle treatment. B, in vitro protein-protein interactions between PPAR␣ D/E and both GST/NCoR (2110 -2453) and GST/p300 (39 -221). Assays were carried out as described in the legend to Fig. 2B, and ligand concentrations were the same as those indicated in Fig. 3A. C, in vitro proteinprotein interactions between PPAR␣ and both GST/NCoR and GST/ p300. Assays were carried out as described above (Fig. 6B). Shown in B and C are representative GST pulldown experiments that were replicated four to six times. cells but fails to influence this interaction in yeast. Neither ligand significantly affected NCoR-PPAR␣ interactions in vitro under the conditions examined in these studies. ETYA does, however, promote interaction of the receptor with the coactivator, p300, in yeast, suggesting that the strain of yeast used for these studies is permeable to ETYA. Thus, the simplest explanation for these results is that ETYA may be metabolized in yeast to compounds which, while capable of promoting p300⅐PPAR␣ complex formation, are incapable of inducing dissociation of NCoR⅐PPAR␣ complexes. It is also possible that ETYA may elicit production of an endogenous yeast compound that promotes PPAR␣-p300 interaction but fails to influence PPAR␣-NCoR complexes. Furthermore, we cannot rule out the possibility that a combination of these two possibilities may be responsible for our observations in yeast using ETYA. Given that treatment of transfected HEK293 cells with ETYA activates a PPRE-containing reporter construct, the anomalous results that we obtained in yeast with this compound do not appear to have direct applicability to function of PPAR␣ in mammalian cells.
We previously hypothesized the existence of putative, endogenous agonists in yeast that may disfavor PPAR␣-corepressor interactions (39). However, if putative, endogenous ligands contribute to the constitutive PPAR␣-coactivator interactions observed in yeast (39), the presence of these agonists clearly does not preclude PPAR␣ interaction with NCoR. In this case, endogenous yeast agonists may function in a manner similar to that of ETYA or metabolic derivatives thereof which promote PPAR␣-p300 association but do not inhibit PPAR␣-NCoR in-teraction. These hypotheses may provide an explanation for how NCoR was unexpectedely isolated as a PPAR␣-interacting protein in our yeast two-hybrid screen.
Previously we demonstrated that the coactivators, p300 and SRC-1, require 21 carboxyl-terminal residues of the PPAR␣ LBD for interaction with PPAR␣ (39). By analogy with other nuclear receptors, this region of PPAR␣ is predicted to contain the putative core of the ligand dependent-transcriptional activation function (57). We demonstrate herein that the PPAR␣ truncation mutant, PPAR␣⌬448, which lacks the 21 carboxylterminal amino acids encompassing the putative AF-2 core, interacts with NCoR but not p300. Therefore, the corepressor, NCoR, and the coactivators, p300 and SRC-1, appear to interact with the receptor in mechanistically distinct manners that utilize different regions of PPAR␣ as protein-protein interaction surfaces. However, simultaneous interaction between the receptor and both NCoR and p300/SRC-1 are unlikely, because both types of interaction require common amino acid residues within the hinge region of the receptor.
Deletion of the hinge region of PPAR␣ (amino acids 166 -281) abolished NCoR-PPAR␣ interaction, and amino acids 166 -179 within the amino-terminal portion of the PPAR␣ hinge region were common to all receptor fragments that exhibited interaction with NCoR. The PPAR␣ mutant, PPAR␣⌬180 (amino acids 91-179), which interacted efficiently with NCoR, lacks the entirety of both the putative CoR box (29) and the ligand binding domain. These results suggest that PPAR␣ amino acids 166 -179 may mediate interactions with NCoR. However, extensive analyses in yeast and in vitro have failed to demonstrate that PPAR␣ 166 -179 are sufficient to mediate interaction with NCoR (data not shown), possibly indicating that additional contacts are required for efficient interaction. Nonetheless, our results suggest that PPAR␣ likely contains a NCoR interaction surface that is clearly not contained within the LBD of the receptor and, thus, may be distinct from that present in either RAR or TR (28,29).
Zamir and collaborators (44) have shown that PPAR␥ can interact with the corepressors, SMRT and NCoR, in solution, but weakly if at all when bound to DNA. Similarly, we were unable to demonstrate the formation of a DNA bound PPAR␣⅐RXR␣⅐NCoR complex in vitro (data not shown). However, in contrast to the findings of Zamir and colleagues (44) who observed no corepressor-dependent PPAR␥-mediated repression, our transient transfection studies clearly demonstrate a functional interaction between a PPRE-bound, PPAR␣⅐RXR␣ complex and NCoR. Such discrepant results could simply be a result of either differing PPAR subtypes (␣ versus ␥) or cell lines (HEK 293 versus 293T) or a combination of these two possibilities. The inability to observe a DNA-bound PPAR␣⅐RXR␣⅐NCoR complex in vitro may be due to an inherent instability of such complexes, and indeed, other groups have reported that cross-linking reagents are required to stabilize similar complexes in vitro (58). NCoR clearly associates with and represses the transcriptional activity of PPRE-bound, PPAR␣⅐RXR␣ heterodimeric complexes in HEK293 (Fig. 7). However, we cannot exclude the possibility that additional cellular factor(s) present in HEK293 cells, but lacking in vitro, are required for the assembly of a DNA-bound, PPAR␣⅐RXR␣⅐NCoR complex.
Finally, it is conceivable that interaction of NCoR or SMRT with either PPAR␣ or PPAR␣⅐RXR␣ complexes may influence other signaling pathways by titration of limiting amounts of these corepressors. This form of receptor cross-talk may serve to relieve transcriptional repression mediated by other nuclear receptors, such as RAR, TR, or RevErb, that utilize common corepressors. Results presented herein raise the possibility FIG. 8. Concentration-response curves for activation of PPAR␣⅐RXR␣ heterodimeric complexes by WY-14,643, 9-cis-RA, and ETYA in HEK293 cells. HEK293 cells were cotransfected with 2 g of the (PPRE) 3 -tk-CAT reporter plasmid and expression vectors for NCoR (2 g) and PPAR␣/RXR␣ (0.5 g of each). Cells were treated with increasing concentrations of WY-14,643 (A), 9-cis-RA (B), or ETYA (C) for 24 h, and CAT activity was determined as described in the legend of Fig. 7. The curves shown for each agonist were obtained using an iterative, curve-fitting routine (GraphPad Prism). Shown is a representative CAT experiment that was replicated three (9-cis-RA (9cRA)) or five (WY-14,643 and ETYA) times. The following EC 50 values represent the mean Ϯ S.E. of these multiple determinations: WY-14,643, 288 Ϯ 81 nM; 9-cis-RA, 40 Ϯ 8 nM; ETYA, 189 Ϯ 108 nM. that PPAR interactions with corepressors in solution or on DNA may play a prominent role in regulating PPAR-dependent transcriptional regulation of target genes.