Association of Peroxisome Proliferator-activated Receptor and Hsp72*

In an effort to understand the relationship between a 72-kDa heat shock protein (Hsp72) and peroxisome pro-liferator-activated receptors (PPARs), we have charac- terized their interaction using clofibric acid-Sepharose chromatography and co-immunoprecipitation with an- tisera raised against either rat PPAR (rPPAR) or Hsp72. First, we observed that both rPPAR and Hsp72 elute in a clofibrate-dependent manner from the clofibric acid-Sepharose matrix. Second, we found that immunopre- cipitation of either protein from solution resulted in the precipitation of the other. This result was obtained from rat liver cytosol, from Spodoptera frugiperda (Sf91 in- sect cells expressing rPPAR, and from reticulocyte lysate rPPAR expression systems. These results suggest that Hsp72 and rPPAR form a complex in vivo and that Hsp72 may play a role in the folding, subcellular local- ization, andor signaling pathway of PPARs.


Association of Peroxisome Proliferator-activated
In an effort to understand the relationship between a 72-kDa heat shock protein (Hsp72) and peroxisome proliferator-activated receptors (PPARs), we have characterized their interaction using clofibric acid-Sepharose chromatography and co-immunoprecipitation with antisera raised against either rat PPAR (rPPAR) or Hsp72. First, we observed that both rPPAR and Hsp72 elute in a clofibrate-dependent manner from the clofibric acid-Sepharose matrix. Second, we found that immunoprecipitation of either protein from solution resulted in the precipitation of the other. This result was obtained from rat liver cytosol, from Spodoptera frugiperda (Sf91 insect cells expressing rPPAR, and from reticulocyte lysate rPPAR expression systems. These results suggest that Hsp72 and rPPAR form a complex in vivo and that Hsp72 may play a role in the folding, subcellular localization, andor signaling pathway of PPARs. Many hypolipidemic drugs, leukotriene D4 antagonists, herbicides, industrial solvents, and plasticizing agents induce a characteristic pleiotropic response in the liver of rodents that includes hepatomegaly, an increase in the number of peroxisomes in liver parenchymal cells, and increases in the expression of enzymes involved in the peroxisomal P-oxidation of lipids. Structurally diverse compounds such as clofibrate, nafenopin, ciprofibrate, Wy-14,643, phthalate esters like di-(2ethylhexyl)phthalate, and the adrenal steroid dehydroepiandrosterone that induce this response are collectively termed "peroxisome proliferators" (PPs)' (1)(2)(3). Chronic exposure of rodents to PPs results in the development of hepatocellular carcinomas. Given that these compounds do not covalently modify DNA, they have been classified as nongenotoxic carcinogens (1, 3). Induction of the enzyme activities of peroxisomal P-oxidation system has been shown to be due to an increase in the transcription rates of the genes encoding their respective mRNAs (4). The coordinate increases in the expression of the three genes of the peroxisomal P-oxidation system (i.e. fatty acyl-CoA oxidase; enoyl-CoA hydratasel3-hydroxyacyl CoA-dehydrogenase bifunctional enzyme, and thiolase) suggested a common mechanism of induction and led to the speculation * This work was supported by National Institutes of Health Grant GM23750 and by the Joseph L. Mayberry Sr. Endowment Research Fund. 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. that a peroxisome proliferator-activated receptor (PPAR) might exist (5). Using transient expression systems, a mouse cDNA encoding a novel member of steroid hormone receptor superfamily was shown to be required for PP-induced expression of reporter genes linked to elements present in the 5"flanking region of the bifunctional enoyl-CoA hydratasel3-hydroxyacyl-CoA dehydrogenase and fatty acyl-CoA oxidase genes (6,7). Given this activity, the protein encoded by this cDNA has been referred to as a bona fide PPAR. Recently, It has been shown that PPAR forms a heterodimeric complex with the 9-cis-retinoic acid receptor. The PPAR-9-cis-retinoic acid receptor complex may activate gene expression, in a PP-dependent manner by binding to a direct repeat in the peroxisome proliferatorresponsive element of the fatty acyl-CoA oxidase gene (8,9). In addition to the murine form, PPAR has also been isolated from the rat (rPPAR) (101, human (hPPAR) (111, and three closely related members (xPPARa, xPPAR0, xPPARy) from Xenopus laevis (12). The considerable homology of the various PPAR forms suggests that these receptors have been conserved throughout evolution and that more than one PPAR may exist in a given species. In keeping with this concept, a murine PPAR that is homologous to xPPARy has recently been cloned by this laboratory and has been designated mPPARy (13).
Despite the compelling evidence to implicate PPARs in signal transduction by compounds like clofibrate, there is presently no evidence to indicate that any PP actually binds to any of the currently identified PPARs. In our previous attempts to purify PP-binding proteins, we observed that heat shock protein (Hsp72) has a high affinity for clofibric acid-Sepharose and can be specifically eluted from such columns (14,15). In this report, we attempt to elaborate on our previous clofibrate binding studies and provide evidence that PPARs and Hsp72 form strong associations in vivo.
EXPERIMENTAL PROCEDURES Generation of Polyclonal Antibodies against rPPAR-The full-length coding region of rPPAR cDNA (gift from Dr. Frank Gonzalez, National Institutes of Health) was cloned into the BamHI site of the bacterial expression vector (PGEX-2T). After transformation into Escherichia coli JM109 and induction by isopropyl-1-thio-p-D-galactopyranoside, GST-PPAR fusion protein was expressed and subsequently purified (16). The purified fusion protein was used to raise polyclonal antibodies against rPPAR in rabbits by standard immunization protocols (17). For purification of the anti-serum, the PGEX-ST parent vector was used to overexpress GST protein and the bacterial lysates passed through a glutathione-agarose column. The column was washed with phosphatebuffered saline to remove unbound proteins and then used to remove antibodies against GST protein. Briefly, the anti-PPAR fusion protein antiserum was passed through the column. The unbound fractions were then passed once again through the affinity column. The unbound fractions, after the second pass, which were essentially free of anti-GST antibodies were collected and used for the experiments.
Expression of rPPAR in Insect Cells-The full-length rPPAR was modified using the polymerase chain reaction to add an oligonucleotide encoding six histidine residues and a 5'-ATG start codon to the 5'-end of the cDNA (coding strand, 5'-AAT GCG GCC GCT ATG CAT CAC CAT __ 8493 CAC CAT CAC) (9). The modified rPPAR cDNA was then cloned into the Not1 and BamHI sites of baculovirus transfer vector PvL1392 (Invitrogen) and used to generate a recombinant virus. This resultant virus stock was used to infect Spodoptera frugiperda (Sf9) insect cells, and the recombinant rPPAR was expressed under the control of the polyhedrin promoter (18). The infected sf9 cells were metabolically labeled with 200 pCi of I:'%lrnethionine for 4 h before they were harvested.
Purification of Clofihric Acid-binding Proteins-Clofibric acid was linked to Sepharose by coupling its -COOH group to the amino group of AH-Sepharose through the carbodiimide reaction (14,15). Following coupling, the affinity column was washed extensively and resuspended in HEDG buffer (IO mu HEPES, pH 7.5, 1 mb! EDTA, 1 m i dithiothreitol, 10% glycerol) a t 4 "C.
F344 male rats, weighing about 120 g. were starved for 12 h, after which they were euthanized with CO,. The livers were perfused via the portal vein with ice-cold saline and homogenized in 4 volumes of HEDG buffer containing 0.3 mx? phenylmethylsulfonyl fluoride, 0.15 Y KCI, and 0.1% Triton X-100. The liver homogenates were gently stirred a t 4 "C for 30 min and then centrifuged for 30 min a t 100,500 x g. Using a batch procedure, the 100,500 x g supernatant was added to 10 ml of clofibric acid-Sepharose and gently shaken overnight a t 4 "C. After extensive washing, clofibric acid-binding protein(s) were eluted by incubation of affinity matrix with 3 my clofibric acid a t room temperature for 1 h, as described previously (14,15). Alternatively, small scale purification of clofibric acid-binding proteins were employed in some experiments using 1 ml of rat liver cytosol mixed with 100 pl clofibric acid-Sepharose in 2-ml Eppendorftubes and gently shaken overnight a t 4 "C. The resin was then first washed 4 times with ice-cold HEDG buffer containing 0.4 JI KC1 and then at room temperature four times with HEDG buffer plus 1 >I KC1 and twice with HEDG buffer containing 0.4 XI KCl. The retained proteins were eluted a t room temperature for 20 min with 200 pl of 1 my clofibric acid (or other compounds) in HEDG containing 0.4 51 KCI.
Immunoprecipitation of rPPm by Monoclonal Antibodies Raised against Hsp7O-An eukaryotic expression vector containing the fulllength rat PPAR (pSVsport-PPAR) was constructed by cloning rPPAR cDNA into the SmaI site of pSVsport. I:'"SlMethionine-labeled rPPAR was generated by translation of the cDNA in a rabbit reticulocyte lysate system from the Sp6 promoter (TNT-translation system, Promega). Immunoprecipitation reactions (500 1. 11) included 10-20 p1 of liver cytosol (approximately 10 pg of rat liver cytosolic protein) and 10 pl of in vitro synthesized I:'5S1methionine-labeled rPPAR in 20 my Tris-HCI, pH 8.0. The protein mixture was incubated for 30 min on ice before the addition of 5 pl of a monoclonal antibody (3A3) raised against Hsp7O (Affinity Bioreagents). After overnight incubation a t 4 "C, antigen-antibody complexes were collected by the addition of 50 p1 of 50% protein A-agarose (Sigma) and the immune complexes washed four times with 500 pI of RIPA buffer (10 m M Tris-HCI, pH 8.0, 150 m M NaCI, 1% Triton X-100, 1% sodium deoxycholate). After the final wash, the complexes were resolved by SDS-PAGE on 10% gels, stained with Coommasie Blue in 5 0 9 methanol and 7% acetic acid, destained, dried, and subjected to autoradiography a t -70 "C overnight (8).

RESULTS
Characterization of Polyclonal Anti-rPPM Antibodies-In a n effort to obtain a PPAR-specific reagent that was useful in both Western blot analysis and immunoprecipitation experiments, we raised rabbit antisera against a recombinant GST-PPAR fusion protein. To characterize the purified antibodies, we performed Western blot analysis and/or immunoprecipitations on PPAR that had been expressed in bacteria, baculovirus, and reticulocyte lysates (Fig. 1). In each case, the antibodies recognized a 55-kDa band. The presence of a single crossreacting band in rat liver cytosol (Fig. 1, lane 2 ) indicates that the purified antibodies can detect the endogenous rPPAR in this tissue. As a n additional proof of antibody specificity and to demonstrate that the addition of a polyhistidine tag to the PPAR would lead to a rapid purification of PPAFt, we also performed Western blot analysis on the baculovirus-expressed rPPAR-6his protein that had been purified by Ni-NTA chroma- tography. Again the antibodies recognized the 55-kDa protein with no cross-reactivity under the conditions employed (Fig. 1, l a n e 7).
Binding of rPPAR to Clofibric Acid-Sepharose Affinity Matrix-SDS-PAGE of rat liver proteins that bound to clofibric acid-Sepharose yielded a major 72-kDa protein and minor bands migrating at 55 (Fig. 2 A ) and 31 kDa (14). Since we had previously identified the 72-kDa protein as Hsp72 (15), we first confirmed the identity of this protein with anti-Hsp70 monoclonal antibodies (Fig. 2B, lane 3 ) and then focused on the identities of those proteins that displayed migrations similar to our recombinant rPPAR (i.e. 55 kDa). Western blot analysis demonstrated that our anti-PPAR antibodies recognized a 55-kDa protein that was specifically eluted from the column with clofibric acid (Fig. 2B, lane 2 ). Since preimmune serum did not react with this protein (Fig. 2B, lane 1 ), the results suggest that PPAR from rat liver cytosol is retained on the clofibric-acid Sepharose.
To further examine the elution specificity from the clofibric acid-Sepharose column, various elution conditions were studied. Western blot analysis indicated that both Hsp72 and rPPAR were co-eluted by 1 mM clofibric acid, ciprofibrate, and nafenopin, but were not eluted using either dimethyl sulfoxide or phenobarbital, a compound that does not produce peroxisome proliferation in vivo (Fig. 3). These data suggest that both Hsp72 (Fig. 3A) and rPPAR (Fig. 3B) are specifically eluted from clofibric acid-Sepharose in a stereospecific manner by a number of PPs.
Co-immunoprecipitation of Hsp72 a n d PPM-The elution of Hsp72 and rPPAR from the clofibric acid affinity chromatography suggested to us that rPPAR and Hsp72 form a complex and that this complex binds to clofibric acid-Sepharose. Therefore, we attempted to determine whether rPPAR and Hsp72 are associated in solution. To test this, we performed a series of co-immunoprecipitation experiments using rat liver cytosol (15), rPPAR that had been translated in a reticulocyte lysate (known to be rich in Hsp72) and PPAR-6his protein that was expressed in baculovirus-infected Sf9 cells. As shown in Fig. 4 (lanes 2 and 3 ) a monoclonal antibody that recognizes all three members of the Hsp70 family (the constitutive, inducible, and GRP78), efficiently co-immunoprecipitated [35Slmethionine-labeled rPPAR from Sf9 cells and reticulocyte lysate, indicating the association of these two proteins in vitro, as well as in vivo.
As control experiments, we demonstrated that neither protein A-agarose (data not shown) nor nonspecific mouse serum were able to precipitate the in vitro translated rPPAR (Fig. 4, lane 1 ). It should be noted that the antibodies against Hsp7O precipitated three radioactive bands when PPAR was translated in vitro. When subjected to Western blot analysis, all three bands reacted with the anti-rPPAR antibodies (Fig. 4, lane 4 ) , indicating that they may either be partially synthesized or de- FIG. 4. Co-immunoprecipitation of PPAR with Hsp72. rPPAR was transcribed and translated in rabbit reticu1oc.yte lysate as described under "Experimental Procedures." Co-immunoprecipitation reaction (500 yl) containing 10 p1 of in vifro translated I""Slmethionine rPPAR and 10 pg of rat liver cytosolic protein was incubated on ice for 30 min. The sf9 cells infected with PPAR baculovirus were labeled with I:'5Slmethionine for 4 h before the cellular extracts were prepared. Following incubation 5 pl of anti-Hsp antibody was added to these mixtures and the reaction incubated overnight at 4 "C. The immune complex was washed and electrophoresed in 10% SDS-polyacrylamide gels. After electrophoresis, the gels were dried and bands visualized by autoradiography. Lane 1, immunoprecipitation with nonimmune mouse serum. Lane 2, immunoprecipitation with anti-Hsp70 antibodies. Lane 3 , Sf9 insect cells infected with rPPAR-recombinant baculovirus, which were labeled with I:'"Slmethionine. lysed, and incubated with anti-Hsp70 antibody. The immune complexes were precipitated with protein A-agarose, washed, and electrophoresed in 10% polyacrylamide gels. Lone 4 , immunoblot analysis of the i n vitro translated protein precipitated in lane 2 using anti-PPAR antibodies.
graded products of rPPAR.
Association of Hsp72 with Baculovirus-expressc.d rPPAR-To confirm the association of Hsp72 with PPAR, baculovirus-expressed rPPAR-6his protein was purified on a Ni-NTA affinity column. The column was washed with 50 m%t imidazole containing 300 mM NaCl and then eluted with 500 mM imidazole, 300 mxt NaCI. The eluate was subjected to Western blot analysis using either anti-rPPAR (Fig. 5A, lane I ) or anti-Hsp70 antibodies. As seen in Fig. 5, the anti-Hsp antibodies raised against the conserved region of the Hsp7O family were capable of recognizing the insect cell Hsp72 analogue, which co-purified with rPPAR (Fig. 5B, lane 2 ) . When extracts of uninfected insect cells were passed through the Ni-NTA affinity matrix, washed, and eluted, no Hsp72 was detected in the eluate (Fig.  5B, lane 1 ), verifying that the presence of Hsp72 is through its association with rPPAR and not the Ni-NTA resin.
In addition, immunoprecipitation with anti-Hsp antibody of Sf9 insect cell lysates infected with recombinant rPPAR-baculovirus resulted in co-precipitation of rPPAR and Hsp72 in the immune complex (Fig. 6). Conversely, anti-rPPAR antibody coprecipitated the insect analogue of Hsp72, once again confirming the association of these two proteins. Proteins that were immunoprecipitated with anti-Hsp70 when immunoblotted with anti-rPPAR (Fig. 6 A ) revealed the presence of rPPAR (Fig.  6A, lane 2 ). Likewise, proteins that were immunoprecipitated with anti-rPPAR revealed the presence of Hsp72 (Fig. 6B, lane  3 ) . These results clearly indicate that both Hsp72 and rPPAR are in complex in Sf9 cells expressing this recombinant protein.

DISCUSSION
We have previously isolated a clofibric acid-binding protein from rat liver cytosol by clofibric acid-Sepharose chromatography and found it to be a member of the Hsp70 protein family (1,14). In  sess an apparent molecular masses in the range of 52-55 kDa, we attempted to determine if any PPARs corresponded to any of the major proteins that elute from our clofibric acid-Sepharose columns (15). Western blot analysis, using anti-PPAR antibodies, indicated that rPPAR was in fact specifically eluted from this column (Figs. 2 and 3).
The observation that both Hsp72 and PPAR were eluted from a clofibric acid-Sepharose column by a number of known PPs led us to ask if PPAR and Hsp70 eluted from the column independently or if they eluted as a complex. We found by immunoprecipitation studies that antibodies specific to either protein could elute a complex of the two proteins from rat liver cytosol, a reticulocyte lysate system, and Sf9 cell extracts. These data, coupled with the co-elution studies, cited above, strongly suggest, but do not prove, that a PPAR.Hsp70 complex has a high affinity for peroxisome proliferators. Interestingly, our results do not allow us to determine which component of the PPAR.Hsp70 complex has the binding site for PPs or in this case specifically clofibric acid. Given that Coomassie Blue staining of eluants suggest that a greater molar quantity of Hsp72, over PPAR, is associating with clofibric acid-Sepharose, it is tempting to speculate that it is Hsp72 that is actually binding clofibric acid. Such a possibility would suggest that Hsp72 may actually harbor the ligand binding site for PPs.
Our speculation for a PP binding site on Hsp72 is tentative and should be confirmed by affinity labeling or more complete structure activity studies. In this regard, it is important to note that many attempts to establish the direct binding of PPAR to PPs have failed. The reasons have been suggested to be due to the low affinity of PPs for their binding site or existence of high concentration of endogenous ligands in vivo (6). In many respects, the association of Hsp72 with PPAR is not surprising. Hsp70 is known to function as an intracellular chaperone and to facilitate the folding and unfolding of proteins (20,21). It has been shown that heat shock proteins play a role in the functions of steroid hormone receptors. For example, both the Hsp9O and Hsp70 proteins were found in a complex with the glucocorticoid and progesterone receptors (22-24). Hsp9O was also found to form a complex with the dioxin receptor (AhR) and to modulate its ligand and DNA binding functions (27). Recently, members of the Hsp superfamily, including Hsp56 and Hsp70, were found in an association with a group of immunophilins such a s cyclosporin, FK506, rapemycin, and deoxypersgualin-binding proteins (26)(27)(28). Hsp7O was also found to interact with tumor suppressor gene products, including the retinoblastoma gene product pRb 110 (29) and p53 gene (30).
Our results showing that PPAR forms a complex with a member of the Hsp70 family suggest that the heat shock protein may participate in post-translational modification process such as the folding of PPAR. Alternatively, Hsp72 may act as a carrier or molecular chaperon to translocate PPAR from the cytoplasm to the nucleus to accomplish its transactivation function. Support for this comes from the demonstration that the constitutive form Hsp72 is responsible for the translocation of certain nuclear proteins (21). However, the intracellular translocation of PPAR a s well as Hsp72 responding to peroxisome proliferator administration remains largely unknown. Understanding the relationships between the binding proteins and putative receptors should help elucidate the mechanisms by which these xenobiotics induce certain gene activation and liver tumors in rats and mice.