Ubiquitin Ligase NEDD4 Regulates PPARγ Stability and Adipocyte Differentiation in 3T3-L1 Cells

Peroxisome proliferator–activated receptor-γ (PPARγ) is a ligand-activated nuclear receptor which controls lipid and glucose metabolism. It is also the master regulator of adipogenesis. In adipocytes, ligand-dependent PPARγ activation is associated with proteasomal degradation; therefore, regulation of PPARγ degradation may modulate its transcriptional activity. Here, we show that neural precursor cell expressed developmentally down-regulated protein 4 (NEDD4), an E3 ubiquitin ligase, interacts with the hinge and ligand binding domains of PPARγ and is a bona fide E3 ligase for PPARγ. NEDD4 increases PPARγ stability through the inhibition of its proteasomal degradation. Knockdown of NEDD4 in 3T3-L1 adipocytes reduces PPARγ protein levels and suppresses adipocyte conversion. PPARγ correlates positively with NEDD4 in obese adipose tissue. Together, these findings support NEDD4 as a novel regulator of adipogenesis by modulating the stability of PPARγ.

Scientific RepoRts | 6:38550 | DOI: 10.1038/srep38550 degradation, while ubiquitin chains linked through other lysine residues (K6, K11, K27, K29, K33 and K63) or mixed chains may be involved in various cellular signaling and pathways not related to protein degradation 22 . To test if NEDD4 promotes the assembly of non-K48 linkages to PPARγ , we employed an HA-tagged K48R mutant ubiquitin with a single lysine to arginine mutation at position 48, to prevent the assembly of K48-ubiquitin chain.  PPARγ levels is on the right. Variable amounts of pRc-CMV-T7-NEDD4 expression vector from 1 to 3 μ g were transfected with the PPARγ into HEK293 cells. NEDD4 plasmid expression saturates at 2 μ g per well of a 6-well plate. (B) 1 μ g of either FLAG-tagged SIAH1 or HA-tagged SIAH2, together with 1 μ g of T7-tagged NEDD4, were transfected into CHO cells stably expressing FLAG-tagged PPARγ 2. The relative PPARγ levels are quantified on the right. (C) NEDD4 overexpression enhances in vivo ubiquitination of PPARγ . T7-tagged NEDD4, FLAG-tagged PPARγ 2, and HA-tagged ubiquitin (Ub) were transiently expressed in HEK293 cells at a ratio of 1.5:1:2 in combinations as indicated above the blots. (D) NEDD4 knockdown reduces in vivo ubiquitination of PPARγ . HEK293 cells seeded on 6-well plates were transfected with 1 μ g of FLAG-tagged PPARγ 2, 2 μ g of HA-tagged Ub, along with 1.5 μ g of plasmid containing NEDD4-targeting shRNA or nontargeting control shRNA per well. Cells were harvested for Western analysis 48 hr after the transfection. (E) NEDD4 overexpression enhances lysine-48 independent ubiquitination of PPARγ . T7-tagged NEDD4, FLAG-tagged PPARγ 2, along with HA-tagged wild-type (WT) Ub or HA-tagged K48R Ub were transfected into HEK293 cells at a ratio of 1.5:1:2 in combinations as indicated above the blots. Western assays were performed at 48 hr after the transfection. (F) In vitro ubiquitination assay between NEDD4 and escalating amounts of recombinant His-GST tagged PPARγ 2. (G) In vitro PPARγ 2 ubiquitination by NEDD4 in the presence of wild-type ubiquitin (WT-Ub), K48-only ubiquitin (K48O-Ub), or K63-only (K63O-Ub). Data represent mean ± SEM of 3 independent experiments; *P < 0.05, **P < 0.01. Full-length blots are presented in Supplementary Fig. S4.
Scientific RepoRts | 6:38550 | DOI: 10.1038/srep38550 The anti-HA immunoblotting of PPARγ 2 immunoprecipitated from HEK293 cell lysates containing co-expressed FLAG-tagged PPARγ 2 and T7-tagged NEDD4 in the presence of either HA-tagged wild-type ubiquitin or K48R mutant revealed robust ubiquitinated forms of PPARγ 2 at comparable levels (Fig. 3E). The basal ubiquitinated forms of PPARγ 2 were slightly less in the presence of the K48R mutant compared to that in the presence of wild-type ubiquitin (lane 5 vs. lane 1 in Fig. 3E), perhaps because of a reduced assembly efficiency of the K48R ubiquitin. To test if NEDD4 directly ubiquitinates PPARγ , we performed PPARγ in vitro ubiquitination assay. Consistent with the in vivo ubiquitination data, we found that PPARγ was efficiently ubiquitinated by NEDD4 in a dose-dependent manner (Fig. 3F). Using wild-type (WT), K48-only (K48O), and K63-only (K63O) ubiquitin, we observed strong ubiquitination signals in each reaction, suggesting that NEDD4 catalyzes both K48 and K63 linkages (Fig. 3G).

NEDD4 increases PPARγ protein stability.
PPARγ is a short-lived protein (t 1/2 = 2 hr) 23 . Consistent with other studies, we found that PPARγ was rapidly degraded upon inhibition of de novo protein synthesis by cycloheximide. Pretreatment of the proteasomal inhibitor MG132, but not the lysosomal inhibitor chloroquine, reversed the degradation of PPARγ ( Supplementary Fig. S1). Next, we measured the half-life of PPARγ by cycloheximide chase assay with overexpressed or downregulated NEDD4. Interestingly, overexpression of NEDD4 increased PPARγ half-life from 2 hr to > 4 hr, while knockdown of NEDD4 destabilized PPARγ , with its half-life decreasing to 75 min (Fig. 4A,B).
The human and rat NEDD4 share 80% homology in the protein sequences. We compared the effect of NEDD4 from the two species on PPARγ protein degradation. Our results indicate that the overexpression of either human or rat NEDD4 protected against PPARγ degradation upon treatment with cycloheximide (Fig. 4C). Next, we tested whether NEDD4 was able to hinder the reduction of PPARγ caused by SIAH1 and SIAH2. MG132 treatment for 6 hr largely rescued SIAH2-mediated reduction, but, to our surprise, only slightly rescued SIAH1-mediated reduction. NEDD4 overexpression largely rescued SIAH2-mediated reduction of PPARγ to the level comparable with MG132 treatment, but had no effect on SIAH1-mediated reduction (Fig. 4D). NEDD4 overexpression combined with MG132 treatment did not further restore the decreased steady-state level of PPARγ mediated by SIAH2 or after addition of cycloheximide (Fig. 4D). NEDD4 overexpression combined with MG132 treatment did not further restore the decreased steady-state level of PPARγ by the addition of cycloheximide either (Fig. 4E). Our data suggest that NEDD4 blocks the PPARγ decrease by mechanisms similar to that of MG132, which is likely through inhibiting PPARγ proteasomal degradation.

Knockdown of NEDD4 reduces PPARγ expression and inhibits adipogenic response in 3T3-L1
adipocytes. PPARγ is arguably the central regulator of adipogenesis. The stabilization of PPARγ by NEDD4 raises the question of whether NEDD4 regulates adipogenesis. To address this hypothesis, we employed adeno-associated virus (AAV)-delivery of specific shRNA to knockdown NEDD4 in 3T3-L1 cells, which are a well-established culture model for studying adipogenesis 24 . NEDD4 was reduced by > 60% beyond day 5 post-differentiation (Fig. 5B,G). The knockdown of NEDD4 in 3T3-L1 cells significantly impeded the efficiency of the preadipocyte-to-adipocyte conversion as revealed by the oil red O staining at day 5 and 10 post-differentiation (Fig. 5A). To examine whether NEDD4 reduces PPARγ levels in 3T3-L1 adipocytes, we measured PPARγ expression these cells during the adipogenic conversion. Consistent with our hypothesis, PPARγ 2 expression was significantly reduced beyond day 5 post-differentiation (P < 0.05). Significant reduction in PPARγ 1 was only detected at day 6 post-differentiation (P < 0.05), possibly due to potential feedback and compensation effects on the γ 1 which is more labile than γ 2 13 (Fig. 5B,C). Protein expression of C/EBPα , another adipogenic factor, was also reduced in the shNEDD4 AAV-infected 3T3-L1 adipocytes (Fig. 5B). The mRNA levels for aP2 and C/EBPα, two PPARγ downstream target genes, were decreased by NEDD4 knockdown (Fig. 5D,E). Despite the decrease in PPARγ protein, its mRNA levels were not altered in comparison to those infected with non-targeting control shRNA AAV (Fig. 5F), confirming that NEDD4 regulates PPARγ at protein level. Of note, the expression of NEDD4 per se was not changed during adipocyte differentiation (Fig. 5B,G, Supplementary Fig. S2). We also examined whether NEDD4 influences PPARγ transcriptional activity. We performed PPARγ luciferase reporter assay. To our surprise, PPARγ transactivation by rosiglitazone was not altered by NEDD4 knockdown or NEDD4 overexpression (Fig. 5H). These data indicate that NEDD4 alters adipocyte differentiation primarily by regulating PPARγ abundance at the protein level, but not its transcriptional activation.

NEDD4 is not essential for ligand-dependent degradation of PPARγ in 3T3-L1 adipocytes.
Consistent with previous findings 12,20 , treatment of differentiated 3T3-L1 cells with the PPARγ agonist rosiglitazone induced both PPARγ 1 and γ 2 protein downregulation which was blocked by pretreatment of the proteasome inhibitor MG132 (Fig. 6A). To test if NEDD4 has an effect on ligand-induced degradation of PPARγ , we knocked down NEDD4 expression in 3T3-L1 cells and incubated the cells with or without rosiglitazone. Rosiglitazone induced significant downregulation in PPARγ 1 and γ 2 protein. Quantified results showed that the γ 1and γ 2 protein levels were significantly decreased after rosiglitazone treatment. Knockdown of NEDD4 further downregulated the expression of γ 1 and γ 2. However, the percentages of decrease were similar to those in shControl AAV infected cells (Fig. 6B), suggesting that NEDD4 is not essential for ligand-dependent degradation of PPARγ . We further examined the levels of ubiquitinated endogenous PPARγ in AAV infected 3T3-L1 cells with or without rosiglitazone treatment. As the cells treated with rosiglitazone were associated with lower total PPARγ protein levels, the amount of protein immunoprecipitated down from ligand-treated cells was also less. Our results showed that shNEDD4 AAV infected cells contained lower levels of ubiquitinated PPARγ compared with that in shControl AAV infected cells in the presence or absence of ligand (Fig. 6C). The expression of both PPARγ isoforms during time-dependent treatment of rosiglitazone in shNEDD4 AAV infected cells was not different from the control (Fig. 6D). We conclude that NEDD4 promotes ubiquitination of endogenous PPARγ in differentiated 3T3-L1 cells but is not essential in the ligand-dependent decrease of PPARγ protein.

NEDD4 positively correlates with PPARγ protein levels in obese adipose tissue.
To study the correlation between NEDD4 and PPARγ expression in obese adipose tissue, we analyzed NEDD4 and PPARγ protein expression in mixed epididymal fat tissue samples from wild-type and Nedd4 heterozygous (Nedd4 +/− ) mice fed a HFD or from aged wild-type and Nedd4 +/− mice. The samples were arranged in increasing order of CHO cells stably expressing FLAG-tagged PPARγ 2 were transfected with indicated plasmids. Two days after transfection, cells were treated with cycloheximide for the indicated times. Quantification of data is shown below. (C) The human (h) or rat (r) NEDD4 cDNAs were expressed in CHO cells stably expressing FLAGtagged PPARγ 2. Cycloheximide was added to the media 2 hr before cell harvesting. Quantification of data is on the right. (D,E) CHO cells stably expressing FLAG-tagged PPARγ 2 were co-transfected with the indicated plasmids and treated with or without the indicated reagents. Cycloheximide and MG132 were added to the media 15 hr before cell harvesting. Cycloheximide was used at 20 μ M. MG132 was used at 10 μ M. Data represent mean ± SEM of 3-4 independent experiments; *P < 0.05, **P < 0.01, ***P < 0.001. CHX: cycloheximide. Fulllength blots are presented in Supplementary Fig. S4. Luciferase activity was normalized to Renilla activity as a control for transfection efficiency. IDM: 0.5 mM IBMX, 1 μ M dexamethasone, and 1.5 μ g/mL insulin. Data represent mean ± SEM of 3 independent experiments; *P < 0.05, **P < 0.01, n.s., no significance. Full-length blots are presented in Supplementary Fig. S4.

Discussion
We have shown here that the ubiquitin ligase NEDD4 stabilizes PPARγ , promoting adipogenesis in 3T3-L1 cells. The evidence supporting this conclusion is as follows. First, NEDD4 interacts with and ubiquitinates PPARγ . Second, up-or down-regulation of NEDD4 prolongs or shortens the PPARγ protein half-life respectively. Third, knockdown of NEDD4 in 3T3-L1 cells reduces PPARγ expression and blocks adipocyte differentiation. Finally, the protein levels of NEDD4 and PPARγ are positively correlated in obese adipose tissue. PPARγ plays a central role in adipocyte differentiation. Increasing attention has been paid to PPARγ posttranslational modification, which includes phosphorylation, sumoylation and ubiquitination 23 . Despite the fact that PPARγ contains a typical PPxY binding motif for the WW domains of NEDD4, we did not observe interaction through this motif, as the amount of NEDD4 in the complex with PPxY mutant PPARγ was the same found in complex with wild-type PPARγ . Previous reports showed that WW-domain containing E3 ligases can be recruited by substrates in a PPxY-independent manner [25][26][27] . Our data indicated that NEDD4 binds the hinge/ LBD domain of PPARγ , and PPARγ missing the hinge and LBD domains was cleaved into smaller fragments, suggesting that these two domains are critical for the stability of PPARγ . The hinge region has been reported to regulate the subcellular distribution and trafficking of many nuclear receptors 28 . This region has also been shown to serve as an interaction site for modulators of nuclear receptors by numerous studies. Such a function has been demonstrated for androgen receptor [29][30][31] , estrogen receptor 32-36 , glucocorticoid receptor 37,38 , progesterone receptor 39 , PPARα 39 , and PPARγ [40][41][42] . The LBD has been more extensively studied in structure and function. The activation function 2 (AF-2) region within LBD can recruit co-activators via the amino acid LxxLL motif 43 . In line with these published reports, our findings add to the evidence that hinge and LBD domains of nuclear receptors are critical for nuclear receptor modulator and co-regulator recruitment.
Our data suggest that NEDD4 facilitates both K48 and K63-linked polyubiquitination of PPARγ in vitro, although the exact ubiquitin-chain type(s) formed in vivo may be influenced by other binding factors or different E2-E3 interactions. For example, (1) other binding factors may position PPARγ differently in E3-substrate complex. In this case, lysine preference in a given ubiquitination reaction would be affected. Alternatively, (2) the polyubiquitin chain is initially built by one or more E2s on the HECT cysteine residue. In this respect, the linkage specificity would be determined by the E2. NEDD4 has been shown to preferentially synthesize K63-linked ubiquitin chains in vivo and in vitro in numerous reports 44 . In our case, ubiquitin proteasome pathway is the major pathway for PPARγ degradation. We did not observe degradation of PPARγ when overexpressing NEDD4, suggesting that the ubiquitin chains added to PPARγ by NEDD4 fails to target PPARγ to the proteasomes. Therefore, it is less likely that K48-linked polyubiquitination is the major chain type formed by NEDD4 in PPARγ in cells. It is possible that NEDD4 preferentially mediates non-K48-linked polyubiquitination (eg. K63 linkage) in vivo which competes with the K48-linked polyubiquitination in PPARγ and therefore protects PPARγ from proteasomal degradation. It is also possible that a certain type of uncharacterized ubiquitin chain masks the recognition of PPARγ by proteasomes. Despite the fact that NEDD4 ubiquitinates PPARγ , we cannot rule out the possibility that direct binding of NEDD4 to PPARγ also contributes to the stabilization of PPARγ protein. Future studies could consider replacing the active-site cysteine residue in NEDD4 HECT domain with serine residue by genome editing, and see if PPARγ stability is affected. However, doing so might affect other substrates of NEDD4 as well which may exert feedback regulations on PPARγ . Although differential posttranslational regulation of the two PPARγ isoforms has not been previously demonstrated, functional differences between PPARγ 1 and γ 2 have been shown 9 . That the PPARγ isoforms could be differentially regulated is not surprising. In the case of our study, NEDD4 affects PPARγ 1 and γ 2 differently. PPARγ 1 expression decreases when NEDD4 is knocked down but to a lesser extent than the PPARγ 2 expression. Previously published data showed that there are no ubiquitination sites in the A/B domain. Neither the PPARγ 1 nor γ 2A/B domain are modified by ubiquitin. However, those domains do seem to influence stability and proteasome degradation independent of ubiquitin modification 21 . The turnover rate of PPARγ 1 and γ 2 are also differentthe half-life of γ 1 is shorter than γ 2 20 , so the N-terminal extension has an effect on stability whether or not a ligand is bound in the LBD. There is a previous study showing that the A/B domain can influence ligand binding through interdomain communication 45 , so it is possible that other posttranslational modifications of the γ 2A/B domain affect NEDD4-mediated ubiquitination and the stability of PPARγ , especially considering the fact that the N-terminal 30 amino acids of the PPARγ 2 contain several serine and threonine residues.
Polyubiquitinated proteins are normally degraded by the 26S proteasome 46 . It is easy to postulate that knockdown of PPAR-γ E3 ligase reduces ubiquitination of PPAR-γ and thus there is less PPAR-γ degradation and more transactivation leading to increased adipogenesis. In fact, the relationships between PPARγ ubiquitination, degradation and transcriptional activation are not always straight-forward. The E3 ligase SIAH2 has been shown to ubiquitinate PPARγ for degradation and to decrease its activity 16 . Depletion of SIAH2 in 3T3-L1 preadipocytes, however, prohibits adipocyte differentiation 16 . Another example is the E3 ligase TRIM23 which has recently been shown to ubiquitinate PPARγ , not for degradation, but rather, stabilization 17 . TRIM23 knockdown inhibits adipogenesis, but it does not appears to affect PPARγ transcriptional activity 17 . Similar to TRIM23's effects, we found that NEDD4 stabilizes PPARγ . Using a reporter gene assay, we demonstrated that the transcriptional activity of PPARγ is not altered by either NEDD4 overexpression or knockdown. Other substrates and signaling pathways that are regulated by NEDD4 may account for this puzzling outcome. It has been reported that NEDD4 can ubiquitinate tumor suppressor phosphatase and tensin homolog (PTEN), protein kinase B (PKB/AKT) and insulin signaling molecules [47][48][49] , all of which are involved in lipid metabolism. However, the regulation of PPARγ activity by these insulin signaling molecules in adipocytes has yet to be clarified. The final outcome of PPARγ activation, therefore, may be a balance between different NEDD4 substrates' regulation on PPARγ activity. These observations reveal differential roles for different E3 ligases in adipogenesis and PPARγ regulation. Because most mechanistic studies were performed with ectopically overexpressed PPARγ , PPARγ may stand out among other substrate candidates in this scenario. When performing adipogenesis or animal studies, other substrate(s) may be more preferentially regulated than the endogenous PPARγ by the E3 ligases. Therefore, whether the phenotypes in adipocytes are direct effects through PPARγ per se or mixed effects involved other pro-/anti-adipogenic molecules are unclear. Moreover, a remaining unsolved question is what are the functionally important regions and residues within PPARγ that are regulated by NEDD4. Delineating such regions may advance in the development of novel pharmacological agents for selective PPARγ stability modulators as alternatives to PPARγ agonists.

Methods
Immunoblot Analysis. Cells or tissues were lysed in radioimmunoprecipitation (RIPA) buffer supplemented with protease and phosphatase inhibitors. Cell debris in lysates was removed by centrifugation at maximum speed at 4 °C. Protein concentrations were determined by the bicinchoninic acid (BCA) method (Thermo Fisher Scientific, Waltham, MA). Sample were re-suspended in sodium dodecyl sulfate (SDS) sample buffer and denatured in boiling water for 5 min, then electrophoresed on a Novex 4-20% Tris-Glycine gel (Invitrogen, Carlsbad, CA). After transferring the proteins to a polyvinylidene difluoride (PVDF) membrane (Millipore, Billerica, MA), the membrane was blocked at room temperature in 1 × Tris-buffered saline plus 0.05% Tween-20 (TBST) containing 5% non-fat dry milk for 1 hr. After blocking, the membrane was incubated overnight at 4 °C in primary antibody diluted in TBST/5% BSA/0.02% NaN 3 . On the second day, the membrane was washed thoroughly with TBST and incubated in secondary horseradish peroxidase (HRP) linked antibody (GE Healthcare Life Sciences, Pittsburgh, PA). The blot was washed three times with TBST followed by incubation with enhanced chemiluminescence (ECL) for signal development. For data presented in Fig. 7, samples were run on different gels. Each gel had an equal number of wild-type samples and Nedd4 +/− samples. The intensity of protein NEDD4 and β -Actin bands was quantified by ImageJ. Then each of the densitometry values of NEDD4 was normalized to value of β -Actin of the same sample. The NEDD4/β -Actin ratio values of all samples were averaged from each blot to obtain a ratio value specific to each blot. The values of NEDD4/β -Actin on other blots were then adjusted to the first blot by the average ratio values of each blot. An average NEDD4/β -Actin value (X) for all samples was calculated. NEDD4/β -Actin ratio value of each sample was normalized to X, and was arranged in increasing order of NEDD4 expression. Same quantification method was applied for PPARγ 1 and γ 2. Adeno-associated Virus (AAV)-mediated Knockdown. The pAAV2.1.CMV.EGFP-U6. shRNA plasmid vector (a kind gift from Dr. Tonia Rex at Vanderbilt University, Nashville, TN, USA) was used to generate AAV virus. Of the three different short hairpin RNA (shRNA) constructs tested, the one showing most effective downregulation of mouse NEDD4 protein and mRNA (> 50%) was chosen for subsequent studies. The targeting sequence was 5′ -TGGCGATTTGTGAACCGTA-3′ . The non-targeting control sequence was 5′ -CAACAAGATGAAGAGCACCAA-3′ . Virus production was performed by the Gene Transfer Vector Core at the University of Iowa.
Differentiation of 3T3-L1 Cells and Oil Red O Staining. The 3T3-L1 preadipocytes were cultured in high glucose DMEM supplemented with 10% bovine calf serum. Cells were split 1:10 every 2 days to prevent reaching confluency. Cells were incubated in 95% O 2 and 5% CO 2 at 37 °C. For adipocyte differentiation, the preadipocytes were allowed to grow to confluency in 10% bovine calf serum. Two days post confluency, cells were switched to DMEM supplemented with 10% fetal bovine serum (FBS) and stimulated with a differentiation cocktail (IDM) containing 0.5 mM 3-isobutyl-1-methylxanthine (IBMX), 1 μ M dexamethasone, and 1.5 μ g/mL insulin. Two days after IDM induction, cells were fed with DMEM plus 10% FBS and 1.5 μ g/mL insulin. Three days later, cells were cultured in DMEM plus 10% FBS and media was replaced every 2-3 days until full differentiation was achieved. For oil red O staining, a stock oil red O solution was prepared by adding 150 mg of oil Red O powder to 50 mL of isopropanol. Then, 3 parts of oil Red O stock solution were mixed with 2 parts distilled water and incubated for 10 min at room temperature. The oil red O working solution was filtered through Whatman filter paper. Differentiated 3T3-L1 adipocytes cultured on 6-well dishes were first fixed in 10% formalin for 30-60 min. Cells were then briefly rinsed with 3 times with distilled water and incubated in 60% isopropanol for 5 min. Freshly prepared oil red O working solution was added to the cells for 5 min. Cultures were rinsed with room temperature tap water until the water rinses were clear. The stained lipid droplets were photographed with a digital camera. Bright-field images were captured with the IX50 inverted system microscope (Olympus corporation of the Americas) using magnification of × 40. A non-stained area was chosen to set the white balance.
Co-Immunoprecipitation. Cells were collected in a solution containing 25 mM Tris-HCl, 150 mM NaCl, and 1% NP-40 supplemented with protease and phosphatase inhibitors. Lysates were passed 5 times through a 25 G syringe to increase protein extraction. The lysates were centrifuged to obtain soluble proteins. After determining the protein concentrations, ~500 μ g of total protein was used for a co-immunoprecipitation (co-IP) assay. Then, 1 μ g of antibody or IgG, and 30 μ l of protein A/G beads (Santa Cruz Biotechnology, Dallas, TX) were added to each sample for overnight incubation with gentle shaking at 4 °C. On the second day, beads were centrifuged at 5,000 rpm for 2 min, washed three times with the co-IP buffer, and finally resuspended in 1 × SDS sample buffer and denatured in boiling water for 5 min. Samples were centrifuged again at maximum speed and the resulting supernatant was loaded directly on a Tris-Glycine gel or stored at − 80 °C until use.

In Vitro
Dual Luciferase Reporter Assay. HEK293 cells of relatively equal numbers were plated on 24-well plates.
On the second day, seeded cells were transiently co-transfected with 250 ng 3 × PPRE-Luc, 250 ng PPARγ 2-FLAG, 10 ng Renilla luciferase control reporter vector pRL-SV40 and 250 ng of other plasmids using Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, CA). At 32 hr after the transfection, cells were treated with rosiglitazone (10 μ M) for 16 hr. The luciferase assay was performed by a dual luciferase assay system (Promega, Madison, WI) according to the manufacturer's instructions. The luciferase activity was adjusted by the Renilla luciferase activity to remove the variation caused by transfection efficiency.
Statistical Analysis. All data represent mean ± SEM. Two-tailed, paired Student's t-test was used to analyze the statistical significance of data. Statistical significance was determined by P value of less than 0.05 (P < 0.05).