Activation of Mitogen-activated Protein Kinase by the A2A-adenosine Receptor via a rap1-dependent and via a p21 ras -dependent Pathway*

The A2A-adenosine receptor, a prototypical Gs-coupled receptor, activates mitogen-activated protein (MAP) kinase in a manner independent of cAMP in primary human endothelial cells. In order to delineate signaling pathways that link the receptor to the regulation of MAP kinase, the human A2A receptor was heterologously expressed in Chinese hamster ovary (CHO) and HEK293 cells. In both cell lines, A2A agonist-mediated cAMP accumulation was accompanied by activation of the small G protein rap1. However, rap1 mediates A2A receptor-dependent activation of MAP kinase only in CHO cells, the signaling cascade being composed of Gs, adenylyl cyclase, rap1, and the p68 isoform of B-raf. This isoform was absent in HEK293 cells. Contrary to CHO cells, in HEK293 cells activation of MAP kinase by A2A agonists was not mimicked by 8-bromo-cAMP, was independent of Gαs, and was associated with activation of p21 ras . Accordingly, overexpression of the inactive S17N mutant of p21 ras and of a dominant negative version of mSos (the exchange factor of p21 ras ) blocked MAP kinase stimulation by the A2A receptor in HEK 293 but not in CHO cells. In spite of the close homology between p21 ras and rap1, the S17N mutant of rap1 was not dominant negative because (i) overexpression of rap1(S17N) failed to inhibit A2A receptor-dependent MAP kinase activation, (ii) rap1(S17N) was recovered in the active form with a GST fusion protein comprising the rap1-binding domain of ralGDS after A2A receptor activation, and (iii) A2A agonists promoted the association of rap1(S17N) with the 68-kDa isoform of B-raf in CHO cells. We conclude that the A2A receptor has the capacity two activate MAP kinase via at least two signaling pathways, which depend on two distinct small G proteins, namely p21 ras and rap1. Our observations also show that the S17N version of rap1 cannot be assumed a priori to act as a dominant negative interfering mutant.

Adenosine is ubiquitously released from metabolically active cells and is rapidly generated in the extracellular space by dephosphorylation of ATP that has been released from nerve terminals as a cotransmitter. Hypoxia leads to accumulation of excessive amounts of the nucleoside. In the extracellular space, adenosine serves as an autacoid that regulates the function of virtually every organ and tissue via four different classes of G protein-coupled receptor subtypes, designated A 1 -, A 2A -, A 2B , and A 3 -adenosine receptor (1). These differ by their affinity for the endogenous agonist as well as by their pharmacological specificity. In addition, they operate through distinct signaling mechanisms. The A 1 -and A 3 -adenosine receptors control most, if not all, their cellular effectors via pertussis toxin-sensitive G proteins of the G i and G o family; in contrast, both A 2A -and A 2B -adenosine receptors couple to G s and thereby stimulate cAMP formation (2).
In primary cultures of endothelial cells from various species and vascular beds, adenosine stimulates proliferation (3,4), an observation that is consistent with a possible role of adenosine in angiogenesis (5). In human endothelial cells derived from the umbilical vein, the mitogenic effect is mediated by the A 2Aadenosine receptor (6); however, the available evidence indicates that growth stimulation is not mediated by cAMP, because forskolin, an activator of adenylyl cyclase, and membrane-permeable cAMP analogs such as 8-Br-cAMP 1 inhibit endothelial cell growth (6,7). In addition, the endothelial A 2A -adenosine receptor stimulates phosphorylation of the MAP kinase isoforms erk1 and erk2 (8) and of p70S6 kinase (9), effects that cannot be mimicked by 8-Br-cAMP.
It has long been appreciated that elevated levels of cAMP inhibit the proliferation of many cells (10). This effect is thought to arise, at least in part, from inhibition of the interaction of p21 ras with the downstream kinase raf-1, an effect that leads to cAMP-mediated suppression of the MAP kinase pathway (11)(12)(13)(14). However, in some cells activation of PKA does lead to activation of MAP kinase, the best studied example being PC12 cells; in this pheochromocytoma cell line, the stimulation of MAP kinase by cell-permeable analogs of cAMP has been linked to activation of B-raf via rap1, another member of the ras-like family of small monomeric GTP-binding proteins (15). These findings predict that receptor-induced adenylyl cyclase activation ought to activate MAP kinase in a manner dependent on rap1 but independent of p21 ras . This prediction has been verified in S49 mouse lymphoma cells where the ␤ 2 -adrenergic receptor requires G s and rap1, but not p21 ras , to signal to MAP kinase (16). In contrast, in HEK293 cells, the ␤ 2 -adrenergic receptor has recently been shown to activate MAP kinase via a complex pathway that involves PKA-dependent phosphorylation of the receptor; this modification causes the G protein specificity of the receptor to switch from G s to G i ; the resulting activation of G i , in turn, generates enough free G protein ␤␥-dimers to support the stimulation of MAP kinase via tyrosine kinase-dependent activation of p21 ras (17). It is not clear if this model of signal transduction is generally applicable to all G s -coupled receptors. In the present work, we have investigated the mechanism by which the A 2A -adenosine receptor, a typical G s -coupled receptor, regulates MAP kinase after heterologous expression in two cell lines, i.e. HEK293 and CHO. In both cell lines, cAMP was generated and MAP kinase was activated in response to agonist stimulation. However, MAP kinase activation was achieved via distinct signaling pathways in the two cell lines. In CHO-A 2A cells, MAP kinase activation involved G␣ s , cAMP, PKA, rap1, and B-raf; in contrast, in HEK-A 2A cells, MAP kinase activation was independent of G s and cAMP and required p21 ras .

EXPERIMENTAL PROCEDURES
Materials-Guanine nucleotides, adenosine deaminase, basic fibroblast growth factor (bFGF), the C12A5 anti-hemagglutinin mouse monoclonal antibody, and Complete ® protease inhibitor tablets were from Roche Molecular Biochemicals. CGS21680 was from Tocris Cookson Ltd. (Bristol, United Kingdom). Hepes was from Biomol (Munich, Germany); rolipram and XAC were obtained from Research Biochemicals (Natick, MA). The materials required for SDS-polyacrylamide gel electrophoresis were from Bio-Rad. Fetal calf serum was from PAA Laboratories (Linz, Austria), Dulbecco's modified Eagle's medium, nonessential amino acids, ␤-mercaptoethanol, and G418 (Geneticin) were obtained from Life Technologies, Inc. Ham's F-12 medium was from BioConcept (Allschwil, Switzerland). cAMP, 8-Br-cAMP, protein A-Sepharose, CPA, DNase I, cholera toxin, forskolin, pertussis toxin, L-glutamine, PDBu, penicillin G, streptomycin, soybean trypsin inhibitor, lysozyme, Triton X-100, and thrombin were purchased from Sigma. The Mek1 inhibitor PD98059 and affinity-purified rabbit antisera recognizing the dually phosphorylated forms of erk1 and erk2 or all forms of erk1/erk2 was from New England Biolabs Inc. (Beverly, MA). The inhibitor of protein kinase A H89 was from Alexis Corp. (Laeufelfingen, Switzerland). The Micro BCA ® protein assay reagent kit was from Pierce. Buffers and salts were from Merck (Darmstadt, Germany); [ 3 H]Adenine was from NEN Life Science Products. Glutathione-Sepharose and protein G-Sepharose was from Amersham Pharmacia Biotech (Uppsala, Sweden). Centrifuge tubes and tissue culture plates were from Greiner (Vienna, Austria) and from Cornig Costar (Acton, MA). SuperFect ® polycationic transfection reagent and plasmid preparation kits were from Qiagen (Hilden, Germany). The cDNA coding for the human A 2A -adenosine receptor in the plasmid vector pWS-1253e-A 2A -R and the HEK293 (HEK-A 2A ) as well as the CHO (CHO-A 2A ) cell lines stably expressing the human A 2A -adenosine receptor (18), were kindly provided by M. J. Lohse (University of Wuerzburg, Wuerzburg, Germany). The following plasmids were generous gifts: the (Rous sarcoma virus-based) plasmid for mammalian expression of a HA-tagged p44 MAP kinase (HA-Erk1) from U. Rapp and J. Troppmeier (University of Wuerzburg, Wuerzburg, Germany), the plasmid for expression of dominant negative mSos from Y. Daaka and R. J. Lefkowitz (Duke University, Durham, NC), the plasmid driving bacterial expression of the minimal ras binding domain of raf1 (raf-RBD) fused to GST as well as (cytomegalovirus-based) plasmids for mammalian expression rap1A(S17N) and Ras(S17N) from D. Vogt and A. Wittinghofer (Max Planck Institute, Dortmund, Germany); the plasmids for bacterial expression of a GST fusion protein of the rap1 binding domain of ralGDS (ral-RBD) and for mammalian expression of a HA-tagged version of rap1A(S17N) from B. Franke and J. L. Bos (University of Utrecht, Utrecht, Netherlands). The vectors pEGFP-C1 and pRc-CMV were obtained from CLONTECH (Palo Alto, CA). Mouse derived pan-ras monoclonal antibody was purchased from Oncogene Research Products (Cambridge, MA), affinity-purified rabbit antiserum against rap1/ Krev-1 was from Santa Cruz Biotechnology Inc. (Santa Cruz, CA), and horseradish peroxidase-conjugated anti-mouse-and anti-rabbit immunoglobulin antibodies were from Amersham Pharmacia Biotech (Buckinghamshire, United Kingdom). The immunoreactive bands on nitrocellulose blots were detected by chemiluminescence using SuperSignal chemiluminescence substrate from Pierce.
Cell Culture and Cellular Transfections-HEK293 cells were maintained in Dulbecco's modified Eagle's medium, CHO cells were grown in Ham's F-12 nutrient mixture at 5% CO 2 and 37°C. Culture media were supplemented with 10% fetal calf serum, 2 mM L-glutamine, ␤-mercaptoethanol, non-essential amino acids, 100 units/ml penicillin G, 100 g/ml streptomycin. Media for the culture of stably transfected cells were supplemented with 0.2 mg/ml Geneticin (G418) in order to maintain the selection pressure. For transient transfections, HEK-A 2A cells were plated at a density of 8 ϫ 10 5 cells/60-mm dish and transiently transfected with 5-10 g of the desired cDNAs using the calcium phosphate precipitation method. CHO-A 2A cells were plated at a density of 2.5 ϫ 10 6 cells/10-cm dish transfected using the polycationic Super-Fect ® reagent. Co-transfection of pEGFP-C1, a vector carrying a redshifted variant of wild-type green fluorescent protein cDNA from the jellyfish Aequoria victoria, served as a control to monitor transfection efficiency by fluorescence microscopy. When required in cotransfections with several plasmids, the appropriate empty vectors were added to keep the amount of DNA/dish constant. The medium was changed to remove excess DNA precipitates 3-18 h after transfection. Serum starvation and incubation with adenosine deaminase was initiated on day 1 or 2 after transfection, prior to the MAP kinase assay; at this time point, transfected (and control) CHO-A 2A cells were also replated; cells from one transfection on a 10-cm dish were divided onto four 6-cm dishes. The subsequent starvation under serum-free conditions lasted for 24 and 48 h for HEK-and CHO-A 2A , respectively; thereafter, MAP kinase stimulation was done as outlined below. Primary cultures of human endothelial cells were isolated from umbilical veins and propagated as described previously (7).
Stimulation of MAP Kinase Phosphorylation and Immunoblots-Cells were grown to confluence on 60-mm dishes, and rendered quiescent by serum starvation for 24 -48 h prior to MAP kinase assays in order to minimize basal activity. The starving medium was supplemented with 1 IU/ml adenosine deaminase to remove any endogenously produced adenosine. Pretreatment of cells with cholera toxin was also done for 48 h in starving medium containing adenosine deaminase. If not otherwise indicated, cells were subsequently stimulated by addition of medium containing or lacking agonists at 22°C for 5 min. Control incubations were carried out in order to verify that the carry-over of dimethyl sulfoxide, which resulted in final concentrations of Յ0.1%, neither affected the basal levels of MAP kinase phosphorylation nor the response to agonists. The exposure to agonists or vehicle was terminated by rapidly rinsing with ice-cold phosphate-buffered saline; thereafter, the dish was immediately immersed in liquid nitrogen. After rapid thawing, cells were lysed by addition of 80 l of lysis buffer (50 mM Tris, 40 mM ␤-glycerophosphate, 100 mM NaCl, 10 mM EDTA, 10 mM p-nitrophenol phosphate, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na 3 VO 4 , 10 mM NaF, pH adjusted to 7.4 with HCl, 1% Nonidet P-40, 0.1% SDS, 250 units/ml aprotinin, 40 g/l leupeptin). The cellular debris was removed by centrifugation at 10,000 ϫ g for 10 min, and the total protein content was measured photometrically using bicinchoninic acid (Micro-BCA kit, Pierce). Aliquots corresponding to 2.5-5 ϫ 10 4 cells (10 -30 g of protein) were dissolved in Laemmli sample buffer containing 30 mM dithiothreitol and applied to SDS-polyacrylamide gels (monomer concentration 10 -15% acrylamide, 0.26 -0.4% bisacrylamide). MAP kinase activation was assayed by incubating nitrocellulose blots with an antiserum that recognizes only the phosphorylated forms of p42 and p44 MAP kinase; in order to rule out that the differences observed were due to the application of unequal amounts of lysates, control blots were also probed with an antiserum recognizing both the unphosphorylated (inactive) and phosphorylated (active form). The immunoreactive bands were visualized by enhanced chemiluminescence using horseradish peroxidase-linked secondary antibodies. Immunodetection of the other proteins was performed in an analogous manner, using the appropriate antibodies or antisera.
Pull-down Assays for the Determination of p21 ras and rap1 Activation-GST fusion protein of the minimal ras-binding domain of raf1 (19) as well as of the rap1-binding domain of ralGDS (ral-RBD, Ref. 20) were expressed in Escherichia coli (strain BL21DE3); following induction by isopropyl-1-thio-␤-D-galactopyranoside, bacterial lysates were prepared as described. GST fusion proteins were immobilized by incubating bacterial lysates for 1 h at 4°C with glutathione-Sepharose (GSH-Sepharose) preequilibrated in pull-down buffer (50 mM Tris, 200 mM NaCl, 2 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 10% glycerol, 1% Nonidet P-40, 2 g/ml aprotinin, 1 g/ml leupeptin, 10 g/ml soybean trypsin inhibitor). Sepharose beads were washed three times in order to remove excess GST fusion protein. Cells were prepared for the assay in a manner similar to that outlined above for MAP kinase assays (serum starvation and incubation with adenosine deaminase for 24 -48 h,); if not otherwise stated, incubation with agonists was carried out for 5 min, followed by rapidly rinsing with ice-cold phosphate-buffered saline and addition of pull-down buffer to achieve cell lysis. Cell lysates were cleared by centrifugation (10,000 ϫ g for 10 min). The resulting supernatants were incubated together with the GSH-Sepharose beads (50 l of a 1:1 slurry containing about 10 g of immobilized GST fusion protein) for 1 h to allow for the association of activated p21 ras or rap1 with the effector-GST fusion protein. Samples were washed three times in modified radioimmune precipitation buffer, resuspended in Laemmli sample buffer, and applied to SDS-polyacrylamide gels; p21 ras or rap1 were visualized using specific antibodies in a dilution of 1:500.
Agonist and Cholera Toxin-mediated Cellular cAMP Accumulation-Cells were grown to confluence in six-well plates. [ 3 H]Adenine (0.1 mCi/ml) was added 18 h prior to stimulation of the cells and was maintained throughout the subsequent incubation at the same concentration. Rolipram (100 M) and adenosine deaminase (1 unit/ml) were added 60 min before termination of the assay; agonists were added for the last 30 min. Cholera toxin (0.03-1 g/ml) was added for the time periods indicated in the figure legends. Assays were performed in triplicate. The formation of [ 3 H]cAMP was quantified according to Salomon (21).
Each experiment reported was carried out at least three times.  (Fig. 1, B and C), CGS 21680 stimulated MAP kinase activity over a concentration range that was reasonably similar to that required for adenylyl cyclase activation and that was comparable to that required for MAP kinase activation in primary human endothelial cells (Fig. 1B, OE).

Activation of Adenylyl Cyclase and of MAP Kinase by the Heterologously Expressed Human
Activation of MAP kinase was mediated by the A 2A -adenosine receptor, because (i) the A 1 -selective agonist CPA was substantially (100 -1000-fold) less potent (shown for CHO-A 2A in Fig. 1, B and C, left panel) and (ii) the adenosine receptor antagonist XAC prevented the activation of MAP kinase over a wide concentration range of CGS21680 (shown for CHO-A 2A in Fig. 1D); (iii) finally, MAP kinase stimulation by CGS21680 was not observed in untransfected control cells (data not shown). In the presence of PD098059, an inhibitor of MAP kinase kinase 1 (Mek1; see Ref. 22), the stimulation of MAP kinase by CGS21680 was suppressed (shown for HEK-A 2A in Fig. 1C, right panel); hence, Mek1 is an upstream regulator of MAP kinase in the signaling cascade that links the A 2A -adenosine receptor to MAP kinase.
Role of G Proteins-Cholera toxin-induced ADP-ribosylation of G␣ s impairs the ability of the protein to cleave GTP. This leads to irreversible activation of G␣ s , resulting in a pronounced stimulation of downstream effectors; in most cells, the effect is reversed due to down-regulation of G␣ s after long term toxin treatment (23). We exposed CHO-A 2A and HEK-A 2A cells to cholera toxin for time spans varying from 10 min to 48 h and measured the generation of [ 3 H]cAMP ( Fig. 2A). Interestingly, a different time course was observed in the two cell lines. In HEK-A 2A cells (Fig. 2A, right panel), accumulation of cAMP rose rapidly; the peak was reached after 3 h of exposure to cholera toxin and was followed by a decrease of cAMP accumulation such that at 12 h after addition of cholera toxin cAMP had returned to basal levels; these remained unchanged over the next 36 h. Immunoblots with a G␣ s -specific antiserum confirmed that the levels of G␣ s were greatly reduced (data not shown). In contrast, in CHO-A 2A cells, the initial peak of cAMP production at 3 h was rather small; unlike HEK-A 2A cells, cAMP levels increased subsequently and a pronounced elevation that lasted from 24 to 48 h was observed ( Fig. 2A, left  panel). A down-regulation of G␣ s was also not detected in CHO-A 2A cells by immunoblots, even if cells were exposed for 48 h to cholera toxin concentrations up to 1 g/ml (data not shown). Accordingly, agonist-induced generation of cAMP after 48 h of pretreatment with cholera toxin remained unaffected in CHO-A 2A (cf. filled bars in the left panel of Fig. 2B), but was virtually abolished in HEK-A 2A cells (Fig. 2B, right panel). We are currently unable to explain the resistance of G␣ s in CHO-A 2A cells to down-regulation by cholera toxin. More importantly, the loss of G␣ s and of the receptor-mediated cAMP increase in HEKA 2A was exploited to determine, if MAP kinase stimulation by the A 2A receptor required G␣ s . As can be seen in the right panel of Fig. 2C, the response of MAP kinase to the A 2A agonist was readily detectable in HEK-A 2A cells that had been preincubated for 48 h with cholera toxin. This suggests that in HEK cells the A 2A -adenosine may recruit other signaling mechanisms in addition to or in the place of G s that trigger the MAP kinase cascade.
In contrast, after pretreatment with cholera toxin, the basal level of MAP kinase phosphorylation was substantially increased in CHO-A 2A cells and CGS21680 failed to induce any further stimulation of MAP kinase (Fig. 2C, left panel). Activation of PKC by the phorbol ester ␤-PDBu still augmented MAP kinase phosphorylation in cholera toxin-treated CHO-A 2A cells (Fig. 2C, left panel); hence, we rule out that the absence of an A 2A agonist effect reflects a general unresponsiveness of the cells. It is more likely that the high basal level of MAP kinase phosphorylation is caused by the persistent elevation of cAMP levels, which occlude any further stimulation of MAP kinase by CGS21680 on MAP kinase (see below).
In HEK293 cells, the ␤ 2 -adrenergic receptor switches its G protein coupling specificity from G s to G i upon PKA-dependent phosphorylation; this allows for pertussis toxin-sensitive activation of MAP kinase in response to ␤-adrenergic agonists (17). We have therefore tested whether the A 2A -adenosine receptormediated MAP kinase activation was dependent on pertussis toxin-sensitive G proteins. However, ADP-ribosylation of G␣ i by pertussis toxin pretreatment of CHO-A 2A (Fig. 3A, middle  panel) and of HEK-A 2A cells (Fig. 3A, bottom panel) had no effect on MAP kinase phosphorylation following agonist exposure. In experiments carried out in parallel under identical conditions, the incubation of stably tranfected cells with pertussis toxin completely suppressed signaling by prototypical G i -coupled receptors such as the A 1 -adenosine and D 2 -dopamine receptors and more than 99% of the G␣ i subunits were found to be ADP-ribosylated (data not shown; see also Ref. 24). HEK-A 2A cells were also incubated with the combination of cholera toxin (300 ng/ml, 48 h) and pertussis toxin (100 ng/ml, 24 h). This harsh treatment caused many cells to detach from the plastic support; nevertheless, in the remaining adherent cells MAP kinase stimulation was still seen upon CGS21680 stimulation (data not shown). Finally, PKC isoforms can be activated by G i -coupled receptors via release of ␤␥-dimers, which lead to activation phospholipase C␤2 and related isoforms (25) as opposed to stimulation of inositol trisphosphate by G␣ q /G␣ 11 ; hence, we have also incubated HEK-A 2A cells with GF109203X, an inhibitor of Ca 2ϩ -dependent PKC isoforms. Although the ␤-PDBu-dependent phosphorylation of MAP kinase was abolished by 1 M GF109203X, the stimulation by CGS21680 remained unaffected (Fig. 3B). We therefore rule out that any G i -dependent pathway is involved in the MAP kinase response following stimulation of the A 2A -adenosine receptor in HEK293 cells.
cAMP-induced MAP Kinase Activation-As mentioned above, pretreatment of CHO-A 2A cells with cholera toxin sub-stantially increased the phosphorylation of MAP kinase in the absence of any agonist. If this effect were related to cAMP-dependent activation of PKA (as opposed to a result of ␤␥-dimers released from persistently activated G␣ s ), a membrane-permeable cAMP analogue ought to mimic the effect of cholera toxin. This was the case. Addition of 8-Br-cAMP to CHO-A 2A cells resulted in MAP kinase activation (Fig. 3A); the stimulation of MAP kinase phosphorylation induced by both CGS21680 and 8-Br-cAMP was blocked by the protein kinase A inhibitor H89. In contrast, addition of 8-Br-cAMP to HEK-A 2A did not per se stimulate MAP kinase up to concentrations of 0.5 mM (Fig. 3, A  and B), although activation by a low concentration of CGS21680 was readily detectable (Fig. 3B). The absence of any stimulatory effect is not due to the inability of 8-Br-cAMP to permeate into HEK-A 2A cells, because the nucleotide blunted the effect of bFGF (Fig. 3B) Activation of rap1 and p21 ras -The small monomeric GTPase p21 ras plays a crucial role as an upstream regulator of MAP kinase. In ras-dependent MAP kinase cascades, cAMP exerts an inhibitory effect by suppressing the activation of raf-1 (11)(12)(13)(14). However, in some cell types (15,16,26), cAMP stimulates MAP kinase and this effect is thought to be mediated via rap1, a small monomeric GTPase of the ras family (27). We have therefore compared the ability of the A 2A -adenosine receptor to activate rap1 in CHO-A 2A and HEK-A 2A . The assay employed for detection of GTP-bound rap1 is based on the highly specific binding to the rap-binding domain of ralGDS (19); ralGDS binds to GTP-bound rap1 with an affinity of 10 nM, while its affinity for rap1.GDP is lower by 2-3 orders of magnitude (28). A GST fusion protein comprising the ral-RBD served as specific bait to pull down activated rap1 from control cells (i.e. cells maintained in the presence of adenosine deaminase) and from cells stimulated with agonists. Activation of rap1 was seen after stimulation with CGS21680, 8-Br-cAMP and thrombin; the extent of stimulation was comparable in both cell lines (Fig.  4A). This observation is consistent with a role of rap1 in linking the A 2A -adenosine receptor to MAP kinase in CHO-A 2A cells. In contrast, 8-Br-cAMP (and forskolin) failed to activate of MAP kinase in HEK-A 2A cells (see Fig. 3B); hence, it appears unlikely that rap1 participates in A 2A agonist-induced MAP kinase activation, although the protein is expressed and is activated in HEK-A 2A cells.
Because p21 ras plays a pivotal role in the activation of MAP kinase by both tyrosine kinase receptors and G protein-coupled receptors, we have searched for increase in GTP-bound p21 ras following activation of the receptor in both CHO-A 2A and HEK-A 2A -cells. The minimal ras-binding domain of raf-1 (amino acids 51-131, raf-RBD) fused to GST was employed as a bait to trap activated p21 ras (19,29). In contrast to rap1, activation of p21 ras was observed only in HEK-A 2A and not CHO-A 2A cells upon stimulation with CGS21680 (Fig. 4B) ative p21 ras only blunted the activation of MAP kinase by bFGF, whereas the stimulation elicited by A 2A agonist and 8-Br-cAMP remained unaffected (Fig. 5A, left panel). In contrast, expression of ras(S17N) abolished the MAP kinase response elicited by the A 2A agonist and bFGF in HEK-A 2A cells (Fig. 5A, right panel). Signaling via p21 ras is important for cell survival after transfection; thus, coexpression of ras(S17N) may result in negative selection. However, immunoblots carried out in parallel showed that ras(S17N) was overexpressed in both cell lines (data not shown). Furthermore, while the cotransfection with ras(S17N) reduced the levels of accumulation of HA-tagged erk1, it is evident from the data shown in Fig. 5B that epitope-tagged enzyme was still abundantly expressed; similarly, the variability in the expression levels in individual transfections cannot account for the difference in phosphorylated MAP kinase recovered in the immunoprecipitates shown in Fig. 5A. More importantly, cotransfection of the cells with a dominant negative version of mSos (17), the exchange factor for p21 ras , recapitulated the effect of ras(S17N). Transient expression of mSosPro in CHO-A 2A cells inhibited MAP kinase activation by bFGF but not by CGS 21680 and 8-Br-cAMP (Fig. 5A, left panel). In contrast, in HEK-A 2A cells cotranfected with mSosPro, the response to both CGS21680 and bFGF was suppressed (Fig. 5A, right panel).
A dominant negative version of rap1 ought to disrupt A 2A receptor-dependent regulation of MAP kinase in CHO-A 2A cells. We have therefore employed a plasmid encoding rap1(S17N); however, transiently expressed rap1(S17N) did not prevent the activation of MAP kinase in CHO-A 2A cells irrespective of the activator tested (Fig. 5, left panel). The dominant negative action of rap1(S17N) has been questioned recently (30). Our observations are also inconsistent with a dominant negative effect of rap1S17N for the following reasons; if cells transiently expressing a HA-tagged version of rap1(S17N) were stimulated with agonists, HA-rapS17N was readily pulled down with ral-RBD-GST from lysates of stimulated cells. The electrophoretic mobility of HA-rapS17N is lower than that of endogenous rap1. Furthermore, HA-rapS17N can easily be discriminated from endogenous rap1 by employing the antibody directed against the HA epitope for immunodetection (Fig. 6A). The extent of activation of HA-rapS17N and of endogenous rap1 was comparable (cf. Fig. 4A). We have also corroborated that rap1(S17N) is not inactive by assessing its ability to associate with an endogenous effector, namely with B-raf (see below).
Activation of B-raf in CHO-A2A Cells-Activation of B-raf is thought to be essential in linking cAMP-dependent activation of PKA to stimulation of MAP kinase (15,31). Provided that the expression of B-raf is distinct in CHO-A 2A and HEK-A 2A cells, this may account for the difference in signaling pathways controlled by the A 2A -adenosine receptor in the two cell lines. Immunoblotting of whole cell lysates with a polyclonal antibody directed against the carboxyl terminus of B-raf showed that the 68-kDa isoform of B-raf was expressed in CHO-A 2A but not in HEK-A 2A cells (lanes lys. in Fig. 6B). The functional role of this B-raf isoform was assessed as follows. HA-tagged rap1(S17N) was transiently expressed in CHO-A 2A and HEK-A 2A cells; cells were stimulated by the A 2A agonist CGS21680 (or with thrombin as a positive control), and the level of rap1associated B-raf was detected by immunoprecipitation with an antibody directed against the HA epitope followed by immunoblotting for B-raf (Fig. 6B, lanes A, C, and T). In samples from CHO-A 2A , association of B-raf to HA-rap1(S17N) was stimulated by CGS21680. In HEK-A 2A cells, neither CGS21680 nor thrombin induced an association of the p68 isoform of B-raf with HA-rap1(S17N) (Fig. 6B), although both compounds strongly activate rap1 in HEK-A 2A (see Fig. 4A). Multiple isoforms (up to 10) can be generated from the B-raf gene by alternative splicing (32). However, we have not detected any additional isoform (other than p68 B-raf) in the immunocomplex with HA-tagged rap1. We note that, after stimulation of CHO-cells with thrombin, the levels of p68 B-raf complexed to HA-rap1(S17N) were not significantly increased (left panel in Fig. 6B). If a HA-tagged version of wild type rap1 was employed, comparable levels of p68 B-raf were recovered in the immunoprecipitates after stimulation with thrombin and CGS 21680 (data not shown). It is also evident that the levels of active HA-rap1(S17N) formed after thrombin stimulation were substantially lower than those observed with endogenous rap1 (cf. Figs. 4A and 6A). Taken together, these observations indicate that the rap1(S17N) exerted a dominant negative effect on the signal generated by the thrombin receptor but not on that of the A 2A receptor. DISCUSSION G protein-coupled receptors control the activity of the MAP kinase cascade via several mechanisms; these include a ␤␥dimer-mediated activation of non-receptor tyrosine kinases (33), Ca 2ϩ mobilization (34) and activation of PKC isoforms (35), signaling via phosphatidylinositol 3-kinase (36), and cAMPdependent activation of PKA (15,16,26). In the present study, we show that the A 2A -adenosine receptor controls at least two distinct signaling pathways that lead to MAP kinase activation; it is evident from our observations that the cellular complement of signaling components determines which pathway is utilized. In CHO cells, the A 2A -adenosine receptor regulates MAP kinase phosphorylation via a cascade composed of G␣ s , adenylyl cyclase, PKA, rap1, p68 B-raf, and Mek1. In contrast, in HEK293 cells, G␣ s , cAMP, and rap1 do not participate in the MAP kinase response because the p68 isoform of B-raf is not available; the A 2A -adenosine receptor rather relies on activation of p21 ras . The ␤ 2 -adrenergic receptor, which is endogenously expressed in HEK 293 cells, activates p21 ras via G i ; this results from a PKA-dependent phosphorylation of the receptor, which switches its G protein-specificity from G s to G i ng/ml bFGF ϩ 5 units/ml heparin (F), 0.1 mM 8-Br-cAMP or vehicle (A). Cell lysates (600 g) were subjected to immunoprecipitation with a monoclonal anti-hemagglutinin antibody. Immunodetection of activated HA-erk1 was accomplished with the antiserum directed against phosphorylated MAP kinase as in Fig. 1; n.d., not determined. The diagrams shown under the immunoblots were obtained by densitometric quantification of the immunoreactivity (n ϭ 3-5; error bars indicate S.E.). Panel B, the expression levels of HA-tagged erk1 was determined in cells cotransfected with the control vector or with plasmids encoding dominant versions of p21 ras (ras(17N)) and of mSos (mSosPro) as indicated; aliquots of the lysates (30 g) employed for the immunoprecipitation shown in A were applied onto SDS-polyacrylamide gels; nitrocellulose blots were probed with an antiserum directed against erk1 and erk2. The bottom row of cells were transfected only with the empty vector. Data are representative for two additional experiments that gave comparable results. ras(S17N). Cells were transiently cotransfected with a plasmid encoding HA-tagged erk1 in combination with the empty vector kRSPA (vector) or with plasmids encoding the dominant negative mutant of p21 ras (ras(S17N)), the dominant negative version of mSos (mSosPro) or rap(S17N). After serum starvation, cells were incubated for 10 min with 1 M CGS21680 (C), 1 M PDBu (P), 5 FIG. 6. Panel A, pull-down of endogenous rap1 and transiently expressed HA-rap1(S17N) with a ralGDS-RBD/GST fusion protein after A 2A -adenosine receptor-induced activation. HEK-A 2A cells were transiently transfected with a plasmid encoding hemagglutinin-tagged rap1(S17N) (HA-rapS17N) and rendered quiescent by serum starvation. Subsequently, the stimulation was done with 1 M CGS21680 (CGS), 0.1 unit/ml thrombin (T), or vehicle (A) for 10 min; after cell lysis, endogenous rap1 and HA-rap1(S17N) were precipitated with the ralGDS-RBD/GST fusion protein as described in the legend to Fig. 4; HA-rap1(S17N) was detected with the monoclonal anti-hemagglutinin antibody; a 5% aliquot of the cell lysate was applied to one lane of the gel (lane Std). Panel B, co-immunoprecipitation of B-raf with HA-rap1(S17N) after stimulation with CGS21680 and thrombin. Cellular  (17) and requires internalization of the receptor (37). However, several findings in the present study rule out that this mechanism operates on the A 2A receptor. Neither pretreatment of HEK-A 2A cells with pertussis toxin (to block G i ) nor with cholera toxin (to deplete G␣ s ) affects the MAP kinase response, which is, in addition, resistant to the PKA inhibitor H89. This is also consistent with the fact that the A 2A -adenosine receptor lacks typical phosphorylation sites for PKA (38). Finally, in HEK-A 2A cells, classical and novel isoforms of protein kinase C are not involved in MAP kinase activation by the A 2A receptor. This is also true for endothelial cells (9), where the mitogenic effect of adenosine cannot be accounted for by activation of protein kinase C because this leads to inhibition of cell growth (39).
The small GTP-binding protein rap1 (ras proximate) was originally identified as product of the Krev1 gene, because it was capable of reverting the transformed phenotype induced by oncogenic Ki-ras (27). This action presumably reflects the ability of rap1 to displace p21 ras from raf-1 (40) and other effectors (41,42). Hence, a constitutively activated form of rap1 (rap1V12G) effectively blunts stimulation of MAP kinase via p21 ras (43). However, rap1 can per se also mediate MAP kinase activation (15,16). Accordingly, when microinjected into or overexpressed in Swiss 3T3 cells, rap1 acts as an oncogene (44,45). Two mechanisms lead to activation of rap1; GDP release mediated by an exchange factor (30) and phosphorylation by Ser/Thr kinases (19); the relation between the two mechanisms of regulation is not clear at present (46). The S17N mutant of rap1 is thought to act in a dominant negative manner; this assumption is based on the obvious analogy with the corresponding mutation in p21 ras (47) and the high degree of homology that the two proteins share. In p21 ras , the replacement of Ser 17 by Asn interferes with the coordination of Mg 2ϩ upon binding of GTP and presumably thereby impedes dissociation of the protein from the exchange factors such as cdc25 and m-Sos (30,48,49). However, in the first report that employed rap1(S17N), transfection of this mutant elicited the same effect as a constitutively active version of rap1 (50). Similarly, recent experiments show that, in vitro, purified rap1(S17N) is activated by and released from the exchange factor C3G; the mutant also fails to prevent activation of wild type rap1 (30). These data strongly suggest that, upon transfection, rap1(S17N) does not block cellular signaling. Our data are in line with this prediction, because we failed to observe a dominant negative of rap1(S17N). More importantly, we found that, upon stimulation of transfected cells by the A 2A agonist, active GTP-liganded rap1(S17N) was generated that is capable of associating with both, an exogenously added effector binding domain (derived from ralGDS) and an endogenous effector (i.e. the p68 isoform of B-raf). In contrast, rap1(S17N) has been reported to disrupt cAMP-dependent activation of MAP kinase (15,16). The source of this discrepancy is unknown. Additional exchange factors for rap1 (other than C3G) have recently identified (51,52); rap1(S17N) may fail to dissociate from some of these. The inability of thrombin to promote the association of rap1(S17N) with p68 B-raf supports this conjecture.
In HEK-A 2A activation of MAP kinase by the A 2A -adenosine receptor recapitulates essential features of the response elicited in endothelial cells (8), because this signaling is independent of G␣ s and cAMP but it requires p21 ras . The nature of components that link activation of p21 ras to the A 2A receptor is not known. The only G protein, which the A 2A -adenosine receptor has been unequivocally demonstrated to couple to, is G s (53). However, it has been recently appreciated that components other than G protein subunits may mediate signaling by heptahelical receptors via a direct interaction with the carboxyl terminus of the receptors (54,55). The A 2A -adenosine receptor has an extended carboxyl terminus, which plays an ill defined role in signaling (56,57). Its potential role in MAP kinase activation is currently being explored.