Cross-talk between Phorbol Ester-mediated Signaling and Tyrosine Kinase Proto-oncogenes

In the accompanying paper (Emkey, R., and Kahn, C. R. (1997) J. Biol. Chem. 272, 31172–31181), we demonstrated that phorbol 12-myristate 13-acetate (PMA) treatment of Fao cells induces tyrosine phosphorylation of several proteins including ErbB2 and ErbB3. In the present study we show that sphingomyelinase also results in the enhanced tyrosine phosphorylation of ErbB2 and ErbB3 in these cells. In contrast to activation by PMA, the sphingomyelinase-induced phosphorylation of these proteins is independent of protein kinase C. However, both agents stimulate tyrosine phosphorylation of the kinase Pyk2 suggesting that it may be involved in the PMA and sphingomyelinase activation of these ErbB proto-oncogenes. Insulin plays a negative regulatory role in the ligand and non-ligand-induced phosphorylation of the ErbB proto-oncogenes via two mechanisms. Prolonged insulin treatment resulted in decreased expression of both ErbB2 and ErbB3. Insulin also appears to negatively regulate the protein tyrosine kinase responsible for phosphorylating ErbB2 in PMA-stimulated cells. The former effect of insulin was relieved by treatment with inhibitors of phosphatidylinositol 3-kinase. The similarities in PMA and sphingomyelinase-induced effects and the negative regulatory role of insulin suggest a mechanism by which multiple ligands can synergize with or protect against the tumorigenic effects of phorbol esters.

In the accompanying paper (Emkey, R., and Kahn, C. R. (1997) J. Biol. Chem. 272, 31172-31181), we demonstrated that phorbol 12-myristate 13-acetate (PMA) treatment of Fao cells induces tyrosine phosphorylation of several proteins including ErbB2 and ErbB3. In the present study we show that sphingomyelinase also results in the enhanced tyrosine phosphorylation of ErbB2 and ErbB3 in these cells. In contrast to activation by PMA, the sphingomyelinase-induced phosphorylation of these proteins is independent of protein kinase C. However, both agents stimulate tyrosine phosphorylation of the kinase Pyk2 suggesting that it may be involved in the PMA and sphingomyelinase activation of these ErbB proto-oncogenes. Insulin plays a negative regulatory role in the ligand and non-ligand-induced phosphorylation of the ErbB proto-oncogenes via two mechanisms. Prolonged insulin treatment resulted in decreased expression of both ErbB2 and ErbB3. Insulin also appears to negatively regulate the protein tyrosine kinase responsible for phosphorylating ErbB2 in PMAstimulated cells. The former effect of insulin was relieved by treatment with inhibitors of phosphatidylinositol 3-kinase. The similarities in PMA and sphingomyelinase-induced effects and the negative regulatory role of insulin suggest a mechanism by which multiple ligands can synergize with or protect against the tumorigenic effects of phorbol esters.
Metabolism of membrane phospholipids has been shown to play an important role in cellular proliferation and differentiation (1,2). The newly described sphingomyelin cycle is a signal transduction pathway that has been implicated in the differentiation of HL-60 cells (3,4), the potentiation of plateletderived growth factor-stimulated proliferation of Swiss 3T3 cells (5), suppression of insulin-induced tyrosine phosphorylation of the major insulin receptor substrate-1 (6,7), and the actions of tumor necrosis factor-␣ (TNF-␣) 1 on apoptosis in tumor cells (8,9) and the proinflammatory response (10).
The sphingomyelin pathway is mediated via activation of sphingomyelinase and can be initiated by TNF-␣, interleukin-1, progesterone, or ␥-interferon (11)(12)(13). Two forms of sphingomyelinase have been identified based on their pH optima as follows: an acidic sphingomyelinase (pH optimum ϳ4.5) found in lysosomes (14) and plasma membrane (15), and a neutral sphingomyelinase (pH optimum ϳ7.4) found associated with membranes, as well as in the cytosol (13). Both sphingomyelinases hydrolyze membrane sphingomyelin to yield ceramide and phosphocholine. Ceramide has been shown to function as a second messenger that activates several targets including ceramide-activated protein kinase, a membrane-associated proline-directed serine/threonine protein kinase that has subsequently been identified as the kinase suppressor of Ras (16), a cytosolic protein phosphatase (11,17), the mitogenactivated protein kinase cascade (18), and the stress-activated protein kinases (19). Activation of the sphingomyelin pathway in Swiss 3T3 cells by treatment with bacterial sphingomyelinase (containing acidic and neutral sphingomyelinase) or cellpermeable analogs of ceramide stimulates the tyrosine phosphorylation of focal adhesion kinase (FAK) and paxillin (20). In the preceding paper (21), we demonstrated that phorbol 12myristate 13-acetate (PMA) treatment of well differentiated rat Fao hepatoma cells with PMA resulted in enhanced tyrosine phosphorylation of several proteins including FAK, paxillin, ErbB2, and ErbB3. ErbB2 and ErbB3 are members of the epidermal growth factor family of tyrosine kinase receptors (22)(23)(24)(25)(26). These receptors are activated and autophosphorylate after binding to their respective ligand and then are capable of homodimerization, as well as heterodimerization with other ErbB family members. Once activated, the receptors associate with the src homology 2 (SH2) domains of several signaling proteins that initiate a variety of signaling cascades (27)(28)(29)(30)(31)(32)(33).
In the present study, we have explored the mechanisms of PMA and sphingomyelinase effects on the ErbB proto-oncogenes. We show that treatment of Fao cells with sphingomyelinase results in enhanced tyrosine phosphorylation of ErbB2 and ErbB3 similar to that observed with PMA. The sphingomyelinase-induced tyrosine phosphorylation of these receptors, however, was independent of protein kinase C (PKC). We present preliminary data suggesting that one of the kinases involved in ErbB phosphorylation may be the recently described member of the FAK family, Pyk2 (34). We also show that PMA-stimulated phosphorylation is negatively regulated by prolonged treatment with insulin by mechanisms involving down-regulation of the ErbB proteins, as well as the tyrosine kinase(s) which phosphorylate them. from Sigma; bisindolylmaleimide (BIM) and LY294002 from Calbiochem; Protein A-Sepharose 6MB from Pharmacia Biotech Inc.; antibodies against ErbB2 and ErbB3 from Santa Cruz Biotechnology; anti-Pyk2 and anti-phosphotyrosine (PY20) antibodies from Transduction Laboratories; rabbit anti-mouse IgG (H ϩ L) from Jackson ImmunoResearch Laboratories; and insulin from Lilly. Monoclonal anti-phosphotyrosine antibody (4G10) was obtained from Dr. M. White (Joslin Diabetes Center, Boston). Purified recombinant heregulin-␤1 (amino acids 177-244, rHRG-␤1 177-244 ) was the generous gift of Dr. M. X. Sliwkowski (Genentech, Inc., San Francisco, CA).
Cell Culture, Stimulation, Immunoprecipitation, and Immunoblotting-Fao cells were maintained in RPMI 1640 supplemented with 10% fetal calf serum at 37°C, 5% CO 2 . Cells were grown to Ͼ70% confluence, washed once with phosphate-buffered saline, and placed in RPMI 1640 lacking serum overnight (ϳ16 h). Cells were stimulated with either 100 nM insulin for 5 min, 1 g/ml PMA for 30 min, 0.1 unit/ml sphingomyelinase for 30 min, or 50 nM recombinant heregulin-␤1 amino acids 177-244 for 8 min unless indicated otherwise. Alternatively, cells were maintained in serum-free medium containing 100 nM insulin and/or 20 M LY294002 or pretreated with 10 M BIM for 90 min prior to stimulation with PMA, sphingomyelinase, or heregulin as indicated. Cells were washed twice with phosphate-buffered saline and lysed in 50 mM HEPES, pH 7.4, 140 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 10 mM Na 4 P 2 O 7 , 100 mM NaF, 2 mM EDTA, 10% glycerol, 1% Triton X-100, 2 mM Na 3 VO 4 , 2 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, and 10 mM benzamidine. Lysates were cleared by centrifugation for 10 min in a microcentrifuge at 4°C. Protein concentration was determined using the Bio-Rad Protein Assay (Bio-Rad). An equal concentration of protein from each lysate was immunoprecipitated with the indicated antibody by incubating the lysate with the antibody for 2 h at 4°C with mixing. Protein A-Sepharose was added to each sample and incubated an additional hour at 4°C with mixing. The beads were collected, washed three times in lysis buffer, boiled in Laemmli sample buffer, separated by SDS-PAGE on a 6% acrylamide gel, and transferred to nitrocellulose. The blots were blocked in 3% bovine serum albumin in Tris-buffered saline with 0.05% Tween 20 (TBS-Tween) for at least 30 min at room temperature, incubated with the indicated primary antibody for 90 min, washed with TBS-Tween, incubated with 125 I-protein A for 1 h, washed with TBS-Tween, and visualized by autoradiography. Alternatively, blots were exposed in a PhosphorImager cassette, scanned on Molecular Dynamics Phosphor-Imager, and quantitated using ImageQuant software.

Sphingomyelinase Induces Tyrosine Phosphorylation of
ErbB2 and ErbB3 in Fao Cells-In the accompanying paper (21) we reported that treatment of the rat hepatoma cell line Fao with the tumor-promoting phorbol ester phorbol 12-myristate 13-acetate (PMA) resulted in the tyrosine phosphorylation of several proteins including focal adhesion kinase (FAK) paxillin, ErbB2, ErbB3, and unidentified proteins of 120 -130 and ϳ70 kDa. Since FAK and paxillin have been shown to undergo tyrosine phosphorylation in Swiss 3T3 cells treated with either PMA or sphingomyelinase (20), we examined whether these two agents also induced tyrosine phosphorylation of the same proteins in Fao cells. Anti-phosphotyrosine immunoprecipitates were prepared from control unstimulated Fao cells and cells treated for 30 min with either PMA (1 g/ml) or sphingomyelinase (0.1 unit/ml). The immunoprecipitates were subjected to SDS-PAGE, and tyrosine-phosphorylated proteins were visualized by anti-phosphotyrosine immunoblot analysis (Fig. 1A).
As previously noted, unstimulated cells contain a major phosphorylated band at 120 kDa consisting of FAK and p120 and a minor band at 70 kDa (paxillin and p70) (21). Stimulation by either PMA or sphingomyelinase resulted in minor increases in phosphorylation of proteins at 120 kDa and at 70 kDa and a marked increase in tyrosine phosphorylation in a 190-kDa band that we have identified in PMA-treated cells as ErbB2 and ErbB3 (21). Immunoprecipitation of lysates with antibodies against either ErbB2 or ErbB3 revealed that these proteins are also tyrosine-phosphorylated by stimulation with sphingomyelinase ( Fig. 1, B and C). The tyrosine phosphoryl-ation of ErbB2 and ErbB3 in sphingomyelinase-treated cells was increased 4.6-and 3.4-fold, respectively. Therefore, both sphingomyelinase and PMA stimulate the tyrosine phosphorylation of ErbB2 and ErbB3 in Fao cells.
Although sphingomyelinase mimicked the effect of PMA on tyrosine phosphorylation, it did not mimic the effect on serine/ threonine phosphorylation. Thus, PMA treatment results in a retarded migration of ErbB2 and paxillin on SDS-PAGE consistent with the fact that PMA induces serine/threonine phosphorylation of these proteins (Fig. 1). By contrast, sphingomyelinase had no effect on the mobility of either of these proteins, although it clearly stimulated the tyrosine phosphorylation of both ( Fig. 1). There was also a subtle difference in the time and dose dependence of tyrosine phosphorylation of ErbB2/ErbB3 with sphingomyelinase and PMA. Immunoprecipitation with an anti-phosphotyrosine antibody followed by immunoblotting FIG. 1. Sphingomyelinase-induced tyrosine phosphorylation in Fao cells. Fao cells were maintained in serum-free medium overnight (Control) followed by stimulation with 1 g/ml PMA for 30 min (PMA) or 0.1 unit/ml sphingomyelinase (SMase) for 30 min. A, tyrosinephosphorylated proteins were immunoprecipitated (IP), separated by 6% SDS-PAGE, transferred to nitrocellulose, and immunoblotted (IB) with an anti-phosphotyrosine (PY) antibody as described under "Experimental Procedures." B and C, cell lysates were immunoprecipitated with antibodies against either phosphotyrosine, ErbB2, or ErbB3 as indicated. The immunoprecipitates were separated by 6% SDS-PAGE, transferred to nitrocellulose, and immunoblotted with either an anti-ErbB2 (B) or anti-ErbB3 antibody (C). These results are representative of at least five experiments.
with the same antibody revealed that tyrosine phosphorylation of ErbB2/ErbB3 obtained with sphingomyelinase was more transient than the PMA-induced phosphorylation as reported previously (compare Fig. 2, A and B, with Fig. 3, A and B, in the accompanying paper). Thus, phosphorylation of ErbB2/ErbB3 peaked at 20 -30 min of sphingomyelinase treatment and gradually declined over the next 30 min, whereas PMA-induced phosphorylation remained at maximal levels even after 60 min of stimulation (21). To ensure maximal stimulation, cells were treated with 0.1 unit/ml sphingomyelinase for 30 min.
The Role of PKC in Tyrosine Phosphorylation of ErbB2 and ErbB3 Induced by PMA, Sphingomyelinase, and Heregulin-To explore the role of protein kinase C in the actions of both PMA and sphingomyelinase and the natural ligand heregulin to phosphorylate and activate ErbB2 and ErbB3, cells were pretreated with phorbol ester or with the PKC inhibitor bisindolylmaleimide (BIM) prior to stimulation with either PMA, sphingomyelinase, or the native ligand for ErbB3, heregulin ␤1 (HRG␤1, amino acids 177-244). Cell lysates were then subjected to immunoprecipitation using either anti-phosphotyrosine, anti-ErbB2, or anti-ErbB3 antibodies followed by immunoblotting with the indicated antibody.
As shown in the accompanying paper (21), the PMA-induced tyrosine phosphorylation of ErbB2 and ErbB3 was dependent on PKC, since pretreatment with BIM abolished the PMAinduced phosphorylation of the ErbB receptors without altering the expression level of the receptors (Fig. 3). Similar results were obtained by down-regulating PKC by prolonged treatment with phorbol ester (data not shown). In contrast, the phosphorylation of ErbB2 and ErbB3 by heregulin, a natural ligand for ErbB3, was not altered by pretreatment of cells with BIM or down-regulation of PKC (Fig. 3). This is not surprising since heregulin is known to bind ErbB3 and induce the formation of ErbB2/ErbB3 heterodimers thereby activating the intrinsic kinase activity of ErbB2 and enabling it to phosphorylate ErbB3 (35). Interestingly, sphingomyelinase-induced phosphorylation was also independent of PKC. Thus, pretreatment with BIM or down-regulation of PKC had no effect on the ability of sphingomyelinase to stimulate phosphorylation of ErbB2 and ErbB3 (Fig. 3). These results indicate that sphingomyelinase-and heregulin-induced tyrosine phosphorylation of ErbB2 and ErbB3 occur in a PKC-independent manner.
As mentioned earlier, in addition to stimulating tyrosine phosphorylation, PMA also induced a mobility shift of ErbB2 on SDS-PAGE characteristic of serine/threonine phosphorylation, whereas sphingomyelinase had no effect on the mobility of either ErbB2 or ErbB3. By contrast, heregulin treatment of cells resulted in both tyrosine phosphorylation and a reduced migration of ErbB2 on SDS-PAGE (Fig. 3C). The PMA-induced mobility shift of ErbB2 was dependent on PKC since pretreatment with an inhibitor of PKC, BIM, abolished the shift ( Fig.  3C and accompanying paper (21)). In contrast, the heregulininduced mobility shift of ErbB2 was not inhibited by BIM (Fig.  3C). Taken together, these results suggest that at least two independent mechanisms exist for the tyrosine and serine/ threonine phosphorylation of ErbB2, or at the very least, the two pathways converge downstream of PKC.
The identity of the tyrosine kinase responsible for phosphorylating ErbB2 in PMA-treated cells is unknown; however, an excellent candidate is the recently identified tyrosine kinase Pyk2 (34). Pyk2 is a member of the FAK family of kinases and is activated by extracellular signals that increase the intracellular calcium concentration, activators of protein kinase C, TNF-␣, and stress signals such as ultraviolet irradiation and changes in osmolarity (34,36,37). Therefore, we examined whether Pyk2 was activated in Fao cells treated with either PMA, sphingomyelinase, or heregulin.
Fao cells were serum-starved overnight and then stimulated with either 50 nM heregulin ␤1 (HRG␤1, amino acids 177-244), 1 g/ml PMA, or 0.1 unit/ml sphingomyelinase. Cell lysates were immunoprecipitated with an anti-phosphotyrosine antibody, subjected to SDS-PAGE, and immunoblotted with an anti-Pyk2 antibody. Fig. 4A demonstrates that more tyrosinephosphorylated Pyk2 was immunoprecipitated from cells stimulated with each of these three agents than from control unstimulated cells. Quantitation of three independent experiments reveals about a 2-2.5-fold greater amount of Pyk2 in anti-phosphotyrosine antibody immunoprecipitates from stimulated cells compared with control cells (Fig. 4B). These results do not prove that Pyk2 itself is tyrosine-phosphorylated, since we did not have a Pyk2 antibody that was suitable for immunoprecipitation, but do demonstrate that either Pyk2 is tyrosine-phosphorylated in Fao cells stimulated with heregulin, PMA, or sphingomyelinase or that Pyk2 is associated with a tyrosine-phosphorylated protein under these conditions. Thus far attempts to detect an association of Pyk2 with ErbB2 or ErbB3 in PMA-treated Fao cells by co-immunoprecipitation have been unsuccessful. Hence, it is uncertain whether Pyk2 is an important component in the activation of ErbB2 and ErbB3 by PMA in Fao cells or the activation of Pyk2 is mediated by ErbB2 and/or ErbB3. Additional studies are required to distinguish between these possibilities.

PMA, Sphingomyelinase, and Heregulin-induced Tyrosine
Phosphorylation of ErbB2 and ErbB3 Is Abolished by Pretreatment with Insulin-Carver et al. (38) have demonstrated that exposure of cultured primary hepatocytes to insulin results in a marked reduction of ErbB3 protein. This down-regulation of ErbB3 is due to an effect at the transcriptional level (38). We were interested in determining what effect, if any, insulin had on the expression and tyrosine phosphorylation of ErbB3 in Fao cells. In addition, unlike the primary hepatocytes that do not contain ErbB2 (38), Fao cells do express ErbB2 thereby enabling us to examine the effect of insulin on ErbB2 expression and phosphorylation.
As shown in Fig. 5, treatment of Fao cells overnight with 100 nM insulin abolished the ability of PMA, sphingomyelinase, and heregulin to induce the tyrosine phosphorylation of both ErbB2 and ErbB3 (Fig. 5, A, B, and D). Similar to the observation in cultured hepatocytes (38), prolonged exposure of Fao cells to insulin resulted in a marked (Ն90%) reduction in ErbB3 expression (Ն90% decrease) as determined by immunoprecipitation and Western blot analysis (Fig. 5E). This decrease of ErbB3 protein could account for the loss of tyrosine phosphorylation of ErbB3 in response to the various agents following pretreatment with insulin (Fig. 5D). We were also unable to detect any tyrosine phosphorylation of ErbB2 in stimulated cells after insulin pretreatment (Fig. 5B). Interestingly, however, this marked decrease in phosphorylation was accompanied by only a modestly diminished level of ErbB2 expression (10 -40% decrease) (Fig. 5C). Thus, the decrease in phosphorylation of ErbB2 cannot be explained simply by an alteration in the ErbB2 protein, suggesting that insulin must have an additional effect to either down-regulate or inhibit the kinase responsible for phosphorylation of ErbB2 or stimulate a phosphotyrosine phosphatase that would dephosphorylate the ErbB2.
To explore further the mechanism of insulin's effects on ErbB2 and ErbB3, the time course of this effect was studied. The initial observation of insulin's consequences on the tyrosine phosphorylation and expression of ErbB2 and ErbB3 in Fao cells was made following an overnight (ϳ16 h) exposure to insulin; however, effects of insulin were observed at much shorter times. The PMA-induced tyrosine phosphorylation of ErbB2 was severely reduced by 2 h of insulin treatment, despite the fact that the level of ErbB2 had not changed at this time (Fig. 6, A, B, and E). This was even more apparent after 6 h of insulin pretreatment where there was virtually no detectable tyrosine phosphorylation of ErbB2 induced by PMA, while ErbB2 protein expression had decreased only ϳ40%. In contrast, the expression level of ErbB3 correlated with the PMA-induced tyrosine phosphorylation of ErbB3 at all times of insulin pretreatment (Fig. 6, C, D, and F). ErbB3 expression was decreased by as early as 2 h of insulin treatment and was virtually undetectable after 6 h of treatment. Similarly, the tyrosine phosphorylation of ErbB3 was reduced after 1 h of insulin treatment and was nearly abolished following 6 h of insulin treatment. These time courses support our hypothesis that insulin's negative regulation of the tyrosine phosphorylation of ErbB2 involves a mechanism other than by decreased expression of ErbB2 protein, whereas the effect of insulin on ErbB3 phosphorylation could be explained by effects on the level of protein alone.

The Role of Phosphatidylinositol 3-Kinase in Insulin's Ability to Inhibit PMA-induced Tyrosine Phosphorylation of ErbB2
and ErbB3-One of the signaling proteins activated upon stimulation with insulin which is involved in many of its metabolic actions is phosphatidylinositol 3-kinase (PI 3-kinase). To examine the role of PI 3-kinase in insulin's ability to decrease the expression and PMA-induced tyrosine phosphorylation of ErbB2 and ErbB3, Fao cells were incubated in serum-free medium overnight in the presence or absence of 100 nM insulin and/or 20 M LY294002, a specific antagonist of PI 3-kinase. The cells were then stimulated with PMA (1 g/ml for 30 min), cell lysates were prepared, subjected to immunoprecipitation with either anti-ErbB2 or anti-ErbB3 antibodies, and analyzed by immunoblotting with antibodies against either phosphotyrosine, ErbB2, or ErbB3 (Fig. 7). Inhibition of PI 3-kinase with LY294002 had no effect on the ability of PMA to stimulate tyrosine phosphorylation of either ErbB2 or ErbB3 (Fig. 7, B,  C, E, and F). Similarly, the PMA-induced mobility shift of ErbB2 was also insensitive to LY294002 (Fig. 7B). Inhibition of PI 3-kinase did, however, abolish insulin's ability to decrease the expression levels of ErbB2 and ErbB3 (Fig. 7, B, C, E, and  F). Inclusion of LY294002 with insulin overnight restored ErbB2 and ErbB3 expression levels to 90 and 50%, respectively, that of control unstimulated or PMA-stimulated cells.
Interestingly, inhibition of PI 3-kinase during prolonged insulin treatment restored the PMA-induced tyrosine phosphorylation of ErbB3 to nearly the same extent as it restored ErbB3 expression levels (Fig. 7, D, E, and F). This observation further supports the hypothesis that the observed inhibition of PMA-induced tyrosine phosphorylation of ErbB3 following long term exposure to insulin is due to the markedly decreased expression of ErbB3 which occurs under these conditions and that PI 3-kinase plays an important role in this effect of insulin.
Insulin appears to utilize a different mechanism to inhibit PMA-induced tyrosine phosphorylation of ErbB2 which appears to be independent of PI 3-kinase. Thus, inhibition of PI 3-kinase during prolonged insulin treatment was unable to restore the PMA-induced tyrosine phosphorylation of ErbB2 despite the fact that ErbB2 protein levels were identical to levels in unstimulated control cells (Fig. 7, A, B, and C). This result indicates that elevation of ErbB2 protein expression to control levels was insufficient to restore PMA-induced tyrosine phosphorylation of ErbB2 following chronic treatment with insulin. This suggests that insulin employs a mechanism that is independent of PI 3-kinase and ErbB2 expression to inhibit tyrosine phosphorylation of ErbB2 in PMA-treated cells.
Taken together, these results support our hypothesis that insulin inhibits the tyrosine phosphorylation of ErbB3 by severely decreasing the expression level of ErbB3 and that this effect of insulin relies on activation of PI 3-kinase. By contrast, the ability of insulin to abolish the PMA-induced tyrosine phosphorylation of ErbB2 does not appear to utilize the same mech- anism. Insulin has a much more modest effect on the expression level of ErbB2 compared with ErbB3. In addition, the inclusion of LY294002, which increases ErbB2 levels to the same as that from control unstimulated cells, was unable to restore the PMA-induced tyrosine phosphorylation of ErbB2. This suggests that insulin is performing some other regulatory role in the cell which inhibits the tyrosine phosphorylation of ErbB2. Furthermore, this mechanism is independent of PI 3-kinase. One possibility is that insulin inactivates or downregulates the tyrosine kinase responsible for phosphorylating ErbB2 in PMA-stimulated cells. DISCUSSION The primary targets of tumor-promoting phorbol esters are members of the protein kinase C (PKC) family of serine/threonine kinases (39,40). Recently, we (21) and others (41)(42)(43) have shown that phorbol esters such as PMA also stimulate the tyrosine phosphorylation of a number of cellular proteins, including the membrane tyrosine kinase proto-oncogenes ErbB2 and ErbB3 (21). In the present study we have shown that sphingomyelinase treatment of hepatoma cells also induces tyrosine phosphorylation of these proteins and that stimulation by both of these agents, as well as the native ligand heregulin, is blocked by insulin pretreatment. This allows us to begin to examine the mechanism(s) of regulation of ErbB2 and ErbB3 phosphorylation utilized by these agents and develop a potential molecular model of this process and the enzymes involved (summarized in Fig. 8).
Phosphorylation of ErbB2 and ErbB3 is increased by treatment of hepatoma cells with PMA, sphingomyelinase, and heregulin in a manner that is qualitatively and quantitatively identical. As noted in the accompanying paper, PMA-induced tyrosine phosphorylation of ErbB2 and ErbB3 is dependent on PKC (21). In contrast, sphingomyelinase and heregulin are still able to stimulate tyrosine phosphorylation of ErbB2 and ErbB3 when PKC is inhibited by pretreatment with bisindolylmale-imide or is down-regulated by chronic PMA treatment. It is likely that the action of sphingomyelinase proceeds through ceramide, a product formed upon hydrolysis of sphingomyelin by sphingomyelinase (44). Cell-permeable analogs of ceramide stimulate the tyrosine phosphorylation of FAK and paxillin in Swiss 3T3 cells to the same extent as sphingomyelinase treatment (20). Production of ceramide by activation of the sphingomyelinase pathway has also been shown to mediate many effects of the TNF-␣ signaling cascade (2), and thus ceramide is believed to function as a second messenger capable of activating several target proteins (11). Whether TNF-␣ also stimulates tyrosine phosphorylation of ErbB2 and ErbB3 remains to be determined, but treatment of Fao cells with TNF-␣, sphingomyelinase, or ceramide has been shown to inhibit, rather than mimic, insulin-induced tyrosine phosphorylation of the insulin receptor and its substrate insulin receptor substrate-1 (6,45).
We believe that both the PMA and sphingomyelinase pathways lead to activation of one or more protein tyrosine kinases that are responsible for phosphorylating and activating ErbB2 and ErbB3. An excellent candidate for one of these kinases is the recently identified tyrosine kinase Pyk2 (34). Pyk2 is a member of the FAK family of non-receptor tyrosine kinases and has been shown to be activated by a variety of factors including PMA. We have seen an increased amount of Pyk2 in antiphosphotyrosine immunoprecipitates from Fao cells stimulated with either PMA or sphingomyelinase. Although this does not prove that Pyk2 is the kinase involved in phosphorylation of ErbB2 and ErbB3, it is consistent with such an hypothesis. It is not clear, however, whether the PMA and sphingomyelinase pathways converge downstream of PKC to activate Pyk2 or whether these pathways are parallel and involve distinct kinases. It is important to note that heregulin also seems to stimulate activation of Pyk2. This suggests that ErbB2 and/or ErbB3 mediate activation of Pyk2 rather than Pyk2 phospho- rylating the ErbB proteins. Since, for reasons discussed below, it appears that different kinases may be responsible for phosphorylating ErbB2 and ErbB3, in our model (Fig. 8) we have termed these hypothetical kinases ErbB2 kinase (EB2 kinase) and ErbB3 kinase (EB3 kinase).
Insulin treatment provides an interesting tool to dissect the mechanism of ErbB2 and ErbB3 phosphorylation. Carver et al. (38) have shown that chronic insulin treatment results in decreased expression of ErbB3 in cultured primary hepatocytes, and we find a similar effect of insulin in the Fao hepatoma cells with a Ͼ90% decrease in ErbB3 at the protein level. Insulin treatment of Fao cells also results in a decrease in ErbB2 protein but only 10 -40%. Furthermore, the time course of the effects on ErbB2 and ErbB3 is slightly different. Expression of ErbB2 is decreased after 4 h of exposure to insulin and remains at this level for up to 8 h of treatment, while ErbB3 protein level begins to decrease after 2 h of insulin treatment and declines to nearly undetectable levels within 6 h of exposure to insulin.
These effects of insulin are dependent to some degree on the activation of phosphatidylinositol 3-kinase (PI 3-K). Inhibition of PI 3-kinase with LY294002 completely abolishes the ability of insulin to decrease the expression of ErbB2 and restores the expression of ErbB3 to 50% the level in control unstimulated cells. Therefore, insulin's ability to down-regulate ErbB2 and ErbB3 proceeds through PI 3-kinase, although it appears that other factors may be involved in the down-regulation of ErbB3.
Prolonged treatment with insulin also inhibits the tyrosine phosphorylation of ErbB2 and ErbB3 induced by either PMA, sphingomyelinase, or the native ligand for ErbB3, heregulin. For ErbB3, inhibition of PI 3-kinase during the insulin pretreatment increases protein levels to 50% that of control cells and produces a similar effect on insulin's inhibition of PMAinduced tyrosine phosphorylation of ErbB3. This supports the hypothesis that insulin's inhibition of PMA-induced tyrosine phosphorylation of ErbB3 is due to its ability to down-regulate ErbB3 expression. For ErbB2, on the other hand, the mechanism for insulin's inhibition of PMA-induced tyrosine phosphorylation is not as easily explained. Thus, inhibition of PI 3kinase completely abolishes insulin's ability to down-regulate ErbB2 protein but does not reverse insulin's effect on PMAinduced phosphorylation of ErbB2. This indicates that insulin utilizes two separate mechanisms to decrease ErbB2 expression and phosphorylation. It decreases expression of ErbB2 by a PI 3-kinase-dependent pathway but inhibits ErbB2 phosphorylation by a PI 3-kinase-independent pathway. There are at least two possible explanations for insulin's inhibition of PMA-induced phosphorylation of ErbB2. First, insulin may activate or increase the expression of a protein tyrosine phosphatase. This seems unlikely since inclusion of vanadate did not prevent insulin's inhibitory effect (data not shown). The second possibility is that this effect of insulin is due to inactivation of the tyrosine kinase responsible for phosphorylating ErbB2. Together with the other data, this suggests that the PMA-induced tyrosine phosphorylation of ErbB2 may be due to some extrinsic kinase such as Pyk2, rather than ErbB2 autophosphorylation.
The apparently different mechanisms of insulin's inhibition of ErbB2 and ErbB3 phosphorylation allows us to eliminate some potential modes of ErbB2 and ErbB3 activation by PMA and to hypothesize about other possible mechanisms. First, although heregulin-induced phosphorylation of ErbB3, which has low intrinsic kinase activity (35), depends on transphosphorylation by ErbB2 (46 -51), this does not appear to be the mechanism of activation employed by PMA. Thus, pretreatment of cells with insulin and an inhibitor of PI 3-kinase (LY294002) is able to restore PMA-induced phosphorylation of ErbB3, at a time when PMA-induced phosphorylation of ErbB2 is reduced. This would suggest that ErbB2 is not responsible for phosphorylating ErbB3 in the PMA-treated cells. In addi- FIG. 8. Model for activation of ErbB2 and ErbB3 by PMA and sphingomyelinase. A model depicting activation of ErbB2 and ErbB3 by PMA and sphingomyelinase (SMase) based on the data presented here. Briefly, PMA and sphingomyelinase stimulate the tyrosine phosphorylation of ErbB2 and ErbB3. PMA, but not sphingomyelinase-induced phosphorylation, is dependent on protein kinase C (PKC). It appears as though phosphorylation of ErbB2 and ErbB3 is mediated by different protein tyrosine kinases, ErbB2 kinase(s) (EB2 kinase(s)) and ErbB3 kinase(s) (EB3 kinase(s)), respectively. Insulin has a 2-fold negative effect on this pathway. First, prolonged treatment with insulin results in decreased expression of ErbB2 and ErbB3. This effect of insulin is dependent on activation of phosphatidylinositol 3-kinase (PI 3-K). Second, insulin inhibits the tyrosine phosphorylation of ErbB2 and ErbB3 induced by PMA and sphingomyelinase. In the case of ErbB3 this effect is dependent on PI 3-K and appears to be due to the decreased expression of ErbB3. Insulin's inhibition of ErbB2 phosphorylation is independent on PI 3-K and appears to occur via a mechanism distinct from down-regulation of ErbB2. This model is examined more closely under "Discussion." TNFR, TNF receptor. tion, this observation suggests that different kinases are responsible for phosphorylating ErbB2 and ErbB3. EB2 kinase(s) appears to be negatively regulated by insulin in a PI 3-Kindependent manner. Although we cannot determine if EB3 kinase(s) is also negatively regulated by insulin, since insulin's inhibition of ErbB3 phosphorylation may be accounted for by insulin's ability to decrease ErbB3 expression, however, if insulin does inhibit or down-regulate EB3 kinase(s) it is likely to be dependent on activation of PI 3-K.
Although many questions remain regarding the ligand-independent activation of ErbB2 and ErbB3 by PMA and sphingomyelinase, some general conclusions can still be drawn. First, it is clear that there is extensive cross-talk between several very different signaling pathways including the protein kinase C pathway, the sphingomyelinase pathway, and the ErbB2/ ErbB3 tyrosine kinase pathway. Although the mechanisms of activation appear to be different, the final result on ErbB2 and ErbB3 phosphorylation and activation is similar. Second, the insulin signaling pathway can play a negative regulatory role in this activation process. It has previously been shown that PMA (52) and TNF-␣ (6, 7) have a negative effect on the insulin signaling pathway. In the current study, we have demonstrated that this negative cross-talk also proceeds in the opposite direction, insulin inhibits activation of ErbB2 and ErbB3 by PMA, as well as by sphingomyelinase and heregulin. The potential physiological and pathological consequences of these pathways of positive and negative cross-talk in tumor growth and cellular regulation remain to be determined.