Differential activation of mitogen-activated protein kinase and S6 kinase signaling pathways by 12-O-tetradecanoylphorbol-13-acetate (TPA) and insulin. Evidence for involvement of a TPA-stimulated protein-tyrosine kinase.

AG-18, an inhibitor of protein-tyrosine kinases, was employed to study the role of tyrosine-phosphorylated proteins in insulin- and phorbol ester-induced signaling cascades. When incubated with Chinese hamster ovary cells overexpressing the insulin receptor, AG-18 reversibly inhibited insulin-induced tyrosine phosphorylation of insulin receptor substate-1, with minimal effects either on receptor autophosphorylation or on phosphorylation of Shc64. Under these conditions, AG-18 inhibited insulin-stimulated phosphorylation of the ribosomal protein S6, while no inhibition of insulin-induced activation of mitogen-activated protein kinase (MAPK) kinase or MAPK was detected. In contrast, 12-O-tetradecanoylphorbol-13-acetate (TPA)-induced activation of MAPK kinase and MAPK and phosphorylation of S6 were inhibited by AG-18. This correlated with inhibition of TPA-stimulated tyrosine phosphorylation of several proteins, the most prominent ones being pp114 and pp120. We conclude that Tyr-phosphorylated insulin receptor substrate-1 is the main upstream regulator of insulin-induced S6 phosphorylation by p70s6k, whereas MAPK signaling seems to be activated in these cells primarily through the adaptor molecule Shc. In contrast, TPA-induced S6 phosphorylation is mediated by the MAPK/p90rsk cascade. A key element of this TPA-stimulated signaling pathway is an AG-18-sensitive protein-tyrosine kinase.

p70 s6k is not activated by MAPK and appears to lie on a separate signaling pathway (16). One of its upstream activators is IRS-1, whose phosphorylation by insulin receptor kinase creates a binding site for the SH2 domains of the p85 regulatory subunit of phosphatidylinositol 3-kinase (PI3K) (17,18). The association between p85 and IRS-1 results in activation of PI3K (17,18) via a mechanism independent of the direct activation of PI3K by Ras (19). Activation of PI3K then stimulates p70 s6k by an as yet unknown mechanism (20,21).
A large variety of extracellular signals, aside from insulin, lead to ribosomal S6 phosphorylation. One example is the tumor-promoting phorbol ester (TPA) that activates the Ca 2ϩand phospholipid-dependent protein kinase C (22). Although protein kinase C has been implicated as playing an important role in insulin-induced activation of the MAPK cascade (23), other studies suggest that insulin (24) and insulin-like growth factor I (25) activate the MAPK cascade independent of protein kinase C.
To study the relative contribution of MAPK/p90 rsk and PI3K/ p70 s6k to insulin-and TPA-induced stimulation of S6 phosphorylation, we made use of tyrphostins, synthetic competitive inhibitors of several tyrosine kinases (see Ref. 26 for review) that inhibit insulin receptor kinase activity in vitro (27) and block insulin-induced lipogenesis and anti-lipolysis in fat cells (28). Employing Chinese hamster ovary (CHO) cells that overexpress the wild-type insulin receptor gene (CHO.T) (29), we found that AG-18 effectively inhibits insulin-induced IRS-1 phosphorylation as well as S6 kinase activity. On the other hand, insulin induction of the MAPK cascade is not affected by AG-18. The phorbol ester TPA also stimulates S6 kinase activity that is inhibited by AG-18, but unlike the insulin stimulation, this inhibition is correlated with the inhibition of the MAPK cascade. These results implicate a bifurcation in insulin signaling where IRS-1 mediates, to a large extent, S6 phosphorylation via p70 s6k , while Shc may mediate a signaling pathway leading to MAPK/p90 rsk activation. In contrast, the TPA effect on S6 phosphorylation seems to be transmitted via the MAPK/ p90 rsk pathway. Most important, a protein-tyrosine kinase, whose activity is inhibited by AG-18, is one of the elements linking protein kinase C to the MAPK cascade.
Cell Cultures-Chinese hamster ovary cells, transfected with a wildtype human insulin receptor gene (CHO.T (29)), were a generous gift of Dr. William J. Rutter (University of California, San Francisco). The cells were grown in F-12 medium as described previously (29).
Purification of Insulin Receptor-Insulin receptors were partially purified from rat liver plasma membranes. The preparation of membranes, solubilization in Triton X-100, and affinity chromatography of insulin receptors on wheat germ agglutinin coupled to agarose were carried out as described previously (32).
Phosphorylation of Exogenous Substrates-The reaction was carried out as described previously (33). Briefly, partially purified insulin receptors (50 -200 g/ml) were incubated in the presence or absence of 10 Ϫ7 M insulin (in 50 mM Hepes, 0.1% bovine serum albumin, 0.1% Triton X-100, pH 7.6, for 30 min at 22°C) and the indicated concentrations of AG-18 (dissolved in 10% Me 2 SO). Control tubes were incubated with Me 2 SO at a final concentration of 1%. Phosphorylation in a final volume of 100 l was initiated with 20 l of a reaction mixture to yield the following final concentrations: 50 M [␥-32 P]ATP, 1 mM CTP, 40 mM magnesium acetate, 0.1% Triton X-100, and 0.2 mg/ml poly(Glu,Tyr) (4:1). Reactions were allowed to proceed for 10 min at 22°C and were terminated by spotting 80 l onto Whatman No. 3MM filter papers that were extensively washed in 10% trichloroacetic acid, rinsed in ethanol, dried, and counted by liquid scintillation in a Betamatic counter.
Treatment of Intact Cells with Tyrphostins-Cells, grown in 60-mm plates, were deprived of serum for 16 h before each experiment. 2 h before treatment, the medium was replaced with fresh medium supplemented with 1 mg/ml bovine serum albumin (Sigma radioimmunoassay-grade). Cells were then incubated with or without tyrphostins for an additional 4 h. Tyrphostins were supplemented as 100 ϫ concentrated solutions dissolved in 50 mM Hepes, pH 7.6, and 10% Me 2 SO. Final Me 2 SO concentrations during the incubation did not exceed 0.1%. In some of the experiments, when tyrphostins were present for periods of 16 h, they were added at the time of serum deprivation and were kept throughout the incubations. Following stimulation with the appropriate ligands, cells were washed three times with ice-cold PBS and frozen in liquid nitrogen. Preliminary experiments indicated that this concentration of Me 2 SO did not have any adverse effects on the biological activities under study.
Western Blotting-Samples were boiled in Laemmli sample buffer (34) containing 20 mM dithiothreitol, resolved by SDS-PAGE, and transferred to nitrocellulose. Electrophoretic transfer from the gels to the nitrocellulose papers and incubation with anti-Tyr(P) antibodies were carried out as described previously (33). Detection of bound antibodies was carried out with an ECL kit (Amersham) according to the manufacturer's instructions.
S6 Kinase Assay-The 40 S ribosomal subunit of Artemia salina was prepared as described previously (35). CHO.T cells, grown in 60-mm plates, were incubated for 16 h in serum-free medium in the absence or presence of AG-18. Insulin (0.1 M) or TPA (0.4 g/ml) was added to the cells, and incubation was continued for an additional 10 -60 min as indicated. Following treatment, cells were lysed in 1 ml of buffer II (50 mM Hepes, pH 7.6, 30 mM ␤-glycerophosphate, 15 mM MgCl 2 , 10 mM EGTA, 2 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, and 25 g/ml aprotinin) by freezing and thawing (three times). The extracts were centrifuged at 12,000 ϫ g for 15 min at 4°C, and the supernatants were assayed for S6 kinase activity using the 40 S ribosomal protein as substrate (35).
Determination of MAPK and MAPKK Activities-Determination of these activities was carried out as described previously (36). Briefly, cells were harvested in buffer H (50 mM ␤-glycerophosphate, pH 7.3, 1.5 mM EGTA, 1 mM EDTA, 1 mM dithiothreitol, 0.1 mM sodium vanadate, 1 mM benzamidine, 10 g/ml aprotinin, 10 g/ml leupeptin, and 2 g/ml pepstatin A) and disrupted by 2 ϫ 7-s sonication (50 watts) on ice, followed by centrifugation at 20,000 ϫ g for 15 min at 4°C. The supernatant contained the cytosolic extracts to be examined. All subsequent steps were performed at 4°C. Cytosolic extracts so obtained (0.5 ml) were fractionated on 0.35-ml DEAE-cellulose minicolumns. The flowthrough and wash in 0.02 M NaCl in buffer A (50 mM ␤-glycerophosphate, pH 7.3, 1.5 mM EGTA, 1 mM EDTA, 1 mM dithiothreitol, and 0.1 mM sodium vanadate) were collected and measured for MAPKK activity toward extracellular signal-regulated kinase-2 in a double coupled assay (36). Elution with 0.75 ml of buffer A (including 0.22 M NaCl) contained Ͼ85% of MAPK activity measured against myelin basic protein as described (36).
Intracellular ATP Content-The intracellular ATP content was determined as described (37). Briefly, CHO.T cells, grown in 10-cm plates, were incubated for 2-16 h in serum-free medium in the absence or presence of AG-18 (0 -200 mM). Following treatment, cells were lysed in 1 ml of buffer, and the ATP content of the 12,000 ϫ g supernatant was determined by the luciferin/luciferase assay (37). The chemiluminesence was monitored using a Lumac 3M Luminometer (Model M2010A). Results are the means of duplicate measurements that did not vary by Ͼ5%. Each experiment was carried out at least three times with essentially similar results.

Effects of AG-18 on Insulin Receptor Kinase Activity in Vitro and Protein Tyrosine Phosphorylation in Intact
Cells-It has been previously shown that several tyrphostins inhibit the kinase activity of the insulin receptor in vitro. Among various tyrphostins studied in the present as well as previous work (27), AG-18 was found to be the most potent inhibitor. Although AG-18 (0 -150 M) failed to inhibit the autophosphorylation of the insulin receptor kinase under our in vitro assay conditions (data not shown), it inhibited insulin receptor kinase activity toward exogenous substrates with a K i of 0.5 mM (Fig. 1) and was therefore employed to study the effects of tyrphostins on insulin-and TPA-induced tyrosine phosphorylations and biological responses in CHO.T cells.
As shown in Fig. 2 (A and B) and consistent with previous studies (38,39), incubation of CHO.T cells with insulin induced tyrosine phosphorylation of two major proteins: the 95-kDa ␤-subunit of the insulin receptor (insulin receptor kinase) and one of its major cellular targets, IRS-1 (4,18,38,40,41). Additional proteins that underwent enhanced Tyr phosphorylation in response to insulin were pp60/pp62 (42)(43)(44) and two (out of the three) isoforms of Shc (Shc64 and Shc54), whose identity was verified by immunoprecipitation from cell extracts with Shc antibodies (data not shown).
Incubation of the cells for 16 h with increasing concentrations of AG-18 resulted in a dose-dependent inhibition of insulin-stimulated phosphorylation of several proteins (Fig. 2, A  and B). This inhibition could not be attributed to an inhibitory effect of AG-18 on insulin binding since incubation of CHO.T cells with this drug had no effect on the number of insulin receptors expressed on the cell surface nor did it affect the affinity of insulin for its receptors in these cells (data not shown). AG-18 inhibited insulin-induced Tyr phosphorylation of IRS-1, with a half-maximal effect at ϳ50 M. Inhibition of pp60/pp62 phosphorylation was also readily detected (Fig. 2, A  and B). In contrast, autophosphorylation of pp95, the ␤-subunit of the insulin receptor, as well as Tyr phosphorylation of Shc64 were largely unaffected, whereas phosphorylation of Shc54 was only partially inhibited. Failure of AG-18 to inhibit Shc phosphorylation was also reflected by the inability of the drug to inhibit insulin-induced complex formation of Shc and Grb2. This was demonstrated when insulin-treated cells were preincubated for 16 h either in the presence or absence of 150 M AG-18. Under these conditions, similar amounts of Grb2 (2.3 versus 1.9% of the total) were found in Shc immunoprecipitates.
Interestingly, AG-18 was a more potent inhibitor when applied to intact cells (Fig. 2) compared with its inhibitory effects on substrate phosphorylation of the insulin receptor kinase in a cell-free system (Fig. 1). These differences could be accounted for by AG-18 accumulation within the cell, making its effective intracellular concentration higher than that applied extracellularly. Alternatively, the native conformation of the insulin receptor kinase maintained in vivo could be more susceptible to the inhibitory effects of the drug.
Effect of AG-18 on Intracellular ATP Content-Compounds with structural similarity to AG-18, such as SF 6847, could act as inhibitors of oxidative phosphorylation (45). To rule out the possibility that the inhibitory effects of AG-18 on Tyr phosphorylation are simply due to a marked reduction in intracellular energy charge, its effects on intracellular ATP content were studied. Incubation of CHO.T cells for 16  Inhibition of Insulin-and TPA-induced S6 Phosphorylation by AG-18 -Insulin induces S6 phosphorylation in a timedependent manner. A maximal effect is attained by 10 min and is persistent for at least 1 h (data not shown). As shown in Fig.  3 (left), a 16-h incubation of CHO.T cells with increasing concentrations of AG-18 yielded a dose-dependent inhibition of insulin-stimulated phosphorylation of S6. A similar extent of inhibition was obtained whether the cells were stimulated with insulin for 10, 20, or 60 min. The effects of AG-18 were rather specific (Fig. 3, right) since only insulin-mediated phosphorylation was inhibited, whereas insulin-independent phosphorylation of several other proteins remained unaffected. Half-maximal inhibition of S6 phosphorylation required incubation with 50 M AG-18, a concentration that was similar to that required for half-maximal inhibition of IRS-1 phosphorylation, while nearly maximal inhibition was obtained at 200 M. S6 phosphorylation could also be induced upon stimulation of the cells with TPA (Fig. 4). Here again, the stimulatory effect of TPA was inhibited upon preincubation of the cells with increasing concentrations of AG-18, with a half-maximal effect being obtained at 75 M.
Effects of AG-18 on Insulin-and TPA-induced Activation of the MAPK Cascade-Phosphorylation of the ribosomal protein S6 is mediated by at least two different protein kinases known as p90 rsk and p70 s6k (2). Since p90 rsk is activated by MAPKs (10,15,16,46,47), we studied the effects of AG-18 on insulinand TPA-stimulated MAPKs. Consistent with previous studies (48), insulin added to CHO.T cells markedly enhanced MAPK activity; however, we failed to detect significant inhibition of this activity when the cells were preincubated with AG-18 (Fig.  5A). In contrast, AG-18 at 100 M inhibited by 50% the maximal stimulatory effect of TPA on MAPK activity (Fig. 5B). This concentration is similar to that required to inhibit TPA-stimulated phosphorylation of S6 by 50%. Interestingly, even in the presence of 200 M AG-18, no more than 50% of the TPAstimulated MAPK activity was inhibited. This suggests the presence of both AG-18-sensitive and AG-18-insensitive pathways leading to MAPK activation upon TPA stimulation. Similar results were obtained when we studied the activity of MAPKK, the dual specificity Tyr/Thr kinase that phosphorylates and activates MAPK (11,13). While TPA-stimulated MAPKK was partially inhibited by AG-18 (Fig. 6B), no such inhibitory effect was observed in insulin-stimulated cells (Fig. 6A).
AG-18 Inhibits TPA-stimulated Protein Tyrosine Phosphorylation-Since AG-18 is thought to be a selective inhibitor of protein-tyrosine kinases (26), the above results suggest the involvement of a TPA-activated and an AG-18-inhibited protein-tyrosine kinase in mediating the activation of the MAPK cascade. To directly address this possibility, the effects of TPA and AG-18 on protein tyrosine phosphorylation were evaluated. Incubation of CHO.T cells with TPA resulted in a timedependent enhancement of tyrosine phosphorylation of several proteins, the most prominent ones having molecular masses of 114 and 120 kDa (Fig. 7, upper). Tyrosine phosphorylation of these proteins was inhibited in a dose-dependent manner by AG-18 in vivo. Furthermore, there was a good correlation between the concentrations of AG-18 required to inhibit TPAinduced tyrosine phosphorylation (of pp114 and pp120), activation of the MAPK cascade, and phosphorylation of ribosomal S6 protein (Fig. 7, lower). These results are compatible with a model in which protein-tyrosine kinases medi-ate at least part of the effects of TPA on MAPK activation and S6 phosphorylation. DISCUSSION A protein-tyrosine kinase inhibitor from the tyrphostin family (AG-18) was used to distinguish between the pathways leading to the activation of p70 s6k and the activation of the MAPK/p90 rsk cascade. AG-18 effectively inhibits insulin-induced activation of S6 kinase while having no inhibitory effect on insulin-induced activation of the MAPK cascade. These results indicate that activation of MAPK per se is not sufficient for stimulation of S6 phosphorylation and suggest that insulininduced activation of S6 kinase(s) may occur through an alternative pathway. In this respect, our results complement studies demonstrating that MAPK activation by the insulin receptor is not required for insulin-induced metabolic processes such as glucose transport or glycogen synthase in 3T3-L1 adipocytes (49).
Inhibition of S6 phosphorylation by AG-18 largely parallels the inhibitory effects of AG-18 on insulin-induced phosphorylation of IRS-1 and is compatible with the notion that IRS-1 phosphorylation mediates many insulin responses (41,50,51), including the stimulation of S6 phosphorylation. The latter presumably involves activation of PI3K (17,18) and subsequent activation of p70 s6k (20,21,52), which occurs independent of activation of p21 ras and the MAPK cascade (53). Conversely, we have shown that inhibition of IRS-1 and S6 phosphorylation occurs without inhibition of the MAPK cascade. Although we cannot rule out the possibility that AG-18 fails to inhibit phosphorylation of IRS-1 at Tyr 895 , which is part of the Grb2-binding site (54), our findings support the view that there are alternative pathways for insulin-induced activation of the MAPK cascade, independent of IRS-1 (55,56). This conclusion is supported by the observation that association of Grb2/Sos with IRS-1 plays little if any role in MAPK activation in L6 myoblasts (57). A likely candidate to stimulate the MAPK cascade is Shc, which serves as a downstream effector of the insulin receptor (5) and acts to activate the Ras/MAPK pathway (8, 55, 58 -60). Indeed, AG-18 failed to inhibit insulininduced Tyr phosphorylation of Shc64, and phosphorylation of Shc54 was only partially inhibited. Similarly, AG-18 failed to inhibit insulin-induced complex formation between Shc and Grb2. Hence, although our results clearly support the involvement of Shc in MAPK stimulation, different Shc isoforms might play different roles in insulin signal transduction, and further studies are required to address this possibility.
Activation of p90 rsk as a result of Shc phosphorylation, together with activation of the MAPK cascade, could account for the residual S6 phosphorylation observed in the presence of 150 M AG-18 in insulin-treated cells. The fact that this residual S6 phosphorylation is rather low (ϳ20%) suggests, however, that the predominant mode of insulin-activated S6 phosphorylation (at least in CHO cells) occurs through the IRS-1/ PI3K/p70 s6k signaling pathway. Taken together, our findings are consistent with a model (Scheme 1) in which IRS-1 mediates insulin-induced activation of p70 s6k , whereas the signals leading to the activation of the MAPK/p90 rsk cascade are transmitted via the adaptor molecule Shc.
Tyrphostins were previously shown to inhibit insulin-stimulated lipogenesis in fat cells, while they failed to inhibit the anti-lipolytic effect of the hormone (28). These differences could be accounted for by the different potency of tyrphostins to inhibit phosphorylation of insulin receptor kinase substrates that could mediate these processes (IRS-1 and Shc). Accordingly, we suggest that IRS-1 is more prone to inhibition by these competitive inhibitors because it is less abundant and/or has a lower affinity toward insulin receptor kinase when compared with other insulin receptor substrates (e.g. Shc) that mediate activation of the MAPK cascade. This assumption is supported by recent findings (56), where cells expressing insulin receptor mutants (of Tyr autophosphorylation sites within the kinase region) maintained insulin-induced phosphorylation of Shc, whereas phosphorylation of IRS-1 was largely reduced. Alternatively, some of the insulin-induced Tyr-phosphorylated proteins (e.g. Shc64) could serve as substrates for intermediary protein-tyrosine kinases rather than as substrates for the insulin receptor kinase itself. These intermediary protein-tyrosine kinases could undergo activation upon insulin receptor autophosphorylation, which is not inhibited by AG-18 (see above). Activation could involve, for example, binding of SH2 domains of these putative intermediary protein-tyrosine kinases to unique Tyr(P) residues within the cytoplasmic portion of the insulin receptor.
A different picture emerges when the effects of TPA on S6 phosphorylation are studied. A good concordance exists between the inhibitory effects of AG-18 on TPA-induced MAPKK and MAPK activity and S6 phosphorylation, which suggests that protein kinase C induces S6 phosphorylation preferentially through the MAPK signaling pathway. Moreover, the difference in the effects of AG-18 on insulin-versus TPA-activated MAPK suggests that, in these cells, insulin-induced activation of the MAPK cascade occurs via a protein kinase Cindependent pathway. Hence, p70 s6k appears to be mainly responsible for insulin-induced S6 phosphorylation (20), while p90 rsk could mediate the effects of TPA (61).
Since AG-18 is a very poor inhibitor of Ser/Thr protein kinases, including protein kinase C (62), the inhibitory effects of AG-18 suggest that one or more AG-18-sensitive protein- FIG. 7. Effect of AG-18 on TPA-induced protein tyrosine phosphorylation. Upper, confluent CHO.T cells were incubated for 16 h in serum-free medium with the indicated concentrations of AG-18. At the end of incubation, 0.4 g/ml TPA was added for a 30-min incubation period. Cells were then washed three times with ice-cold PBS, and cell extracts were prepared, resolved by 10% SDS-PAGE, and immunoblotted with anti-Tyr(P) antibodies. Lower, the intensity of the bands corresponding to pp120 (E) and pp114 (q) was quantitated by densitometry, and the percent inhibition induced by AG-18 was calculated. The effects of AG-18 on TPA-induced S6 phosphorylation (f) and MAPK activity ( ) are also presented for comparison. SCHEME 1. Tentative model illustrating the signaling pathways induced by insulin and TPA that stimulate S6 phosphorylation. Many proteins known to take part in other aspects of insulin and TPA-mediated responses were eliminated for simplicity. The expected sites of action of AG-18 are indicated. PM, plasma membrane; IR, insulin receptor; PKC, protein kinase C; PTK, protein-tyrosine kinase, Erk, extracellular signal-regulated kinase. tyrosine kinases mediate the effects of protein kinase C on the MAPK pathway and S6 phosphorylation. This conclusion is supported by the facts that (i) TPA induces protein tyrosine phosphorylation in CHO.T cells, and (ii) AG-18 inhibits both TPA-stimulated tyrosine phosphorylation and TPA-stimulated MAPK activity with a similar dose-response curve. Although we cannot rule out the possibility that insulin and protein kinase C activate different isoforms of the dual specificity MAPKK, our results are most consistent with the hypothesis that the TPA-activated protein-tyrosine kinase presumably differs from the dual specificity kinase, MAPKK. This conclusion is primarily based on the fact that insulin-induced activation of MAPKK is insensitive to the presence of AG-18.
Several studies implicate protein kinase C, the direct effector of TPA, as a mediator of protein tyrosine phosphorylation events. In rat basophilic leukemia cells, Tyr phosphorylation of a 110-kDa protein occurs secondary to calcium influx and protein kinase C activation (63). Activation of protein kinase C and/or the induction of calcium influx was implicated in immunoglobulin E receptor-induced Tyr phosphorylation of focal adhesion-associated tyrosine kinase (pp125 FAK ) in fibronectin-adherent rat basophilic leukemia cells (64,65). Similarly, protein kinase C was shown to mediate carbachol-stimulated tyrosine phosphorylation in human SH-SY5Y neuroblastoma cells (66). Our results suggest that a protein kinase C-stimulated protein-tyrosine kinase should be present upstream of MAPKK in the protein kinase C signaling pathway, leading to the activation of p90 rsk .
The nature of the TPA-activated protein-tyrosine kinase is presently unknown, but among its potential substrates, we find pp114 and pp120, whose inhibited phosphorylation correlates with inhibition of MAPK activity. Hence, we can formulate a tentative signaling cascade (Scheme 1) in which a TPA-activated protein-tyrosine kinase stimulates the common Grb2/Sos and the Ras signaling pathway (8, 55, 58 -60) and in such a way leads to activation of MAPK and S6 phosphorylation (13). Further studies are required, however, to figure out the role of pp114/pp120 and to determine whether the TPA-activated protein-tyrosine kinase indeed utilizes the Grb2/Sos/Ras signaling elements to induce activation of the MAPK cascade.