Obesity and insulin resistance are often associated with lower circulating adiponectin concentrations and elevated serum interleukin-6 (IL-6) and/or tumor necrosis factor-α (TNF-α). Adiponectin suppresses activation of nuclear factor-κB (NF-κB) in aortic endothelial cells and porcine macrophages. Accordingly, we hypothesized that adiponectin is an anti-inflammatory hormone and suppresses activation of NF-κB in adipocytes. Because peroxisome proliferator-activated receptor γ2 (PPARγ2) antagonizes the transcriptional activity of NF-κB, we determined whether adiponectin alters PPARγ2 expression in pig adipocytes. In addition, we determined whether interferon-γ alters the expression of PPARγ2 in the presence or absence of adiponectin. Primary adipocytes from pig subcutaneous adipose tissue were treated with or without lipopolysaccharide (LPS; 10 μg/ml) and adiponectin (30 μg/ml), and nuclear extracts were obtained for gel shift assays to assess nuclear localization of NF-κB. Whereas LPS induced an increase in NF-κB activation, adiponectin suppressed both NF-κB activation and the induction of IL-6 expression by LPS (P < 0.05). Similar results were obtained in 3T3-L1 adipocytes. In addition, adiponectin antagonized LPS-induced increase in TNF-α mRNA expression (P < 0.05) and tended (P < 0.065) to diminish its accumulation in the culture media in 3T3-L1 adipocytes. Adiponectin also induced an upregulation of PPARγ2 mRNA (P < 0.05). Although IFN-γ did not reduce the basal expression of PPARγ2, it suppressed PPARγ2 induction by adiponectin (P < 0.05). These findings indicate that adiponectin may be a local regulator of inflammation in the adipocyte and adipose tissue via its regulation of the NF-κB and PPARγ2 transcription factors.
- nuclear factor-κB
- peroxisome proliferator-activated receptor γ
recent advances in adipose biology have provided convincing evidence that the adipocyte does not simply store energy but also secretes multiple proteins that influence metabolism in peripheral tissues (1, 7, 14). Adiponectin is such a protein and acts directly to regulate metabolic pathways in adipose tissue (10), liver (31), and skeletal muscle (30). Adiponectin also enters the brain and regulates energy expenditure through central mechanisms (20).
Adiponectin acts as an anti-inflammatory protein and suppresses cytokine production by activated macrophages (28, 32). Mechanistically, adiponectin has been shown to disrupt the activation of NF-κB, a major transcriptional regulator of proinflammatory cytokine expression, in an aortic endothelial cell model (16). This finding was recently extended to activated porcine macrophages in that adiponectin attenuated the nuclear translocation of NF-κB and attenuated the induction of TNF-α and IL-6 by LPS (28).
Increased adipose expression of TNF-α and IL-6 and increased circulating concentrations of one or both of these cytokines are commonly associated with obesity and insulin resistance (11, 21). There is now compelling evidence that obesity results in a marked accumulation of macrophages in the adipose tissue, and transcript profiles in adipocytes and adipose-derived macrophages reflect an active inflammatory state (27, 29). Considering that LPS induces NF-κB activation in adipocytes (2), coupled with the anti-inflammatory activities of adiponectin in macrophages, we hypothesized that adiponectin functions through an autocrine pathway to suppress inflammation in adipocytes. Consequently, we conducted experiments to establish whether adiponectin suppresses LPS-induced activation of NF-κB and proinflammatory cytokine production in two adipocyte models. Because peroxisome proliferator-activated receptor γ2 (PPARγ2) exerts anti-inflammatory actions and antagonizes the transcriptional activity of NF-κB, and because overexpression of adiponectin has previously been shown to induce the expression of PPARγ2 in adipose tissue of mice (10), we also addressed the possibility that adiponectin acts directly to upregulate the expression of PPARγ2 in pig adipocytes.
MATERIALS AND METHODS
All animal handling procedures were approved by the institutional animal care and use committee and were described previously (4). Adipocytes from the subcutaneous fat depot of castrate male pigs weighing ∼100 kg were isolated by collagenase digestion as described previously (2). The isolated cells were diluted to approximate 20% cell suspensions with DMEM (Sigma, St. Louis, MO) containing 3% fatty acid-free serum albumin. The cell suspensions were gassed initially and at 2-h intervals with a mixture of air and CO2 and were incubated in a gyratory floor incubator at 37°C for the selected duration. For experiments with pig adipocytes in which media and mRNA were recovered for cytokine expression analyses, cells were pretreated for 2 h with adiponectin (30 μg/ml) and then treated with LPS (10 μg/ml) or porcine IFN-γ (50 ng/ml) for an additional 3 h. Reactions were terminated by allowing adipocytes to float on the surface and carefully aspirating the medium from the bottom of the vials. Total RNA was isolated from the cells by using TRIzol reagent (Invitrogen, Carlsbad, CA). All experiments were repeated at least three times with cells obtained from different pigs. Within each experiment (pig), each treatment was replicated at least three times.
Gel shift assay.
Cells were treated with LPS for 2 h before nuclear extracts were recovered for this assay. The recovery of nuclear extracts and mobility shift assays were performed and validated using supershift analysis as described previously (2). A total of 15 μg of nuclear protein were used for each mobility shift assay.
3T3-L1 adipocyte culture.
3T3-L1 preadipocytes (ATCC, Manassas, VA) were seeded in 24-well plates and cultured according to standard conditions. Briefly, cells were grown in 5% CO2 in medium containing 10% calf serum (GIBCO, Carlsbad, CA) in the presence of 1% penicillin-streptomycin mixture (Invitrogen). Two days postconfluence (day 0), cells were induced to differentiate with a medium containing 10% fetal bovine serum (GIBCO), 1.7 μM insulin, 1 μM dexamethasone, and 0.5 mM IBMX for 48 h. Thereafter, fresh medium containing only insulin was added every 2 days for another 6 days. On day 8, cells were fully differentiated and medium was changed to insulin-free medium containing 10% fetal bovine serum.
Cell transfection and reporter assay.
On day 5 after the induction of differentiation, cells were transfected with Fugene 6 reagent (Roche, Indianapolis, IN) for 48 h with pNF-κB-Luc plasmid (0.25 μg/well), which contains five repeats of the NF-κB enhancer element (TGGGGACTTTCCGC) driving the expression of a firefly luciferase gene (Stratagene, La Jolla, CA). Additional wells were also transfected (0.25 μg/well) with positive control (pFC-MEKK) and negative control (pLuc-MCS) plasmids (Stratagene) for validation of the assay. Transfection efficiency was normalized by cotransfecting cells (0.05 μg/well) with pRL-CMV vector (Promega, Madison, WI), a plasmid encoding the Renilla luciferase gene under the cytomegalovirus promoter. Fresh medium was added to cells after 48 h (day 7 of differentiation). Experiments on the fully differentiated cells having the reporters were conducted on day 8 by pretreating cells for 6 h with recombinant porcine adiponectin (30 μg/ml). Thereafter, cells were treated with LPS (100 ng/ml) for 18 h, after which medium and the cell lysate were recovered for ELISA and the luciferase assay, respectively. The Dual-Luciferase Reporter Assay System (Promega) was used to quantify the expression of the firefly luciferase and the Renilla luciferase. The firefly luciferase was normalized to the Renilla and presented relative to the controls.
RNase protection assays.
The RNase protection assay for IL-6 mRNA was conducted, as described previously (2), using a pig-specific multiprobe RNase Protection System (Pharmingen, San Diego, CA). Generally, 50 μg of total RNA were incubated at 56°C with [32P]UTP-labeled antisense RNA. RNase protection assay for PPARγ2 was carried out as described previously (3). Briefly, 20 μg of total RNA were hybridized to a PPARγ riboprobe as described previously (15). Protected bands obtained after RNase digestion were resolved on a 5% denaturing acrylamide gel. Autoradiographs were prepared from dried gels, and signal intensity was quantified using a Digital Science Imaging System (v. 2.0.1; Kodak, New Haven, CT).
Media concentrations of IL-6 and TNF-α were measured using mouse-specific ELISA kits from Pierce Endogen (Rockford, IL). The assay sensitivity was <7 pg/ml. Both intra-assay and interassay coefficients of variation were <10%.
Real-time quantitative PCR.
Total RNA recovered from cells as described above were DNase treated (Turbo DNase; Ambion, Houston, TX) and reverse transcribed using the Superscript First-Strand cDNA Synthesis kit (Invitrogen). Primer sequences for mouse IL-6 were 5′-AACGATGATGCACTTGCAGA-3′ and 5′-GAGCATTGGAAATTGGGGTA-3′ for the sense and antisense primers, respectively (12). Primers for β-actin and TNF-α were, for β-actin, sense, 5′-ATGGGTCAGAAGGACTCCTACG-3′ and antisense, 5′-AGTGGTACGACCAGAGGCATAC-3′, and for TNF-α, sense, 5′-TCCCCAAAGGGATGAGAAGTTC-3′ and antisense, 5′-TCATACCAGGGTTTGAGCTCAG-3′ as reported previously (16). Thermal cycling conditions for the PCR reactions were 94°C for 5 min followed by 40 cycles of 94°C for 45 s, 61°C for 30 s, and 72°C for 30 s. Polymerase reaction products amplified by these primers (283, 308, and 411 bp for IL-6, β-actin, and TNF-α, respectively) were cloned into pGEMT vector (Promega) and sequenced for verification. Real-time reactions were carried out on an iCycler real-time machine (Bio-Rad, Hercules, CA) using the IQ SYBR Green Supermix kit (Bio-Rad, Hercules, CA). Abundance of each gene product was calculated by regression against the standard curve generated in the same reaction by their respective plasmids. The IL-6 and TNF-α values were normalized against β-actin for each sample.
All data were examined for normality and analyzed using the mixed-model analysis of a split-plot design. The fixed effect was the treatment, and the random effect was the replicate. The main effects (treatment and replicate) were tested against the treatment × replicate interaction term. When protected by a significant F-test, mean separation was accomplished using the least-squares mean separation (pdiff) procedure (22).
Adiponectin attenuates NF-κB activation in adipocytes.
Using a gel shift assay that we have previously validated by supershift analysis (2), we determined that adiponectin downregulates NF-κB activation by LPS in pig adipocytes (Fig. 1A). Whereas there was a clear induction of nuclear translocation in response to LPS, adiponectin attenuated this response. To confirm these results quantitatively, we transfected 3T3-L1 adipocytes with Renilla and firefly luciferase constructs and performed reporter gene assays. First, we validated the assay system by using both the positive and negative control plasmids pFC-MEKK and pLuc-MCS, respectively. As presented in Fig. 1B, luciferase activation was obtained only in the cells transfected with the positive control and the NF-κB (pNF-κB-Luc) plasmids, and it was clear that LPS treatment activated only the cells with the pNF-κB-Luc plasmid (P < 0.05). As shown in Fig. 1C, LPS doubled reporter gene activity (P < 0.001), whereas the MEK inhibitor U0126 blocked this induction, as expected. Consistent with the results in primary pig adipocytes, adiponectin caused a 22% reduction (P < 0.001) in the stimulation of luciferase activity by LPS (Fig. 1C).
Adiponectin suppresses LPS-induced IL-6 and TNF-α mRNA expression and tends to suppress media TNF-α protein level.
We have shown previously (2) that LPS induces IL-6 expression in pig adipocytes and that this increase occurred with substantial accumulation of IL-6 in the media. Consequently, we wanted to determine whether the anti-inflammatory activity of adiponectin encompasses IL-6 in adipocytes. Adiponectin reduced (P < 0.05) the LPS-mediated increase in IL-6 expression in pig adipocytes (Fig. 2A). However, we were unable to detect a reduction in the media IL-6 concentration (data not shown), presumably because of the short duration of incubation that is feasible for primary pig adipocytes. Accordingly, we also tested this hypothesis in 3T3-L1 adipocytes under longer term experimental conditions. Whereas LPS induced the release of IL-6 into the culture media of these cells, adiponectin suppressed (P < 0.05) the magnitude of the response (Fig. 2B). However, under these conditions, adiponectin had no effect (P > 0.05) on IL-6 mRNA abundance (Fig. 2C).
Because TNF-α has been implicated in obesity-related inflammation, we assessed the potential regulatory effect of adiponectin on its expression in endotoxin-treated 3T3-L1 adipocytes. As presented in Fig. 2D, adiponectin suppressed TNF-α mRNA expression (P < 0.05) and tended (P < 0.065) to reduce accumulation of the protein in the media (Fig. 2E). In the presence of the MEK inhibitor (U0126), the endotoxin-induced increase in IL-6 and TNF-α mRNA abundance and media protein content was markedly reduced (P < 0.05; Fig. 2, B–E).
Adiponectin regulates PPARγ2 expression in pig adipocytes.
Finally, we conducted experiments to determine whether LPS or IFN-γ acts directly on pig adipocytes to regulate PPARγ expression and whether adiponectin would alter this response. As indicated in Fig. 3, there was no significant effect of LPS on PPARγ2 expression. However, adiponectin caused an appreciable increase (P < 0.05) in the expression of the PPARγ2 transcript. As with LPS, IFN-γ alone did not alter PPARγ2 expression, but it did block the increase (P < 0.05) caused by adiponectin (Fig. 4).
In this report we provide the first evidence that adiponectin regulates inflammation in the adipocyte itself and that this anti-inflammatory activity includes partial inhibition of the NF-κB transcription factor. This finding is of considerable significance for several reasons. First, in addition to obesity and insulin resistance, chronic inflammation accompanies the prediabetic condition known as the metabolic syndrome (33). Furthermore, recent evidence indicates that obesity is accompanied by a marked infiltration of adipose tissue with macrophages (27, 29) and that transcript profiles in both adipocytes and macrophages reflect active inflammatory states. We have shown adiponectin to suppress both TNF-α and IL-6 production in macrophages (28), and we now extend this role to include the regulation of both cytokines in adipocytes, albeit in different adipocyte models.
In LPS-induced inflammation, the adipocyte is a far more significant source of IL-6 than of TNF-α, as previously demonstrated (2), and the same is true of obesity-related inflammation (27). Consistent with prior findings in pig adipocytes (2), we found a fourfold greater increase in endotoxin-induced media IL-6 accumulation than in TNF-α (200 pg/ml for IL-6 vs. 50 pg/ml; Fig. 2, B and E) in 3T3-L1 adipocytes. Also, adiponectin exerted its suppressive effect on IL-6 largely at the protein level and attenuated the TNF-α response at the mRNA level and tended to attenuate its protein accumulation in the media. The finding that adiponectin suppressed the LPS-induced increase in IL-6 mRNA abundance in pig adipocytes but not in 3T3-L1 adipocytes is somewhat perplexing. Whether this reflects species and/or cell type-dependent variability remains a significant question. Nevertheless, the implication from the collective data is that adiponectin acts as a local regulator of inflammation in adipose tissue, suppressing cytokine production by both adipocytes and macrophages. Thus the factors and underlying mechanisms that regulate adiponectin production and signaling and whether obesity and hyperinsulinemia result in adiponectin resistance, as suggested by Tsuchida et al. (24), are critical questions for obesity and diabetes researchers.
To date, there is little information available as to the signaling pathway used by adiponectin to suppress cytokine production. Using aortic endothelial cells as a model system, Ouchi et al. (19) found that adiponectin caused an accumulation of cAMP and blocked TNF-α signaling in these cells by preventing the activation and translocation of NF-κB. We have documented a similar action of adiponectin previously in pig macrophages (28) and now in pig adipocytes. Because NF-κB is a major regulator of TNF-α and IL-6 expression (9, 25), these data have led us to hypothesize that the anti-inflammatory actions of adiponectin in adipocytes and macrophages are mediated in part by inhibition of NF-κB activation, but this must be confirmed directly.
The upregulation of PPARγ2 by adiponectin in pig adipocytes, as indicated in two separate sets of experiments in the current study, is consistent with results published by other investigators (10) showing that increasing the circulating concentration of adiponectin was associated with increased PPARγ expression in adipose tissue. A growing body of literature shows that the PPAR transcription factors are potent inhibitors of inflammation in immune cells (5, 8, 13, 23). In addition to the control of gene expression, PPARγ has been linked to the regulation of NF-κB through a physical interaction that blocks its transcriptional activity (26). Whether the induction of PPARγ by adiponectin in our study contributed to its suppression of NF-κB activation or regulation of IL-6 is not yet certain. However, the expression of both adiponectin receptors is markedly upregulated in macrophages treated with ligands selective for either PPARα or PPARγ (8). These data, coupled with the anti-inflammatory actions of adiponectin noted above, indicate that a significant portion of the anti-inflammatory activity of these ligands and the PPARs in vivo may be linked to an amplification of the signaling potential for adiponectin via an upregulation of its receptors. This concept is consistent with the fact that the reduction in cholesteryl ester content of macrophages was greater when adiponectin was combined with the PPARα ligand than when either factor was used alone (8).
Finally, the fact that IFN-γ blocked the ability of adiponectin to increase PPARγ expression indicates the complexity of the regulation of pro- and anti-inflammatory pathways in adipocytes. We have thus far not detected an increase in this cytokine in pig adipocytes stimulated with LPS or other proinflammatory factors (2). However, we have shown that IFN-γ does indeed target the adipocyte to increase IL-15 expression (2), and systemic inflammation certainly includes increased circulating concentrations of this cytokine in the pig (6). Thus the ability of adiponectin to regulate PPARγ expression in the adipocyte in vivo may be impacted by IFN-γ.
In summary, we have shown in the present study that adiponectin acts directly on the adipocyte to suppress inflammation and, specifically, that this adipocyte-derived hormone suppresses LPS-induced NF-κB activation and IL-6 and TNF-α production. The apparent reciprocal regulation between adiponectin and TNF-α is particularly intriguing. As established earlier (18), mice in which the adiponectin gene is disrupted express high levels of adipose tissue TNF-α mRNA that are corrected by viral adiponectin expression, and TNF-α treatment significantly reduces the expression and secretion of adiponectin in human adipocytes (17). The present data indicate that adiponectin directly targets the adipocyte to suppress TNF-α mRNA expression and largely support earlier observations. Our findings extend the anti-inflammatory activity of adiponectin to include adipocytes, in addition to macrophages, and may be particularly relevant to obesity-related inflammation in adipose tissue. Furthermore, the direct induction of PPARγ2 by adiponectin in porcine adipocytes indicates the possibility that the PPARγ transcription factor participates in the suppression of NF-κB activation and IL-6 production and that IFN-γ may oppose this action in vivo.
This work was funded in part by grants to M. E. Spurlock from the Purdue Research Foundation and the United States Department of Agriculture, Competitive Grants Program.
This study is Purdue Agriculture Research Program Journal article No. 17473 of the Purdue University Agricultural Experiment Station, West Lafayette.
We are grateful to Drs. Paul Collodi and Shawn Donkin for review of the manuscript and to Dr. Jill Hutchcroft for helpful discussions of the data. We also appreciate the time and efforts of Dick Byrd and the Purdue University farm crew for arranging the transportation of the pigs to the campus facilities.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society