alpha 2B-adrenergic receptor activates MAPK via a pathway involving arachidonic acid metabolism, matrix metalloproteinases, and epidermal growth factor receptor transactivation.

We have investigated the mechanisms whereby alpha(2B)-adrenergic receptor (alpha(2B)-AR) promotes MAPK activation in a clone of the renal tubular cell line, LLC-PK1, transfected with the rat nonglycosylated alpha(2)-AR gene. Treatment of LLC-PK1-alpha(2B) with UK14304 or dexmedetomidine caused arachidonic acid (AA) release and ERK2 phosphorylation. AA release was abolished by prior treatment of the cells with pertussis toxin, quinacrine, or methyl arachidonyl fluorophosphonate but not by the addition of the MEK inhibitor U0126. The effects of alpha(2)-agonists on MAPK phosphorylation were mimicked by cell exposure to exogenous AA. On the other hand, quinacrine abolished the effects of UK14304, but not of AA, suggesting that AA released through PLA2 is responsible for MAPK activation by alpha(2B)-AR. The effects of alpha(2)-agonists or AA were PKC-independent and were attenuated by indomethacin and nordihydroguaiaretic acid. Treatment with batimastat, CRM 197, or tyrphostin AG1478 suppressed MAPK phosphorylation promoted by alpha(2)-agonist or AA. Furthermore, conditioned culture medium from UK14304-treated LLC-PK1-alpha(2B) induced MAPK phosphorylation in wild-type LLC-PK1. Based on these data, we propose a model whereby activation of MAPK by alpha(2B)-AR is mediated through stimulation of PLA2, AA release, generation of AA derivatives, activation of matrix metalloproteinases, release of heparin-binding EGF-like growth factor, transactivation of epidermal growth factor receptor, and recruitment of Shc. Whether this pathway is particular to alpha(2B)-AR and LLC-PK1 or whether it can be extended to other cell types and/or other G-protein-coupled receptors remains to be established.

The ␣ 2 -adrenergic receptors (␣ 2 -ARs) are members of the G-protein-coupled receptor superfamily that mediate physiological responses to the endogenous catecholamines, such as reduction of blood pressure, sedation, platelet aggregation, and inhibition of renin release or insulin secretion. Three subtypes of ␣ 2 -ARs (namely ␣ 2A , ␣ 2B , and ␣ 2C ) have been identified (1). Although recent studies, conducted on mice with genetic alterations of ␣ 2 -AR expression, have clarified the respective roles of ␣ 2A -and ␣ 2B -ARs in the mediation of the cardiovascular and sedative effects of ␣ 2 -agonists, the precise functions of each subtype are far from being definitively elucidated (2). Until recently, the effects of ␣ 2 -ARs were generally considered as exclusively due to the modulation of effectors such as adenylyl cyclase or phospholipase C␤. There is now accumulating evidence that, in addition to these pathways, ␣ 2 -ARs are also involved in the regulation of cell growth via stimulation of mitogen-activated protein kinases (MAPKs). The phosphorylation of MAPKs has been observed in transfected cells (3,4) as well as in various types of cells spontaneously expressing ␣ 2 -ARs (5,6). The three receptor subtypes promoted phosphorylation of ERK1 and ERK2 in Chinese hamster ovary cells (3). According to results obtained in HEK 293 and COS cells (7,8), this effect is independent of receptor internalization via clathrin-coated pits.
Like for other ARs (9), it is probable that the mechanisms whereby ␣ 2 -ARs promote MAPK activation are highly dependent upon the subtype considered and the particular cell type it is expressed in. So far, the precise pathways of the mitogenic signal transmission were exclusively examined for ␣ 2A -AR. In HEK 293 cells (10), activation of ERK1/2 by ␣ 2A -AR is primarily triggered through release of ␤␥ subunits from pertussis toxin-sensitive G proteins, stimulation of phospholipase C␤, phosphoinositide hydrolysis, increase of intracellular Ca 2ϩ , and successive activation of Pyk2 and Src. Activation of Src causes the formation of Shc-Grb2-Sos complex, which leads to ERK phosphorylation via the Ras/Raf/MEK cascade. In COS cells (11), ␣ 2A -AR-induced phosphorylation of ERK2 proceeds via two distinct pathways, which are dependent ("transactivation pathway") or not ("direct pathway") on the tyrosine kinase activity of the EGF receptor (EGF-R). The early steps of both pathways involve the release of ␤␥ subunits from G i proteins and the activation of Src by an unknown process that is independent of inositol 1,4,5-trisphosphate production (12). Then, the phosphorylation of MAPKs occurs either directly through recruitment of the MEK cascade via phosphorylation of the adapter protein Shc or indirectly through activation of unidentified matrix metalloproteinases, release of heparin-binding EGF-like growth factor (HB-EGF), and subsequent transactivation of EGF-R.
Recent experiments carried out on rat proximal tubule cells in primary culture and on LLC-PK1 cells transfected with the rat nonglycosylated ␣ 2 -AR (RNG) gene (LLC-PK1-␣ 2B ) have shown that ␣ 2B -ARs promote MAPK activation and arachidonic acid (AA) release (13). The sequential relationship between PLA2 and MAPK activation was not investigated. As demonstrated in eosinophils during the process of adhesion to fibronectin (14), PLA2 activation may result from its phosphorylation by MAPKs. Conversely, as shown in rabbit renal epithelial cells for angiotensin II receptor, the activation of MAPK could be the consequence of AA release (15). Based on the use of different inhibitors, the present work demonstrates that activation of MAPK by ␣ 2B -AR is, in LLC-PK1-␣2B, primarily mediated by a pathway involving stimulation of PLA2, generation of AA derivatives, activation of matrix metalloproteinases, release of HB-EGF, and transactivation of EGF-R.  (CRM 197), and U0126 were obtained from Calbiochem. Indomethacin, ketoconazole, nordihydroguaiaretic acid (NDGA), phorbol 12myristate 13-acetate (PMA), tyrphostin AG1478, EGF, staurosporine, 1,10-phenanthroline, and all other chemicals were from Sigma. Fetal calf serum was purchased from Invitrogen. Anti-ERK1 and anti-ERK2 polyclonal Ab were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), and anti-active MAPK was from Promega (Madison, WI). Anti-Shc polyclonal Ab and fluorescein-conjugated goat anti-rabbit IgG were respectively purchased from Upstate Biotechnology, Inc. (Lake Placid, NY) and Nordic Immunological Laboratories (Tilburg, The Netherlands).
Culture of LLC-PK1-␣ 2B Cells-The clone of the renal tubular cell line, LLC-PK1, permanently expressing the rat ␣ 2B -AR was obtained by transfection with a pcDNA3 vector containing the coding region of the RNG gene. LLC-PK1-␣ 2B cells were routinely grown in Dulbecco's modified Eagle's medium containing 25 mM glucose, 100 g/ml streptomycin, 100 IU/ml penicillin, and 5% fetal calf serum. Binding experiments with [ 3 H]RX821002 showed that the level of receptor expression was 730 Ϯ 51 fmol/mg of protein.
Detection of ERK1/2 and Shc-Three days postseeding, cells were placed for 24 h in culture medium free of serum. They were then exposed to the compound to be tested, rapidly rinsed with ice-cold phosphate-buffered saline, and harvested in 1 ml of radioimmune precipitation buffer (10 mM Tris-HCl, pH 7.4, 1% Triton-X100, 1% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 2 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, and 0.5 mM aprotinin). Soluble proteins were extracted by centrifugation (15,000 ϫ g, 15 min at 4°C), separated by SDS-PAGE, and blotted onto a nitrocellulose membrane. Phosphorylated forms of MAPKs were revealed by chemiluminescence using anti-active MAPK Ab. Shc phosphorylation was determined after immunoprecipitation. Briefly, 500 l of cell lysate were incubated overnight at 4°C with 5 g of rabbit polyclonal Shc-Ab and 50 l of protein A-agarose beads. Immune complexes were extensively washed with ice-cold radioimmune precipitation buffer, dried, and denatured in Laemmli buffer. Samples were subjected to SDS-PAGE, transferred onto a nitrocellulose membrane, and probed with horseradish peroxidase-conjugated anti-phosphotyrosine Ab. In all experiments, the membranes were stripped of Ig and reprobed using either a mixture of anti-ERK1 Ab and anti-ERK2 Ab or anti-Shc Ab. Films were analyzed by densitometry, and the extent of phosphorylation was normalized to protein loading.
Measurement of AA Release-Cells rendered quiescent by a 24-h period of serum deprivation were labeled for 10 h with 1 Ci/ml [ 3 H]AA. They were carefully washed in Dulbecco's modified Eagle's medium containing 10 mM Hepes and 0.2% fatty acid-free bovine serum albumin and then exposed to the drug to be tested. Aliquots of the incubation medium were collected every 10 min over a period of 30 min and centrifuged (20,000 ϫ g, 10 min, 4°C), and the radioactivity was measured in the supernatant.
Immunofluorescence Microscopy-Cells plated on glass coverslips were grown, rendered quiescent as indicated above, and exposed or not to the compound to be tested. They were fixed in 4% paraformaldehyde (15 min) and treated with 50 mM NH 4 Cl in phosphate-buffered saline (10 min). The cells were permeabilized first in phosphate-buffered saline buffer containing 0.05% saponin and 0.2% bovine serum albumin (15 min) and then in methanol (10 min at Ϫ20°C). All subsequent steps were carried out in permeabilization buffer and were separated by several washes. The cells were incubated with ERK2 polyclonal Ab (1:40) and then with fluorescein-conjugated goat anti-rabbit IgG (1: 400). The coverslips were finally washed in phosphate-buffered saline, mounted in fluorescent mounting medium (Dako Corp., Carpinteria, CA), and examined under epifluorescence illumination. Digital images were captured using the software CoolSNAP (Roper Scientific GmbH, Munich, Germany) and processed with Adobe Photoshop 4 (Adobe Systems Inc., San Jose, CA).
Statistical Analysis-Results are expressed as mean Ϯ S.E. for the number of experiments indicated (n). The data were analyzed using Student's t test, and a p value Ͻ0.05 was considered statistically significant.

AA Release Is Involved in ␣ 2B -AR-induced MAPK Phospho-
rylation-A previous study from our group has shown that exposure of proximal tubule cells to ␣ 2 -agonists resulted in activation of MAPK and in an increase of AA release (13). As depicted in Fig. 1A, the treatment of LLC-PK1-␣ 2B with 1 M UK14304 induced a time-dependent increase of the tyrosine phosphorylation of p42 MAPK. The effect is maximal between 10 and 20 min and persists for at least 40 min. In addition, cell exposure to UK14304 caused an acceleration of AA release, which was abolished by 20 M quinacrine or 50 M methyl arachidonyl fluorophosphonate (not shown). Effects of the ␣ 2agonist on MAPK phosphorylation and AA release were abolished by pretreatment of the cells with pertussis toxin (Fig. 1, B and C). On the other hand, the addition of the MEK inhibitor, U0126, blunted phosphorylation of MAPK but did not affect the augmentation of AA release induced by UK14304. Similar results were obtained using dexmedetomidine, suggesting that stimulation of PLA2 activity by ␣ 2 -agonists is not the consequence of MAPK activation.
In primary culture of rabbit proximal tubule cells, activation of MAPK by angiotensin II receptor is the consequence of AA release (16). In a first step to evaluate the putative role of AA as an intermediary between activated ␣ 2B -AR and MAPK, LLC-PK1-␣ 2B cells were treated with AA. As shown in Fig. 2A, AA induced a dose-dependent increase in MAPK phosphorylation. The effect of AA was detectable at 200 nM and reached a maximum at 20 M. Such concentrations are in the physiological range, since the level of free AA was estimated to be 5 M in the rat kidney (17). Exposure to AA also resulted in the redistribution of ERK2 from the cytoplasm to the nucleus (Fig.  2B), showing that phosphorylation of MAPK could be correlated with the translocation of ERK2. However, in contrast to that for UK14304, the effect of AA was not affected by pretreatment of the cell with pertussis toxin (Fig. 2C). In a second step, the role of endogenous AA release in the activation of MAPK by ␣ 2 -agonist was evaluated using PLA2 inhibitors. Of the compounds tested, quinacrine was the only one with no side effect; all others, including methyl arachidonyl fluorophosphonate and AACOCF3, caused by themselves a significant increase in ERK2 phosphorylation. Such an undesirable effect was previously reported for methyl arachidonyl fluorophosphonate in macrophage (18). Preincubation of LLC-PK1-␣ 2B for 5 min with 20 M quinacrine totally abolished MAPK phosphorylation induced by UK14304 but not by AA (Fig. 3A). Again, results from Western blotting were confirmed by examination of the subcellular distribution of ERK2. Indeed, translocation to the nucleus following exposure to UK14304 is abrogated by quinacrine pretreatment (Fig. 3B). All together, these results are therefore consistent with the implication of PLA2 and AA generation in MAPK activation induced by ␣ 2B -AR.
MAPK Phosphorylation Depends on Generation of AA Derivatives-In rabbit proximal tubule, the effect of AA on MAPK depends on the generation of an epoxy metabolite (16). The production of AA derivatives results, in most mammalian cells, from the activity of three distinct enzymatic systems, namely the lipoxygenase, cyclooxygenase, and cytochrome P450dependent epoxygenase. To evaluate the respective contribution of these pathways, MAPK activation was examined in LLC-PK1-␣ 2B treated with inhibitors prior to stimulation with UK14304. As shown in Fig. 4, the lipoxygenase inhibitor NDGA (10 M) significantly diminished the MAPK phosphorylation induced by UK14304. A decrease was also observed with the cyclooxygenase inhibitor indomethacin (50 M) but not with the epoxygenase inhibitor ketoconazole (30 M). Of note, the combined pretreatment with NDGA and indomethacin completely abolished the phosphorylation of ERK induced by UK14304 or exogenous AA. These findings strongly suggest that cyclooxy-genase and/or lipoxygenase activities are essential in the mediation of the effects of ␣ 2B -AR.
Although metabolic products are responsible for many of the indirect effects of AA, some are the direct consequence of PKC activation. To determine whether PKC participated in MAPK activation by AA, experiments were carried out in the presence of staurosporine (Fig. 5). Treatment of the cells with 200 nM staurosporine totally abolished the phosphorylation of MAPK induced by PMA, proving that the different isoforms of PKC were truly inhibited. By contrast, it did not prevent AA-induced MAPK phosphorylation. Thus, as previously found for ␣ 2 -agonists (13), the effect of AA is independent of PKC.
MAPK Phosphorylation Requires Metalloproteinase Activity and EGF-R Transactivation-Previous studies carried out on rabbit proximal tubule have shown that cell treatment with AA resulted in a significant increase of EGF-R phosphorylation and its subsequent association with Shc (19). In our model, the role of EGF-R activation was first investigated using the specific inhibitor of EGF-R tyrosine kinase activity, tyrphostin AG1478. As shown in Fig. 6A, the preincubation of LLC-PK1-␣ 2B cells in culture medium containing 100 nM tyrphostin AG1478 prevented the phosphorylation of MAPK caused by UK14304, dexmedetomidine, or AA. Thus, the transactivation of EGF-R plays a critical role in the mediation of the effect of ␣ 2B -AR on MAPK. According to recent evidence, EGF-R transactivation by G-protein-coupled receptor requires the cleavage of pro-HB-EGF by matrix metalloproteinases. Therefore, we next investigated whether MAPK phosphorylation was sensitive to the inhibitor of the matrix metalloproteinases, batimastat (Fig. 6B). Pretreatment of the cells with 5 M batimastat neither affected the basal level of MAPK phosphorylation nor inhibited the response to EGF but resulted in a blockade of the effect of UK14304 or AA. Similar results were obtained with 1,10-phenanthroline (not shown). The implication of HB-EGF release was examined using the diphtheria toxin mutant, CRM 197. Pretreatment of LLC-PK1-␣ 2B cells with CRM 197 (200 ng/ml) had no effect on the ability of exogenous EGF to activate MAPK (not shown) but strongly inhibited the effect UK14304 or AA (Fig. 6C). The release of a factor, with EGF activity and acting in an autocrine/paracrine mode, was confirmed by experiments in which the effect of conditioned medium from LLC-PK1-␣ 2B was assayed on wild-type LLC-PK1 (Fig. 7). Incubation of wild-type cells in medium collected from nonstimulated LLC-PK1-␣ 2B or their direct treatment with UK14304 did not cause any change in the extent of MAPK phosphorylation (not shown). In contrast, a clear increase was observed when medium came from LLC-PK1-␣ 2B treated with UK14304. As expected, this response was blocked by tyrphostin AG1478 but was unaffected by the addition of quinacrine or batimastat or by the prior treatment of wild-type LLC-PK1 with pertussis toxin. It is well established that activation of MAPK by EGF-R occurs via the tyrosine phosphorylation of adapter proteins such as Shc and the recruitment of Grb2-Sos complexes. Previous experiments on rat proximal tubule cells in primary  culture have shown that ␣ 2B -AR stimulation resulted in tyrosine phosphorylation of the p46 and p52 isoforms of Shc (13). Therefore, we examined whether AA induces phosphorylation of Shc in LLC-PK1-␣ 2B . As depicted in Fig. 8, treatment of the cells with AA caused a time-dependent increase in the phosphorylation of the p52 isoform of Shc.

DISCUSSION
The phosphorylation of MAPKs (ERK1/2) by ␣ 2 -agonists has been reported in a variety of cells, including Chinese hamster ovary cells transfected with the RNG gene (3) and COS or HEK 293 cells transfected with the ␣2C2 gene (7,8), which encode the rat and human ␣ 2B -AR subtypes, respectively. More recently, this effect has also been observed in rat proximal tubule cells in primary culture as well as in LLC-PK1-␣ 2B (13). In these two models, receptor stimulation resulted in an acceleration of cell proliferation, suggesting that the action of catecholamines on ␣ 2B -AR may play, in rat, a role in the adaptive response to acute renal tissue injury. Additionally, ␣ 2B -AR is known to enhance Na ϩ reabsorption as a consequence of increased activity of the Na ϩ /H ϩ exchanger, NHE3 (20). Since NHE3 was recently found to be controlled by MAPK in mouse proximal tubule cells (21), it is possible that MAPK activation is also involved in the regulation of NHE3 by ␣ 2B -AR.
The mechanisms whereby G-protein-coupled receptors activate the MAPK cascade are highly dependent upon the receptor considered and the cell type it is expressed in. Although previous studies have shown that the action of the ␣ 2B -AR is independent of receptor internalization (7,8), the signaling pathway(s) accounting for the phosphorylation of MAPK by this receptor subtype remains poorly defined. The results obtained in this study provide substantial evidence that, in LLC-PK1-␣ 2B , the activation of ERK by ␣ 2 -agonists is triggered via a mechanism comprising the activation of matrix metalloproteinases, the release of HB-EGF, and the subsequent activation of the EGF-R. This cascade was demonstrated by the following observations. First, UK14304-induced phosphorylation of MAPK is totally abrogated in the presence of the matrix metalloproteinase inhibitors (batimastat or 1,10-phenanthroline) and by prior treatment of the cells with CRM 197. Second, conditioned medium from LLC-PK1-␣ 2B cells treated with UK14304 causes activation of MAPK in wild-type LLC-PK1, even in the presence of batimastat. Third, the consequences of LLC-PK1-␣2B exposure to ␣ 2 -agonists or of wild-type LLC-PK1 exposure to conditioned medium are abolished by prior treatment of the cells with the inhibitor of EGF-R tyrosine kinase activity, tyrphostin AG1478. Previous studies of lysophosphatidic acid receptor or ␣ 2A -AR have demonstrated that the con- tribution of the EGF-R transactivation pathway is largely dependent on the cell type (10,22). In HEK 293, the major pathway of MAPK activation by ␣ 2A -AR is via the activation of Pyk2, a calcium-dependent tyrosine kinase of the focal adhesion kinase family (10). On the other hand, the effects of ␣ 2A -AR in COS cells are mediated by both EGF-R transactivation and direct recruitment of the MEK cascade by Src (9,23). According to our results, the transactivation pathway is predominant in LLC-PK1-␣ 2B ; whether it is exclusive awaits definitive demonstration.
An other major effort of the present work was to define the pathway leading from stimulated ␣ 2B -AR to transactivation of EGF-R. Because ␣ 2 -agonists activate AA release in LLC-PK1-␣ 2B and because AA is responsible for MAPK phosphorylation following angiotensin II treatment in rabbit proximal tubule cells (16), we sought to evaluate its implication. Consistent with a crucial role of the lipid second messenger, exposure of LLC-PK1-␣ 2B to exogenous AA resulted in tyrosine phosphorylation of p52 Shc and in activation of MAPK with time courses that show a striking parallel to those observed with ␣ 2 -agonists. Like for ␣ 2 -agonists, the action of AA was prevented by batimastat, 1,10-phenanthroline, CRM 197, or tyrphostin AG1478. However, a major divergence was that, unlike the effects of ␣ 2 -agonists, those of AA are resistant to pretreatment with pertussis toxin or to the addition of quinacrine. Additional support for implication of AA in the mediation of ␣ 2B -AR signal came from the study of the effects of inhibitors of AA metabolism. According to these experiments, phosphorylation of MAPK by UK14304 was strongly inhibited by NDGA and indomethacin. Ketoconazole was by contrast ineffective, indicating that AA products generated by lipoxygenase and/or cyclooxygenase, but not epoxygenase, are involved in ␣ 2 -agonist effect. The observation that AA causes MAPK phosphorylation in our model is in opposition with results obtained on LLC-PK1/C14 (24). In this clone, ERK phosphorylation was observed in response to epoxyeicosatrienoic acids but not to AA. Moreover, AA became efficient after cell transfection with an active form of cytochrome P450 epoxygenase of bacterial origin, indicating eicosanoid-dependent activation of MAPK. The reason for these discrepancies is enigmatic. However, activation of MAPK by AA was repeatedly reported in rabbit proximal tubules as well as in various cell types, including vascular smooth muscle cells and neutrophils. In vascular smooth muscle cells, efficacy of AA was dependent on its conversion into 15-hydroxyeicosa-tetraenoic acid and on PKC activation (25). Dependence on lipoxygenase and PKC activity was also found in human neutrophils (26). In this cell type, the effects of AA engaged a membrane receptor linked to G i/o proteins (27). This is not the case in LLC-PK1-␣ 2B , since neither staurosporine nor pertussis toxin treatment abolished ERK phosphorylation caused by AA. Regarding these points, LLC-PK1-␣ 2B resembles rabbit renal epithelial cells. However, it is epoxy derivatives that mediate the effects of AA on MAPK phosphorylation in these cells (16). Whereas involvement of the cytochrome P450 pathway can be excluded in our model, the respective contribution of COX and LOX is still unclear, since it is difficult to reconcile why products from either pathway could function similarly. The implication of COX activity is beyond doubt, because the effects of ␣ 2 -agonists and AA were also blocked by aspirin (not shown). By contrast, that of LOX is more questionable, since NDGA can also interfere with COX activity and act as an antioxidant. Alternatively, the possibility that prostaglandins and leukotrienes act in concert cannot be definitively ruled out. Indeed, the combined action of COX and LOX was already demonstrated to be necessary for some of the effects of angiotensin II in rat kidney and bovine bronchi (28,29). It is therefore clear that the identification of the AA metabolites responsible for MAPK activation in LLC-PK1-␣2B will require future study. In addition, the mechanism whereby these products may affect matrix metalloproteinase activity has yet to be defined. In line with the existence of a relationship between the two phenomena, constitutive expression of cyclooxygenase-2 in human colon cancer cells results in increased activation of MMP-2 (30), whereas inhibitors of PLA2 and cyclooxygenase-2 reduce the release of matrix metalloproteinases in prostate tumor cells (31).
In conclusion, our results provide evidence for a pathway by which ␣ 2B -AR activates MAPK through stimulation of PLA2, generation of AA metabolites by cyclooxygenase and/or lipoxygenase, stimulation of matrix metalloproteinases, release of HB-EGF, and transactivation of the EGF-R (Fig. 9). Whether this scenario is particular to ␣ 2B -AR in LLC-PK1 or whether it can be extended to other cell types and/or other G-proteincoupled receptors remains to be established.