Tumor necrosis factor-α (TNFα) can decrease adipose tissue mass, but in obesity, adipose tissue hypertrophy persists despite increased TNFα expression. The hormonal milieu of obesity may antagonize the adipostat effects of TNFα. We examined the effects of insulin and the synthetic glucocorticoid, dexamethasone (Dex), on TNFα-induced apoptosis and gene expression in human adipocytes and preadipocytes. Using RT multiplex PCR, the expression of the proapoptotic genes interleukin-1β (IL-1β)-converting enzyme (ICE) and TNFα and the antiapoptotic genes bcl-2, nuclear factor-κB (NFκB), and NFκB inhibitory subunit, IκB, were examined. The expression and release of IL-1β, a postulated downstream effector of ICE-mediated apoptosis, were also determined. TNFα increased the messenger ribonucleic acid levels of ICE, TNFα, IL-1β, bcl-2, and NFκB in preadipocytes and adipocytes (P < 0.01). Dex inhibited TNFα-induced messenger ribonucleic acid expression of ICE, TNFα, and IL-1β (P < 0.01), but not that of bcl-2 and NFκB. TNFα stimulated IL-1β release from preadipocytes and adipocytes up to 20-fold, but the effect was abrogated by Dex. Apoptosis induced by TNFα was reduced to control levels (P < 0.01) by Dex. Insulin had no significant effect on TNFα-induced apoptosis and gene expression. In obesity, glucocorticoids may reduce TNFα actions in adipose tissue by inhibiting TNFα-induced apoptosis, IL-1β release, and TNFα expression.
TUMOR NECROSIS factor-α (TNFα) could reduce adipose tissue mass due to its ability to inhibit adipocyte differentiation and lipogenesis (1), stimulate lipolysis (2), and induce apoptosis (3). Circulating levels and adipose tissue production of TNFα are increased in obesity (4), but despite this, adipose tissue hypertrophy persists, suggesting that the hormonal milieu of obesity may modify the adipostat effects of TNFα and lead to TNFα resistance (5). Insulin and glucocorticoids are known to promote adipogenesis through inducing adipocyte differentiation and increasing lipogenesis. Insulin excess is associated with weight gain, and elevated glucocorticoid levels, found in Cushing’s syndrome and in experimental animal models, are associated with obesity. TNFα up-regulates the enzyme 11β-hydroxysteroid dehydrogenase enzyme 1 that is responsible for the generation of cortisol from inactive cortisone (6), an enzyme present in fat tissue and elevated in central obesity (7). We hypothesized that in obesity glucocorticoids and insulin antagonize the apoptotic effects of TNFα, leading to reduced control of adipose tissue mass.
In human preadipocytes and mature adipocytes, two TNF receptors (p55 and p80) exist. Long-term inhibition of insulin-stimulated glucose transport and antiadipogenic effects are mediated through down-regulation of the glucose transporter GLUT4 via the p55 TNF receptor (8). TNFα is known to induce apoptosis by activating apoptosis executors, such as interleukin-1β (IL-1β)-converting enzyme (ICE) and ICE-like cysteine proteases, referred to as caspases. ICE cleaves the 33-kDa pro-IL-1β into the 17.5-kDa, biologically active IL-1β (9). Overexpression of ICE causes apoptosis (10, 11), whereas mutations involving the active site for catalytic activity (10) or antisense treatment (12) eliminate the ability of ICE to induce apoptosis. Treatment with an ICE-specific inhibitor (13) or the product of crmA, a cowpox virus- encoded cytokine response modifier gene that inhibits ICE and ICE-related proteases (13–15), inhibits TNFα-induced apoptosis. IL-1β, which is produced by macrophages and other antigen-presenting cell types, plays a regulatory role in ICE-mediated apoptosis (16). Pro-IL-1β accumulates in the cell cytosol (17), and mature IL-1β is released during apoptosis after cleavage by ICE (14, 16, 18). When released, IL-1β mediates TNFα-induced apoptosis (16).
TNFα undermines its own killing powers by inducing the expression of antiapoptotic mediators, such as bcl-2 (19) and nuclear factor-κB (NFκB) (20). bcl-2 inhibits apoptosis in response to many different death-inducing signals, such as overexpression of ICE (10), and has been shown to activate transcription factor NFκB through degradation of the cytoplasmic inhibitor NFκB inhibitory subunit (IκB) (21). NFκB prevents TNFα-induced apoptosis (22, 23), and knockout mice missing NFκB die before birth, apparently of massive apoptosis of liver cells (22). In immune cells TNFα expression is increased by lipopolysaccharide and inhibited by glucocorticoids (24, 25). Similarly, in adipose tissue, TNFα release is stimulated by lipopolysaccharide (26), but the effects of glucocorticoids and other hormonal factors, such as insulin, on the expression of this cytokine are unknown.
Glucocorticoids and insulin regulate apoptosis in a cell-specific manner. Dexamethasone (Dex), a synthetic glucocorticoid, inhibits TNFα-induced cytotoxicity/apoptosis in TA1 preadipocytes (27), mouse L-929 cells, human mammary carcinoma MCF-7 cells, and murine tumorigenic fibroblasts L-M cells (28–31), but in mouse lymphoma WEHI7.2 cells (32), human leukemic 6TG1.1 T cells (33), and human osteosarcoma U20S cells (34), Dex induces apoptosis. Insulin has been shown to rescue many cells from apoptosis, but in myeloma U266 cells, insulin increases apoptosis by activating caspase-3 (35).
To elucidate the mechanisms by which Dex and insulin may regulate TNFα effects, we examined DEX or insulin-mediated changes in TNFα-induced messenger ribonucleic acid (mRNA) expression of ICE, TNFα, bcl-2, NFκB, IκB, and IL-1β. The release of IL-1β into the medium after TNFα treatment and the effects of DEX and insulin on this release were studied. We used both preadipocytes and mature adipocytes from sc tissue for this study. Adipose tissue is comprised of a heterogeneous population of preadipocytes and mature fat cells with large lipid stores, both of which express receptors for TNFα.
Subjects and Methods
Human subjects
Subcutaneous adipose tissues were obtained from 16 patients undergoing either elective abdominal surgery (n = 5) or cosmetic abdominal liposuction (n = 11) in accordance with the guidelines of the local ethical committee. Twelve women (age, 27–64 yr; body weight, 65.2–82.0 kg) and four men (age, 35–72 yr; body weight, 76.1–86.7 kg) were fasted for at least 6 h preoperatively, and all underwent general anesthesia. None of the patients had diabetes or severe systemic illness, and none was taking medications known to influence adipose tissue mass, distribution, or metabolism. Variations in body weight, age, and sex did not influence the general conclusions reached, so the data were pooled.
Adipocyte culture
Adipose tissues were transferred to the laboratory within 1 h of removal and used immediately. Mature adipocytes (2 × 106 cells), prepared as described previously (36), were cultured in suspension in 10 mL DMEM/Ham’s F-12 with or without varying concentrations (0.3–10 nmol/L) of recombinant human TNFα (PeproTech, London, UK), Dex (10−11–10−6 mol/L), and insulin (10−10–10−6 mol/L) for the indicated periods of time (1–48 h). Cells were used for total RNA extraction. Conditioned media were stored at −80 C and used for analysis of IL-1β release by enzyme-linked immunosorbent assay (ELISA). To study adipocyte apoptosis, isolated mature adipocytes were cultured using a modified ceiling culture method as previously described (36). Adipocytes were then incubated for 24 h in serum-free DMEM/Ham’s F-12 with or without 6 nmol/L recombinant human TNFα in the presence or absence of 10−8 and 10−7 mol/L Dex and 10−7 mol/L insulin.
Preadipocyte culture
Preadipocytes that formed pellets after centrifugation at 250 × g (36) were resuspended in red blood cell lysing buffer (154 mmol/L NH4Cl, 5.7 mmol/L K2HPO4, and 0.1 mmol/L ethylenediamine tetraacetate, pH 7.0) and incubated for 10 min. Cells were centrifuged at 150 × g for 5 min and resuspended in DMEM/Ham’s F-12 medium supplemented with 15% bovine FCS (First Link UK Ltd., Brierley Hill, UK). Cells (2.5 × 104/cm2) were cultured for 72 h in six-well plates to reach confluence and were incubated (1–24 h) in serum-free DMEM/Ham’s F-12 medium with or without varying concentrations of recombinant human TNFα (0.3–10 nmol/L), Dex (10−11–10−7 mol/L), and insulin (10−10–10−6 mol/L). Cells were used for total RNA extraction. Conditioned media were stored at −80 C and used for measurement of IL-1β by ELISA. For apoptosis studies, preadipocytes were cultured on 20 × 20-mm coverslips in six-well plates and treated with or without 6 nmol/L TNFα in the presence or absence of Dex (10−8–10−7 mol/L) and insulin (10−7 mol/L).
RNA extraction
Total RNA was extracted from cultured preadipocytes and adipocytes using TRIzol reagent (Life Technologies, Inc., Gaithersburg, MD) according to the manufacturer’s protocol. RNA was treated with RQ-1 deoxyribonuclease (Promega Corp., Madison, WI) and quantified by absorbance at 260 nm in a spectrophotometer (Ultrospec III, Pharmacia LKB, Piscataway, NJ). The integrity of the RNA was verified by ethidium bromide staining of ribosomal RNA (rRNA) bands on a 1% agarose gel.
Analysis of TNFα-induced gene expression by RT-multiplex PCR (RT-MPCR)
RT-MPCR amplifies multiple genes in a single PCR reaction. Therefore, variations in RNA quality, initial quantitation errors, and random tube to tube variations in RT-PCR reactions can be compensated. Additionally, the expression of the gene of interest can be normalized against the amplified internal control. The mRNA expression of ICE, TNFα, NFκB, bcl-2, IκB, and the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were analyzed by RT-MPCR (Maxim Biotech, Inc., San Francisco, CA) according to the manufacturer’s protocol with modifications. Two micrograms of total RNA were reverse transcribed for 1 h at 42 C in 25 μL buffer[ 50 mmol/L Tris-HCl (pH 8.3), 50 mmol/L KCl, 10 mmol/L MgCl2, and 10 mmol/L dithiothreitol] containing 50 U AMV reverse transcriptase, 1 μg random hexamer, 1 mmol/L deoxy-NTP, and 40 U RNasin ribonuclease inhibitor. All reagents were supplied by Promega Corp. complementary DNA (cDNA) was denatured for 5 min at 94 C, and 2.5 μL cDNA from each sample were used for subsequent PCR amplification in a final volume of 25 μL buffer containing 2.5 U Taq DNA polymerase (Promega Corp.), and supplied sense and antisense primers (Maxim Biotech, Inc., USA). To determine the linearity of amplification for quantitation, PCR was performed for 26–40 cycles, and amplification of all genes was linear between 26–32 cycles. For further PCR, reactions were performed at 96 C for 1 min and 56 C for 4 min for 2 cycles and at 94 C for 1 min and 56 C for 2.5 min for 29 cycles. The amplified fragments correspond to 921 bp (GAPDH), 658 bp (ICE), 535 bp (TNFα), 409 bp (the p65 subunit of NFκB), 235 bp (bcl-2), and 158 bp (the L-factor subunit of IκB). PCR was negative when reactions were performed without AMV reverse transcriptase during RT reactions, indicating the absence of DNA contamination in RNA samples.
To provide an additional control for each RT-MPCR, a parallel 18S rRNA RT-PCR using the same cDNA, PCR conditions, and 18S primers was performed using the QuantumRNA 18S kit, which contains both 18S rRNA primers and competimers (AMS Biotechnology Europe Ltd., Abingdon, UK). 18S rRNA competimers are modified at their 3′-ends to block extension by DNA polymerase. By mixing 18S primers with increasing concentrations of 18S competimer, the overall PCR amplification efficiency of 18S cDNA was reduced without the primers becoming limiting due to abundance of 18S rRNA. In this study, a 3:7 ratio of 18S primers to competimers was chosen, and PCR was performed at 56 C for 16–36 cycles to determine the linear range of amplification. The linear range was 18–26 cycles, and 23 cycles were chosen for further PCR to offset any variations in RT-MPCR. The 488-bp 18S PCR product was separated on 2% agarose gels stained with ethidium bromide.
Analyses of IL-1β mRNA by RT-MPCR
RT-MPCR was performed to analyze IL-1β mRNA using 18S rRNA as an internal standard. QuantumRNA 18S primers and competimers at a ratio of 2:8 were added to the PCR reaction with two IL-1β primers that were designed using the Primer3 software available at http://www-genome.wi.mit.edu/genome_software/other/primer3.html. The primer sites are located at nucleotides 591–610, and 955–974 of human IL-1β complete cDNA (GenBank accession no. M15330). The primer sequences are: upstream, 5′-AATGACAAAATACCTGTGGC-3′; and downstream, 5′-AAACCTTTCTGTTCCCTTTC-3′. RT was performed using the method described above. PCR was carried out at 94, 55, and 70 C, 1 min each, for 29 cycles (within the linear range of amplification). The amplified IL-1β (384-bp) and 18S (488-bp) cDNAs were separated on 2% agarose gels.
Images of UV-illuminated agarose gels were captured using UVP ImageStore 5000 (UV Products Ltd., San Gabriel, CA). The density of cDNA bands were analyzed using Gelbase/Gelblot software (UV Products, Ltd.). Relative changes in the mRNA levels of apoptosis-related genes were assessed by comparing the density ratios of the gene of interest against both GAPDH and 18S. Changes in IL-1β mRNA expression were assessed by comparing the density ratios of IL-1β against 18S cDNA.
Quantitation of IL-1β release with ELISA
IL-1β release was measured using a Quantikine immunoassay kit that detects active IL-1β (the mature cleaved form) in cell culture supernatants (R\|[amp ]\|D Systems Europe Ltd., Abingdon, UK). Briefly, recombinant human IL-1β standards and cell culture media were added to the microtiter plate that was precoated with a monoclonal antibody specific for human IL-1β. Bound IL-1β was detected with a horseradish peroxidase-conjugated polyclonal antibody using hydrogen peroxide and chromogen as substrates. Chemiluminescence, developed after the addition of sulfuric acid, was measured at 450 nmol/L and for correction at 540 nmol/L using a Labsystems bichromatic multiskan microplate reader (Labsystems, Farnborough, UK).
Apoptosis detection by terminal deoxynucleotidyl transferase-mediated deoxy-UTP-fluorescein nick-end labeling and propidium iodide staining (TUNEL)
Apoptosis was detected by TUNEL and propidium iodide (PI) staining of nuclear DNA using methods previously described (36). Apoptotic features shown by both PI and FITC fluorescence of cell nuclei were identified, and apoptotic indexes were obtained by counting 250 cells in at least 4 fields of view.
Statistics
All experiments in the study were performed using adipose tissue from at least three patients (n ≥ 3). Data from representative preparations are shown. One-way ANOVA and paired Student’s t test were used for data analysis in the study. Data are the mean ± sem.P < 0.05 was considered significant.
Results
TNFα increased the mRNA expression of ICE, TNFα, NFκB, and bcl-2 in human preadipocytes and adipocytes
After RT-MPCR and parallel 18S rRNA RT-PCR, regulation of the expression of apoptosis-related genes by TNFα was analyzed by comparing their density ratios against those of the housekeeping gene GAPDH and 18S rRNA. The data are shown in Fig. 1A. TNFα treatment (8 h) induced dose-dependent increases in the expression of ICE, TNFα, NFκB, and bcl-2 in human preadipocytes (Fig. 1A; P < 0.01 at 1 nmol/L or more), whereas it had no significant effect on the expression of IκB. A concentration of 6 nmol/L (∼100 ng/mL) TNFα was chosen for subsequent experiments, because significant effects were found at this concentration. The use of GAPDH as an internal control for RT-MPCR in the study was validated by plotting the density ratios of each gene against GAPDH and 18S rRNA independently, and this revealed closely comparable changes (variation <10%) in the expression of each gene.
A, The effects of TNFα on apoptosis-related gene expression in human preadipocytes. Preadipocytes were treated with TNFα (0–10 nmol/L) for 8 h. mRNA expression was analyzed by RT-MPCR and normalized against GAPDH (internal control) and 18S rRNA (external control from parallel 18S-PCR). The upper panel shows a representative RT-MPCR. P < 0.01 at 1 nmol/L and above for each gene. B, The time course of TNFα (6 nmol/L) treatment of preadipocytes. Maximum increases in the expression of bcl-2, ICE, and NFκB were observed at 8 h (P < 0.01), and a maximum increase in TNFα expression was obtained at 1 h (P < 0.01).
Figure 1B shows the time course of the effects of TNFα treatment on preadipocytes. ICE mRNA was increased 2.5-fold (P < 0.01), with a maximum increase at 8 h. TNFα mRNA was increased up to 3-fold at 1 h (P < 0.01), followed by a gradual decline to the control level at 16 h. There were gradual increases in NFκB and bcl-2 mRNA, rising to 2-fold at 8 h (P < 0.01).
In mature human adipocytes, dose-dependent increases in the gene expression of ICE, TNFα, NFκB, and bcl-2 were also observed after treatment with TNFα (Fig. 2A). At 6 nmol/L, TNFα increased the expression of ICE 3-fold (P < 0.01), that of TNFα 2.5-fold (P < 0.01), that of NFκB 2.1-fold (P < 0.01), and that of bcl-2 2.7-fold (P < 0.01). Figure 2B shows that these increases were observed after 24-h treatment. They coincide with the general decline in these mRNAs in control (nontreated) samples during 2- to 24-h incubation (data not shown).
The effects of TNFα on apoptosis-related gene expression in human mature adipocytes. A, Adipocytes were treated with TNFα (0–10 nmol/L) for 24 h. mRNA expression was normalized against GAPDH and 18S after RT-MPCR and parallel 18S-PCR. P < 0.01 at 3 nmol/L and above for each gene. B, The time course of TNFα effects. P < 0.01 at 24 h.
Dex inhibited TNFα-induced ICE and TNFα mRNA expression
To determine the effects of Dex and insulin on TNFα-induced apoptosis-related gene expression, preadipocytes were treated with 6 nmol/L TNFα for 8 h in the presence/absence of varying concentrations of Dex and insulin. Figure 3A shows the dose-dependent inhibitory effect of Dex on TNFα-induced gene expression. Dex (10−7 mol/L) significantly reduced TNFα-induced ICE mRNA by 56% (P < 0.01) and TNFα mRNA by 54% (P < 0.01). The reductions in TNFα-induced expression of NFκB (by 16%) and bcl-2 (by 14%) were not statistically significant. Insulin treatment (10−10–10−6 mol/L) had no significant effect on TNFα- induced gene expression (data not shown).
A, The effects of Dex on TNFα-induced gene expression in preadipocytes. Preadipocytes were treated with 6 nmol/L TNFα and the indicated concentrations of Dex for 8 h. Significant inhibition of ICE and TNFα expression were observed at 10−10 mol/L and above (P< 0.01). B, The effects of Dex and insulin on TNFα-induced gene expression in mature adipocytes. Adipocytes were treated with 6 nmol/L TNFα in the presence or absence of 10−7 mol/L Dex, 10−7mol/L insulin, or both for 24 h. mRNA expression was analyzed by RT-MPCR. Dex inhibited TNFα-induced mRNA expression of ICE and TNFα (P < 0.01). Insulin alone had no significant effect on gene expression, but combined treatment with Dex and insulin synergistically inhibited TNFα-induced ICE, TNFα, and NFκB (P < 0.01).
Figure 3B shows the effects of Dex and insulin on TNFα-induced gene expression in mature human adipocytes. Dex treatment (10−7 mol/L for 24 h) decreased TNFα-induced mRNA expression of ICE by 29% (P < 0.05) and TNFα by 36% (P < 0.01), but had no significant effect on NFκB and bcl-2 expression. Insulin alone (10−7 mol/L for 24 h) had no significant effect on TNFα-induced gene expression. Combined treatment with Dex (10−7 mol/L) and insulin (10−7 mol/L), however, synergistically inhibited TNFα induced mRNA expression of ICE by 52% (P < 0.01), TNFα by 57% (P < 0.01), NFκB by 37% (P < 0.01), and bcl-2 by 12% (P > 0.05). Similarly, in preadipocytes the combination of Dex and insulin synergistically inhibited TNFα-induced expression of ICE by 67% (P < 0.01), TNFα by 69% (P < 0.01), NFκB by 35% (P < 0.01), and bcl-2 by 29% (P < 0.01).
Dex inhibited TNFα-induced IL-1β mRNA and protein release in human preadipocytes and mature adipocytes
The expression and regulation of IL-1β by TNFα, Dex, and insulin in human adipose cells were investigated. In human preadipocytes, treatment with 6 nmol/L recombinant human TNFα increased IL-1β mRNA expression 5.3-fold (P < 0.01) at 1 h and 4.2-fold (P < 0.01) at 8 h. Preincubation with 10−7 mol/L Dex for 8 h reduced TNFα-induced IL-1β mRNA to almost control levels at both time points. Coincubation with TNFα and Dex (10−7 mol/L) inhibited TNFα-induced IL-1β mRNA by 93% at 1 h (P < 0.01) and 83% at 8 h (P < 0.01; Fig. 4). Coincubation with 10−7 mol/L insulin inhibited TNFα-induced IL-1β expression by 11%, but the inhibition was not significant (data not shown).
The effects of Dex on the mRNA expression of IL-1β in preadipocytes. Preadipocytes were incubated with or without 6 nmol/L TNFα and 10−7 mol/L Dex (or pretreatment for 8 h) for 1–8 h. IL-1β mRNA expression was analyzed by RT-MPCR using 18S as an internal control. The upper panel shows a representative RT-MPCR. *, P< 0.01 compared with control; **, P < 0.001 compared with TNFα-treated cells.
IL-1β release from preadipocytes was increased 5-fold by treatment with 6 nmol/L TNFα for 24 h (Fig. 5). Preincubation with 10−7 mol/L Dex (12 h) and cotreatment with Dex and TNFα reduced TNFα-induced IL-1β release to control levels. Treatment with Dex alone (10−7 mol/L) reduced IL-1β release to 60.6% of the control level. Treatment with insulin (10−7 mol/L) produced no effect on IL-1β release. Combined treatment with insulin (10−7 mol/L) and Dex (10−7 mol/L) had no further effect on TNFα-induced IL-1β release than Dex alone.
The effects of Dex on IL-1β release from preadipocytes. Preadipocytes were treated with Dex (10−7 mol/L), TNFα (6 nmol/L), and the combination of Dex (including pretreatment for 8 h, pre-Dex) and TNFα for 24 h. IL-1β release was measured by ELISA. *, P < 0.01 compared with control; **, P < 0.001 compared with TNFα-treated cells.
In mature adipocytes, TNFα (6 nmol/L for 2–24 h) had no significant effect on IL-1β mRNA expression (Fig. 6). Coincubation with 6 nmol/L TNFα and 10−7 mol/L Dex for 24 h reduced IL-1β mRNA to 30% of that in TNFα (alone)-treated samples (P < 0.01). Coincubation with 10−7 mol/L insulin decreased IL-1β mRNA by 27% (P < 0.01).
The effects of Dex on the mRNA expression of IL-1β in adipocytes. Adipocytes were treated with 6 nmol/L TNFα in the presence of 10−7 mol/L insulin, 10−7 mol/L Dex, and their combination. IL-1β mRNA expression was analyzed by RT-MPCR using 18S as an internal control. The upper panel shows a representative RT-MPCR.* , P < 0.01 compared with levels in TNFα-treated samples.
TNFα increased IL-1β release from mature adipocytes 1.5-fold after 8 h (25.5 vs. 37.2 pg/mL; P < 0.01), 5.2-fold after 8–24 h (9.2 vs. 48 pg/mL; P< 0.001), and 20-fold after 24–48 h of treatment (3.4 vs. 68.1 pg/mL; P < 0.001) as shown in Fig. 7. Treatment with Dex (10−10–10−6 mol/L) decreased TNFα-induced IL-1β release in a dose-dependent manner (data not shown). At 10−7 mol/L, it completely abrogated TNFα- induced IL-1β release (P < 0.001; Fig. 7). In contrast, insulin (10−7 mol/L) had no effect on IL-1β release. Combined treatment with insulin and Dex had no further effect on TNFα-induced IL-1β release than Dex alone.
The effects of Dex on IL-1β release from adipocytes. Adipocytes were treated with Dex (10−7 mol/L), TNFα (6 nmol/L), or their combination for 8–48 h. IL-1β release was measured by ELISA. *, P < 0.01; **, P < 0.001 (compared with control). ***, P < 0.001 (compared with TNFα-treated samples).
Dex inhibited TNFα-induced apoptosis of human preadipocytes and adipocytes
After 24-h treatment with 6 nmol/L TNFα with or without treatment with Dex (10−8 and 10−7 mol/L) and insulin (10−7 mol/L), apoptosis of preadipocytes and mature adipocytes was assayed by TUNEL and PI staining of nuclear DNA (36). Abnormal nuclear changes characteristic of apoptosis, including clumping of chromatin beneath the nuclear envelope, convolution of the nuclear outline, and disintegration of the cell nucleus, were observed. After 24 h in serum-free conditions, 10.2% of preadipocytes were undergoing apoptosis (Fig. 8A). Treatment with Dex alone had no significant effect. Treatment with TNFα (6 nmol/L) increased apoptosis almost 2-fold to 18.6%, but preincubation with 10−7 mol/L Dex for 8 h before treatment with TNFα completely blocked TNFα-induced apoptosis (mean ± sem, 10.7 ± 2.1%; P < 0.01 vs. TNFα-treated). Cotreatment with Dex (10−7 and 10−8 mol/L) also effectively inhibited TNFα-induced apoptosis (P < 0.01 vs. TNFα-treated). Insulin alone (10−7 mol/L) and in combination with TNFα had no significant effect on apoptotic indexes. Combined treatment with insulin and Dex had no additional effect in preventing TNFα-induced apoptosis than Dex alone.
The effects of Dex and TNFα on apoptosis of preadipocytes (A) and adipocytes (B). Preadipocytes and adipocytes were cultured in serum-free medium and treated with TNFα (6 nmol/L), Dex (10−8 or 10−7 mol/L), or their combination for 24 h. Apoptosis was detected by TUNEL and propidium iodide staining. *, P < 0.01 compared with control samples; **, P < 0.01 compared with TNFα-treated samples.
In mature adipocytes, apoptotic indexes of 20.4 ± 2.1 (±sem) were observed after cells were serum deprived for 24 h (Fig. 8B). Treatment with Dex alone during serum deprivation did not significantly affect the apoptotic index. Treatment with TNFα increased apoptosis almost 2-fold to 38.6 ± 2.5% (P < 0.01). Coincubation with 10−7 mol/L Dex significantly decreased apoptosis to 25.4 ± 1.7% (P < 0.01 vs. TNFα-treated) and to 31.5 ± 2.6% (P < 0.01 vs. TNFα-treated) with 10−8 mol/L Dex. Insulin alone (10−7 mol/L) and in combination with TNFα had no significant effect on apoptotic indexes. Combined treatment with insulin and Dex had no further effect in preventing TNFα-induced apoptosis than Dex alone.
Discussion
Consistent with the general belief that TNFα induces both pro- and antiapoptotic mediators, TNFα treatment of human sc preadipocytes and mature adipocytes led to dose-dependent increases in the gene expression of ICE, TNFα, NFκB, and bcl-2. In preadipocytes, the increase in TNFα mRNA was rapid (maximum at 1 h), whereas the increases in ICE, NFκB, and bcl-2 mRNAs occurred later (1–8 h). Rapid induction of endogenous TNFα mRNA and subsequent protein translation by exogenous TNFα have been reported in human myosarcoma (KYM-S) cells, murine tumorigenic fibroblasts (L-M) (37), and human ZR-75–1 breast carcinoma cells (38) and have been associated with acquired resistance to TNFα-induced cytotoxicity in these cell lines. We confirmed by immunostaining (data not shown) that adipocytes produce detectable levels of TNFα. The local concentrations of TNFα in adipose interstitial fluid are unknown, but could conceivably reach the concentrations used in this study. This and the findings that exogenous TNFα induces apoptosis in both human preadipocytes and mature adipocytes (3) questions the association of overexpression with resistance in human adipose cells. Locally produced TNFα may potentiate its adipostat effects by autocrine/paracrine effects on its own production.
Dex treatment inhibited TNFα-induced mRNA expression of ICE and TNFα in both sc preadipocytes and adipocytes. Insulin in combination with DEX produced a synergistic inhibition of TNFα-induced mRNA expression of ICE, TNFα, NFκB, and bcl-2. The selective inhibitory effects of Dex may offset the balance between TNFα-induced proapoptotic and antiapoptotic mediators and thereby modify the apoptotic effects of TNFα. The mRNA expression of IκB in preadipocytes and adipocytes was not affected by treatment with TNFα, Dex, and/or insulin, but the expression of this gene was low in all conditions.
In mature sc adipocytes, TNFα-mediated increases in mRNA expression were observed only after 24-h treatment compared with 1–8 h for the preadipocytes. The differences in the time course of TNFα-mediated changes in the two cell populations may relate to the culture conditions. Preadipocytes were cultured in serum-containing medium for 72 h (to reach confluence) before TNFα treatment, whereas mature adipocytes were treated immediately after isolation. The mRNAs of ICE, TNFα, NFκB, and bcl-2 in control (nontreated) adipocytes decreased with incubation time (2–24 h), suggesting that the isolation procedure itself may have increased the expression of these genes and overshadowed the early effects of TNFα, such as the rapid induction of endogenous TNFα mRNA seen in preadipocytes.
TNFα (6 nmol/L) increased IL-1β mRNA expression 5-fold in human sc preadipocytes, but had no significant effect on mature adipocytes. Dex decreased TNFα-induced IL-β mRNA expression to almost control levels in preadipocytes and to 70% of the control levels in mature adipocytes. Dex has been shown to decrease IL-1β mRNA expression in the human promonocytic cell line U-937 by inhibiting IL-1β gene transcription and decreasing the stability of IL-1β mRNA (39). Furthermore, Dex is known to down-modulate (via glucocorticoid receptor protein) the activity of the activator protein-1 complex, a transcription factor that regulates the expression of various cytokines, including TNFα and IL-1β (40, 41). The inhibition of IL-1β and TNFα mRNAs by Dex in the present study may be mediated by similar mechanisms.
Dex also abrogated TNFα-induced IL-1β protein release from human sc preadipocytes and adipocytes. The decrease in IL-1β release may result from decreased IL-1β mRNA expression in treated cells. However, in mature adipocytes the discrepancy between the effects of Dex on IL-1β mRNA (no effect at 8 h and only 30% reduction at 24 h) and the effects on IL-1β release (40% decrease in total IL-1β release after 8 h and 91% decrease after 8–24 h) suggests that decreased IL-1β release may also result from inhibition of ICE activation. The time-dependent (8–48 h) decrease in IL-1β release from control mature adipocytes may correlate with their recovery from the isolation procedure. TNFα treatment increased IL-1β protein release from both preadipocytes (5 fold) and adipocytes (1.5- to 20-fold), indicating ICE activation in these cells. Dex inhibits IL-1β release from murine peritoneal macrophages (42) and decreases circulating IL-1β levels in mice (43). The inhibition of IL-1β release from both adipose cells and macrophages may thus reflect a general effect of glucocorticoids on production of this cytokine.
IL-1β plays a pivotal immunological role in host defense against infections. It also stimulates lipolysis and inhibits lipogenesis by inhibiting the expression of fatty acid transport protein and fatty acid translocase in adipose tissue (44). In this study we have shown mRNA expression of IL-1β and the release of assayable protein. The finding that IL-1β is released by human sc adipose cells and that it is regulated by TNFα may have significant in vivo implications. Firstly, increased TNFα production in obesity, infection, or malignancy may act on adipose tissue to increase the expression and release of IL-1β, which, in turn, regulates lipid metabolism and synergizes with other effects of TNFα. Secondly, adipose tissue constitutes up to 10–30% of the total body weight and produces a range of cytokines, including TNFα (4, 5); IL-6 (45); complement factors D, B, and C3 (46); and macrophage colony-stimulating factor (47). The finding that this large depot also produces and releases IL-1β in response to TNFα suggests a possible immunological role for adipose tissue. Conceivably, elevated TNFα during infection may stimulate IL-1β release from adipose tissue, potentiating the host immune response. These effects of TNFα and IL-1β may be modulated by glucocorticoids in obesity or therapeutically administered for combating inflammatory diseases. Whether glucocorticoids exert the same antagonistic effects on TNFα-induced apoptosis and interleukin-1β release in adipocytes from visceral fat and other fat depots remains to be determined.
The outcome of TNFα activation, i.e. cell death or cell survival, depends on the balance of pro- and antiapoptotic mediators. Dex selectively inhibited TNFα-induced gene expression of ICE, TNFα, IL-1β, and IL-1β release in human sc adipocytes and preadipocytes. Combined treatment with Dex and insulin synergistically inhibited TNFα-induced gene expression, although insulin alone had no significant effect. Dex treatment decreased TNFα-induced apoptosis to control levels, but in apoptosis assays, there was no synergism with insulin. This may be because insulin in combination with Dex inhibited not only the proapoptotic genes, ICE and TNFα, which were induced by TNFα, but also the antiapoptotic genes, NFκB and Bcl-2. Consistent with the lack of synergy of insulin with Dex on apoptosis are the data showing no synergistic effect on IL-1β release, which supports the role of this cytokine in mediating the effects of TNFα. ICE is only one member of the large caspase family. The effects of Dex on other TNFα-induced apoptosis mediators, such as other ICE-related proteases, may be involved in its inhibition of apoptosis.
Glucocorticoids and TNFα exert opposing effects on food intake and on adipose differentiation and lipogenesis. Both are produced in adipose tissue and may thus regulate adipose tissue mass in an autocrine/paracrine as well as a neuroendocrine manner. Our findings that Dex inhibited TNFα-induced apoptosis, IL-1β release, and TNFα expression in adipose tissue provide the first direct in vitro evidence that glucocorticoids inhibit TNFα actions in adipose tissue and delineate a possible mechanism by which TNFα resistance occurs in obesity.
Acknowledgements
We thank all the operative surgeons and theater staff at the University Hospitals Trust, and Mr. P. L. Levick at the Priory Hospital in Birmingham for their assistance.
This work was supported by the University of Birmingham, the British Diabetic Association, and Eli Lilly & Co. UK.
Current address: Jean Mayer U.S. Department of Agriculture Human Nutrition Research Center on Aging, Tufts University, Boston, Massachusetts 02111-1524.
