Regulation of peroxisome proliferator-activated receptor gamma activity by mitogen-activated protein kinase.

Adipocyte differentiation is regulated both positively and negatively by external growth factors such as insulin, platelet-derived growth factor (PDGF), and epidermal growth factor (EGF). A key component of the adipocyte differentiation process is PPARgamma, peroxisomal proliferator-activated receptor gamma. To determine the relationship between PPARgamma activation and growth factor stimulation in adipogenesis, we investigated the effects of PDGF and EGF on PPARgamma1 activity. PDGF treatment decreased ligand-activated PPARgamma1 transcriptional activity in a transient reporter assay. In vivo [32P]orthophosphate labeling experiments demonstrated that PPARgamma1 is a phosphoprotein that undergoes EGF-stimulated MEK/mitogen-activated protein (MAP) kinase-dependent phosphorylation. Purified PPARgamma1 protein was phosphorylated in vitro by recombinant activated MAP kinase. Examination of the PPARgamma1 sequence revealed a single MAP kinase consensus recognition site at Ser82. Mutation of Ser82 to Ala inhibited both in vitro and in vivo phosphorylation and growth factor-mediated transcriptional repression. Therefore, phosphorylation of PPARgamma1 by MAP kinase contributes to the reduction of PPARgamma1 transcriptional activity by growth factor treatment.

Peroxisome proliferator-activated receptors (PPARs) 1 are members of the nuclear hormone receptor superfamily (1). These receptors heterodimerize with retinoic acid-like receptor, RXR, and become transcriptionally active when bound to ligand. The three PPAR isoforms (␣, ␦, and ␥) differ in their C-terminal ligand binding domains, and each appears to bind and respond to a specific subset of agents including hypolipidemic drugs, long chain fatty acids, aracadonic acid metabolites, and antidiabetic thiazolidinediones (2)(3)(4). PPAR␥ is expressed predominantly in mouse white and brown fat, with lower levels in liver, whereas PPAR␣ is present in heart, kidney, and liver (5,6). PPAR␦ expression is ubiquitous (7,8).
Ectopic expression of either PPAR␣ or PPAR␥ in NIH-3T3 cells is sufficient to induce adipocyte differentiation in the presence of PPAR␥ activators (9,10). The rapid induction of PPAR␥ during adipocyte differentiation and its enriched expression in adipose tissues suggest that PPAR␥ is responsible for the initiation and maintenance of the adipocyte phenotype in vivo (9). Previously two isotypes of PPAR␥ (PPAR␥1 and PPAR␥2) have been identified in 3T3-L1 adipocytes (11). Zhu et al. (12) have demonstrated that these two isotypes are derived from a single PPAR␥ gene by alternative promoter usage and RNA splicing. However, thus far, no functional difference has been found between the two isotypes.
Adipogenesis is a complex process; multiple hormones and factors regulate the conversion of progenitor cells to adipocytes. Insulin and/or insulin-like growth factor enhance the ability of PPAR ligand to induce differentiation of both 3T3-L1-and PPAR␥-overexpressing cell lines (9,13). In contrast, growth factors such as epidermal growth factor (EGF), platelet-derived growth factor (PDGF), and fibroblast growth factor inhibit adipocyte conversion (14 -18). In this report, we find that activation of EGF and PDGF receptors and subsequent phosphorylation of PPAR␥1 by the MAP kinase signaling pathway decreases PPAR␥1 transcriptional activity. This repression is mediated by MAP kinase phosphorylation of Ser 82 on PPAR␥1. These studies identify PPAR␥1 as a substrate of MAP kinase and provide evidence for regulation of PPAR␥1 activity by phosphorylation.

EXPERIMENTAL PROCEDURES
Chemicals and Materials-Cell culture reagents were purchased from Life Technologies, Inc. The ECL detection system and carrier-free [ 32 P]orthophosphate were obtained from Amersham Corp. The PDGF was purchased from Intergen, while EGF was from Harlan. PD98059 and BRL49653 were synthesized at Parke-Davis Pharmaceutical Research Division of Warner-Lambert Co.
Vector Constructs and Transient Transfection-For eukaryotic expression of PPAR␥1 and RXR␣, the entire PPAR␥1 or RXR␣ cDNA was inserted 3Ј to the cytomegalovirus promoter in pSG5 (Stratagene). Constitutively active MAP kinase kinase (CA-MEK), which contains mutations at Ser 218 to Glu and Ser 222 to Glu was obtained from Dr. S. Decker (Parke-Davis). Site-directed mutagenesis of PPAR␥1/pSG5 was conducted using the MORPH site-specific plasmid DNA mutagenesis system (5 Prime 3 3 Prime, Inc., Boulder, CO). The oligonucleotide used in mutagenesis was CAAAGTAGAACCTGCAGCTCCACCTTAT-TATTCTGAAAAGACCC and changed Ser 82 to Ala. The reporter construct used in the transfections contained three copies of the PPRE site from the aP2 enhancer (ARE7) inserted upstream of a minimal thymidine kinase (TK) promoter in the pGL3 basic luciferase vector (a gift from Dr. R. Wyborski). All constructs were sequenced prior to use. For the transient transfection, NIH 3T3 cells were grown in 10% fetal calf serum/Dulbecco's modified Eagle's medium and co-transfected with various expression plasmids and pCMV ␤-galactosidase plasmid (Clontech) using Lipofectamine (Life Technologies, Inc.). After recovery, cells were placed in 0.5% bovine serum albumin/Dulbecco's modified Eagle's medium for 5 h and then treated with 25 M BRL49653 and/or 100 ng/ml PDGF for 16 h. Luciferase and ␤-galactosidase activities were determined using a Luciferase assay (Promega) and the Galacto-light system (Tropix, Inc.).

Production of PPAR Fusion Proteins and in Vitro Phosphorylation
* 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.
Assay-To express the maltose-binding protein (MBP) fusion proteins in Escherichia coli, the coding regions of PPAR␥1, PPAR␦, and RXR␣ were inserted downstream of the isopropyl-␤-D-thiogalactopyranosideinducible MalE-lacZ␣ gene fusion in the pMAL-C2 plasmid (New England Biolabs). Protein expression was induced with isopropyl-␤-D-thiogalactopyranoside, and the fusion proteins were partially purified by amylose affinity chromatography (19). In vitro phosphorylation of MBP, MBP-PPAR␥1, and MBP-PPAR␦ by MAP kinase was performed as described previously (20) using a bacterially expressed glutathione S-transferase fusion protein of 44-kDa MAP kinase (GST-MAP kinase) and the 45-kDa MEK (GST-MEK1). Using a PPAR␥-specific polyclonal antibody (produced using the MBP-PPAR␥ fusion protein), 2 in vitro translated PPAR␥1 and the mutant PPAR␥1 (S82A) were immunoprecipitated and phosphorylated by active GST-MAP kinase as described above.
Cell Transfection and in Vivo Radiolabeling-293T cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal calf serum (Life Technologies, Inc.) and transfected using a calcium phosphate transfection protocol according to the manufacturer (Stratagene). For in vivo labeling, transfected cells were serum-starved overnight in 0.5% bovine serum albumin/Dulbecco's modified Eagle's medium, pretreated with phosphate-free medium for 1 h, and subsequently incubated in 0.8 mCi of [ 32 P]orthophosphate at 37°C for 3 h. Cells were preincubated with either BRL49653 (25 M) or PD98059 (40 M) for 15 min followed by the addition of EGF (100 ng/ml). EGF stimulation proceeded for 5 or 15 min prior to removal of the media and cell lysis. Cells were harvested in radioimmune precipitation lysis buffer (10% glycerol, 137 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 20 mM Tris, pH 8.0, 2 mM EDTA, complete protease inhibitors, and 20 mM NaVO 4 ). Whole cell extracts were immunoprecipitated with anti-PPAR␥ antibody and protein A-Sepharose (Life Technologies, Inc.) for 16 h at 4°C and resolved in 10% SDS-PAGE. To detect MAP kinase activity in 293T cells, whole cell lysates were prepared and subjected to Western blot analysis using the anti-active MAP kinase antibody (Promega) and ECL system (Amersham).

Growth Factors
Decrease the Transcriptional Activity of PPAR␥1-Transcription reporter assays were used to determine the effect of growth factors on the transcriptional activity of PPAR␥1. The luciferase reporter constructs used in NIH3T3 cells contained the TK promoter (TKpGL3) or three copies of ARE7 PPRE elements upstream of the TK promoter (ARE7-TKpGL3). In the absence of co-transfected PPAR␥1 and RXR␣ expression plasmids, no PPAR␥ ligand (BRL49653)-dependent transcription was observed from either the TkpGL3 or ARE7-TKpGL3 (Fig. 1A). In the presence of PPAR␥1 and RXR␣, a 2-fold increase in transcription was observed from ARE7-TK reporter after 16 h of treatment with BRL49653. The addition of 100 ng/ml PDGF to these cells decreased both the basal and BRL49653-activated transcription from the ARE7. This suggests that at least a fraction of the activity from the ARE7-TKpGL3 plasmid in the absence of exogenously added ligand is due to the activation of the PPAR␥1⅐RXR␣ heterodimer by endogenous ligands. This activity was also reduced by PDGF treatment.
Close examination of the PPAR␥ amino acid sequence revealed that PPAR␥1 contains one serine residue, Ser 82 , whose surrounding amino acids correspond to the consensus phospho-rylation site for MAP kinase ( Fig. 2A) (22). This site is absolutely conserved between human and mouse PPAR␥1. A variation of the MAP kinase consensus site is also found in mouse PPAR␣ at a similar position in the amino acid sequence. PPAR␦ lacks this site altogether ( Fig. 2A). Since both EGF and PDGF are known to activate MAP kinase in vivo, a CA-MEK that constitutively activates MAP kinase was co-transfected with ARE7-TKpGL3, PPAR␥1, and RXR␣ expression plasmids. As shown in Fig. 1B  kinase in vitro, partially purified MBP, MBP-PPAR␥1, or MBP-PPAR␦ fusion proteins were incubated with preactivated GST-MAP kinase and [␥-32 P]ATP under conditions that phosphorylate myelin basic protein, a known MAP kinase substrate. As shown in Fig. 2B, MAP kinase efficiently phosphorylated PPAR␥1 but not MBP-PPAR␦ or maltose-binding protein (data not shown). Coomassie staining verified that nearly equal amounts of intact proteins were loaded (Fig. 2C). To determine if Ser 82 is the residue phosphorylated in vitro, a mutation was introduced into PPAR␥1 that changed Ser 82 to Ala. Both the wild type PPAR␥1 and the mutant PPAR␥1 (S82A) were in vitro translated and immunoprecipitated with a polyclonal anti-PPAR␥ antibody. The immunoprecipitated products were used as substrates in the in vitro MAP kinase assay. Mutation at Ser 82 to Ala completely abolished the MAP kinase-dependent phosphorylation of PPAR␥1 (Fig. 2D), indicating that the Ser 82 is the only amino acid in PPAR␥1 that is phosphorylated by MAP kinase. Fig. 2E represents the PPAR␥ antibody Western blot of the in vitro translated proteins and shows that both proteins were expressed in the reticulocyte extracts.
PPAR␥1 Is Phosphorylated by EGF Treatment-To determine if PPAR␥1 is phosphorylated by growth factor treatment, 293T cells were transfected with PPAR␥1, serum-starved for 24 h, and incubated with [ 32 P]orthophosphate. To maintain more physiologic conditions, no phosphatase inhibitors were added to the cells prior to lysis. Whole cell lysates were prepared after EGF treatment and immunoprecipitated with a PPAR␥-specific antibody. PPAR␥1 was weakly phosphorylated in the absence of sera and growth factors; however, treatment with 100 ng/ml of EGF for 5 or 15 min increased PPAR␥1 phosphorylation 1.5-and 1.8-fold, respectively (Fig. 3A, lanes 3  and 4). In 293T cells, EGF treatment stimulated MAP kinase activity as determined by Western blot analysis with the antiactive MAP kinase antibody (Fig. 3B, lanes 1 and 2). To determine if the MAP kinase signaling pathway is involved in the phosphorylation of PPAR␥1, the transfected cells were pretreated with 40 M PD98059, a specific MEK inhibitor (20) for 15 min prior to EGF treatment. PD98059 prevented EGFstimulated phosphorylation of PPAR␥1 (Fig. 3A, lane 6), suggesting that MAP kinase activation is involved in the phosphorylation of PPAR␥1. At this concentration, PD98059 inhibited MAP kinase activation by EGF (Fig. 3B, lane 3). Interestingly, pretreatment of the cells with 25 M BRL49653 for 15 min also reduced the EGF-dependent phosphorylation of PPAR␥1 (Fig.  3A, lane 5) without affecting the ability of EGF to stimulate MAP kinase activity (Fig. 3B, lane 4). This implies that occupation of the ligand binding domain may inhibit the ability of MAP kinase to recognize and/or phosphorylate PPAR␥1.

Mutation of Ser 82 to Ala in PPAR␥1 Prevents in Vivo Phosphorylation of PPAR␥1 and Transcriptional Repression by PDGF-To determine if
Ser 82 is the residue-phosphorylated in vivo in response to EGF treatment, the Ser 82 3 Ala PPAR␥1 mutant was introduced into 293T cells, and in vivo labeling was performed in the presence and absence of 100 ng/ml EGF (Fig.  4A). Although phosphorylation of the wild type PPAR␥1 was enhanced by EGF treatment as before, phosphorylation of the mutant was unaffected. Similar amounts of both mutant and wild type protein were expressed in the transfected cells, as shown by Western blot analysis (Fig. 4B). Since all PPAR␥1 phosphorylation was abolished by this mutation, this result demonstrates that the MAP kinase site at Ser 82 is the only phosphorylation site on PPAR␥1.
To verify that the negative regulation of PPAR␥1 by growth factors was dependent upon PPAR␥1 phosphorylation, NIH 3T3 cells were co-transfected with either the wild type PPAR␥1 or Ser 82 3 Ala PPAR␥1 mutant and ARE7-TKpGL3. Transfected cells were then treated with BRL49653 in the presence or absence of PDGF. Neither basal nor BRL49653-stimulated activity was affected by the Ser 82 3 Ala mutant. In contrast, the activity of the Ser 82 3 Ala mutant PPAR␥1 was resistant to PDGF-mediated repression (Fig. 5).
Phosphorylation of PPAR␥1 Does Not Alter Its DNA Binding Activity-To determine if phosphorylation affects PPAR␥1 DNA binding, a mobility shift assay was performed on a labeled double-stranded oligonucleotide containing the ARE7 PPRE with both in vitro phosphorylated and unphosphorylated MBP-PPAR␥1. As previously reported, PPAR␥1 (Fig. 6, lane 2) alone did not bind to the ARE7 element (6). However, in the presence of RXR␣, both the phosphorylated and unphosphorylated forms of PPAR␥1 bound equally well to the ARE7 probe (Fig. 6, lanes  3 and 5). In addition, phosphorylation of preformed PPAR␥1⅐RXR␣ heterodimer prior to mobility shift assay did not alter PPAR␥1 DNA binding. DISCUSSION The complexity of gene expression requires the utilization of multiple regulatory mechanisms to control both the quantity and activity of all components of the transcription machinery including upstream enhancer proteins. In this study, we have shown that activation of the MAP kinase signaling pathway by EGF and PDGF induces the phosphorylation of PPAR␥1 on Ser 82 and that this event decreases the ability of PPAR␥1 to activate transcription. Mutation of the phosphorylated residue (Ser 82 ) prevents PPAR␥1 phosphorylation as well as the growth factor-mediated repression of PPAR␥1-dependent transcription. This phosphorylation-mediated transcriptional repression is not due to a reduced capacity of the PPAR␥1⅐RXR␣ complex to heterodimerize or recognize its DNA binding site but is due to its ability to become transcriptionally activated by ligand.
The activity of several nuclear hormone receptors is regulated by phosphorylation. Okadaic acid-induced phosphorylation of the human ␤1 thyroid receptor enhances the DNA binding capacity of the protein and increases the ligand-mediated transcription (23). Phosphorylation of retinoic acid receptor ␣ and RXR␣ modulates heterodimerization of the receptors and consequently increases DNA binding activity (24). In addition, the MAP kinase-dependent phosphorylation of Ser 118 on the estrogen receptor causes a 1.8 -2.3-fold increase in transcriptional activation by the AF1 domain (25). Taken together, these data suggest that in general phosphorylation of nuclear receptors enhances their transcriptional activity. In contrast, our data suggest that MAP kinase phosphorylation of PPAR␥1 negatively regulates its function.
EGF, PDGF, and fibroblast growth factor inhibit the conversion of 3T3-L1 preadipocytes to adipocytes (15,17,18). Moreover, primary rat adipogenic precursor cells are also inhibited from becoming adipocytes in the presence of EGF (14), and EGF-treated animals show retardation of the development of adipose tissue (16). Although the precise mechanism of this inhibition is unknown, growth arrest is required for adipogenesis. It is presumed that activation of the intracellular signaling cascades by growth factors must interfere with the activity of the factors involved in differentiation. We suggest that this interference occurs with the activation of MAP kinase. The activation of MAP kinase by EGF or PDGF induces the phosphorylation of PPAR␥1, which negatively regulates its activity, thereby preventing the progression of adipocyte differentiation.
The one piece still missing in this puzzle is how insulin promotes adipocyte differentiation. Insulin, like other growth factors, induces MAP kinase activity in 3T3-L1 adipocytes. In fact, two recent publications suggest that insulin stimulation does induce the PPAR␥1, PPAR␥2, and PPAR␣ phosphorylation (26,27). However, in contrast to our data, both groups present data suggesting that the insulin induced phosphorylation enhances the transcriptional activity of the PPARs. The use of different growth factors and different cell lines may explain this discrepancy. Yet, Zhang et al. (27) reported that mutation of the phosphorylated serine does not prevent the activation of PPAR␥ by insulin. In addition, expression of dominant negative MEK blocks the activity of PPAR␦ that is not phosphorylated by MAP kinase. This suggests that the activation of transcription by insulin in their system occurs through a mechanism independent of the MAP kinase-induced phosphorylation of PPAR␥. Tontonoz et al. (9) have shown that deletion of the N-terminal portion of PPAR␥2, which lacks Ser 82 , enhances the ability of PPAR␥ to induce adipocyte differentiation. Moreover, recently Hu et al. (28) demonstrated that the ectopic expression of a mutant PPAR␥2 (a serine to alanine mutation at position 112 in PPAR␥2, which is equivalent to Ser 82 of PPAR␥1) enhanced sensitivity to ligand-induced adipogenesis. These results strongly support the conclusion of the present paper.
Additional studies on adipocyte function show that although insulin activates MAP kinase in 3T3-L1 adipocytes, insulin-dependent metabolic responses such as glucose uptake, glycogen synthesis, and lipogenesis are unaffected by the inhibition of MAP kinase with the MEK inhibitor PD98059 (29,30). In addition, the MEK inhibitor does not prevent or delay 3T3-L1 adipocyte differentiation (data not shown). Since many of the effects of insulin in adipocytes do not utilize the MAP kinase signaling cascade, we suggest that other signaling events induced by insulin during adipogenesis more strongly regulate PPAR␥ activity than direct phosphorylation by MAP kinase.
The molecular mechanism of inhibition of PPAR␥ via phosphorylation is yet to be determined. Data presented here show that under equilibrium conditions DNA binding of recombinant PPAR␥1⅐RXR␣ complexes is unaffected by phosphorylation, implying that heterodimerization of the complex is also unaffected. This suggests that transcriptional activation by PPAR␥1 is regulated by phosphorylation. Transcriptional activation by nuclear receptors is modulated upon the association of the receptors with co-activators (31, 32) and co-repressors (21,(33)(34). Because of allosteric changes in the receptor, ligand-bound receptor has a greater affinity for the co-activator than the co-repressor and thus enhances transcription (21,33). Since pretreatment with BRL49653 decreased receptor phosphorylation in cell culture, we speculate that phosphorylation, possibly by hindering ligand binding or preventing changes in receptor conformation, plays a role in the selectivity and/or affinity of PPAR␥ for the cofactors. FIG. 5. Mutant Ser 82 3 Ala PPAR␥1 is resistant to growth factor-mediated transcription repression. NIH 3T3 cells were cotransfected with reporter (ARE7-TKpGL3) and either PPAR␥1 or Ser 82 3 Ala PPAR␥1 expression plasmids as described in Fig. 1. Serumstarved cells were stimulated with appropriate treatments as indicated. Open bar, Me 2 SO; hatched bar, BRL49653; solid bar, BRL49653 plus PDGF. WT, wild type.