Stimulation of the ERK Pathway by GTP-loaded Rap1 Requires the Concomitant Activation of Ras, Protein Kinase C, and Protein Kinase A in Neuronal Cells*

The small GTPases Ras or Rap1 were suggested to mediate the stimulatory effect of some G protein-coupled receptors on ERK activity in neuronal cells. Accordingly, we reported here that pituitary adenylate cyclase-activating polypeptide (PACAP), whose G protein-coupled receptor triggers neuronal differentiation of the PC12 cell line via ERK1/2 activation, transiently activated Ras and induced the sustained GTP loading of Rap1. Ras mediated peak stimulation of ERK by PACAP, whereas Rap1 was necessary for the sustained activation phase. However, PACAP-induced GTP-loading of Rap1 was not sufficient to account for ERK activation by PACAP because 1) PACAP-elicited Rap1 GTP-loading depended only on phospholipase C, whereas maximal stimulation of ERK by PACAP also required the activity of protein kinase A (PKA), protein kinase C (PKC), and calcium-dependent signaling; and 2) constitutively active mutants of Rap1, Rap1A-V12, and Rap1B-V12 only minimally stimulated the ERK pathway compared with Ras-V12. The effect of Rap1A-V12 was dramatically potentiated by the concurrent activation of PKC, the cAMP pathway, and Ras, and this potentiation was blocked by dominant-negative mutants of Ras and Raf. Thus, this set of data indicated that GPCR-elicited GTP loading of Rap1 was not sufficient to stimulate efficiently ERK in PC12 cells and required the permissive co-stimulation of PKA, PKC, or Ras.

phenotype characterized by neurite outgrowth (1). The effect of NGF is dependent on a long lasting activation of the Ras-Raf-MEK-ERK pathway (3). Activation of the cAMP pathway was also reported to induce PC12 differentiation (4,5). In line with the demonstrated role of ERK activation in NGF-induced PC12 differentiation, cAMP analogues, and forskolin, a direct activator of adenylate cyclase (AC) was shown to stimulate ERK activity in this cell line (5,6). Similarly, mobilization of calcium and/or stimulation of the diacylglycerol (DAG) production results in ERK activation and eventually neurite outgrowth (7,8).
The mechanisms responsible for the control of the ERK pathway by receptor tyrosine kinases, cAMP analogues, and calcium/DAG were intensively investigated, and Ras-like small GTP-binding proteins emerged as key elements in this pathway. The products of the H-Ras gene was shown to stimulate the activity of the MEK kinase Raf-1 following activation of receptor tyrosine kinases (9). Cyclic AMP and calcium were also suggested to control the activity of Ras in some cell types (7,10), resulting in ERK activation. More recently, the Ras superfamily member Rap1 and the protein kinase B-Raf were suggested to link PKA activation by cAMP analogues to MEK1 stimulation in neuronal cells (11). On the other hand, several mechanisms have been proposed for calcium-induced ERK activation, including activation of the calcium/calmodulindependent kinase (12)(13)(14), the Pyk2 tyrosine kinase (15,16), and the cAMP/Rap1/B-Raf pathway (17).
Based on the above-mentioned data, one would predict that activation of GPCRs positively coupled to AC or PLC-␤ would lead to PC12 differentiation. However, in contrast to the effect of cAMP analogues or forskolin, stimulation of the endogenous A 2A adenosine receptor (18) or of the transfected ␤ 1 adrenergic receptor (19), two GPCRs positively coupled to AC, does not result in neurite outgrowth. Similarly, the situation is more complex than anticipated for PLC-coupled GPCRs as exemplified by the differential effects of the three closely related, PLC-coupled, ␣ 1 adrenergic receptor subtypes (8). These contradictory data led us to study in detail the mechanisms involved in the stimulation of the ERK pathway by the GPCR for PACAP. This neuropeptide belongs to the vasoactive intestinal peptide-secretin-glucagon family of peptides and was originally isolated through its ability to stimulate the adenylate cyclase activity of pituitary cells in vitro (20). PACAP neurotrophic activity was first reported for PC12 cells in which the 38-amino acid form of PACAP was shown to promote neurite outgrowth (21). Indeed, the receptor for PACAP is the only endogenous GPCR to stimulate neurite outgrowth in PC12 cells. This neurotrophic activity was extended to the protection of cerebellar granule (22,23) and dorsal root ganglion neurons (24) from apoptosis, and of cortical (25) and hippocampal (26) neurons from ischemia-induced cell death. PACAP was also shown to control the proliferation and differentiation of cortical (27)(28)(29) and cerebellar neuroblasts (30). PACAP neurotrophic activity is mediated by the activation of Pac1, the PACAP-specific GPCR. Pac1 displays a complex pattern of alternative splicing that modulates its ligand binding and signaling properties (31)(32)(33)(34) resulting in the modulation of PACAP physiological effects (28,35). We and others (21,31) showed that Pac1 potently stimulated the AC and PLC activities as well as the ERK pathway that was implicated in its neurotrophic activity (23,36). In this context, we selected the PC12 cell line to assess the involvement of the small GTP-binding proteins Rap1 and Ras in the signaling pathways responsible for PACAP-induced ERK activation in this system.

EXPERIMENTAL PROCEDURES
Materials-PACAP and mouse 2.5 S NGF were purchased from Neosystem (Strasbourg, France) and Promega, respectively. Forskolin and phorbol 12-myristate 13-acetate (PMA) were purchased from Sigma. The PKA inhibitors H89 and (R p )-cAMP, the PKC inhibitor bisindolylmaleimide-1, the calcium chelator BAPTA-AM, and the calmodulin inhibitor W13 were purchased from Alexis Corp. (San Diego, CA). The PLC inhibitor U-73122 and its inactive analogue, U-73343, were from Calbiochem. Monoclonal antibodies to Rap1 and Ras were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), and monoclonal antibody to HA (12CA5) was from Roche Molecular Biochemicals. Phosphorylationspecific and total ERK antibodies were purchased from New England Biolabs (Beverly, MA).
Cell Culture-Wild-type and PKA-deficient PC12 cells (A126-1B2) were kindly provided by Prof. J. A Wagner (Harvard Medical School, Boston) and were grown on tissue culture dishes (Falcon) coated with poly-L-lysine in Opti-MEM/Glutamax medium supplemented with 5% fetal calf serum and 5% horse serum (Invitrogen). Culture medium was further supplemented with 10 units/ml penicillin and 10 g/ml streptomycin. Cells were maintained at 37°C in a saturating humidified atmosphere of 95% air and 5% CO 2 .
Endogenous ERK1/2 Activities-PC12 cells were plated in 6-well plates in complete medium until they reached 80% confluence. Cells were then washed and cultured for an additional 16 -18 h in RPMI 1640/Glutamax medium supplemented with 1% fetal calf serum (low serum medium) before treatment with PACAP (100 nM), forskolin (20 or 50 M), PMA (0.3 or 1 nM), or NGF (40 ng/ml) for the indicated time. Before stimulation, the cultures were exposed to different inhibitors as mentioned in the figure legends. Cells were then rapidly rinsed in ice-cold PBS and lysed in Laemmli's sample buffer. Cell lysates were boiled for 5 min, and proteins were resolved by SDS-PAGE, blotted onto nitrocellulose membrane, and probed with phospho-specific and phosphorylation state-independent antibodies following the manufacturer's instructions (New England Biolabs).
Endogenous Rap1 and Ras GTP Loading-Rap1 and Ras GTP loading were measured with a non-radioactive method as described previously (37,38). Briefly, after stimulation, cells were rinsed rapidly in ice-cold phosphate-buffered saline at pH 7.4 and solubilized at 4°C for 10 min in 0.3 ml of lysis buffer (10% glycerol, 1% Nonidet P-40, 50 mM Tris, pH 7.4, 200 mM NaCl, 2.5 mM MgCl 2 , 250 M phenylmethylsulfonyl fluoride, 1 M leupeptin, 0.1 M aprotinin, 10 mM NaF, and 1 mM Na 3 VO 4 ). Lysates were clarified by centrifugation at 10,000 ϫ g for 10 min, and supernatants were incubated with glutathione-Sepharose beads (Amersham Biosciences) freshly coupled to 10 g of GST-Ral-GDS-RBD to isolate Rap1GTP or GST-Raf1-RBD to isolate Ras-GTP. Protein complexes were allowed to form for 1 h at 4°C. Precipitates were washed 3 times with lysis buffer. Finally, precipitates were resuspended in Laemmli's sample buffer, and denatured proteins were loaded on a 12% SDS-PAGE. The proteins were transferred onto polyvinylidene difluoride Immobilon-P transfer membrane filters (Millipore, Bedford, MA). Immunodetection was performed using an anti-Rap1 or anti-Ras antibody and an anti-mouse IgG coupled to horseradish peroxidase as a secondary antibody. Blots were developed with an enhanced chemiluminescence Western blot detection system (Pierce).
Transfections-Cells were transfected with LipofectAMINE 2000 according to the manufacturer's instructions (Invitrogen) and lysed 28 h after transfection.
Gal4-Elk1-dependent Luciferase Activity-PC12 cells were grown to 70% confluence in a 24-well plate prior to transfection as described above. In all experiments, 2 g of (Gal4) 5 -E1B-TATA-luciferase reporter plasmid and 2 g of a plasmid encoding the Gal4-(Elk1 transactivation domain) fusion protein (kindly provided by R. Hipskind, Institut de Genetique Moleculaire; Montpellier, France) were co-transfected with various amounts of a test plasmid as indicated in the figure legends. Rap1A-V12, Rap1A-N17, and Ras-V12 cDNAs were provided by J. de Gunzburg (INSERM U248, Institut Curie, Paris, France). Ras-N17 and Rap1B-V12 were obtained from F. Zwartkruis (Utrecht University, The Netherlands) and P. J. Stork (The Vollum Institute, Portland), respectively. The N-Raf1 dominant-negative mutant of Raf-1, lacking the C-terminal kinase domain, was described previously (39). The total amount of transfected DNA was kept constant with addition of pRK5-CAT. Eight hours after transfection, cells were placed for 18 h in low serum medium before stimulation with PACAP, forskolin, PMA, or NGF. Luciferase activity was assayed 5 h later as described previously (31).
Infection with Adenoviruses-Adenoviruses expressing Gap1m (Ras-GAP), Rap1-GAP, or LacZ (as a negative control) were kindly provided by Dr. S. Hattori (National Institute of Neurosciences, Tokyo) (40). Cells were infected in 6-well plates at a multiplicity of infection of 100 in complete medium. Forty eight hours post-infection, endogenous ERK activities were determined as indicated above.
Neurite Outgrowth and Immunostaining-PC12 cells were transfected with HA-Ras-V12, HA-Rap1A-V12, or EGFP-N1 as a control. Twenty eight hours after transfection, cells in low serum medium for 18 h were rinsed with PBS and fixed in 4% paraformaldehyde for 20 min at room temperature. Cells were then rinsed three times with PBS/ glycine 0.1 M, permeabilized with PBS/Triton X-100 0.1% for 10 min at room temperature, and incubated for 30 min in PBS/bovine serum albumin 1%. Fixed cells were stained with a monoclonal antibody to HA (12CA5, Roche Molecular Biochemicals) for 1 h at room temperature, rinsed in PBS, incubated with Alexa fluor 488-conjugated goat antimouse antiserum (1:1000) (Molecular Probes), and examined with a Zeiss Axiovert microscope.

RESULTS
Sustained Activation of ERK1/2 by PACAP-Because quantitative and qualitative differences were reported among PC12 cells used in different laboratories, we verified ERK regulation by PACAP in the PC12 cells used throughout this study. As expected, PACAP-38 induced a biphasic phosphorylation of both ERK1 and -2 at Thr-202 and Tyr-204 (Fig. 1A). ERK phosphorylation was maximally stimulated at 5 min, and then gradually diminished but remained above resting level for several hours. This kinetic profile was qualitatively similar to the one obtained with NGF (Fig. 1A). The effect of NGF was however more pronounced and lasted longer, in agreement with the reported more robust effect of NGF on neurite outgrowth.
PACAP Stimulates Endogenous Rap1 and Ras-We then evaluated the enhancement of Rap1 and Ras GTP loading induced by PACAP-38 which activates AC and PLC in PC12 cells (21). Both monomeric G proteins were rapidly activated upon PACAP-38 addition (Fig. 1, B and C). Rap1 activation was sustained, comparable in amplitude to the one obtained with forskolin and superior to the one induced by NGF (Fig. 1B). On the other hand, PACAP-induced Ras stimulation was transient and weaker than that obtained with NGF (Fig. 1C).
Rap1A Is Critical for Long Term PACAP-induced Stimulation of ERK-To assess the role of Rap1 and Ras in PACAPinduced ERK activation, we co-transfected dominant-negative mutants of both G proteins, Ras-N17 and Rap1A-N17, respectively, with a plasmid encoding a Gal4-Elk1 fusion protein, and we measured the activity of a luciferase reporter gene under the control of a minimal promoter incorporating five Gal4responsive elements. In this system, ERK activation induces Gal4-Elk1 phosphorylation and the subsequent transcriptional regulation of the reporter gene, providing a read-out for long term stimulation of the ERK pathway. The induction of the luciferase activity by PACAP was completely abolished upon preincubation with the MEK-specific antagonist U0126 (10 M, 30 min), attesting for the specificity of the reporter system ( Fig.  2A). PACAP-induced Elk1-transactivating activity was blocked more efficiently by Rap1A-N17 than by Ras-N17 (Fig. 2B), whereas it was the reverse for NGF (Fig. 2C). Together with the data presented in Fig. 1 which indicate that Rap1 stimulation is sustained and ample whereas Ras stimulation is more transient and weaker, these results indicated that both Ras and Rap1 were involved in PACAP-elicited ERK activation and that Rap1 played a prominent role in long term ERK activation.
To confirm the relative role of Ras and Rap1 in PACAPinduced ERK stimulation, we used adenoviruses encoding LacZ, Gap1m (Ras-Gap), or Rap1GAP (40) (Fig. 3). As expected from the data obtained using the Elk1 reporter system, deactivation of Ras or Rap1 by the GTPase-activating activity of the GAPs led to the attenuation of PACAP-elicited endogenous ERK1/2 phosphorylation. Interestingly, Ras-GAP was more efficient than Rap-GAP at 5 min following PACAP addition, whereas it was the reverse at 15 min. These observations suggested that Ras was prominently involved in the peak stimulation of ERK phosphorylation by PACAP, whereas the sustained stimulation of the ERK pathway required Rap1 activation.
PACAP-elicited Rap1 Stimulation Is Dependent on PLC but Independent of Calcium, PKC, and PKA-To determine FIG. 1. PACAP-induced ERK phosphorylation and GTP loading of endogenous Rap1 and Ras. A, PACAP-induced ERK-1 and -2 phosphorylation. PC12 cells were treated with 100 nM PACAP or 40 ng/ml NGF for the indicated times. Cells were harvested and lysed, and samples were analyzed by Western blotting using a phosphospecific ERK1/2 antibody (P-ERK1/2) (upper panels). Equal loading was verified by probing the blots with an anti-ERK1/2 antibody (ERK1/2) (lower panels). The data are representative of three independent experiments. B, PACAP induced Rap1 GTP loading. Following addition of PACAP as in A, cells were harvested and lysed. GTP-loaded Rap1 was affinityprecipitated using GST-RalGDS (Rap-binding domain) and immunoblotted with an anti-Rap1 antibody (upper panels). The total amount of Rap1 protein is shown in the lower panels. To evaluate the amplitude of Rap1 stimulation by PACAP, the same experiment was performed with 50 M forskolin (Forsk.) or 40 ng/ml NGF. C, PACAP-stimulated Ras GTP loading. GTP-loaded Ras was affinity-precipitated using GST-Raf1 (Ras-binding domain) and immunodetected with a Ras antibody. As a positive control, cells were stimulated with NGF (40 ng/ml) for 5 min. whether Rap1 GTP loading elicited by PACAP was indeed sufficient to account for PACAP-induced ERK stimulation, we compared the signaling pathways involved in Pac1-elicited Rap1 and ERK stimulations. We first tested the effects of pharmacological inhibitors of different signaling pathways on Rap1 activation. The only effective molecule was U73122 (Fig.  4A), a specific blocker of PLC. Significantly, the inactive analogue U73343 had no effect (Fig. 4A). PKC, calcium, and calmodulin were also found dispensable because bisindolylmaleimide-1, EGTA, BAPTA-AM, and W13 were ineffective (Fig.  4A). PKA was also not involved because H89 and (R p )-cAMP did not alter Rap1 activation by PACAP (Fig. 4A). This finding was further confirmed using the A126-1B2 variant of PC12 cells, which displays only 10% of the PKAII activity (41), and in which PACAP stimulated Rap1 GTP loading (Fig. 4B). A role for a cAMP-GEF could also be excluded because the stimulation of Rap1 by forskolin was abolished in A126-1B2 cells (Fig. 4B).
PLC, PKA, PKC, Calcium, and CaM Are Necessary for ERK Activation by PACAP-We then tested whether PLC-elicited Rap1 activation is indeed sufficient for PACAP-induced ERK activation. As expected U73122 blocked ERK phosphorylation induced by PACAP (Fig. 4C) confirming the involvement of the PLC pathway. Surprisingly, H89 efficiently attenuated PACAP-induced ERK phosphorylation (Fig. 4C). The involvement of PKA was further confirmed using A126-1B2 cells in which PACAP-elicited ERK phosphorylation was notably attenuated (Fig. 4D). Interestingly, in wild-type PC12 cells, forskolin, a direct activator of AC, was able to induce ERK phosphorylation less efficiently than PACAP (Fig. 4D), whereas it was the reverse for the stimulation of the cAMP production (data not shown). As expected the effect of forskolin was completely abolished in A126-1B2 cells, whereas the effect of NGF was not affected (Fig. 4D). Altogether, these data indicate that PKA is necessary but not sufficient for PACAP-induced ERK activation.
Bisindolylmaleimide-1, the most specific PKC blocker (42), strongly diminished PACAP-induced ERK phosphorylation (Fig. 4C), showing that the activation of PKC is necessary for PACAP to stimulate efficiently ERK. Similarly, a rise in intra-cellular calcium is necessary as addition of EGTA to the incubation medium or of the cell-permeant BAPTA-AM attenuated PACAP-induced ERK phosphorylation (Fig. 4C). Moreover, the effect of calcium was at least partially mediated by CaM because addition of W13, a specific calmodulin blocker, strongly diminished PACAP-induced ERK1/2 phosphorylation (Fig. 4C).
Activation of Rap1 Is Not Sufficient for Efficient Stimulation of ERK Activity-Because there was an apparent discrepancy between the pathway involved in Rap1 GTP loading and those necessary for potent ERK stimulation by PACAP, we tested whether activated Rap1 is sufficient to stimulate ERK efficiently. . Forty eight hours later, the cells were incubated with PACAP (100 nM) for the indicated times. Cells were harvested and lysed, and samples were analyzed by Western blotting using a phosphospecific ERK1/2 antibody (P-ERK1/2) (upper panels). The same Western blot was exposed for different times to provide optimal sensitivity for each time point. Equal loading was verified by probing the blot with an anti-ERK1/2 antibody (ERK1/2) (lower panels). A constitutively active mutant of Ras, Ras-V12, was exquisitely potent in inducing Gal4-Elk1 phosphorylation-dependent luciferase activity (Fig. 5A; note the bi-logarithmic scale). Surprisingly, the corresponding constitutively active mutant of Rap1A, Rap1A-V12, only minimally stimulated Elk1 phosphorylation-induced luciferase activity (Fig. 5A). The observed weak activation of the ERK pathway was not specific for the Rap1A isoform because a constitutive Rap1B mutant was also ineffective (Fig. 5A). Because the effect of Rap1-GTP depends on the presence of B-Raf (11,43), we monitored the PC12 cells used in the present study for the expression of B-Raf by Western blotting and evidenced the expected 95-and 68-kDa bands (data not shown).
To exclude that the difference in the potency of Ras-V12 and Rap1A/B-V12 was because of a difference in the protein levels, we performed similar experiments with HA-tagged constructs and monitored the expression levels of the mutants by Western blotting. As shown in Fig. 5B, transfection of various amount of the different constructs resulted in similar levels of Ras-V12, Rap1A-V12, and Rap1B-V12. In these conditions, Ras-V12 was considerably more effective than the corresponding Rap1A/B mutants ( Fig. 5B; note the logarithmic scale).
To confirm further the results obtained with the Elk reporter system, we examined ERK activation following transient transfection of constitutively activated mutants of Rap1A, Rap1B, and Ras. As shown in Fig. 5C, HA-Ras-V12 efficiently stimulated ERK phosphorylation, whereas similar levels of HA-Rap1A-V12 and HA-Rap1B-V12 were inefficient.
These data indicated that GTP loading of Rap1 was not sufficient to induce effectively ERK activation. These results were further confirmed at a more physiological level. We monitored neurite outgrowth in transfected PC12 cells and could demonstrate that Ras-V12 efficiently induced neurite outgrowth, in line with its potency in stimulating the ERK pathway. In contrast, Rap1A-V12 was completely inefficient (Fig. 5D).
Activation of Ras, the cAMP Pathway, and PKC Synergize with Constitutively Active Rap1A in ERK Activation-We reasoned that because multiple signaling pathways are involved in ERK activation by PACAP, these signals may be required to potentiate the effect of GTP-loaded Rap1A. In line with this were transfected as in A with the indicated amount of each plasmid. Luciferase activity was determined as in A. Expression levels of constitutively active mutants were verified by Western blotting using an anti-HA monoclonal antibody. C, PC12 cells were transfected with the indicated amounts of plasmids encoding HA-tagged Rap1A-V12, Rap1B-V12, or Ras-V12. Equal amounts of cell lysate were assayed for Western blotting using a phosphospecific ERK1/2 antibody (P-ERK1/2; upper panel) and an ERK1/2 antibody (middle panel). The amount of each transfected protein was monitored by Western blotting using an anti-HA monoclonal antibody. As expected from experiments performed in B, Ras-V12 induced ERK phosphorylation, whereas constitutively activated mutants of Rap1 proteins were ineffective. D, PC12 cells were transfected with plasmids encoding green fluorescent protein (GFP) as a negative control, HA-Ras-V12 and HA-Rap1A-V12. Twenty eight hours later, transfected cells were visualized with a fluorescent microscope, directly (green fluorescent protein) or following immunohistochemistry with an anti-HA monoclonal antibody. Ras-V12 (1 g)-transfected cells displayed long neurites, whereas the morphology of Rap1A-V12 (5 g)-transfected cells did not differ from control, green fluorescent protein (5 g)-transfected, cells. hypothesis, we found that Elk1-transactivating activity is synergistically activated by Rap1A-V12 and minute amounts of Ras-V12 (Fig. 6A). Although the synergy of the constitutively active Rap1A and Ras mutants was apparently weak (Fig. 6A), one should take into account that it resulted from a limited number of transfected cells, due to the small amount of Ras-V12 used in this experiment.
Similarly, efficient Elk1 phosphorylation-dependent luciferase activation was obtained upon incubation of Rap1A-V12transfected cells with doses of forskolin and PMA, which by themselves did not produce a significant effect (Fig. 6B).

Dominant-negative Mutants of Ras and Raf Prevent the Potentiation of Rap1-V12-elicited ERK Stimulation by the cAMP
Pathway and PKC-To elucidate further the relative roles of Ras, PKC, and the cAMP pathway, we used dominant-negative mutants of Ras and its downstream kinase, Raf-1. Ras-N17 moderately attenuated the potentiation by PMA and forskolin of Rap1A-V12-elicited stimulation of Gal4-Elk1-dependent luciferase activity (Fig. 7). On the other hand, the N-terminal fragment of Raf-1 (10), N-Raf-1, strongly attenuated the effect of Rap1A-V12, PMA, and forskolin. This fragment of Raf-1 includes the Ras binding domain of Raf-1 and hence behaves as a blocker of Ras signaling. Consistently, similar results were obtained with the N-terminal fragment of B-Raf (data not shown). The more pronounced inhibition of the potentiation by PMA and forskolin of Rap1A-V12-elicited stimulation of Elk1dependent luciferase activity induced by N-Raf-1 compared with Ras-N17 suggested a Ras-independent activation of Raf by PMA and/or forskolin. In this context, the phosphorylation of Raf-1 by PMA-activated PKC was a plausible candidate pathway (44). Altogether, these data indicated that PMA and forskolin potentiated the stimulatory effect of Rap1A-V12 through both Ras-dependent (blocked by Ras-N17 and N-Raf-1) and -independent (blocked by N-Raf-1 only) pathways. DISCUSSION Collectively, our data indicate that PACAP activated the ERK pathway through the pleiotropic stimulation of several signaling pathways acting synergistically. We showed that PACAP enhanced GTP loading of two small GTP-binding proteins, Rap1 and Ras (Fig. 1). The stimulation of Rap1 GTP loading was more pronounced and lasted longer than that of Ras. Experiments using adenoviruses encoding Ras-or Rap1-GAP indicated that Ras is primarily involved in the initial peak stimulation of ERK, whereas Rap1 is involved in sustained ERK activation (Fig. 3). This is reminiscent of reports by York and co-workers (45) and Garcia and co-workers (39), suggesting that NGF-induced and thrombopoietin-mediated sustained activation of ERK in PC12 and UT7-Mpl cells, respectively, also involved the successive stimulation of the ERK pathway by Ras and Rap1. Hence, consecutive stimulation of Ras and Rap1 may represent a common theme in sustained, biphasic activation of the ERK pathway, whatever the nature of the triggering receptor.
Although Rap1 activation was shown to be essential for PACAP-induced ERK stimulation (Figs. 2 and 3), we found that it is not sufficient. First, although PACAP-induced enhancement of Rap1 GTP-loading is blocked exclusively by a PLC blocker (Fig. 4A), PACAP-induced ERK activation is blocked or strongly attenuated by inhibitors of other signaling pathways (Fig. 4C). Second, whereas Ras-V12, a constitutively active mutant of Ras, induced a robust stimulation of an Elk1 reporter plasmid, the same dose of Rap1A-V12 minimally stimulated this construct (Fig. 5, A and B). These two lines of evidence indicated that stimulation of Rap1 GTP-loading was not sufficient to induce a potent ERK activation in PC12 cells. The lack of effect of Rap1A-V12 was unexpected given the reported effect of Rap1B-GTP␥S (43) on B-Raf activity in vitro and of Rap1B-V12 on an Elk1 reporter system (11). To test for a difference in the activity of Rap1A and Rap1B, we transfected the Rap1B-V12 cDNA used by Vossler and co-workers (11) and evidenced a stimulation of the Elk1 reporter system similar to the one obtained with the Rap1A-V12 cDNA (Fig. 5A). We concluded that, in the PC12 cells used in the present study, an additional signaling pathway should be activated in order to potentiate the ERK stimulation by activated, GTP-loaded, Rap1. Constitutive activation of this pathway at a permissive level in the cell clone used by Vossler and co-workers (11) is a possible explanation for the observed differences.
Because PACAP-induced Rap1 GTP-loading was dependent on PLC activity only, whereas ERK activation was dependent on multiple signaling pathways, we sought to determine more precisely the role of Ras, AC, and PKC in ERK activation by PACAP. One alternative was that Ras, AC, or PKC activation worked in parallel to PLC-dependent Rap1 activation, and the other was that these signaling pathways synergized with Rap1 activation. Data presented in Fig. 6 support the second alternative, because Ras, cAMP, and PKC strongly enhanced Rap1A-V12-induced Gal4-Elk1-dependent luciferase activity. This synergistic effect may take place at different levels. First, Rap1 may require, in addition to GTP loading, a post-translational modification to potently stimulate the B-Raf-MEK-ERK pathway. One obvious candidate is phosphorylation by PKA because this kinase was reported to phosphorylate and activate Rap1 (46,47). Phosphorylation by PKA may thus facilitate Rap1 GTP loading and/or provide an additional signal required for a productive interaction with B-Raf. Rap1 phosphorylation by PKA may also help to alleviate the inhibitory effect of Rap1 on Raf-1 (48) and hence favor Raf-1 interaction with Ras (see below). Second, the activation of Ras, the cAMP pathway, or PKC may be required to sensitize the B-Raf-MEK-ERK pathway to GTP-loaded Rap1. Qiu and co-workers (49) demonstrated that the Rap1-B-Raf-MEK-ERK pathway is modulated by the amount of 14-3-3 scaffolding protein associated with B-Raf. In cells where little 14-3-3 protein is associated with B-Raf, cAMP inhibits ERK activity, whereas it stimulates ERK in cells where 5-fold more 14-3-3 is found associated with B-Raf. One may therefore suggest that the cAMP pathway or PKC may regulate the association of 14-3-3 with B-Raf and hence modulate the potency of GTP-loaded Rap1 to stimulate B-Raf activity.
Data presented in Fig. 7 suggested a direct role for PKC and/or the cAMP pathway in Ras activation, because the potentiation of Rap1A-V12-activated Gal4-Elk1-dependent luciferase activity by PMA and forskolin was partially blocked by a dominant-negative mutant of Ras. Furthermore, PMA and/or forskolin probably also have a Ras-independent effect on Raf, because an N-terminal fragment of Raf-1 was more efficient than Ras-N17 in blocking the potentiating effect of PMA and forskolin. Besides the well documented activation of Raf-1 by GTP-loaded Ras, some PKC isoforms were also reported to phosphorylate and activate Raf-1 (44), suggesting a possible mechanism for Ras-independent activation of Raf-1 by PKC.
The requirement of the Ras-Raf-1 module for the efficient stimulation of the ERK pathway by GTP-loaded Rap1 is surprising, considering the reported inhibitory effect of activated Rap1 on Raf-1 activity (11). One may argue that differentiation-permissive activation of the ERK pathway by extracellular stimuli was consistently found to be biphasic in PC12 cells (3) and mediated by the successive activation of Ras and Rap. However, we showed that Rap1 GTP loading is an early event that is concomitant of Ras activation (Fig. 1). Hence, a mechanism certainly exists that delays ERK stimulation by activated Rap1. In this view, the prerequisite for Raf-1 activation for efficient ERK stimulation by GTP-loaded Rap1 would warrant that the Rap1-B-Raf module is activated after the Ras-Raf-1 module. Furthermore, Ras is found at the plasma membrane, whereas Rap1 is found co-localized with a perinuclear compartment (50). A recent study (51) suggested that the re-cruitment of Raf-1 to the plasma membrane by activated Ras leads to its phosphorylation on tyrosine 341 and serine 338. Interestingly, phosphorylated Raf-1 is activated by Rap1, in contrast to dephosphorylated Raf-1 which is inhibited by Rap1 (51). This observation, together with the present data, suggest a mechanism in which Ras activation leads to the recruitment of Raf-1 to the plasma membrane and to its subsequent phosphorylation. Phosphorylated Raf-1 would then interact with intracellular, GTP-loaded Rap1 and allow subsequent ERK activation by activated Rap1. This mechanism would result in a compartmentalized activation of the ERK pathway at the plasma membrane before activation at intracellular membrane domains. Hence, the requirement for Raf-1 activation for GTPloaded Rap1 to stimulate efficiently the ERK pathway may be involved in the spatio-temporal control of ERK activation during differentiation, which is reminiscent of the observed pattern of Ras and Rap1 activation in NGF-stimulated PC12 cells (50).
We also presented data on the signaling pathways involved in PACAP-elicited Rap1 GTP loading. Because Pac1 is coupled to AC and PLC, the signaling pathways downstream of these effectors are candidates for the control of Rap1 GTP loading. Although forskolin stimulates Rap1 GTP loading, in a strictly PKA-dependent manner (Fig. 4B), PKA activity is dispensable for PACAP-induced Rap1 activation (Fig. 4, A and B). This observation is somehow unexpected considering the tight and efficient coupling of Pac1 to AC and the reported effect of forskolin or constitutively activated PKA on Rap1 activation in PC12 cells (11). One possible interpretation is that the physiological level of cAMP elicited by Pac1 activation is not sufficient for a direct activation of Rap1 in PC12 cells. This is in line with the reports by Huang an co-workers (18) and Williams and co-workers (19) indicating that the stimulation of cAMP production at physiological levels by GPCRs is not sufficient to trigger neurite outgrowth in PC12 cells. Likewise, PKC or calmodulin are not directly involved in Rap1 activation because the specific blockers, bisindolylmaleimide-1 and W13, respectively, were ineffective (Fig. 4A). Because Pac1 is coupled to both AC and PLC, cAMP-GEFs and CalDAG-GEFs are ideal candidates to mediate the PACAP-induced Rap1 activation. Our data exclude a role for the cAMP-GEFs. Rap1 stimulation by forskolin is completely abolished in PKA-deficient cells (Fig.  4B) indicating that elevation of cAMP level is not sufficient per se to activate Rap1 independently of PKA in this cell line. Only U73122, a specific PLC blocker, was effective in preventing PACAP-induced Rap1 GTP loading, indicating that the PLC activity is indispensable for Rap1 stimulation by PACAP (Fig.  4A). To explain these observations, one should postulate the existence of either a DAG-GEF, which would not require calcium to be activated, an IP 3 -GEF, or a PLC with a Rap1-GEF activity. No DAG-GEF was characterized so far. However, Johan de Rooij 2 (52) indicates that CalDAG-GEFIII translocates to membranes upon addition of 12,13-tetradecanoyl phorbol acetate and is insensitive to increased levels of calcium. Similarly, no IP 3 -GEF has been described so far. However, an IP 4 -GAP (GAP1 IP4BP ) protein was isolated (53), indicating that proteins controlling Rap1 activity in response to inositol phosphates binding do exist. The existence of PLC⑀, a PLC with a Rap1-GEF activity, was recently reported (54 -57). PLC⑀ is the founding member of a novel family of polyphosphoinositidespecific phospholipases that integrate multiple signaling pathways. Its PLC activity is controlled by G␣ 12 and Ras, and PLC⑀ bears a GEF activity toward Ras and Rap1. Hence PLC⑀ was a good candidate to mediate the observed PACAP-induced Rap1 GTP loading. By using two different primer pairs for rat PLC⑀, we performed RT-PCR and evidenced PLC⑀ expression in rat kidney but not in PC12 cells (data not shown).
Altogether, the present data indicate that activation of the PACAP receptor triggered a pleiotropic signaling. The proximal effectors of Pac1 are AC and PLC whose activation resulted in the stimulation of a set of kinases directly controlled by second messengers, i.e. PKA, PKC, and calcium/calmodulin-dependent kinase. Activation of the PACAP receptor also resulted in Ras activation through a mechanism that was not studied in detail in this report. Ras activation was primarily involved in the initial peak stimulation of ERK activity. In parallel, PACAPelicited PLC activation resulted in Rap1 GTP loading which was, per se, not sufficient to induce an efficient ERK stimulation. The activation of other signaling pathways such as those controlled by Ras, cAMP, and PKC, including Raf-1, were required to potentiate the effect of GTP-loaded Rap1. Hence, the present work revealed an unforeseen difference between Ras and Rap1 in the stimulation of the ERK pathway. Whereas GTP-loaded Ras was fully competent to stimulate the ERK pathway, this was not the case for Rap1 which required additional signals to become effective. This finding will hopefully help to solve some of the "apparent discrepancies" (52) found in the literature on the mechanism of stimulation of the ERK pathway by Rap1 in neuronal cells. These data also reconciled previous reports with respect to the role of the cAMP pathway in PACAP-versus forskolin-induced ERK activation in PC12 cells (5,6,36,58). Finally, our data shed new light on the control of Rap1 activity by GPCRs, an important event in the context of neuronal differentiation.