INTRODUCTION

Tumor necrosis factor- (TNF-)¹ is a multifunctional cytokine that was originally identified as a tumoricidal protein (1). Subsequent investigations have shown that TNF- was fundamentally involved in control of the growth, differentiation, and metabolism in several normal cells and tissues.

A sum of work has been focused on the role of TNF- in the regulation of adipocyte development and metabolism (1, 2). Thus, TNF- strongly inhibits adipose conversion and even causes a dramatic dedifferentiation of adipocytes in culture (3-6). Moreover, it is also documented that TNF- decreases the synthesis and activity of several proteins essential for lipogenesis and triglyceride accumulation in adipocytes. These include lipoprotein lipase (5, 7-9), acetyl-coenzyme A carboxylase (4, 10, 11), acyl-coenzyme A synthetase (12), stearoyl-coenzyme A desaturase (12), and the insulin-sensitive glucose transporter GLUT4 (13, 14).

In addition to the above effects on adipogenesis and reduction of de novo fatty acid synthesis, TNF- also depletes triglycerides from adipocytes by directly increasing lipolysis (15-17). So far, the molecular mechanism by which the cytokine potentiates lipolysis remains unclear (17).

Through activation of three -AR subtypes, catecholamines exert a key role in the regulation of lipid metabolism in adipocytes. They modulate cAMP-dependent processes such as lipolysis and the genetic control of the lipogenic and thermogenic pathways. As regards the pleiotropic effects of TNF- on several metabolic pathways in adipocytes, and to the key role of the -AR system in lipid mobilization, it is possible that an interplay between TNF- and -ARs contributes to the biological effects of the cytokine. The present study was initiated to investigate the regulation by TNF- of the three -AR subtypes. Using the model of the 3T3-F442A cell line that presents a coordinate expression of the three -AR subtypes in the mature adipose phenotype (18, 19), we demonstrate that TNF- differentially regulates ₁-, ₂-, and ₃-AR gene expression and responsiveness.

MATERIALS AND METHODS

3T3-F442A cells (20) were grown and differentiated as described (18). At day 8 after confluence, more than 90% of the cells had the morphology of mature adipocytes. After two washes, cells were kept for 24 h in defined medium consisting of Dulbecco's modified Eagle's medium/Ham's F-12 medium (2:1, v/v) and 0.1% bovine serum albumin. Thereafter, the cells were maintained either in the absence or in the presence of TNF-.

Total RNA was extracted from 3T3-F442A adipocytes by the method of Cathala et al. (21). For Northern blot analysis RNA samples were electrophoresed through a 1.5% agarose, 2.2 M formaldehyde gel, and transferred to nylon Nytran-plus membranes (Schleicher & Schuell). After RNA fixation, prehybridization was carried out at 65 °C for 30 min in the presence of 0.5 M sodium phosphate (pH 6.8), 7% SDS, 1% bovine serum albumin, and 1 mM EDTA (22). Hybridization was performed in the same buffer in the presence of the heat-denatured probe (2-3 × 10⁶ cpm/ml). Membranes were washed twice for 30 min at 65 °C in 2 × SSC (1 × SSC: 150 mM NaCl, 15 mM sodium citrate), 0.1% SDS, and then once in 0.2 × SSC, 0.1% SDS for 15 min at 65 °C. Probes were labeled by random priming with [-³²P]dCTP (ICN Radiochemicals). The ₁- and ₂-AR probes have been described elsewhere (23). The ₃-AR probe is a 305-bp amplification product of the cloned murine ₃-AR gene (24). -Actin probe (pAct-1) (25) was generously given by Dr. Howard Green (Harvard Medical School, Boston, MA).

For RT-PCR analysis of -AR subtype expression, total RNA was digested for 15 min at 37 °C with 0.1 unit of RNase-free DNase I (RQ1 DNase, Promega)/µg of nucleic acid in 40 mM Tris-HCl (pH 7.9), 10 mM NaCl, 6 mM MgCl₂, 10 mM CaCl₂ in the presence of 1 unit/µl ribonuclease inhibitor (RNAguard, Pharmacia Biotech Inc.). After phenol/chloroform extraction and ethanol precipitation, RNA (0.25-1 µg) was reverse-transcribed with MMLV-RT (200 units/µg) (Life Technologies, Inc.) in the presence of 10 µM random hexanucleotides (Pharmacia), 1 unit/µg ribonuclease inhibitor, 400 µM of each dNTP in a final volume of 20 µl consisting of 50 mM Tris-HCl (pH 8.3), 75 mM KCl, 3 mM MgCl₂, and 10 mM dithiothreitol. After a 1-h incubation at 42 °C, MMLV-RT was heat-inactivated. To ensure that subsequent amplification did not derive from contaminant genomic DNA, a control without MMLV-RT was included for each RNA sample. cDNA were denatured for 5 min at 95 °C and submitted to 20-25 cycles of amplification (1 cycle: 94 °C, 30 s; 60 °C, 1 min; 72 °C, 1 min) followed by a final extension of 5 min at 72 °C in a DNA thermal cycler 9600 (Perkin Elmer). PCR was performed in a 25-µl reaction containing 1 unit of Taq polymerase (Life Technologies, Inc.), 125 µM each dNTP, 5% formamide, 125 nM both sense and antisense oligonucleotides. The buffer consisted of 20 mM Tris-HCl (pH 8.4), 50 mM KCl, and 2 mM MgCl₂. Sequences of the sense and antisense oligonucleotides were: 5-GGATCCAAGCTTTCGTGTGCACCGTGTGGGCC-3 and 5-GGATCCAAGCTTAGGAAACGGCGCTCGCAGCTGTCG-3 for the ₁-AR; 5-GCCTGCTGACCAAGAATAAGGCC-3 and 5-CCCATCCTGCTCCACCT-3 for the ₂-AR; 5-ATGGCTCCGTGGCCTCAC-3 and 5-CCCAACGGCCAGTGGCCAGTCAGCG-3 for the ₃-AR; 5-GAGACCTTCAACACCCC-3 and 5-GTGGTGGTGAAGCTGTAGCC-3 for -actin. These oligonucleotides were derived from the sequences of the corresponding genes and cDNAs (24, 26-28). Amplification products had expected sizes of 286, 329, 308, and 236 bp for ₁-, ₂- and ₃-ARs and -actin, respectively. They were separated on a 2% agarose gel and then visualized by ethidium bromide staining or transferred to nylon Nytran-plus membranes (Schleicher & Schuell). Prehybridization and hybridization with specific ₁-, ₂-, and ₃-AR and -actin [-³²P]dCTP random-primed DNA probes were carried out in a sodium phosphate buffer (22). Final washes were performed in 0.2 × SSC, 0.1% SDS for 15 min at 65 °C. Preliminary experiments were carried out with various amounts of cDNA to determine nonsaturating conditions of PCR amplification for ₁-, ₂-, and ₃-AR and -actin. Thus, cDNA amplification was performed in comparative and semiquantitative conditions. For quantitation, gels or autoradiograms were analyzed by videodensitometric scanning (Vilber Lourmat Imaging), and -AR signals were normalized to those of -actin.

For transcription analysis, preparation of the nuclei, elongation assay, DNase I, and proteinase K treatments were performed as described previously (29-31). Equal amounts (~2 × 10⁷ cpm) of labeled nascent RNAs were hybridized for 40 h at 60 °C in a sodium phosphate buffer (22) to nitrocellulose filters spotted with linearized probes (23, 29-31). Membranes were washed twice in 2 × SSC for 30 min at 60 °C and treated for 30 min at 37 °C in the same buffer containing RNase A at 5 µg/ml. Final washing was performed for 20 min at 65 °C in 0.2 × SSC, 0.1% SDS.

Cell extracts were prepared as described previously (18). Protein content was assayed (32) using bovine serum albumin as a standard.

Adenylyl cyclase activity (EC 4.6.1.1) was measured as described (18). Briefly, the reaction was carried out for 10 min at 35 °C and was initiated by the addition of crude membranes to a standard buffer containing 0.1 mM [-³²P]ATP (1 µCi) (ICN), 1 mM cAMP, 10 mM phosphocreatine, 0.5 unit of creatine phosphokinase, 100 µM GTP (Boehringer Mannheim) (except when GTPS, NaF, or forskolin were used), 10 mM MgCl₂, 0.2 mM EDTA, and 50 mM Tris-HCl (pH 7.5), with or without a -adrenergic effector.

G3PDH activity was measured in the cytosolic fraction as described (33).

For [¹²⁵I]CYP binding experiments, cells were harvested and homogenized at 4 °C in 1 mM EDTA, 25 mM Tris-HCl (pH 7.5). Homogenates were centrifuged for 30 min at 4 °C at 100,000 × g. The pellet was resuspended in the homogenization buffer and stored at 80 °C until it was used. Membrane aliquots (40 µg of protein) were incubated for 30 min at 37 °C with [¹²⁵I]CYP with or without competing ligand in a final volume of 250 µl consisting of 50 mM Tris-HCl (pH 7.4), 10 mM MgCl₂, 1 mM ascorbic acid, and 100 µM GTP. After dilution with 3 ml of ice-cold 10 mM MgCl₂, 50 mM Tris-HCl (pH 7.5), separation of bound from free radioligand was achieved by vacuum filtration over Whatman GF/C filter paper, followed by three washes with the same buffer. Saturation experiments were performed with [¹²⁵I]CYP concentrations ranging from 5 to 4000 pM. Competition experiments were carried out at 30 pM [¹²⁵I]CYP. Nonspecific binding was determined in the presence of 10 µM CGP12177 and was usually ~20% of total binding at 30 pM [¹²⁵I]CYP. Data from saturation and competition experiments were analyzed with the EBDA and LIGAND programs (Biosoft-Elsevier, Cambridge, United Kingdom (UK)).

Lipolysis was assessed as glycerol release from adherent cells in 24-well plates. Adipocyte monolayers were washed with Krebs-Ringer phosphate buffer (pH 7.4), supplemented with 2% fatty acid-free bovine serum albumin, 4.5 g/liter glucose, and 50 µg/ml Na₂S₂O₅ as antioxidant. Cells were then incubated with or without -adrenergic effectors for 2 h at 37 °C. Aliquots of the incubation medium (300 µl) were removed to determine glycerol (34). Reduction of NAD to NADH was recorded at 340 nm in the presence of glycerol dehydrogenase (EC 1.1.1.6) from Enterobacter aerogenes (Boehringer Mannheim). Reaction buffer consisted of 130 mM (NH₄)₂SO₄, 10 µM MnCl₂, 3.3 mM NAD, and 125 mM K₂CO₃/NaHCO₃ (pH 10).

[¹²⁵I]CYP was obtained from Amersham and [-³²P]ATP from ICN Radiochemicals. CGP12177 and CGP20712A were generous gifts from Ciba-Geigy (Basel, Switzerland). ICI118551 was provided by Imperial Chemical Industries (Macclesfield, UK), BRL37344 by Smith Kline Beecham Pharmaceuticals (Epsom, UK). ISO, ()-norepinephrine, ()-epinephrine, GTPS, NaF and forskolin were Sigma products. Murine recombinant TNF- was purchased from Genzyme (Cambridge, MA).

Results are presented as means ± S.E. The level of significance between groups was assessed using either paired or unpaired Student's t test.

RESULTS

Differential Modulation by TNF- of ₁-, ₂- and ₃-AR mRNA Levels

Since the ₃-AR is the main -AR subtype expressed in rodent adipocytes (18, 19, 35), we first evaluated the effect of TNF- on ₃-AR gene expression. 3T3-F442A mature adipocytes were exposed or not to various TNF- concentrations for 6 h, and total RNA was extracted. Then ₃-AR mRNA levels were studied both by Northern blotting and RT-PCR analyses. These two approaches converged to demonstrate that TNF- down-regulated ₃-AR mRNA levels (Fig. 1). Repression of the three ₃-AR mRNA species (2.3, 2.8, and 4.4 kilobases) was detectable at a cytokine concentration of 10 pM and was maximal at 100 pM, the half-maximal effect being between 10 and 30 pM.

Effect of TNF- concentration on ₃-AR mRNA levels.

The time dependence of ₃-AR mRNA down-regulation was also studied in 3T3-F442A adipocytes untreated or treated for 1-48 h with 1 nM TNF-. Accumulation of -actin mRNA remained relatively constant over the time course of the treatment. Conversely, ₃-AR mRNA levels decreased as early as 2 h (Fig. 2) after the onset of TNF- treatment, with a maximal repression occurring by 6 h (8.5 ± 1.6-fold decrease, p < 0.001).

Time-dependent down-regulation by TNF- of ₃-AR mRNA expression in 3T3-F442A adipocytes.

Modulation of -AR mRNA levels by TNF- was not limited to the ₃-AR subtype. Exposure of 3T3-F442A adipocytes to increasing concentrations of TNF- for 6 h slightly decreased ₁-AR mRNA abundance (Fig. 3, B and C). This effect was detectable only at a concentration of 100 pM and was maximal at 1 nM. By contrast, TNF- potently induced ₂-AR mRNA levels (Fig. 3, A-C). This increase was significant at 3 pM TNF- and was maximal at 300 pM, giving a half-maximal action between 3 and 10 pM. TNF--induced repression of ₁-AR transcripts was observed after a 1-h exposure to 1 nM cytokine and was maximal at 6 h (~1.6-fold decrease, p < 0.01) (data not shown). Induction of ₂-AR mRNA content by TNF- was very rapid. As compared with control levels, the increase in ₂-AR mRNA abundance was observed as early as 1 h in the presence of 1 nM TNF-. This effect was maximal at 6 h (6-fold increase, p < 0.005) and then became less pronounced (4-fold induction at 48 h).

Dose-dependent effect of TNF- on ₁- and ₂-AR mRNA levels.

Regulation by TNF- of -AR Subtype Density

Saturation and competition binding experiments were carried out on membrane fractions to determine the levels of -AR subtype density after treatment of 3T3-F442A adipocytes for 24 h with 1 nM TNF-. In agreement with the results derived from Northern blotting and RT-PCR analyses, TNF- down-regulated the ₃-AR (Table I). The density of ₃-ARs, corresponding to the B_max of the low affinity sites for [¹²⁵I] CYP (18), was reduced by 2-fold after TNF- exposure (n = 4, p < 0.001). By contrast, the number of the high affinity binding sites that represents the sum of ₁- and ₂-ARs was moderately induced in the presence of TNF- (1.8-fold increase, n = 4, p < 0.05). No significant difference in the K_D values of the two binding classes for the radioligand could be detected between control and TNF--treated cells.

Table I. Characterization of [125I]CYP binding sites in membranes from control and TNF--treated 3T3-F442A adipocytes Membranes from control and TNF--treated (1 nM for 24 h) 3T3-F442A adipocytes were tested in [125I]CYP saturation binding experiments using a wide range of concentrations (5-4000 pM) of the radioligand. Scatchard analysis of the data with the EBDA/LIGAND program was used to calculate the KD and Bmax values of the high (1- and 2-ARs) and the low (3-AR) affinity sites for [125I]CYP. Results are expressed as mean ± S.E. of four separate experiments. The percentage of each affinity binding class is indicated in parentheses after the Bmax values. *, p < 0.05; **, p < 0.001, TNF--treated versus control adipocytes. Culture condition [125I]CYP binding sites High affinity (1- and 2-ARs) Low affinity (3-AR) KD Bmax KD Bmax pM fmol/mg pM fmol/mg Control 26.4  ± 3.4 4.7  ± 1.0 (2.1%) 2930  ± 490 215.4  ± 20.1 (97.9%) TNF- 29.7  ± 4.8 8.4  ± 1.7 (7.3%)* 2520  ± 500 106.7  ± 16.8 (92.7%)**

[¹²⁵I]CYP competition binding experiments against -AR subtype-selective ligands were also performed to determine the relative proportions of ₁- and ₂-ARs. These competition studies were carried out at a low (30 pM) [¹²⁵I]CYP concentration. As regards the low affinity of the ₃-AR for [¹²⁵I]CYP, this concentration of radioligand did not allow a significant occupancy of the ₃-AR by [¹²⁵I]CYP. This was confirmed by the competition of [¹²⁵I]CYP at 30 pM with increasing concentrations of the ₃-AR-selective agonist BRL37344. Analysis of [¹²⁵I]CYP displacement curve by BRL37344 indicated the absence of any detectable high affinity ₃-AR component (data not shown). Under these conditions, competition of [¹²⁵I]CYP with ₁- or ₂-AR subtype-selective antagonists allowed to estimate the relative proportions of each -AR subtype, with no significant interference of the ₃-AR population. In agreement with molecular results, analysis of the displacement curves of [¹²⁵I]CYP by the ₁-AR-selective antagonist CGP20712A or by the ₂-AR-selective antagonist ICI118551 demonstrated that TNF- caused a 3.3-5.6-fold induction in ₂-AR density (n = 5, p < 0.002) (Table II). TNF- only induced a weak and nonsignificant reduction in the ₁-AR population.

Table II. Competition of [125I]CYP against -AR subtype-selective ligands in membranes from control and TNF--exposed cells Membranes were prepared from control and TNF--treated (1 nM for 24 h) 3T3-F442A adipocytes. Competition binding experiments were performed at 30 pM [125I]CYP in the absence or in the presence of various concentrations of CGP20712A or ICI118551, 1- and 2-AR-selective antagonists, respectively. Data from displacement of [125I]CYP binding by these subtype-selective ligands were used to calculate the Ki values for each affinity component. The corresponding Bmax values were derived from total -AR density (1- plus 2-ARs) drawn from [125I]CYP saturation experiments (see Table I), taking into account the percentage of each affinity component (indicated in parentheses after the Bmax values) obtained from competition experiments. Results are expressed as mean ± S.E. of five separate experiments. *, p < 0.002, TNF--treated versus control adipocytes. Ligand Selectivity Binding affinity (related -AR subtype) Culture conditions Control TNF- Ki Bmax Ki Bmax nM fmol/mg nM fmol/mg CGP20712A  1 High (1-AR) 4.1  ± 1.3 3.0  ± 0.5 (64%) 4.9  ± 1.6 2.75  ± 0.85 (33%) Low (2-AR) 2000  ± 550 1.7  ± 0.5 (36%) 3300  ± 780 5.65  ± 0.85 (67%)* ICI118551  2 High (2-AR) 3.4  ± 1.8 0.95  ± 0.30 (20%) 3.1  ± 0.4 5.35  ± 0.35 (64%)* Low (1-AR) 627  ± 142 3.75  ± 0.30 (80%) 750  ± 182 3.05  ± 0.35 (36%)

Functional Consequences of the Differential -AR Subtype Regulation by TNF-

To determine the functional consequences of the differential regulation of the three -AR subtypes by TNF-, adenylyl cyclase activity in response to ISO or CGP12177 was measured on membranes from control and TNF--treated adipocytes. While ISO is a nonselective -AR agonist, CGP12177 is a ₁-/₂-AR antagonist but has agonistic properties at the ₃-AR site (36), and thus specifically addresses ₃-AR coupling to the adenylyl cyclase system. Exposure of 3T3-F442A adipocytes to TNF- provoked a time- and dose-dependent decline in adenylyl cyclase activity in response to a maximal dose (100 µM) of ISO or CGP12177. As compared with the kinetics of the TNF--induced regulation of -AR subtype mRNA levels, the effect of the cytokine on -adrenergic responsiveness was slightly delayed (Fig. 4). A significant decrease in ISO- and CGP12177-stimulated adenylyl cyclase activity was detectable after a 10-h exposure to 1 nM TNF-, and was near maximal after a 24-h treatment (2.5- and 10-fold decreases for ISO- and CGP12177-stimulated adenylyl cyclase activity, respectively; n = 5, p < 0.0001). Interestingly, repression of -AR effector-stimulated adenylyl cyclase activity was more pronounced for the ₃-AR agonist CGP12177 than for ISO. This inhibitory effect of TNF- on -AR coupling was maintained after a 5-day exposure to the cytokine.

Time dependence of TNF--induced decrease in ISO- or CGP12177-stimulated adenylyl cyclase activity.

Furthermore, when 3T3-F442A adipocytes were treated with increasing concentrations of TNF- for 24 h, a dose-dependent reduction in ISO- and CGP12177-stimulated adenylyl cyclase activity was clearly observed, while no significant change was detectable in G3PDH activity (Fig. 5). Repression of adenylyl cyclase activity stimulated by an optimal concentration (100 µM) of ISO or CGP12177 was detectable at 10 pM TNF- and was maximal between 0.3 and 3 nM. The EC₅₀ value for this effect was 39 ± 4.3 and 10.4 ± 0.3 pM TNF- for ISO and CGP12177, respectively.

Dose-dependent effect of TNF- on G3PDH activity and on ISO- or CGP12177-stimulated adenylyl cyclase activity.

To further analyze the modulation of -AR sensitivity by TNF-, the relative potencies of norepinephrine and epinephrine were compared between control 3T3-F442A adipocytes and cells exposed for 24 h to 1 nM TNF- (Fig. 6). As compared with control cells, maximal catecholamine-induced adenylyl cyclase activities were reduced in TNF--treated adipocytes. The functional switch in -AR subtype expression induced by TNF- was illustrated by the changes in the relative potencies of norepinephrine and epinephrine to stimulate adenylyl cyclase activity. In agreement with previous observations (18), norepinephrine was slightly more potent than epinephrine to activate adenylyl cyclase in control adipocytes. In TNF--treated adipocytes, epinephrine stimulated cAMP production with a higher potency than norepinephrine, suggesting the functional prevalence of the ₂-AR subtype.

Effect of TNF- on ()-norepinephrine- and ()-epinephrine-stimulated adenylyl cyclase activity.

In control and TNF--treated adipocytes, adenylyl cyclase activity was also measured in response to maximal concentrations of -adrenergic, G-protein, or adenylyl cyclase effectors (Table III). As mentioned above, in TNF--treated cells, there was a 2.6- and 2.7-fold reduction in norepinephrine and epinephrine-stimulated maximal adenylyl cyclase activity, respectively. Interestingly, response to the ₃-AR agonist CGP12177 was decreased by 8-fold in parallel. Additionally, TNF- exposure caused a 1.8- and 2.2-fold reduction in GTPS- and NaF-stimulated adenylyl cyclase activity, respectively. By contrast, TNF- did not cause a significant reduction in cAMP production in response to an optimal concentration of forskolin.

Table III. Effect of TNF- on adenylyl cyclase activity stimulated by -AR, G-protein, or adenylyl cyclase effectors Membranes were obtained from 3T3-F442A adipocytes treated or not with TNF- (1 nM for 24 h). Adenylyl cyclase activity in response to an optimal concentration of each indicated effector was determined. The percentage of reduction in stimulated adenylyl cyclase activity measured in TNF--exposed cells as compared to control adipocytes is indicated in parentheses after the Vmax value. Results are expressed as means ± S.E. of four to five independent experiments. Basal adenylyl cyclase activity was 17.7 ± 3.2 and 28.8 ± 4.4 pmol of cAMP/min/mg of protein in control and TNF--treated cells, respectively. *, p < 0.02; **, p < 0.005; ***, p < 0.0001, TNF--treated versus control cells. Effector Adenylyl cyclase activity (over basal) Control TNF- pmol cAMP/min/mg protein Norepinephrine (100 µM) 94.8  ± 9.8 36.4  ± 3.4* (62%) Epinephrine (100 µM) 98.6  ± 13.0 36.8  ± 2.9* (63%) CGP12177 (100 µM) 29.5  ± 2.8 3.7  ± 1.1*** (87%) GTPS (100 µM) 80.1  ± 1.1 44.7  ± 10.2* (44%) NaF (10 mM) 71.4  ± 5.7 32.5  ± 7.1** (54%) Forskolin (100 µM) 302.3  ± 16.4 262.0  ± 20.3 (13%)

Finally, though alterations in G-protein function and/or expression could play a significant role in the TNF--induced uncoupling to the adenylyl cyclase system, our results indicate that the modulation by the cytokine in -AR sensitivity involves, at least in part, receptor-mediated events. Thus, during TNF- exposure, adenylyl cyclase activity in response to the ₃-AR agonist CGP12177 is much more suppressed than the enzyme activity measured in the presence of the nonselective -AR agonists norepinephrine and epinephrine, suggesting that ₃-AR down-regulation induced by the cytokine contributes to the reduced -AR sensitivity. Overall, the differential regulation of -AR subtypes by TNF- results in shifts in the potencies of norepinephrine and epinephrine to stimulate adenylyl cyclase.

We next decided to investigate the interference of TNF- exposure on lipolysis, a key biological process of adipocytes that is under the control of -adrenergic/cAMP pathway. After a 24-h treatment by 1 nM TNF-, glycerol production was tested in response to increasing concentrations of norepinephrine and epinephrine. As previously well documented (15-17), TNF- enhanced by ~4-fold basal lipolytic activity (Table IV). Maximal lipolytic activity in response to norepinephrine or epinephrine was only weakly reduced by TNF- treatment. However, catecholamine-stimulated glycerol production over basal production was reduced by about 2-fold after cytokine exposure. Interestingly, the essential consequence in -AR subtype expression switch was the decreased potency of norepinephrine to activate glycerol production in TNF--treated cells (EC₅₀ values were 7.7 and 45.2 nM for control and TNF--treated cells, respectively).

Table IV. Effect of TNF- treatment on the lipolytic activity of 3T3-F442A adipocytes Lipolysis experiments were performed directly on adherent mature 3T3-F442A adipocytes exposed or not to 1 nM TNF- for 24 h. Glycerol production was tested in response to increasing concentrations of norepinephrine or epinephrine. EC50 value (in nM) corresponds to the concentration giving half-maximal stimulation. Vmax value (nmol of NADH/2 h/well) is the maximal agonist-stimulated lipolytic activity. Catecholamine-stimulated glycerol production over basal glycerol production is indicated in parentheses after each Vmax value. Results are expressed as mean ± S.E. of four separate experiments. *, p < 0.05; **, p < 0.001, TNF--treated versus control adipocytes. Culture condition Basal Agonist Norepinephrine Epinephrine EC50 Vmax EC50 Vmax Control 24.5  ± 2.4 7.72  ± 0.93 246  ± 13 (221.5) 16.25  ± 2.70 241  ± 15 (216.5) TNF- 97.2  ± 3.7** 45.20  ± 9.68* 210  ± 5* (112.8) 29.50  ± 7.95 192  ± 3.8* (94.8)

Effect of TNF- on -AR Subtype Gene Transcription and mRNA Stability

The differential regulation by TNF- of -AR subtype mRNA levels resulted either from modulation of the transcription rate of the related genes, from the control of transcript stability or combination of both. To assess the molecular mechanisms at the basis of these heterologous regulations, nuclear run-on experiments were performed in control and TNF--treated adipocytes. In nuclei isolated from 3T3-F442A adipocytes exposed for 30 min, 1 h, or 2 h to TNF-, transcription rates (after standardization to -actin signals) of the ₂-AR gene were respectively increased by 2.1-, 8.9-, and 6.2-fold as compared with the rates of control cells (Fig. 7). By contrast, in the presence of the cytokine, the transcription rate of the ₃-AR gene was quickly and potently reduced by 98% at the same intervals.

Effect of TNF- on -AR subtype gene transcription.

We also examined ₁-, ₂-, and ₃-AR mRNA half-life in 3T3-F442A adipocytes exposed to an inhibitor of transcription, actinomycin D, in the absence or in the presence of TNF-. The kinetics of -AR subtype mRNA disappearance was then assayed by RT-PCR analysis over a 6-h period (Fig. 8). It was noticeable that in control adipocytes, mRNA half-life of the three -ARs was short, being in the 80-100-min range. Half-life of ₁- and ₃-AR mRNA in TNF--treated cells (t1/2 = 91 ± 26 min and 111 ± 7 min for ₁- and ₃-AR mRNA, respectively) was not different from that measured in control adipocytes (t1/2 = 81 ± 13 min and 102 ± 12 min for ₁- and ₃-AR mRNA, respectively). By contrast, the half-life of ₂-AR mRNA was 2.5-fold longer in TNF--treated cells than that observed in control adipocytes (t1/2 = 84 ± 16 min and 215 ± 38 min in control and TNF--exposed cells, respectively; n = 3, p < 0.02).

Effect of TNF- on ₁-, ₂- and ₃-AR mRNA stability.

DISCUSSION

Exposure to TNF- has been demonstrated to cause a suppression of fat cell gene expression and, in some cases, a dedifferentiation of 3T3-L1 adipocytes (13, 37), TA1 adipocytes (3), or human preadipocytes in culture (6). This effect of TNF- on the differentiation process has been related to its inhibitory action on transcription factors that trigger adipogenesis, like C/EBP- (13, 14, 38, 39) or PPAR- (40). Since the ₃-AR can be considered as a marker of adipose conversion of murine 3T3 cells (18, 19), it is questionable whether the potent down-regulation of this receptor subclass by TNF- reflects a dedifferentiation of 3T3-F442A cells. Several lines of evidence indicate that ₃-AR modulation by the cytokine actually corresponds to a specific modulation of this receptor. First, TNF- effect on the ₃-AR mRNA or coupling is detectable at a low dose of 10 pM, a concentration that does not cause phenotypic changes in 3T3-F442A adipocytes, even after a chronic exposure (41). Second, ₃-AR mRNA repression occurs within few hours, a long time before the onset of adipocyte delipidation and dedifferentiation, and even before PPAR- and C/EBP- down-regulation reported by others (13, 14, 39, 40). This is confirmed by the invariance after a 24-h exposure to TNF- of G3PDH activity, a marker reflecting the magnitude of adipose conversion (42). Finally, the pattern of -AR subtype regulation by TNF- is not univocal; in contrast to the ₃-AR, the ₂-AR, which also slightly emerges during the differentiation process (43), is strongly induced during TNF- exposure, thus underlining that the cytokine specifically regulates -AR subtype expression.

In our study, run-on assays and measurement of mRNA stability demonstrate that TNF- regulates ₁- and ₃-AR mRNA content by a transcriptional mechanism. By contrast, the increase in ₂-AR mRNA transcript abundance is related both to an increase transcriptional activity of the ₂-AR gene and a stabilization of the corresponding mRNA. Further studies have to be performed to determine the molecular mechanisms at the basis of this transcriptional regulation by TNF-.

Since species and cell type differences can influence the regulation of receptors, we cannot yet extrapolate to human adipocytes the results presently observed in murine adipocytes. For instance, the pattern of ₃-AR desensitization appears to be cell-specific (44, 45). TNF- is able to slightly induce ₂-AR expression in human lung cells (46), but its effect on the -adrenergic system of human adipocytes is still unknown. Moreover, species differences have been described in the regulation of the ₃-AR by glucocorticoids, cAMP, or phorbol esters (47). Otherwise, species differences also exist between the levels of expression of the three -AR subtypes. It must be emphasized that, while the ₃-AR plays a pivotal role in rodent white or brown adipose tissue, it is expressed at much lower levels in humans (48-50). Recent studies performed in vivo have shown that the ₃-AR is able to mediate lipolysis in subcutaneous fat in humans (51, 52), but its role appears secondary as compared with that of ₁- or ₂-ARs. It is thus likely that a potential regulation by TNF- of the ₃-AR in human adipocytes would not cause marked functional consequences. By contrast, an induction of the ₂-AR by TNF- in human adipocytes might potentiate the -adrenergic control of all cAMP-dependent biological processes. In other terms, the initial expression level of each -AR subtype in white or brown adipose tissue is certainly a major determinant that influences the final effect of TNF- on -AR responsiveness. In white fat cells, catecholamines and cAMP acutely activate lipolysis but also exert a negative control on several enzymes of the lipogenic pathway (53), thus decreasing lipid stores. Up-regulation of the ₂-AR occurring in diseases accompanied by TNF- overproduction, such as cancer or infection, could thus contribute to the wasting effect of the cytokine. Further investigations are now required to evaluate -AR subtype expression in adipose tissue of subjects displaying high plasma levels of TNF- or other cytokines, and to determine the direct effect of TNF- on human preadipocytes or adipocytes in culture.

Several recent studies have demonstrated a connection between TNF- and insulin resistance observed in obesity and non-insulin-dependent diabetes mellitus. The cytokine is overexpressed in adipose tissues from genetically obese rodents (41, 54, 55) and in human obesity (56, 57). The causal relationship between TNF- and insulin resistance has been demonstrated in vivo by TNF- neutralization that increases insulin sensitivity in the obese insulin-resistant Zucker fa/fa rat (41). Subsequent studies have indicated that, through the p55 TNF receptor (58), TNF- can interfere with the insulin signaling pathway by impairing insulin-induced tyrosine phosphorylation of both the insulin receptor and insulin receptor substrate-1 (IRS-1) (59-63). It has been shown that a serine phosphorylation of IRS-1 (64, 65) converts this protein into an inhibitor of the insulin receptor tyrosine kinase activity. In contrast with these investigators, Stephens et al. (66) have very recently suggested that the primary mechanism for TNF--induced insulin resistance was rather the reduced mRNA transcription of several genes involved in the insulin-stimulated glucose transport. On the other hand, several rodent models of obesity express low levels of ₃-AR mRNA (35, 67-70) in white or brown adipose tissue. Moderately low levels of ₁-AR mRNA are also found in white adipose tissue of the congenitally obese ob/ob mouse (35). By contrast, in this mouse model of obesity ₂-AR mRNA abundance remains unchanged in white adipose tissue and increases in brown adipose tissue (35). This pattern of -AR subtype regulation in animal models of obesity and insulin resistance is reminiscent of the differential regulation of ₁-, ₂-, and ₃-ARs by TNF- observed presently in 3T3-F442A adipocytes. Thus, it is tempting to speculate that the cytokine is involved, at least in part, in the peculiar profile of -AR subtype expression observed in rodent obesity. Studies of ₁-, ₂-, and ₃-AR expression after immunoneutralization of TNF- in an animal model of obesity and insulin resistance would be helpful to ascertain the physiological relevance of this observation.

As regards the central role of -ARs as mediators of catecholamine-stimulated lipolysis and thermogenesis and of catecholamine-inhibited lipogenesis, it is possible that during obesity, the decreased expression of ₁- and ₃-ARs could in turn exacerbate the obese phenotype. Otherwise, it has been suggested that a -adrenergic stimulation, mediated by ₂-ARs, is partially responsible for the peripheral insulin resistance in animals infused with TNF- (71). Thus, the increase in ₂-AR expression caused by the cytokine in adipocytes could contribute to the reduction in insulin-stimulated glucose uptake. Whatever the role of the -adrenergic system in the onset of the TNF--induced insulin resistance, the increase in ₂-AR expression enhances the potency of epinephrine as compared with norepinephrine for stimulating the adenylyl cyclase system. As regards the dual innervation and vascularization of white and brown adipose tissues, this switch in -AR subtype expression may privilege the vascular (i.e. by epinephrine) over the nervous control (i.e. by norepinephrine) of energy expenditure in situations known to increase levels of cytokine expression.