Activation of Peroxisome Proliferator–Activated Receptor β/δ Inhibits Lipopolysaccharide-Induced Cytokine Production in Adipocytes by Lowering Nuclear Factor-κB Activity via Extracellular Signal–Related Kinase 1/2

The PPARβ/δ ligand GW501516 was from Biomol Research Labs (Plymouth Meeting, PA). Other chemicals were from Sigma (St. Louis, MO).

3T3-L1 preadipocytes (American Type Culture Collection) were grown to confluence in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% bovine calf serum. Differentiation of the 3T3-L1 cells was induced 2 days after confluence (day 0) in DMEM containing 10% fetal bovine serum, 500 μmol/l methylisobutylxanthine, 0.25 μmol/l dexamethasone, and 10 μg/ml insulin for 48 h. The cells were then incubated in 10% FBS/DMEM with insulin for 8 days. Medium was changed every 2 days. Fat droplets were observed in >90% of cells after day 10. Adipocytes were then incubated for 96 h with 0.5 μmol/l GW501516 and then with 100 ng/ml LPS for either 1 or 24 h. After the incubation, RNA, total proteins, and nuclear extracts were extracted from adipocytes as described below. Inhibitors were added 30 min before the incubation with LPS. Culture supernatants were collected, and the secretion of IL-6 was assessed by ELISA (Amersham Biosciences, Little Chalfont, U.K.).

Male obese ZDF rats (ZDF/Gmi, fa/fa) and lean litter mates (fa/+ or +/+) were used. Both strains were maintained under standard light-dark cycle (12-h light/dark cycle) and temperature (21 ± 1°C) and fed with the Purina 5008 chow. Epidydimal white adipose tissue was rapidly removed, frozen in liquid nitrogen, and stored at −80°C. Blood samples were collected in EDTA tubes, and plasma was obtained by centrifugation at 2,200 × g for 10 min at 4°C. Plasma glucose (Sigma), triglycerides (Sigma), and nonesterified fatty acids (Wako, Germany) levels were determined with a colorimetric test. Insulin (Amersham Biosciences) was determined by radioimmunoassay. All procedures were conducted in accordance with the principles and guidelines established by the University of Barcelona Bioethics Committee, as stated in Law 5/1995, 21st July, passed by the Generalitat de Catalunya. The generation of PPARβ/δ-null mice has been described previously (20).

Levels of mRNA were assessed by the RT-PCR as previously described (21). Total RNA was isolated using the Ultraspec reagent (Biotecx, Houston, TX). The total RNA isolated by this method is nondegraded and free of protein and DNA contamination. The sequences of the sense and antisense primers used for amplification were: IL-6, 5′-TCCAGCCAGTTGCCTTCTTGG-3′ and 5′-TCTGACAGTGCATCATCGCTG-3′; Mcp-1, 5′-GGGCCTGTTGTTCACAGTTGC-3′ and 5′-GGGACACCTGCTGCTGGTGAT-3′; Pdk-4, 5′-AGGTCGAGCTGTTCTCCCGCT-3′ and 5′-GCGGTCAGGCAGGATGTCAAT-3′; Cpt-I, 5′-TTCACTGTGACCCCAGACGGG-3′ and 5′-AATGGACCAGCCCCATGGAGA-3′; Pparβ/δ 5′-GAGGAAGTGGCCACGGGTGAC-3′ and 5′-CCACCTGAGGCCCCATCACAG-3′; Socs-3, 5′-TTTTCGCTGCAGAGTGACCCC-3′ and 5′-TGGAGGAGAGAGGTCGGCTCA-3′; and adenosyl phosphoribosyl transferase (Aprt), 5′-GCCTCTTGGCCAGTCACCTGA-3′ and 5′-CCAGGCTCACACACTCCACCA-3′. Amplification of each gene yielded a single band of the expected size (IL-6, 229 bp; Mcp-1, 157 bp; Pdk-4, 167 bp; Cpt-I, 222 bp; Pparβ/δ, 151 bp; Socs-3, 250 bp; and Aprt, 329 bp). Preliminary experiments were carried out with various amounts of cDNA to determine nonsaturating conditions of PCR amplification for all the genes studied. Therefore, under these conditions, relative quantification of mRNA was assessed by the RT-PCR method used in this study (22). Radioactive bands were quantified by video-densitometric scanning (Vilbert Lourmat imaging). The results for the expression of specific mRNAs are always presented relative to the expression of the control gene (Aprt).

Nuclear extracts were isolated as previously described (21). Cells were scraped into 1.5 ml cold PBS, pelleted for 10 s, and resuspended in 400 μl cold buffer A (10 mmol/l HEPES, pH 7.9, at 4°C, 1.5 mmol/l MgCl₂, 10 mmol/l KCl, 0.5 mmol/l dithiothreitol [DTT], 0.2 mmol/l phenylmethylsulfonyl fluoride [PMSF], and 5 μg/ml aprotinin) by flicking the tube. Cells were allowed to swell on ice for 10 min and then vortexed for 10 s. Samples were then centrifuged for 10 s, and the supernatant fraction was discarded. Pellets were resuspended in 50 μl cold buffer C (20 mmol/l HEPES-KOH, pH 7.9, at 4°C, 25% glycerol, 420 mmol/l NaCl, 1.5 mmol/l MgCl₂, 0.2 mmol/l EDTA, 0.5 mmol/l DTT, 0.2 mmol/l PMSF, 5 μg/ml aprotinin, and 2 μg/ml leupeptin) and incubated on ice for 20 min for high-salt extraction. Cellular debris was removed by centrifugation for 2 min at 4°C and the supernatant fraction (containing DNA-binding proteins) was stored at −80°C. Nuclear extract concentration was determined by the Bradford method.

Electrophoretic mobility shift assay (EMSA) was performed using double-stranded oligonucleotides (Promega, Madison, WI) for the consensus binding site of the NF-κB nucleotide (5′-AGTTGAGGGGACTTTCCCAGGC-3′) and PPAR (5′-CAAAACTAGGTCAAAGGTCA-3′). Oligonucleotides were labeled in the following reaction: 2 μl oligonucleotide (1.75 pmol/μl), 2 μl 5× kinase buffer, 1 μl T4 polynucleotide kinase (10 units/μl), and 2.5 μl [γ-³²P]ATP (3,000 Ci/mmol at 10 mCi/ml) incubated at 37°C for 1 h. The reaction was stopped by adding 90 μl Tris-EDTA buffer (10 mmol/l Tris-HCl, pH 7.4, and 1 mmol/l EDTA). To separate the labeled probe from the unbound ATP, the reaction mixture was eluted in a Nick column (Amersham) following the manufacturer's instructions. Eight micrograms of crude nuclear protein was incubated for 10 min on ice in binding buffer (10 mmol/l Tris-HCl, pH 8.0, 25 mmol/l KCl, 0.5 mmol/l DTT, 0.1 mmol/l EDTA, pH 8.0, 5% glycerol, 5 mg/ml BSA, and 50 μg/ml poly(dI-dC)), in a final volume of 15 μl. Labeled probe (∼60,000 cpm) was added, and the reaction was incubated for 15 min at 4°C (NF-κB) or room temperature (peroxisome proliferator response element). Where indicated, specific competitor oligonucleotide was added before the labeled probe and incubated for 10 min on ice. p65 antibody was added 15 min before incubation with the labeled probe at 4°C. Protein-DNA complexes were resolved by electrophoresis at 4°C on a 5% acrylamide gel and subjected to autoradiography.

To obtain total protein, cells and adipose tissue were homogenized in cold lysis buffer (5 mmol/l Tris-HCl, pH 7.4, 1 mmol/l EDTA, 0.1 mmol/l PMSF, 1 mmol/l sodium orthovanadate, and 5.4 μg/ml aprotinin). The homogenate was centrifuged at 16,700 × g for 30 min at 4°C. Protein concentration was measured by the Bradford method. Proteins (30 μg) were resolved by SDS-PAGE on 10% separation gels and transferred to Immobilon polyvinylidene difluoride membranes (Millipore, Bedford, MA). Western blot analysis was performed using antibodies against total and phospho-ERK1/2 (Cell Signaling) and total (Santa Cruz) and phospho-STAT3 (Tyr⁷⁰⁵) and Ser⁷²⁷) (Cell Signaling). Detection was achieved using the EZ-ECL chemiluminescence kit (Amersham). Size of detected proteins was estimated using protein molecular-mass standards (Invitrogen, Barcelona, Spain).

Results are expressed as means ± SD of five separate experiments. Significant differences were established by Student's t test or one-way ANOVA, according to the number of groups compared, using the GraphPad Instat program, v2.03 (GraphPad Software, San Diego, CA). In the latter case, when significant variations were found, the Tukey-Kramer multiple comparisons test was applied. Differences were considered significant at P < 0.05.

Differentiated 3T3-L1 adipocytes were exposed for 24 h to 0.5 μmol/l GW501516, a selective ligand for PPARβ/δ with an apparent K_d value of 1 nmol/l and 1,000-fold higher affinity toward PPARβ/δ than PPARα and PPARγ (23). Under these conditions, we evaluated the expression of two well-known PPARβ/δ-target genes, pyruvate dehydrogenase kinase 4 (Pdk4) and carnitine palmitoyltransferase 1 (Cpt-I) (24). Ligand treatment for 24 h caused a slight increase in the mRNA levels of both Pdk4 (100%, P < 0.01) and Cpt-I (45%, P < 0.05) (Fig. 1A). When adipocytes were treated for 96 h with 0.5 μmol/l GW501516, a greater increase was observed in the transcript levels of Pdk4 (170%, P < 0.001) and Cpt-I (300%, P < 0.01) (Fig. 1B). Therefore, we selected this longer exposure to the PPARβ/δ ligand in further experiments. No changes were observed either in the expression of well-known PPARγ-target genes (i.e., Ob: data not shown), confirming that the drug treatment was selective for PPARβ/δ, or in the expression of markers of adipocyte differentiation (i.e., Pref-1; data not shown), indicating that this latter process was unaffected. After a 96 h-incubation with GW501516, cells were exposed to 100 ng/ml LPS for 24 h to induce inflammation. LPS activates NF-κB through TLRs in adipocytes (9,25) and induces insulin resistance (12,26). When we evaluated the mRNA levels of Mcp-1 and IL-6, two genes that are under the control of the pro-inflammatory transcription factor NF-κB, we observed that LPS treatment strongly increased the former (22-fold induction, P < 0.001), whereas cells co-incubated with GW501516 showed a significant reduction (51% reduction, P < 0.01 vs. LPS-exposed cells) (Fig. 2A). Furthermore, when LPS-exposed cells were co-incubated with 10 μmol/l parthenolide, which specifically inhibits the activation of NF-κB (27), the effect of LPS was suppressed. Treatment with LPS also led to enhanced IL-6 mRNA levels (4.2-fold induction, P < 0.05), but this increase was blocked in the presence of GW501516 and parthenolide (Fig. 2B). Likewise, incubation with LPS for 24 h caused a 4.2-fold increase in the levels of IL-6 protein secreted into the culture media (control 10.5 ± 3.7 vs. LPS 55 ± 11 ng/ml; Fig. 2C), whereas pretreatment with GW501516 (56% reduction, P < 0.001) and parthenolide (32% reduction, P < 0.05) significantly reduced the secretion of this cytokine. Several cytokines, including IL-6, stimulate Socs-3 expression in adipocytes (28), a protein that inhibits insulin signaling in these cells (29). The elevated levels of mRNA encoding Socs-3 and the increase in Stat-3 phosphorylation (Fig. 2D and E) suggested that the increase in IL-6 release caused by LPS treatment might have local actions (30). Interestingly, GW501516 treatment prevented the increase in Socs-3 expression and Stat-3 phosphorylation, a fact that is consistent with the reduction in IL-6 levels after drug treatment.

Activation of NF-κB plays a crucial role in IL-6 production in adipocytes, as demonstrated by the effects of parthenolide. To test whether GW501516 prevented LPS-induced IL-6 expression by reducing NF-κB activity, we performed EMSA. NF-κB formed three complexes with nuclear proteins (complexes I–III) (Fig. 3A). The specificity of the three DNA-binding complexes was assessed in competition experiments by adding an excess of unlabeled NF-κB oligonucleotide. NF-κB binding activity, mainly of specific complex II, increased in nuclear extracts from LPS-treated cells (Fig. 3B). In contrast, in the presence of LPS and the PPARβ/δ ligand, the binding activity of NF-κB was reduced (Fig. 3A). Addition of antibody against the p65 subunit of NF-κB completely supershifted complex II, thereby indicating that this band comprised mainly this subunit (Fig. 3C).

We hypothesized that a reduction in PPARβ/δ activity and the consequent increase in NF-κB activity is responsible, at least in part, for the enhanced expression of IL-6 in white adipose tissue of animal models of obesity and insulin resistance. To check this hypothesis, we evaluated the expression of PPARβ/δ and its activity in white adipose tissue of ZDF (fa/fa) rats. Male ZDF and lean rats were killed at 12 weeks of age. By this age, ZDF rats presented established diabetes compared with lean animals, as demonstrated by increased plasma levels of glucose (505 ± 67 vs. 167 ± 37 mg/dl, P < 0.0001), triglycerides (649 ± 298 vs. 92 ± 31 mg/dl, P < 0.01), nonesterified fatty acids (0.62 ± 0.17 vs. 0.26 ± 0.09 mEq/l, P < 0.01), and insulin (19.3 ± 7.9 vs. 10.4 ± 2.0 ng/ml, P < 0.05). White adipose tissue of ZDF rats showed a significant reduction in Pparβ/δ transcripts (50% reduction, P < 0.05) (Fig. 4A). Similarly, the mRNA levels of its target gene, Pdk4, were also reduced (41%, P < 0.05) (Fig. 4B), suggesting that the activity of this PPAR isotype was reduced. In contrast, the expression of IL-6 was significantly enhanced (60%, P < 0.05) (Fig. 4C). EMSA showed that the DNA-binding activity of PPAR was reduced in white adipose tissue from ZDF rats compared with lean animals (Fig. 4D). In contrast, NF-κB DNA-binding activity was increased in ZDF rats compared with lean animals (Fig. 4E). These data suggest that the low-grade chronic systemic inflammatory associated with the development of obesity and insulin resistance involves a decreased PPARβ/δ activity and the subsequent activation of NF-κB.

To clearly demonstrate the involvement of PPARβ/δ in the control of NF-κB–mediated Il-6 expression in white adipose tissue, we used the PPARβ/δ-null mice. The absence of Pparβ/δ in white adipose tissue (Fig. 5A) was associated with a reduction in the mRNA levels of Pdk4 (65%, P < 0.001) (Fig. 5B) compared with wild-type mice, whereas IL-6 expression was enhanced (fourfold induction, P < 0.05) (Fig. 5C). In agreement with the enhanced expression of IL-6 in white adipose tissue of PPARβ/δ-null mice, NF-κB binding activity was higher in this tissue compared with that of wild-type mice (Fig. 5D).

Next, we attempted to determine the molecular mechanism by which PPARβ/δ activation inhibits NF-κB. Although it has been reported that PPARβ/δ activation prevents LPS-induced degradation of inhibitor of κBα (IκBα), thus suppressing LPS-induced NF-κB in cardiomyocytes (31), we did not observe changes in the protein levels of this NF-κB inhibitor (data not shown) after GW501516 treatment. On the other hand, ERK1/2 activation leads to NF-κB activation in adipocytes (9). In agreement with this hypothesis, the adipose tissue of ZDF rats showed enhanced phospho-ERK1/2 levels in the adipose tissue of ZDF rats compared with lean rats (Fig. 6A). This observation supports a potential relationship between activation of ERK1/2 and NF-κB. To demonstrate the involvement of the ERK-MAPK cascade in LPS-induced NF-κB activation in adipocytes, we used U0126, a potent and specific ERK1/2 inhibitor, which binds to MEK, thereby inhibiting its catalytic activity and phosphorylation of ERK1/2. Cells exposed to LPS showed increased mRNA levels of both Mcp-1 (31-fold induction, P < 0.001) and IL-6 (fourfold induction, P < 0.05), whereas co-incubation with U0126 caused a 35% (P < 0.001) and 72% (P < 0.05) reduction in the LPS-mediated induction of these two genes, respectively (Fig. 6B and C). As a control, we verified that U0126 treatment inhibited LPS-induced ERK1/2 activation (Fig. 6D).

Finally, we examined whether the PPARβ/δ ligand prevented LPS-induced NF-κB activation by reducing ERK1/2 phosphorylation in adipocytes. Immunoblotting detection of total and phosphorylated ERK1/2 (Fig. 6E) revealed that LPS treatment activated this pathway, whereas in the presence of the PPARβ/δ agonist, this pathway was blocked. These observations suggest that activation of this transcription factor prevents NF-κB activation by inhibiting ERK1/2 phosphorylation. To demonstrate that PPARβ/δ regulates the ERK1/2 pathway, we explored the expression of this kinase in adipose tissue of the PPARβ/δ-null mice. These mice showed enhanced phospho-ERK1/2 protein levels compared with wild-type mice (Fig. 6F), indicating that PPARβ/δ regulates this pathway.

Obesity, insulin resistance, and type 2 diabetes are closely associated with low grade of chronic inflammation characterized by abnormal cytokine production (32). The adipocyte plays a crucial role in this process, because this cell is a source of cytokines (TNF-α, IL-6, and MCP-1), which are secreted as a result of the activation of several signaling cascades involved in obesity-induced insulin resistance (33). A number of studies have implicated chronic activation of the pro-inflammatory transcription factor NF-κB as one of these signaling pathways (34) that links inflammation with obesity and type-2 diabetes. For instance, overexpression of the NF-κB activator IKKβ in mice results in increased inflammatory cytokine production and the onset of diabetes (35). Furthermore, in human adipose tissue, inhibition of NF-κB suppresses the release of pro-inflammatory cytokines (36). Activation of NF-κB can be achieved by either LPS or FFAs through TLR4 (26). Several studies have proposed that LPS participates in initiating the subclinical inflammation observed in obesity and type 2 diabetes. Thus, circulating LPS levels are higher in type 2 diabetic patients than in lean healthy subjects, and LPS stimulates the secretion of pro-inflammatory cytokines in human adipocytes (37).

Here, we demonstrate for the first time that PPARβ/δ activation in adipocytes inhibits LPS-induced cytokine expression and secretion by preventing NF-κB activation. We (38) and others (31) have previously proposed that PPARβ/δ activation inhibits NF-κB activation in cardiac cells either by promoting a protein-protein interaction between PPARβ/δ and the p65 subunit of NF-κB or by increasing the expression levels of the NF-κB inhibitor IκBα. In adipocytes, we did not observe changes either in the PPARβ/δ-p65 interaction (data not shown) or in IκBα protein levels after GW501516 treatment. These observations suggest that these mechanisms are not involved in the anti-inflammatory effects of this drug in adipocytes. In contrast, it has been previously reported that ERK1/2 activation is crucial for the induction of inflammatory changes in adipocytes (34) and leads to enhanced NF-κB activity (9). In agreement with this role of ERK1/2 in inflammation in adipocytes, we observed a reduction in the expression of pro-inflammatory cytokines in these cells when exposed to LPS in the presence of the MAPK pathway inhibitor, U0126. This finding is consistent with previous studies reporting reduced cytokine expression levels in the presence of this inhibitor in human adipocytes (9). These data demonstrate that ERK1/2 inhibition prevents NF-κB activation and the subsequent increase in cytokine expression. In addition, we demonstrate that PPARβ/δ activation inhibits ERK1/2 phosphorylation, which is consistent with the reduction in NF-κB activation after GW501516 treatment. Furthermore, the increase in phospho-ERK1/2 levels observed in white adipose tissue from PPARβ/δ-null mice is consistent with enhanced NF-κB activity. The involvement of PPARβ/δ in ERK1/2 regulation has been previously reported in other cell types. In fact, Burdick et al. (39) recently reported that the PPARβ/δ agonist GW0742 reduces phospho-ERK1/2 levels in human keratinocytes. In addition, phospho-ERK1/2 levels are higher in the skin of PPARβ/δ-null mice treated with 12-O-tetradecanoylphorbol-13-acetate than in the skin of wild-type animals (40). Overall, these data confirm that PPARβ/δ regulates phospho-ERK1/2 levels in several tissues. Additional studies are required to elucidate the molecular mechanism by which PPARβ/δ controls ERK-1/2 activation.

Here, we demonstrate a correlation between enhanced ERK1/2 phosphorylation and NF-κB activation in the white adipose tissue of a genetic model of obesity and diabetes, the ZDF rat. Given the low level of Pparβ/δ in the white adipose tissue of these animals, we hypothesize that a decrease in the expression of this nuclear receptor is one of the underlying mechanisms of this correlation. On the basis of our data, we propose that in this animal model, PPARβ/δ downregulation leads to enhanced ERK1/2 phosphorylation, which in turn activates NF-κB in white adipose tissue and, as a result, induces pro-inflammatory cytokine expression. This hypothesis is supported by the observation that white adipose tissue from the PPARβ/δ-null mice shows increased phospho-ERK1/2 levels and NF-κB activity and higher expression of IL-6 compared with wild-type mice. Although further studies are required to demonstrate this relationship between PPARβ/δ and NF-κB in white adipose tissue, our data provide further insight into the role of obesity-related insulin resistance and low-grade inflammation.

In summary, on the basis of our findings in adipocytes, we propose that PPARβ/δ activation prevents LPS-induced NF-κB activation via ERK1/2, thereby reducing the production of pro-inflammatory cytokines involved in the development of insulin resistance. This action of PPARβ/δ may contribute to preventing obesity-induced insulin resistance.

R.R.-C. has received a grant from the Fundación Ramón Areces. T.C. has received a grant from the Spanish Ministerio de Educación y Ciencia. X.P. has received a grant from the Spanish Ministerio de Educación y Ciencia. This study has received funds from the Swiss National Science Foundation, Generalitat de Catalunya (SGROS-00833), Fundación Ramón Areces, the Spanish Ministerio de Educación y Ciencia (SAF2006-01475), Instituto de Salud Carlos III-RETIC RD06/0,015/ERDF, European Union ERDF funds, and the Fundació Privada Catalana de Nutrició i Lípids.

We thank the University of Barcelona's Language Advisory Service for its help.