Transactivation of ErbB2 and ErbB3 by tumor necrosis factor-alpha and anisomycin leads to impaired insulin signaling through serine/threonine phosphorylation of IRS proteins.

The cellular pathways involved in the impairment of insulin signaling by cellular stress, triggered by the inflammatory cytokine tumor necrosis factor-alpha (TNF) or by translational inhibitors like cycloheximide and anisomycin were studied. Similar to TNF, cycloheximide and anisomycin stimulated serine phosphorylation of IRS-1 and IRS-2, reduced their ability to interact with the insulin receptor, inhibited the insulin-induced tyrosine phosphorylation of IRS proteins, and diminished their association with phosphatidylinositol 3-kinase (PI3K). These defects were partially reversed by wortmannin and LY294002, indicating that a PI3K-regulated step is critical for the impairment of insulin signaling by cellular stress. Induction of cellular stress resulted in complex formation between PI3K and ErbB2/ErbB3 and enhanced PI3K activity, implicating ErbB proteins as downstream effectors of stress-induced insulin resistance. Indeed, stimulation of ErbB2/ErbB3 by NDFbeta1, the ErbB3 ligand, inhibited IRS protein tyrosine phosphorylation and recruitment of downstream effectors. Specific inhibitors of the ErbB2 tyrosine kinase abrogated the activation of ErbB2/ErbB3 and in parallel prevented the reduction in IRS protein functions. Taken together, our results suggest a novel mechanism by which cellular stress induces cross-talk between two different signaling pathways. Stress-dependent transactivation of ErbB2/ErbB3 receptors triggers a PI3K cascade that induces serine phosphorylation of IRS proteins culminating in insulin resistance.

Insulin resistance is a common pathological state in which target cells fail to respond to ordinary levels of circulating insulin (6 -8). It is associated with obesity, type 2 diabetes, and conditions of acute and chronic stress like sepsis, advanced cancer, burn injury, and muscle damage (7)(8)(9)(10)(11)(12). The inflammatory cytokine tumor necrosis factor-␣ (TNF) has been implicated as the mediator of insulin resistance under these pathological conditions (10,13). Several studies have demonstrated that TNF confers insulin resistance by promoting phosphorylation of serine residues on IRS-1 and IRS-2 (14 -16). This impairs the interaction of the IRS proteins with the IR and diminishes their ability to undergo insulin-induced tyrosine phosphorylation (16). Moreover, serine-phosphorylated IRS-1 was found to inhibit insulin-induced tyrosine phosphorylation of the IR itself (15). The effects of TNF are mimicked by treating cells with sphingomyelinase (SMase) or after addition of cell-permeable ceramide analogs suggesting that TNF may utilize the sphingomyelin pathway to impair insulin action (16 -18).
In addition to TNF, other agents and conditions that induce cellular stress impair insulin action. These include oxidative agents (19), osmotic shock (20), and translational inhibitors like anisomycin (AN) (21). Recently it has been demonstrated that AN, similar to TNF, mediates serine phosphorylation of IRS-1 and a concomitant reduction of insulin signaling (22). However, the nature of the IRS kinases, stimulated by TNF or other stress inducers, remains unknown. TNF and AN activate several serine/threonine kinases, including ERK1/2, JNK, PI3K and IKK␤ (23,24), which were all implicated in mediating serine/threonine phosphorylation of IRS proteins. TNFinduced IRS phosphorylation in 3T3-L1 adipocytes is mediated by kinases of the MEK1/2-ERK1/2 pathway (22,25), whereas kinases of the PI3K/Akt/mTOR pathway mediate IRS phosphorylation in myotubes (26). TNF has also been shown to induce activation of IKK␤. Accordingly, inhibition of IKK␤ prevents serine phosphorylation of IRS proteins induced by high fat diet, TNF or phosphatase inhibitors, whereas it improves insulinstimulated tyrosine phosphorylation of IRS proteins, indicating that IKK␤ or its downstream effectors serve as IRS kinases (27). Finally, TNF induces complex formation between JNK and IRS-1, suggesting that JNK might also mediate TNFinduced serine phosphorylation of IRS proteins and insulin resistance (21,22). Thus, it appears that cellular stress stimulates multiple pathways leading to the induction of insulin resistance.
In the present study we provide evidence that TNF and other inducers of cellular stress like AN and cycloheximide utilize a common mechanism for the induction of insulin resistance, which involves a ligand-independent transactivation of ErbB2/ ErbB3, two members of the EGF receptor family (28,29). This activates PI3K and its downstream serine/threonine kinases, induces serine/threonine phosphorylation of IRS proteins, and promotes insulin resistance. Because NDF␤1, the natural ligand of ErbB3, acts in a similar manner, by stimulating PI3K activity and impairing insulin signaling, it suggests that transactivation of receptor tyrosine kinases by cellular stress might be the underlying cause for the induction of insulin resistance.  (30,31). Polyclonal IRS-2 antibody and the polyclonal antibody directed against the p85 regulatory subunit of PI3K (p85-PI3K) were obtained from Upstate Biotechnology Inc. (Lake Placid, NY). Monoclonal anti-phosphotyrosine antibody (PY-20) was obtained from Transduction Laboratories (Lexington, KY). Polyclonal anti-phospho-Akt (Ser 473 ) antibody was from New England BioLabs (Beverly, MA). Polyclonal anti-active ERK1/2 antibody was from Promega Corp. (Madison, WI). Polyclonal ErbB1, ErbB2, and ErbB3 antibodies were obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA).

Materials-Recombinant
Treatment of Fao Cells with Cellular Stressors-Rat hepatoma Fao cells were grown in RPMI 1640 medium supplemented with 10% fetal calf serum (Biological Industries, Beth Haemek, Israel). Confluent monolayers, grown in 60-mm dishes, were deprived of serum for 16 h prior to each experiment. Medium was removed, and the cells were incubated at 37°C with or without SMase, TNF, or the protein synthesis inhibitors CHX and AN in serum-free medium at the indicated concentrations and time intervals. Cells were then stimulated with or without 100 nM insulin for 1 min at 37°C. In some experiments, cells were treated with or without different signaling inhibitors (10 M PD98059, 100 nM wortmannin, 25 M LY294002, 2 M PD168393, or the tyrphostins AG825, AG213, and AG18 at the indicated concentrations) for 30 min prior to incubation with the stress stimuli and insulin. Medium was removed, and the cells were washed three times with phosphate-buffered saline (PBS) and frozen with liquid nitrogen. Cells were solubilized at 4°C with 0.4 ml/dish buffer A (25 mM Tris-HCl, pH 7.4, 2 mM sodium orthovanadate, 0.5 mM EGTA, 10 mM NaF, 10 mM sodium pyrophosphate, 80 mM ␤-glycerol phosphate, 25 mM NaCl, 10 mM MgCl 2 and 1 l/ml protease inhibitor mixture). The cells were scraped and after three freeze-thaw cycles were centrifuged at 12,000 ϫ g for 30 min at 4°C, and the supernatants were collected as the cytosolic fraction. The pellets were washed with buffer A, solubilized with 0.1 ml of buffer B (50 mM Tris-HCl, pH 7.4, 1% Nonidet P-40, 0.25% sodium deoxycholate, 150 mM NaCl, 1 mM EGTA, 1 mM sodium orthovanadate, 1 mM NaF, and 1 l/ml protease inhibitor mixture). After centrifugation at 12,000 ϫ g for 15 min, the supernatants were collected as the particulate fraction. Aliquots of the supernatants (cytosolic or particulate fractions) were normalized for protein, mixed with concentrated (5ϫ) Laemmli sample buffer (250 mM Tris-HCl, pH 6.8, 10% SDS, 10% ␤-mercaptoethanol, 40% glycerol, and 0.01% bromphenol blue), and boiled for 10 min. The solubilized proteins were separated by SDS-PAGE. After transfer to nitrocellulose membrane, the proteins were detected by immunoblotting with the indicated antibodies using ECL. In some experiments, cells were solubilized at 4°C with 0.4 ml/dish of buffer B, and the total cell extracts were treated as described above.
Immunoprecipitation from Fao Cell Extracts-Cells were treated with cellular stress stimuli as described above and solubilized with buffer B. Following centrifugation for 15 min at 12,000 ϫ g, aliquots (1 mg) of the supernatants were incubated with the indicated antibodies for 2 h at 4°C. The immunocomplexes were absorbed on a protein Gand protein A-Sepharose beads mixture (1:1) for 1 h of incubation at 4°C. The beads were collected, washed three times with buffer B, and boiled for 10 min in Laemmli sample buffer. The immunocomplexes were resolved on SDS-PAGE and immunoblotted with the appropriate antibodies as described above.
Treatment with Alkaline Phosphatase-Fao cells were treated with CHX, AN, SMase, or TNF as described above. Cytosolic extracts were prepared in buffer C (25 mM Tris-HCl, pH 7.4, 2 mM sodium orthovanadate, 0.5 mM EGTA, 25 mM NaCl, 10 mM MgCl 2 , and 1 l/ml protease inhibitor mixture). Aliquots (300 g/100 l) were incubated with 1000 units of alkaline-phosphatase for 1 h at 37°C. Following incubation, samples were mixed with concentrated (5ϫ) Laemmli sample buffer, boiled for 10 min, resolved by means of SDS-PAGE, and subjected to immunoblotting with IRS-1 or IRS-2 antibodies.
Binding of IRS-1 and IRS-2 to Immobilized Insulin Receptor-IR was purified from insulin-treated Fao cells on a WGA column as previously described (31). Immobilized IR was washed with buffer A prior to use. Cytosolic extracts derived from Fao cells treated with CHX, AN, SMase, or TNF were prepared in buffer A as described above. Supernatants were then incubated for 2 h at 4°C with the immobilized IR. Subsequently, the beads were washed three times with buffer A and were suspended in Laemmli sample buffer, boiled for 10 min, resolved on 7.5% SDS-PAGE, and subjected to western immunoblotting with IRS-1 and IRS-2 antibodies.
In Vitro Tyrosine Phosphorylation of IRS-1 by Insulin Receptor Kinase-IR derived from Fao cells was purified on a WGA column and eluted by 0.5 M N-acetyl-D-glucosamine as described previously (32). The soluble IR was mixed with concentrated (2ϫ) reaction mixture (40 mM Hepes, pH 7.5, 20 mM MgCl 2 , 20 mM MnCl 2 ) and treated for 30 min with 100 nM insulin prior to use. AN-treated Fao cells were extracted by buffer B and immunoprecipitated with IRS-1 antibodies as described above. The immunoprecipitated IRS-1 was resuspended in 100 l of reaction mixture (20 mM Hepes, pH 7.5, 10 mM MgCl 2 , 10 mM MnCl 2 , containing 1 mM ATP) and then incubated for 10 min at 22°C with in vitro insulin-pretreated IR. The reaction was stopped by addition of 2 l of 1 M EDTA. The beads were then washed three times with buffer B, mixed with Laemmli sample buffer, boiled for 10 min, resolved on 7.5% SDS-PAGE, and subjected to western immunoblotting with PY-20 antibodies.

Insulin Resistance Induced by Transactivation of ErbB2/ErbB3
further investigate the effect of cellular stress agents on insulin action, Fao cells were incubated with the translational inhibitors AN (50 ng/ml) or CHX (3 g/ml) for 30 min. Such treatment reduced the extent of insulin-induced tyrosine phosphorylation of IRS-1 and IRS-2 ( Fig. 1, A and B), reduced the association of IRS proteins with the p85 regulatory subunit of PI3K (Fig. 1, A and B) and reduced PI3K activity (not shown). The effect of AN was apparent at 25 ng/ml and reached a maximal effect at 100 ng/ml (Fig. 1C, upper panel), whereas the effect of CHX was apparent at 1 g/ml and reached a maximal effect at 10 g/ml (Fig. 1C, lower panel). The inhibitory effects of AN and CHX on insulin-stimulated tyrosine phosphorylation of the IRS proteins were rapid and could be detected already after 10-to 15-min incubation with 50 ng/ml AN or 3 g/ml CHX, respectively (Fig. 1D). Similar kinetics was demonstrated for TNF and SMase (14,17). In contrast, AN and CHX, similar to TNF and SMase (14,17), had only a small inhibitory effect on tyrosine phosphorylation of IR in Fao cells (Fig. 1E).
AN and CHX reduced the rate of migration of IRS-1 and IRS-2 during SDS-PAGE (Figs. 1 and 2A), and treatment of these cell lysates with alkaline phosphatase restored to normal the rate of migration of the IRS proteins ( Fig. 2A). Moreover, IRS-1, derived from AN-treated cells, regained its ability to undergo in vitro tyrosine phosphorylation by partially purified IR following treatment with alkaline phosphatase (Fig. 2B). These findings demonstrate that AN and CHX promote serine phosphorylation of IRS-1 and IRS-2 that inhibits their insulininduced tyrosine phosphorylation. In a previous study we have shown that enhanced serine phosphorylation of IRS proteins, induced by TNF or SMase, resulted in a reduction in their ability to interact with the IR (16). Similarly, CHX and AN caused a marked reduction (Ͼ40%) in the ability of IRS proteins to interact in vitro with the IR (Fig. 2C). These results suggest that serine phosphorylation of IRS proteins, induced either by inflammatory cytokines or by translational inhibitors, impairs the ability of IRS proteins to associate with the IR and thereby diminishes their insulin-induced tyrosine phosphorylation. It is noteworthy that at the low concentrations of AN and CHX used in this study, the levels of IRS proteins, analyzed with specific antibodies, remained unaltered (Figs. 3A and 5C). These findings are in accordance with previous reports, which have demonstrated that at low concentrations AN activates multiple serine/threonine kinases without inhibition of protein synthesis (33).

Insulin Resistance Induced by TNF or Anisomycin in Fao Cells Does Not Involve Activation of the MAPK Cascade-Re-
cent studies in 3T3-L1 adipocytes have shown that the MEK1/ 2-ERK1/2 pathway is involved in TNF-induced serine phosphorylation of IRS-1 that leads to its impaired tyrosine phosphorylation (22,25). To determine whether this pathway contributes to the serine phosphorylation of IRS proteins, induced by inflammatory cytokines or translational inhibitors, Fao cells were treated with the different stress inducers, with or without the addition of PD98059, a specific inhibitor of MEK, prior to incubation with insulin. As demonstrated in Fig.  3 (A and B, lower panels), AN and SMase and to a lesser extent CHX and TNF activated ERK1/2, and PD98059 effectively inhibited this ERK1/2 activation (Fig. 3, A and B, lower panels). However, PD98059 did not affect the reduction in phosphotyrosine content of IRS proteins or their mobility shift induced by the stress inducers. In contrast, PD98059 restored to normal the phosphotyrosine content of IRS proteins (Fig. 3C) and their mobility that were impaired upon treatment of Fao cells with TPA, a known inducer of MAPK-mediated serine phosphorylation of IRS proteins (16). These results indicate that in hepatoma Fao cells the inhibitory effect on insulin signaling, induced by TPA, is mediated through the MEK1/2-ERK1/2 pathway whereas the inhibition of insulin signaling by TNF and AN is mediated by different pathways. (26, 34 -36). Therefore, we tested in Fao cells whether these kinases are induced in response to cellular stressors. As shown in Fig. 4A, PI3K activity, assayed in p85-PI3K immunoprecipitates, increased 2-fold after 30-min incubation with TNF, AN, and CHX. These agents also induced the activation of Akt as indicated by the increase in its phosphoserine content (Fig. 4B). Accordingly, treatment of Fao cells with the specific PI3K inhibitors wortmannin (Fig. 5A) or LY294002 (Fig. 5B) increased the insulin-induced tyrosine phosphorylation of IRS-1 and IRS-2 and attenuated the ability of TNF, SMase, AN, and CHX to inhibit this process. Moreover, PI3K inhibitors attenuated the shift in the electrophoretic mobility of IRS-1 and IRS-2, as demonstrated for AN in Fig. 5 (B and C). These results indicate that inflammatory cytokines and translational inhibitors activate PI3K as part of the pathway leading to serine phospho-FIG. 1. Effect of AN and CHX on early events in insulin signaling. Fao cells were treated without or with 50 ng/ml AN or 3 g/ml CHX for 30 min at 37°C and then stimulated without or with 100 nM insulin for 1 min. Cell extracts were subjected to immunoprecipitation with IRS-1 (A and B, left), IRS-2 (A and B, right) or IR (E) antibodies. Immunocomplexes were resolved by means of 7.5% SDS-PAGE and immunoblotted with an anti-phosphotyrosine (P-Tyr) antibody (A and B, upper panels and E). The p85 subunit of PI3K that coprecipitated with IRS-1 or IRS-2 was detected by immunoblotting with a specific antibody (p85) (A and B, lower panels). Fao cells were incubated with the indicated doses of AN and CHX for 30 min (C) or with 50 ng/ml AN and 3 g/ml CHX for the indicated times at 37°C (D), prior to 1-min stimulation with 100 nM insulin. Cytosolic extracts were analyzed by Western immunoblotting using P-Tyr antibodies. Blots are representative of three independent experiments.

The PI3K Pathway Is Activated by TNF and Anisomycin in Fao Cells and Mediates the Inhibition of Insulin-induced Tyrosine Phosphorylation of IRS-1 and IRS-2-TNF and SMase activate PI3K and its downstream target Akt in various cells
rylation of IRS proteins and inhibition of their tyrosine phosphorylation.
The p85 Regulatory Subunit of PI3K Associates with ErbB2 and ErbB3 following Stimulation with Cellular Stressors-Activation of PI3K occurs following association of its regulatory subunit with phosphotyrosine residues of either activated growth factor receptors (37), or adaptor proteins like IRS proteins (2,5). To identify the cellular components that couple stress conditions to PI3K activation, Fao cells were treated for 30 min with AN, CHX, and SMase, after which PI3K was immunoprecipitated from cell lysates. Following SDS-PAGE, PI3K-associated proteins were probed with phosphotyrosine antibodies. One or more tyrosine-phosphorylated 200-kDa proteins could be detected in PI3K immunoprecipitates of the FIG. 2. AN and CHX induce serine phosphorylation of IRS proteins and impair their ability to interact with IR and to undergo tyrosine phosphorylation. Fao cells were incubated without (C) or with 50 ng/ml AN, 3 g/ml CHX, 300 mU/ml SMase (SM) or 5 nM TNF for 30 min at 37°C. Cytosolic extracts were prepared, and samples (300 g) were incubated with 1000 units of alkaline-phosphatase for 1 h at 37°C. Following incubation, the samples were subjected to 7.5% SDS-PAGE, and blots were probed with antibodies to IRS-1 or IRS-2. The blots are representative of three independent experiments (A). AN-treated cell extracts were subjected to immunoprecipitation with antibodies to IRS-1 as described in Fig. 1. The immunocomplexes were incubated without or with alkaline-phosphatase as described in A. Following extensive washes, immobilized IRS-1 was subjected to in vitro phosphorylation with insulin treated-IR as described under "Experimental Procedures." The immunocomplexes were separated by 7.5% SDS-PAGE, and blots were probed with P-Tyr antibodies (B, upper panels). Quantification of tyrosine-phosphorylated IRS-1 levels (mean Ϯ S.E.) based on scanning densitometry of four independent immunoblots is represented by the bar graphs (B, lower panels). Cytosolic proteins from Fao cells treated with AN, CHX, SMase, or TNF as described in A, were bound to immobilized IR as described under "Experimental Procedures." IR/IRS complexes were resolved by means of 7.5% SDS-PAGE and immunoblotted with IRS-1 or IRS-2 antibodies (C, upper panels). Quantification of IR/IRS complexes (mean Ϯ S.E.) based on scanning densitometry of three independent immunoblots is represented by the bar graphs (C, lower panel).

FIG. 3. Effect of PD98059 on translational inhibitors and inflammatory cytokines induced inhibition of insulin signaling.
Fao cells were treated with 10 M PD98059 for 30 min prior to incubation with 50 ng/ml AN, 3 g/ml CHX, 5 nM TNF (A), 300 mU/ml SMase (B), or 100 nM TPA (C) for 30 min at 37°C. Cells were further incubated with 100 nM insulin for 1 min. The cytosolic proteins were analyzed by 7.5% SDS-PAGE and immunoblotted with P-Tyr, IRS-1, or IRS-2 antibodies. The activation of ERK1 and ERK2 was analyzed by Western immunoblotting using phosphospecific-ERK1/2 antibodies. Blots are representative of three independent experiments.

FIG. 4. Inflammatory cytokines and translational inhibitors activate the PI3K pathway.
Fao cells were incubated with either 100 nM insulin for 1 min or 3 g/ml CHX, 50 ng/ml AN, 5 nM TNF for 30 min at 37°C. Immunoprecipitation was performed using p85-PI3K antibodies and the PI3K activity was measured as described under "Experimental Procedures." PI3K activity was quantitated by densitometry (A). In parallel, the activation of Akt was analyzed by Western immunoblotting with phosphospecific Akt antibodies (B). The blots are representative of three independent experiments. treated cells but not in control cells (Fig. 6A). These proteins were distinct from the 180-to 185-kDa IRS proteins that bind to PI3K in response to insulin (Fig. 6A). To further characterize the 200-kDa proteins, their intracellular localization was examined. The tyrosine-phosphorylated 200-kDa proteins were found in the membrane fraction (Fig. 6B) and were absent from the cytosolic faction (Fig. 1, C and D) of Fao cells treated with CHX, AN, TNF, and SMase. Their tyrosine phosphorylation was apparent already after 5-to 10-min incubation with 50 ng/ml AN, 300 milliunits (mU)/ml SMase (Fig. 6C), or with 3 g/ml CHX and 5 nM TNF (not shown). Interestingly, tyrosine phosphorylation of the 200-kDa protein(s) by cellular stressors (Fig. 6, C and D) and the inhibition of IRS proteins phosphorylation (Fig. 1, C and D) occurred with a similar time courses and dose responses.
Previous studies in Fao cells have demonstrated that SMase induces tyrosine phosphorylation of 200-kDa proteins (17,38) that were identified as ErbB2 and ErbB3 (38). Immunoprecipitation with specific antibodies revealed that ErbB2 and ErbB3 were indeed the 200-kDa proteins, which underwent tyrosine phosphorylation in response to cellular stress (Fig. 7A). Furthermore, AN and SMase induced the association of ErbB2 and ErbB3 receptors with the p85 subunit of PI3K (Fig. 7B), and this association led to a marked augmentation in PI3K activity (Fig. 7C). The preferred association of the p85 subunit of PI3K with ErbB3, compared with ErbB2, is consistent with the higher number of p85-binding sites on ErbB3 (28,29). Taken together, these results indicate that inflammatory cytokines or translational inhibitors, which induce cellular stress, transactivate ErbB2/ErbB3 receptors leading to activation of PI3K.
NDF␤1, the Natural Ligand of ErbB3, Impairs Insulin Sig-naling-The transactivation of ErbB2 and ErbB3 and the impairment of insulin signaling induced by cellular stressors raised the possibility that ErbB2/ErbB3 might mediate the inhibition of insulin action induced by cellular stress. To test this possibility, the effect of NDF␤1, a natural ligand of ErbB3, on insulin-induced tyrosine phosphorylation of IRS proteins, was studied. To this end, Fao cells were treated with 50 ng/ml NDF␤1 for 30 min prior to 1-min stimulation with 100 nM insulin. As expected, NDF␤1 treatment, which induces heterodimerization of ErbB3 with ErbB2 (28, 29), increased the tyrosine phosphorylation of ErbB2/ErbB3 proteins (Fig. 7A) and enhanced the PI3K activity associated with ErbB2/ErbB3 immunoprecipitates (Fig. 7D). Furthermore, NDF␤1 pretreatment reduced the electrophoretic mobility of IRS-1 and IRS-2 and inhibited their insulin-induced tyrosine phosphorylation and their association with the p85 subunit of PI3K (Fig. 8, A  and B). The reduced electrophoretic mobility of IRS-1 and IRS-2 could be attributed to NDF␤1-induced serine phosphorylation of these proteins, which was partially reversed by the PI3K inhibitor, LY294002 (Fig. 8C). Accordingly, LY294002 partially reversed the decrease in the insulin-induced tyrosine phosphorylation of IRS-1 and IRS-2 (Fig. 8C). These results support the notion that ErbB2/ErbB3-mediated activation of PI3K, induced either by cellular stress or by the natural ligand NDF␤1, leads to desensitization of insulin signaling.

Inhibition of ErbB2/ErbB3 Activation Repairs Stress-induced Defect in Insulin
Signaling-To further support the role of ErbB2/ErbB3 in impairment of insulin signaling, we studied the ability of AG825 and PD168393, two specific inhibitors of ErbB2 tyrosine kinase (39,40), to prevent the stress-induced impairment in IRS protein phosphorylation. Indeed, AG825

FIG. 5. Effect of wortmannin and LY294002, on stress-induced inhibition of insulin signaling.
Fao Cells were treated with 100 nM wortmannin for 30 min at 37°C prior to incubation with AN, CHX, SMase, or TNF as described in Fig. 2A. Cells were further incubated with 100 nM insulin for 1 min at 37°C. The cytosolic proteins were analyzed by 7.5% SDS-PAGE, and the blots were probed with P-Tyr antibodies (A). Cells were preincubated for 30 min at 37°C with 25 M LY294002 before stimulation with 50 ng/ml AN for 30 min and 100 nM insulin for 1 min at 37°C. Total cell extracts were subjected to immunoprecipitation using IRS-1 or IRS-2 antibodies, and the immunocomplexes were analyzed by Western immunoblotting using P-Tyr antibodies. The levels of tyrosine-phosphorylated IRS-1 and IRS-2 were quantitated by densitometry (B). In parallel, total cell extracts were analyzed by Western immunoblotting using IRS-1 and IRS-2 antibodies (C) or phosphospecific Akt antibodies (D). Blots are representative of four independent experiments.

FIG. 6. Association of PI3K with phosphotyrosine containing proteins following stimulation with inflammatory cytokines and the translational inhibitors.
Fao cells were incubated with either 100 nM insulin for 1 min or 3 g/ml CHX, 50 ng/ml AN, 300 mU/ml SMase for 30 min at 37°C. Total cells extracts were subjected to immunoprecipitation with p85-PI3K antibodies. Immunocomplexes were analyzed by 7.5% SDS-PAGE, and the blots were probed with P-Tyr antibodies (A). Fao cells were incubated with 3 g/ml CHX, 50 ng/ml AN, 5 nM TNF, or 300 mU/ml SMase for 30 min at 37°C. Proteins in the particulate fraction extracts were analyzed by Western immunoblotting using P-Tyr antibodies (B). Cells were incubated with 50 ng/ml AN or 300 mU/ml SMase at 37°C for the indicated times. Proteins in the particulate fraction extracts were analyzed as described above (C). Cells were stimulated with the indicated concentrations of AN for 30 min at 37°C. Particulate fractions extracts were prepared and analyzed as described above (D). Blots are representative of three independent experiments. blocked the serine phosphorylation and the reduction in electrophoretic mobility of IRS-1 and IRS-2 induced by AN and blocked the inhibitory effect of AN on insulin-induced tyrosine phosphorylation of the IRS proteins (Fig. 9A). The effects of AG825 were dose-dependent; they were observed with 10 M AG825 and reached maximal effect with 50 M (Fig. 9B). At these concentrations AG825 inhibited the phosphorylation of ErbB2/ErbB3 (Fig. 9C), the recruitment of PI3K to ErbB2/ ErbB3 (Fig. 9D), and the AN-induced Akt activation (not shown) but did not affect the insulin-induced tyrosine phosphorylation of IRS-1 and IRS-2 (Fig. 9, A and B). Other tyrphostins like AG18 or AG213, which are not specific for ErbB2, did not reverse the AN-induced defect on insulin signaling (not shown). Similar results were observed with PD168393, another inhibitor of the ErbB2 tyrosine kinase. At 2 M, PD168393 blocked AN-and SMase-induced activation of ErbB2/ErbB3 (Fig. 9E) and concomitantly attenuated the defect in the tyrosine phosphorylation of IRS proteins (Fig. 9F). The ability of selective inhibitors of ErbB2 to abrogate tyrosine phosphorylation of ErbB2 and ErbB3 proteins induced by cellular stress supports the notion that cellular stress induces heterodimerization of ErbB2/ErbB3 culminating in their tyrosine phosphorylation and the recruitment of PI3K, primarily to ErbB3. This induces serine phosphorylation of IRS-1 and IRS-2 proteins, which uncouples them from the insulin signaling machinery.

DISCUSSION
The present study provides evidence that cellular stress, triggered by the inflammatory cytokine TNF or by translational inhibitors such as AN and CHX, activates the ErbB2/ ErbB3 receptors, thereby leading to the activation of a PI3Kregulated pathway. This induces serine phosphorylation of IRS-1 and IRS-2 proteins and causes an impairment of insulin action.
Several lines of evidence support such a mechanism. First, we could demonstrate that TNF, SMase, CHX, or AN treatment activated ErbB2 and ErbB3. Moreover, activation of ErbB2 and ErbB3 paralleled the suppression in insulin signaling. Tyrosine-phosphorylated ErbB2/ErbB3 proteins could be detected within 10 min of TNF, SMase, CHX, and AN treatment, whereas the inhibition of insulin-induced tyrosine-phosphorylated IRS-1 and -2 followed a similar time course. Second, specific inhibitors of the ErbB2 tyrosine kinase, AG825 and PD168393, prevented both the stress-induced activation of ErbB2 and ErbB3 as well as the suppression of insulin signaling. The inhibition of tyrosine phosphorylation of ErbB3 by ErbB2 inhibitors indicates that ErbB2 activates both ErbB3 and PI3K. Thus, in response to cellular stress, ErbB2 transphosphorylates the kinase-impaired ErbB3 through heterodimerization, leading to recruitment and activation of PI3K (Fig. 7). Likewise, activation of ErbB2/ErbB3 heterodimers by NDF␤1, the natural agonist of ErbB3, mimics the inhibitory effect of TNF, SMase, and AN on insulin signaling. NDF␤1, similar to stress stimuli, induced PI3K activity through association of the p85 subunit of PI3K with tyrosine-phosphorylated sites, which reside predominantly on ErbB3 within the ErbB2/ErbB3 heterodimers. Inhibition of PI3K activity, either indirectly by specific inhibitors of ErbB2 or directly by wortmannin and LY294002, effectively prevented the serine phosphorylation of IRS proteins and the decrease in their insulininduced tyrosine phosphorylation. Taken together, these findings underline the significance of the ErbB2/ErbB3-stimulated PI3K pathway to the induction of insulin resistance by TNF and other stress stimuli.
IRS proteins play a central role in regulation of insulin action. Although phosphorylation of IRS proteins on tyrosine residues propagates insulin's signals (2,4), serine phosphorylation of IRS proteins has a dual function in serving either as a positive (32) or a negative modulator of insulin signaling (14 -16). Serine phosphorylation can induce the dissociation of IRS FIG. 7. Cellular stressors activate ErbB signaling pathways. Total cell extracts from 50 ng/ml AN, 300 mU/ml SMase, or 50 ng/ml NDF␤1-treated Fao cells were subjected to immunoprecipitation with ErbB1, ErbB2, or ErbB3 antibodies and analyzed by immunoblotting with P-Tyr (A) or p85-PI3K (B) antibodies. Corresponding immunocomplexes were analyzed for PI3K activity as described under "Experimental Procedures" (C and D). The blots are representative of four independent experiments.
FIG. 8. Effect of NDF␤1 on insulin signaling. Cells were incubated with 50 ng/ml NDF␤1 for 30 min at 37°C prior to 1-min stimulation with 100 nM insulin. IRS-1 and IRS-2 were immunoprecipitated with specific antibodies, and the immunocomplexes were resolved by means of 7.5% SDS-PAGE and immunoblotted with P-Tyr and p85-PI3K antibodies (A). In parallel, total cell extracts were analyzed by Western immunoblotting using IRS-1 and IRS-2 antibodies (B). Cells were preincubated with 10 M SB203580, 10 M PD98059, or 25 M LY294002 for 30 min at 37°C prior to stimulation with NDF␤1 and insulin as described above. Cytosolic proteins were resolved by means of 7.5% SDS-PAGE and immunoblotted with P-Tyr antibodies (C). Blots are representative of three independent experiments. proteins from the insulin receptor (IR), hinder tyrosine phosphorylation sites, release the IRS proteins from intracellular complexes that maintain them in close proximity to the receptor, induce IRS proteins degradation, or turn IRS proteins into inhibitors of the IRK. This inhibits insulin signaling and may induce a state of insulin resistance (reviewed in Ref. 41).
A major role to the induction of insulin resistance is attributed to cytokines like TNF (13) or to other cellular stressors such as AN, whose common feature is their ability to enhance serine phosphorylation and inhibit insulin-stimulated tyrosine phosphorylation of IRS proteins (14 -16, 21, 22, 25). This feature is also shared by several growth factors that inhibit insulin signaling through serine phosphorylation of IRS-1. Platelet-derived growth factor inhibits insulin-induced tyrosine phosphorylation of IRS-1 in 3T3-L1 adipocytes, and PI3K inhibitors abrogate this effect (42)(43)(44). In the same cells EGF reduces tyrosine phosphorylation of IRS-1 via an MEK-ERK1/ 2-dependent pathway (25). Recently, communication between cytokines and growth factors signaling systems has emerged as an important mechanism that enables cells to integrate a multitude of signals from its environment (28,29). Several studies have demonstrated that TNF transactivates the EGF receptor and employs its signaling pathways to transduce the cytokine's effects. For example, transactivation of the EGF receptor by TNF plays a pivotal role in TNF-induced NF-B activation (45), whereas TNF and interleukin-1␤ inhibit apolipoprotein B secretion via the EGF receptor signaling pathways (46).
The EGF/ErbB family of receptor tyrosine kinases includes the EGF receptor (ErbB1), ErbB2, ErbB3, and ErbB4 (28,29). Signaling requires the formation of homodimeric or heterodimeric complexes of these receptors. ErbB2 does not have a ligand-binding capacity, but it is the preferred dimerization partner for the other members. ErbB3 has a severely impaired intrinsic kinase activity; therefore heterodimerization is a crucial step for ErbB3-mediated signal transduction. It is noteworthy that ErbB3 has six putative phosphotyrosine binding sites for PI3K and potently activates this enzyme (28,29). The present study demonstrates that TNF and AN transactivate ErbB2 and ErbB3 and utilize their signaling pathways to in-hibit insulin signaling in Fao cells. The activation of ErbB2 and ErbB3 leads to stimulation of a PI3K-regulated pathway that mediates the inhibition of insulin signaling through serine phosphorylation of IRS-1 and IRS-2. This phosphorylation impairs the association of the IRS proteins with the IR and thereby diminishes their ability to undergo insulin-induced tyrosine phosphorylation and to further recruit downstream effectors.
The nature of the IRS kinases located downstream of PI3K that induce insulin resistance is under extensive investigation. The PI3K/Akt/mTOR pathway was shown to mediate TNF inhibition of insulin signaling through phosphorylation of serines 636 and 639 of IRS-1 (26). In addition, mTOR-mediated phosphorylation on serines 632, 662, and 731 of IRS-1, induced by platelet-derived growth factor in 3T3-L1 adipocytes, was found to inhibit insulin-stimulated tyrosine phosphorylation of IRS-1 and its ability to bind PI3K (44). Inhibition of tyrosine phosphorylation of IRS proteins by AN and SMase in Fao cells is blocked by rapamycin, 2 an inhibitor of mTOR, suggesting that a PI3K/Akt/mTOR pathway, stimulated through ErbB2/ ErbB3 activation, is required to mediate the inhibitory effects of these cellular stressors. However, because SMase, a downstream effector of TNF, increases the cellular content of ceramide, TNF might also trigger a ceramide-activated kinase such as PKC that might serve as an IRS kinase (16,17). Indeed, PKC, a downstream effector of PI3K, is activated upon prolonged insulin treatment of cultured cells leading to IRS phosphorylation (47,48), which serves as a negative feedback control mechanism to terminate insulin signaling (41). Recent studies shifted the spotlight to IKK␤, a downstream target of PKC (27,49). Inhibition of IKK␤ by high doses of salicylates prevents serine/threonine phosphorylation of IRS proteins induced by TNF, high fat diet, or phosphatase inhibitors, whereas it improves insulin-stimulated tyrosine phosphorylation of IRS proteins, indicating that IKK␤ or its downstream effectors serve as IRS kinases. The c-Jun NH 2 -terminal kinase 2 R. Hemi and H. Kanety unpublished observations. FIG. 9. Effect of AG825 and PD168393, specific inhibitors of ErbB2 kinase, on stress-induced inhibition of insulin signaling. Cells were preincubated with 50 M AG825 for 30 min at 37°C, prior to stimulation with 50 ng/ml AN for 30 min at 37°C. Cells were then stimulated with 100 nM insulin for 1 min at 37°C. IRS-1 and IRS-2 were immunoprecipitated with specific antibodies and the immunocomplexes were analyzed by Western immunoblotting using P-Tyr antibodies (A). Cells were incubated with increasing doses of AG825 prior to incubation with AN and insulin as described above. Cytosolic proteins were resolved by means of 7.5% SDS-PAGE and immunoblotted with P-Tyr antibodies (B). Cells were preincubated with 50 M AG825 before stimulation with 50 ng/ml AN. ErbB2 and ErbB3 were immunoprecipitated with specific antibodies and the immunocomplexes were analyzed by Western immunoblotting with P-Tyr (C) or p85-PI3K (D) antibodies. Cells were preincubated with 2 M PD168393 for 1 h at 37°C prior to stimulation with 50 ng/ml AN or 300 mU/ml SMase for 30 min at 37°C. Cells were then stimulated with 100 nM insulin for 1 min at 37°C. Particulate (E) and cytosolic (F) extracts (50 g) were resolved by means of 7.5% SDS-PAGE and immunoblotted with P-Tyr antibodies. Blots are representative of four independent experiments.
(JNK) promotes insulin resistance by associating with IRS-1 and phosphorylating Ser-307, which inhibits insulin-stimulated tyrosine phosphorylation of IRS-1 (21). However, JNK itself is unlikely to serve as an insulin-or TNF-stimulated IRS kinase, because its activity is insensitive to inhibitors that block phosphorylation of Ser-307 in response to these stimuli (22). Taken together, these studies suggest that several serine/ threonine kinases, downstream of PI3K, are potential IRS kinases. Activation of PI3K cascades either by growth factors like insulin, platelet-derived growth factor, and NDF␤1 or by inflammatory cytokines and translational inhibitors like TNF and AN, respectively, may lead to insulin resistance through IRS phosphorylation by these kinases.
The mechanism(s) underlying the activation of ErbB proteins by TNF and AN remain poorly understood. TNF transactivates the ErbB1 receptor in a thiol-sensitive manner in NIH3T3 cells, and this enhances NF-kB activity (45). Similarly, we have found that the TNF-and AN-induced transactivation of ErbB2/ErbB3 in Fao cells is suppressed by N-acetyl-L-cysteine. 3 This antioxidant prevented the activation of the PI3K/Akt pathway by the cellular stressors and restored insulin sensitivity. 3 These findings suggest that reactive oxygen intermediates may play a role in the activation of ErbB2/ErbB3 by TNF and AN, thus leading to PI3K activation and IRS proteins phosphorylation. Several studies have suggested that reversible inactivation of protein tyrosine phosphatases by stress stimuli (e.g. UV radiation, hyperosmotic shock, and TNF), due to oxidation of reactive cysteine residues in their catalytic domain, modulate the activation state of ErbB1 or ErbB2 receptors (29,50,51). Alternatively, activation of nonreceptor tyrosine kinases was suggested to mediate the activation of ErbB proteins by different cellular stressors (52). The tyrosine kinase Pyk2 has been implicated in the activation of ErbB2/ErbB3 by SMase in Fao cells (38), whereas Src kinases were suggested to mediate ErbB1 activation by H 2 O 2 (53) and heat shock (54). Attempts to identify the upstream pathways that link TNF and AN to ErbB2/ErbB3 transactivation are currently underway.
Stress conditions like acute infection, burn, or trauma are associated with alterations in carbohydrate metabolism that include enhanced glucose uptake and utilization, increased glucose production, depressed glycogenesis, glucose intolerance, and insulin resistance (55) and are part of the acute phase response to infection and invasive stimuli (56). These effects, which are mediated in part by TNF (10 -13), provide a mechanism by which the energy demands of the wound and tissues active in the immune response to stress are satisfied. They enable the synthesis of acute phase reactant proteins and cellular proliferation within the immune system and injured tissue (55,56). Our findings, that TNF and additional cellular stressors transactivate the ErbB proteins on the one hand and inhibit insulin signaling on the other, demonstrate a possible cellular basis for the metabolic alterations during whole body stress. ErbB stimulation by cellular stress leads to activation of PI3K/Akt signaling, which is important for cell survival and may promote the acute phase response, at least in part through activation of NF-B (35,45). NF-B up-regulates immune and cell survival genes (57). However, activation of PI3K and its downstream serine/threonine kinases, which is important for cell survival, leads to insulin resistance through inhibition of IRS proteins function. Thus, in the presence of inflammation or cellular stress, cells would divert their cell machinery and metabolism from controlled nutrient transport and anabolism to that of survival and repair.