c-Src Is Activated by the Epidermal Growth Factor Receptor in a Pathway That Mediates JNK and ERK Activation by Gonadotropin-releasing Hormone in COS7 Cells*

Key participants in G protein-coupled receptor (GPCR) signaling are the mitogen-activated protein kinase (MAPK) signaling cascades. The mechanisms involved in the activation of the above cascades by GPCRs are not fully elucidated. A prototypic GPCR that has been widely used to study these signaling mechanisms is the receptor for gonadotropin-releasing hormone (GnRHR), which serves as a key regulator of the reproductive system. Here we expressed GnRHR in COS7 cells and found that GnRHR transmits its signals to MAPKs mainly via Gαi, EGF receptor without the involvement of Hb-EGF, and c-Src, but independently of PKCs. The main pathway that leads to JNK activation downstream of the EGF receptor involves a sequential activation of c-Src and phosphatidylinositol 3-kinase (PI3K). ERK activation by GnRHR is mediated by the EGF receptor, which activates Ras either directly or via c-Src. Besides the main pathway, the dissociated Gβγ and β-arrestin may initiate additional, albeit minor, pathways that lead to MAPK activation in the transfected COS7 cells. The pathways detected are significantly different from those in other cell lines bearing GnRHR, indicating that GnRH can utilize various signaling mechanisms for the activation of MAPK cascades. The unique pathway elucidated here in which c-Src and PI3K are sequentially activated downstream of the EGF receptor may serve as a prototype of signaling mechanisms by GnRHR and by additional GPCRs in various cell types.

regulator of the reproductive system. It acts via a specific GPCR (GnRHR) and triggers the synthesis of the common ␣and ␤-chains of the gonadotropins, which in turn, control the function of the gonads and induce steroidogenesis (reviewed in Refs. 29 and 30). In the pituitary-derived ␣T3-1 cells, it was shown that GnRHR transmits its signals primarily via G q , phospholipases, PKCs, and Ca 2ϩ , culminating in the activation of several MAPK cascades (reviewed in Refs. 9 and 31). Studies from our laboratories have shown that the signaling of GnRH in ␣T3-1 cells involves a direct activation of Raf-1 by PKC, and this step is partially dependent on a second pathway consisting of Ras activation downstream of dynamin and c-Src (32)(33)(34). The activation of JNK in these cells is also mediated primarily by PKC that further induces the sequential activation of c-Src and CDC42/RAC (35). Interestingly, few additional signaling pathways that can lead from GnRHR to MAPKs were identified in ␣T3-1 cells as well. Grosse et al. (36) showed that GnRH signals to ERK by activating EGF receptors whereas Mulvaney et al. (37,38) showed that ERK is activated in ␣T3-1 via calcium influx through L-type calcium channels and that JNK activation is PKC-independent but mediated by elevated intracellular calcium. On the other hand, Vasilyev et al. (39) showed that in L␤T4 cells short-term incubation with GnRH leads to induction of LH␤ transcription, whereas continuous long-term incubation leads to repression of the LH␤ transcription. We recently showed that in L␤T2, ERK, and JNK are involved in the expression of the LH␤ subunit promoter (40). GnRHR was found to utilize as yet additional distinct intracellular signaling pathways to activate MAPKs. These include PKA (41), independent G␤␥ subunits of the G i /G o proteins (42), and the EGF receptor (36,43,44). Collectively, these results indicate that GnRHR can utilize several signaling pathways in different cell types and under different conditions to execute a single intracellular effect. Therefore, GnRHR serves as a good experimental model to study signaling pathways that can participate in the activation of MAPK cascades by GPCRs.
In the current study we used GnRHR-expressing COS7 cells and found that both ERK and JNK are activated by GnRH with similar kinetics to that found in ␣T3-1 cells, but the mechanism that mediates this activation is significantly different. Thus, in the GnRHR-expressing COS7 cells, GnRHR transmits its signals to MAPKs mainly by activating the EGF receptor, although a minor contribution was detected also for the dissociated G␤␥ and ␤-arrestin. JNK activation by GnRHR in these cells is fully dependent on a sequential activation of c-Src and PI3K, which operate mainly downstream of the EGF receptor, but can be activated mildly also by G␤␥. On the other hand, ERK activation by GnRH in these cells is fully mediated by the EGF receptor, which activates Ras directly or via c-Src. This activation is not dependent on the secretion of heparin binding (Hb)-EGF as shown for other GPCRs (45). Thus, in transfected COS7 cells GnRH elicits is a unique signaling system in that it places c-Src downstream of EGF receptor in the pathway that leads from GPCR to MAPK cascades.
Transfection, Stimulation, and Harvesting of COS7 Cells-Subconfluent COS7 cells were transfected with 5 g of the GnRHR together with 5 g of either an examined plasmid or vector control. The transfection was carried out using the DE-dextran technique, and the transfection efficiency was 80 -95%. When time courses were determined, the transfected cells were pulled 16 h after transfection, and split again to the assay plate to ensure homogeneous expression of GnRHR in all plates. Thirty-two hours after transfection, the cells were serumstarved for 16 h and incubated for the desired time intervals with GnRH-a in the presence or absence of various inhibitors. After stimulation, cells were washed twice with ice-cold phosphate-buffered saline, once with Buffer A consisting of 50 mM ␤-glycerophosphate (pH 7.3), 1.5 mM EGTA, 1 mM EDTA, 1 mM dithiothreitol, 0.1 mM sodium orthovanadate. The cells were subsequently harvested in ice-cold Homogenization Buffer (Buffer H) consisting of 50 mM ␤-glycerophosphate (pH 7.3), 1.5 mM EGTA, 1 mM EDTA, 1 mM dithiothreitol, 0.1 mM sodium orthovanadate, 1 mM benzamidine, aprotinin (10 g/ml), leupeptin (10 g/ml), and pepstatin (2 g/ml). Cell lysates were centrifuged (20,000 ϫ g, 15 min), and the supernatant was assayed for protein content. For the determination of c-Src and EGF receptor, the cells were lysed with RIPA buffer consisting of 137 mM NaCl, 20 mM Tris, pH 7.4, 10% (v/v) glycerol, 1% Triton X-100, 0.5% (w/v) deoxycholate, 0.1% (w/v) SDS, 2.0 mM EDTA, 1.0 mM phenylmethylsulfonyl fluoride, and 20 M leupeptin. After 10 min in RIPA buffer (4°C), the cell lysates were centrifuged (20,000 ϫ g, 15 min), and the supernatant was subjected to c-Src assay as below.
Western Blot Analysis-Cell supernatants that contained cytosolic proteins were collected, and aliquots from each sample (20 g) were separated on 10% SDS-PAGE followed by Western blotting with the appropriate antibodies. Alternatively, immunoprecipitated samples were boiled in sample buffer and subjected to SDS-PAGE and Western blotting. The blots were developed with alkaline phosphatase or horseradish peroxidase-conjugated anti-mouse or anti-rabbit Fab antibodies (Jackson).
Determination of ERK Activity-Transfected COS7 cells were serumstarved (0.1% FCS, 16 h), and the examined stimulants were added for various time intervals. The cells were then washed (twice with phosphate-buffered saline and once with Buffer A), scraped into 300 l of Buffer H, and sonicated (50 watts, 2 ϫ 7 s), at 4°C. After centrifugation (20,000 ϫ g, 15 min, 4°C), aliquots of the resulting supernatant were subjected to immunoprecipitation with anti-ERK C-terminal antibody using protein A-or protein G-agarose (20 l). ERK activity was determined by the phosphorylation of myelin basic protein (MBP) as described (46). Solid Phase Assay for JNK Activity-Transfected COS7 cells were serum-starved (0.1% fetal calf serum, 16 h), and the examined stimulants were added for various time intervals. The cells were then washed (twice with phosphate-buffered saline and once with Buffer A), scraped into 250 l of Buffer H, and sonicated (50 watts, 2 ϫ 7 s), all at 4°C. After centrifugation (20,000 ϫ g, 15 min, 4°C), aliquots of the resulting supernatant were assayed by the Coomassie protein assay (Pierce) for protein. JNK activity was detected according to Hibi et al. (47). Briefly, aliquots (100 -150 g of protein) of the cell extracts were incubated (2 h, 4°C) with GST-c-Jun-(1-91) to allow JNK to bind to the substrate. After extensive washing, JNK activity was measured by phosphorylation of the GST-c-Jun (1-91), which was mediated by the bound kinase in the presence of 20 mM MgCl 2 , 20 M [␥-32 P]ATP (300 cpm/pmol) for 20 min at 30°C. The reactions were terminated by the addition of sample buffer, and the samples were subjected to SDS-PAGE and autoradiography on Kodak X-100 films. The phosphorylation of GST-c-Jun-(1-91) was quantitated by densitometry (Bio-Rad 690).
Determination of Ras Activity-Subconfluent COS7 cells were cotransfected with 5 g of the GnRHR together with 5 g of either an examined plasmid or vector control. Thirty-two hours after transfection, the cells were serum-starved for 16 h and treated with GnRH-a (10 Ϫ7 M) for various time points in the absence or presence of the tested inhibitors (15 min, 37°C). Following stimulation, the cells were lysed in Ral buffer (50 mM Tris-HCl, pH 7.5, 10% glycerol, 200 mM NaCl, 2.5 mM MgCl 2 , 1% Nonidet P-40, 1 g/ml aprotinin, 1 g/ml leupeptin, 0.1 mg/ml trypsin inhibitor, 250 M phenylmethylsulfonyl fluoride, 10 mM NaF, and 1 mM sodium orthovanadate). After 10 min in the Ral buffer (4°C) the cell lysates were centrifuged (14,000 ϫ g, 10 min) and the supernatant (300 g protein) was subjected for further treatment. The active GTP-bound form of Ras was precipitated by GST-Raf-RBD (20 g) and washed three times in a buffer containing 20 mM Hepes, pH 7.5, 0.15 M NaCl, 0.1% Triton X-100, and 10% glycerol and once with buffer A. The amount of Ras pulled-down was assessed by Western blotting using mouse anti-pan-ras antibody.
Determination of PKB Activity-Transfected COS7 cells were serumstarved (0.1% fetal calf serum, 16 h), and the examined stimulants were added for various time intervals. The cells were then washed (twice with phosphate-buffered saline and once with Buffer A), scraped into RIPA buffer. After centrifugation (20,000 ϫ g, 15 min, 4°C), aliquots of the resulting supernatant were subjected to immunoprecipitation with anti-PKB C-terminal antibody using protein A-agarose or protein Gagarose (20 l). PKB activity was determined by the phosphorylation of histone H2B as described for MBP phosphorylation by ERK above.
c-Src Activity-Cell lysates (400 -500 g of protein in Buffer H containing 1% Triton X-100) were incubated with anti-c-Src-antibody precoupled to protein A-Sepharose and mixed at 4°C. The immunocomplexes were washed once with RIPA, twice with 0.5 M LiCl in 0.1 M Tris-HCl, pH 8.0, and once with Buffer A. The washed immunoprecipitates were resuspended in a kinase assay buffer (35), and the c-Src activity was determined using acid-denatured enolase (3 M) as substrate in the presence of 20 M [␥-32 P]ATP (8,000 cpm/pmol). The enzymatic reactions were terminated by the addition of sample buffer. The samples were then subjected to SDS-PAGE and autoradiography. Alternatively the harvested fraction were separated by SDS-PAGE and subjected to Western blot analysis with anti-active c-Src antibody and anti-general c-Src antibody.

JNK1 Activation by GnRH-a in COS7 Cells-Signaling by
GPCRs is mediated via several distinct pathways that vary among cell types and stimuli. GnRHR has been proven as a good tool in the study of GPCR signaling mechanism toward MAPK cascades (9). To determine the cell type specificity of GnRH signaling and to study the effect of various signaling inhibitors on this activation we used COS7 cells that do not express endogenous GnRHR. These cells were transfected with a plasmid containing mouse GnRHR that yielded a considerable amount of expression of the GnRHR in most cells as judged by a Western blot analysis with anti-GnRHR antibody and expression of an unrelated green fluorescent protein (data not shown). To ensure similar level of expression of the GnRHR in all plates of any experiment, the transfected cells were combined and cut into smaller plates. The cells were serumstarved for 16 h prior to the stimulation, pretreated with various pharmacological inhibitors, and then stimulated with 10 Ϫ7 M GnRH-a. The high yield of transfection in COS7 cells allowed detection of endogenous MAPK activation, without a significant background from non-transfected cells. Thus, when examined with anti-DP-JNK antibody, a gradual change in recognition by the antibody was observed in two endogenous bands at molecular masses of 46 and 54 kDa, which corresponded to JNK1 and JNK2, respectively. Since the relative amount of staining of the 46 kDa JNK1 was stronger than that of the 54 kDa JNK2, we demonstrate here only the results of JNK1. Thus, expression of GnRHR in the COS7 without an addition of GnRH-a did not change the activity of JNK1. However, elevation in JNK1 phosphorylation was detected already 5 min after stimulation with GnRH-a, peaked at 10 -30 min after stimulation and declined thereafter (Fig. 1A). Pretreatment of the cells with the PTK inhibitor, Genistein as well as inhibitors of PI3K (wortmannin), EGF receptor (AG1478), and c-Src (PP1) but not the PKC inhibitor, GF109203X, significantly reduced the activation of JNK1 by GnRH-a. Similar results to those obtained with the anti-DP-JNK antibody were observed when the endogenous JNK activity was measured by an in vitro kinase assay (Fig. 1B). Again, Genistein, wortmannin, AG1478, and PP1 significantly prevented JNK activation by GnRH-a, while GF109203X had no inhibitory effect. This pattern of inhibition is markedly different from that obtained in ␣T3-1 cells where stimulation of JNK activity with GnRH-a was inhibited by GF109203X but not by wortmannin or AG1478 (Ref. 35 and data not shown), indicating that the pathway that leads to JNK activation by GnRH may differ in different cell lines.
Involvement of EGF Receptor, ␤␥ Dimer, c-Src, and ␤-Arrestin in JNK1 Phosphorylation by GnRH-a-To study the possible involvement of additional signaling components in the Gn-RHR to JNK pathway and to confirm the involvement of components that were identified by the inhibitors above, we coexpressed GnRHR together with interfering mutants of various signaling components into COS7 cells. Serum starvation and treatment with GnRH-a were followed as described above. As expected from the inhibition with AG1478 and PP1, the dominant negative form of the EGF receptor as well as Csk, which inhibits the activity of c-Src, nearly abolished the activation of JNK1 by GnRH-a ( Fig. 2A). ␤-Arrestin, which can serve as a mediator of signaling of GPCRs toward MAPKs (3,26), seemed to play a minor role in the GnRHR-JNK signaling. Although the wild-type ␤-arrestin had no significant effect on JNK1 activation by GnRH-a, the dominant negative form of this protein inhibited this activation by ϳ30%. Similar inhibition was exerted by CD8-tagged ␤-ARK, which acts as a scavenger for the dissociated ␤␥ dimer (48). On the other hand, dominant negative Ras, as well as the wild type and the dominant negative forms of FAK and dynamin, did not seem to influence the studied pathway. Taken together, these results suggest a major role for EGF receptor, c-Src, and PI3K, and a minor role for ␤-arrestin and ␤␥ dimer, in the pathway that links the GnRHR to JNK in the transfected COS7 cells. Other known signaling components such as PKC, FAK, dynamin, and Ras do not seem to be involved in this process. As seen in (data not shown) indicating that the amount of receptors in the transfected COS7 cells was not too high, which make the results more reliable.
ERK Activation by GnRH-a in COS7 Cells-We then studied the mechanism of ERK activation by GnRH in transfected COS7 cells. As above, the cells were serum-starved and treated, and anti-DP-ERK antibody was used to detect the phosphorylation of ERK in its activation loop. As with JNK1, the expression of GnRHR in COS7 cells did not change the level of the regulatory phosphorylation of endogenous ERK1 and 2 (Fig.  3A). Addition of GnRH-a to the transfected cells resulted in a substantial phosphorylation of both ERK1 and ERK2, which peaked at 5 min after treatment, remained high for additional 25 min and declined 30 min later. Pretreatment of the cells with the PTK inhibitor Genistein, AG1478, and to some extent also with PP1, inhibited the GnRH-a-induced phosphorylation of ERK. On the other hand, treatment with the PI3K inhibitor wortmannin and the PKC inhibitor GF109203X had no inhibitory effect. Similar results were observed also when the endogenous ERK activity toward MBP was measured by an in vitro kinase assay (Fig. 3B). Again, ERK was transiently activated by GnRH-a with a peak at 5-10 min after stimulation. Genistein, and AG1478 completely prevented the GnRH-a-induced ERK activation, PP1 inhibited this activation by ϳ35%, while GF109203X and wortmannin had no detectable inhibitory effect. As with JNK1, this pattern of inhibition is different from that obtained in ␣T3-1 cells, where stimulation of ERK by GnRH-a was inhibited by GF109203X, but not by AG1478 (33). Thus, the pathway that leads to ERK activation by GnRH-a differs in the different cell lines. Moreover, the sensitivity of ERK activation to inhibitors in the transfected COS7 cells was different from the sensitivity of JNK to the same inhibitors. This indicates that the pathway that leads to ERK activation by GnRH is different, at least in part, from the one leading to the JNK cascade.
To study the possible involvement of additional signaling components in the GnRHR to ERK pathway, we coexpressed the GnRHR together with interfering mutants of various signaling components in COS7 cells. As expected from the study using pharmacological inhibitors, the dominant negative form of EGF receptor significantly attenuated the phosphorylation of ERK1 and ERK2 upon GnRH-a treatment (Fig. 4). In addition, the expression of dominant negative Ras, which acts upstream of the ERK cascade in many systems (5) also caused a substantial reduction in GnRH-a-stimulated ERK phosphorylation. Csk, that inhibits c-Src activity, reduced the GnRHstimulated phosphorylation of ERK by ϳ40%. As observed for JNK1, the dominant negative form of ␤-arrestin inhibited this activation by ϳ30%. On the other hand, the G␤␥ scavenger CD8-␤-ARK, as well as the wild type and the dominant negative forms of dynamin and FAK had no effect. Taken together, the results suggest a role for the EGF receptor, Ras, c-Src, and possibly also ␤-arrestin in the pathway that links the GnRHR to ERK in transfected COS7 cells. Other known signaling components such as PKC, FAK, dynamin, PI3K, and dissociated ␤␥ dimer do not seem to be involved in this process.
Ras Activation by GnRH-a Is Mediated by EGF Receptor c-Src-One of the most important mediators of signals to the ERK cascade is the small GTP-binding protein Ras. Therefore we used the Raf-RBD pull-down assay (49) in order to detect the Ras activation by GnRH. As expected, Ras was activated within 2 min after GnRH-a addition to the transfected COS7 cells, the activity peaked at 5 min and decreased slightly 5 min later. Interestingly, when the cells were cotransfected with dominant negative EGF receptor or preincubated with the EGF receptor inhibitor AG4178, both basal activity and GnRH-astimulated activity were almost completely abolished (Fig. 5A). On the other hand, cotransfection with Csk or preincubation with the c-Src inhibitor PP1 also caused reduction (50 Ϯ 10%, Fig. 5B) in Ras activation, but their effect was significantly lower than that of the inhibitors of the EGF receptor. Therefore, it is reasonable to assume that EGF receptor is the main stimulator of Ras, where c-Src may mediate part of the EGF receptor-induced signal and the well established Grb-Sos pathway probably mediates the other part of the signal (9).

Role of EGF Receptor in the Mediation of GnRH Signals-
Our results show that EGF receptor plays a central role in the transmission of GnRHR-initiated signals toward JNK and ERK. Another signaling component that is involved in this process is c-Src that seems to be a central player in the pathway that leads to JNK activation and participates to some extent also in the GnRHR-ERK pathway. It became important to study the activation of these two components and the interplay between them. Using both anti-pEGF receptor and anti-PY antibodies we showed that the EGF receptor was rapidly activated by GnRH (Fig. 6, A and B), and remained active for more than 60 min. This activation was not affected by inhibitors of c-Src, indicating that this PTK is not located upstream of the FIG. 6. EGFR transactivation by GnRH. A and B, COS7 cells that were transiently transfected and serumstarved as described, were stimulated with GnRH-a (10 Ϫ7 M) for the indicated times and tested for their enhanced tyrosine phosphorylation using anti-phospho-EGF receptor (A) and anti-PY (B) antibodies. C, detection of Hb-EGF shedding in COS7 cells expressing GnRHR. Conditioned medium (CM) was collected and incubated with heparin-agarose to precipitate the Hb-EGF ectodomain. sHb-EGF secreted in the culture medium was detected by Western blot analysis using goat anti-Hb-EGF antibody. D, ERK activation in non-GnRHR-expressing COS7 cells treated with CM from GnRH-treated Gn-RHR-expressing cells: COS7 cells were transfected as described, serum-starved for 16 h, and treated with GnRH-a (10 Ϫ7 M) for 5 and 10 min. CM from transfected cells (lanes 1-6) was collected and transferred to non-transfected COS7 cells (lanes 7-12) that did not express the Gn-RHR and further incubated for the indicated times. Non-transfected cells stimulated (5 min) with CM from untreated transfected cells or cells transfected with vector alone were used as controls (lanes 7 and 8). Phosphorylation of ERK was determined with the appropriate anti-DP and anti-general ERK antibodies. EGF receptor as suggested in several other systems (50,51). A main mechanism for GPCR-induced activation of EGF receptor seems to be through activation of a membrane proteinase (e.g. MMP9), which in turn releases the membrane-bound Hb-EGF to activate the receptor (52). We examined the amount of Hb-EGF released to the medium upon activation of the GnRHRtransfected COS7 cells and found that it could not be detected in the first day after GnRH-a addition, but appeared in later time points after stimulation (Fig. 6C). In order to verify the lack of Hb-EGF, we used CM of GnRH activated COS7 cells to activate the MAPK in non-transfected COS7 cells as described by Pierce et al. (45). Indeed, the CM used did not induce any activation of ERK under any of the conditioned medium (Fig. 6D), indicating that the activation of EGF receptor is not mediated by Hb-EGF.
Activation of c-Src by GnRH-a Is Mediated Mainly by EGF Receptor-It became clear from the results above that c-Src is not localized upstream of the EGF receptor. We then undertook to find whether it may act downstream of this RTK in our system. Thus, addition of either GnRH-a or EGF to these cells triggered a sustained activation of c-Src (Fig. 7A) that was apparent already 2 min upon treatment. Moreover, addition of the EGF receptor inhibitor, AG1478, or cotransfection of the dominant negative form of the EGF receptor inhibited most of the activation of c-Src by GnRH-a (Fig. 7B). Inhibition was also detected with the G␤␥ scavenger, but neither with the dominant negative forms of dynamin or ␤-arrestin (Fig. 7, C and D), nor with the PKC inhibitor GF109203X (data not shown).
These results indicate that c-Src is activated by a unique mechanism that involves mainly activation of the EGF receptor, and this is complemented by a signal via the dissociated G␤␥. PKC, which plays a key role in the activation of c-Src by GnRH in ␣T3-1 cells (35), dynamin, and ␤-arrestin do not seem to participate in c-Src activation in the GnRHR-expressing COS7 cells.
Activation of PI3K/PKB by GnRH-a Occurs Downstream of c-Src-As shown above, inhibition of PI3K by wortmannin influenced the activation of JNK but not ERK by GnRH in transfected COS7 cells. PI3K has been implicated in the signaling of few GPCRs and was shown to be activated mainly by dissociated ␤␥ dimer (53), or activation of several PTKs (9). Since PKB is a known substrate for PI3K, we used the phosphorylation of this protein kinase as readout for the activation of the PI3K by GnRH. Indeed, using either anti-phospho-PKB antibody or in vitro kinase assay toward histone H2B, we found that both phosphorylation and activity of PKB were gradually activated upon GnRH-a treatment (Fig. 8, A and B). Pretreatment with the PTK inhibitor Genistein and the PI3K inhibitor wortmannin abolished the phosphorylation of PKB, whereas the PKC inhibitor GF109203X had no effect on this process (Fig. 8A). We then undertook to study the cross-talk between the EGF receptor, c-Src, and PI3K in the mechanism of JNK stimulation by GnRH-a. Pretreatment of the cells with the c-Src inhibitor PP1, as well as cotransfection of the cells with Csk, resulted in a pronounced inhibition of PKB activation by GnRH-a (Fig. 8C). These results indicate that the majority of the GnRH signal to PI3K is mediated by c-Src in the transfected COS7 cells. Furthermore, dominant negative EGF receptor also inhibited the GnRH-a-induced activation of PI3K/PKB (Fig. 8C), but this inhibition was not complete (ϳ70%). As shown above for the activation of c-Src, the other upstream component that had a partial inhibitory effect on PKB activation was the ␤␥ dimer (Fig. 8D), whereas the dominant negative forms of dynamin and ␤-arrestin did not influence this process (Fig. 8D). These results indicate that PI3K is localized in a pathway that involves sequential activation of c-Src, PI3K, and JNK upon GnRH treatment. These data also support the observation FIG. 7. Mechanism of c-Src activation by GnRH. A, COS7 cells were transfected with a plasmid containing mouse GnRHR. Thirty-two hours after transfection, the cells were serum-starved for 16 h, and one plate was pretreated with PP1 (5 M, 15 min). Then the cells were stimulated either with EGF (50 ng/ml) or GnRH-a (10 Ϫ7 M) for the indicated times. Phosphorylation of c-Src on Tyr-416 (activated c-Src) was detected with anti-phospho-Src antibody (p-Src). The amount of c-Src did not change throughout the experiment as detected by anti-c-Src antibody (Src). These results were reproduced twice. B, COS7 cells were cotransfected with mouse GnRHR together with either K721A-EGF receptor (Dn-EGFR) or with a vector control. One plate was transfected with vector alone (Vec). Thirty-two hours after transfection, the cells were serum-starved for 16 h after which one plate was pretreated with AG1478 (5 M, 15 min). Then the cells were stimulated with GnRH-a (10 Ϫ7 M for the indicated times) and c-Src activity toward denatured enolase was determined as described under "Materials and Methods." The amount of immunoprecipitated c-Src was determined using anti-c-Src antibody (lower panel). The results were reproduced three times. C, COS7 cells were cotransfected with plasmid containing mouse GnRHR together with plasmids containing either CD8-tagged ␤ARK (␤␥ scav); K44A-dynamin (Dn-Dyn); V54D-␤-arrestin2 (Dn-Arr), or no insertion as control. Two days after transfection, the cells were serum-starved for 16 h and then either treated with GnRH-a (10 Ϫ7 M; 10 min) or left untreated (0). Activation of c-Src was determined by Western blot analysis using anti-phospho c-Src antibody (p-Src) or antigeneral Src antibody (Src). These results were reproduced twice. D, the amount of activated c-Src was determined by densitometry and plotted as a bar graph of percent activation from that of GnRH-a-stimulated cells that were cotransfected with GnRHR and vector control in each experiment. These are averages and S.E. of three experiments.
above that c-Src activation by GnRH in the transfected COS7 cells is mediated mainly by the EGF receptor, and is complemented by the dissociated ␤␥ dimmer.
Involvement of G␣ i in GnRH-a-mediated Signaling-Our results clearly show that EGF receptor is the main upstream mediator of GnRH signaling; however, it is not clear how this receptor is activated by the GnRHR. It is well established that in ␣T3-1 cells, GnRHR transmits its signal via G␣ q and PLC␤ (9). However, in the GnRHR-transfected COS7 cells we did not find any role for PKC in the activation of MAPKs, suggesting the involvement of other G proteins. To test the possible involvement of other G␣ isoforms we used pertussis toxin that serves as a selective inhibitor of G␣ i (54). The serum-starved GnRHR-transfected COS7 cells were pretreated with the toxin for 5 h, after which they were stimulated with GnRH-a and the phosphorylation of JNK, ERK, c-Src, and EGF receptor was . Cellular extracts of these cells were subjected to a Western blot analysis using either anti-phospho-PKB antibody (Ser-473; p-PKB) or anti-PKB antibody (PKB). These results were reproduced twice. B, COS7 cells were treated as in A and then subjected to immunoprecipitation with anti-PKB antibody followed by an in vitro phosphorylation reaction with histone H2B as a substrate (p-H2B). The phosphorylation was detected by autoradiography, and the amount of immunoprecipitated PKB was detected with anti-PKB antibody (PKB). These results were reproduced twice. C, COS7 cells were cotransfected with plasmid containing mouse GnRHR together with plasmid containing either Csk; K721A-EGF receptor (Dn-EGFR), or no insertion as vector control. Thirty-two hours after transfection, the cells were serum-starved for 16 h after which two plates were pretreated with the c-Src inhibitor PP1 (5 M, 15 min). Then the cells were stimulated with GnRH-a (10 Ϫ7 M for the indicated times) and phosphorylation of PKB was detected with either anti-phosphorylated PKB antibody (p-PKB) or with anti-PKB antibody (PKB). These results were reproduced three times. D, COS7 cells were cotransfected with plasmid containing mouse GnRHR together with plasmids containing either CD8-tagged ␤ARK (␤␥ scav); K44A-dynamin (Dn-Dyn); or V54D-␤-arrestin2 (Dn-Arr); or no insertion as vector control. Two days after transfection, the cells were serum-starved for 16 h and then either treated with GnRH-a (10 Ϫ7 M; 30 min) or left untreated (0). Activation of PKB was determined by Western blot analysis as in C. The results were reproduced twice. E, activation of PKB was determined by densitometry and plotted as a bar graph of percent activation from that of GnRH-a-stimulated COS7 cells that were cotransfected with GnRHR and vector control in each experiment. These are averages and S.E. of three experiments. examined using the appropriate antibodies. We found that the GnRH-a-stimulated phosphorylation of all these signaling components was significantly inhibited by the pertussis toxin (Fig.  9). No effect of the pertussis toxin on the amount of the MAPKs, c-Src, or EGF receptor or the viability of the transfected COS7 cells was observed in the course of this experiment. Therefore, the results suggest that G␣ i is a main intermediate in the GnRHR to MAPKs signaling pathway in COS7 cells, although additional components may be involved in the pathway to JNK. These results are best explained by a model in which ERK activation is mediated mainly by the EGF receptor and Ras, with some involvement of c-Src but not of PI3K, while JNK activation is mediated by G␣ i , EGF receptor, and PI3K (Fig.  10). ␤-arrestin appears to be involved in the activation of both cascades and to operate via an independent pathway. On the other hand, other signaling molecules such as FAK, dynamin, and PKC do not seem to be involved in the GnRH-MAPK signaling in GnRHR-expressing COS7 cells. DISCUSSION The receptor for GnRH serves as a good model in the study of GPCR signaling toward MAPKs (9). Here we studied the signaling processes that lead from GnRHR to ERK and JNK in GnRHR-transfected COS7 cells. The pathways that we have identified in these cells differ substantially from those observed in the pituitary-derived ␣T3-1 cells (32,33,35). Nevertheless, the signaling mechanisms observed here are important for several reasons: (i) the unique mechanism of JNK and ERK activation by GnRH reported here can serve as a model for other GPCRs activation of the MAPK cascades; (ii) a second GnRHR has been recently described (55,56), and therefore it is important to characterize signaling mechanisms specific for type 1 GnRHR; and (iii) aside from its pituitary functions, GnRH exerts also extrapituitary effects upon the gonads and gonadal steroid-dependent tumor cells (42) where the GnRHR may couple to different signaling components. Therefore, the findings here may serve as a model for the extrapituitary functions that are elicited by GnRH.
Several distinct signaling pathways have previously been demonstrated to lead signals from the endogenous GnRHR to the ERK cascades in ␣T3-1 cells. Thus, activation of ERK in these cells involves simultaneous activation of a pathway that involves activation of Raf-1 by PKC (33), together with a supportive pathway that includes dynamin, c-Src, and Ras (33). In addition, Ca 2ϩ from various pools has been implicated in the differential activation of ERK and JNK by GnRH in ␣T3-1 cells (32,37). The three pathways do not seem to contradict, but rather to complement each other to form a signaling network, which is essential to achieve the full GnRH effect on MAPKs. Furthermore, it is tempting to suggest that changes in the frequency of GnRH pulses or in the condition of the cells might recruit selective signaling components, and thus change their relative contribution to the MAPK activation. Notably, G␣ q and PKC are central components in the signaling network initiated by GnRHR in the ␣T3-1 cells (9). However, these components do not seem to play a role in the GnRHR-transfected COS7 cells, where G␣ i and EGF receptor seem to play the major role, with a smaller contribution by ␤-arrestin and to some extent also the dissociated ␤␥ dimer (Fig. 10). Interestingly, under different experimental systems, GnRHR was found to utilize as yet additional distinct intracellular signaling pathways to activate MAPKs. Thus, in GGH3 cells, ERK activity is mediated by PKA as well as by PKC (41); in Caov-3 cells, GnRHR operates via G␣ as well as by independent G␤␥ of the G i /G o proteins (42), in L␤T2 cells GnRH activation of JNK is not dependent on PKC (57), and in immortalized GT1-7 neurons it signals through PKC that transactivate EGF receptor (43,44). Collectively, these results indicate that GnRH can utilize several signaling pathways in different cell types and under different conditions to execute a single intracellular effect.
Additional variability in the mechanism of GnRHR signaling was detected even within a particular cell line. We demonstrate here the involvement of G␣ i , EGF receptor, and to some extent also of c-Src and ␤-arrestin in the activation of ERK. On the other hand, Grosse et al. (36), provided evidence for the involvement of G q and PKC in the same pathway and the same cells. Grosse et al. (36) also showed that GnRH signals to ERK is mediated by EGF receptor in ␣T3-1 cells, while we found no role for EGF receptor in this process in the same cells (33). Moreover, Mulvaney and Roberson (38) showed that JNK activation in ␣T3-1 cells is mediated by elevated intracellular Ca 2ϩ , with no role for PKC, whereas our results (35) show that PKC plays an important role in the GnRH signaling to JNK in the same cells. The reason for the discrepancies between the results is not clear. However, we do believe that all the signaling pathways that were published are correct for the same ␣T3-1 and COS7 cells which are modified under different growing conditions used in the different laboratories. We found that maintenance of ␣T3-1 cells for more than three months in culture, gradually modulates their signaling properties in that they are less dependent on PKC for the activation of ERK (data not shown). We also found that appropriate serum-starvation for a minimal time of 14 h is necessary to obtain reproducible signaling results in both cells, probably because this time is required for a complete removal of MAPK phosphatases (data not shown). Thus, GnRHR may utilize more than one pathway to transmit its signals in different cells and under different conditions. The pathways used are probably dependent on the repertoire of signaling components that are available to the receptor under each growing condition. Despite of the different signaling pathways, the downstream processes that are activated, such as the MAPK cascades or even gene expression, are remarkably similar. This can be explained by the centrality of the MAPK cascades that can receive input from various sources, and direct them to their right destination under varying experimental conditions. We show here the involvement of both G␣ i and EGF receptor in the transduction of GnRH signals. These two components seem to be important in the signaling of many GPCRs (9). The mechanism by which G␣ i induces the activation of the EGF receptor seems to vary in different systems. This activation was shown to be mediated by c-Src, by the GPCR itself, by the dissociated G␤␥, by the release of Hb-EGF, and more (45). The system examined here seems to be unique in that the EGF receptor is activated in a mechanism that does not involve any of the above mechanisms and is probably mediated directly by the dissociated G␣ i upon the engagement of the GnRHR with its ligand.
c-Src is a central component in most GPCR signaling (58) operating either as the main signal transducer to MAPK cascades or in cooperation with additional signaling components (2,9,59,60). Several mechanisms for c-Src activation by GPCRs have been identified. It has been reported that c-Src may be activated by a PKC-dependent mechanism (61), via a dissociated ␤␥ dimer (62), or by a direct interaction with the ␣ subunit of G proteins (63). Interestingly, activation of c-Src may be mediated also by its recruitment of this protein kinase to the GPCRs themselves, and this can occur by a direct interaction of the c-Src with proline-rich motifs in the GPCR (64), or by an interaction with the scaffold protein ␤-arrestin, which often interacts with GPCRs (65,66). These interactions usually occur specifically through the SH3 domain of c-Src, and often lead to the c-Src activation.
c-Src is involved in the activation of MAPKs by GnRH in several cellular systems. We have previously shown that in ␣T3-1 cells, c-Src is activated via a mechanism that is partially (ϳ70%) dependent on PKC (35), but may involve also dynamin (33). However, this is clearly not the situation here, since the selective PKC inhibitor GF109203X had no influence on the activation of c-Src, ERK or JNK upon GnRH stimulation. Instead, the activation of c-Src observed here is mainly dependent on stimulation by the EGF receptor (Fig. 7). In addition, the dissociated G␤␥ slightly contributes to c-Src activation similar to other systems (2), but the receptor itself, ␤-arrestin or dynamin does not influence this process. To our knowledge this is the first demonstration of a direct c-Src activation by RTKs upon stimulation by GPCRs. Hence, c-Src activation by GPCRs can occur not only before but also through transactivation of RTKs in different cellular systems.
PI3K is another key player in GPCR signaling (67), and it has recently been implicated in the control of cell growth, survival, and malignant transformation (68). Several mechanisms of PI3K activation by GPCRs have been elucidated over the past few years (9). These include activation by the dissociated ␤␥ dimer (69,70), which is probably mediated by a direct interaction of the ␤␥ dimer with two domains of the catalytic p110 subunit (71). G␣ i and G␣ q have been implicated in the activation of PI3K in transfected COS7 cells (69), and activation of FAK has been implicated in PI3K signaling downstream of gastrin receptor (72). In addition, PI3K was shown to act downstream of PTKs as it was shown that tyrosine phosphorylation of IRS-1 by Src family kinases leads to the recruitment and the activation of PI3K in response to gastrin (73). The EGF receptor was also implicated in a similar process (18), and in this case the activation of PI3K can occur either directly by the receptor or via the small GTPase Ras (74). In our hands, none of these processes seem to play an important role in the activation of PI3K by GnRH. Since the activation of PKB, which serves as a readout for PI3K was markedly inhibited by Csk and the dominant negative EGF receptor, we conclude that c-Src is responsible for the activation of PI3K in our system. Another signaling component that was shown to contribute to the activation of JNK and ERK by GnRH in the current system is ␤-arrestin. However, the activation of c-Src and PI3K was not affected by the dominant negative ␤-arrestin indicating that ␤-arrestin may utilize an unrelated signaling pathway to activate the MAPKs. In this context it is important to mention that a direct association of ␤-arrestin with ERK and JNK was reported to occur in some systems (25,26), and this might play a role also in the ␤-arrestin-mediated signals shown here. On the other hand, despite their significant expression in COS7 cells, several signaling components that have previously been reported to participate in GPCR signaling are not involved in the GnRHR-MAPK pathway here. Thus, G␣ q , PKC, dynamin, and FAK do not seem to play a role in the examined pathways.
In summary, we report here that in GnRHR-expressing COS7 cells, G␣ i and EGF receptor play a central role in the transmission of GnRH signals to JNK and ERK in an Hb-EGFindependent manner. The main pathway that leads to JNK activation downstream of the EGF receptor involves activation of c-Src and PI3K, whereas ERK activation is mediated by EGF receptor, which transmits its signals to the Ras/ERK pathway either directly or via the activation of c-Src. A minor contribution to this machinery was detected also for the dissociated ␤␥ dimer that may participate in the activation of c-Src, and ␤-arrestin that seems to operate via an independent mechanism. The pathways detected here are different from those in other cell lines, indicating that in different cellular context, GnRHR can utilize distinct signaling machinery to achieve similar activation of the MAPK cascades. The results might be applied to analysis of signaling mechanisms involved in MAPK activation by GnRH in extrapituitary tissues such as gonadal steroid-dependent tumors.