Tumor Necrosis Factor-α Stimulates Lipolysis in Differentiated Human Adipocytes Through Activation of Extracellular Signal-Related Kinase and Elevation of Intracellular cAMP

All chemicals were from Sigma Chemical (St. Louis, MO) unless otherwise indicated. BSA (fraction V) was from Intergen (Purchase, NY), fetal bovine serum (FBS) was from Hyclone (Logan, UT), PD98059 was from Calbiochem (San Diego, CA), and U0126 was from Promega (Madison, WI). Tissue culture reagents were purchased from GIBCOBRL (Rockville, MD) and Hyclone; type I collagenase was from Worthington Biochemical (Lakewood, NJ); enhanced chemiluminescence (ECL) Western blotting detection reagents, horseradish-linked anti-rabbit and anti-mouse IgG were from Amersham Life Sciences (Arlington Heights, IL); [¹²⁵I]-Protein A was from PerkinElmer Life Sciences (Boston, MA); and rosiglitazone (BRL 49653) was kindly provided by GlaxoSmithKline (from Stephen A. Smith and Thomas Leonard; King of Prussia, PA).

Subcutaneous adipose tissues were obtained from male or female patients undergoing abdominal plastic surgery in accordance with the approval of the Tufts New England Medical Center Human Investigation Review Committee. Patients ranged in age from 24 to 64 years and body weight from 65 to 80 kg; all were fasted overnight and underwent surgical anesthesia. None of the patients had diabetes or severe systemic illnesses. Preadipocytes were isolated using the method of Rodbell (29) with modifications. In brief, freshly diced tissue pieces were digested in Hank’s balanced salt solution (HBSS) containing 2 g/l type I collagenase and 4% BSA for 1 h in a 37°C water bath and shaken at 100 cycles/min. The disrupted tissue was filtered through double-layered cotton mesh and centrifuged at 250g for 5 min.

Preadipocytes were resuspended in erythrocyte-lysing buffer (154 mmol/l NH₄Cl, 5.7 mmol/l K₂HPO4, and 0.1 mmol/l EDTA, pH 7.0) and incubated for 10 min to remove contaminating erythrocytes. After centrifugation at 150g for 5 min, preadipocytes were resuspended in Dulbecco’s modified Eagle’s medium (DMEM):F12 medium containing 10% FBS and inoculated into 24-well plates at a density of ∼50,000 cells/cm². After overnight incubation, preadipocytes were washed in HBSS and cultured in DMEM:F12 containing 3% FBS, 100 nmol/l human insulin, 1 μmol/l dexamethasone, and for the first 3 days of culture, 0.25 mmol/l 3-isobutyl-1-methylxanthine and 5 μmol/l rosiglitazone. Medium was changed every 2–3 days, and cells were used after 2–3 weeks of incubation. As judged by microscopy, >80% of the cells were differentiated.

Differentiated human adipocytes from seven healthy female patients (aged 25–60 years, BMI <25 kg/m²) were obtained from Zen-Bio (Research Triangle, NC). Cells cultured by our laboratory and Zen-Bio responded similarly to treatments despite variations in body weight and age of patients. Therefore, data were pooled.

Differentiated human adipocytes were incubated in serum-free DMEM containing 5 mmol/l glucose and 2% BSA for 24 h before treatments. Glycerol content in the incubation medium was used as an index for lipolysis and was measured using a colorimetric assay (GPO-Trinder; Sigma). Results were corrected for cellular proteins, which were quantified using the bicinchoninic acid protein assay kit (Pierce, Rockford, IL) and were expressed as micromoles of glycerol per milligrams of protein or percentage of control.

Cells were washed briefly with PBS and collected with boiling lysis buffer containing 62.5 mmol/l Tris-HCl (pH 6.8), 6% SDS, 0.5 mmol/l Na₃Vo₄, and 0.5 μmol/l dithiothreitol. After boiling at 95°C for 10 min, cells were briefly sonicated. Equal amounts of protein were subjected to ECL-based immunoblotting as described (6). A rabbit polyclonal anti-perilipin antibody was generated using peptide MSMNKGPTLLDGDLPEQENVL and used at 1:3,000. Rabbit antibodies specific for individual Gα subunits Gα_s, Gα_i1, Gα_i2, and Gα_i3 were obtained from Santa Cruz Biotechnology (Santa Cruz, CA) and used at 1:1,000 for immunoblotting of plasma membrane proteins. The expression of PDE3B was examined by subjecting equal amounts of plasma membrane proteins to SDS-PAGE, immunoblotting with a rabbit polyclonal antibody to PDE3B generated using a COOH-terminal peptide (QWLEAEEDNDTESGDDEDG), and as a loading control, immunoblotting with an antibody to Gα_s. Protein expression was detected by ECL and for densitometry by reaction with [¹²⁵I]-protein A. Binding of [¹²⁵I]-protein A to rabbit IgG was detected by autoradiography at −80°C with Kodak XRP film and a DuPont Lightening Plus intensifying screen. Quantification of the autoradiograms was by scanning densitometry.

Activation of ERK kinase was determined by measuring the phosphorylation state of ERK1/2 (30). Aliquots (50 μg protein/sample) of cellular proteins were resolved by SDS-PAGE. ERK1/2 phosphorylation was detected by immunoblotting using a 1:1,000 dilution of rabbit polyclonal phospho-MAP kinase-specific antibody (Cell Signaling Technology, Beverly, MA) with a 1:5,000 dilution of horseradish peroxidase-conjugated anti-rabbit IgG as secondary antibody. Quantitation of ERK 1/2 phosphorylation was performed using ChemiImager 4000 (Alpha Innotech, San Leandro, CA). To assess ERK1/2 expression, an equal amount of proteins (50 μg protein/sample) was used for immunoblotting using a 1:1,000 dilution of rabbit polyclonal anti-ERK1/2 IgG (Cell Signaling Technology).

After 24 h treatment, intracellular cAMP concentrations in differentiated human adipocytes were measured using an enzyme immunoassay kit from Amersham Pharmacia Biotech. Results were corrected for cellular protein content and expressed as femtomoles of cAMP per milligrams of protein.

After 24 h treatment, plasma membranes were prepared as described (31). Equal amounts of proteins were used for Western analyses of G proteins and PDE3B.

Results were expressed as means ± SE. One-way ANOVA was used for statistical analyses. P values <0.05 were considered statistically significant.

All experiments were performed with differentiated human adipocytes. After 2–3 weeks in culture, >80% of the preadipocytes were differentiated as assessed by microscopic lipid accumulation.

To investigate the mechanism(s) by which TNF-α regulates lipolysis in human adipocytes, we first examined the concentration dependency of TNF-α on lipolysis. TNF-α increased lipolysis in a dose-dependent manner beginning at 10 ng/ml. The lipolytic effect was evident only after 6 h, suggesting that this effect does not occur via a direct activation of the lipolytic machinery (Fig. 1). This finding is consistent with that of Hauner et al. (4) in human adipocytes. At 100 ng/ml, TNF-α (at 24 h) increased lipolysis by ∼1.7-fold. Because of the potentially high concentrations of adipose-derived TNF-α in the adipose interstitial fluid in obese subjects (7–10), we used 100 ng/ml in subsequent experiments.

As a first step to determine whether the MAPK pathway is involved in TNF-α-mediated lipolysis, the time course of the effects of TNF-α on ERK1/2 activation was examined. TNF-α treatment increased ERK1/2 phosphorylation, whereas it did not affect the expression of total ERK (Fig. 2). The increase was maximal at 15–30 min, and the effect remained detectable for 24 h.

ERK1/2 activity is regulated by the upstream kinase MEK1/2, which phosphorylates ERK1/2 on threonine and tyrosine residues (32,33). To study the possible role of the ERK1/2 pathway in lipolysis, two specific MEK1/2 inhibitors, PD98059 (34) and U0126 (35), were used to block activation of the MEK/ERK pathway. PD98059 inhibits MEK activation by binding to inactive MEK and preventing the phosphorylation by upstream kinases such as Raf-1. U0126 inhibits the catalytic activity of the active enzyme MEK1/2. As shown in Fig. 3A, treatment with PD98059 (6 and 24 h) inhibited TNF-α-induced and basal ERK1/2 phosphorylation without altering total ERK protein expression. Concomitant with its inhibitory effect on ERK activation, PD98059 treatment decreased lipolysis in TNF-α-treated cells to below control levels, and in non-TNF-α-treated cells (PD98059 alone), to ∼50% of control (Fig. 3B).

Treatment with U0126 resulted in more potent inhibitory effects on ERK1/2 phosphorylation and lipolysis. It decreased lipolysis in both TNF-α-treated and non-TNF-α-treated cells (U0126 alone) to 9.4 and 13.6% of control, respectively (Fig. 3C) and reduced ERK1/2 phosphorylation to an undetectable level, as assessed by Western blotting (Fig. 3D).

TNF-α has been shown to induce lipolysis, in part, by decreasing perilipin expression in 3T3-L1 adipocytes (6,23). To determine whether TNF-α stimulates lipolysis in human adipocytes by a similar mechanism, Western blotting was performed. As shown in Fig. 4A and B, TNF-α treatment did not significantly decrease perilipin expression but caused an electrophoretic shift of perilipin from 65 to 67 kDa, consistent with PKA activation and hyperphosphorylation of perilipin. Coincubation with PD98059 or U0126 diminished this effect of TNF-α and resulted in perilipin migrating as a single 65-kDa band. These results led us to hypothesize that the MEK inhibitors PD98509 and U0126 inhibit TNF-α-induced lipolysis, at least in part, by inhibiting hyperphosphorylation of perilipin.

The rate of lipolysis in adipocytes, in general, is critically dependent on the intracellular concentration of cAMP. To examine the possibility that TNF-α-induced lipolysis could involve the regulation of intracellular cAMP, differentiated human adipocytes were treated in the absence or presence of TNF-α for 24 h. TNF-α treatment resulted in an ∼1.6-fold increase in intracellular cAMP; incubation with PD98059 abolished this effect of TNF-α while decreasing basal intracellular cAMP by ∼26% (Fig. 5A). These effects of TNF-α and PD98059 were consistent with their effects on lipolysis, i.e., stimulation by TNF-α and inhibition by PD98059.

To further verify whether TNF-α-stimulated lipolysis is mediated by activation of cAMP-dependent PKA, cells were incubated in the presence or absence of TNF-α and the PKA inhibitor H89 (20 μmol/l) for 24 h. Treatment with H89 abrogated TNF-α-induced lipolysis (Fig. 5B), and in non-TNF-α-treated cells (H89 alone), it reduced lipolysis to ∼50% of control. Concomitant with this effect, H89 reversed the TNF-α-induced electrophoretic shift of perilipin and caused it to migrate as a single 65-kDa perilipin band, suggesting that this protein was not hyperphosphorylated by PKA in the presence of H89 (Fig. 5C).

In adipocytes, intracellular concentration of cAMP is determined by its synthesis controlled by adenylyl cyclase and its degradation, regulated by the hydrolytic activity of cyclic nucleotide phosphodiesterases (PDEs), especially PDE3B. To determine the mechanism(s) by which TNF-α increases intracellular cAMP in differentiated human adipocytes, Western analyses of G protein expression were performed. In rat adipocytes, TNF-α has been suggested to stimulate lipolysis by decreasing the expression of G_i, which inhibits adenylyl cyclase and, hence, the production of cAMP (31). Treatment of differentiated human adipocytes with TNF-α, however, did not significantly decrease expression of Gα_i1, Gα_i2, and Gα_i3, nor did it have significant effect on Gα_s expression (data not shown). We next examined the expression of PDE3B by Western analyses using Gα_s as a loading control and both ECL- and [¹²⁵I]-protein A-based detection systems (Fig. 6). Densitometry using [¹²⁵I]-protein A detection indicates that TNF-α treatment for 24 h decreased expression of PDE3B by ∼50%. Cotreatment with PD98059 abrogated this decrease, and treatment with PD98059 alone increased PDE3B by ∼30%. Thus, the TNF-α-induced increase in intracellular cAMP is likely to be mediated by downregulation of PDE3B expression, and the regulation is downstream of ERK1/2 activation.

In the present study, we show that TNF-α-increased lipolysis in human adipocytes correlates with ERK activation. The data are the first demonstration to link TNF-α activation of ERK to stimulation of PKA, resulting in increased adipocyte lipolysis.

The MEK1/2 inhibitors PD98059 and U0126 inhibited TNF-α-induced lipolysis. The reduction in lipolysis paralleled the reduction in ERK1/2 activation. In addition, PD98059 and U0126 reduced basal ERK1/2 activation and basal lipolysis, indicating that MEK1/2-ERK1/2 regulate TNF-α-induced as well as basal lipolysis in these cells. In many cells, activation of ERK triggers its translocation into the nucleus, where it phosphorylates and activates transcription factors (36). In turn, these activated transcription factors regulate the expression of specific genes. For example, the ERK pathway has been shown to mediate phosphorylation of peroxisome proliferator-activated receptor-γ, a major transcription factor in adipocytes, and reduce its transcriptional activity (37). In our study, TNF-α required 6 h to stimulate lipolysis, but the peak of ERK activation occurred within 15–30 min. The temporal discrepancy suggests that the mechanism of ERK1/2 action is more complicated than direct phosphorylation of HSL (19) and that alterations in gene and protein expression may be involved. Concomitant with the increase in lipolysis, TNF-α decreased PDE3B protein expression and increased intracellular cAMP. Both effects were restored by cotreatment with the MEK1/2 inhibitor PD98059, indicating that TNF-α may decrease PDE3B expression and thereby increase cAMP concentrations and lipolysis. We propose that TNF-α increases ERK activation within 15–30 min, which subsequently results in the alteration of gene and protein expression over the course of hours, leading to decreased PDE3B protein expression, increased cAMP levels, and ultimately increased lipolysis. Our hypothesis is supported by previous observations that in 3T3L1 adipocytes, treatment of 3T3-L1 adipocytes with TNF-α for 24 h reduced PDE3B mRNA and protein expression and concurrently increased lipolysis (28).

TNF-α has been shown to stimulate lipolysis in isolated rat adipocytes by decreasing the expression of G_i protein (31). In our present study, we did not observe a significant effect of TNF-α on expression of immunoreactive G_i (data not shown). The discrepancy between the two studies may be due to the differential effects of TNF-α in rat versus human adipocytes.

Increases in intracellular cAMP and resulting activation of PKA are the major pathway by which lipolytic agents stimulate lipolysis, which occurs at the surface of the intracellular lipid droplets where perilipins are specifically located (21,38). In our study, treatment with TNF-α resulted in an increase in the molecular weight of perilipin in SDS-PAGE, as evidenced by the electrophoretic shift from 65 to 67 kDa. The increase in perilipin migration was concomitant with the increase in intracellular cAMP and lipolysis. The involvement of cAMP in TNF-α-mediated lipolysis was further evidenced by the fact that H89, a selective and potent cAMP-dependent PKA inhibitor, inhibited TNF-α-induced lipolysis and reversed the TNF-α-induced electrophoretic shift of perilipin, causing it to migrate as a 65-kDa band. In view of the fact that hyperphosphorylation of perilipin by PKA facilitates lipase-mediated lipolysis (24), we speculate that in human adipocytes, TNF-α increases intracellular cAMP levels and PKA activation, resulting in hyperphosphorylation of perilipin and ultimately increased lipolysis. Activated PKA is known to simultaneously phosphorylate perilipin A, and the major lipase in adipocytes, HSL (39–43). Whether, the TNF-α-induced increase of intracellular cAMP in our study leads to phosphorylation of HSL awaits further study.

The TNF-α-induced electrophoretic shift of perilipin and stimulation of lipolysis were abrogated by PD98059 and U0126, suggesting that regulation of the two events is downstream of the activation of the ERK1/2-MEK1/2 pathway. In contrast to the finding of Souza and colleagues (6,23), that TNF-α increased lipolysis in 3T3-L1 adipocytes by decreasing perilipin expression, we did not observe significant reduction of perilipin by TNF-α. The difference between the two studies may reflect the differential regulatory effects of TNF-α on perilipin(s) in the two cell types, with downregulation of perilipin expression in 3T3-L1 adipocytes and the hyperphosphorylation of perilipin in human adipocytes. Both effects can result in reduced perilipin coating of the lipid droplet, leading to increased access of lipase(s) to the lipid droplet surface and increased lipolysis.

Based on our findings, we suggest a model for TNF-α-stimulated lipolysis in human adipocytes, which includes the activation of the MEK1/2-ERK1/2 pathway, resulting in downregulation of PDE3B expression with a subsequent increase in intracellular cAMP and activation of PKA, which hyperphosphorylates perilipin, leading to a modification of the lipid droplet surface and allowing lipase(s) to hydrolyze triacylglycerol. The model may represent only one of the pathways by which TNF-α stimulates lipolysis in human adipocytes, but certainly in our cell culture system, this is a major mechanism.

This work was funded in part by the U.S. Department of Agriculture (USDA), under agreement no. 581950-9-001,DK 50647, by research awards from the American Diabetes Association, and in part by GlaxoSmithKline. Data from this manuscript were presented in part at the 61st Annual Meeting of the American Diabetes Association, Philadelphia, PA, June 2001.

We thank Dr. Cary Cannoun for obtaining human adipose tissue and Dr. Martin Obin for critically reading the manuscript.