Background
Obesity is now known to play a causal role in the complex disease state of metabolic syndrome, as well as being a significant risk factor for cardiovascular disorders and diabetes [
1,
2]. Although once thought to serve as a simple storage depot for excess fats, adipose tissue also regulates organismic metabolism through a variety of signaling mechanisms including autonomic nervous stimulation and secreted hormones [
3,
4]. When in proper balance, these regulatory mechanisms effectively control energy preservation (lipogenesis) during the post-prandial period and energy mobilization (lipolysis) during times of increased energy expenditure.
In addition to these mechanisms of metabolic regulation, adipose tissue is also capable of producing proteins that are classical mediators of the inflammatory response. In the early 1990's, it was discovered that adipocytes synthesize and secrete the pro-inflammatory cytokine, Tumor Necrosis Factor-alpha (TNF-α) [
5]. Since then, it has been shown that a number of acute phase reactants and inflammatory mediators are made by adipocytes including plasminogen activator inhibitor-1, IL-1β, IL-6, IL-8, IL-10, IL-15, hepatocyte growth/scatter factor and prostaglandin E
2 (PGE
2) [
6]. In fact, enough of these factors are secreted by adipocytes that overall systemic levels are significantly elevated in obese subjects [
7] and a number of studies have now identified a direct correlation between body mass index (BMI) and systemic levels of inflammatory proteins [
8]. These clinical observations provide key evidence linking obesity with cardiovascular disorders and begin to shed light on how low-level, chronic inflammation adversely affects cardiovascular function in obese subjects.
Recent evidence suggests that cytokine expression in adipose tissue is initiated by crosstalk occurring between adipocytes and macrophages [
8‐
11]. Macrophages typically account for 5–10% of cells within adipose tissue obtained from non-obese donors; however, in diet-induced obesity, macrophage infiltration can account for up to 60% of all cells in adipose tissue [
12]. Cytokines secreted by macrophages, including TNFα, IL-1β and IL-6, are known to stimulate cytokine expression in adipocytes [
13‐
15] and establish a paracrine loop between these two cell types [
16]. This paracrine stimulation in turn elevates systemic cytokine levels observed in obese individuals. In
bone fide inflammatory cells, cytokine gene expression is activated following activation of the Nuclear Factor-kappaB (NF-κB) signal transduction pathway [
17]. Activation of the NF-κB pathway is mediated by a variety of signals including those initiated from the TNFα receptor and Toll-like receptor family. NF-κB itself is a heterodimeric transcription factor that is retained in the cytosol in its inactive state by complexing with a set of inhibitory proteins designated IκB. Upon receptor activation of NF-κB signaling the IκB complex is phosphorylated by IκB kinase (IKK). This in turn leads to its dissociation from NF-κB and rapid degradation by the proteosome. Free NF-κB is then able to translocate to the nucleus where it binds to specific promoter elements resulting in the activation of a battery of genes, including those encoding for inflammatory proteins.
In adipocytes, both expression and activity of NF-κB increase during differentiation [
18] suggesting that it is a key player in mediating adipose-specific cytokine expression. Moreover, excessive NF-κB activity has been associated with the development of type 2 diabetes as obese individuals have high circulating levels of TNF-α, IL-1β and IL-6 that, like cardiovascular risk, directly correlate with insulin resistance [
7,
19,
20]. Collectively, these observations suggest that therapeutic targeting of the NF-κB signaling pathway in adipose tissue represents a logical pursuit to reduce systemic cytokine levels and reverse their negative influence on cardiovascular function and diabetic progression.
There are no shortages of reported inhibitors of NF-κB activation as they now number in the hundreds [
21]. Unfortunately, the targets for many of these inhibitors are not necessarily restricted to the NF-κB pathway raising concerns of potential side effects due to off-target inhibition. As an alternative, we have investigated the effectiveness of certain natural products to inhibit the NF-κB signaling pathway [
22]. Many natural products have been shown to possess low level toxicity and potent anti-inflammatory properties by targeting similar pathways as non-steroidal anti-inflammatory drugs (NSAIDs). Two natural polyphenols of particular interest are curcumin and resveratrol. These natural products are modest inhibitors of NF-κB activation and inflammatory gene expression [
22‐
30] and have proven safe in human clinical trials [
31‐
33]. In the present study, we examined if curcumin and resveratrol are also able to inhibit NF-κB activation in adipocytes and in doing so inhibit cytokine expression in these cells. We believe that using natural products to inhibit the chronic inflammatory response of adipose tissue may provide a novel approach to reduce systemic cytokine levels which in turn is expected to improve cardiovascular health and insulin sensitivity.
Methods
Reagents
TNFα was obtained by R&D Systems, Minneapolis, MN and was used in all experiments at a final concentration of 20 ng/ml. Lipopolysaccharide (LPS) was purchased from Sigma, St. Louis, MO and used to activate BV-2 murine macrophages at a final concentration of 20 μg/ml. Curcumin was synthesized in the lab [
34] and resveratrol was purchased from A.G. Scientific Inc., San Diego, CA.
Cell culture and adipocyte differentiation
For our studies, we utilize an
in vitro cell culture system that has been extensively characterized for adipocyte differentiation, namely mouse 3T3-L1 fibroblasts [
35]. Following induction into the differentiation pathway, 3T3-L1 cells undergo growth arrest, become spherical, and form large intracellular lipid droplets. Subcutaneous implantation of these cells in mice results in tissue masses that are histologically indistinguishable from white adipose tissue [
36,
37]. 3T3-L1 cells were obtained from American Type Culture Collection (Manassas, VA) and grown in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen, Carlsbad, CA) supplemented with 10% (v/v) fetal calf serum (Irvine Scientific, Santa Ana, CA), 1 mM sodium pyruvate, 0.1 mM non-essential amino acids, 2 mM L-glutamine, 100 μg/ml streptomycin sulfate, and 100 units/ml penicillin. Cells were cultured at 37°C with 10% CO
2 and passaged twice weekly. To differentiate 3T3-L1 cells into adipocytes, cells were incubated with 250 nM dexamethasone, 450 μM 3-isobutyl-1-methylxanthine, and 167 nM insulin for 2 days, followed by 167 nM insulin for an additional 3 days.
BV-2 murine macrophages (a gift from Dr. Paul Stemmer, Wayne State University, Detroit, MI) were grown in RPMI-1640 (Hyclone®, Logan, UT) supplemented with 10% (v/v) fetal calf serum, 1 mM sodium pyruvate, 2 mM L-glutamine, 100 μg/ml streptomycin sulfate, and 100 units/ml penicillin. Cells were cultured at 37°C with 5% CO2 and passaged twice weekly.
qRT-PCR and RT-PCR analysis
Total RNA was purified from cells using RNeasy (Qiagen, Valencia, CA) and converted to cDNA using TaqMan
® Reverse Transcriptase (Applied Biosystems, Branchburg, NJ). Cyclooxygenase-2 (COX-2), IL-1β, IL-6, TNFα, and β-actin expression levels were measured by quantitative Real-Time PCR analysis (qRT-PCR) of cDNA samples. Gene and primer information can be found in Table
1. Amplification of leptin and macrophage specific markers F4/80 and Mac-1 was performed by reverse transciptase-PCR (RT-PCR).
Table 1
Gene and primer information used in this study.
COX-2 | NM_011198 | TGGGGTGATGAGCAACTATT AAGGAGCTCTGGGTCAAACT | 7 | 132 |
IL-1β | NM_008361 | GACCTTCCAGGATGAGGACA AGCTCATATGGGTCCGACAG | 3 | 183 |
IL-6 | NM_031168 | AGTTGCCTTCTTGGGACTGA CAGAATTGCCATTGCACAAC | 2 | 191 |
TNFα | NM_013693 | ACGGCATGGATCTCAAAGAC GTGGGTGAGGAGCACGTAGT | none | 116 |
β-actin | NM_007393 | CCTGAACCCTAAGGCCAACC CAGCTGTGGTGGTGAAGCTG | 3 | 287 |
leptin | NM_008493 | TGACACCAAAACCCTCATCA TCATTGGCTATCTGCAGCAC | 2 | 213 |
F4/80 | NM_010130 | GCTGTGAGATTGTGGAAGCA CTGTACCCACATGGCTGATG | 15 | 135 |
Mac-1 | NM_008401 | AAGGATTCAGCAAGCCAGAA TAGCAGGAAAGATGGGATGG | 20 | 136 |
qRT-PCR was performed using ABsolute QPCR SYBR Green Mix (Fisher Scientific, Atlanta, GA) with the following cycling parameters: 1 cycle, 95°C, 15 min; 40 cycles, 95°C, 15 sec, 63°C, 1 min. Changes in gene expression were determined by the Comparative CT method. Since β-actin gene expression is unaffected by TNFα treatment, β-actin mRNA levels were quantified in each sample using identical cycling conditions and used to normalize values obtained for COX-2, IL-1β, IL-6, and TNFα expression. Amplified products were separated on 3% agarose gels and stained with Gel Star® (Cambrex, Rockland, ME).
Immunoblotting
Cell lysates were prepared using 1× Laemmli sample buffer (Sigma-Aldrich). After heating samples at 95°C for 10 min, they were vortexed on high for 20 s to shear DNA and reduce viscosity. Proteins were then separated by SDS-PAGE and transferred to PVDF membrane (0.2 μm, BioRad, Hercules, CA) using a wet tank transfer system (BioRad). Membranes were blocked with 20 mM Tris, pH 7.4, 150 mM NaCl (TBS) containing 0.1% (v/v) Tween-20, 5% (v/v) calf serum for 30 minutes at 23°C and incubated with either anti-IκB monoclonal antibody (2 μg/ml, Imgenex, San Diego, CA) or anti-β-actin monoclonal antibody (1:500, no. A-4700, Sigma-Aldrich) for 24 h at 23°C. Membranes were washed three times (10 min each) with TBS, 0.1% (v/v) Tween-20, and bound antibodies were detected with goat anti-mouse HRP-conjugated secondary antibody (1:3000, BioRad) followed by chemiluminescence detection with Immobilon™ Western according to the manufacturer's instructions (Millipore, Billerica, MA). Images were captured using a Syngene GeneGnome system equipped with a Peltier-cooled 16-bit CCD camera and saturation detection. Densitometry was performed using ImageJ software (version 1.37; National Institutes of Health,
http://rsb.info.nih.gov/ij/).
NF-κB nuclear localization assay
BV-2 murine macrophages were cultured in the absence or presence of LPS (20 μg/ml) for 24 h. 3T3-L1-derived adipocytes were cultured in the absence or presence of TNFα (20 ng/ml) or incubated with TNFα together with curcumin or resveratrol or vehicle alone (dimethylsulfoxide at 0.1% final concentration) for 62 h. Nuclear localized NF-κB was quantified using a Transcription Factor ELISA Kit to detect activated p65 subunit of NF-κB (Panomics, Fremont, CA). All reagents required for preparing nuclear extracts and performing ELISA assays were included and their use was described by the manufacturer.
Cytotoxicity assay
Cells were grown in 96-well plates to 80–90% confluency. Media was replaced with fresh complete media containing the indicated concentrations of curcumin or resveratrol, or vehicle alone (dimethylsulfoxide at 0.1% final concentration). After a 24 h incubation, WST-1 (Roche Molecular Biochemicals, Indianapolis, IN) was added to the cultures to a final concentration of 10% (vol/vol). Following an additional incubation at 37°C for 60 min, absorbance was recorded for each well (450 nm; reference wavelength, 690 nm).
Cytokine and PGE2 ELISA
Quantitation of cytokine protein levels from cell culture supernatants was done by ELISA Ready-SET-Go! kit (eBioscience, San Diego, CA) per manufacturer's instructions. Parameter™ PGE2 competitive binding ELISA kit (R&D systems, Minneapolis, MN) was used to measure PGE2 levels.
Statistical analyses
All experimental protocols were done in at least triplicate points and error bars represent standard deviations of mean values. Student's t-test was performed on some figures using data sets composed of a minimum of triplicate values. Comparison of data sets resulting in p values < 0.05 were considered statistically significant.
Conclusion
Increased adiposity is now a well established risk factor for developing complications related to metabolic syndrome and type II diabetes mellitus. Mounting evidence indicates that low level, chronic inflammation resulting from cytokines secreted by adipose tissue may play a significant role in causing, or at the very least aggravating, the inflammatory component of cardiovascular disease and in desensitizing cells to insulin leading to high circulating glucose levels. These observations suggest a hypothesis that reducing or preventing the inflammatory properties of adipose tissue represents a novel and promising therapeutic approach to curb the progression of cardiovascular disease and to restore insulin sensitivity in type II diabetics.
Macrophage infiltration has recently been postulated to be a primary stimulus for fueling the inflammatory properties of adipose tissue [
44]. Monocyte chemoattractants [
9,
11], such as monocyte chemoattractant protein-1 (MCP-1) which is synthesized and secreted by adipocytes [
45], are thought to mediate macrophage infiltration and intensify macrophage expression of TNFα [
8]. TNFα has pleiotropic effects on adipocyte physiology including an induction of lipolysis to increase the mobilization of free fatty acids [
46,
47], activating cytokine expression [
8] and promoting insulin resistance [
5,
48]. Observations such as these provide sufficient evidence suggesting that TNFα is the predominant factor that mediates the crosstalk between macrophages and adipocytes and that elevated TNFα levels found in obese individuals establishes a paracrine loop in adipose tissue [
16] that is responsible for the elevated systemic levels of cytokines seen in obesity.
TNFα mediates its affects on adipocytes by activating the NF-κB signaling pathway [
11,
18]; a signaling event that has been studied extensively in the innate immune response. In conventional immune cells, activation of the NF-κB signaling pathway requires relocation of the NF-κB heterodimer from the cytoplasm to the nucleus where it functions as part of a multi-protein transcription complex controlling the expression of most inflammatory mediators. In adipose tissue, low level NF-κB activation has been identified in vivo [
49] suggesting that, like in conventional immune cells, NF-κB is largely responsible for cytokine gene expression in adipocytes. Only recently has the role of NF-κB in adipose function come under scrutiny. Berg, et al., examined NF-κB expression and activity during adipocyte differentiation and found both parameters to be elevated in fully differentiated adipocytes [
18]. Consistent with their findings, we were able to activate NF-κB signaling in differentiated adipocytes with TNFα treatment and in doing so demonstrate an increase in NF-κB nuclear translocation. However, to extend these observations we also examined the upstream signaling event that is directly responsible for NF-κB activation, namely IκB degradation. We found that IκB was rapidly degraded in adipocytes following TNFα treatment and with kinetics similar to those measured for true immune cells [
41]. These data provide compelling evidence that NF-κB signaling in adipocytes shares a similar time course of activation as inflammatory cells.
Because the NF-κB signaling pathway is such a pleiotropic pro-inflammatory and pro-survival factor in a wide range of disorders, it has been an attractive target for small-molecule inhibitor development. Thus far almost 800 compounds have been reported to inhibit NF-κB activation [
21]. A large fraction of these inhibitors include natural products that are capable of targeting multiple checkpoints in the NF-κB activation pathway. Of particular interest are the polyphenolic natural compounds, curcumin and resveratrol. Curcumin is derived from the spice turmeric, which comes from the root of
Curcuma longa of the ginger family. It is an established inhibitor of NF-κB activation [
29] and has recently been shown to specifically target IKK [
24]. Inhibitors targeting IKK have so far proven to be the most effective compounds for preventing the activation of NF-κB [
50‐
54] by directly preventing the phosphorylation of IκB, and as a consequence, block NF-κB translocation to the nucleus. Important for clinical drug development, curcumin has also been found safe in six human trials at oral doses up to 8 g/day administered for 3 months [
31,
32]. The other natural product that has been a focus of our laboratory is resveratrol [
22]. A product of red grapes, resveratrol possesses multiple biological activities including anti-oxidant and anti-cancer activities, and like curcumin, is an inhibitor of NF-κB activation [
55] through targeted inhibition of IKK [
56]. In addition, although the extent of its bioavailability is still under investigation [
57,
58], resveratrol has been shown to be quite safe in preclinical trials [
33]. In the present study, we examined if curcumin and resveratrol might represent promising therapeutics to combat the chronic inflammatory properties of adipose tissue by exploring their effects on NF-κB activation and inflammatory cytokine expression in adipocytes. We first identified that curcumin and resveratrol are able to inhibit NF-κB translocation to the nucleus in TNFα-stimulated adipocytes. Moreover, we also found that both natural products are able to prevent IκB degradation. These data establish that curcumin and resveratrol carry out their inhibitory functions either at the level of IκB phosphorylation by IKK or upstream from this checkpoint in the NF-κB activation pathway. We next examined the effects of curcumin and resveratrol on downstream gene regulation in adipocytes since NF-κB activation is largely responsible for mediating inflammatory gene expression in immune cells. Indeed, treatment of TNFα-stimulated adipocytes with curcumin or resveratrol resulted in a significant reduction in TNFα, IL-1β, IL-6, and COX-2 gene expression. The IC
50 values measured for inhibition of IL-1β, IL-6, and COX-2 gene expression by either compound were found to be < 2 μM; for TNFα gene expression, the IC
50 value was ~8 μM.
During the course of identifying inhibitors for NF-κB signaling, many studies will limit their analysis to measuring the effects of inhibitors on the transcriptional status of cytokine genes. Although these studies provide a wealth of data regarding the direct control of cytokine gene expression by the state of NF-κB activity, they fall short of identifying additional mechanisms of regulation at post-transcriptional levels. Limiting inhibitor identification to effects on transcriptional levels in bona fide immune cells may be acceptable since the NF-κB signaling pathway that mediates these immunological responses has been well studied [
59]. However, because much less is known about potential multi-level regulatory elements in non-immune cells that may affect NF-κB signaling, cytokine expression analyses should include a quantitative assessment of secreted cytokines to identify possible post-transcriptional control of cytokine expression. By extending our analysis to measuring levels of secreted cytokines, we have identified unique expression patterns that may have significant impact on our understanding of adipocyte contributions to systemic inflammation. First, although adipocytes express TNFα mRNA, we were unable to measure any secreted TNFα by ELISA. This observation suggests that the major source of circulating TNFα found in obese subjects arises from adipose-infiltrating macrophages rather than adipocytes. A similar observation was made by Fain and colleagues when comparing isolated adipocytes to stromal vascular cells obtained from human adipose explants [
6]. In this study the authors found significant amounts of TNFα secreted by stromal vascular cells, with little or no detectable TNFα secreted by adipocytes. One caveat of this study stems from the fact that the adipocytes were removed from the in vivo environment where they are exposed to macrophage-derived TNFα. Removal of TNFα-stimulation from the isolated adipocytes would discontinue signaling events that arguably might be necessary to sustain TNFα secretion by adipocytes. Our study clearly addresses this concern by demonstrating the lack of TNFα secretion in TNFα-stimulated adipocytes.
We also found that preadipocytes express the gene for IL-1β, yet differentiated adipocytes show no mRNA expression. Interestingly, TNFα treatment was able to re-activate IL-1β mRNA expression in differentiated adipocytes; however, in spite of this re-activation we were unable to detect any secreted IL-1β from treated adipocytes indicating that post-transcriptional mechanisms are in place to prevent expression of IL-1β protein. These observations may be interpreted based on the effects of long-term treatment of adipocytes with IL-1β. Such treatment has been shown to inhibit insulin receptor substrate -1 (IRS-1) expression [
60] and activation [
61] thereby inducing insulin resistance. By repressing IL-1β transcription during adipocyte differentiation, insulin responsiveness can be maintained for proper glucose homeostasis. Furthermore, because expansion of adipose tissue is accompanied by accelerated macrophage infiltration providing a substantial source of secreted TNFα, which we show can activate IL-1β gene expression, additional levels of regulation become necessary to prevent secretion of IL-1β protein by adipocytes. Collectively, these observations indicate that multiple regulatory checkpoints are in place to prevent IL-1β expression and ensure proper insulin responsiveness by adipocytes.
In contrast to the results obtained for measurements of secreted TNFα and IL-1β, TNFα-stimulation of adipocytes did have a pronounced effect on secreted levels of IL-6 and PGE2. We found little or no IL-6 secreted by unstimulated, fully differentiated adipocytes; however, when stimulated with TNFα, a significant level of secreted IL-6 was measured. In spite of a lack of secreted IL-6, we found that the IL-6 gene is expressed in unstimulated adipocytes and is responsive to TNFα stimulation as mRNA levels increased by 6-fold. These data indicate that TNFα stimulation of adipocytes not only increases transcriptional activity of the IL-6 gene, but also activates post-transcriptional events to produce secreted IL-6. Secreted PGE2 levels were also measured as a direct assessment of COX-2 activity. We found that TNFα-stimulation modestly increased COX-2 gene expression by 2-fold and increased secreted PGE2 by 3-fold over basal levels found in unstimulated adipocytes. Notably, both curcumin and resveratrol treatment of TNFα-stimulated adipocytes significantly reduced secreted levels of IL-6 and PGE2 in a dose-dependent manner. IC50 values for curcumin and resveratrol inhibition of IL-6 are estimated to be ~20 μM. By contrast, IC50 values for inhibition of PGE2 differ for each compound; ~2 μM for curcumin and > 20 μM for resveratrol. These IC50 values determined for secreted levels of IL-6 and PGE2 are noticeably higher than what was measured for inhibition of IL-6 and COX-2 gene expression. These differences are most likely due to previously unidentified effects of curcumin and resveratrol on post-transcriptional events and highlight the importance of measuring the final product in addition to transcriptional levels when identifying the quantitative effects of potential inhibitory compounds.
Whenever a compound is being developed as a potential therapeutic, issues involving in vivo bioavailability must be addressed. In this regard, data thus far presented on the pharmacokinetics of curcumin [
62,
63] and resveratrol [
64] have been confusing and often times contradictory. Both polyphenols have relatively short half-lives in vivo as they are rapidly metabolized to their glucuronide and sulfated forms. These metabolites, readily found in the circulation, typically demonstrate very low cell permeability and questionable bioactivity when compared to their unmetabolized forms. In spite of these hurdles, the in vivo efficacies of curcumin and resveratrol have been reproducibly shown by numerous investigators. Many challenges lie ahead in order to systematically and quantitatively address the pharmacokinetics of these natural products. Immediate questions that need to be addressed to improve on in vivo efficacy include, 1) do the metabolites of curcumin and resveratrol have comparable bioactivity with the parent compounds, 2) does the circulating pool of metabolites represent a source of inhibitor that can be modified to their more active forms, and 3) can chemical substitutions be made to the base structures of curcumin and resveratrol making them more active and less susceptible to conjugation.
Most importantly for our hypothesis, the results presented here provide proof-of-principle evidence that use of curcumin and resveratrol represents a promising new therapeutic approach to reduce both local and systemic inflammatory contributions by adipose tissue. At present, we believe that the low μM IC
50 values of curcumin and resveratrol together with their positive in vivo effects make these natural products excellent lead compounds to guide the development of more potent inhibitors of NF-κB activation and inflammatory gene expression. Toward this goal, we have recently developed chemical libraries of synthetic analogs based on the chemical structures of curcumin [
23] and resveratrol [
22]. Studies are currently underway to identify if these novel structural analogs improve upon the inhibitory properties of the parent compounds while also critically addressing the challenges of bioavailability and in vivo metabolism.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AMG carried out the experiments in this study and participated in the experimental design, RAO provided the original conceptual framework for the study, assisted with the experimental design and finalized the manuscript for submission. All authors read and approved the final version.