Prostaglandins Promote and Block Adipogenesis through Opposing Effects on Peroxisome Proliferator-activated Receptor γ*

Fat cell differentiation is a critical aspect of obesity and diabetes. Dietary fatty acids are converted to arachidonic acid, which serves as precursor of prostaglandins (PGs). PGJ2 derivatives function as activating ligands for peroxisome proliferator-activated receptor γ (PPARγ), a nuclear hormone receptor that is central to adipogenic determination. We report here that PGF2α blocks adipogenesis through activation of mitogen-activated protein kinase, resulting in inhibitory phosphorylation of PPARγ. Both mitogen-activated protein kinase activation and PPARγ phosphorylation are required for the anti-adipogenic effects of PGF2α. Thus, PG signals generated at a cell surface receptor regulate the program of gene expression required for adipogenesis by modulating the activity of a nuclear hormone receptor that is directly activated by other PG signals. The balance between PGF2α and PGJ2 signaling may thus be central to the development of obesity and diabetes.

Fat cell differentiation is a critical aspect of obesity and diabetes. Dietary fatty acids are converted to arachidonic acid, which serves as precursor of prostaglandins (PGs). PGJ2 derivatives function as activating ligands for peroxisome proliferator-activated receptor ␥ (PPAR␥), a nuclear hormone receptor that is central to adipogenic determination. We report here that PGF2␣ blocks adipogenesis through activation of mitogen-activated protein kinase, resulting in inhibitory phosphorylation of PPAR␥. Both mitogen-activated protein kinase activation and PPAR␥ phosphorylation are required for the anti-adipogenic effects of PGF2␣. Thus, PG signals generated at a cell surface receptor regulate the program of gene expression required for adipogenesis by modulating the activity of a nuclear hormone receptor that is directly activated by other PG signals. The balance between PGF2␣ and PGJ2 signaling may thus be central to the development of obesity and diabetes.
Altered levels of free fatty acids or their metabolites commonly occur in obesity and diabetes (1,2), and fatty acid uptake is increased in these disorders (3). Levels of arachidonic acid (AA), 1 which is derived from dietary essential fatty acids, are high relative to other fatty acids in obesity and diabetic states (4), and high levels of AA may exacerbate diabetes by negatively regulating glucose uptake (5). AA serves as precursor for eicosanoid signaling molecules including leukotrienes, hydroxyeicosatetraenoic acids, and prostaglandins (PGs) (6). Many eicosanoids signal via cell surface G-protein-coupled re-ceptors (GPCRs) (7). Others including 8 S hydroxyeicosatetraenoic acids, leukotriene B4, and a number of PGs including PGJ2 and derivatives such as 15-deoxy-⌬12,14-PGJ2 (15d-PGJ2) bind and activate members of the nuclear hormone receptor superfamily (8) called peroxisome proliferator-activated receptors (PPARs) ␣ and ␥ (9 -12).
Obesity is due to increased size and number of adipocytes. PPAR␥, the nuclear receptor for PGJ2 derivatives, plays a central role in adipogenesis (12)(13)(14). PPAR␥ is the target of thiazolidinediones, an exciting new class of antidiabetic drugs that function as direct ligands for PPAR␥ and have also been shown to be adipogenic (10,(15)(16)(17). An endogenous PPAR␥ ligand is therefore likely to be an important metabolic regulator.
The rate-limiting step in PGJ2 biosynthesis is catalyzed by cyclooxygenase (COX) (18). The actions of different enzymes upon the COX product PGH2 lead to numerous PGs that have different effects on growth, differentiation, and function of many tissues, including PGF2␣ (7). PGF2␣ is known to be synthesized by preadipocytes but does not activate PPAR␥ (9) and by contrast has a potent inhibitory effect upon adipocyte differentiation (19,20). Thus, products of AA metabolism downstream of COX have opposing effects upon adipogenesis. Although the adipogenic effects of PGJ2 derivatives involve direct activation of nuclear PPAR␥, PGF2␣ utilizes a specific GPCR on the cell surface to initiate intracellular signal transduction (21)(22)(23)(24). We and others recently showed that activation of the MAP kinase pathway phosphorylates PPAR␥ and inhibits adipogenesis (25)(26)(27). Although insulin can induce phosphorylation of ectopic PPAR␥ in cells expressing ectopic insulin receptor (27), insulin is adipogenic, and its ability to activate MAP kinase does not play an important role during adipogenesis (28). Indeed, the physiological inducer of PPAR␥ phosphorylation is unknown. Thus we considered whether PGF2␣ might work through this pathway. Here we show that PGF2␣ induces phosphorylation of PPAR␥ via activation of MAP kinase and that this is required for inhibition of 3T3-L1 adipogenesis by PGF2␣.

MATERIALS AND METHODS
Cell Culture and Differentiation-3T3-L1 cells were obtained from American Type Culture Collection (Rockville, MD). Cells were cultured in Dulbecco's modified Eagle's medium containing 10% bovine calf serum (HyClone). Cell were differentiated into mature adipocytes as described previously (29). In some experiments PGs (Cayman Chemical), PD98059 (Calbiochem), or retinoic acid (RA) (Sigma) were added simultaneously with the differentiation medium at day 0 or as indicated and maintained throughout. Compounds were dissolved in either ethanol or Me 2 SO according to manufacturer instructions. For acute treatment with PGs, cells were treated for 30 min and then harvested for protein. Retroviral gene transduction of 3T3-L1 cells with PPAR␥2 or PPAR␥2 (S112A) were performed as described previously (26). Briefly, infected 3T3-L1 cells were selected in G418 and grown to confluence in growth medium (Dulbecco's modified Eagle's medium containing 10% iron-enriched fetal bovine serum). 2 days post-confluency (day 0) vehicle control or PGs were added to medium and maintained throughout. Morphology and RNA were analyzed at day 10.
Northern Analysis-Isolation of total RNA and Northern blots were performed as described previously (29). The cDNA probes for aP2, adipsin, and actin were labeled with 32 P using random hexamers.
Western Analysis-Whole cell extracts were prepared from 3T3-L1 cells using RIPA lysis buffer supplemented with pepstatin (5 g/ml), leupeptin (5 g/ml), and phenylmethylsulfonyl fluoride (2 mM). Extracts were incubated at 4°C for 30 min and then centrifuged 30 min at 4°C. * This work was supported by National Institutes of Health Grant DK49780. 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.

RESULTS
PGF2␣ Activates MAP Kinase-We first determined whether PGF2␣ activates MAP kinase. Fig. 1A shows that indeed, treatment of 3T3-L1 cells with PGF2␣ leads to a dose-dependent increase in activated MAP kinase. Similar MAP kinase activation was observed with fluprostenol, a specific ligand for the PGF2␣ receptor (FP receptor) (23), but not with 15d-PGJ2. Note that PD98059, the specific inhibitor of MAP kinase kinase (30), also prevented the activation of MAP kinase by PGF2␣ and fluprostenol.
MAP Kinase Activation Is Required for Inhibition of Adipogenesis by PGF2␣-We next tested whether MAP kinase activation was required for inhibition of adipogenesis by PGF2␣. Fig. 1B shows the morphology of 3T3-L1 cells 7 days after exposure to differentiation medium. As expected, PGF2␣, fluprostenol, and RA blocked acquisition of the adipocyte phenotype, whereas 15d-PGJ2 did not. The MAP kinase inhibitor PD98059 had little effect on its own. However, it completely prevented inhibition of adipogenesis by PGF2␣ and by fluprostenol, indicating that MAP kinase activation was required for the anti-adipogenic effects of these compounds. This effect of PD98059 was specific for inhibition by PGF2␣ and fluprostenol, because PD98059 had no effect on the ability of RA to inhibit adipogenesis, consistent with evidence that RA acts by antagonizing the effects of C/EBP (29), a mechanism that would be predicted to be independent of MAP kinase. We also assessed the effects of MAP kinase inhibition on the induction of fatty acid binding protein aP2 and adipsin, two adipocyte-specific genes that serve as markers of the differentiation process (31)(32)(33). Fig. 1C shows that induction of aP2 and adipsin was blocked by PGF2␣ and by RA. Consistent with the cellular morphology, MAP kinase inhibition prevented inhibition of the adipocyte-specific genes by PGF2␣ but not by RA. Indeed, MAP kinase inhibition caused a reproducible increase in aP2 expression, consistent with a recent report that MAP kinase activation is anti-adipogenic (28).
PGF2␣ Induces Phosphorylation of PPAR␥-Given the ability of PGF2␣ to activate MAP kinase and the inhibitory effects of MAP kinase phosphorylation on the activity of PPAR␥, we next tested the effects of PGF2␣ on PPAR␥ phosphorylation. Phosphorylation of PPAR␥2 on serine 112 can be detected by differences in electrophoretic mobility of the phosphorylated and nonphosphorylated forms. Fig. 2A shows that 3T3-L1 cells on day 4 of a standard differentiation protocol express abundant PPAR␥1 and PPAR␥2 protein, with about equal amounts of the hypophosphorylated and phosphorylated forms of each. Treatment with PGF2␣ increased the phosphorylation state of  Fig. 1B. Cell lysates were separated by SDS-PAGE and analyzed by Western mobility shift assay using antibody to PPAR␥. B, phosphorylation of PPAR␥ during adipocyte differentiation, and the effects of treatment with PGF2␣. PGF2␣ (100 nM) was added to differentiating 3T3-L1 cells at day 3 and maintained throughout. Protein extracts were harvested at day indicated and analyzed by Western blot. Arrows indicates phosphorylated PPAR␥ (P). C, addition of PGF2␣ on day 3 inhibits induction of adipocyte-marker aP2. Northern analysis of aP2 mRNA at day 7 after exposure to indicated treatments. Ethidium bromide staining of 28 S is shown to confirm loading of equal RNA amounts. the PPAR␥ isoforms in a concentration-dependent manner, and the phosphorylation was blocked by the MAP kinase inhibitor. Similar results were obtained with fluprostenol. By contrast, 15d-PGJ2 had no effect upon the phosphorylation state of PPAR␥1 or PPAR␥2, and PD98059 had no independent effect. Prostacyclin (PGI2) also had no effect on PPAR␥ phosphorylation (data not shown). Thus PGF2␣ activation of MAP kinase resulted in phosphorylation of endogenous PPAR␥ in 3T3-L1 cells. PGF2␣-induced phosphorylation of PPAR␥ inhibited its transcriptional activity as previously shown for phosphorylation of PPAR␥ by other agents (data not shown).
We next tested whether PGF2␣-induced phosphorylation of PPAR␥ correlated with inhibition of adipogenesis. Fig. 2B shows that PPAR␥ protein was not reproducibly detectable on day 0, and the differentiation process involves amplification of PPAR␥ expression by a feed-forward mechanism. The ability of PPAR␥ ligands to induce adipogenesis (14,34) strongly suggests that functional PPAR␥ is present in the preadipocyte, and thus hyperphosphorylation by PGF2␣ would inhibit their function. However, PGF2␣ inhibition of adipogenesis also blocked the feed-forward induction of PPAR␥, preventing analysis of PPAR␥ phosphorylation due to technical limitations (data not shown). Nevertheless, we were able to test the correlation between inhibition of adipocyte differentiation and PPAR␥ phosphorylation by addition of PGF2␣ at times when PPAR␥ protein was expressed. Fig. 2B shows that PPAR␥ protein was induced between days 1 and 2 of adipogenesis and reached maximum levels by days 3 and 4. Both PPAR␥ and PPAR␥2 were present in nonphosphorylated and phosphorylated states, with similar ratios of nonphosphorylated to phosphorylated PPAR␥ throughout differentiation. Addition of PGF2␣ at day 3 caused a marked reduction in adipocyte differentiation, as indicated by a dramatic (70%) reduction in aP2 gene expression on day 7 (Fig. 2C) as well as by a similar reduction in adipogenesis as assessed by cell morphology (data not shown). PPAR␥ induction is itself a marker of adipocyte differentiation, and indeed PPAR␥ levels were lower in the PGF2␣-treated cells (Fig. 2B). Thus the ability of PGF2␣ to inhibit adipogenesis correlated with the phosphorylation of endogenous PPAR␥.
PGF2␣ Selectively Inhibits Adipogenesis Due to Phosphorylated Form of PPAR␥-Thus far we have shown that the ability of PGF2␣ to activate MAP kinase is essential for its ability to inhibit adipogenesis, that MAP kinase activation by PGF2␣ is sufficient to phosphorylate PPAR␥ in 3T3-L1 cells, and that this phosphorylation correlates with inhibition of differentiation. We next tested whether the ability to phosphorylate PPAR␥ was necessary for PGF2␣ to inhibit adipogenesis. We previously showed that ectopic expression of PPAR␥2 results in adipocytic differentiation of 3T3-L1 cells (35). We hypothesized that PGF2␣-induced phosphorylation of the ectopic PPAR␥2 would inhibit its ability to induce adipogenesis, whereas adipocytic differentiation induced by PPAR␥2 (S112A) would be unaffected by activation of MAP kinase by PGF2␣. Indeed, Fig.  3A shows that PGF2␣ completely prevented adipogenesis due to PPAR␥2 expression but had little or no effect upon adipogenesis caused by PPAR␥2 (S112A). In contrast, 15d-PGJ2 potentiated the adipogenicity of both wild type and mutant PPAR␥. Fig. 3B shows that the wild type PPAR␥ was nearly completely phosphorylated by PGF2␣, whereas the PPAR␥2 (S112A) was not phosphorylated at all. Fig. 3C shows that this difference in susceptibility to the effects of PGF2␣ was reflected in the expression of the adipocyte marker adipsin. PGF2␣ completely prevented the ability of PPAR␥2 to induce adipsin but had no effect on adipsin induction by the nonphosphorylatable form of PPAR␥2. Again, 15d-PGJ2 functioned as an activating ligand of both forms of PPAR␥.

DISCUSSION
Our results shed important new light on the role of cell surface and nuclear receptors for eicosanoids. Some eicosanoids, such as leukotriene B4, interact with both cell surface receptors and nuclear PPAR␣, and these signaling pathways are cooperative (9 -12, 36 -38). In contrast, PGF2␣ and PGJ2 derivatives are different products of AA metabolism with opposing biological effects mediated by cell surface and nuclear receptors, respectively. These converge on the PG nuclear receptor PPAR␥ as in the model shown in Fig. 4. 3T3-L1 cells are known to produce PGF2␣, and in accordance with recent suggestions our model proposes that a PPAR␥ ligand related to PGJ2 is endogenously produced during adipogenesis. However, the expression of PGD synthase in adipocytes has not been established, and it is likely that prostaglandins or other PPAR␥ ligands derived from paracrine, endocrine, as well as pharmacological sources outside the fat cell could also influence this process. In any case, in a single cell type, PGs can function both by binding to GPCRs (PGF2␣) or to nuclear hormone receptors (PGJ2 and derivatives). Despite the different types of receptors bound by PGF2␣ and PGJ2 derivatives, both the inhibition of PPAR␥ activity by PGF2␣-driven phosphorylation as well as activation of PPAR␥ by noncovalent PG ligand binding could be physiologically relevant not only in adipocytes but also in other cell types that express PPAR␥, such as lung (39), colon (40,41), and hematopoietic cells (42). Moreover, because other eicosanoids function via cell surface and nuclear receptors, similar opportunities exist for convergent regulation of biological processes by eicosanoids that either bind directly to or modulate the phosphorylation state of nuclear hormone receptors. This experiment was repeated three times with similar results. B, PGF2␣ treatment leads to phosphorylation of ectopic PPAR␥2 but not PPAR␥2 (S112A). Cells were harvested 1 h after treatment with PGF2␣ (100 nM) on day 0. C, PGF2␣ inhibits adipsin induction due to PPAR␥2 but not PPAR␥2 (S112A). Actin and 28 S rRNA are shown as loading controls.
Chronic exposure of adipocytes to AA causes a state of insulin resistance associated with down-regulation of the insulinresponsive glucose transporter (43). Interestingly, AA-derived PGF2␣ levels are increased in diabetic humans (44,45). The link between free fatty acid levels and the pathogenesis of noninsulin-dependent diabetes has been recognized for many years (46). Our results provide a potential molecular mechanism relating fatty acids to both increased and decreased activity of the adipogenic nuclear receptor PPAR␥. Because thiazolidinediones that activate PPAR␥ are potent antidiabetic compounds (47), the ability of PGF2␣ to inhibit the activity of PPAR␥ suggests that the FP receptor may be a novel therapeutic target for diabetes. FIG. 4. Model of opposing actions of PGF2␣ and PGJ2 derivatives in adipocyte differentiation. PGF2␣ interacts with the FP receptor, which activates MAP kinase resulting in phosphorylation of PPAR␥ and inhibition of its activity. PGJ2 derivatives are depicted as being endogenously produced but could also be derived from extracellular paracrine, endocrine, or pharmacological sources. Thiazolidinediones could also serve as PPAR␥ ligands. In any case, these ligands interact directly with PPAR␥ to stimulate its activity. The net effects on PPAR␥ activity result in repression or stimulation of the adipogenic gene program. Similar antagonistic effects on PPAR␥ function could also alter PPAR␥ activity in the mature adipocyte.