Arrestin Serves as a Molecular Switch, Linking Endogenous α2-Adrenergic Receptor to SRC-dependent, but Not SRC-independent, ERK Activation*

Our previous studies have demonstrated that neither receptor endocytosis nor arrestin is required for ERK activation by the α2-adrenergic receptor (Wang, Q., Zhao, J., Brady, A. E., Feng, J., Allen, P. B., Lefkowitz, R. J., Greengard, P., and Limbird, L. E. (2004) Science 304, 1940–1944). The present studies address whether arrestin plays a role in determining the route of α2AR-evoked ERK signaling activation, taking advantage of endogenous expression of the α2AAR subtype in mouse embryonic fibroblasts (MEFs) and the availability of MEFs without arrestin expression (derived from Arr2,3–/– mice). Our data demonstrate that the endogenous α2AAR evokes ERK phosphorylation through both a Src-dependent and a Src-independent pathway, both of which are G protein dependent and converge on the Ras-Raf-MEK pathway. Arrestin is essential to recruit Src to this process, as α2AAR-mediated ERK signaling in Arr2,3–/– MEFs does not involve Src. Stimulation of α2AAR enhances arrestin-Src interaction and promotes activation of Src. α2 agonists have similar potencies in stimulating Src-dependent and Src-independent ERK phosphorylation in wild-type and Arr2,3–/– cells, respectively. However, Src-independent α2AAR-mediated ERK stimulation has both a longer duration of activation and a more rapid translocation of pERK into the nucleus when compared with Src-dependent activation. These data not only affirm the role of arrestin as an escort for signaling molecules such as Src family kinases but also demonstrate the impact of arrestin-dependent modulation on both the temporal and spatial properties of ERK activation.

Mitogen-activated protein (MAP) kinases are important effectors for G protein-coupled receptors (GPCRs) 2 to regulate cell growth, survival, and movement (1). GPCRs activate the MAPK cascades through divergent pathways involving tyrosine kinases, small GTPases, and other effectors (1). Recent studies have suggested that activation of extracellular signalregulated kinases 1 and 2 (ERK1/2) by GPCRs can occur through G protein-dependent and G protein-independent pathways and that arrestin is the essential mediator of the latter pathway (1).
Arrestin is a family of proteins with multifunctions in GPCR regulation (2,3). As originally revealed in the visual system (4 -7), arrestin associates with the agonist-evoked conformations of GPCRs, especially when the receptor is phosphorylated by G protein receptor kinases, and mediates homologous desensitization of GPCR pathways (4, 5, 8 -11). In non-sensory systems, ubiquitously expressed arrestin 2 (␤-arrestin 1) and 3 (␤-arrestin 2) also serve as adaptors for linking GPCRs to endocytosis machinery by directly binding to both the heavy chain of clathrin (12) and to ␤-adaptin (13). Depending on their association with arrestin post-endocytosis, GPCRs are classified into two families (3,14,15). Class A receptors dissociate from arrestin after internalization and are rapidly recycled back to the cell surface. By contrast, class B receptors stay associated with arrestin after internalization and are eventually subjected to lysosomal degradation.
In addition to its functions in mediating desensitization and internalization, arrestin has been implicated as scaffolding protein linking receptors to downstream signaling pathways including the ERK pathway (16 -19) in a G protein-independent manner (20). Arrestin-mediated ERK activation has been mostly demonstrated by class B receptors, and this process occurs on endocytic vesicles (17)(18)(19). Compared with G protein-dependent activation of ERK signal, which is transient and translocated into nuclei, ERK signal activated through an arrestin-dependent, G protein-independent pathway is sustained and restricted to the cytosol (20,21). Endocytosis is not always required for ERK activation (22)(23)(24)(25)(26), and arrestin also seems to be dispensable for ERK activation by some GPCRs, including the gonadotropin-releasing hormone receptor (27), the D2 and D3 dopamine receptors (24), and the ␣ 2 AR (26). The regulatory role of arrestin in ERK activation by these receptors, if any, has not been fully clarified to date.
Our previous studies have demonstrated that the ␣ 2 -adrenergic receptor (AR), a class A receptor, evokes phosphorylation of ERK1/2 through G protein-dependent pathways and this process does not require arrestin-mediated receptor endocytosis (22,26). Nonetheless, arrestin does regulate the time course of ␣ 2 AR-mediated ERK activation, and the presence of arrestin accelerates the rate of ERK activation by the ␣ 2 AR (26).
To further explore the role of arrestin in the ␣ 2 AR-mediated G protein-dependent activation of ERK, we analyzed ERK phosphorylation stimulated by endogenous ␣ 2A AR in mouse embryonic fibroblasts (MEFs) with or without (Arr2,3 Ϫ/Ϫ ) arrestin expression. Here we report that the ␣ 2 AR mediates ERK activation through Src-dependent and Src-independent pathways. Even though arrestin is dispensable for ␣ 2 AR-mediated ERK activation per se, it is required for Src kinase involvement in this process. Our data reveal that arrestin3 forms a complex with c-Src and agonist stimulation of the ␣ 2 AR promotes both arrestin-Src interaction and Src phosphorylation at Tyr-416 (representing active Src). Our findings also reveal that the Src-independent activation of ERK signaling by the endogenous ␣ 2A AR in Arr2,3 Ϫ/Ϫ MEFs is of longer duration and phospho-ERK (pERK) is more rapidly translocated into nuclei of Arr2,3 Ϫ/Ϫ cells than in wild-type (WT) cells. Cell Culture-MEFs and human embryonic kidney 293 cells were cultured in Dulbecco's modified Eagle's medium supple-mented with 10% fetal calf serum, 2 mM glutamine, 100 units/ml penicillin, and 10 g/ml streptomycin at 37°C/5% CO 2 .

Materials
Determination of Receptor Density-Endogenous ␣ 2A AR receptor density in MEFs was assayed using standard saturation binding protocols with 3 H-Rauwolscine (PerkinElmer) as described in Ref. 28 and non-linear regression analysis using Graph-Pad Prism.
Measurement of ERK Activation-ERK activation (active/total) was measured as described previously (22). Active (pERK) and total ERKs were detected by Western analysis using corresponding antibodies (Cell Signaling). For pertussis toxin treatment, 200 ng/ml was added to culture for 20 h before and during the time of stimulation of cells. For other inhibitor treatment, drugs in the amounts indicated in figure legends were added to culture for 1 h before and during the time of stimulation of cells.
Immunofluorescence of pERK in MEFs-pERK in MEFs was detected by pERK antibody (Cell Signaling) using the immunostaining protocol provided by the manufacturer. Briefly, MEFs cultured on coverslips were fixed with 3% paraformaldehyde for 30 min at 4°C followed by three washes in TBST (Trisbuffered saline (50 mM Tris-HCl, pH 7.4, 150 mM NaCl) and 0.1% Triton X-100). After one wash in TBS buffer, cells were further fixed/permeabilized with 100% MeOH for 5 min at Ϫ20°C. Following three washes with TBST, cells were blocked with 5% horse serum in TBST for 1 h at room temperature and then incubated with pERK antibody (diluted 1:400 in TBST containing 5% bovine serum albumin) overnight at 4°C. AlexaFluor-488-conjugated anti-mouse antibody (Molecular Probes) was used as the secondary antibody at 1:1000 (diluted in TBST containing 3% bovine serum albumin) for 1 h at room temperature. Fluorescent images were taken using a LSM510 confocal microscope (in the Vanderbilt University shared imaging facility).

RESULTS
Endogenously Expressed ␣ 2A ARs in MEFs Activate ERK through the G i/o Family of G Proteins-To explore the function of arrestin on endogenous ␣ 2 AR-mediated signaling, we evaluated MEFs, which express the ␣ 2A AR subtype (but not the ␣ 2B AR or ␣ 2C AR subtypes) endogenously, as revealed by reverse transcription PCR (Fig. 1A). Functional ␣ 2 AR expression in these cells was documented by [ 3 H]Rauwolscine binding (74.3 Ϯ 10.7 fmol/mg protein detected at the K D for [ 3 H]Rauwolscine; data not shown). Stimulation of endogenous ␣ 2A AR in MEFs with ␣ 2 agonists allows an assessment of ␣ 2 AR activation of ERK in native cells (Fig. 1B). For the studies in Fig. 1B, we used the endogenous ␣ 2 agonist epinephrine in the presence of propranolol to block ␤AR and prazosin to block ␣ 1 AR (see "Experimental Procedures") to stimulate the ␣ 2A AR. ERK activation by the ␣ 2A AR in MEFs can also be evoked by clonidine, UK 14,304 (data not shown), and dexmedetomidine ( Fig. 1D) and can be blocked by the ␣ 2 antagonist yohimbine (Fig. 1B).
As discussed earlier, there is evidence for GPCR activation of ERK by G protein-dependent and -independent pathways (19 -21, 32). ERK activation by ␣ 2 ARs in MEFs is eliminated by pretreatment of MEFs by pertussis toxin (Fig. 1, B and C, PTx), demonstrating that endogenous ␣ 2A AR activation of ERK in MEFs is mediated by G proteins of the G i /G o subtype, consistent with the properties of ␣ 2A AR activation of ERK in heterologous systems (33)(34)(35). Furthermore, the G protein dependence of ERK activation by endogenous ␣ 2A AR in MEFs occurs both in WT cells, which express arrestins, and in cells derived from mice null for arrestin 2 and 3 (Arr2,3 Ϫ/Ϫ ).
The data in Fig. 1, B and C, emphasize that arrestin-dependent receptor endocytosis is not a requirement for ␣ 2A AR-mediated ERK signaling, although arrestin-dependent endocytosis, followed by recycling, leads to an arrestin-dependent acceleration of ␣ 2 AR activation of ERK (26). We asked whether Total RNA was extracted from MEFs, and reverse transcription PCR was performed using primers specific for amplification of the ␣ 2A , ␣ 2B AR, or ␣ 2C AR subtype as described under "Experimental Procedures." PCR products of plasmids (plsd) containing each subtype were loaded as positive controls, and reverse transcription PCR reactions without reverse transcriptase were run as negative controls. B, stimulation of endogenous ␣ 2A ARs activates ERK signaling. WT and Arr2,3 Ϫ/Ϫ MEFs were treated with ␣ 2 AR agonist (10 Ϫ4 M epinephrine ϩ 10 Ϫ6 M propranolol to block ␤ARs ϩ 10 Ϫ6 M prazosin to block ␣ 1 ARs) in the presence or absence of yohimbine (an ␣ 2A AR antagonist) or pertussis toxin (PTx) for 5 min. For pertussis toxin treatment, cells were pretreated with 200 ng/ml pertussis toxin for 20 h before and during stimulation. Total cell lysates were separated on SDS-PAGE and blotted for pERK and total ERK as outlined under "Experimental Procedures." C, endogenous ␣ 2A AR-evoked ERK activation is pertussis toxin sensitive. ERK activity (active/total) was quantitated from five independent experiments, and percent increase of ERK activity over basal (ϮS.E.) was plotted using GraphPad Prismா. *, p Ͻ 0.01. D, ERK phosphorylation in response to increasing doses of ␣ 2 AR agonists in WT versus Arr2,3 Ϫ/Ϫ MEFs. Top, representative blots for pERK following 5 min of treatment of ␣ 2 AR agonist dexmedetomidine (Dex) at indicated doses. Bottom, dose response curves of ERK activation by Dex in WT versus Arr2,3 Ϫ/Ϫ MEFs. ERK activation (active/total) was quantified from five independent experiments and plotted using GraphPad Prism. The EC 50 values for Dex to induce ERK phosphorylation in WT and Arr2,3 Ϫ/Ϫ MEFs are 8.4 ϫ 10 Ϫ9 M and 4.0 ϫ 10 Ϫ8 M, respectively. arrestins would alter the potency of ␣ 2 AR agonists in activating ERK as a result of the rapid regeneration of native ␣ 2A AR in arrestin-expressing cells. As shown in Fig. 1D, the EC 50 of the ␣ 2 AR agonist/partial agonist dexmedetomidine in activating ERK in Arr2,3 Ϫ/Ϫ MEFs (4.0 ϫ 10 Ϫ8 M) is comparable with that in WT cells (0.84 ϫ 10 Ϫ8 M). In addition, the maximal activation of ERK by dexmedetomidine in Arr2,3 Ϫ/Ϫ MEFs is similar to that in WT cells (Fig. 1D). A similar potency for ERK activation in WT versus Arr2,3 Ϫ/Ϫ MEFs was also observed in response to the ␣ 2 AR partial agonist UK 14,304 (data not shown). We chose partial agonists for these studies as they are most sensitive to revealing difference in coupling efficiency between receptors and effectors (36,37). These data indicate that the presence of arrestins, although accelerating the rate of ERK activation in MEFs (Fig. 2, below, and Ref. 26), does not significantly change the potency or the efficacy of ␣ 2 agonists in activating ERK.
Lack of Arrestins Alters the Temporal Pattern of ␣ 2A AR-elicited ERK Activation-Although the data in Fig. 1, B-D, indicate that arrestins are not required for ␣ 2A AR activation in MEFs, we nonetheless wanted to explore whether arrestin had any regulatory impact on endogenous ␣ 2A AR-evoked, G proteindependent ERK signaling. We noted, as shown in Fig. 2, that arrestin alters both the time of onset and the duration of ␣ 2A AR-evoked activation of ERK in MEFs. In WT cells, endogenous ␣ 2A AR-evoked ERK activation peaks at 2 min and then declines, due to desensitization of receptor-mediated signaling. Virtually no stimulation is detected after 20 min of treatment with ␣ 2 AR agonists in WT MEFs that express arrestin (Fig. 2). In cells without arrestin expression, the time to peak stimulation appears to be delayed (Fig. 2B). In addition, the desensitization of ␣ 2A AR-evoked ERK signal is less rapid and occurs to a lesser extent; thus, ϳ40% of ERK activity remains detectable after 20 min of treatment with the ␣ 2 agonist (Fig. 2). This extended ERK activation in Arr2,3 Ϫ/Ϫ cells lasts for another 3 h in the presence of agonist before becoming fully desensitized  (data not shown). The impaired desensitization of ␣ 2A AR signaling seen in Arr2,3 Ϫ/Ϫ cells is consistent with the well appreciated roles of arrestins in the receptor desensitization process (38).
Enrichment of pERK in the Nucleus following Agonist Treatment Is Independent of Arrestin Expression-We examined the morphological distribution of pERK following ␣ 2 agonist activation of MEFs. Detection of pERK in WT MEFs first occurs in cytosol (see the 1-min time point in Fig. 3), but pERK becomes enriched in nuclei after agonist stimulation for 10 min (Fig. 3). In Arr2,3 Ϫ/Ϫ MEFs, the ␣ 2A AR-evoked pERK signal is highly enriched in nuclei at all time points examined, including the earliest time point measured, i.e. 1 min (Fig. 3). Whether pERK is activated in the nucleus, without required cytosolic to nuclear transport, is not known, but cytosolic pERK was barely detectable at the earliest time points in our studies of Arr2,3 Ϫ/Ϫ MEFs.
Endogenous ␣ 2A AR-evoked ERK Activation Is Ras-Raf Dependent Both in the Presence and Absence of Arrestins-It has been reported that the ␣ 2 AR activates ERK signaling through a G␤␥-and a Ras-Raf-dependent pathway (39). Here we examined whether this is also true in cells without arrestin expression. As shown in Fig. 4, inhibitors of Ras (Ftase inhibitor I) and Raf (Raf inhibitor II) blocked ϳ80% of ERK activation elicited by endogenous ␣ 2A AR in both WT and Arr2,3 Ϫ/Ϫ cells, indicating that the ␣ 2A AR still activate ERK through the G protein-Ras-Raf pathway even without arrestin expression. It has been reported that G␤␥ activation of Ras is sensitive to phosphoinositide 3 kinase (PI3K) inhibitors (40,41), suggesting that PI3K can serve as an intermediate linking GPCRs to ERK signaling. Therefore, we examined the effect of PI3K inhibitor treatment in ERK activation mediated by the endogenous ␣ 2A AR in MEFs. The PI3K inhibitor LY294002 failed to alter ␣ 2A AR-evoked ERK activation in either WT or Arr2,3 Ϫ/Ϫ MEFs (Fig. 4, A and B), although control studies confirmed that LY294002 was active, as it blocked serum-induced phosphorylation of Akt (a known downstream target of PI3K) in parallel experiments (Fig. 4, C and D). These data indicate that PI3K is not involved in this process.
Lack of Arrestins Eliminates Sensitivity of ␣ 2A AR Activation of pERK to Src Family Kinase Inhibitors-Because arrestins have been reported to serve as a scaffold for the ␤ 2 AR, MEK, ERK, and Src (16), we explored whether activation of Src kinase is a critical link between ␣ 2A AR activation and ERK phosphorylation and whether arrestin is necessary for this role of Src, if B, summary of quantitative data. ERK signals (active/total) were quantitated from five independent experiments, and the percent increase of ERK activity over basal (ϮS.E.) was plotted using GraphPad Prismா. *, p Ͻ 0.01. C and D, LY294002 blocks serum-induced Akt activation but has no effect on ␣ 2A ARevoked ERK activation in parallel experiments. WT MEFs were pretreated with or without 50 mM LY294002 for 1 h and then stimulated with ␣ 2 agonist (10 Ϫ4 M epinephrine ϩ 10 Ϫ6 M propranolol ϩ 10 Ϫ6 M prazosin) for 5 min (C ) or with serum for 15 min (D) in the presence or absence of LY2994002. Total cell lysates were blotted for pERK and total ERK (C ) or for pAkt and total Akt (D). it occurs. The overall involvement of a tyrosine kinase (or kinases) in ␣ 2A AR activation of ERK in WT cells is evident in Fig. 5, where we observed elimination of agonist-stimulated ERK phosphorylation in the presence of the general tyrosine inhibitor genistein. This involvement of tyrosine kinases is not due to transactivation of the epidermal growth factor (EGF) receptor tyrosine kinase, because ␣ 2A AR activation was insensitive to inhibition by the EGF receptor kinase inhibitor, AG1478 (Fig. 5), whereas this agent completely eliminated effects of EGF on ERK phosphorylation in parallel control experiments (data not shown). For WT MEFs, the Src family tyrosine kinase inhibitor PP2 eliminated the ␣ 2A AR activation of ERK (Fig. 5), suggesting that sensitivity to genistein may be accounted for by the role of Src family kinase in linking ␣ 2A AR activation to ERK phosphorylation. The involvement of Src kinase in ␣ 2A AR-mediated ERK activation is confirmed by the fact that expression of a kinase-inactive dominant negative Src (SrcK297R) (42) remarkably reduced ERK phosphorylation following ␣ 2A AR stimulation (Fig. 6). However, a different pathway appears to link the ␣ 2A AR to ERK phosphorylation in cells lacking arrestin expression, because stimulation of p-ERK accumulation in Arr2,3 Ϫ/Ϫ MEFs is independent of either genistein or Src kinase inhibitors (Fig. 5). Thus, in the absence of arrestin expression, the ␣ 2A AR appears to engage an additional signaling pathway to phosphorylate ERK, a Src kinaseindependent pathway.
We further examined ␣ 2A AR-mediated ERK activation in cells lacking expression of the Src family kinases using Src-null SYF cells derived from Src, Yes, and Fyn triple knock-out mouse embryos. Consistent with the above findings, the ␣ 2A AR activated ERK in SYF cells, affirming its ability to do so in a Src-null, thus Src-independent, environment (Fig. 7A). Interestingly, similar to findings in Arr2,3 Ϫ/Ϫ MEFs, the ␣ 2A AR-evoked pERK signal in SYF cells is highly enriched in nuclei at the earliest time point of measurement (Fig. 7B). In contrast, when c-Src is reconstituted in SYF cells (i.e. SYFϩc-Src cells), ␣ 2A ARevoked pERK signal is mainly distributed in cytosol at this early 2-min time point (Fig. 7B).
Agonist Stimulation of the ␣ 2A AR Enhances Arrestin3 Interaction with c-Src as Well as Src Activation-Our data suggest that arrestin may serve as a scaffold linking Src kinase to ␣ 2A AR-evoked ERK signaling pathways. Direct interaction between c-Src and arrestin has been reported previously (16,43). Here we examined whether Src and arrestin interact under our experimental conditions and whether activation of ␣ 2 AR has any impact on this interaction. As shown in Fig. 8, agonist occupancy of the ␣ 2A AR remarkably enriched the amount of arrestin3 in the HA-Src immunoisolated complex by 3.1-fold, indicating that activation of ␣ 2A AR enhances formation of the arrestin-Src complex. Moreover, the amount of active Src (detected by antibody against p-Tyr-416) in cells was increased  following agonist stimulation, even though the amount of total Src (detected by the anti-HA antibody) is not changed (Fig. 8). These data strongly suggest that stimulation of ␣ 2A AR promotes arrestin-Src interaction and Src activation.

DISCUSSION
Arrestin regulates GPCR-mediated signaling in a variety of ways. Arrestin binding to activated and phosphorylated receptors uncouples G proteins from these receptors, thus desensitizing G protein-dependent signaling (2). Arrestin also is required for GPCR internalization (12,13), which leads either to receptor degradation or to recycling (3,14,15). Arrestin-dependent endocytosis and subsequent recycling enhances the initial rate of GPCR-mediated signaling, at least in some settings (26). Furthermore, arrestin can serve as a scaffold to directly link receptors to downstream signaling effectors in a G protein-independent manner (20). Our current studies demonstrated that in the process of G protein-dependent ERK activation by the ␣ 2 AR, arrestin not only is critical for desensitization of this process (Fig. 2) but also is required for involvement of the Src family of tyrosine kinases in this event (Figs. 5 and 9).
As shown schematically in Fig. 9, ERK stimulation by the ␣ 2 AR requires the G i / o subfamily of G proteins. With arrestin present, G proteins activate Src tyrosine kinase, which subsequently activates the Ras-Raf pathway and results in ERK phosphorylation. In this process, arrestin serves as a scaffold recruiting Src to the plasma membrane for activation by G proteins (Figs. 8 and 9), by analogy to its role in ERK activation by ␤ 2 AR (16). In contrast to ␤ 2 AR-mediated ERK activation, which appears to require arrestin-dependent endocytosis via clathrincoated pits (44), ␣ 2 AR-evoked ERK activation does not require endocytosis (22,26). Furthermore, even without arrestin expression (in Arr2,3 Ϫ/Ϫ MEFs), ␣ 2 AR agonists are still able to evoke ERK phosphorylation and do so with a potency similar to that observed in WT cells (Fig. 1D). Arrestin-independent ERK activation occurs through a G protein-dependent but Src-independent pathway that requires Ras activation (Figs. 4, 5, and 9). Therefore, our data reveal that arrestin serves as a molecular switch determining the signal transduction route of ERK activation by the ␣ 2 AR.
Our data also demonstrate an important role of arrestin in modulating the temporal and spatial profiles of ␣ 2 AR-mediated ERK activation. In cells without the arrestin-mediated desensitization process, the duration of the ␣ 2 AR-evoked pERK signal is considerably extended (Fig. 2), lasting up to 5 h (data not shown). In Arr2,3 Ϫ/Ϫ cells, ␣ 2 AR-evoked phosphorylation of ERK is detected principally in the nucleus, without detectable prior phosphorylation in the cytosol (Figs. 3 and 9), indicating a role of arrestin in retaining ERK localization in the cytosol. The ERK signaling pathway is critical for multiple cellular processes, including proliferation, differentiation, survival, and migration, and the duration and subcellular localization of pERK are the major factors determining what cellular response active ERK signaling would trigger (45,46). Therefore, the changes in duration and localization of ␣ 2 AR-evoked pERK due to absence of arrestin observed in this study may result in significant alteration of ␣ 2 AR-mediated cellular responses. For example, the sustained nuclear-localized pERK in the absence of arrestin may lead to transcription of certain responsive genes and sub- FIGURE 8. Agonist stimulation of the ␣ 2A AR promotes arrestin-Src interaction and Src activation. CosM6 cells cotransfected with cDNA encoding ␣ 2A AR, HA-Src, and GFP-arrestin3 were treated with or without 100 M epinephrine for 5 min at 37°C in the presence of 1 mM propranolol as indicated. Lysates were prepared as described under "Experimental Procedures." Top, proteins immunoprecipitated by protein G-agarose-rat anti-HA antibody were separated by SDS-PAGE and transferred to nitrocellulose membrane. Membrane was blotted with anti-GFP antibody to detect GFP-arrestin, with anti phospho-Src (Tyr 416) antibody to detect active Src, and with anti-HA antibody to detect total Src. Bottom, total lysates were separated, transferred, and blotted for GFP-arrestin and HA-Src. FIGURE 9. Postulated model for ERK activation by the ␣ 2 AR in the presence and absence of arrestin. Stimulation of the ␣ 2 AR evokes ERK signaling through the G i/o subfamily of G proteins. Arrestin plays multiple functions in this process. First, arrestin is important for desensitizing the ␣ 2 AR-evoked ERK phosphorylation, presumably by competing for receptor-G protein interactions upon receptor phosphorylation (not shown). Arrestin also mediates receptor endocytosis and permits receptor recycling, thus accelerating agonist-mediated simulation (26). Third, arrestin serves as a molecular switch determining whether or not Src family kinase is involved in the signal transduction pathway of this process. In the presence of arrestin, the Src family of tyrosine kinases is recruited and activated in a G protein-dependent manner, and this may result from direct interactions of arrestin and Src kinase (16). Subsequently, Src family kinase activates Ras, which then activates the Raf-MEK-ERK cascade. The presence of arrestin also delays the appearance of pERK in the nucleus, perhaps also due to direct arrestin-pERK interactions that tether pERK in the cytosol, at least temporarily. In the absence of arrestin, the ␣ 2 AR-activated G proteins still activate the Ras-Raf pathway but in a Src-independent manner and with virtually instantaneous appearance of pERK in the nucleus.
sequently activate cellular response pathways mediated by these genes.
Our data also eliminate PI3K and transactivation of EGF receptor tyrosine kinase as contributors to ␣ 2 AR-mediated ERK activation in native cells. Our present findings revealed that ␣ 2A AR-evoked ERK phosphorylation in MEFs is not altered by PI3K inhibitor LY290004 (Fig. 4), indicating that PI3K is not involved in the ␣ 2A AR-stimulated G protein-Ras-Raf signaling pathway. This is in contrast to previous findings that PI3K is employed by some GPCRs such as dopamine D2 and D3 receptors to activate ERK (24). Similarly, although some GPCRs activate ERK by transactivation of receptor tyrosine kinases, such as the EGF receptor (32), our findings revealed that ␣ 2 AR activation of ERK signaling is not significantly altered by the EGF receptor inhibitor AG1478 in either WT or Arr2,3 Ϫ/Ϫ MEFs (Fig. 5).
Taken together, our findings provide the first evidence of the role of arrestin as a determinant of Src kinase-dependent versus Src kinase-independent activation of p42/44 ERK by endogenous ␣ 2A AR despite the absence of an obligatory role of arrestin in linking ␣ 2A AR to ERK activation. These findings add arrestin to the list of "molecular switches" in GPCR signaling. Furthermore, the data demonstrate a possible role for arrestin and Src kinase-dependent pathway of ERK activation in dictating the rate and extent of nuclear localization of pERK.