Tumor Necrosis Factor α Inhibits Insulin-induced Mitogenic Signaling in Vascular Smooth Muscle Cells

Tumor necrosis factor α (TNFα) interferes with insulin signaling in adipose tissue and may promote insulin resistance. Insulin binding to the insulin receptor (IR) triggers its autophosphorylation, resulting in phosphorylation of Shc and the downstream activation of p42/p44 extracellular signal-regulated kinase 1/2 mitogen-activated protein kinase (ERK1/2), which mediates insulin-induced proliferation in vascular smooth muscle cells (VSMC). Since insulin resistance is a risk factor for vascular disease, we examined the effects of TNFα on mitogenic signaling by insulin. In rat aortic VSMC, insulin induced rapid phosphorylation of the IR and Shc and caused a 5.3-fold increase in activated, phosphorylated ERK1/2 at 10 min. Insulin induced a biphasic ERK1/2 activation with a transient peak at 10 min and a sustained late phase after 2 h. Preincubation (30–120 min) with TNFα had no effect on insulin-induced IR phosphorylation. In contrast, TNFα transiently suppressed insulin-induced ERK1/2 activation. Insulin-induced phosphorylation of Shc was inhibited by TNFα in a similar pattern. Since mitogenic signaling by insulin in VSMC requires ERK1/2 activation, we examined the effect of TNFα on insulin-induced proliferation. Insulin alone induced a 3.4-fold increase in DNA synthesis, which TNFα inhibited by 48%. TNFα alone was not mitogenic. Inhibition of ERK1/2 activation with PD98059 also inhibited insulin-stimulated DNA synthesis by 57%. TNFα did not inhibit platelet-derived growth factor-induced ERK1/2 activation or DNA synthesis in VSMC. Thus, TNFα selectively interferes with insulin-induced mitogenic signaling by inhibiting the phosphorylation of Shc and the downstream activation of ERK1/2.

Insulin has pleiotropic actions to regulate cell growth, differentiation, and metabolism (1). These varied biological effects of insulin result from the activation of a wide array of intracellular signaling proteins involved in multiple postreceptor pathways (for reviews, see Refs. 1 and 2). Binding of insulin to its cognate receptor (IR) 1 triggers autophosphorylation of the IR, which activates a tyrosine kinase domain in the cytoplasmic tail of the ␤-subunit. The activated IR tyrosine kinase associates with and phosphorylates a variety of intracellular substrates including Shc and IRS-1, which function as docking sites for Src homology 2 domain-containing proteins that propagate insulin's signal downstream. In most cell types, insulin signaling branches off into two pathways that lead to either the activation of ERK1/2 or phosphatidylinositol 3-kinase (PI 3-kinase), respectively. Activation of either pathway has been implicated for the mitogenic effect of insulin in different cell types, whereas metabolic responses elicited by insulin are more closely linked to the PI 3-kinase pathway (3).
The ERK1/2 pathway appears to be critical for insulin's actions in the vasculature, where insulin enhances the mitogenic activity of other growth factors and is itself a mitogen for VSMC (4 -7). Insulin induces ERK1/2 activation in VSMC through the activation of the Raf-1 3 MAPK/ERK kinase 3 ERK1/2 cytosolic protein kinase cascade, which we have recently shown to be required for insulin-stimulated VSMC growth (7). These findings are supported by studies in other cell types showing that insulin-induced mitogenic signaling is linked to the activation of the ERK1/2 pathway (8,9).
TNF␣ has been shown to inhibit insulin-induced ERK1/2 activation in L6 skeletal muscle cells, C2C12 muscle cells, and 3T3L1 adipocytes (10 -12). In addition to its effects on insulin signaling through the ERK1/2 pathway, TNF␣ has been demonstrated to induce serine/threonine phosphorylation of IRS-1 and a concomitant inhibition of insulin-stimulated IR and IRS-1 tyrosine phosphorylation and PI 3-kinase activation in adipocytes (11,(13)(14)(15)(16). The inhibitory effects of TNF␣ on insulin signaling in adipocytes appear to be more relevant for metabolic events than mitogenesis in these terminally differentiated cells. In other cell types with greater proliferative capacity than adipocytes, TNF␣ might interfere with mitogenic signaling by insulin. VSMC proliferation is a critical event in the development and progression of vascular lesion formation. Since insulin stimulates VSMC proliferation, we examined the effect of TNF␣ on insulin signaling in VSMC.

EXPERIMENTAL PROCEDURES
Materials-Dulbecco's modified Eagle's medium, glutamine, antibiotics, HEPES, Dulbecco's modified Eagle's medium, and monoclonal antibody against smooth muscle ␣-actin were obtained from Sigma; rat recombinant tumor necrosis factor ␣ was from R&D systems (Minne-* This work was supported by Deutsche Forschungsgesellschaft Grant DFG GO 800/1-1 and National Institutes of Health Grant HL58328-03. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Cell Culture-Rat aortic smooth muscle cells were prepared from thoracic aortas of 2-3-month-old Harlan Sprague-Dawley rats using the explant technique (17) and were cultured as described previously (18). When used, PD98059 or wortmannin were added 30 min prior to the following treatment protocol. For all data shown, each individual experiment represented in the n value was performed using an independent preparation of VSMC.
Western Blot Analysis-For protein analysis, early passaged (5 or less) VSMC were grown to 60 -70% confluency and then starved for 24 h in 0.4% fetal bovine serum/Dulbecco's modified Eagle's medium. After various incubation periods with insulin (1 mol/liter), TNF␣ (100 units/ ml), or PDGF-BB (10 ng/ml), cells were lysed, and protein concentrations were determined as described previously (18). For immunoprecipitations, 200 g of protein in 200 l of radioimmune precipitation buffer were incubated overnight with rabbit polyclonal antibodies (1:100) against insulin receptor ␤, IRS-1, or Shc at 4°C. Then 40 l of protein-Sepharose G beads (Amersham Pharmacia Biotech) were added to each sample, followed by incubation for 2 h at 4°C. After that, the samples were centrifuged at 14,000 rpm for 30 s, and pellets were washed several times with radioimmune precipitation buffer and resuspended in 50 l of 1ϫ SDS buffer. Equal amounts of protein were then separated by SDS-polyacrylamide gel electrophoresis and immunoblotting against tyrosine-phosphorylated protein or phosphorylated ERK1/2 was performed as previously reported (18). All Western blot experiments were repeated at least three times with a different cell preparation.
BrdUrd Incorporation-To examine DNA synthesis, incorporation of the thymidine analogue BrdUrd was measured using the BrdUrd labeling and detection kit II from Roche Molecular Biochemicals (18). Quiescent VSMC were incubated with insulin (1 mol/liter), PDGF (20 ng/ml), or TNF␣ (100 units/ml) in serum-free medium. When used as inhibitors, TNF␣ (100 units/ml), PD98059 (30 mol/liter), or wortmannin (50 nmol/liter) was added 30 min prior to stimulation with insulin or PDGF. Controls were kept in serum-free medium. Cell nuclei incorporating BrdUrd appeared brown and were counted in 4 -6 different high power fields/well.
Statistics-Analysis of variance was performed for statistical analysis, and p values less than 0.05 were considered to be statistically significant. Data are expressed as mean Ϯ S.E.

TNF␣ Transiently Inhibits Insulin
Signaling through ERK1/2 in VSMC-In adipocytes, skeletal muscle cells, and other cell types, TNF␣ has been shown to interfere with insulin signaling, leading to inhibition of ERK1/2 and PI 3-kinase activation. In those studies, suppression of insulin signaling was observed after cells were pretreated with TNF␣ for periods ranging from 10 min to 4 days (10, 11, 13-16, 19, 20).
To investigate the effects of TNF␣ on insulin signaling in VSMC, quiescent cells were stimulated either directly with TNF␣ (100 units/ml) and/or insulin (1 mol/liter) or with insulin after preincubation with TNF␣ for various times. Activation and phosphorylation of ERK1/2 was assessed by immunoblotting with a phosphospecific ERK1/2 antibody. Quiescent cells in the control group exhibited low ERK1/2 activity as evidenced by the faint bands detected at 44 and 42 kDa, corresponding to the ERK2 and ERK1 MAPKs (Fig. 1). Stimulation for 10 min with either TNF␣ or insulin alone led to a 2.1-or 5.3-fold induction of ERK1/2 activity (ERK1 ϩ ERK2), respectively. When added simultaneously, TNF␣ plus insulin activated ERK1/2 by 7.1-fold compared with control, demonstrating their additive effect on this pathway. Pretreatment with TNF␣, however, led to a transient suppression of insulin-induced ERK1/2 activation by 94% at 30 min and 89% at 60 min. After longer exposure to TNF␣, ERK1/2 activation in response to insulin was restored (4.8-fold after 90 min, 5.1-fold after 120 min). Since insulin at 1 mol/liter may bind to both IR and IGF-1 receptors and induce downstream signaling events, we also tested the effect of TNF␣ pretreatment on ERK1/2 activity after stimulation with IGF-1. After pretreatment with TNF␣, we observed a comparable inhibition of IGF-1-induced ERK1/2 activation as in insulin-stimulated cells (data not shown), demonstrating that TNF␣ suppresses ERK1/2 activation resulting through both the IR and the IGF-1 receptor.

TNF␣ Does Not Affect Tyrosine Phosphorylation of the Insulin Receptor (IR) or IRS-1 but Inhibits Insulin-induced Shc
Phosphorylation in VSMC-Insulin-induced ERK1/2 activation is a signaling event that occurs downstream of the IR tyrosine kinase autophosphorylation. In adipocytes, hepatoma cells, and fibroblasts, TNF␣ blocks insulin signaling by inhibiting IR tyrosine kinase activity, which leads to impaired tyrosine phosphorylation of IRS-1, thereby blocking downstream signaling events (11,13,15,16,19,20). The IR tyrosine kinase also directly phosphorylates Shc, which in turn associates with the adaptor proteins Grb2 and SOS that activate p21 ras and thereby initiates the sequential activation of the Ser/Thr kinases Raf-1, MAPK/ERK kinase, and ERK1/2 (2). We therefore examined the effects of TNF␣ on insulin-induced tyrosine phosphorylation of the IR, IRS-1, and Shc.
VSMC were stimulated with insulin and/or TNF␣ for 5 min with or without pretreatment with TNF␣ (30 -120 min). The insulin receptor ␤-subunit (IR␤), IRS-1, or Shc was immunoprecipitated, and phosphotyrosine-specific antibodies were used for immunoblotting to detect phosphorylated tyrosines on IR␤, IRS-1, or Shc. Levels of tyrosine phosphorylation of IR␤ or IRS-1 were below detection in quiescent cells or in VSMC stimulated with TNF␣ alone (Fig. 2). Treatment with insulin induced an increase in tyrosine-phosphorylated IR␤ and IRS-1 by 1.9-and 2.3-fold, respectively, that was not affected by costimulation or preincubation with TNF␣. The suppression of insulin signaling by TNF␣ in VSMC, therefore, occurs down- stream of the tyrosine phosphorylation of the IR. This is in contrast to the reported effects in adipocytes and skeletal muscle, suggesting that TNF␣ suppresses insulin signaling in VSMC by a distinct and novel mechanism.
In contrast to the lack of an effect on IR␤-and IRS-1-phosphorylation, we observed a dramatic inhibition of insulin-induced Shc tyrosine phosphorylation after pretreatment with TNF␣ (Fig. 3). While quiescent cells in the control group did not show detectable levels of tyrosine-phosphorylated Shc, stimulation with either TNF␣ or insulin alone for 5 min resulted in a 2.3-or 2.5-fold induction of Shc phosphorylation, respectively. Interestingly, an additive stimulation of Shc phosphorylation was observed when TNF␣ and insulin were added simultaneously (3.8-fold as compared with control), similar to their effect on ERK1/2 activation. Pretreatment with TNF␣ led to a transient suppression of insulin-induced tyrosine phosphorylation of Shc with a pattern similar to that observed for insulinstimulated ERK1/2 activity, causing an inhibition by 71% at 30 min and 47% at 60 min. After exposure to TNF␣ for 90 min, insulin-stimulated Shc phosphorylation was restored.
The suppression of insulin-stimulated ERK1/2 activation by TNF␣ in VSMC, therefore, occurs at the level of Shc phosphorylation, which represents a new target for TNF␣ in interfering with insulin signaling. Shc protein levels did not change in response to insulin or TNF␣, alone or in combination, for incubation periods of 5-90 min (Fig. 3B). Moreover, there was no apparent shift in the mobility of Shc, which suggests that TNF␣ did not induce substantial phosphorylation at sites other than tyrosine (Fig. 3B).
TNF␣ Inhibits Insulin-induced VSMC Proliferation-To determine the effect of TNF␣ on mitogenic signaling by insulin, quiescent cells were stimulated for 24 h with insulin (1 mol/ liter) and TNF␣ (100 units/ml) either alone or in combination, and VSMC proliferation was measured by BrdUrd incorporation (Fig. 4). Insulin alone induced a 3.37-Ϯ 0.22-fold increase in VSMC DNA synthesis, which was inhibited by 47.8 Ϯ 7.2% when TNF␣ was added 30 min before insulin and maintained throughout the experiment (p Ͻ 0.05). TNF␣ alone had no significant effect on mitogenesis. Blocking the ERK1/2 pathway with PD98059 (30 mol/liter) also inhibited insulin-stimulated BrdUrd incorporation (57.2 Ϯ 9.3% inhibition; p Ͻ 0.05), whereas wortmannin had only a modest effect that did not reach statistical significance. These results are in accord with our previous observations that insulin-induced mitogenic signaling in VSMC occurs predominantly through the ERK1/2 pathway (7). In combination, these data indicate that TNF␣ suppression of insulin-induced DNA synthesis probably occurs through its blockade of ERK1/2 activation. TNF␣ Suppresses the Biphasic ERK1/2 Activation by Insulin-Insulin-induced ERK1/2 activity in VSMC is biphasic. The first peak of ERK1/2 activity occurs as a rapid, transient activation of ERK1/2 within 10 min, which declines at 20 min and reaches a plateau activity above control at 30 min (Fig. 5A). Two hours after insulin stimulation, there is a late phase with a second, sustained peak of ERK1/2 activity (Fig. 5A), which remains elevated above quiescent cell levels for up to 6 h (data not shown). A variety of growth factors induce a biphasic activation of ERK1/2 in VSMC (21)(22)(23)(24), and a recent study has linked PDGF-induced VSMC proliferation to the late phase of ERK1/2 activity (25).
The effect of TNF␣ on the biphasic induction of ERK1/2 by insulin is shown in Fig. 5B. Cells were pretreated with TNF␣ for 30 min, and insulin-induced ERK1/2 activation followed from 0 to 180 min. Pretreatment with TNF␣ for 30 min suppresses not only the early phase but rather the full biphasic ERK1/2 response induced by insulin. Suppression of the late phase of ERK1/2 activation is consistent with our data showing that TNF␣ inhibits mitogenic signaling by insulin.
Stimulation of ERK1/2 and DNA Synthesis by PDGF Is Not Affected by TNF␣-To examine whether TNF␣ suppresses signaling from another tyrosine kinase receptor family member FIG. 2. TNF␣ does not inhibit insulin-induced tyrosine phosphorylation of IR␤ and IRS-1. Quiescent VSMC were stimulated with insulin (1 mol/liter) and/or TNF␣ (100 units/ml) directly for 5 min or after pretreatment with TNF␣ for various times (30 -120 min). The IR␤ or IRS-1 was then immunoprecipitated from cell lysates, and Western blot analysis with phosphotyrosine-specific antibodies was performed for detection of tyrosine-phosphorylated IR␤ or IRS-1. Immunoblotting to detect tyrosine phosphorylated IR␤ and IRS-1 was done in three separate experiments.

FIG. 3. TNF␣ inhibits insulin-stimulated tyrosine phosphorylation of Shc.
Quiescent VSMC were stimulated with insulin and/or TNF␣ as described in Fig. 2. Then Shc protein was immunoprecipitated followed by immunoblotting with phosphotyrosine-specific antibodies (A). For analysis of total Shc, equal amounts of protein were immunoblotted with an antibody directed against Shc protein (B). The Western blots shown are representatives of three experiments using different cell preparations. The densitometric data are expressed as mean Ϯ S.E., *, p Ͻ 0.05 versus insulin alone. that also activates ERK1/2, we investigated the effects of the cytokine on PDGF-induced signal transduction in VSMC. PDGF also induces a biphasic activation of ERK1/2. In addition,PDGF-inducedmitogenesisinVSMCisknowntobeERK1/2dependent (25). Quiescent VSMC were stimulated with TNF␣ (100 units/ml) and/or PDGF-BB (20 ng/ml) for 10 min with or without preincubation with TNF␣. As demonstrated in Fig. 6A, TNF␣ had no effect on PDGF-induced ERK1/2 activity. Consistent with this result, PDGF-induced VSMC proliferation (5.8 Ϯ 0.55-fold versus control) was also not affected by TNF␣ (Fig. 6B). Similar to our observations for insulin, PD98059 attenuated PDGF-stimulated BrdUrd incorporation by 43.4 Ϯ 8.5% (p Ͻ 0.05), demonstrating that mitogenic signaling through the PDGF pathway also requires activation of ERK1/2. These data demonstrate that TNF␣ suppression of insulin signaling is not the result of a general effect common to all tyrosine kinase growth factor receptor pathways.

DISCUSSION
The main finding of this study is that TNF␣ suppresses insulin-induced Shc phosphorylation and the downstream ERK1/2 activation, an important postreceptor branch of the insulin signaling cascade involved in cell growth. Our data provide the first evidence that mitogenic signaling by insulin is suppressed by TNF␣. This is probably a direct consequence of TNF␣ blocking insulin-induced ERK1/2 activation at the level of Shc phosphorylation by the IR. We have recently shown that stimulation of VSMC proliferation by insulin is ERK1/2-dependent (7). This finding is confirmed in this study, where insulin-stimulated mitogenesis was significantly inhibited by the pharmacological MAPK kinase inhibitor PD98059. TNF␣ suppressed insulin-induced VSMC proliferation to a similar extent as it inhibited insulin-activated ERK1/2.
A variety of growth factors induce a biphasic ERK1/2 activation in VSMC (21)(22)(23)(24). PDGF-induced VSMC proliferation has recently been linked to the late phase of ERK1/2 activation (25). Consistent with these findings, we have previously shown that TNF␣ stimulates a rapid and transient activation of ERK1/2 but fails to stimulate DNA synthesis in VSMC (18). In . Cell extracts were prepared, and equal amounts of protein (30 g) were separated on 7.5% SDS-PAGE gels, followed by immunoblotting with a phosphospecific ERK1/2 MAPK antibody. To better illustrate the biphasic kinetics of ERK1/2 activation, the autoradiograms depicted in Fig. 6, A and B, were exposed for longer periods of time than the one shown in Fig. 1A. The Western blots shown are representative of three independently performed experiments.
FIG. 6. TNF␣ has no effect on PDGF-induced signaling in VSMC. A, cells were made quiescent by serum starvation and were stimulated for 10 min with PDGF-BB (20 ng/ml) and/or TNF␣ (100 units/ml) with or without TNF␣ pretreatment. Cell extracts were prepared and samples that had been stimulated for 10 min were analyzed for activated, phosphorylated ERK1/2 by immunoblotting with a phosphospecific ERK1/2 antibody. B, VSMC proliferation in response to PDGF (20 ng/ml) was assessed by BrdUrd incorporation as described in the legend to this study, we demonstrate that pretreatment with TNF␣ suppresses the entire biphasic ERK1/2 response to insulin, which probably accounts for its inhibition of insulin-stimulated growth. This inhibitory effect of TNF␣ on insulin-stimulated mitogenic signaling, however, does not reflect a general antimitogenic action of the cytokine in VSMC, since PDGF-induced ERK1/2-dependent DNA synthesis was not suppressed.
Two major pathways have been identified that can lead to the activation of ERK1/2 after insulin binds to the IR. Tyrosine phosphorylation of either IRS-1 or Shc by the IR creates docking sites for the Src homology 2 domain in the adaptor protein Grb2. IRS-1⅐Grb2 or Shc⅐Grb2 complexes then bind to the Ras effector protein SOS, a guanyl nucleotide exchange protein, through the Grb2 Src homology 3 domains that recognize proline-rich motifs. SOS activates Ras by exchanging GTP for GDP bound to the G protein, which can trigger the downstream Raf-1 3 MAPK/ERK kinase 3 ERK1/2 protein kinase cascade.
Depending on the cell type or tissue examined, IR-mediated tyrosine phosphorylation of either IRS-1 or Shc can activate ERK1/2. Recent findings in IRS-1-deficient mice report that insulin signaling to ERK1/2 in skeletal muscle is mediated primarily through tyrosine phosphorylation of IRS-1 by the insulin receptor (26). Strikingly different results were obtained using fibroblasts from the same animal model where insulininduced ERK1/2 activation was unaffected by the absence of IRS-1 (27). In these cells, IR phosphorylation of Shc was proposed to be the major route for insulin's activation of the Ras 3 ERK1/2 pathway. The degree to which insulin activates ERK1/2 through either IRS-1 or Shc or through both may vary significantly among different cell types. Postreceptor mechanisms for insulin-induced ERK1/2 activation are poorly characterized for VSMC. Our finding that TNF␣ did not affect insulin-stimulated tyrosine phosphorylation of IRS-1 but inhibited phosphorylation of Shc and activation of ERK1/2 suggests that Shc predominantly regulates the Ras 3 ERK1/2 pathway in VSMC. The underlying mechanism for the transient suppression of insulin signaling to ERK1/2 via Shc by TNF␣ in VSMC remains to be determined. TNF␣-receptor down-regulation, desensitization, and/or decay of inhibitory signaling molecules could account for this phenomenon.
The inhibition of insulin-induced mitogenic signal transduction in VSMC by TNF␣ underscores the pluripotency of the cytokine in inhibiting insulin signaling in different tissues. This is even more interesting, since TNF␣ appears to affect different steps in the signaling cascades that are activated by insulin in different cell types, thereby targeting different cell functions. Studies in C2C12 muscle cells and L6 skeletal muscle cells reported that TNF␣ inhibited both insulin-stimulated ERK1/2 activation and glucose uptake. However, no activation of ERK1/2 by TNF␣ was observed, indicating a potentially different mechanism of action of TNF␣ in these cells (10 -12). Our experiments in VSMC did not reveal any impairment of insulin-induced tyrosine phosphorylation of IRS-1 by TNF␣, which is in marked contrast to several studies in adipocytes, hepatoma cells, skeletal muscle cells, and fibroblasts (11-16, 19, 20). In combination, these data indicate that although TNF␣ can interfere with insulin signaling in multiple cell types, different targets in the signaling cascades may be involved.
TNF␣ is a cytokine that is up-regulated in the vasculature under pathological conditions. Increased expression of the cytokine has been reported after balloon injury (28,29) and in restenotic lesions (30). Moreover, the presence of TNF␣ has been demonstrated in intimal VSMC (31) and plaques of atherosclerotic arteries (32,33), and although several TNF␣ activities in the vasculature are considered proatherogenic (34), a recent study identified accelerated atherosclerosis in mice lack-ing the p55-TNF␣ receptor, suggesting a protective effect of the cytokine in the development of atherosclerosis (35). VSMC proliferation is a critical step in the progression of restenosis and is probably also an important contributor to the pathogenesis of atherosclerosis. Hyperinsulinemia may be a risk factor for cardiovascular disease, possibly through direct effects on the vasculature (36). Here we show that TNF␣ interferes with mitogenic signaling of insulin through the ERK1/2 pathway and suppresses insulin-induced VSMC growth. This raises the possibility that under certain conditions the release of TNF␣ in the local tissue may actually exert protective effects on insulininduced changes in the arterial wall.