Ligand-induced Peroxisome Proliferator-activated Receptor α Conformational Change

Structurally diverse peroxisome proliferators and related compounds that have been demonstrated to induce the ligand-dependent transcriptional activation function of mouse peroxisome proliferator-activated receptor α (mPPARα) in transfection experiments were tested for the ability to induce conformational changes within mPPARα in vitro. WY-14,643, 5,8,11,14-eicosatetraynoic acid, LY-171883, and clofibric acid all directly induced mPPARα conformational changes as evidenced by a differential protease sensitivity assay. Carboxyl-terminal truncation mutagenesis of mPPARα differentially affected the ability of these ligands to induce conformational changes suggesting that PPAR ligands may make distinct contacts with the receptor. Direct interaction of peroxisome proliferators and related compounds with, and the resulting conformational alteration(s) in, mPPARα may facilitate interaction of the receptor with transcriptional intermediary factors and/or the general transcription machinery and, thus, may underlie the molecular basis of ligand-dependent transcriptional activation mediated by mPPARα.

Structurally diverse peroxisome proliferators and related compounds that have been demonstrated to induce the ligand-dependent transcriptional activation function of mouse peroxisome proliferator-activated receptor ␣ (mPPAR␣) in transfection experiments were tested for the ability to induce conformational changes within mPPAR␣ in vitro. WY-14,643, 5,8,11,14-eicosatetraynoic acid, LY-171883, and clofibric acid all directly induced mPPAR␣ conformational changes as evidenced by a differential protease sensitivity assay. Carboxylterminal truncation mutagenesis of mPPAR␣ differentially affected the ability of these ligands to induce conformational changes suggesting that PPAR ligands may make distinct contacts with the receptor. Direct interaction of peroxisome proliferators and related compounds with, and the resulting conformational alteration(s) in, mPPAR␣ may facilitate interaction of the receptor with transcriptional intermediary factors and/or the general transcription machinery and, thus, may underlie the molecular basis of ligand-dependent transcriptional activation mediated by mPPAR␣.
Peroxisome proliferator-activated receptors (PPARs) 1 are members of a large family of ligand-inducible transcription factors that includes receptors for retinoids, vitamin D, and thyroid and steroid hormones (1)(2)(3)(4)(5). The mammalian PPAR family is composed of at least three genetically and pharmacologically distinct subtypes, PPAR␣, -␥, and -␦ (reviewed in Ref. 6). Murine PPAR ␣ (mPPAR␣) was originally isolated from a mouse liver cDNA library by Issemann and Green (7) who demonstrated that the receptor was activated in transfection experiments by a group of compounds known to induce peroxisome proliferation in rodents. A number of structurally diverse compounds have subsequently been demonstrated to activate PPAR␣ in transient transfection experiments. Particularly noteworthy among these compounds are: 1) lipids such as arachidonic acid (8 -11) and its synthetic analog 5,8,11,14eicosatetraynoic acid (ETYA, Refs. 9, 11-13), 8-[S]-hydroxyeicosatetraenoic acid (14), a lipoxygenase metabolite of arachidonic acid, and linoleic acid (8 -11, 14, 15); 2) fibric acid antihyperlipidemic drugs (WY-14,643, clofibric acid, gemfibrozil, ciprofibric acid; Refs. 10,16,17) that represent a class of therapeutic agents useful in the treatment of hypertriglyceridemia (18); and 3) a leukotriene D4 antagonist, LY-171883 (19). Many of these compounds, together with phthalate ester plasticizers (di(-2-ethylhexyl)-phthalate) and herbicides (2,4,5trichlorophenoxyacetic acid), are known collectively as peroxisome proliferators (reviewed in Ref. 20). While chemically distinct, most of these compounds have been demonstrated to induce proliferation of peroxisomes leading to hepatic hyperplasia and hepatocarcinogenesis in many species (20). Peroxisome proliferator-induced alteration of hepatocyte phenotype is believed to result from activation of PPAR␣ and subsequent modulation of gene expression downstream of this nuclear receptor (reviewed in Refs. 6, 20; see below). The central role of PPAR␣ in xenobiotic-induced peroxisomal proliferation was recently demonstrated by the absence of hepatomegaly and peroxisome proliferation in mice null for expression of this gene (21).
PPARs modulate expression of target genes by binding to response elements comprised of a degenerate direct repeat of the hexameric nucleotide sequence, TGACCT, separated by one base pair (DR1). PPAR has been shown to bind cognate response elements with high affinity only in the context of a heterodimeric complex with the retinoid X receptor (RXR, Refs. 11,17,[22][23][24]. PPAR⅐RXR heterodimeric complexes appear to be responsive to both PPAR activators and 9-cis-retinoic acid, the endogenous ligand for RXR (11,17,(22)(23)(24).
PPAR response elements (PPREs) have been identified in the 5Ј regions of several mammalian genes coding for proteins involved in lipid metabolism such as acyl-CoA oxidase (17,25), bifunctional enzyme (26,27), malic enzyme (16), liver fatty acid binding protein (28), 3-hydroxy-3-methylglutaryl-CoA synthase (15), and cytochrome P450 fatty acid -hydroxylase (29). Such findings indicate a prominent regulatory role for the PPAR receptor family in lipid metabolism and homeostasis. In addition, overexpression of PPAR␣ and -␥ in cultured fibroblasts and subsequent exposure to PPAR ligands has been shown to confer adipogenicity (30,31), further illustrating the central regulatory role of PPAR family members in lipid homeostasis.
In contrast to many other receptors in the retinoid/thyroid hormone receptor superfamily, functional domains of PPARs and critical amino acid residues within such putative domains have not been extensively characterized. Two previous studies with PPAR␣ have identified: 1) a Glu 282 3 Gly point mutation in mPPAR␣ that ameliorates transcriptional responses to WY-14,643 and ETYA (13), and 2) a Leu 433 3 Arg point mutation in human PPAR␣ (hPPAR␣) that abolishes heterodimerization with RXR (32). The present studies were undertaken to identify mPPAR␣ carboxyl-terminal receptor regions that are important for both ligand responsiveness and heterodimerization with mRXR␣ and to determine if structurally diverse PPAR ligands induce similar conformational changes within mPPAR␣. To our knowledge, these studies provide the first direct biochemical evidence demonstrating that peroxisome proliferators induce conformational changes within mPPAR␣. Ligand-induced stabilization of particular mPPAR␣ conformational states likely underlies the molecular basis for the ability of these compounds to activate the receptor and to modulate expression of mPPAR␣ target genes including those implicated in peroxisome proliferation. (7)  A mPPAR␣ amino-terminal truncation mutant was constructed by polymerase chain reaction using a 5Ј primer (ML023) that introduced an EcoRI site, favorable Kozak sequence, and an initiator methionine fused to Asp 91 of mPPAR␣ and a 3Ј primer that introduced a BamHI site 3Ј of the mPPAR␣ natural stop codon. The resulting fragment was appropriately digested and subcloned into the eukaryotic expression vector, pTL1 (33), yielding PPAR⌬AB. PPAR⌬AB is transcribed/translated in vitro at least 10-fold more efficiently than full-length receptor and exhibits DNA binding and heterodimerization activities that are indistinguishable from full-length receptor (data not shown). The carboxyl-terminal truncation mutants, PPAR⌬AB/⌬448 and PPAR⌬AB/ ⌬425 (Fig. 1A), were prepared by polymerase chain reaction using ML023 as the 5Ј primer and a 3Ј primer that introduced stop codons at positions 448 and 425, respectively, preceding a BamHI site. Both of the resulting fragments were appropriately digested and subcloned into pTL1 as described above. PPAR⌬AB, PPAR⌬AB/⌬448, and PPAR⌬AB/ ⌬425 were transcribed/translated in vitro with equal efficiencies (data not shown).

Plasmids and Receptor Constructs-Full-length mPPAR␣
GST-mRXR␣ was prepared by polymerase chain reaction amplification of full-length mRXR␣ using a 5Ј primer that introduced a HincII site immediately upstream of the natural initiator methionine and a 3Ј primer that introduced an EcoRI site 3Ј of the natural stop codon of mRXR␣. The resulting fragment was appropriately digested and subcloned into a EheI/EcoRI-digested GST fusion vector (pGEX-cs).
In Vitro Transcription/Translation-Proteins were prepared by in vitro transcription/translation using rabbit reticulocyte lysate as described previously (3,33). Translation reactions were carried out in the presence of [ 35 S]methionine for production of radioactively labeled proteins used in DPSAs and GST-pull down experiments, whereas receptor proteins used in electrophoretic mobility shift assays were translated in the presence of unlabeled methionine. Unprogrammed lysates were generated identically using equal amounts of linearized pTL1 in place of receptor-coding templates.
Receptor proteins (10 and 20 fmol of PPAR⌬AB and mRXR␣, respectively) were preincubated on ice for 15 min prior to addition of a mix containing ϳ50,000 cpm of Klenow end-filled DR1 or ACO-PPRE probes. Components of the probe mix were (in mM) HEPES-NaOH, pH 7.5, 10; EDTA, 1; dithiothreitol, 1; and NaCl, 150. The mix was supplemented with 10% glycerol, 1 g/l bovine serum albumin, and poly[d(I⅐C)] (2 g/tube). The amount of lysate in each binding reaction was held constant by addition of unprogrammed reticulocyte lysate. Samples were loaded on a 5% polyacrylamide gel, electrophoresed, and gels were dried and subjected to autoradiography as described previously (33,36).
Bacterial Expression and Purification of GST and GST-mRXR␣ Fusion Proteins-GST-mRXR␣ expression in the DH5␣FЈ strain of Escherichia coli was induced by addition of isopropyl ␤-D-thiogalactopyranoside (1 mM final concentration) to the growth media and cultured for an additional 2 h. Bacterial extracts were prepared using standard methods. The fusion protein was purified on a glutathione-Sepharose 4B column as per the manufacturer's (Pharmacia Biotech Inc.) recommendations.
GST Pull-down Experiments-Glutathione-Sepharose 4B (Pharmacia) was washed extensively in phosphate-buffered saline (PBS) and resuspended in a volume of PBS sufficient to generate a 50% slurry. This slurry (1 ml) was mixed with 2 volumes of PBS (2 ml) containing either no protein, purified GST, or purified GST-mRXR␣ (proteins were at a concentration of ϳ1 mg/ml) and incubated with rotation at 4°C overnight. The resin slurry was gently centrifuged, washed 5 times in 2 volumes of PBS (2 ml) to remove all unbound protein, and finally resuspended in 1 volume of binding buffer (same as EMSA buffer but without poly[d(I⅐C)]). GST pull-down experiments were conducted using 20 l (ϳ100 fmol) of in vitro translated 35 S-PPAR⌬AB, 35 S-PPAR⌬AB/ ⌬448, or 35 S-PPAR⌬AB/⌬425, and 40 l of GST bound, GST-mRXR␣ bound, or unbound glutathione-Sepharose slurries. After an overnight incubation at 4°C with continuous rotation, samples were gently centrifuged and washed 10 times using 250 l of binding buffer. After the final wash the resin was resuspended in 30 l of 2 ϫ loading buffer (125 mM Tris-HCl, pH 6.8; 4% (w/v) SDS; 1.4 M ␤-mercaptoethanol; 25% (v/v) glycerol; 0.1% (w/v) bromphenol blue) of which one-half was electrophoresed on 12.5% denaturing gels and processed as described previously (33,36). Some GST pull-down experiments were conducted as described above but with the addition of unlabeled, annealed oligonucleotides corresponding to ACO and DR1 PPREs to all incubations and wash buffers at a final concentration of 5 fmol/l. Differential Protease Sensitivity Assays-Two l (ϳ10 fmol) of in vitro translated 35 S-PPAR⌬AB, 35 S-PPAR⌬AB/⌬448, or 35 S-PPAR⌬AB/ ⌬425 were preincubated in 7 l of binding buffer (as described under "GST Pull-down Experiments") containing either WY-14,643 (pirinixic acid), ETYA, LY-171883 (5-[4Ј-(4Љ-acetyl-3Љ-hydroxy-2Љ-propylphenoxy) butyl]tetrazole), clofibrate (2-(4-chlorophenoxy)-2-methylpropanoic acid ethyl ester), clofibric acid (2-(4-chlorophenoxy)-2-methylpropanoic acid), or an equal volume of vehicle for 30 min at 22°C. The final concentration of vehicle did not exceed 0.15% (v/v) in any experiment conducted. Stock solutions of all ligands were prepared on the day of the experiment in dimethyl sulfoxide (WY-14,643, and LY-171883) or ethanol (ETYA, clofibric acid, and clofibrate). DPSAs were initiated by addition of 1 l of 10 ϫ stock solution of chymotrypsin in water and were allowed to proceed for 20 min at 22°C. Reactions were terminated by addition of 1 volume of 2 ϫ loading buffer (as described under "GST Pull-Down Experiments"). Electrophoresis, autoradiography, and densitometric quantification were carried out as described previously (33,36). DPSAs conducted to determine the effect of heterodimerization with mRXR␣ were carried out essentially as described above except that [ 35 S]methionine-labeled PPAR preparations were incubated with a 2-fold molar excess of in vitro translated mRXR␣ or unprogrammed lysate for 10 min at 22°C prior to addition of ligand.

RESULTS
Carboxyl-terminal truncation mutants of PPAR⌬AB were constructed to define regions of the receptor required for interaction with RXR and to determine if diverse ligands require distinct mPPAR␣ structural features. Based on the crystal structures of RXR␣ (37,38) and retinoic acid receptor ␥ (RAR␥, Ref. 38) LBDs and the predicted structural similarity of these receptors to mPPAR␣ (39 and data not shown) two PPAR⌬AB carboxyl-terminal truncation mutants were prepared as follows: 1) PPAR⌬AB/⌬448 that lacks a portion of putative helix H11 and all of helix H12, and 2) PPAR⌬AB/⌬425 that lacks putative helices H10-H12 (see Fig. 1A). Because both carboxylterminal truncation mutants lack the core of the putative liganddependent transcriptional activation function (AF-2, Ref. 39), neither would be expected to activate transcription in a liganddependent manner.
Protein-protein interaction experiments were carried out to investigate the ability of PPAR␣ carboxyl-terminal truncation mutants to interact with RXR independently of DNA binding. GST-mRXR␣ fusion protein, immobilized on glutathione-Sepharose, was used in standard GST pull-down experiments for this purpose. In vitro translated 35 S-PPAR⌬AB and 35 S-PPAR⌬AB/⌬448 both interacted with GST-mRXR␣ (Fig. 3,  lanes 4 and 5) while an interaction between 35 S-PPAR⌬AB/ ⌬425 and GST-mRXR␣ was not detected (Fig. 3, lane 6). The efficiency of 35 S-PPAR⌬AB/⌬448 interaction with GST-mRXR␣ was reduced approximately 2-fold relative to that of 35 S-PPAR⌬AB with GST-mRXR␣ in agreement with DNA binding experiments described above. No interactions between any of the PPAR receptor proteins and an immobilized GST protein (Fig. 3, lanes 7-9) or glutathione-Sepharose alone were ob-served (data not shown). Additionally, results from experiments conducted in the presence of unlabeled response elements, identical to those used in DNA binding assays (see above), were indistinguishable from those described above (data not shown).
The Sensitivity of mPPAR␣ to Chymotryptic Digestion Is Altered by Interaction with Ligands That Activate the Receptor-We have adapted a differential protease sensitivity assay (DPSA, Ref. 36) for use with 35 S-PPAR⌬AB to address the possibility that peroxisome proliferators and related compounds (see Fig. 1B) interact directly with and alter the protease sensitivity of the receptor. Digestion of 35 S-PPAR⌬AB with increasing concentrations of chymotrypsin in the presence of 100 M LY-171883, ETYA, or WY-14,643 (Fig. 4A, lanes 11-13,  14 -16, and 17-19, respectively) resulted in the appearance of protease-resistant fragments of approximately 33, 31, and 27 kDa, referred to hereafter as PF33, PF31, and PF27, respectively. Clofibric acid and clofibrate, when examined at concentrations of 100 M, resulted in very weak signals (data not shown); therefore, these PPAR ligands were examined at concentrations of 1 mM. While clofibric acid clearly induced forma-  (38) and hRXR␣ (37). The carboxyl-terminal mPPAR␣ residues of PPAR⌬AB/⌬448 and PPAR⌬AB/⌬425 (Ile 447 and Pro 424 , respectively) are indicated by arrows. The linear alignment of mPPAR␣ LBD with that of hRAR␥ and hRXR␣ was adapted from Wurtz et al. (39). B, structures of the PPAR activators used in this study. tion of PF33, PF31, and PF27 (Fig. 4A, lanes 5-7), clofibrate only weakly induced formation of these proteolytic fragments (Fig. 4A, lanes 8 -10). The glucocorticoid receptor ligand, dexamethasone, had no effect on the proteolytic sensitivity of 35 S-PPAR⌬AB at concentrations up to 1 mM (data not shown). Moreover, none of the mPPAR␣ activators examined affected the protease sensitivity of other nuclear receptors such as mRXR␣ (data not shown), indicative of the specificity of these observations. These results suggest that mPPAR␣ undergoes a ligand-induced conformational change upon interaction with compounds previously demonstrated to activate the receptor in transient transfection experiments (7,13,19,44). Chymotrypsin-resistant fragments induced by clofibric acid, clofibrate, LY-171883, ETYA, and WY-14,643 appear to be indistinguishable suggesting that a similar change within mPPAR␣ may be induced by all five PPAR activators. The following rank order of efficacy of the five PPAR activators for induction of PFs within PPAR⌬AB, at concentrations of 100 M, was determined using quantitative densitometric scanning of autoradiographs from DPSAs (as described previously in Ref. 36; data not shown): WY-14,643 Ͼ Ͼ ETYA Ͼ LY-171883 Ͼ Ͼ clofibric acid Ͼ clofibrate.
These results suggest that the most carboxyl-terminal 21 mPPAR␣ residues (448 -468 corresponding to all of putative helix 12 and a portion of helix 11; see Fig. 1A) are important for mPPAR␣ responsiveness to ETYA. With the possible exception of WY-14,643, 35 S-PPAR⌬AB/⌬425 did not exhibit a differential proteolytic pattern in the presence of any PPAR ligands examined (Fig. 4C) suggesting that the extreme carboxyl-terminal mPPAR␣ amino acids may be required for responsiveness to many PPAR ligands (see below).
Induction of Proteolytic Fragments Is Dependent on Ligand Concentration-DPSAs were conducted using 35 S-PPAR⌬AB at a constant chymotrypsin concentration and increasing concentrations of PPAR ligands (WY-14, 643, ETYA, LY-171883, CFA; see Fig. 1B) to determine the dependence of PF33, PF31, and PF27 on ligand concentration. Induction of all proteolytic fragments from 35 S-PPAR⌬AB was clearly ligand-dependent in all cases (Fig. 5A-D), and the relative potencies with which these compounds induced 35 S-PPAR⌬AB conformational change in vitro was generally consistent with previously reported transcriptional activation studies (Refs. 7, 13, 19; see "Discussion").  (Fig. 6A, compare lanes 2 and 6) or 35 S-PPAR⌬AB/⌬448 (Fig. 6B, compare  lanes 2 and 6). In addition, the concentration dependence of WY-14,643 on the induction of proteolytic fragments derived from either receptor did not differ noticeably in the presence of mRXR␣ (compare lanes 2-5 with lanes 6 -9 of Fig. 6A and B, respectively). Similar results were observed for both 35 S-PPAR⌬AB and 35 S-PPAR⌬AB/⌬448 when using the PPAR ligands clofibric acid, clofibrate, LY-171883, and ETYA (data not shown). Moreover, the rank order of efficacy of the five compounds tested for induction of PFs within PPAR⌬AB and PPAR⌬AB/⌬448 did not differ from that stated above (data not shown). Therefore, heterodimerization with mRXR␣ does not appear to influence, positively or negatively, the capacity of mPPAR␣ to bind PPAR activators and undergo ligand-induced conformational changes. 35 S-PPAR⌬AB/⌬425 was not examined in these experiments due the inability of this receptor mutant to interact with mRXR␣ ( Figs. 2 and 3) or bind ligand (Fig. 4C). DISCUSSION Our results suggest that the extreme carboxyl-terminal amino acids of mPPAR␣ are required for formation of PPAR⅐RXR heterodimeric complexes both in solution and bound to DR1 and ACO-PPRE probes. This finding is in agreement with a previous study that characterized a hPPAR␣ point mutation (Leu 433 3 Arg corresponding to the same residue in mPPAR␣) which abolished heterodimerization with RXR (32) thus illustrating a critical role for this region (which is deleted in PPAR⌬AB/⌬425). A putative leucine zipper-like heptad repeat, located between residues 426 -433 of mouse, human, and rat PPAR␣, has been postulated to mediate heterodimerization with RXR (32). Truncation of mPPAR␣ to amino acid 447 (PPAR⌬AB/⌬448) gave rise to a receptor protein that was capable of interacting with RXR and binding to DR1 and ACO-PPRE probes, albeit at a 2-fold decreased efficiency as com-  4-6) or GST (lanes 7-9) immobilized on glutathione-Sepharose beads was incubated with 35 S-PPAR⌬AB, 35 S-PPAR⌬AB/⌬448, and 35 S-PPAR⌬AB/⌬425 (ϳ100 fmol) and extensively washed as described under "Materials and Methods." The beads were resuspended in 30 l of 2 ϫ SDS sample buffer and boiled, and 15 l were loaded on a 12.5% SDS-polyacrylamide gel. Input lanes represent ϳ10 fmol of 35 S-PPAR⌬AB, 35 S-PPAR⌬AB/⌬448, and 35 S-PPAR⌬AB/⌬425 (lanes 1-3, respectively). Electrophoresis and gel processing were carried out as described under "Materials and Methods." The positions of Bio-Rad prestained low molecular mass standards are indicated.

PPAR Activator-induced Alteration in Chymotryptic Sensitivity of mPPAR␣ Is Not Altered by Heterodimerization with mRXR␣-Because
pared with a receptor protein with an intact carboxyl terminus (PPAR⌬AB). Considered together, these results suggest that the mPPAR␣ dimerization interface contains at least Leu 433 , which is 100% conserved across all PPAR subtypes (32, data not shown), and extends through at least Ile 447 . In addition to heterodimerizing with RXR, PPARs have been reported to interact with thyroid hormone receptor (Ref. 45), and more recently, Miyata et al. (46) reported that mPPAR␣ interacts with a third member of the nuclear receptor superfamily, the orphan receptor LXR␣. Therefore, it appears that there may be phys- FIG. 4. Ligand-induced mPPAR␣ conformational change. A, 35 S-PPAR⌬AB subjected to DPSA. 35 S-PPAR⌬AB (ϳ10 fmol) was preincubated for 30 min at room temperature with either vehicle (lanes 1-4), 1 mM clofibric acid (CFA, lanes 5-7), 1 mM clofibrate (CLO, lanes 8 -10), 100 M LY-171883 (lanes 11-13), 100 M ETYA (lanes 14 -16), or 100 M WY-14,643 (lanes 17-19) before addition of chymotrypsin (final concentrations of 75, 150, and 300 g/ml, respectively, in lanes 2 -4, 5-7, 8 -10, 11-13, 14 -16) or water (lane 1). Proteolytic digestions were carried out at room temperature for 20 min, after which time samples were denatured and electrophoresed on a 12.5% SDS-polyacrylamide gel. Gels were processed as described under "Materials and Methods." B, 35 S-PPAR⌬AB/⌬448 subjected to DPSA. Preincubations, electrophoresis, and gel processing were carried out as described in A. Final concentrations of chymotrypsin were 20, 50, and 100 g/ml, respectively, in lanes 2 -4, 5-7, 8 -10, 11-13, 14 -16. C, 35 S-PPAR⌬AB/⌬425 subjected to DPSA. Preincubations, protease concentrations, electrophoresis, and gel processing were carried out as described in B. Arrows throughout the figure indicate positions of proteolytic fragments and migration of Bio-Rad prestained low molecular mass standards. Note that unproteolyzed receptor preparations incubated with vehicle alone (lane 1) were indistinguishable from those incubated with all ligands tested (data not shown). Clofibric acid and clofibrate are abbreviated as CFA and CLO, respectively. iologically relevant cross-talk between PPARs and signaling pathways mediated by other nuclear receptors. It will be of interest to determine if other nuclear receptors interact with PPARs through distinct or common heterodimeric protein interfaces and if these protein-protein interactions and/or the functional capacities of the involved receptors are allosterically regulated by DNA binding as previously demonstrated for other nuclear receptors (47,48).
PPARs, like other receptor proteins within the nuclear receptor superfamily, exhibit a conserved subdivision of receptor regions referred to as A/B, C, D, and E/F (49, reviewed in Refs. [2][3][4][5]. Experiments conducted with various chimeric receptor proteins composed of putative PPAR␣ ligand binding domains (LBDs) fused to heterologous DNA binding domains from estrogen (7,50) and glucocorticoid (8) receptors, bacterial tetracycline repressor (14), and GAL4 (44,51,52) have demonstrated the requirement for a large portion of the carboxyl terminus of PPARs (D and E/F regions as defined in Ref. 7) for ligand-responsive transcriptional activation.
PPAR activating ligands constitute a chemically diverse group of compounds in which the most obvious common structural elements are an acidic group (free carboxyl group, a metabolically labile derivative thereof, or a bioisostere such as a tetrazole or sulfonamide moiety) and a electron-rich region (aromatic ring or series of alkenes or alkynes) (53). When considering the structural diversity exhibited by these compounds, it seems possible that the molecular determinants of mPPAR␣ interaction with each ligand or class of ligands may be distinct. Indeed, our results indicate that distinct mPPAR␣ regions are required for responsiveness to different PPAR activators. While PPAR⌬AB is responsive to WY-14,643, ETYA, LY-171883, clofibric acid, and clofibrate, as detected by DPSAs, deletion of mPPAR␣ residues 448 -468 (PPAR⌬AB/⌬448) severely compromises responsiveness to ETYA but not other PPAR ligands. The distal carboxyl-terminal amino acids of mPPAR␣ that are deleted in PPAR⌬AB/⌬448 correspond to part (H12) of the region that has been proposed to stabilize ligand-receptor interactions with hRAR␥ by functioning as a "lid" on the ligand binding cavity (38,39). The greatly reduced efficacy with which ETYA induced PPAR⌬AB/⌬448 conformational change relative to that of PPAR⌬AB suggests that the hydrophobicity of putative H12 may play a critical role in the stabilization of ETYA binding, perhaps by stabilizing an extended conformation of this compound. Truncation of mPPAR␣ residues 425-468 (PPAR⌬AB/⌬425) gave rise to a receptor protein which was slightly responsive to WY-14,643 but unresponsive to all other PPAR ligands examined. In addition to deletion of putative helix H12, PPAR⌬AB/⌬425 also lacks putative helices H10 and H11, encompassing a region that has been proposed to form one side of the nuclear receptor ligand binding pocket (39), which may explain the inactivity of this mutant in DPSAs. However, we cannot presently rule out the possibility that the inactivity of PPAR⌬AB/⌬425 is due to improper protein folding and/or detrimental structural distortions outside the deleted region. Nonetheless, it is clear that mPPAR␣ residues 448 -468 are important for ligand binding and/or conformational change induced by ETYA while being dispensable for responsiveness to other PPAR ligands support-  -9), was added and the reaction allowed to proceed for 20 min at room temperature. Electrophoresis and gel processing were as described in Fig. 4. B, 35  ing the hypothesis that distinct mPPAR␣ receptor regions may be required for interaction with structurally dissimilar PPAR activators.
Recently, several synthetic antidiabetic thiazolidinediones (44,51,52) and 15-deoxy-⌬ 12,14 -prostaglandin J 2 (44,52) have been shown to bind directly to the ligand binding domain of the mouse PPAR␥. Direct binding of any compounds to PPAR␣ subtypes, however, has not been demonstrated. Issemann et al. (7) specifically report the lack of [ 3 H]nafenopin binding by mPPAR␣ which may be due to both a low affinity of nafenopin for mPPAR␣ and a large amount of endogenous binding activity in the cell lines tested (7). In the current study, DPSA methodology was adapted to facilitate detection of ligand interactions with mPPAR␣. DPSAs have been demonstrated to be a useful method for detection of ligand-induced conformational change within the nuclear receptor superfamily (36, 54 -64). The differential sensitivity of liganded and unliganded receptors is likely related to ligand-induced stabilization of one or more receptor conformations that exhibit sensitivity to proteolytic digestion different from that of unliganded receptor. Although DPSAs yield data that are somewhat less amenable to quantitative analyses than radioligand binding experiments, an advantage of the former is the ability to detect low affinity ligand-receptor interactions. All mPPAR␣ activators tested induced mPPAR␣ conformational change in vitro, albeit it with varying potencies. However, the relative potencies of these compounds with regard to induction of PPAR⌬AB conforma-tional change in vitro is generally consistent with previously reported transcriptional activation studies utilizing mPPAR␣: WY-14,643 Ͼ Ͼ ETYA Ͼ LY-171883 Ͼ Ͼ clofibric acid (7,13,19). ETYA, a potent activator of Xenopus PPAR␣ (11), was a substantially weaker ligand than WY-14,643 in the in vitro studies employing PPAR⌬AB described herein (Fig. 5). Hsu and coworkers (13) also reported that ETYA was approximately 10fold weaker than WY-14,643 as an activator of mPPAR␣ in transient transfection experiments, suggesting that the enhanced potency of ETYA reported by Keller and co-workers (11) may be conferred by species-specific receptor activity, cell-specific factors, and/or mechanism(s) other than direct binding of this arachidonic acid analog to the receptor.
It has been hypothesized that Hsp72, which has been demonstrated to interact directly with rat PPAR␣, may bind PPAR activators and, in turn, allosterically activate the associated receptor protein (65). Presently, we cannot exclude this possibility; however, results from DPSAs in which mPPAR␣ carboxyl-terminal truncation mutants were used suggest an alternative molecular mechanism. For example, deletion of 21 mPPAR␣ carboxyl-terminal amino acids (PPAR⌬AB/⌬448) compromises the responsiveness of the receptor to ETYA but not other PPAR ligands. If Hsp72 binds these ligands directly and allosterically transduces a signal that alters the conformation of mPPAR␣, it seems unlikely that this process would be attenuated by carboxyl-terminal truncation of mPPAR␣ unless the deleted amino acids are required for mPPAR␣-Hsp72 interaction. In such a case, one would expect this truncation to abolish responsiveness to all ligands. Of course, it is possible that some ligands selectively interact with mPPAR␣, Hsp72, or both proteins in the context of an mPPAR␣⅐Hsp72 complex. The latter possibility would be reminiscent of ecdysone receptor in which ecdysone binding activity is associated with a complex of ecdysone receptor and ultraspiracle (61). In any event, ligandinduced conformational change likely underlies the molecular basis of ligand activation of the putative mPPAR␣ transcriptional activation function, AF-2, and identification of ligandinduced mPPAR␣ proteolytic fragments will be of critical importance to our understanding of the dynamic process of PPAR activation by ligands and interaction of liganded receptor with putative transcriptional intermediary factors.