Identification of a Novel Prostaglandin Reductase Reveals the Involvement of Prostaglandin E2 Catabolism in Regulation of Peroxisome Proliferator-activated Receptor γ Activation*

This report identifies a novel gene encoding 15-oxoprostaglandin-Δ13-reductase (PGR-2), which catalyzes the reaction converting 15-keto-PGE2 to 13,14-dihydro-15-keto-PGE2. The expression of PGR-2 is up-regulated in the late phase of 3T3-L1 adipocyte differentiation and predominantly distributed in adipose tissue. Overexpression of PGR-2 in cells decreases peroxisome proliferator-activated receptor γ (PPARγ)-dependent transcription and prohibits 3T3-L1 adipocyte differentiation without affecting expression of PPARγ. Interestingly, we found that 15-keto-PGE2 can act as a ligand of PPARγ to increase co-activator recruitment, thus activating PPARγ-mediated transcription and enhancing adipogenesis of 3T3-L1 cells. Overexpression of 15-hydroxyprostaglandin dehydrogenase, which catalyzes the oxidation reaction of PGE2 to form 15-keto-PGE2, significantly increased PPARγ-mediated transcription in a PGE2-dependent manner. Reciprocally, overexpression of wild-type PGR-2, but not the catalytically defective mutant, abolished the effect of 15-keto-PGE2 on PPARγ activation. These results demonstrate a novel link between catabolism of PGE2 and regulation of ligand-induced PPARγ activation.

tion factor in the nuclear receptor family, binding to the promoter of target genes by forming heterodimer with retinoid X receptor (5). Upon ligand binding, PPAR␥ releases bound corepressors and recruits co-activator for transcriptional activation (6,7). High-affinity synthetic agonists of PPAR␥, thiazolidinediones, have been widely used as antidiabetic drugs because of their effects in the regulation of lipid metabolism and their anti-inflammatory effects in adipose tissue (8,9). Several naturally occurring ligands, many associated with the promotion or resolution of inflammation (10 -13), including 15-deoxy-⌬ 12,14 prostaglandin J 2 (15d-PGJ 2 ) (14,15), components of oxidized low density lipoprotein such as 9-hydroxyoctadecadienoic acid (HODE), 13-HODE, and 15-hydroxyeicosatetraenoic acid (10,16), lysophosphatidic acid (17), and nitrolinoleic acid (11), have the capacity to activate PPAR␥. The production of prostaglandin E 2 (PGE 2 ) is elevated in many syndromes of inflammation (13,18). However, little is known about whether the catabolism of PGE 2 is associated with modulation of PPAR␥ activity (19). PGE 2 , a short-lived mediator, is inactivated via an oxidation reaction catalyzed by NAD ϩ -dependent 15-hydroxyprostaglandin dehydrogenase (PGDH), which generates 15-keto-PGE 2 , which is, in turn, further catabolized by a reaction catalyzed by NADPH/NADH-dependent 15-oxoprostaglandin-⌬ 13 -reductase (PGR) (20). It has been shown that adipose tissue possesses high activity of both PGDH and PGR, indicating that PGE 2 catabolism is highly active in adipocytes (21). 3T3-L1 preadipocyte cell line has been used as a model for characterizing the events responsible for adipocyte differentiation (22), and PGE 2 is the most abundant prostaglandin produced in 3T3-L1 fibroblasts through the release of arachidonic acid from endogenous phospholipids stores or upon addition of exogenous arachidonic acid (23). Although the levels of PGE 2 decreased upon stimulation of 3T3-L1 adipocyte differentiation, PGE 2 is a major prostaglandin produced in adipocytes (23,24).
In this report, we used a differential display to identify a novel gene encoding prostaglandin reductase, designated as PGR-2. It is highly expressed in the late phase of 3T3-L1 adipocyte differentiation and is also abundant in adipose tissues. PGR-2 is capable of catabolizing 15-keto-PGE 2 , and its overexpression represses the transcriptional activity of PPAR␥. Following these agene). A PCR reaction was carried out for the construction of pFLAG-PGR-2 using pTE-PGR-2 as a template and two primers (forward primer, 5Ј-AAC TGA AGC TTC AAG TGA TGA  TCA TA-3Ј, where the start codon is underlined; and reverse  primer, 5Ј-AGC TCT CCC ATA TGG TCG ACC T-3Ј) to generate a HindIII-SalI DNA fragment of PGR-2. This DNA fragment was cloned into the HindIII-SalI sites of pCMV-Tag2B, yielding pFLAG-PGR-2. The HindIII-XhoI fragment of the pFLAG-PGR-2 was inserted into the SmaI-XhoI sites of pGEX-4T-3 vector (Amersham Biosciences) to yield the pGEX-PGR-2 construct, which was used for the generation of the GST-PGR-2 fusion protein. The HindIII-SalI fragment of the pFLAG-PGR-2 was inserted into the EcoRV-SalI sites of pEGFP-C1 vector (BD Biosciences-Clontech) to create the pEGFP-PGR-2 construct. Site-directed mutagenesis of pGEX-PGR-2, pFLAG-PGR-2, and pEGFP-PGR-2 was performed to generate the PGR-2/Y259F catalytically defective mutant using the QuikChange kit (Stratagene) with the following mutagenic primers (mutated sites are underlined): forward primer, 5Ј-GGT CAG ATT TCT CAG TTC AGT AAC GAT GTG CCC-3Ј; reverse primer, 5Ј-GGG CAC ATC GTT ACT GAA CTG AGA AAT CTG ACC-3Ј.
Cloning of Mouse PGDH cDNA-NAD ϩ -dependent PGDH was cloned from the cDNA of mouse kidney using PCR amplification (sense primer, 5Ј-AGT CGG ATC CAT GCA CGT GAA CG-3Ј, where the start codon is underlined; antisense primer, 5Ј-CAG TCT CGA GTT ATG GAG CTT TTA C-3Ј) to generate a BamHI-XhoI DNA fragment of PGDH. This DNA fragment was cloned into the BamHI-XhoI sites of pCMV-Tag2B to yield the pFLAG-PGDH construct.
Expression and Purification of Recombinant PGR-2-An Escherichia coli strain XL1-blue was transformed with the plasmid pGEX-PGR-2 to generate the recombinant protein of PGR-2 for the enzyme assay, and the recombinant protein was induced with 0.1 mM isopropyl-1-thio-␤-D-galactoside at 25°C overnight. GST-PGR-2 recombinant protein was purified with a glutathione-Sepharose column according to the manufacturer's instructions (Amersham Biosciences).
Enzymatic Assay for PGR-2-PGR-2 activity assay was carried out in a mixture containing 0.1 M Tris-HCl (pH 7.4), 0.5 mM NADPH, and the substrate 15-keto-PGE 1 , 15-keto-PGE 2 , 15-keto-PGF 1␣ , or 15-keto-PGF 2␣ in a total volume of 100 l. The reaction was started by adding 5 g of purified GST-PGR-2 protein and incubated at 37°C for 30 min. NADPH remaining after the reaction was oxidized by adding 200 l of color reagent (790 M indonitrotetrazolium chloride, 60 M phenazene methosulfate, and 1% Tween 20) at 37°C in the dark for 10 min followed by adding 700 l of phthalate buffer (pH 3.0) to stop the reaction. The colorimetric reaction is based on the fact that indonitrotetrazolium can react with NADPH in the presence of phenazene methosulfate to produce formazans (25). The absorbance of formazans was measured at 490 nm with a spectrophotometer. One unit of the enzyme was defined as the amount of enzyme catalyzing the production of 1 mol NADP ϩ /min. The apparent K m and V max values were based on Michaelis-Menten kinetics and calculated by Eadie-Hofstee regression using GraphPad Prism software (version 4.0). All values presented are the means of three or more measurements.
Prostaglandin Extraction-Samples after in vitro PGR-2 enzyme reaction were acidified to pH 4.0 by 0.5 M citric acid and extracted on C 18 solid-phase extraction cartridges (Cayman Chemical). After the cartridges were washed with water and hexane, prostaglandins were eluted with ethyl acetate containing 1% methanol. The 293T cells expressing PGDH and PGR-2 were incubated in 10 M PGE 2 and 10 M [ 2 H 4 ]PGE 2 for 18 h to extract intracellular PGE 2 metabolites. Cells were washed twice with phosphate-buffered saline, harvested in homogenization buffer (0.1 M sodium phosphate, pH 7.4, containing 1 mM EDTA and 10 M indomethacin), and then freeze-thawed in liquid nitrogen. After pH adjustment to pH 4.0 and protein precipitation, the supernatants were loaded onto the C 18 solid-phase extraction cartridges for prostaglandin elution.
Prostaglandin Analysis-The in vitro PGR-2 reaction products were analyzed by nano-ESI-MS and MS/MS. The analytes were directed to the homemade nanosprayer applied with Ϫ3.5 kV on the QSTAR-XL hybrid quadrupole time-of-flight mass spectrometer (Applied Biosystems/MDS Sciex). All of the data were acquired and processed using AnalystQS 1.1 with Bioanalyst 1.1 extension. The instrument was calibrated using the fragment ions that resulted from the collision-induced dissociation of Glu-fibrinopeptide B. Full scan mass spectra (MS) were recorded in the negative ion mode in the range of m/z 100 -500. Product ion mass spectra (MS/MS) were obtained for m/z 351 (13,14-dihydro-15-keto-PGE 2 ) and m/z 349 (15-keto-PGE 2 ) for further identification of the analytes.
Samples were dried under nitrogen, dissolved in 200 l of 20% acetonitrile, and filtered before analysis for characterization of intracellular PGE 2 metabolites. The instrument used was a Thermo-Finnigan Quantum Ultra AM tandem quadrupole interfaced to a Shimadzu Prominence HPLC system. A 150 mm ϫ 2 mm ϫ 3 m Luna C 18 (2) 100A HPLC column (Phenomenex) was maintained at 40°C. The mobile phase was generated from HPLC-grade water (A) and 5% methanol/95% acetonitrile (B), each containing 0.005% acetic acid adjusted to pH 5.7 with ammonium hydroxide. The flow rate was 200 l/min using a gradient starting at 20% B and ramping to 60% B in 20 min. Transfection and Reporter Assays-One day before transfection, 3T3-L1 fibroblasts were plated at 70 -80% confluence, and 293T cells were plated at 30% confluence. After overnight growth at 37°C, cells were incubated with transfection mixture containing plasmid DNA and Lipofectamine at a 1:6 ratio (Invitrogen). After 24 h of transfection, cells were harvested for determination of luciferase and ␤-galactosidase activity. Each experiment was repeated three or more times.
Ligand Binding Assays-Ligand binding assays were performed with GST-mPPAR␥2 fusion protein, which was expressed in E. coli and purified by glutathione-Sepharose affinity chromatography. The protein-bound Sepharose beads were used directly in a binding reaction with 100 nM [ 3 H]BRL49653 in a buffer containing 10 mM Tris-HCl (pH 7.4), 50 mM KCl, and 10 mM dithiothreitol. Competitors or solvent (dimethyl sulfoxide) were added as indicated. Following incubation for 2 h at 4°C, the PPAR␥-Sepharose beads were washed with 10 volumes of ice-cold binding buffer and pulled by centrifugation. The amount of [ 3 H]BRL49653 bound to the beads was measured with a scintillation counter. Each experiment was repeated three or more times. Amino acid sequence alignment of PGR-2 and LTB4DH/PGR. Human and mouse sequences for PGR-2 and LTB4DH/PGR were aligned with ClustalW and Prettybox. Human and mouse sequences for PGR-2 and LTB4DH/PGR proteins share 86 and 80% identity, respectively. The orthologue of LTB4DH/PGR in pig was cloned separately from kidney and lung (GenBank TM accession numbers NM_214385 and U87622, respectively) (26,27). The Tyr residue (denoted by an asterisk), which is the catalytic site of PGR, is conserved in both PGR-2 and LTB4DH/PGR (28). A proline-rich region with a possible Src homology 3 domain recognition site is indicated by a bold line.

Coactivator Recruitment by in Vitro GST Pulldown Assay-GST
and GST-SRC1 were expressed in E. coli. and purified on the glutathione-Sepharose beads. The purified His-tagged hPPAR␥-LBD was preincubated with ligands for 30 min on ice and then mixed with immobilized GST or GST-SRC1 in the pulldown buffer (1ϫ phosphatebuffered saline, 10% glycerol, 0.5% Nonidet P-40). The reactions were incubated overnight at 4°C after which beads were washed three times in pulldown buffer and boiled in 2ϫ sample buffer. Proteins bound to beads were separated by 11% SDS-PAGE. Blots were then developed with antibodies against PPAR␥ or GST. Post-confluent 3T3-L1 preadipocytes were induced (day 0) to differentiate using the standard differentiation mixture as described. During differentiation, total RNA was isolated from 3T3-L1 cells each day from day 0 to day 10. mRNA expression of PGR-2 was analyzed by reverse transcription-PCR with 36B4 expression as an internal control. B, expression of PGR-2 and PPAR␥ is induced during adipocyte differentiation in different kinetics. Total cellular lysates prepared from post-confluent 3T3-L1 preadipocytes and at various times during differentiation into adipocytes were subjected to Western blot analyses with anti-PGR-2, anti-actin, and anti-PPAR␥ antibodies. C, tissue distribution of mouse PGR-2. Total RNAs from various organs of the mice were extracted. Expression of mRNA of the PGR-2 gene was studied with reverse transcription-PCR using 36B4 expression as an internal control.

Identification of PGR-2 by Differential Display during 3T3-L1
Adipocyte Differentiation-The genes differentially expressed in 3T3-L1 cells at day 10 after the induction of adipocyte differentiation were screened using differential display. Among those genes up-regulated, one gene shares 100% identity to NM_029880 as well as AK021033 or AK020666 in the Gen-Bank TM . Further blast search of the data base revealed that gene corresponding to NM_029880 encodes a protein sharing a 40% amino acid sequence identity with two paralogous sequences, human NM_012212 and mouse NM_025968, both of which encode the bifunctional leukotriene B 4 -12-hydroxydehydrogenase/15-oxoprostaglandin-⌬ 13 -reductase (LTB4DH/PGR or PGR-1) (26 -28) (Fig. 1). Based on this similarity, this new gene was designated PGR-2.
Expression of PGR-2 was increased during the differentiation of 3T3-L1 cells at both the mRNA ( Fig. 2A) and protein levels, with a time lag of ϳ1 day following the expression of PPAR␥ (Fig. 2B). Among the mouse tissues examined, adipose tissue showed the highest expression of PGR-2 mRNA (Fig.  2C). This expression pattern was distinct from that of PGR-1, which is most highly expressed in liver and kidney (26).
We also analyzed the in vivo catabolites of PGE 2 through the action of PGDH and PGR-2 in intact cells. To this end, 293T cells were transfected with a control plasmid (Tag2B) or expression vectors of PGDH and PGR-2 and treated with PGE 2 (Fig. 4A). Deuterium-labeled PGE 2 was included in the medium for in vivo uptake to assure the identity of lipid metabolites for mass spectrometric analysis. Intracellular lipid extracts were prepared from these cells for mass spectrometric analysis. We found that both 15-keto-PGE 2 and 13,14-dihydro-15-keto-PGE 2 could be detected in 293T cells expressing PGDH and PGR-2 (Fig. 4B). However, 13,14-dihydro-15-keto-PGE 2 was undetectable in cells transfected with the control vector (data not shown).
The Role of PGR-2 Expression in Adipogenesis-Because the expressed level of PGR-2 was not increased until 3T3-L1 cells had attained late stage differentiation to adipocytes, we then determined whether constitutive expression of PGR-2 would influence adipocyte differentiation. 3T3-L1 cells were transiently transfected with a wild-type expression vector and a catalytically defective GFP-PGR-2, and differentiation was then induced. Based on structural prediction, the conserved Tyr-259 residue in PGR-2 ( Fig. 1) might function as the Tyr-245 of PGR-1 that participates in the hydrogen bond network around the 2Ј-hydroxyl group of the nicotine amide ribose, which interacts with two water molecules in stabilizing an enolate intermediate for the catalysis of 15-keto-PGE 2 reduction (28). The Y259F mutant of PGR-2 was then generated and showed a significant decrease in catalytic efficiency in the reduction reaction of 15-keto-PGE 2 (Fig. 5A). We found that ectopic expression of the wild type but not Y259F of GFP-PGR-2 dramatically inhibited adipocyte differentiation, as evaluated by the accumulation of lipid droplets with Oil-Red O staining (Fig. 5B), and decreased intracellular TG synthesis by 53% as compared with control cells (Fig. 5C). Furthermore, ectopic expression of wild-type GFP-PGR-2, but not the Y259F mutant, significantly reduced the induction of the PPAR␥ target gene, aP2, whereas the expression level of PPAR␥ remained unaltered (Fig. 5D). These results clearly demonstrate that constitutive expression of functional PGR-2 impairs PPAR␥-mediated adipocyte differentiation without affecting the protein expression level of PPAR␥.
15-Keto-PGE 2 Is a PPAR␥ Ligand-The effect of overexpression of PGR-2 on the suppression of adipocyte differentiation (Fig. 5, B-D) and the induction pattern of PGR-2 expression (Fig. 2, A and B) indicates that catabolism of

. Enforced expression of PGR-2 in the early phase of adipogenesis inhibits the differentiation of 3T3-L1 cells.
A, Michaelis-Menten plot of wild-type PGR-2 (Wt) and PGR-2/Y259F mutant (mt) enzyme reaction with various concentration of 15-keto-PGE 2 (15k-PGE 2 ). Post-confluent 3T3-L1 preadipocytes were transfected with expression vector of GFP, GFP-PGR-2 (Wt), and GFP-PGR-2/Y259F (mt) as indicated. Following the transfection, cells were induced to differentiate with the standard differentiation mixture as described under "Experimental Procedures." B, accumulation of lipid droplets was shown by staining with Oil-Red O at 9 days after induction. C, quantitative analyses of total cellular TG content were assayed 9 days after the induction. ***, p Ͻ 0.001. D, after induction, total cellular lysates prepared from day 1 (24 h), day 2 (48 h), and day 3 (72 h) during differentiation were subjected to Western blot analyses with anti-aP2, anti-PPAR␥, and anti-actin antibodies. Cell lysates of preadipocyte (Day 0) and adipocytes (Day 6) were loaded in parallel as controls.
15-keto-PGE 2 by PGR-2 needs to be suppressed during the early phase of adipocyte differentiation. Because PGE 2 is a major prostaglandin produced in preadipocytes (23,24) and there is a temporal difference between the expression of PGR-2 and PP〈R␥ during adipogenesis of 3T3-L1 cells (Fig.  2B), we raised the question of whether 15-keto-PGE 2, rather than PGE 2 , has a functional role in modulating PPAR␥ activity during adipocyte differentiation. We transiently transfected the PPRE-based reporter gene (PPRE-TK-Luc) into 3T3-L1 that had been induced for differentiation for 2 days (PPAR␥ expression was induced and the level of PGR-2 was still low) to determine the effect of 15-keto-PGE 2 on PPREdependent transcription in the early differentiation stage of 3T3-L1 cells. The addition of 15-keto-PGE 2 to the culture medium of transfected cells was capable of activating transcription of the PPRE-based reporter gene as efficiently as BRL49653, a synthetic ligand of PPAR␥ (Fig. 6A). In parallel experiments, the control reporter gene (TK-Luc) did not respond to 15-keto-PGE 2 or BRL49653. The results suggest that 15-keto-PGE 2 can activate PPRE-dependent transcription.
Because ligand-induced transactivation of PPAR␥ is achieved by the recruitment of coactivators such as SRC1 and TIF2 (30), we performed an in vitro coactivator-dependent receptor ligand assay by incubating the purified recombinant GST-SRC1 with PPAR␥-LBD proteins. Similar to BRL49653 as a positive control, inclusion of 15-keto-PGE 2 in the assay enabled PPAR␥-LBD to be pulled down by GST-SRC1, indicating that 15-keto-PGE 2 is capable of acting as a ligand in inducing the interaction between SRC1 and PPAR␥-LBD (Fig. 6D). The same effect of 15-keto-PGE 2 could be seen when the coactivator TIF2 was used for the interaction assay (data not shown). No interaction was detected when GST protein was used for the binding assay. Taken together, these results indicate that 15-keto-PGE 2 is a PPAR␥ ligand that activates its transcriptional function by promoting co-activator recruitment.
Catabolic Conversion of PGE 2 by Expression of Prostaglandin Dehyrogenase Activates PPAR␥-To substantiate the potential role of 15-keto-PGE 2 generation in PPAR␥ activation, we next tested whether overexpression of PGDH could convert PGE 2 to an activator for PPAR␥-mediated transactivation. 293T cells were transfected both with the GAL4 luciferase reporter plasmid (UAS G ) and a vector expressing the ligand-binding domain (LBD) of the mouse PPAR␥ fused to yeast GAL4 DNA-binding domain with or without the PGDH expression vector (Fig. 7A). Overexpression of PGDH had no effect on the transactivation activities of GAL4-PPAR␥. Treatment of cells with 10 M PGE 2 significantly increased GAL4-PPAR␥-mediated transcriptional activity in cells expressing PGDH, but not in the cells transfected with the control vector (Fig. 7B). Consistent with the catabolic function of PGDH for PGE 2 , treatment with 15-keto-PGE 2 also activated GAL4-PPAR␥ in the cells regardless of PGDH expression (Fig. 7B). Furthermore, activation of PPAR␥ activity in cells expressing PGDH was increased with increasing concentration of PGE 2 (Fig. 7C). These results support our notion that metabolic conversion of PGE 2 to 15-keto-PGE 2 is able to activate PPAR␥.
It has been reported that treatment of 3T3-L1 preadipocytes with dexamethasone and insulin (DI) is unable to induce adipocyte differentiation unless methylisobutylxanthine (MIX) is added to DI to stimulate the generation of endogenous PPAR␥ ligands via the cAMP signaling pathways (31,32). This system allows us to evaluate the effect of supplementation of PPAR␥ ligands on the promotion of adipogenesis in the absence of methylisobutylxanthine. As shown in Fig. 8C, differentiation did not occur when preadipocytes were incubated in the presence of DI or MIX alone. As expected, the addition of MIX into DI medium gave rise to adipocyte differentiation of 3T3-L1 cells, as manifested by the accumulation of lipid droplets with Oil-Red O staining (Fig. 8C) and the quantitative analysis of the intracellular TG content (Fig. 8D). Interestingly, the addition of 15-keto-PGE 2 or BRL49653 into DI medium also resulted in adipocyte differentiation to a degree comparable with the effect of MIX in DI medium (Fig. 8, C and D). We further examined the ability of 15-keto-PGE 2 to induce expression of PPAR␥ target genes during adipogenesis. Similar to MIX treatment, both 15-keto-PGE 2 and BRL49653 treatment strongly induced the expression of aP2, an adipocyte-specific marker, after 6 days of adipocyte differentiation (Fig. 8E). Previous studies have shown that PPAR␥ ligands are involved in a positive feedback loop to maintain a relatively high expression of PPAR␥ (31). Consistently, expression of PPAR␥ was increased in cells cultured in a DI medium supplemented with 15-keto-PGE 2 , BRL49653, or MIX. Moreover, the addition of DI with 15-keto-PGE 2 or BRL49653 already induced a prominent aP2 expression in 2 days of treatment, in contrast to a rather weak aP2 induction following the addition of MIX to DI medium (Fig. 8F). The differences in aP2 induction in response to 15-keto-PGE 2 , BRL49653, and MIX treatment during the early phase of induction of adipogenesis suggest that direct ligand-induced activation of PPAR␥ by 15-keto-PGE 2 and BRL49653 gave an earlier induction of aP2 than MIX treatment, which requires more time for endogenous ligand generation. Therefore, these data support the proposition that 15-keto-PGE 2 can act as a PPAR␥ ligand for adipocyte differentiation.

DISCUSSION
The current study presents the first evidence linking the metabolism of 15-keto-PGE 2 to modulation of PPAR␥-dependent transcription and adipogenesis. We have shown that catabolic conversion of PGE 2 to 15-keto-PGE 2 by overexpression of PGDH enables PPAR␥ activation. In addition, we identified PGR-2 as an enzyme that catalyzes the reaction in converting 15-keto-PGE 2 to 13,14-dihydro-15-keto-PGE 2 . The amount of PGR-2 expression is abundant in adipose tissue and is up-regulated during 3T3-L1 adipocyte differentiation. Ectopic overexpression of PGR-2 in 3T3-L1 cells dramatically inhibits adipocyte differentiation. These data illuminate the role of PGE 2 catabolism in PPAR␥-dependent transcription. PGE 2 is the major prostaglandin produced by preadipocytes (23), and its catabolism in adipose tissue is highly active (21). Using in vitro ligand binding assays, we have proved that 15-keto-PGE 2 , a metabolite of PGE 2 , is able to compete for binding of a synthetic ligand, BRL49653, to PPAR␥. Furthermore, the interaction between SRC1 and PPAR␥ is induced upon binding of 15-keto-PGE 2 to PPAR␥, indicating its ligand function in coactivator recruitment. In addition, in vivo transactivation assays have demonstrated that 15-keto-PGE 2 preferentially activates PPAR␥-LBD-mediated transcription as compared with PPAR␣ and PPAR␦. More importantly, the addition of 15-keto-PGE 2 to 3T3-L1 cells can replace the agonist of PPAR␥, thereby inducing differentiation of adipocytes efficiently. Altogether, these results show that 15-keto-PGE 2 can function as a PPAR␥ ligand.
PPAR␥ was suggested as a nuclear prostanoid receptor when 15d-PGJ 2 , the dehydration product of PGD 2 , was first identified as a possible endogenous PPAR␥ ligand (14,15). Indeed, 15d-PGJ 2 can drive PPAR␥-derived adipocyte differentiation but only at concentrations considerably in excess of those formed endogenously (24,32). Other compounds formed naturally also possess the capacity to activate PPAR␥; among them are 9-HODE, 13-HODE, and 15-hydroxyeicosatetraenoic acid (10,16), lysophosphatidic acid (17), and nitrolinoleic acid (11). However, it remains to be determined whether the concentrations necessary (IC 50 values in the in vitro ligand binding assay ranging from 1 to 50 M) are ever attained in vivo. PGE 2 is of particular interest, as it is formed in adipose tissues (33) and is the most abundant prostaglandin formed in preadipocytes. Because a high activity of PGDH has been detected in adipose tissue (21), it seems likely that 15-keto-PGE 2 would be generated in adipocytes.
Here, we have provided diversified lines of evidence consistent with the capacity of 15-keto-PGE 2 to activate PPAR␥-dependent adipogenesis. An outstanding question is whether this occurs at concentrations of the metabolite actually formed in vivo. At present, it remains a technical challenge to quantify 15-keto-PGE 2 in adipocytes. However, it has been reported that lipid metabolites containing the ␣,␤-unsaturated ketone, including 15-keto-PGE 2 , bind to PPAR␥ covalently, activating PPAR␥. Accordingly, it is possible that this covalent binding mechanism enables 15-keto-PGE 2 to activate PPAR␥ in vivo at lower concentrations than would otherwise be the case (34). Further studies will be needed to determine whether PPAR␥ is actually covalently modified by endogenous concentrations of 15-keto-PGE 2 and whether this modified form accumulates during 3T3-L1 adipocyte differentiation. This is one of the few examples of a biological functionality attributable to a prostaglandin metabolite rather than to the parent compound. For example, the role of PGDH in closure of the ductus arteriosus (35) is attributable to its capacity to inactivate the parent moiety. The present observations raise the possibility that the increased PGDH but unaltered PPAR␥ expression in white adipose tissue of ob/ob mice (36, 37) might reflect a role for 15-keto-PGE 2 in adipogenesis in this model, whereas the recognition that PPAR␥ can exert its anti-inflammatory effects by ligand-dependent transrepression of inflammatory response genes (38,39) raises the possibility that macrophage-derived 15-keto-PGE 2 might also modulate adipocyte differentiation and function. Because PGDH is responsible for terminating pro-inflammatory PGE 2 signaling, it will be interesting to know whether the anti-inflammatory capacity of PPAR␥ in adipose tissue is regulated by catabolic conversion of PGE 2 into 15-keto-PGE 2 .
In summary, this study demonstrates for the first time a new link between the catabolism of PGE 2 and the regulation of the PPAR␥ function. Future studies will determine the importance of this system in the regulation of adipogenesis in vivo.