Transactivation of the Epidermal Growth Factor Receptor in Endothelin-1-induced Mitogenic Signaling in Human Ovarian Carcinoma Cells¹ Free

ETs³are three closely related peptides, ET-1, ET-2, and ET-3, which were originally identified by the potent vasoconstrictor activity of ET-1 on vascular smooth muscle cells (1, 2). ETs have since been found to have a wide range of biological actions, including mitogenic effects, in many mammalian cell types (3). The biological actions of ETs are mediated through binding to ET_A and ET_Breceptors which are members of the seven transmembrane GPCR family. ET_A receptors display higher affinity for ET-1 and ET-2, whereas ET_B receptors show comparable binding affinities for the three isopeptides; the receptor subtypes also differ in their intracellular regions and in their binding specificities toward different G proteins (4). Mitogenic actions of ETs have been observed in cell types of different origin, including rat vascular smooth muscle cells (5),rat and murine fibroblasts (6, 7), rat renal mesangial cells (8), and human melanocytes (9),keratinocytes (10), and astrocytes (11).

ETs also exert paracrine and autocrine mitogenic actions in several types of tumor cells (12). The latter effects include ET-1-driven positive feedback loops that are mediated by ET_A receptors in cultured human ovarian(13, 14) and prostate (15, 16) carcinoma cells, and may contribute to tumor cell growth in vivo. ET-1-induced mitogenic signals include rapid and transient induction of early response genes, namely c-fos, c-jun,and c-myc (5, 17). However, the complex array of pathways that mediate these nuclear responses has not been clearly defined. Among downstream events after ET_A or ET_B receptor activation,Ca²⁺ release from intracellular stores(18), activation of PKC (6), phospholipase C(5), and phospholipase D (19), increased cAMP levels (20) and tyrosine kinase activities(21) have been implicated in different cell types showing a proliferative response to ET-1. These signals probably act in concert through a complex interplay between the individual pathways(22), but their relative importance in contributing to the mitogenic response has not been fully clarified (23). After the recognition of a prominent role for tyrosine kinase activities in ET-1 mitogenic signaling, agonist binding to ET receptors and other GPCRs was found to be associated with activation of MAPK family members, including Erk (24), Jun kinase(25), and p38 (26). Tyrosine kinase-dependent activation of the ras/MAPK pathway is an important step in ET-1-induced mitogenic signaling (27). Intermediate signaling molecules involved in ET-1-stimulated tyrosine kinase pathways include the adaptor proteins Shc and Grb2 (28, 29), and the nonreceptor kinases Src (30),pp125^FAK (31), and PI3K(32).

Recently, a novel aspect of the role of tyrosine kinase signaling in ET action has been revealed by the finding in Rat-1 fibroblasts that the ET_A receptor-mediated mitogenic response, as well as other GPCR-mediated responses, is accompanied by transactivation of the EGF-R. This event leads, through the formation of Shc/Grb-2/Sos complexes, to activation of the ras/MAPK pathway and transcription of early response genes. (33) This pathway has been postulated to have a pivotal role in the activation of MAPK after ligand binding to GPCRs and to possibly be of relevance to the pathogenesis of diseases such as cancer. We, therefore, analyzed the mitogenic actions of ET-1 in ovarian carcinoma cells to define the role of transactivation of receptor tyrosine kinase pathways in this process. In these cells, the mitogenic action of ET-1 is related,at least in part, to increased tyrosine kinase activity because it is reduced by enzyme inhibitors such as genistein and herbimicin(34). The only tyrosine kinase activity identified in ET-1-stimulated ovarian carcinoma cells to date is the cytoplasmic nonreceptor kinase pp125^FAK (34),the ability of which to activate the mitogenic ras/MAPK pathway is still controversial. We have previously observed that hEGF is a potent mitogenic stimulus in OVCA 433 cells (34). The present studies were performed to analyze the activation of the EGF-R kinase and downstream events in response to EGF and ET-1, and to determine the extent to which transactivation of the EGF-R mediates the mitogenic response to ET-1 in OVCA 433 cells. In this cell line, the ability of ET-1 to behave as an autocrine growth factor suggests that it may contribute to the progression of human ovarian tumors.

The human ovarian carcinoma cell line OVCA 433 (20) was a generous gift of Dr. Giovanni Scambia (Catholic University School of Medicine, Rome, Italy). Cells were cultured in DMEM (Whitthaker Bioproducts, Inc., Walkersville, MD) containing 2 mml-glutamine, 1% penicillin-streptomycin, and 10% FCS in 75-cm² plastic flasks at 37°C under 5%CO₂-95% air. To perform experiments for analysis of tyrosine phosphorylation, cells were plated in 100-mm Petri dishes. When the cells reached ∼80% confluence, the cultures were serum-starved for 24 h in DMEM not supplemented with FCS to reach quiescence. Quiescent cells were then stimulated with agonists after pretreatment with inhibitors or antagonists as appropriate.

ET-1 peptide and the ET_A selective inhibitor BQ123 were obtained from Peninsula Laboratories (Belmont, CA), and hrEGF was from Collaborative Biomedical Products, (Bedford, MA). The EGF-R kinase inhibitor tyrphostin AG1478 and the PKC inhibitor GF109203X were purchased from Calbiochem (La Jolla, CA); the PI3 kinase inhibitor wortmannin was purchased from Sigma (St. Louis, MO). Monoclonal and polyclonal Abs to Shc, Grb2, and Paxillin were obtained from Transduction Laboratories (Lexington, KY); monoclonal Ab to EGF-R was purchased from UBI (Lake Placid, NY), and rabbit polyclonal Ab to Erk-2 MAPK was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit MBP was purchased from Sigma.

Cells were seeded in 96-well plates (1 × 10⁴ cells/well), grown to ∼80% confluence, and then incubated in serum-free medium for 24 h to induce quiescence. Mitogenic stimuli were then added at time 0, after—when required—a short period of pretreatment with tyrphostin AG1478 (15 min), BQ123 (30 min), wortmannin (30 min), or GF109203X (30 min). After 18 h, 1μCi (37 kBq) [methyl-³H]thymidine(6.7 Ci/mmol); DuPont, New England Nuclear Research Products,Wilmington, DE) was added to each well. Six h later the culture media were removed, and the cells were washed three times with PBS, treated for 15 min with ice-cold 10% trichloroacetic acid, and washed twice with 100% ethanol; labeled material was then solubilized by treatment with 0.4 n sodium hydroxide at 37°C for 30 min. Cell-associated radioactivity was then quantitated by liquid scintillation counting. Responses to the mitogenic stimuli were assayed in sextuplicate, and results were expressed as the means of these samples; means of three separate experiments have been used to plot the graph (Fig. 8).

OVCA 433 cells were grown to ∼80% confluence in 100-mm tissue-culture-treated Petri dishes and then were serum-starved for 24 h. After the addition of agonists or antagonists to the dishes for the proper time period, the cells were rapidly washed with ice-cold PBS and lysed with 0.8 ml of ice-cold lysis buffer [50 mmTris-HCl (pH 7.4), 100 mm NaCl, 50 mm sodium fluoride, 5 mm EDTA, 1 mm orthovanadate, 0.06 units of aprotinin, 1 mm phenylmethylsulfonyl fluoride, and 10 μg/ml leupeptin]. After centrifugation for 10 min at 14,000 rpm in Eppendorf centrifuge to remove insoluble material,the lysates were precleared for 30 min at 4°C by incubation with protein A-Sepharose CL-4B (Pharmacia, Uppsala, Sweden), and immunoprecipitation was performed by incubation for 1.5 h at 4°C with Abs insolubilized on protein A-Sepharose CL-4B. The immunoprecipitates were washed six times with lysis buffer, solubilized in 2% SDS Laemmli buffer under reducing conditions, and analyzed by electrophoresis on 7.5 or 10% polyacrylamide gels (SDS-PAGE) followed by immunoblotting. Total cell lysates (50 μg) in Laemmli buffer were loaded on a 7.5% acrylamide gel to analyze tyrosine phosphorylated proteins or on a 11.25% acrylamide gel(acrylamide:bisacrylamide, 30:0.2) to analyze Erk 2 mobility shift.

Blotting to polyvinylidene difluoride of total cell lysates or immunoprecipitates after SDS-PAGE was performed at 70V for 3–4 h in 10 mm CAPS, pH 11, and 20% (v/v) methanol, or in conventional Tris/glycine/methanol transfer buffer (EGF-R immunoprecipitates). Nonspecific binding of Abs was prevented by incubating the blotted polyvinylidene difluoride membrane in 1% BSA in TBS-T (Tris-buffered saline, 0.5% Tween 20) for 1 h at room temperature. The blots were then incubated for 1 h with antiphosphotyrosine monoclonal Ab (0.5 μg/ml; clone 4G10; Upstate Biotechnology Incorporated, Lake Placid, NY), or with Abs to specific proteins appropriately diluted in TBS-T and 1% BSA. Membranes were then washed three times for 5 min each at room temperature with TBS-T and were subsequently incubated for 45 min at room temperature with peroxidase-labeled affinity-purified goat antimouse or antirabbit Ab(Bio-Rad Laboratories, Hercules, CA). After three washes for 5 min with TBS-T, immunostained bands were visualized using the Enhanced ChemiLuminescence (ECL) detection system according to the manufacturer’s instructions (Amersham) using Kodak AR-5 or Amersham ECL autoradiographic film. Quantitative densitometric analysis was performed by ImageQuant software. All of the results shown are representative of at least three separate experiments.

OVCA 433 cell lysates from 100-mm Petri dishes were immunoprecipitated using anti-Erk 2 rabbit polyclonal Ab previously linked on protein A-Sepharose CL-4B. The immunoprecipitates were washed four times with lysis buffer and twice with kinase buffer (35 mm Tris/HCl(pH 7.5), 15 mm MgCl₂, and 1 mm MnCl₂) and then were incubated in this same buffer containing 5 μg of MBP and 3 μCi of[γ-³²P]ATP (10 μm) for 30 min at 30°C. The reaction was blocked by the addition of hot Laemmli buffer, and the samples were analyzed on 15% SDS-PAGE. Gels were dried, and phosphorylated MBP was visualized by autoradiography.

The effects of EGF and ET-1 on tyrosine phosphorylation in OVCA 433 cells are shown in Fig. 1. Cells were stimulated for 5 min with 10 ng/ml EGF (Lanes 2and 3) or 20 nm ET-1 (Lanes 4-6) in the absence or presence of the specific EGF-R kinase inhibitor, tyrphostin AG1478 (Ref. 35; Lanes 3 and 5) or of the ET_A-selective antagonist BQ123 (Ref. 36; Lane 6), at a concentration 125 nm and 1μ m, respectively. EGF-stimulated cells(Lane 2) showed marked increases in several phosphorylated proteins, including a 170-kDa band (possibly EGF-R), a 105-kDa band, and 66-, 52-, and 46-kDa bands that correspond in position to the three Shc isoforms (Fig. 1,A). All of these changes were prevented by pretreatment with tyrphostin AG1478(Lane 3). It is evident that some of the phosphorylated components induced by EGF stimulation (Lane 2) are also present in ET-1-stimulated cells (Lane 4). These include the 170-kDa band, which is clearly visible in Fig. 1 B after a longer exposure time of the same blot; two bands whose electrophoretic mobility corresponds to the 52- and 46-kDa isoforms of the adaptor molecule Shc; and a minor phosphorylated band at ∼30 kDa. The phosphorylation pattern of ET-1-treated cells also includes an increase in phosphorylation in the 60- to 65-kDa range, two discrete bands at 78 and 95 kDa, and a broad band in the 120- to 130-kDa range that encompasses the location of pp125FAK, a cytoplasmic kinase we have previously shown to be phosphorylated in response to ET-1 in these cells (34). In ET-1-stimulated cells, tyrphostin AG1478 inhibited phosphorylation only of those bands that were common to both ET-1- and EGF-stimulated cells (Lane 5). As expected, all of the ET-1-induced bands were abolished in cells stimulated in the presence of BQ123 (Lane 6). The complex phosphorylation pattern observed in ET-1-stimulated cells was consistent with activation of both tyrphostin AG1478-sensitive and tyrphostin AG1478-insensitive tyrosine kinase activities. Furthermore, the electrophoretic mobilities of the tyrphostin AG1478-sensitive bands suggested that phosphorylation of the 170-kDa EGF-R and of the 46- and 52-kDa Shc isoforms was induced by ET-1 stimulation.

In view of the reported role of EGF-R transactivation in ET-1 mitogenic signaling in Rat-1 fibroblasts (33) and the phosphorylation patterns observed in ET-1-stimulated OVCA 433 cells, we determined whether mitogenic concentrations of ET-1 could induce EGF-R phosphorylation in these cells. As illustrated in Fig. 2,A, the immunoprecipitated 170-kDa EGF-R was highly phosphorylated after exposure to 10 ng/ml EGF (Lane 3). Cells stimulated for 5 min with 20 nm ET-1 (Lane 4) showed a 2.2-fold increase in EGF-R phosphorylation, as evaluated by quantitative densitometric analysis, that was prevented by the ET_A-specific antagonist, BQ123 (Lane 8). Furthermore, EGF- and ET-1-induced EGF-R phosphorylation was greatly reduced or abolished in the presence of tyrphostin AG1478(Lanes 5 and 6), as suggested by the total phosphorylation patterns shown in Fig. 1. A nonspecific inhibitory effect of tyrphostin AG1478 on EGF-R phosphorylation was excluded by the following experiment in which a different pattern of inhibitory actions of tyrphostin AG1478 and BQ123 was observed when the same cell extracts were immunoprecipitated with a monoclonal Ab to the cytoplasmic protein, paxillin, phosphorylation of which is induced by ET-1 (37). As shown in Fig. 2 B, ET-1-induced paxillin phosphorylation in OVCA 433 cells (Lane 4) was unaffected by tyrphostin AG1478 pretreatment (Lane 6) but was abolished in the presence of BQ123 (Lane 8). Furthermore, EGF stimulation also induced a moderate degree of paxillin phosphorylation, in accord with recent observations of pp125FAK phosphorylation in response to EGF in Rat-1 cells (38). These results indicate that the inhibition of ET-1-stimulated EGF-R phosphorylation by tyrphostin AG1478 is highly specific, because ET-1-induced phosphorylation of paxillin, which appears to be independent of EGF-R kinase activation, was unaffected by this agent.

EGF-R transactivation by ET-1 through the ET_Areceptor has been reported to be accompanied by phosphorylation of the adaptor protein Shc in Rat-1 fibroblasts (33), and ET_B-mediated Shc phosphorylation has been described in ET-1-treated rat astrocytes (29). As shown in Fig. 3, immunoprecipitation of extracts from EGF-stimulated OVCA 433 cells with an anti-Shc Ab revealed increased phosphorylation of all three of the Shc isoforms (Lane 2), and this was prevented by tyrphostin AG1478 (Lane 3). The ET-1-treated cells also showed increases in Shc phosphorylation, most prominently of the 52-and 46-kDa isoforms (Lane 4), which were inhibited by BQ123 as well as by tyrphostin AG1478, suggesting activation of EGF-R kinase activity after ET_A engagement. Furthermore, we observed that a phosphorylated 170-kDa band, migrating with the same mobility as EGF-R, was coimmunoprecipitated by anti-Shc Abs, both in EGF-stimulated and, to a much lesser extent, in ET-1-stimulated cells(not shown). A kinetic analysis of Shc phosphorylation in response to ET-1 exposure (Fig. 4,A) revealed that phosphorylation of Shc was a rapid event,peaking at between 2 and 5 min, and preferentially involved the 52-kDa isoform with a minor increase in the 46-kDa isoform. After longer exposure times, a small degree of phosphorylation was also detectable in the 66-kDa isoform (Fig. 4,B). A decrease was observed after 15 min, and phosphorylation returned to basal levels at between 30 and 60 min (Fig. 4, A and B)

After activation of the EGF-R and other receptor tyrosine kinases, Shc phosphorylation creates a docking site for the SH2 domain of the adaptor protein Grb2, which in turn interacts through its SH3 domain with the Ras nucleotide exchanger, Sos, leading to the activation of Ras. To determine whether ET-1-induced phosphorylation of Shc was accompanied by its recruitment into functional complexes, the association of Grb2 molecules with phosphorylated Shc was analyzed by Shc immunoprecipitation from cells stimulated with ET-1, followed by immunoblotting with an anti-Grb2 monoclonal Ab. As shown in Fig. 5, control immunoprecipitates from unstimulated cells contained only a small amount of Grb-2. This increased progressively after 5, 15, and 30 min of exposure to ET-1, and returned to basal levels after 60 min. This profile was in accord with the kinetics of Shc phosphorylation as determined by hybridization of the upper part of the blot with antiphosphotyrosine Ab (not shown). Stimulation with EGF (10 ng/ml for 5 min) caused a major increase in Grb2 binding to Shc, in accordance with the more pronounced Shc phosphorylation observed in EGF-treated cells. The specificity of the immunoreactivty was confirmed by the identical electrophoretic mobility of the Grb2 band detected in unstimulated cell extracts. Thus, Shc is not only transiently phosphorylated in response to ET-1 stimulation but also becomes functionally complexed with Grb2 with kinetics paralleling its degree of phosphorylation, as previously observed in rat fibroblasts,astrocytes, and renal glomerular mesangial cells (28, 33, 39).

ET-1 binding to the ET_A receptor leads to activation of MAPK, in particular of Erk 2, in several cell types. We have previously shown that ET-1-induced mitogenic signaling in ovarian carcinoma cells is accompanied by phosphorylation and activation of Erk 2 (34). An experiment to determine whether this event is related to EGF-R transactivation, using Western-blot analysis of whole cell extracts, is shown in Fig. 6,A. Phosphorylation of MAPK Erk 2 is indicated by the appearance of an electrophoretic band of reduced mobility. As expected,the EGF-induced activation of Erk 2 (Lane 2) was prevented by tyrphostin AG1478 (Lane 3). However, ET-1-induced activation (Lane 4) was only partly inhibited by tyrphostin AG1478 (35% reduction in the phosphorylated:unphosphorylated Erk 2 ratio; Lane 5), but was almost abolished by BQ123(Lane 6). The activation of Erk 2 and the effects of tyrphostin AG1478 and BQ123 were confirmed performing an in vitro kinase assay using MBP as a substrate for the immunoprecipitated Erk 2 (Fig. 6,B). These results indicate that OVCA 433 cells possess specific ET-1-induced signaling pathways,in addition to the EGF-R/Shc/Grb2 pathway, that converge with the latter at the level of Erk 2 MAPK and promote its activation independently of EGF-R transactivation. The kinetics of Erk 2 phosphorylation in response to 20 nm ET-1 (shown in Fig. 6,C), exhibited a time-delay as compared with Shc phosphorylation (Fig. 4), being minor at 2 min and maximal at 5–10 min, and returned to near-basal levels at between 30 and 60 min.

The above indications of a role for EGF-R transactivation in ET-1-induced mitogenic signaling prompted a more detailed analysis of the response to ET-1 in the absence or presence of tyrphostin AG1478. This agent should inhibit the cascade of phosphorylation-dependent events downstream of the EGF-R while not interfering with pathways dependent on other tyrosine kinase activities. To this end, a kinetic analysis of EGF-R, Shc, and Erk 2 phosphorylation, and of Grb2 recruitment in complexes with Shc, was performed in cells stimulated with ET-1 in the absence or presence of tyrphostin AG1478. Cell extracts were divided into aliquots and processed in parallel to minimize experimental variations, and the effects of inhibitor were compared in ET-1- and EGF-stimulated cells (Fig. 7). ET-1-induced EGF-R transactivation was evident at 5–15 min, and receptor phosphorylation returned to near basal levels within 60 min. Shc phosphorylation followed the same kinetics as EGF-R phosphorylation and was inhibited, as well as the latter, by tyrphostin AG1478. Its recruitment in complexes with Grb2 exhibited a similar pattern,depending on its degree of phosphorylation. Finally, the activation of Erk 2 MAPK by ET-1 was significantly but incompletely reduced by tyrphostin AG1478, whereas EGF-induced activation was completely abolished. These results indicate that dual signaling pathways converge on Erk 2 MAPK activation, one dependent on EGF-R kinase activity(tyrphostin-sensitive) and the other independent of EGF-R, Shc, and Grb2 (tyrphostin-insensitive).

The above observations prompted us to analyze the contribution of the tyrphostin AG1478-sensitive pathway to the mitogenic effect of ET-1 in OVCA 433 cells by performing thymidine incorporation assays (Fig. 8). In the presence of 62.5 nm tyrphostin AG1478, a concentration that did not significantly impair thymidine incorporation in response to 10% FCS, the mitogenic effect of 10 ng/ml hrEGF was abolished, whereas the proliferative response to 10 nmET-1, was significantly (P < 0.01) impaired,with a 38% reduction in the rate of thymidine incorporation. On the other hand, 1 μm BQ123, an ET_A antagonist abolished ET-1-induced thymidine incorporation but had no significant effect on EGF-induced proliferation. These results confirm the contribution of EGF-R transactivation pathway to ET-1 mitogenic activity in human ovarian carcinoma cells, in accord with the above analysis of tyrosine phosphorylation events and MAPK activation. Furthermore, they suggest that different kinases contribute to ET-1-induced Erk 2 activation and proliferation in OVCA 433 cells.

Because the G_q-coupled α1B adrenergic and M1 muscarinic receptors can activate p42 and p44 MAPK by a ras-independent and PKC-dependent pathway (40), we determined whether PKC contributes to ET-1-induced Erk 2 phosphorylation in OVCA 433 cells. To this end, Erk mobility was analyzed in extracts from cells stimulated for 5 min with 20 nm ET-1 or 60 nm TPA in the absence or presence of the PKC inhibitor GF109203X at a concentration 800 nm(41). As shown in Fig. 9 A, the ET-1-induced Erk 2 mobility shift was unaffected by the PKC inhibitor, whereas a marked inhibition of Erk 2 activation was observed in control TPA-stimulated cells. These data indicate that although PKC activity contributes to ET-1-stimulated mitogenic signaling in OVCA 433 cells (34), this action does not involve ras-independent Erk 2 activation.

In some GPCR-coupled signaling pathways, PI3K has been implicated by studies using wortmannin, a specific inhibitor of this lipid kinase(42), in signaling that leads to Shc phosphorylation. In experiments performed to evaluate the role of PI3K in ET-1-induced signaling in ovarian carcinoma cells, 80 nm wortmannin reduced both ET-1- and EGF-induced Erk 2 activation, as indicated by the electrophoretic mobility shift of the phosphorylated form (Fig. 9,B). Quantitative densitometric analysis indicated that wortmannin caused a 32% reduction in the phosphorylated:unphosphorylated Erk 2 ratio in ET-1-stimulated cells. To determine whether PI3K activation accounts for the tyrphostin AG1478-insensitive Erk2 activation induced by ET-1, additional experiments were performed in the presence of both tyrphostin AG1478 and wortmannin. As shown in Fig. 9 C, ET-1-induced Erk 2 phosphorylation was only partially blocked in the presence of both inhibitors. Furthermore, the degree of inhibition (36%) was similar to that caused by tyrphostin AG1478 or wortmannin alone. These results indicate that the tyrphostin AG1478-insensitive activation of Erk 2 is only partially attributable to the contribution of a PI3K-dependent pathway.

The possibility that PI3K could be involved upstream of EGF-R transactivation was also examined. However, as shown in Fig. 10, A and B, ET-1-induced EGF-R and Shc phosphorylation were not significantly inhibited by wortmannin. In fact, basal and EGF- or ET-1-induced EGF-R phosphorylation were increased in cells treated with wortmannin. This effect was observed even in the presence of tyrphostin AG1478, and counteracted the inhibitory effect of wortmannin on basal and induced phosphorylation levels (Fig. 10,A). Treatment with wortmannin also had minor effects on EGF-induced Shc phosphorylation, modifying the phosphorylation pattern of isoforms and coprecipitated bands, but no changes were observed in ET-1-stimulated cells (Fig. 10 B). Even in the presence of these interfering effects of wortmannin, it was evident that ET-1-induced EGF-R transactivation and Shc phosphorylation were not impaired by the PI3K inhibitor, whereas Erk 2 activation was significantly reduced.

These studies have demonstrated that phosphorylation of the EGF-R is induced during the ET-1-stimulated growth response of ovarian carcinoma cells. Such transactivation of the EGF-R is accompanied by a coordinate increase in the phosphorylation of the adaptor molecule Shc,and its recruitment in complexes with the SH2/SH3 adaptor, Grb2. These findings suggest the existence of an EGF-R-dependent route leading to the ras/MAPK activation pathway. Furthermore, the ET-1-induced phosphorylation of Erk 2 MAPK is partially dependent on EGF-R transactivation, as indicated by the effects of EGF-R kinase inhibition by tyrphostin AG1478. There is now abundant evidence that activation of GPCRs, in particular by ligands that elicit mitogenic responses, can induce transactivation of receptor tyrosine kinases. These transactivations include platelet-derived growth factor-R and EGF-R activation by angiotensin II in vascular smooth muscle cells (43); EGF-R and p185^neu/ErbB2 activation by lysophosphatidic acid (LPA), ET-1, and thrombin in Rat-1 fibroblasts (33); EGF-R activation by LPA in Cos-7 cells (44); EGF-R activation by m1 muscarinic receptor (m1 mAChR) agonists in Cos-7 cells(45); and EGF-R activation by bradykinin in PC12 cells (46). These observations indicate that transactivation of receptor tyrosine kinases by GPCRs is not unusual and may be a significant factor in the cascade of events that follows activation of heterotrimeric G proteins. The present findings support this hypothesis by demonstrating EGF-R transactivation during the mitogenic response to ET-1 in human ovarian carcinoma cells, in which ET_A receptors mediate autocrine growth actions in vitro and possibly in vivo (13, 34, 47).

ET-1 has been found to elicit mitogenic responses in several human tumor cell types in vitro (12), and to exert autocrine mitogenic actions in human ovarian carcinoma cell lines(13, 14). Furthermore, analysis of tissue sections from a high percentage of human primary and metastatic ovarian carcinomas has revealed prominent in vivo expression of ET-1 and ET_A receptors that is restricted to the tumor cells (47). These observations, together with data from prostate carcinomas (15), astrocytomas (48),meningiomas (49), and melanomas (50), have indicated the relevance of ET-1 autocrine/paracrine circuits to the growth of several neoplasms. Transactivation of cell surface receptors for growth factors could represent one of the basic mechanisms underlying the mitogenic activity of ET-1, as originally suggested by the studies of Daub et al. (33) in rat fibroblasts. The present observations indicate that transactivation of a growth factor receptor by a GPCR, namely ET_A,can indeed determine mitogenic effects in tumor cells, in which growth factor overexpression is a common finding and is frequently related to abnormal growth-promoting activity.

In particular, elevated expression of EGF-R has been observed, and is frequently associated with a poorer prognosis, in tumors where autocrine or paracrine mitogenic effects of ETs have been demonstrated. These include ovarian carcinoma (51), glioblastoma multiforme (52), astrocytoma (53), and prostatic carcinoma (54). Furthermore, overexpression of the ErbB2 proto-oncogene, a member of the EGF-R family,represents a major prognostic factor in human ovarian carcinoma(55). In Rat-1 fibroblasts, ET-1 can transactivate not only endogenous or expressed human EGF-R but also endogenous p185^neu, the rat homologue of human ErbB2(33). It is possible that the effectiveness of ET-1 autocrine circuits is augmented in the presence of receptor tyrosine kinase overexpression. The finding that EGF-R transactivation in ovarian carcinoma cells is in part responsible for the mitogenic effect of agonist-induced ET_A receptor activation, and our previous demonstration that ET-1 exerts additive proliferative effects in the presence of maximally effective EGF concentrations (34), suggest that the coexistence of ET-1 and EGF/transforming growth factor α autocrine circuits in tumor cells could provide maximal growth advantage.

As indicated by the present and previous studies (34), the mitogenic effects of ET-1 in human ovarian carcinoma cells have an absolute requirement for tyrosine kinase activities. EGF-R transactivation-dependent and -independent pathways appear to converge at the level of Erk 2 MAPK, contributing both to its activation and to the mitogenic response as indicated by the decreased proliferative effects in the presence of tyrphostin AG1478. It should be noted that tyrosine kinase activities are not exclusively responsible for the mitogenic effects of ET-1, and that cross-talk with other signaling pathways could be a relevant characteristic of ET-1-induced proliferation. This is exemplified by the contribution of Ca²⁺ influx and release from intracellular stores to the ET-1-induced activation of Erk MAPK (56).

In all instances in which GPCR-induced EGF-R transactivation has been described to date, the formation of molecular complexes containing Shc,Grb2, and Sos has suggested the involvement of ras activation. In OVCA 433 cells, transient Shc phosphorylation is indeed induced in response to ET-1, as well as the recruitment of Shc in complexes with the adaptor Grb2. Both of these events are inhibited, as previously reported in rat fibroblasts (33), by the EGF-R kinase inhibitor tyrphostin AG1478. This indicates that both of these events are secondary to the activation of EGF-R kinase activity. Thus, EGF-R transactivation could play a critical role in the initiation of this signaling cascade leading to ras activation. However, in contrast to the findings in Rat-1 fibroblasts,transactivation of EGF-R is not the only mechanism leading to Erk 2 MAPK activation during ET-1 mitogenic signaling in OVCA 433 cells. We observed that although tyrphostin AG1478 completely inhibits the activation and recruitment of EGF-R, Shc, and Grb2 in response to both EGF and ET-1, as well as the Erk 2 phosphorylation and activation in response to EGF, it only partially inhibits the response to ET-1. The incomplete inhibition of ET-1-induced Erk 2 activation is paralleled by the effects of tyrphostin AG1478 on cell proliferation in response to EGF and ET-1 in OVCA 433 cells. Whereas ET-1-induced proliferation was only moderately affected (with an ∼38% reduction in [³H]thymidine uptake) by the presence of tyrphostin AG1478 during the 24-h stimulation period, EGF-induced proliferation was completely abolished. The present data, although demonstrating the role of Erk 2 activation in EGF-induced as well as in ET-1-induced mitogenic responses, indicate that an alternative pathway is also involved in ET-1 mitogenic signaling and Erk 2 activation. Its insensitivity to inhibition of EGF-R kinase activity prevents complete blockade of ET-1-induced Erk 2 activation by tyrphostin AG1478.

In an attempt to identify possible alternative pathways in ET-1 signaling leading to Erk 2 activation, we investigated two different mechanisms by using inhibitors of kinase activities. The first was related to the possible role of PKC-dependent MAPK Erk 2 activation in ET-1 mitogenic signaling. Several reports have described PKC-dependent and ras-independent Erk 2 activation pathway in ET-1 and GPCR signaling(24, 57, 58), and we have previously demonstrated that PKC contributes to the proliferative effect of ET-1 in ovarian carcinoma OVCA 433 cells (34). Furthermore, PKC activity has been recently implicated in EGF-R tyrosine phosphorylation in GPCR signaling(45). However, no effect on ET-1-induced Erk 2 activation was observed in the presence of the PKC inhibitor, GF109203X, which suggests that its activation is independent of a PKC-mediated activation of Raf kinase. This result is in accord with recent observations in Rat-1 cells transfected with the G_q-coupled, pertussis toxin-insensitive,bombesin/gastrin releasing peptide receptor and the neuromedin B preferring receptor. In these cells, bombesin- and neuromedin B-stimulated Raf-1 and Erk 2 activities were not inhibited by the specific PKC inhibitor GF109203X or by down-regulation of phorbol ester-sensitive PKC isoforms (59). Furthermore, it should be noted that the above-mentioned evidence of a role for PKC in GPCR-induced Erk activation was obtained in cell lines in which Erk activation appeared to be dependent on a pertussis toxin-sensitive ET_A receptor. This contrasts with our observations in OVCA 433 cells, in which ET-1-stimulated mitogenic effects were insensitive to pertussis toxin (34). These conflicting reports suggest the importance of the cell context in the complex sequence of events that leads to transcriptional activation of early-response genes during ET-1 signaling.

We also investigated the possible involvement of PI3K in ET-1 mitogenic signaling, in view of recent reports that Chinese hamster ovary cells expressing somatostatin receptors (60) and ET_A receptors (32) possess a Ca²⁺ and PKC-independent signaling pathway to MAPK activation that is dependent on PI3K, based on its inhibition by wortmannin. Furthermore, in accordance with its effects on MAPK activation, wortmannin was reported to inhibit ET-1-induced activation of Raf-B but not the effect of TPA thereon (32). Our data indicate that PI3K is not involved in the cascade of events that determines EGF-R transactivation, in accord with observations in GPCR-induced EGF-R transactivation models (46), but contributes to ET-1 signaling leading to Erk 2 activation. The inability of the simultaneous presence of both tyrphostin AG1478 and wortmannin to completely block Erk 2 activation indicates that PI3K probably does not account for the tyrphostin AG1478-insensitive Erk 2 activation. Moreover, because EGF-induced activation of Erk 2 is itself partially inhibited by wortmannin, its inhibition of ET-1 signaling could reflect an effect on the same EGF-R-dependent pathway. Consistent with this hypothesis, Erk 2 activation induced by low EGF concentrations (0.02–0.2 ng/ml) can be blocked by PI3K inhibitors(61). This effect was attributed to a permissive role of PI3K activity on this pathway at two levels, one upstream of Ras,involving Shc-Grb2-Sos complex formation at the plasma membrane, and one downstream of Ras, involving Rafdependent Mek phosphorylation. In our experiments, maximal EGF stimulation was still partially sensitive to wortmannin. PI3K and its lipid products are possibly involved in ET-1 signaling, both upstream of Ras, along the same EGF-R-dependent pathway, and between Ras and Mek, to affect different pathways leading to Erk 2 activation.

In summary, these findings demonstrate a significant but not exclusive role for EGF-R transactivation in ET-1 mitogenic signaling in human ovarian carcinoma cells. Our evidence indicates that EGF-R phosphorylation, Shc phosphorylation, and Grb2 recruitment in complexes with Shc are coordinate events related to EGF-R kinase activity. This pathway appears to converge with other unidentified but non-PKC-dependent pathways at the level of Erk 2 activation. Increased PI3K activity does not explain the EGF-R-independent activation but appears to be involved in ET-1-induced pathways leading to Erk 2.

We acknowledge the excellent technical assistance of Marco Varmi.