Prostaglandin reductase-3 negatively modulates adipogenesis through regulation of PPARγ activity.

Adipocyte differentiation is a multistep program under regulation by several factors. Peroxisome proliferator-activated receptor γ (PPARγ) serves as a master regulator of adipogenesis. However, the endogenous ligand for PPARγ remained elusive until 15-keto-PGE2 was identified recently as an endogenous PPARγ ligand. In this study, we demonstrate that zinc-containing alcohol dehydrogenase 2 (ZADH2; here termed prostaglandin reductase-3, PTGR-3) is a new member of prostaglandin reductase family that converts 15-keto-PGE2 to 13,14-dihydro-15-keto-PGE2. Adipogenesis is accelerated when endogenous PTGR-3 is silenced in 3T3-L1 preadipocytes, whereas forced expression of PTGR-3 significantly decreases adipogenesis. PTGR-3 expression decreased during adipocyte differentiation, accompanied by an increased level of 15-keto-PGE2. 15-keto-PGE2 exerts a potent proadipogenic effect by enhancing PPARγ activity, whereas overexpression of PTGR-3 in 3T3-L1 preadipocytes markedly suppressed the proadipogenic effect of 15-keto-PGE2 by repressing PPARγ activity. Taken together, these findings demonstrate for the first time that PTGR-3 is a novel 15-oxoprostaglandin-Δ13-reductase and plays a critical role in modulation of normal adipocyte differentiation via regulation of PPARγ activity. Thus, modulation of PTGR-3 might provide a novel avenue for treating obesity and related metabolic disorders.

antibiotic selection. The selection was continued for at least 1 month. Individual colonies were isolated to culture cells for further propagation. Detailed protocols for lentivirus production and infection of cells were performed following the procedures of the National RNAi Core Facility of Academia Sinica in Taiwan.

In vitro enzymatic reaction
All reagents were purchased from Sigma-Aldrich (St. Louis, MO). Colorimetric method was used to determine PTGR-3 enzyme activity as described previously ( 2 ). Briefl y, PTGR-3 recombinant protein (OriGene, Rockville, MD) was incubated with 0.5 mM NADPH and 0.6 mM prostaglandins (13,14-dihydro-15keto-PGE 2 , 15-keto-PGE 2 , 15-keto-PGF 2 ␣ , 15-keto-PGF 1 ␣ , or 15-keto-PGE 1 ) in 0.1 M Tris-HCl (pH 7.4) at 37°C for 30 min. After enzymatic reaction, color reagent (790 M indonitrotetrazolium chloride, 60 M phenazene methosulfate, and 1% Tween 20) was added and incubated at 37°C in the dark for 10 min. To stop the reaction, phthalate buffer (pH 3.0) was added. PTGR-3 activity was determined by measuring the absorbance of formazans, an indicator of remaining NAPDH, at 490 nm with a spectrophotometer. One unit of the enzyme was defi ned as the amount of enzyme catalyzing the production of 1 mol NADP + /min. The apparent K m and V max values were determined using the Michaelis-Menten equation and calculated by nonlinear regression.

Western blot
Total protein from tissue or cells was extracted by radioimmunoprecipitation assay buffer with protease inhibitors. The sample was centrifuged at 12,000 rpm for 10 min, and the supernatant The protein structure of PTGR-3 is computationally predicted to contain an oxidoreductase catalytic site located between amino acids 184 and 320 based on sequence homology and shares 23% amino acid homology to PTGR-2. However, the function of PTGR- 3 has not yet been investigated.

Cell culture
All chemical reagents were purchased from Sigma-Aldrich (St. Louis, MO) unless specifi ed otherwise. The 3T3-L1 preadipocytes were cultured in Dulbecco modifi ed Eagle's medium (DMEM) with 10% calf serum (CS) at 37°C in an atmosphere of 5% CO 2 . To induce adipocyte differentiation, confl uent cells were cultured in induction medium (DMEM containing 10% FBS, 1 M dexamethasone, 0.5 mM 3-isobutyl-1-methylxanthine, and 1 g/ml insulin). After 2 days, the cells were maintained in DMEM containing 10% FBS and 1 g/ml insulin. Two days later, cells were cultured in DMEM containing 10% FBS for 8 days, with a medium change every 2 day. After 8 days of culturing, cells on the plates were stained with Oil Red O to measure the degree of adipocyte differentiation. Total protein lysate was extracted to determine the protein levels of adipogenic genes whose expression increases during adipocyte differentiation: adipocyte fatty acidbinding protein (aP2) and adiponectin.

Lentiviral infection
Small hairpin RNA (shRNA) plasmids for the mouse PTGR-3 gene and the control vector (shLuc) were obtained from the National RNAi Core Facility of Academia Sinica in Taiwan. For PTGR-3 knockdown, confl uent 3T3-L1 preadipocytes were infected with lentivirus carrying control shRNA and shRNA targeting PTGR-3, respectively. After lentiviral infection, cells were selected by antibiotic selection using puromycin for 2 days. Two days later, cells were isolated to culture cells for further propagation. For the PTGR-3 overexpression experiment, confl uent 3T3-L1 preadipocytes were infected with lentivirus carrying empty vector (pLVX-IRES-neo, Clontech, Mountain View, CA) and PTGR-3, respectively. After lentiviral infection, cells were maintained in selective medium containing the G418 antibiotics. Cell lines stably expressing empty vector or PTGR-3 were established by PCR with reverse transcription showed increased expression of PPAR ␥ -targeted genes, including Lpl and Cd36 , in PTGR-3-knockdown 3T3-L1 cells after induction (supplementary Fig. II-A). By contrast, adipocyte differentiation was remarkably decreased when we ectopically expressed PTGR-3 protein in 3T3-L1 preadipocytes ( Fig. 2E, F ), which was accompanied by a decrease in intracellular triglyceride content ( Fig. 2G ) and in expression of adipocyte differentiation markers ( Fig. 2H and supplementary Fig.  II-B). Taken together, these results demonstrate PTGR-3 negatively regulates adipocyte differentiation.

Approvals for usage of animals
All animal experiments were performed in accordance with protocols approved by the National Taiwan University Institutional Animal Care and Use Committee (IACUC Approval No. 20100329).

Statistical analysis
Results are expressed as means ± SE. Comparison between two groups were performed using unpaired t -test. P < 0.05 was considered statistically signifi cant. A two-way ANOVA (ANOVA) procedure was used to determine the main effects of PPAR ligands, PTGR-3, and their interaction on intracellular triacylglycerol content and luciferase activity. Multiple comparisons between groups were performed using post-hoc Tukey test (SAS Institute Inc., Cary, NC). A signifi cant difference indicated P р 0.05.

PTGR-3 is a negative regulator of adipocyte differentiation
To explore the physiological function of PTGR-3, we fi rst assayed the tissue expression pattern of PTGR-3. PTGR-3 protein and mRNA are ubiquitously expressed in several tissues, including heart, brown adipose tissue, and white adipose tissue ( We also examined whether PTGR-3 levels were altered in murine models of obesity. We found that PTGR-3 levels were decreased in white adipose tissue of high-fat-diet (HFD) fed mice ( Fig. 1B ) and ob/ob mice ( Fig. 1C and supplementary Fig. I-B). Furthermore, PTGR-3 is specifi cally expressed in stromal vascular fractions (SVF) rather than mature adipocytes ( Fig. 1D ), and PTGR-3 levels remarkably decreased during induced adipogenesis of 3T3-L1 cells ( Fig. 1E ). These results suggest that PTGR-3 may be involved in adipocyte differentiation.
We further tested whether GW9662, an irreversible competitive PPAR ␥ antagonist that covalently binds to a cysteine residue in the PPAR ␥ ligand-binding domain, can erase the enhanced adipocyte differentiation in PTGR-3knockdown 3T3-L1 preadipocytes. After GW9662 treatment, adipocyte differentiation, triglyceride content, adipogenic marker genes, and PPRE-driven luciferase activity were signifi cantly decreased in PTGR-3-knockdown cells compared

PTGR-3 decreases proadipogenic effect of 15-keto-PGE 2 through regulation of PPAR ␥ activity
To assess whether PTGR-3 has a direct effect on 15-keto-PGE 2 -mediated adipogenesis, we treated vector-only and PTGR-3-overexpressing 3T3-L1 cells with or without 15-keto-PGE 2 . Similar to previous results, 15-keto-PGE 2 significantly promoted adipocyte differentiation ( Fig. 5A , two-way ANOVA, P < 0.05). Conversely, overexpression of PTGR-3 suppressed adipocyte differentiation in 3T3-L1 cells ( Fig. 5A , two-way ANOVA, P < 0.05). However, the proadipogenic effect of 15-keto-PGE 2 was remarkably abolished when PTGR-3 protein was overexpressed in 3T3-L1 preadipocytes by signaling transduction. For the past decade, 15-deoxy-⌬ 12.14 -prostaglandin J 2 (15d-PGJ 2 ) has been considered a potent natural ligand for PPAR ␥ ( 1 ). 15d-PGJ 2 is derived from prostaglandin H 2 (PGH 2 ), which is converted by PGD synthase to prostaglandin D 2 and further chemically dehydrated to form 15d-PGJ 2 . Several lines of evidence have demonstrated that inhibition of the COX pathway or dysregulation of genes involved in this series of reactions in 3T3-L1 preadipocytes affects adipocyte differentiation (6)(7)(8)(9). The highest level of 15d-PGJ 2 was detected in 3T3-L1 cells during the maturation phase of adipocytes ( 10 ). In contrast, other groups have demonstrated that no differences in formation of 15d-PGJ 2 are observed during adipocyte differentiation ( 11 ). Thus, it remains uncertain whether 15d-PGJ 2 is truly an endogenous ligand for PPAR ␥ that modulates adipocyte differentiation. Prostaglandin F 2 ␣ (PGF 2 ␣ ) is generated from PGH 2 by PGF 2 ␣ synthase (PGFS), and PGF 2 ␣ is reported to be an inhibitory modulator of adipocyte differentiation ( 12 ). The level of PGF 2 ␣ rapidly increased to a peak 3 h after initiation of differentiation and then declined, indicating that PGF 2 ␣ suppresses in the early phase of adipogenesis ( 13 ). Knockdown of aldo-keto reductase 1B3 (AKR1B3), a PGFS, decreased de novo PGF 2 ␣ biosynthesis in 3T3-L1 cells and then further promoted lipid accumulation ( 13 ). It has been shown that PGF 2 ␣ suppressed adipogensis by enhancing phosphorylation of PPAR ␥ via prostaglandin F receptor-activated MEK⁄ERK cascade ( 14 ). On the other hand, PGE 2 , a precursor of 15-keto-PGE 2 , is the most abundant prostaglandin produced in 3T3-L1 preadipocytes, and the highest level of PGE 2 was detected during the early phase of adipogenesis ( 15,16 ). The activity of 15-hydroxy prostaglandin dehydrogenase, an enzyme catalyzing the conversion of PGE 2 to 15-keto-PGE 2 , was also detected in adipose tissue ( 16 ), implying that PGE 2 metabolites or its catabolizing enzymes may play a potential role in modulating adipocyte differentiation. However, the role of PGE 2 metabolites in lipid metabolism has not been widely studied.
It has been demonstrated that several prostaglandins act as bioactive modulators for regulating adipocyte differentiation through either direct binding on the PPAR ␥ ligand-binding domain or indirect mechanisms mediated from preadipocytes, such as DLK1/PREF1 and Wnt proteins, play an important role in maintaining the undifferentiated state in preadipocytes by extracellular signaling transduction. It has been demonstrated that dysregulation of these genes promotes adipocyte differentiation by infl uencing C/EBP ␣ and PPAR ␥ expression in vivo or in vitro ( 17,18 ). These results suggest that maintenance of preadipocytes in an undifferentiated state requires the activity of extracellular signaling molecules. However, little is known about whether intracellular proteins, such as enzymes involved in proadipogenic molecule metabolism, also modulate maintenance of the preadipose state. Here, we demonstrate that expression of PTGR-3, similar to DLK1/PREF1 and Wnt proteins, is predominantly expressed in preadipocytes and decreases remarkably during adipocyte differentiation. In addition, PTGR-3 protein is abundantly expressed in adipose tissue of lean mice, which contains a relatively high percentage of undifferentiated preadipocytes compared with adipose tissue of obese mice, implying that PTGR-3 may play a role in maintaining the undifferentiated state in preadipocytes. Forced PTGR-3 expression attenuates adipocyte differentiation, whereas knocking down PTGR-3 in 3T3-L1 preadipocytes accelerates adipocyte differentiation. Overall, these observations suggest that PTGR-3 protein may facilitate maintaining the undifferentiated state compared with PTGR-3. Both PTGR-2 and PTGR-3 were expressed abundantly in white adipose tissue (supplementary Fig. I-A) and were downregulated in white adipose tissue of ob/ob mice (supplementary Fig. I-B). Interestingly, expression of PTGR-3 in 3T3-L1 preadipocytes decreased rapidly within 48 h during the course of induced adipocyte differentiation, accompanied by a rise of 15-keto-PGE 2 level and PPAR ␥ expression, implying that PTGR-3 may modulate PPAR ␥ activity by regulation of 15-keto-PGE 2 production in an early stage of adipocyte differentiation. After 2 days of induction, PTGR-2 expression started to increase and endogenous 15-keto-PGE 2 level gradually decreased, indicating PTGR-2 may play a more dominant role in the late stage of adipocyte differentiation than PTGR-3. Cumulatively, these fi ndings demonstrate that these enzymes display different kinetic parameters and different expression patterns during adipogenesis. Whether and how PTGR-2 and PTGR-3 coordinately regulate 15keto-PGE 2 homeostasis and regulate adipogenesis remains to be investigated. Conversion of preadipocytes to adipocytes is a complicated process, which is regulated by several factors, including hormonal and nutrient stimulation. Preadipocytes have to integrate these signals from the environment for initiation of adipocyte differentiation. Several factors secreted  Vector-only and PTGR-3-overexpressing preadipocytes were transiently transfected with reporter vectors (TK-LUC and PPRE-LUC) for 24 h, and then cells were incubated in induction medium with or without 10 M 15-keto-PGE 2 treatment. The activity of fi refl y luciferase was determined and normalized to the activity of renilla luciferase. The bars indicate the means ± SE for three independent replicates (n = 3). Different superscripts indicate signifi cant difference between groups (two-way ANOVA, P р 0.05 with post-hoc Tukey test). shRNA control and PTGR-3-knockdown cells were maintained in induction medium with or without 1 M GW9662 treatment for 2 days. After 8 days of adipogenic stimulation, cells on the plates were stained with Oil Red O (E) and quantifi ed (F) and adipogenic gene (adiponectin and aP2) expression was determined (G). The bars indicate the means ± SE for three independent replicates (n = 3). Different superscripts indicate signifi cant difference between groups (two-way ANOVA, P р 0.05 with post-hoc Tukey test). Expression of adipogenic genes (adiponectin and aP2) in GW9662-treated control and PTGR-3-knockdown cells. The expression of adipogenic markers (adiponectin and aP2) was determined and normalized to ␣ -tubulin. (H) PPRE-driven luciferase activity of shRNA control and PTGR-3-knockdown cells after GW9662 treatment. shRNA control and PTGR-3-knockdown cells were transiently transfected with reporter vectors (TK-LUC and PPRE-LUC) for 24 h and then cells were incubated in induction medium with or without 1 M GW9662 treatment. The activity of fi refl y luciferase was determined and normalized to the activity of renilla luciferase. The bars indicate the means ± SE for three independent replicates (n = 3). Different superscripts indicate signifi cant difference between groups (two-way ANOVA, P р 0.05 with post-hoc Tukey test). (I) Effi ciency of ectopic expression of UAS G ×4-TK-LUC, CMX-GAL4-PPAR ␥ , and Flag-PTGR-3 protein in 293 cells. The expression of PTGR-3 and Gal/PPAR ␥ was determined and normalized to ␣ -tubulin. (J) Transactivation of 15-keto-PGE 2 on PPAR ␥ ligand-binding domain in Flag-PTGR-3-overexpressing 293 cells. 293 cells were transiently transfected with expression vectors (CMX-GAL4, CMX-GAL4-PPAR ␥ , and Flag-PTGR-3) and UAS G ×4-TK-LUC reporter plasmid for 24 h, and then cells were incubated in growth medium with or without 10 M 15-keto-PGE 2 for 2 days. The activity of fi refl y luciferase was determined and normalized to the activity of renilla luciferase. The bars indicate the means ± SE for three independent replicates (n = 3). Different superscripts indicate signifi cant difference between groups (one-way ANOVA, P р 0.05 with post-hoc Tukey test).
in preadipocytes by inhibiting endogenous 15-keto-PGE 2 production and that suppression of PTGR-3 expression is required for normal adipocyte differentiation by regulating PPAR ␥ transcriptional activity.
In conclusion, we provide evidence that PTGR-3 is a novel 15-oxoprostaglandin-⌬ 13 -reductase that affects adipocyte differentiation by regulation of PPAR ␥ activity. PTGR-3 expression is reduced in adipose tissue from murine models of obesity. Modulation of PTGR-3 expression or activity might provide a novel avenue in treating obesity and related metabolic disorders.