Stimulation of the mitogen-activated protein kinase via the A2A-adenosine receptor in primary human endothelial cells.

Adenosine exerts a mitogenic effect on human endothelial cells via stimulation of the A2A-adenosine receptor. This effect can also be elicited by the β2-adrenergic receptor but is not mimicked by elevation of intracellular cAMP levels. In the present work, we report that stimulation of the A2A-adenosine receptor and of the β2-adrenergic receptor activates mitogen-activated protein kinase (MAP kinase) in human endothelial cells based on the following criteria: adenosine analogues and β-adrenergic agonists cause an (i) increase in tyrosine phosphorylation of the p42 isoform and to a lesser extent of the p44 isoform of MAP kinase and (ii) stimulate the phosphorylation of myelin basic protein by MAP kinase; (iii) this is accompanied by a redistribution of the enzyme to the perinuclear region. Pretreatment of the cells with cholera toxin (to down-regulate Gsα) abolishes activation of MAP kinase by isoproterenol but not that induced by adenosine analogues. In addition, MAP kinase stimulation via the A2A-adenosine receptor is neither impaired following pretreatment of the cells with pertussis toxin (to block Gi-dependent pathways) nor affected by GF109203X (1 μM; to inhibit typical protein kinase C isoforms) nor by a monoclonal antibody, which blocks epidermal growth factor-dependent signaling. In contrast, MAP kinase activation is blocked by PD 098059, an inhibitor of MAP kinase kinase 1 (MEK1) activation, which also blunts the A2A-adenosine receptor-mediated increase in [3H]thymidine incorporation. Activation of the A2A-adenosine receptor is associated with increased levels of GTP-bound p21ras. Thus, our experiments define stimulation of MAP kinase as the candidate cellular target mediating the mitogenic action of the A2A-adenosine receptor on primary human endothelial cells; the signaling pathway operates via p21ras and MEK1 but is independent of Gi, Gs, and the typical protein kinase C isoforms. This implies an additional G protein which links this prototypical Gs-coupled receptor to the MAP kinase cascade.

When released into or formed in the extracellular space, adenosine acts as an autacoid via interaction with four types of G protein 1 -coupled receptors, termed A 1 -, A 2A -, A 2B -, and A 3adenosine receptor. These receptors are encoded by distinct genes and can be differentiated based on their affinities for adenosine analogues and methylxanthine antagonists (1,2). In addition, the receptors can be classified based on their mechanism of signal transduction; A 1 -and A 3 -adenosine receptors interact with pertussis toxin-sensitive G proteins of the G i and G o family (3,4,5), whereas A 2A -and A 2B -adenosine receptors stimulate adenylyl cyclase via G s (6,7).
Adenosine is ubiquitously released by hypoxic tissues in large amounts; the nucleoside has therefore been proposed as one of the angiogenic factors that link the altered metabolism in oxygen-deprived cells to the formation of new capillaries (8). Earlier observations suggested that adenosine acts as an endothelial mitogen in vivo (9,10). The mitogenic effect of adenosine has been verified in cultured endothelial cells derived from several vascular beds (11)(12)(13). In human endothelial cells, the proliferative response is mediated by the A 2A -adenosine receptor, an effect mimicked by stimulation of the endothelial ␤ 2 -adrenergic receptor (14). However, the mechanism by which adenosine analogues promote endothelial cell growth is not clear; there is, in particular, the apparent paradox that persistent stimulation of the signaling cascade composed of G s , adenylyl cyclase, and protein kinase A, which is downstream of A 2A -adenosine receptor, inhibits endothelial cell proliferation (14 -16). In the present work, we have therefore searched for additional effector mechanisms. We report that, in human endothelial cells, the A 2A -adenosine receptor stimulates the mitogen-activated protein kinase; this activation is independent of G s , G i , and typical protein kinase C isoforms but is associated with activation of p21 ras .
Cell Culture-Human umbilical venous endothelial cells were isolated according to Jaffe et al. (20). Cords were rinsed twice with phosphate-buffered saline and then incubated for 20 min with 0.2% collagenase in phosphate-buffered saline. The cell suspension was collected and centrifuged, and the cell pellet was resuspended in medium 199 enriched with endothelial cell growth supplement (ECGS), 100 g/ml streptomycin, 100 units/ml penicillin, 0.25 g/ml amphotericin B, 1 IU/ml heparin, and 20% fetal calf serum (FCS) and plated in culture dishes precoated with 1% gelatin. The cells were grown at 37°C in a 5% CO 2 humidified incubator. The following day the medium was changed to remove erythrocytes and was renewed twice a week. After 4 -6 days cells formed confluent monolayers and were further subcultured, and cell plating efficiency after detachment with EDTA 0.02% (Sigma) was Ͼ98%. The thus obtained cell population consisted of Ͼ95% endothelial cells verified by their cobblestone morphology and immunofluorescence staining with antibody against von Willebrand factor antigen (Dako, Dakopatts, Denmark). Cells were used in the second and third passage.
MAP Kinase Tyrosine Phosphorylation and Immunoblots-Endothelial cells were grown to confluence on 60-mm culture dishes; subsequently, serum and growth factors were withdrawn, and the cells were maintained for 6 h in starving medium (medium 199 containing 1% methanol-extracted bovine serum albumin). Thereafter, the starving medium was replaced by medium 199 containing 1 IU/ml heparin. After an additional 30-min equilibration period, 0.1 ml of medium containing or lacking agonists and adenosine deaminase were added. Control incubations were carried out to verify that the carry-over of dimethyl sulfoxide, which resulted in maximum final concentrations of Յ0.1%, neither affected the basal level of MAP kinase phosphorylation nor the response to agonists. The incubation was carried out at 22°C in the presence of agonists, inhibitors, and vehicle as indicated in the figure legend and terminated by rapidly rinsing (Յ10 s) with 10 ml of ice-cold phosphate-buffered saline containing 100 M PMSF and immediately frozen in liquid N 2 . After thawing, the cells were scraped from the dishes in 0.1 ml of lysis buffer (in mM: 50 Tris, 40 ␤-glycerophosphate, 100 NaCl, 10 EDTA, 10 p-nitrophenol phosphate, 1 PMSF, 1 Na 3 VO 4 , 10 NaF, pH adjusted to 7.4 with HCl; 1% Nonidet P-40, 0.1% SDS, 250 units/ml aprotinin, 40 g/ml leupeptin). The unsolubilized material was removed by centrifugation at 50,000 ϫ g for 10 min. Protein content in the lysates was measured colorimetrically using bicinchoninic acid (micro-BCA, Pierce). Aliquots corresponding to 2.5-5⅐10 4 cells (10 -30 g of protein) were dissolved in Laemmli sample buffer containing 40 mM dithiothreitol and were applied to SDS-polyacrylamide minigels (monomer concentration 12% acrylamide, 0.16% bisacrylamide, 6 cm of resolving gel). After transfer to nitrocellulose membranes, immunodetection was performed using an anti-phosphotyrosine antibody (UBI, 16-101) conjugated to horseradish peroxidase (1-2 g/ml). Phosphorylation of MAP kinase was also assessed using an antiserum that specifically recognizes the phosphorylated p42 and p44 isoforms (New England BioLabs, 9101) at a 1:1000 dilution. Recombinant phosphorylated and nonphosphorylated p42 MAP kinase (rat extracellular signal-regulated kinase 2, New England BioLabs, 9103) were used as standards. In order to rule out that changes in immunoreactivity could simply be accounted for by different amounts of protein applied to individual lanes, the antibodies were removed by denaturation and reductive cleavage (70°C for 30 min in 62.5 mM Tris⅐HCl, pH 6.8, 2% SDS, 100 mM mercaptoethanol), and the nitrocellulose blots were reprobed with an antiserum, which recognizes the p42 and p44 isoforms of MAP kinase (UBI, 06-183 or New England BioLabs, 9102) at a 1:1000 dilution. In order to detect the shift in electrophoretic mobility associated with MAP kinase activation, cellular lysates (ϳ30 -40 g) were applied onto large SDSpolyacrylamide gels (12-cm resolving gel) containing 2 M urea.
To determine the levels of G protein subunits, endothelial cells (ϳ5⅐10 5 ) were incubated in the presence of cholera toxin, forskolin, and 8-Br-cAMP as outlined in Fig. 5. After 24 -48 h, the medium was removed, and the monolayer was rinsed twice with phosphate-buffered saline. The cells were detached with EDTA, resuspended in 0.5 ml of phosphate-buffered saline, and immediately frozen in liquid N 2 . Cell lysis was achieved by two freeze-thaw cycles. After centrifugation at 50,000 ϫ g for 20 min, the resulting particulate fraction was dissolved in Laemmli sample buffer containing 40 mM dithiothreitol and 2% SDS; 50% aliquots were then applied to SDS-polyacrylamide gels. After transfer to nitrocellulose membranes, the splice variants of G s ␣ were visualized using the G s ␣-specific antiserum CS1 (18) at a dilution of 1:1000. The purified recombinant splice variants of G s ␣, rG s ␣ -s , and rG s ␣ -L (21) were used as standards. The remainder of the particulate fractions was used to assess the levels of the G protein ␤-subunits.
Immunocytochemical Staining-Endothelial cells were fixed in situ in 12-well dishes on their plastic support using acetone/methanol (1:1) or, alternatively, pure methanol at Ϫ20°C for 20 min. After fixation, the dishes were blocked with PBS containing 3% FCS for 1 h at room temperature, incubated for 2 h with the antiserum against MAP kinase diluted in PBS containing 3% FCS, and subsequently stained with the alkaline phosphatase technique using the LSAB kit (DAKO K676) according to the instructions of the manufacturer. Briefly, cells were incubated with a biotinylated link antibody and streptavidin-coupled alkaline phosphatase; New Fuchsin was used as a chromogen.
Analysis of p21 ras -bound Guanine Nucleotides-The method for determining the relative amount of GDP and GTP bound to p21 ras (23) was adapted as follows. Confluent cells in 3.5-cm dishes were serumstarved for 8 h and then labeled with ϳ0.5 mCi of [ 32 P]orthophosphate for an additional 16 h in serum-free, phosphate-free medium (1 ml containing 0.2% methanol-extracted BSA). The cells were washed three times with phosphate-buffered saline (prewarmed to 37°C) and then gently overlayered with 1 ml of phosphate-free medium containing adenosine deaminase (2 units/ml) and BSA. After an additional 30-min equilibration period, agonists or vehicle (medium containing BSA) were added in 0.1 ml. After agonist stimulation, incubations were terminated by lysing the cells with ice-cold buffer (mM: 50 Tris, pH 7.5, 20 MgCl 2 , 150 NaCl, 1 PMSF, 1 Na 3 VO 4 ; 1% Nonidet P-40, 250 units/ml aprotinin); p21 ras was recovered from the lysate by immunoprecipitation with 1.5 g of monoclonal rat antibody Y13-259 (UBI) and 10 g of linker antibody (rabbit anti-rat IgG) complexed to protein A-Sepharose. Guanine nucleotides bound to p21 ras were eluted from the immunoprecipitate by the addition of 20 l of buffer (mM: 750 KH 2 PO 4 , pH 3.4, 5 EDTA, 1 GDP, 1 GTP) and heating (10 min at 65°C). The eluate (10 l/spot) was analyzed by thin layer chromatography on polyethyleneimine-coated plates (developing buffer ϭ 1 M KH 2 PO 4 , pH 4). The radioactivity labeled and the added unlabeled nucleotides were visualized by autoradiography (Kodak X-Omat AR films, 1-4 days of exposure) and a UV lamp, respectively.
[ 3 H]Thymidine Incorporation-The ability of bFGF, EGF, and the adenosine receptor agonist NECA to stimulate [ 3 H]thymidine incorporation into DNA was determined as described previously (14) with minor modifications; briefly, confluent growth-arrested endothelial cells were detached with EDTA and seeded in 96-well plates (1-2⅐10 4 cells/well) in the presence of medium containing 2.5% FCS. After 5 h, an interval required for the cells to adhere, PD 098059 (final concentration ϭ 20 M), monoclonal anti-EGF receptor antibody (final concentration ϭ 3 g/ml), or vehicle was added; 1 h later, the concentration of FCS was raised to 10% (control), and the cells were incubated with the following compounds or combination of compounds at the indicated final concentrations: FCS ϩ adenosine deaminase (2 units/ml), FCS ϩ adenosine deaminase ϩ NECA (1 M), FCS ϩ bFGF (10 ng/ml), FCS ϩ EGF (10 ng/ml). The final volume was 0.15 ml; 15 h thereafter, 0.1 ml of medium 199 containing [ 3 H]thymidine (0.3 to 0.5 Ci) was added for an additional 4 h. At the end on the incubation, the medium was removed, and the cells were detached by trypsinization and lysed by a freeze-thaw cycle. The particulate material was trapped onto glass fiber filters using a Skatron cell harvester, and the radioactivity retained was measured by liquid scintillation counting. The blank level of [ 3 H]thymidine retained on the filters in the absence of DNA synthesis (determined in the presence of aphidicolin) was Ͻ50 cpm. Assays were done in sextuplicate.
Each experiment reported was carried out at least three times with three different endothelial cell batches.

RESULTS
Tyrosine Phosphorylation and Activation of MAP Kinase in the Presence of Adenosine Analogues and Isoproterenol-In order to search for potential intracellular targets downstream of the A 2A -adenosine receptor, cellular lysates were obtained from endothelial cells incubated with adenosine analogues and isoproterenol; adenosine deaminase was present in the medium to remove endogenously produced adenosine. Immunoblotting with an antibody against phosphotyrosine revealed an increased immunostaining of bands migrating at 42 and 44 kDa in the presence of the nonselective adenosine analogue NECA, the highly A 2A -selective adenosine analogue CGS 21680, and the ␤-adrenergic agonist isoproterenol, all at 1 M (Fig. 1A). The pattern of the other bands visualized by the phosphotyrosine antibody was not affected by the agonists employed; this also holds true for the higher and lower molecular weight range not represented in Fig. 1A. While the increase in tyrosine phosphorylation was consistently observed in the 42-kDa band, the response of the 44-kDa band was more modest (see below). In contrast to the effect of NECA and CGS 21680, the pattern of tyrosine phosphorylation was unaffected in the presence of the A 1 -selective agonist CPA. The phosphotyrosine-directed antibody was removed by reductive cleavage and heat denaturation, and the blots were reprobed with an antiserum directed against mitogen-activated protein kinases (MAP kinase); the 42-and 44-kDa bands, tyrosine phosphorylation of which was regulated by the agonists employed, comigrated with the p42 and p44 isoforms of MAP kinase, respectively (Fig. 1B). This suggested that A 2A -adenosine receptor agonists as well as isoproterenol stimulated tyrosine phosphorylation of MAP kinases. We have verified this interpretation by using an antiserum specific for the phosphorylated forms of p42 and p44 MAP kinase (Fig. 1C). Both, the p42-and the p44-isoform of MAP kinase are present in human endothelial cells in roughly equal amounts (Fig. 1B); phosphorylation of the p42 isoform in response to mitogens was consistently observed; however, the response of the p44 isoform was modest and varied greatly in individual batches of primary cultures (cf. Figs. 1-3).
Incubation of the cells with the adenosine receptor antagonist XAC completely prevented the NECA-induced increase in MAP kinase phosphorylation irrespective of whether immunodetection was performed with the anti-phosphotyrosine antibody (Fig. 1D) or with the antiserum against phosphorylated MAP kinase (Fig. 1E). Similarly, the ␤-adrenergic antagonist L-propranolol blocked the isoproterenol-induced response (Fig. 1, D and E).
We have compared the ability of NECA and isoproterenol to promote tyrosine phosphorylation of the 42-kDa band with that of basic fibroblast growth factor and the phorbol ester TPA (1 M); both compounds are potent and efficient mitogens for human endothelial cells. In most cell batches, incubation of the endothelial cells with the latter two compounds resulted in higher levels of phosphotyrosine incorporation into the 42-and 44-kDa bands than NECA or isoproterenol ( Fig. 2A). This is consistent with the efficiency with which the compounds stimulate proliferation; under our culture conditions, TPA and bFGF are the strongest mitogenic stimuli.
We have verified that the increase in tyrosine phosphorylation resulted in higher MAP kinase activity. Lysates from control and stimulated cells were applied to a polyacrylamide gel containing the MAP kinase substrate myelin basic protein.
Following renaturation, MAP kinase activity was detected in the presence of [␥-32 P]ATP. An increase of myelin basic protein phosphorylation was observed in the 42-/44-kDa region of the gel in response to bFGF, TPA, NECA, and isoproterenol (Fig.  2B). Phosphorylation of MAP kinases slows their migration through polyacrylamide gels. This was not seen with the minigel system (6 cm resolving gel) employed to generate the data depicted in Fig. 1. However, the shift to slower mobility was detectable if a larger gel was employed (Fig. 2C). The isoproterenol-and NECA-induced stimulation of tyrosine phosphorylation of MAP kinase gradually increased, the maximum being reached after 10 -15 min (Fig. 3, A and C) and maintained for at least 50 min (not shown).
Translocation of MAP Kinase-Persistent activation of MAP kinase results in cellular redistribution such that a fraction migrates into the nucleus, where it phosphorylates target proteins (24 -26). We have determined the cellular localization of MAP kinases by immunocytochemistry in order to search for an agonist-induced redistribution. In untreated control cells, immunoreactive material was primarily observed in the cellular periphery yielding a ring-like staining pattern (Fig. 4A). If the endothelial cells were incubated with the phorbol ester TPA (1 M), the immunoreactive material was found almost exclusively in the perinuclear region (Fig. 4C). Addition of NECA (Fig. 4B) or of isoproterenol (not shown) altered the distribution of the immunoreactive material in a similar manner, although the effect was less pronounced than that of TPA. The different extent of translocation correlates with the efficiency with which the compounds stimulate proliferation (see above).
Delineation of the G Protein-dependent Signaling Cascade-The A 2A -adenosine and the ␤-adrenergic receptors are coupled to adenylyl cyclase activation via G s . However, if the cAMP/ protein kinase A signaling cascade was directly activated by incubating the cells with a high concentration of the membrane-permeable cAMP analogue 8 Br-cAMP, no increase in the phosphorylation of the p42 and p44 MAP kinase was seen when compared with the reference incubation in the presence of adenosine deaminase or to NECA-stimulated cells (Fig. 5A); similar results were obtained if the levels of phosphotyrosine were determined (not shown). Thus, cAMP generated by A 2Aadenosine receptor-mediated adenylyl cyclase stimulation is unlikely to account for activation of MAP kinase. In order to determine whether another G s -regulated effector was involved in the activation of MAP kinase, we have exploited the fact that persistent activation of G s ␣ via cholera toxin but not of the downstream signaling cascade cells leads to its down-regula-tion (18,27). This phenomenon was also seen in endothelial cells treated for 48 h with cholera toxin (Fig. 5B); an essentially complete down-regulation of both splice variants of G s ␣ was already observed after a 24-h incubation with cholera toxin, whereas the levels of G protein ␤-subunits were unaffected (not shown). In spite of the profound reduction in the levels of G s ␣, activation of the A 2A -adenosine receptor by NECA still promoted the phosphorylation of MAP kinase (Fig. 5, C and D). In some, but not all cell preparations, the effect of NECA appeared to be augmented by cholera toxin pretreatment (cf. Fig. 5, C and D). In contrast, cholera toxin pretreatment essentially abolished the response to isoproterenol (Fig. 5D). This indicates that the ␤ 2 -adrenergic receptor but not the A 2A -adenosine receptor requires functional G s ␣ for MAP kinase activation.
Under appropriate conditions, seven transmembrane spanning receptors can interact with several distinct classes of G proteins. G i ␣ -2 is the most abundant G protein ␣-subunit in endothelial membranes and is required to support endothelial cell growth in the presence of serum-derived growth factors (14). We have therefore pretreated the endothelial cells with pertussis toxin to block the functional coupling between receptors and G i . This preincubation eliminates the functional response to G i -dependent endothelial mitogens present in FCS (14). However, phosphorylation of MAP kinase in response to activation of the A 2A -adenosine receptor was not reduced but rather enhanced following pertussis toxin-pretreatment (Fig. 6A).
Direct stimulation of protein kinase C isoforms by the phorbol ester TPA results in MAP kinase stimulation in endothelial cells (cf. Figs. 2 and 4). The phospholipase C␤/protein kinase C signaling cascade is subject to two types of regulation by G protein subunits. Free G protein ␤␥-dimers are generated by subunit dissociation from G proteins of the G i /G o class, which in most cells account for the bulk of the G proteins, and thus mediates pertussis toxin-sensitive stimulation of phospholipase C␤. The second pathway involves G protein ␣-subunits of the G q family. We have therefore also determined the effect of the protein kinase C inhibitor GF109203X on tyrosine phosphorylation of MAP kinase in response to A 2A -adenosine and ␤ 2 -adrenergic receptor stimulation. At 1 M, GF109203X inhibited MAP kinase activation induced by the phorbol ester TPA but did not interfere with the NECA-and isoproterenol-induced tyrosine phosphorylation (Fig. 6B). Identical results were obtained, if the cells were pretreated with the phorbol ester TPA for 24 h which resulted in down-regulation of the typical protein kinase C isoforms ␣ and ␤ and to a lesser extent of protein kinase C⑀. Raising the concentration of GF109203X to 30 M blocked the effect of NECA and isoproterenol; under these conditions, however, the level of basal tyrosine phosphorylation was also reduced indicating that the concentration range at which GF109203X was selective had been exceeded (data not shown).
Tyrosine kinases activate MAP kinase via a signaling cascade that results in guanine nucleotide exchange on p21 ras and subsequent stimulation of raf kinase activity. G protein-coupled receptors are also capable of activating p21 ras in several cell lines (28,29). We have therefore stimulated endothelial cells with bFGF and the adenosine receptor agonist NECA and determined the level of GDP-and GTP-bound p21 ras . Under basal conditions, determined in the presence of adenosine deaminase, the immunoprecipitate contained predominantly radiolabeled GDP (Fig. 7). Both, NECA and bFGF, caused a time-dependent increase in the level of GTP bound to p21 ras (Fig. 7).
In p21 ras -dependent mitogenic signaling the activation of the raf kinase by p21 ras is linked to stimulation of MAP kinase via MAP kinase kinase (MEK). To confirm that the A 2A -adenosine receptor-induced activation of p21 ras and MAP kinase were linked, we have used the inhibitor PD 098059, which blocks the activation of MEK1 (17). Preincubation of endothelial cells with PD 098059 blocked the NECA-induced stimulation of MAP kinase and greatly reduced the response to bFGF (Fig. 8A). If the A 2A -adenosine receptor-induced mitogenic response and activation MAP kinase were causally related, PD 098059 ought to block the ability of NECA (and of bFGF) to stimulate DNA synthesis. This was the case (Table I). In the absence of PD 098059, bFGF and NECA stimulated [ 3 H]thymidine incorporation ϳ1.7and ϳ1.6-fold over the respective control incubations (FCS and FCS ϩ adenosine deaminase). Preincubation of the cells with 20 M PD 098059 did not affect cell viability but reduced the levels of [ 3 H]thymidine incorporation observed in the presence of FCS and of FCS ϩ adenosine deaminase. More importantly, in the presence of PD 098059, the response to bFGF was clearly blunted (1.2-fold stimulation), and the effect of NECA was essentially undetectable (1.1-fold stimulation).
Recent experiments in rat-1 fibroblasts indicate that activation of MAP kinase by G protein-coupled receptors depends on a mechanism that required a functional EGF receptor (30). As mentioned earlier, we did not observe any increase in tyrosine phosphorylation in the high molecular weight range of the immunoblots shown in Fig. 1. This, however, may have escaped detection because of an unfavorable signal-to-noise ratio. We have therefore used two approaches to assess the contribution of EGF receptor-dependent pathways to the A 2A -adenosine receptor-induced response. (i) The endothelial cells were preincubated with several tyrosine kinase inhibitors such as genistein, tyrphostin 25, tyrphostin 23, and tyrphostin 51 for up to 2 h. None of these compounds inhibited MAP kinase activation by NECA up to concentrations of 50 M. However, these results were inconclusive since the inhibitors also failed to block the response elicited by bFGF and EGF. This suggested that human endothelial cells did not accumulate significant amounts of these compounds (data not shown). (ii) Alternatively, cells were preincubated with a monoclonal antibody capable of blocking the human EGF receptor. At a concentration of antibody that blocked the increase in MAP kinase phosphorylation induced by 10 ng/ml EGF and substantially reduced the effect of 10-fold higher EGF concentration, no inhibition of the NECA-induced response was seen (Fig. 8B). Similarly, the EGF-induced stimulation of [ 3 H]thymidine incorporation (ϳ1.7-fold over the FCS control, see Table I) was essentially eliminated by the antibody (ϳ1.1-fold of the reference value ϭ FCS ϩ antibody, see Table I). In contrast the stimulation by NECA remained unaffected by the antibody (ϳ1.6-and 1.7-fold stimulation over the corresponding reference values, i.e. FCS ϩ adenosine deaminase in the absence and presence of antibody, respectively, see Table I). DISCUSSION In the present work, we show that the A 2A -adenosine receptor stimulates MAP kinase (predominantly the p42 isoform) in primary cell cultures of human endothelial cells. Our experiments have identified MAP kinase as an intracellular target, which can account for the growth-stimulating activity of adenosine on endothelial cells, with p21 ras and MAP kinase kinase 1 as components of the underlying signaling pathway.
MAP kinases or extracellular signal-regulated kinases are rapidly stimulated by growth promoting factors acting on a variety of cell surface receptors (24). In turn, MAP kinases phosphorylate and regulate key intracellular enzymes and transcription factors required for G 0 /G 1 transition and cellular proliferation. Receptors with tyrosine kinase activity activate MAP kinases in a multistep process that involves a limited number of defined protein-protein interactions, guanine nucleotide exchange on p21 ras , and activation of a downstream protein kinase cascade (24 -26). In contrast, MAP kinase activation by G protein-coupled receptors is less well understood. At least five distinct pathways for G protein-mediated activation of MAP kinase have been documented. (i) Stimulation of G icoupled receptors results in the formation of GTP-bound p21 ras in fibroblast cell lines (28,29). In most cells, pertussis toxinsensitive G proteins of the G i /G o family are the most abundant such that receptor-induced G protein subunit dissociation will generate high levels of free ␤␥-dimers in the membrane. Acti- FIG. 7. Activation of p21 ras in human endothelial cells. Growtharrested human endothelial cells were metabolically labeled with [ 32 P]orthophosphate for 16 h and subsequently equilibrated in medium containing 2 units/ml adenosine deaminase for 30 min. Thereafter (time ϭ 0), NECA and bFGF were added to a final concentration of 1 M and 10 ng/ml, respectively; an equivalent volume of vehicle (0.1 ml of medium containing BSA) was added to the control incubation (Ada). The incubation continued for 2, 5, and 10 min, as indicated. The cells were lysed, and p21 ras was recovered from the lysates by immunoprecipitation, and guanine nucleotides bound to p21 ras were analyzed by thin layer chromatography as described under "Experimental Procedures." vation of p21 ras may actually be accomplished by the ␤␥dimers. Overexpression of ␤␥-subunits rather than ␣-subunits is associated with increased p21 ras -dependent signaling (31); this effect can be reversed by the ␤␥-dimer binding fragment of the ␤-adrenergic receptor kinase 1/G protein-coupled receptor kinase 2 (32) and is mediated by tyrosine kinase-dependent phosphorylation of Shc adapter proteins (33). (ii) MAP kinase activation can also be achieved through G protein-coupled receptors via protein kinase C isoforms in response to increased diacylglycerol levels generated from phospholipid precursors by phospholipase C or phospholipase D (25,34). (iii) Receptorpromoted calcium entry (22,35) and (iv) elevation of cAMP may lead to MAP kinase activation in some cells (34,36). (v) Recently, MAP kinase activation by G protein-coupled receptors (for lysophosphatidic acid, thrombin, and endothelin-1) has been shown to depend on tyrosine phosphorylation of the EGF receptor in rat-1 fibroblasts. This phenomenon has been termed "transactivation" of the EGF receptor (30). Our observations rule out several of these mechanisms in linking the A 2A -adenosine receptor to MAP kinase activation. In particular, raising intracellular cAMP in endothelial cells does not per se induce tyrosine phosphorylation of MAP kinase. Similarly, activation of protein kinase C isoforms via a G q -dependent stimulation of phospholipase C␤ is not involved as the underlying mechanism; the protein kinase C inhibitor GF109203X is inactive at concentrations that essentially eliminate the activation of MAP kinase by the phorbol ester TPA. The ability of GF109203X to block the NECA-induced MAP kinase activation at high concentrations most likely reflects loss of selectivity and is thus impossible to interpret. We have also failed to detect calcium transients in endothelial cells loaded with the indicator dye Fura-2 after addition of NECA or isoproterenol, although transient increases in intracellular calcium were seen in response to activation of the G q -coupled H 1 -histamine receptor. 2 Finally, a functional EGF receptor is not required for activation of MAP kinase via the A 2A -adenosine receptor. Obviously, we cannot rule out that a receptor tyrosine kinase other than the EGF receptor is transactivated by the A 2Aadenosine receptor in endothelial cells. This type of cooperation between tyrosine kinase receptors and G protein-coupled receptors may be a general phenomenon and specific for cell types or receptor pairs, e.g. in rat vascular smooth muscle cells, stimulation of the angiotensin II type 1 receptor causes tyrosine phosphorylation and activation of the platelet-derived growth factor receptor (37). In contrast, transactivation of the platelet-derived growth factor receptor is not seen in rat-1 fibroblasts (28).
Both the A 2A -adenosine receptor and the ␤ 2 -adrenergic receptor are prototypical G s -coupled receptors. Endothelial cells express ␤ 2 -adrenergic receptors (38); although it is doubtful that epinephrine and/or norepinephrine participate in the regulation of angiogenesis in vivo, ␤-adrenergic agonists exert a mitogenic effect in cultured endothelial cells (14). Our observations show that A 2A -adenosine receptor and ␤ 2 -adrenergic receptor use distinct mechanisms to activate MAP kinase in endothelial cells. Tyrosine phosphorylation of MAP kinase in response to ␤ 2 -adrenergic receptor activation is abolished after cholera toxin pretreatment; in contrast, the A 2A -adenosine receptor does not depend on G s ␣ to stimulate MAP kinase. Hence, we conclude that, in endothelial cells, the signaling pathways by which the two receptors impinge on the MAP kinase cascade require different G proteins.
So far, no interaction of the A 2A -adenosine receptor with G proteins other than G s has been reported; in turkey erythrocytes ␤-adrenergic receptors stimulate phospholipase C␤ via G q (39,40), an effect which cannot be mimicked by the A 2A -adenosine receptor (40). G s -coupled receptor such as the ␤ 2 -adrenergic receptor and the TSH receptor have also been shown to interact with the G 12 /G 13 protein class (41,42) and to thereby activate the Na ϩ /H ϩ -antiporter (43). G␣ 12 and G␣ 13 are expressed in the human endothelial cells used in the present study, and the stimulation of the A 2A -adenosine receptor can stimulate intracellular alkalinization which is blocked by the Na ϩ /H ϩ -exchange inhibitor HOE694 (data not shown); thus, the endothelial A 2A -adenosine receptor may couple directly to G 12 and/or G 13 . In addition, ␤-adrenergic receptors are known to directly interact with G i under appropriate conditions (44). In endothelial cells, G i is clearly not involved in MAP kinase activation by A 2A -adenosine and ␤ 2 -adrenergic receptors since pertussis toxin failed to block the NECA-and isoproterenolinduced response. Surprisingly, preincubation with pertussis toxin augmented the response to NECA. This was also seen in some cell batches following cholera toxin pretreatment. We currently cannot provide a mechanistic explanation for this phenomenon. However, we have recently observed that, in sympathetic neurons, down-regulation of G s ␣ causes sensitization of the ␣ 2 -adrenergic autoreceptors that signal via ␣-subunits of the G i /G o subfamily (45). This suggests that changing the availablility of functional G protein ␣-subunits may also affect the efficiency of signal transfer through pathways that are controlled by unrelated G proteins.
To our knowledge, there are only two other cell types in which a growth-promoting effect of the A 2A -adenosine receptor has been documented, namely NIH3T3 cells and thyroid epithelial cells. The ability of adenosine analogues to enhance the proliferation of NIH3T3 cells was thought to arise from elevation of intracellular cAMP levels that support the growth induced by obligatory mitogenic signals such as activation of tyrosine kinase-coupled receptors (46). When placed under the control of the thyroglobulin promoter and expressed ectopically as a transgene in the thyroid gland, the A 2A -adenosine receptor is capable of inducing large hyperfunctioning goiters (47); this 2 S. Boehm and V. Sexl, unpublished observations. was interpreted as evidence for continuous activation of the cAMP-signaling cascade which stimulated thyroid growth such that adenosine substituted for TSH, the physiological regulator of the thyroid. However, in primary cultures of human thyroid epithelial cells, activation of the TSH receptor, a G s -coupled receptor which, nevertheless, has the capacity to activate essentially all G protein ␣-subunits (42) leads to tyrosine phosphorylation of the p42 isoform of MAP kinase, and this effect is neither mimicked by activation of the G s /adenylyl cyclase/protein kinase A cascade nor blocked by pertussis toxin treatment (48). Thus, the TSH receptor and the A 2A -adenosine receptor may share a common signaling mechanism that links them to MAP kinase activation in the appropriate cellular background. In human endothelial cells, stimulation of guanine nucleotide exchange on p21 ras and subsequent activation of MEK1 are part of the intervening signaling cascade.