Tumor Necrosis Factor Receptor-1 Can Function through a Gα_q/11-β-Arrestin-1 Signaling Complex^*

Chronic low grade inflammation is a critical feature of a number of common, non-immune diseases such as insulin resistance/obesity, type 2 diabetes mellitus, certain forms of cancer, and neurodegenerative disorders. The cytokine, TNFα³, is a key signaling molecule that activates proinflammatory effects in target cells. TNFα binds to the TNF receptor-1 (TNF-R1) and receptor-2 (TNF-R2) and stimulates the NFκB pathway through complexes that includes the TRAF, RIP (receptor-interacting protein), and TRADD (TNF-R-associated death domain) families (1, 2). Numerous studies have reported TNFα signaling to NFκB activation; however, the mechanisms underlying TNFα stimulates ERK, cdc42, or PI 3-kinase actually remain incompletely understood. It is well known that chronic TNFα treatment can lead to insulin resistance both in vitro (3) and in vivo (4), and elevated levels of TNFα have been reported in obese and/or diabetic individuals. Thus, further insights into the mechanisms of TNFα signaling are of interest. Recent findings reported that TNF-R1 is observed in plasma membrane microdomains, termed lipid rafts and/or caveolae (5, 6), in which various G-proteins, 7-transmembrane (TM) receptors (7-TMRs), and many other signaling components are collected. For example, the endothelin-1 (ET-1) receptor ETAR, a G_q-coupled 7-TM receptor, is detected in caveolae, and insulin receptors are also found in this membrane structure (7). Interestingly, it has been shown that the insulin receptor can couple into the heterotrimeric G-protein Gα_q/11 (8) and that chronic ET-1 treatment can induce insulin resistance by heterologous desensitization of Gα_q/11 signaling (9, 10).

Based on these facts, we sought to determine whether TNFα signaling could couple with G-protein components to mediate biologic effects. In this report we show that in 3T3-L1 adipocytes, TNFα signals through the TNF-R1·TRAF2 receptor complex and that this can couple into a trimeric G-protein Gα_q/11 and an adaptor protein β-arrestin-1 to mediate TNFα activation of ERK, lipolysis, and stimulation of the proinflammatory pathway.

EXPERIMENTAL PROCEDURES

Materials—Mouse monoclonal anti-phosphotyrosine (PY-20) antibody was purchased from BD Biosciences. Horseradish peroxidase-linked anti-rabbit, -mouse, -goat antibodies, anti-c-Jun N-terminal kinase (JNK), -cdc42, TRAF2, -p110α subunit of PI 3-kinase, and -pan-Src polyclonal antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Phospho-specific anti-ERK and -JNK antibodies were from Cell Signaling, Inc. TNF-R1 antibody and mouse- and human-TNFα were from R&D Systems (Minneapolis, MN). Mouse TNFα was used in the experiments unless otherwise specified. Anti-Gα_q/11 antibody, Src-kinase inhibitor (PP2), PD98059, LY294002, and toxin-B were from Calbiochem. [γ-³²P]GTPγ-azidoanilide was from Affinity Labeling Technologies, Inc. (Lexington, KY). RT-PCR primers were from Invitrogen, and the oligonucleotide sequences used were: matrix metalloproteinase 3 (MMP3) forward, 5′-TGACCCCACTCACTTTCTCC-3′; MMP3 reverse, 5′-GCCTTGGCTGAGTGGTAGAG-3′; monocyte chemoattractant protein-1 (MCP-1) forward, 5′-AGCACCAGCCAACTCTCAC-3′; MCP-1 reverse, 5′-TCTGGACCCATTCCTTCTTG-3′; TNF-R1 forward, 5′-GACCGGGAGAAGAGGGATAG-3′; TNF-R1 reverse, 5′-CACGCACTGGAAGTGTGTCT-3′; TNF-R2 forward, 5′-AAATGCAAGCACAGATGCAG-3′; TNF-R2 reverse, 5′-TCCTGGGATTTCTCATCAGG-3′. Cell culture materials and other radioisotopes were from ICN (Costa Mesa, CA). All other reagents were purchased from Sigma.

siRNA Transfection—3T3-L1 adipocytes (day 8 after differentiation) were transiently transfected by electroporation (GENE PULSER, Bio-Rad) with 2 nmol of siRNA duplex, as previously described (11). 24–120 h after electroporation, the efficiency of each siRNA was analyzed by RT-PCR or Western blotting every 24 h, and the time point for the maximum efficiency of each siRNA was determined. All experiments with siRNA were performed 48–96 h after electroporation or microinjection. SiGENOME SMARTpool siRNA against TRAF2 and all other custom siRNAs were purchased from Dharmacon Research Inc. (Lafayette, CO). The target sequences of the siRNAs were follows: β-arrestin 1-A, 5′-GGCCTGTGGTGTGGATTAT-3′; β-arrestin 1-B, 5′-AGCCTTCTGTGCTGAGAAC-3′; Gα_q-A, 5′-GCTGGTGTATCAGAACATC-3′;Gα_q-B, 5′-TCCATATGTAGATGCAATA-3′;Gα₁₁-A, 5′-ACTCACACTTGGTCGATTA-3′;Gα₁₁-B, 5′-GTTGGTGTACCAGAACATC-3′; cdc42, 5′-GTTATCCACAGACAGATGT-3′; TNF-R1, 5′-GTATGTCCATTCTAAGAAC-3′; TNF-R2, 5′-ACTCCAAGCATCCTTACAT-3′. Each one of the paired siRNAs (A and B) was used separately, and this confirmed the absence of off-target effects. All sequences were confirmed to have no homology to any other genes by BLAST search (NCBI, National Institutes of Health).

Photoaffinity GTP-loading Assay—Photoaffinity GTP-labeling of membrane-associated G-proteins was performed as we described previously (12) with some modifications. Membrane fractions were semi-purified from scraped 3T3-L1 adipocytes (1 × 10⁸ cells) by Dounce homogenization and centrifugation (18,000 × g, 10 min) in a buffer containing 50 mm HEPES (pH 7.4), 1 mm EDTA, 1 mm Na₃VO₄, 15 mm NaF, 1 mm phenylmethylsulfonyl fluoride and then immediately incubated with 10 μm [γ-³²P]GTP-azidoanilide (12.2 mCi/μmol) for 3 min followed by UV irradiation (254 nm) for 1 min on ice. Samples were lysed with the addition of Nonidet P-40 (final concentrations, 1%), then immunoprecipitated with anti-Gα_s, Gα_q/11, or Gα_i antibody. Immunoprecipitates were resolved by SDS-PAGE, and signals were quantitated by PhosphorImager (GE Healthcare).

Inositol Phosphate Production Assay—3T3-L1 adipocytes were incubated with myo-[³H]inositol (ICN, Costa Mesa, CA) (2 μCi/ml) in inositol-free Dulbecco's modified Eagle's medium for 16 h before assay. After washing cells twice with serum-free Dulbecco's modified Eagle's medium, cells were stimulated with TNFα (20 ng/ml) for 15 min. The accumulation of total ³H-labeled (IP1 and IP2 plus IP3) was measured as previously described (13).

cdc42 Assay—cdc42 activity was measured according to the manufacturer's instructions (Upstate Biotechnology, Inc., Lake Placid, NY). 3T3-L1 adipocytes were starved for 16 h and stimulated with 17 nm insulin or 20 ng/ml TNFα for the indicated time periods, washed once with ice-cold phosphate-buffered saline, and lysed with lysis buffer containing 25 mm HEPES (pH 7.5), 150 mm NaCl, 1% IGEPAL CA-630, 10 mm MgCl₂, 1 mm EDTA, 10% glycerol, 1 mm Na₃VO₄, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 25 mm NaF for 15 min at 4 °C. Insoluble materials were removed by centrifugation. For a negative control cell lysate was incubated with 1 mm GDP for 15 min at 30 °C. 5 μg of p21-activated kinase 1-agarose beads, which specifically bound to active cdc42, were added to the cell lysates and incubated for 1 h at 4 °C. Agarose beads were washed with lysis buffer three times and boiled in Laemmli sample buffer. Samples were resolved by SDS-PAGE and immunoblotted with anti-cdc42 antibody.

Lipolysis Assay—Triglyceride hydrolysis (lipolysis) was measured using free glycerol reagent (Sigma) according to the manufacturer's specifications. Briefly, 3T3-L1 adipocytes were washed twice and incubated with L-buffer (Dulbecco's modified Eagle's medium without phenol red plus 0.1% bovine serum albumin) then stimulated with or without TNFα (20 ng/ml) for 16 h. To measure lipolysis, 150 μl of free glycerol assay reagent was incubated with 10 μl of culture supernatants and glycerol standards for 15 min, and the absorbance was read at 540 nm. A standard curve constructed from the glycerol standards was used to calculate the concentration of glycerol in the culture supernatants.

Cell Culture and Other Methods—3T3-L1 adipocyte cell culture (8), immunoprecipitation with dithiobis(succinimidyl propionate) cross-linking (12), Src kinase assay (14), RT-PCR and PI 3-kinase assay (15), arachidonic acid release assay (16), and cyclic AMP assay (17) were performed as we previously described.

RESULTS

The G-protein Signaling Component, β-Arrestin-1, Couples with the TNF Receptor Complex—β-Arrestin-1 is a scaffold protein that functions in a number of 7-TMR (typical G protein-coupled receptor) signaling pathways, and TRAF2 serves as an adaptor protein for TNFα signaling. To assess the relationship between these systems, we examined the association between β-arrestin-1 and TRAF2 in FLAG-tagged β-arrestin-1 expressing 3T3-L1 adipocytes. As shown in Fig. 1A, β-arrestin-1 was detected in anti-TRAF2 antibody immunoprecipitates in the basal state, and this was greatly reduced after TNFα stimulation (^*, p < 0.05; ^**, p < 0.01). This result indicates that β-arrestin-1 associates with TNF receptor complexes in the basal state and dissociates from this complex after TNFα stimulation. In the 7-TMR systems, β-arrestin-1 binds to Src kinase, and ligand stimulation results in Src-kinase activation (18). To see if this was also the case for TNFα, we measured Src-kinase activity, and we detect that, like ET-1, TNFα acutely stimulated Src-kinase activity (2.5-fold above basal), as shown in Fig. 1B. Using RNA interference, electroporation with siRNA against β-arrestin-1 (βar1-A or βar1-B) decreased protein expression in 3T3-L1 adipocytes (Fig. 1C), and we found that acute stimulation of Src kinase activity with TNFα was inhibited by siRNA knockdown of β-arrestin-1 (Fig. 1D). These results suggest that TNFα-induced Src-kinase activation is dependent on β-arrestin-1, similar to 7-TMR signaling.

Umemori et al. (19) reported that the activation of Gα_q/11 is coincident with phosphorylation of tyrosine residue 356 in the C terminus of Gα_q/11. In response to ET-1 stimulation, we have previously reported that Src kinase plays a key role, leading to tyrosine phosphorylation of Gα_q/11 (14). Accordingly, we examined the effects of TNFα on Gα_q/11 and found that TNFα acutely induced tyrosine phosphorylation of Gα_q/11 at 3 min after stimulation and that this effect was inhibited by siRNA-mediated β-arrestin-1 knockdown (Fig. 1E). Because two different siRNA against β-arrestin-1 (βar1-A or βar1-B in the upper panel) showed quite comparable results, this confirms that off-target effects do not contribute to our results. Furthermore, the Src-kinase inhibitor, PP2, also inhibited TNFα-induced Gα_q/11 tyrosine phosphorylation (Fig. 1F), similar to its effects to block ET-1-stimulated phosphorylation (14). Interestingly, PP2 did not inhibit insulin-induced Gα_q/11 phosphorylation, suggesting that the insulin receptor kinase itself can phosphorylate Gα_q/11 independent of Src kinase activity. These results indicate that β-arrestin-1 mediates the effect of TNFα to phosphorylate Gα_q/11 via the activation of Src-kinase.

TNFα Stimulates Gα_q/11 Activity without Leading to Phospholipase Cβ (PLC-β) Activation—To confirm the activation of Gα_q/11, we performed a GTP-loading assay using photoaffinity GTP labeling method, as described under “Experimental Procedures.” As shown in Fig. 2A, TNFα stimulation increased the amount of GTP bound to Gα_q/11 (3-fold) but had little effect on GTP loading of Gα_i or Gα_s. Consistent with this, TNFα stimulation did not lead to changes in the cellular cyclic AMP level (Fig. 2B). It is generally thought that activated Gα_q/11 stimulates PLC-β, leading to the induction of diacylglycerol, inositol phosphate-3, and subsequent arachidonic acid release. We measured the total amount of inositol phosphates (Fig. 2C, left panel) and arachidonic acid release (Fig. 2C, right panel) in 3T3-L1 adipocytes and found that TNFα stimulation interestingly did not increase either of these metabolites, suggesting that TNFα does not activate PLC-β. This is also observed in insulin receptor signaling (Fig. 2C), in which Gα_q/11 is activated (8), and importantly, both receptors do not have a 7-TM type structure. Several studies have revealed that the second and/or third intracellular loops of 7-TM receptors are required for PLC-β activation (20, 21), and these motifs are not present in the insulin and TNF receptors. Therefore, we hypothesized that non-7-TM type receptor cannot associate with PLC-β. We tested the co-immunoprecipitation with receptor antibody and found no co-immunoprecipitation of PLC-β3 (the main isoform of the PLC-β family in 3T3-L1 adipocytes) with anti-TNF-R1 antibody, although PLC-β3 was modestly co-immunoprecipitated with anti-ETAR-Ab after ET-1 treatment (Fig. 2D).

Biological Roles of β-Arrestin-1-Gα_q/11 as Signaling Mediators for TNFα Action—To investigate which biological effects of TNFα are mediated by β-arrestin-1/Src kinase and Gα_q/11, we first focused on JNK, a proinflammatory pathway mediator, since the mechanism of JNK activation is still unclear. Using RNA interference-induced knockdown of TNF-R1, TNF-R2, or TRAF2 (Fig. 3, A and B, left panel), we found that TNFα (20 ng/ml, 8 min) stimulated JNK phosphorylation and that this effect was strongly inhibited by TNF-R1 or TRAF2 knockdown but not by TNF-R2 knockdown (Fig. 3C), consistent with the view that a TNF-R1·TRAF2 complex mediates TNFα signaling to JNK activation. It has been reported previously that activation of JNK is mediated by cdc42 (22) and PI 3-kinase (23) and that TNFα can stimulate cdc42 (24) and PI 3-kinase activities (25), although the mechanisms for activation of these kinases are unclear. Because activated Gα_q/11 can lead to stimulation of cdc42 and cdc42-associated PI 3-kinase (26), we assessed this pathway with respect to TNFα stimulation. Using RNA interference showing in Figs. 1C and 3B, we found that TNFα-stimulated JNK phosphorylation was inhibited by β-arrestin-1, Gα_q/11, or cdc42 knockdown in 3T3-L1 adipocytes (Fig. 3D). Two different siRNAs against β-arrestin-1 (βar1-A in Fig. 3D, left panels, and βar1-B in rightside panels) or Gα_q/11 (G_q-A + G₁₁-A in Fig. 3D, left panels, and G_q-B + G₁₁-B in right side panels) showed quite similar results, indicating that there were no off-target effects in these siRNAs.

To confirm these signaling events, we tested the effect of kinase inhibitors on TNFα signaling. Consistent with the above results, TNFα-stimulated JNK activation was inhibited by pretreatment of the cells with the Src inhibitor (PP2), the PI 3-kinase inhibitor (LY294002, LY), or the cdc42 inhibitor (toxin-B, TB) but not by MEK inhibitor (PD98059, PD) (Fig. 4A). Furthermore, we confirmed that TNFα stimulates cdc42 activity (Fig. 4B) and cdc42-associated PI 3-kinase activity (Fig. 4C), and these effects were inhibited by PP2 treatment (^*, p < 0.01) (Fig. 4, B and C). Taken together, these data indicated that TNFα signaling goes through the TNF-R1·TRAF2 → β-arrestin-1-Src-kinase → Gα_q/11 → cdc42/PI 3-kinase pathway to activate JNK, a proinflammatory pathway in 3T3-L1 adipocytes.

It has been reported that TNFα-induced ERK activation was regulated by Src kinase (27), although the full mechanisms of ERK activation remain unclear. In 3T3-L1 adipocytes TNFα-induced ERK phosphorylation was maximal at 15 min after stimulation, and this effect was significantly inhibited by TNF-R1 or TRAF2 knockdown but not the results for TNF-R2 knockdown (Fig. 5A), similar to the results for JNK activation (Fig. 3B). Previous reports showed that human TNFα can bind to mouse TNF-R1 but not to mouse TNF-R2 (28), and this is fully consistent with the results in Fig. 5A, showing that the effects of human TNFα on ERK phosphorylation are also blocked by siRNA-mediated TNF-R1 knockdown. As seen in Fig. 5B, TNFα-induced ERK phosphorylation was decreased 68% by β-arrestin-1 knockdown (^*, p < 0.01), indicating that β-arrestin-1 mediates the effect of TNF-R1 signaling. Interestingly, neither Gα_q/11 knockdown nor inhibition of Gα_i with pertussis toxin treatment inhibited TNFα-stimulated ERK phosphorylation (Fig. 5B), demonstrating that this signaling pathway does not utilize these G-protein components. This is of particular interest since Figs. 1E and 2A showed that TNFα can lead to Gα_q/11 phosphorylation and activation, indicating a separation of signaling pathways which flow through β-arrestin-1 versus β-arrestin-1-Gα_q/11.

In other systems we (29) and others (30) have shown that β-arrestin-1 is an upstream regulator of ERK activity, and recent reports by Kang et al. (31) revealed that β-arrestin-1 can be translocated to the nucleus after ligand stimulation, where it participates in regulation of gene expression. Accordingly, we measured mRNA expression levels of two TNFα target genes, MMP3 and MCP-1, and found that expression of these genes was strongly stimulated by TNFα treatment (Fig. 5C). Because TNFα stimulation of both ERK and JNK1 appears to be β-arrestin-1-dependent and because both ERK and JNK1 can regulate gene expression, we investigated the effects of β-arrestin-1 knockdown and found that TNFα-induced MCP-1 and MMP3 expression was inhibited by depletion of β-arrestin-1 (^*, p < 0.03).

Another biologic effect of TNFα in adipocytes is stimulation of lipolysis. We found a role for β-arrestin-1 signaling in this process by showing that TNFα-induced lipolysis was partially inhibited by β-arrestin-1 knockdown (^*, p < 0.05) but not by Gα_q/11 knockdown, as seen in Fig. 5D, left panel. Because β-arrestin-1 mediates TNFα-induced ERK activation, we treated the cells with the MEK inhibitor PD98059 and found that this also partially inhibited TNFα-induced lipolysis (^*, p < 0.05) (Fig. 5D, right panel), consistent with a previous report (32).

DISCUSSION

TNFα stimulation induces a wide range of biologic effects related to innate immunity, inflammation, insulin resistance/diabetes, and other conditions. Chronic inflammation is increasingly recognized as a major cause of decreased insulin sensitivity. By activating pro-inflammatory pathways (33), TNFα can cause insulin resistance in vitro (3) and in animals treated with TNFα (4). The molecular mechanisms of TNFα action have been extensively studied, and many elements of this pathway have been defined (2). In this study we have identified new elements of a novel TNFα signaling pathway in adipocytes, which flows through a TNF-R1·TRAF2 complex and includes the use of the classical 7-TMR signaling components, β-arrestin-1, and Gα_q/11. We found that β-arrestin-1 can mediate TNFα effects on Src kinase and ERK activation and that Gα_q/11 can couple TNFα action to cdc42, PI 3-kinase, and JNK activation, leading to the induction of lipolysis and proinflammatory gene expression in 3T3-L1 adipocytes.

It is well known that β-arrestin-1 plays several roles in 7-TMR function, including receptor desensitization and internalization (34). Interestingly, it has also been reported that β-arrestin-1 interacts with non-7-TM type receptors, such as the insulin-like growth factor-I, epidermal growth factor, and insulin receptors (35), and here we show that the TNF-R1·TRAF2 complex can signal through β-arrestin-1 to mediate TNFα biological effects in 3T3-L1 adipocytes. As such, these studies further demonstrate the diverse functions of β-arrestin-1 as an adaptor protein mediating the signaling properties of receptor systems distinct from its classical role in 7-TMR action. Our studies also show that the G-protein component, Gα_q/11, can also function in the TNFα signaling pathway. To accomplish this task, TNFα signaling uses β-arrestin-1 effects to activate Src kinase, causing Gα_q/11 tyrosine phosphorylation and activation. Together these effects can mediate downstream signaling events such as activation of ERK, PI 3-kinase, and lipolysis.

We investigated the downstream TNFα effects transduced by β-arrestin-1 and/or Gα_q/11, and one of them is JNK activation. It is well known that JNK can propagate the effects of TNFα to induce proinflammatory gene expression, and this can result in insulin resistance. Previous studies have shown that TNFα-induced JNK activation is mediated by TRAF2 (36), cdc42 (22), Src (37), and PI 3-kinase (38). There may be multiple ways in which TNFα can cause JNK activation, and consistent with these reports, our findings show that TNFα can signal through β-arrestin-1/Src-kinase → Gα_q/11 → cdc42/PI 3-kinase to activate JNK. Similarly, previous studies revealed multiple mediators for TNFα-induced ERK activation, including TRAF2, MADD, and Src kinase (27, 39). Ligand stimulation can induce diverse signaling pathways leading to ERK activation; e.g. insulin-like growth factor-I receptor can signal through β-arrestin-1/Gα_i as well as SHC to activate ERK (30, 40). Consistent with this formulation, siRNA-induced β-arrestin-1 knockdown did not completely inhibit TNFα-induced ERK phosphorylation (Fig. 5B), indicating the presence of another pathway(s) from the TNF receptor to this biologic end point. Because NFκB pathway may also be involved in lipolytic effects (41), this may also explain why β-arrestin-1 knockdown only partially inhibits TNFα-stimulated lipolysis (Fig. 5D).

Previous studies have shown that β-arrestin-1 is an upstream mediator of ERK activation in various receptor signaling systems (42) and that β-arrestin-1 can translocate into the nucleus to exert transcriptional effects (31). In addition, activated JNK stimulates c-Jun, which translocates into the nucleus as a transcription factor activating pro-inflammatory gene expression (43). These findings suggested that β-arrestin-1 might play a role in TNFα-induced gene expression. Consistent with this, our results show that siRNA-knockdown of β-arrestin-1 exerts inhibitory effects on TNFα-stimulated MCP-1 and MMP3 gene expression.

TNFα stimulation induces a wide range of biologic effects related to innate immunity, inflammation, insulin resistance/diabetes, and other conditions. Chronic inflammation is increasingly recognized as a major cause of decreased insulin sensitivity, and by activating proinflammatory pathways (33), TNFα can cause insulin resistance in vitro (3) and in animals treated with TNFα (4). Last, it is well known that elevated circulating free fatty acid levels can be a cause of insulin resistance (44). In this study we have identified new elements in TNFα signaling pathways in adipocytes, which can lead to inflammatory pathway activation and lipolysis (Fig. 6). This pathway flows through a TNF-R1·TRAF2 receptor complex and includes the use of classical 7-TMR signaling components, β-arrestin-1 and Gα_q/11, showing a novel signaling paradigm for these molecules distinct from their traditional functions mediating signals from an array of 7-TMRs.