J. J. O. Turner, Elsie Bertram Diabetes Centre, Norfolk and Norwich University Hospital, Norwich NR4 7UY, UK. E-mail: firstname.lastname@example.org
The poor prognosis of obesity is now known to involve a proinflammatory state associated with elevated circulating levels of cytokines and with macrophage infiltration of adipose tissue. In particular, Toll-like receptor (TLR)-4-driven adipose inflammation has been implicated recently in obesity and the development of diabetes. Adipocytes are now recognized as an important source of cytokine and chemokine production, including interleukin (IL)-6 and monocyte chemotractant protein (MCP)-1, and this appears to be a key step in the development of the obesity-associated inflammatory state. Interventions targeted at adipocyte inflammation may therefore form novel therapies to treat or prevent medical complications of obesity. We set out to explore whether anti-inflammatory interventions which are effective in conventional immune cells would operate on primary human cultures of in-vitro differentiated adipocytes. IL-10 was ineffective against TLR-4-induced cytokine secretion due to lack of IL-10 receptor on human adipocytes, in contrast to the widely used murine 3T3-L1 adipocyte model, which is known to respond to IL-10. Adenoviral delivery of an IL-10 receptor construct to the cells restored IL-10 responsiveness as assessed by signal transducer and activator of transcription-3 (STAT-3) phosphorylation. However, the small molecule nuclear factor (NF)-κB inhibitors 2-[(aminocarbonyl)amino]-5-(4-fluorophenyl)-3-thiophenecarboxamide (TPCA)-1 and carbobenzoxyl-Ile-Glu(O-t-butyl)-Ala-leucinal (PSI) as well as adenovirally delivered dominant negative inhibitor of IkappaB kinase 2 (IKK2) and wild-type IκBα were effective inhibitors of TLR-4-driven IL-6 and MCP-1 induction. These data identify a central role for canonical NF-κB signalling in adipocyte cytokine induction and indicate that small molecule inhibitors of NF-κB may form the basis of future treatments for obesity-related conditions where adipocyte inflammatory signalling is implicated.
Obesity is a major health problem associated with a variety of pathologies. It is recognized as a proinflammatory state, and is associated with macrophage infiltration of adipose tissue [1,2]. There are elevated circulating levels of inflammatory markers such as C-reactive protein , cytokines including interleukin (IL)-6 and chemokines, including monocyte chemotractant protein-1 (MCP-1) . Furthermore, the activation of inflammatory signalling pathways associated with obesity is implicated in the pathogenesis of insulin resistance and atherosclerotic vascular disease . Adipocytes are thought to be critical for these manifestations, initiating macrophage infiltration and providing a major source of the increased circulating levels of cytokines . This raises the possibility that inhibiting inflammatory signalling in adipocytes will form the basis of novel future therapies for metabolic syndrome and other obesity-associated conditions. Recent reports have highlighted the potential role of Toll-like receptor (TLR)-4 in obesity-associated adipose inflammation and the development of insulin resistance . Following engagement of lipopolysaccharide (LPS) and the co-factor CD14, TLR-4 signals through the adaptor molecule myeloid differentiation primary response gene 88 (MyD88) to initiate nuclear factor (NF)-κB, p38 mitogen-activated protein kinase (MAPK) and c-JUN amino terminal kinase (JNK)-dependent signalling [9,10]. However, the specific roles of these three pathways in human adipocytes are poorly delineated and the relative contribution of the canonical and alternative NF-κB pathways is thus far completely unexplored.
In ‘professional’ inflammatory cells such as monocyte/macrophages this signalling is inhibited by the potent endogenous anti-inflammatory cytokine IL-10 . Recent evidence shows that murine 3T3-L1 adipocytes express the IL-10 receptor and that IL-10 signals via signal transducer and activator of transcription-3 (STAT-3) to suppress MCP-1 expression in these cells . Furthermore, it has been reported that IL-10 levels are lower in obesity , are related to insulin sensitivity  and IL-10 gene polymorphisms are related to risk of developing type 2 diabetes mellitus (T2DM) .
Given that immortalized cells usually have abnormal chromosomes and aberrant gene expression and in the case of 3T3-L1 are of murine origin, we set out to explore the potential roles of NF-κB inhibition by pharmacological and genetic means and of IL-10 as potential therapeutic strategies for reducing the complications due to adipose inflammation in human primary adipocytes.
Materials and methods
Cell culture reagents used were penicillin–streptomycin, RPMI-1640 and Dulbecco's modified Eagle's medium (DMEM) obtained from Cambrex (Verviers, Belgium) and fetal calf serum (FCS) from Gibco (Paisley, UK). Penicillin–streptomycin–anti-mycotic (PSA) was from PAA (Pasching, Austria). Macrophage colony-stimulating factor (M-CSF) was from Peprotech (London, UK). Chloroform extracted Escherichia coli, serotype EH100 (Ra), ‘rough’ LPS was from Alexis (Farmingdale, NY, USA) and a single batch was used throughout. DNAse type I was from Roche (Mannheim, Germany). Bovine insulin, indomethacin, 3-isobutyl-1-methylxanthine (IBMX), dexamethasone, PSI, 2-[(aminocarbonyl)amino]-5-(4-fluorophenyl)-3-thiophenecarboxamide (TPCA)-1) and Clostridium histolyticum collagenase type 1A were from Sigma (Poole, UK). Human recombinant leptin was from R&D Systems (Minneapolis, MN, USA). Tyrosine-705 phosphospecific STAT-3 antibody was from New England Biolabs (Hitchin, UK). The antibodies to IκBα and inhibitor of IkappaB kinase 2 (IKK2) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and anti-mouse horseradish peroxidase (HRP) from Dako (Glostrup, Denmark). Human recombinant IL-10 was a gift from Schering Plough.
Healthy non-diabetic female subjects undergoing elective breast reconstruction surgery who had a body mass index (BMI) between 25 and 30 kg/m2 were recruited. Informed written consent was obtained and the study was approved by the Hammersmith Hospital Research Ethics Committee (reference number 07/Q0406/29).
Primary human pre-adipocyte cultures were established according to standard techniques . Briefly, subcutaneous adipose samples were transported immediately from the operating theatre and processed under sterile conditions. Adipose tissue was dissected free of blood vessels, skin and connective tissue. After mechanical dispersal to a fragment size of ≤1 mm, they were subjected to enzymatic digestion with collagenase (0·5 mg/ml) and DNAse type I (0·1 mg/ml) for 30–45 min at 37°C and constant shaking at 140 rpm. The digested tissue was then passed through a sterile nylon mesh before centrifuging for 5 min at 500 g. The pellet was resuspended and treated with red cell lysis buffer, washed and plated at a plating density of 9000 cells/cm2 in basal media consisting of 4·5 g/l glucose-containing DMEM supplemented with 10% FCS and 1% PSA. Cells were passaged for a maximum of three passages and plated at or near confluence, in the required plating format for each experiment, prior to differentiation. Differentiation was achieved by changing from basal media to adipogenic media 48 h after the cultures had become confluent. Adipogenic medium was changed every 2–3 days for 14 days prior to experimentation. Adipogenic medium contained basal media supplemented with insulin (100 nM), dexamethasone (1 µM), indomethacin (200 µM) and IBMX (500 µM). For studying induction of cytokines, cells were plated at 104 cells per well in 96-well plates. For studying intracellular signalling by Western blot, cells were plated at 105 cells per well in 12-well plates. Differentiation was assessed by light microscopy to visualize lipid droplets and was routinely >75%. For studying the effects of NF-κB inhibition on cytokine induction, cultures were pretreated with the pharmacological inhibitors at the indicated concentrations for 1 h, or virally infected as described below. The cells were then stimulated with 10 ng/ml LPS for the indicated times, after which the supernatants were collected and cytokines assayed by enzyme-linked immunosorbent assay (ELISA). For studying the effects of IL-10 on cytokine induction, cultures were pretreated with IL-10 at the concentrations indicated (10 ng/ml in the majority of instances) for 1 h prior to addition of LPS at the indicated concentrations and duration, after which supernatants were harvested and cytokines assayed by ELISA.
Primary human monocytes were isolated as described previously  by elutriation from single-donor buffy coat preparations obtained from the North London Blood Transfusion Service. Monocytes were differentiated to macrophages by culturing for 72 h in RPMI-1640 with 5% FCS, 1% penicillin–streptomycin and 10 ng/ml M-CSF.
The concentration of MCP-1 and IL-6 in cell culture supernatants was determined by ELISA (BD Pharmingen, San Diego, CA, USA) following the manufacturer's instructions. Absorbance was read and analysed at 450 nM on a spectrophotometric ELISA plate reader (Labsystems Multiskan Biochromic, Thermo labsystems, Altrincham, UK) using the Ascent version 2.4.2 software. Results are expressed as the mean concentration of triplicates ± standard error of the mean (s.e.m.). Both ELISAs give a lower limit of detection of 60 pg/ml.
Adenovirus infection of adipocytes
Adenovirus infections of differentiated cultures of adipocytes were performed for 2 h at a multiplicity of infection (MOI) of 500:1 in serum-free media, after which cells were returned to serum-containing media and were left to over-express for 48 h prior to commencing experimentation. Successful infection and over-expression was confirmed by visualizing green fluorescent protein (GFP) expression under ultraviolet-microscopy and by Western blot for over expressed protein. Infection efficiency assessed by GFP expression was routinely >80%.
Western blot analysis
Western blotting for pSTAT-3 and total STAT-3 was performed by plating 105 adipocytes per well in 12-well plates or 106 macrophages per well in six-well plates. Cells were serum-starved in media without FCS for 2 h prior to commencing the time–course. Additions (IL-10, final concentration 10 ng/ml and leptin, final concentration 100 ng/ml) were made in minimum volumes of media at the indicated time-points and at the end of the time–course cells were washed with ice-cold phosphate-buffered saline (PBS), lysed in 1% Triton lysis buffer with protease inhibitor cocktail and phosphatase inhibitors added and electrophoresis was performed under reducing and denaturing conditions on 10% polyacrylamide gels.
Replication-deficient control adenoviral vectors without insert (Ad0) were provided by A. Byrnes and M. Wood (University of Oxford, Oxford). Adenovirus encoding GFP was from Quantum Biotech (Quebec, Canada). Adenoviruses encoding IκBα (AdIκBα) and dominant-negative IKK-2 (AdIKK-2dn) were provided by R. De Martin (University of Vienna, Vienna). Adenovirus expressing a murine/human chimeric IL-10 receptor alpha construct was produced as described previously .
Where data from multiple donors have been pooled, they were first normalized and then expressed as mean ± s.e.m. and differences were tested for statistical significance using analysis of variance (anova) with Bonferroni's correction.
Small molecule and adenoviral NF-κB inhibitors attenuate IL-6 and MCP1 induction
LPS strongly induced IL-6 and MCP-1 expression in differentiated cultures of adipocytes. Pretreating the cultures with the IKK2 inhibitor TPCA-1, between 0·3 and 10 µM final concentration resulted in a dose-dependent inhibition of IL-6 and MCP-1 induction. This reached a maximum of 73% ± 2·3% (mean ± s.e.m.) inhibition, P < 0·005 for IL-6 and 85% ± 9·3% inhibition, P < 0·0001 for MCP-1 (Fig. 1). Similar results were observed when the cultures were pretreated with the proteasome inhibitor carbobenzoxyl-Ile-Glu(O-t-butyl)-Ala-leucinal (PSI) at concentrations ranging between 0·2 and 5 µM with respect to MCP-1 induction (83% ± 4·5% inhibition, P < 0·0001 at 5 µM PSI) although, interestingly, this was not observed with respect to IL-6 induction [30% ± 8·2% inhibition, P = not significant (n.s.) at 5 µM PSI] (Fig. 1). In order to confirm independently the role of NF-κB inhibition in attenuation of TLR-4-induced cytokine expression by human adipocytes, we went on to test the role of adenoviral constructs expressing dominant negative IKK2 (IKK2 dn) and wild-type IκBα. Successful viral infection was confirmed by visualizing GFP expression and over-expression was confirmed by Western blot analysis (Fig. 2b,c). IKK2 dn and IκBα constructs produced a >50% inhibition of IL-6 induction (P < 0·005) and >70% inhibition of MCP-1 induction (P < 0·005) (Fig. 2d,e).
IL-10 does not attenuate cytokine induction in in-vitro-differentiated human adipocytes
We next tested the feasibility of inhibiting LPS-induced adipocyte cytokine expression by IL-10. We found that at a fixed concentration and time-point of stimulation (10 ng/ml LPS at 6 h), IL-10 (10 ng/ml) exerted no suppressive effect on IL-6 or MCP-1 induction (Fig. 3a,b) in contrast to the significant inhibitory effect exerted by IL-10 (10 ng/ml) on cytokine induction in macrophages (Fig. 3c). We went on to examine this in more detail across a range of duration (0–48 h) and concentrations of LPS (0–1000 ng/ml) stimulation, but IL-10 (10 ng/ml) produced no significant inhibition of IL-6 (Fig. 3d,e).
IL-10 does not signal in human adipocytes due to absence of IL-10 receptor
Because we had not detected any inhibition of cytokine induction by IL-10 we proceeded to investigate whether IL-10 was able to induce STAT-3 phosphorylation in differentiated adipocytes. No induction of STAT-3 phosphorylation was observed in adipocytes following addition of IL-10 (10 ng/ml) at up to 3 h (Fig. 4a), while monocyte-derived macrophages exhibited a potent STAT-3 phosphorylation response to IL-10 commencing within 10 min of stimulation (Fig. 4b). Used as a positive control, leptin induced STAT-3 phosphorylation in the adipocytes (Fig. 4a). Given the lack of evidence of IL-10 signalling in the adipocytes, we investigated whether or not primary human in vitro-differentiated adipocytes expressed the IL-10 receptor by Western blot analysis; no expression could be detected (data not shown). However, when a full-length murine human chimeric IL-10 receptor  was introduced by adenoviral transfection, IL-10 responsiveness of STAT-3 phosphorylation to IL-10 was observed (Fig. 4c).
Obesity has many impacts on health, and recent progress has established that these are not only mechanical. It is associated with activation of inflammatory signalling in adipose tissue [1,2], and furthermore the evidence suggests that this plays a significant role in the development of obesity-associated pathologies. It is thus probable that targeting inflammatory signalling in adipose tissue will form an important strategy for development of novel therapeutics for the prevention and treatment of conditions such as metabolic syndrome, T2DM and atherosclerotic cardiovascular disease arising in the context of obesity.
Recent reports have highlighted a potential role for TLR-4 signalling in the aetiology of adipose inflammation and insulin resistance in association with obesity , that cytokine induction in adipocytes is NF-κB-dependent [19,20] and that LPS may be a significant ligand [21,22].
What approaches might be used to safely down-regulate NF-κB in adipocytes? IL-10 is a potent anti-inflammatory cytokine that in many circumstances has a central role as an endogenous immunoregulator in limiting NF-κB-driven inflammation . There is some evidence to suggest that IL-10 levels fall in association with obesity . There are also reports of macrophage polarization in adipose tissue in obesity away from the ‘M2’, IL-10-secreting phenotype towards an ‘M1’, classically activated phenotype which is more proinflammatory and secretes less IL-10 . Furthermore, it has been reported recently that murine 3T3-L1 adipocytes are responsive to the anti-inflammatory effects of IL-10 .
We have thus explored the use of NF-κB inhibition and IL-10 as potential therapeutic strategies to attenuate the TLR-4-induced inflammatory signalling in primary human adipocytes.
Our results show that global inhibition of NF-κB (and other pathways) signalling by the proteasome inhibitor PSI inhibits MCP-1 induction by LPS but not IL-6, while the more specific IKK2 inhibitor TPCA-1  potently inhibited MCP-1 and IL-6 induction by LPS. This is, to our knowledge, the first report of TPCA-1 inhibition of NF-κB in adipocytes and identifies adipocyte IKK2 as a potential drugable target for future development of novel treatments for use in obesity.
In order to validate further the importance of the NF-κB pathway in adipocytes, and in particular IKK2 as a therapeutic target for the treatment of adipose inflammation, we went on to use adenovirally delivered IκBα and IKK2 dn constructs. Both these adenoviruses have been used extensively and validated in rheumatoid and atherosclerotic tissue cultures [25,26]. They were well taken up and produced >50% inhibition of IL-6 and MCP-1 induction by LPS. Furthermore, taken together, the TPCA-1 data and IKK2 dn data suggest an important role for the canonical pathway of NF-κB activation in adipocyte inflammation, and thus potentially in complications of obesity.
When we tested the effects of IL-10 on TLR-4 induction of cytokines in in vitro-differentiated adipocytes, we found no evidence of an anti-inflammatory effect. This finding contrasts with those reported in 3T3-L1 adipocytes , but is consistent with many previous reports that human non-myeloid cells do not respond to IL-10 . To our knowledge, this is the first time this has been reported in human adipocytes and appears to highlight an important difference between the widely used 3T3-L1 cell line and human primary adipocytes. Furthermore, IL-10 did not induce STAT-3 phosphorylation in the adipocytes; this was due most probably to non-expression of the IL-10 receptor as judged by Western blot. This was verified in adipocytes by over-expressing an IL-10 receptor alpha construct, following which STAT-3 phosphorylation was induced by IL-10 stimulation, thereby confirming further that absence of IL-10 receptor expression in human adipocytes underlies their lack of IL-10 responsiveness.
Our studies have defined a potential therapeutic strategy to reduce the chronic inflammatory state present in obesity, where adipocytes recruit macrophages to adipose tissue, and the adipocytes and macrophages contribute to the inflammation. Inhibition of NF-κB either by non-specific proteasome inhibitors, or the more specific IKK2 inhibitor TPCA-1, or highly specific dn IKK2, reduced the production of important inflammatory mediators including IL-6, relevant to cardiovascular disease, and MCP-1 is important in macrophage recruitment to many sites, including atherosclerotic tissue. Further work is needed to document the importance of this pathway in vivo.
We are indebted to the patients who kindly gave permission for their tissue to be used in this research. This research is funded by the Arthritis Research Campaign UK, programme grants from the MRC (G7811974) and Wellcome Trust (072643/Z/03/Z) and by an EU FP6 Integrated Project Grant LSHM-CT-2003-503041. We are also grateful for support from the NIHR Biomedical Research Centre funding scheme. J.J.O.T. was a BHF intermediate fellow (FS/06/001).
The authors have no conflicts of interest to declare.