MEK 1 Binds Directly to Arrestin 1 , Influencing Both Its Phosphorylation by ERK and the Timing of Its Isoprenaline-stimulated Internalization *

Dong Meng, Martin J. Lynch, Elaine Huston, Michael Beyermann, Jenny Eichhorst, David R. Adams, Enno Klussmann, Miles D. Houslay, and George S. Baillie From Neuroscience and Molecular Pharmacology, Faculty of Biomedical and Life Sciences, Wolfson Building, University of Glasgow, Glasgow G12 8QQ, Scotland, United Kingdom, the Leibniz-Institut für Molekulare Pharmakologie, Campus Berlin-Buch, Robert-Rössle-Strasse 10, 13125 Berlin, Germany, and the Department of Chemistry, Heriot-Watt University, Riccarton Campus, Edinburgh EH14 4AS, Scotland, United Kingdom

The ␤arrestins are multifunctional signal scaffolding proteins that play a pivotal role in the desensitization process that regulates the functioning of many key heptahelical G proteincoupled receptors (GPCRs) 2 (1,2). The ␤ 2 -adrenergic receptor (␤ 2 -AR) has provided a critical functional paradigm in elucidating this fundamental process, where agonist occupancy triggers its phosphorylation by G-protein-coupled receptor kinase, thereby initiating the recruitment of cytosolic ␤arrestins (3,4). By associating with agonist-occupied receptors, ␤arrestins attenuate GPCR functioning by both regulating interaction with signaltransducing G-proteins and facilitating GPCR internalization, leading to either recycling or degradation of the targeted receptor (5). Indeed, the interaction of receptor-recruited ␤arrestin with clathrin cages provides a key part of the paradigm for the deactivation of select ligand-bound GPCRs (6). ␤arrestins can also deliver sequestered cAMP phosphodiesterase-4 isoforms, particularly PDE4D5, to the site of cAMP synthesis associated with the ␤ 2 -AR, thereby contributing a key part of the cellular desensitizing system for cAMP (5,7,8).
In the resting state, cytosolic ␤arrestin1 proteins are constitutively phosphorylated by extracellular signal-regulated kinase (ERK) at Ser 412 , located within their distal C terminus (9,10). The agonist-stimulated recruitment of ␤arrestin to GPCRs, such as the ␤ 2 -AR, leads to the dephosphorylation of ␤arrestin1 at this site. This event acts as a molecular switch, allowing for the internalization of the ␤ 2 -AR-sequestered ␤arrestin complex. Thus, ERK-phosphorylated ␤arrestin1 is unable to associate with clathrin cages, whereas this constraint is removed upon its dephosphorylation (9,11). Dephosphorylation of Ser 412 is also thought to be a determinant for the association of ␤arrestin1 with c-Src and the phosphorylation of dynamin, a key feature in receptor internalization (12). Thus, the ERK2-dependent Ser 412 phosphorylation and dephosphorylation of ␤arrestin1 provides a pivotal molecular switch that determines the association of ␤arrestin1 with the endocytic machinery governing internalization of the ␤ 2 -AR (11).
The activation of ERK critically depends upon its phosphorylation by the MAPK kinase, MEK1. Compartmentalization and fidelity of this action is endowed by the ability of these proteins to interact and dock to each other, where a motif called the CD domain has been shown to play an underpinning role (13). In particular, MEK1 is known to bind to a negatively charged cluster that consists of two aspartate residues separated by any other two amino acids (DXXD) (14). The functioning of MEK1 has been shown to be integral to many cellular processes, such as transcription regulation, proliferation, and differentiation. Therapeutically, inhibitors of MEK1 have been developed as potential therapeutics for cancer and, more recently, have become the focus of development for treating chronic inflammatory disorders, such as rheumatoid arthritis and asthma (15)(16)(17). It has previously been suggested that MEK may form a complex with ERK and with ␤arrestin (10, 18) that creates a "signalosome" capable of disseminating MAPK signals to defined intracellular compartments. In other studies, it has been suggested that constitutive ␤arrestin-ERK1 complexes probably recruit MEK1 through its binding to ERK as a consequence of active Ras-dependent signaling (19 -23). However, a recent report suggests that MEK can bind directly to ␤arrestin1 (24) within sites on both N and C domains. Critically, however, the exact nature of MEK interaction with ␤arrestin1 is unclear. Given that ␤arrestin1 is ERK-phosphorylated in the cytosol (9) and the critical importance of this event to ␤arrestin-mediated GPCR internalization, the nature of MEK1 interaction with ␤arrestin warrants investigation. Here then, we evaluate MEK1 interaction with the signal scaffolding protein, ␤arrestin1, demonstrating that MEK1 binds directly to ␤arrestin1, allowing the identification of a functionally active peptide disruptor of this complex.
Cell Culture and Drug Additions-HEK293 cells and HEK293 cell lines stably overexpressing FLAG-tagged ␤ 2 -AR-GFP (HEKB2 cells) were cultured as described previously (25). Transfection of these cells was done using Polyfect (Qiagen), following the manufacturer's instructions. Pretreatments for control experiments were done 10 min (UO126, 10 M) or 2 h (cell-permeable peptides, 10 M) prior to the addition of isoproterenol (10 M).
Preparation of Peptide Small Molecule Disruptors-Briefly, all stearyl-peptides were prepared by treatment of the peptide resins with stearic acid and diisopropylcarbodiimide in the presence of N-hydroxybenzotriazole, followed by deprotection using trifluoroacetic acid/ethanedithiol (26).
Fluorescent peptides were prepared by adding 2 mg of 5,6carboxyfluorescein N-hydroxysuccinimide ester (in 0.9 ml of DMSO, 100 l of 0.1 M NaHCO 3 , pH 8.5) to 10 mg of peptide to be labeled. After reaction at room temperature in darkness for 20 h, the reaction mix was directly applied to preparative high pressure liquid chromatography. Fractions containing labeled peptide were collected and lyophilized.
Peptide Arrays and Alanine Scans-␤arrestin1 and MEK1 peptide libraries were produced by automatic SPOT synthesis as described previously (27). They were synthesized on continuous cellulose membrane supports on Whatman 50 cellulose membranes using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry with the AutoSpot-Robot ASS 222 (Intavis Bioanalytical Instruments AG, Köln, Germany). Alanine-scanning peptide libraries were constructed by taking the residues in positive spots and sequentially changing each residue to alanine (or, if an alanine was the natural amino acid at that position, to aspartate). The interaction of spotted peptides with purified, recombinant GST and GST-␤arrestin1 and GST-MEK1 fusion proteins was determined by overlaying the cellulose membranes with 10 g/ml recombinant protein. Bound recombinant proteins were then detected following wash steps with rabbit anti-GST, and detection was performed with a secondary anti-rabbit horseradish peroxidase-coupled antibody.
Expression of GST Fusions in Escherichia coli-Cultures of E. coli JM109 containing pGEX-␤-arrestin1 or pGEX were induced with 1 mM isopropyl-␤-D-thiogalactopyranoside (Roche Applied Science) for 4 h at 30°C. Bacteria were harvested by centrifugation at 6,000 ϫ g for 15 min at 4°C, and the bacterial pellet was frozen at Ϫ80°C overnight. The bacterial pellets were resuspended in 10 ml of ice-cold resuspension buffer (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, 10 mM ␤-mercaptoethanol, and complete protease inhibitor mixture) and sonicated with four 30-s bursts at the maximal setting. Triton X-100 was added to a final concentration of 0.02%, and cell debris was then removed by centrifugation at 15,000 ϫ g for 10 min at 4°C. The cleared supernatant was incubated with onetenth volume of pre-equilibrated glutathione-Sepharose beads on an orbital shaker for 30 min at 4°C. The beads were collected by centrifugation at 13,000 ϫ g for 1 min and washed three times with ice-cold resuspension buffer. The fusion proteins were eluted by the addition of 5 mM glutathione, 50 mM Tris-HCl, pH 8.0, and dialyzed three times against 50 mM Tris-HCl, pH 8.0, 100 mM NaCl, and 5% glycerol. The purified fusion proteins were stored at Ϫ80°C until required.
Microscopy-HEKB2 cells were seeded onto poly(L-lysine)treated coverslips at ϳ20% confluence. After treatment with indicated ligands, cells were fixed for 10 min in 4% (w/v) paraformaldehyde followed by three washes with Tris-buffered saline (150 mM NaCl, 20 mM Tris, pH 7.4). The coverslips were mounted to microscope slides with Immunomount. Cells were visualized using the Zeiss Pascal laser-scanning confocal microscope (Zeiss, Oberkochken, Germany). Fluorescent peptide entry into cells was done using an LSM510 laser-scanning microscope (Zeiss).
In Vitro Pull-down Using Purified Proteins-1 nmol of purified GST or GST-␤arrestin1 was mixed with an equal amount of HisMEK1 (Millipore) in 0.5 ml of binding buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM MgCl 2 , 1 mM dithiothreitol, 0.5% Triton X-100). The mixture was incubated for 1 h at 4°C, and then 50 l of anti-polyhistidine-agarose beads (A5713; Sigma) were added in for an overnight incubation. Beads were collected by 10,000 ϫ g av centrifugation for 1 min and washed three times with binding buffer before loading to an SDS-polyacrylamide gel for protein separation.
Western Blotting and Protein Estimation-Immunoblotting was done as previously described (5), using 25-50 g of cellular protein/well. After treatment, HEK␤2 cells or HEK cells were washed twice with PBS before being scraped into 3T3 lysis buffer (25 mM HEPES, 2.5 mM EDTA, 50 mM NaCl, 50 mM NaF, 30 mM sodium pyrophosphate, 10% glycerol, 1% Triton X-100, pH 7.5) with added protease inhibitors (Complete Protease Inhibitor Mixture; Roche Applied Science). Proteins were separated by PAGE and transferred to nitrocellulose for Western blotting. Protein concentrations of cell lysates were determined with bovine serum albumin as a standard using Bradford reagent as previously described (25). Protein was routinely measured by the method of Bradford using bovine serum albumin as a standard.
Immunoprecipitation of Target Molecules-This was done as described previously by us (29,30). Briefly, detergent-soluble proteins were isolated from cells by disruption in lysis buffer (1% (v/v) Triton X-100, 50 mM HEPES buffer, pH 7.2, 10 mM EDTA, 100 mM NaH 2 PO 4 , 2H 2 O) containing complete protease inhibitor mixture (Roche Applied Science) to 8% volume. Detergent-insoluble proteins were removed by centrifugation at 10,000 ϫ g av for 10 min, and the soluble fraction was retained. Equal volumes of cell lysate containing 500 g of protein were cleared by incubation with 30 l of preimmune serum and/or 30 l of protein A slurry. The beads were then removed by centrifugation at 10,000 ϫ g av for 10 min at 4°C, and cleared lysate was incubated at 4°C for 2 h with constant agitation with a volume of antiserum. Immunoglobulins were then isolated by incubation with protein A-coated Sepharose beads for 1 h before retrieval by refrigerated centrifugation at 10,000 ϫ g av for 5 min. Target molecule-immunoglobulin conjugates attached to the beads were then washed in phosphatebuffered saline (PBS) three times.
Internalization of ␤ 2 -AR Assay-Briefly, cells were grown to 80% confluence in Petri dishes and treated with 0.3 mg/ml disulfide-cleavable biotin (Pierce) in PBS at 4°C for 30 min (5 ml/well). Cells were then washed with precooled PBS three times to cease biotinylation. The cells were then treated with the appropriate ligands, with or without peptides for the indicated times. Samples were washed with PBS two times, and the biotinylated receptors were stripped with stripping buffer (0.05 M glutathione, 0.3 M NaCl, 0.075 M NaOH, 1% FBS/newborn calf serum in PBS) at 4°C for 30 min. Cell extracts were resuspended in radioimmune precipitation buffer (150 mM NaCl, 25 mM KCl, 10 mM Tris-HCl, and 0.1% Triton X-100, pH 7.4), and biotinylated receptors were immunologically isolated using immobilized NeatrAvidin (Pierce) beads at 4°C overnight. Samples were then washed three times with radioimmune precipitation buffer or PBS, and samples were analyzed via SDS-PAGE. Total receptor amounts were visualized and quantified using anti-␤ 2 -AR antiserum (sc-569; Santa Cruz Biotechnology).
Analysis by Microscopy-Slides were examined using a fluorescent imaging microscope at a magnification of ϫ43 and imaged for phase contrast, GFP fluorescence, and 4Ј,6-dia-FIGURE 1. MEK1 binds directly to N domain of ␤arrestin1 peptide array. a, immobilized peptide spots of overlapping 25-mer peptides, each shifted by 5 amino acids, covering the entire sequence of ␤arrestin1 were probed for interaction with either GST-MEK1 or GST alone by immunoblotting. Positively interacting peptides are represented by dark spots. Spot 6 covering the sequence Asp 26 -Glu 50 produced the greatest interaction with GST-MEK1 and was selected for alanine-scanning analysis. b, each amino acid from spot 6 was sequentially and individually substituted with an alanine residue and overlaid with GST-MEK as in a. Control represents an identical spot to spot 6, and other spots represent mutants where one indicated residue is substituted with alanine, apart from spots A26 -30 and A46 -50, wherea stretch of 5 amino acids is substituted with alanine. midino-2-phenylindole staining. Images of three random fields of view were taken from each slide, and all cells within these areas were quantified. After correction for background, areas were outlined to define the intracellular compartment of the cell, using phase contrast to define the plasma membrane, and the number of cells present in each field of view was quantified using 4Ј,6-diamidino-2-phenylindole staining of the nuclei of the cells. The image analysis software, Metamorph 7.0, was used to find spots within the intracellular compartment of the cell, and the number of spots per cell was quantified and used as an indication of the internalization of the ␤ 2 -adrenoreceptor. Student's t test was used for statistical analysis.

RESULTS
Purified MEK1-GST Binds Directly to the ␤Arrestin Peptide Array-MEK1 is a 45-kDa MAPK kinase that activates ERK in a classic amplification cascade. Previous studies (10) utilizing multiple transfection of key intermediates have indicated that MEK1 is able to interact with ␤arrestin. However, it is unclear from these reports as to whether this occurs directly or involves a bridging molecule, such as ERK, which is well known to bind to MEK and has been shown to associate directly with ␤-arrestin (19,31,32). Recently, evidence has emerged suggesting that MEK may bind to both the N and C domains of ␤arrestin1, although the exact nature of the binding sites remains to be determined (24).
In order to determine whether MEK1 has the potential to interact directly with ␤arrestin, we first employed peptide array analysis. This provides a novel and powerful technology for gaining insight into the basis of specific protein-protein interactions. Indeed, we have used this to considerable advantage to exhaustively map sites of interaction between both of the scaffolding proteins, ␤arrestin (27) and RACK1 (7), with the cAMPhydrolyzing PDE4D5 isoform. These were confirmed by mutagenesis analyzed through both pull-down and two-hybrid studies (25). Here we have generated a library of overlapping peptides (25-mers), each shifted by 5 amino acids, which spans the entire sequence of ␤arrestin1. These were spot-synthesized on cellulose membranes to generate an immobilized peptide library that was then probed with a purified, recombinant GST fusion protein of active MEK1. Binding of MEK1 to individual peptide spots was assessed immunologically, with positive interactions identified as dark spots (Fig. 1a). A cluster of peptides (spots 3-6) was observed for GST-MEK1 binding but not GST alone, and that yielding the strongest signal (spot 6; Fig. 1a) was evaluated further.
In order to gain insight into the amino acids involved in allowing MEK1 to bind to this region of ␤arrestin1, we generated a family of peptides derived from the "spot 6" 25-mer parent peptide whose sequence reflected amino acids Asp 26 -Glu 50 of ␤arrestin1. The 25-mer peptide progeny of this parent peptide each had a single substitution, to alanine, of successive amino acids in the sequence to form a scanning peptide array (Fig. 1b). Using this analysis, we observed that MEK1 binding to the 25-mer Asp 26 -Glu 50 ␤arrestin peptide was ablated upon alanine substitution of Asp 26 and Asp 29 (Fig. 1b). Binding of MEK1 was also ablated when the various amino acids in the region Asp 26 to His 30 were all replaced with alanine residues but not when the run of amino acids from Tyr 46 to Glu 50 were similarly substituted with alanine residues.
To reinforce our findings from peptide array analysis, we attempted to show that MEK1 and ␤arrestin1 could directly interact in vitro. After mixing purified fusion proteins of His-MEK1 with GST-␤arrestin1 or GST alone, we pulled down the His tag on MEK1 and evaluated co-immunoprecipitating species by Western blotting (Fig. 2a). GST-␤arrestin1, but not GST, co-purified with His-MEK1, confirming the notion that the MEK1-␤arrestin1 association is direct and is not dependent on other putative members of the signaling complex. In order to determine if this DXXD motif does indeed play an important role in defining the interaction between MEK1 and ␤arrestin, we mutated Asp 26 and Asp 29 to alanine in a FLAG epitopetagged ␤arrestin1 construct and evaluated whether this altered the ability of MEK1 to co-immunoprecipitate with ␤arrestin in cell lysates where they are co-expressed (Fig. 2b). Doing this, we see that although anti-FLAG immunoprecipitates of FLAGtagged ␤arrestin1 pulled down endogenous MEK1, only 8.3 Ϯ 3.9% (mean; n ϭ 3) of MEK remains associated with the FLAGtagged D26A/D29A-␤arrestin1 mutant construct (Fig. 2b). There is specificity in this interaction, since both the wild-type and D26A/D29A mutant forms of FLAG-tagged ␤arrestin1 acted to pull-down equal amounts of PDE4D (Fig. 2), whose binding site on ␤arrestin does not involve either Asp 26 or Asp 29 FIGURE 2. Mutation of ␤arrestin1 Asp 26 and Asp 29 to Ala ablates MEK1 binding to ␤arrestin1. a, fusion proteins of His-MEK1 and GST-␤arrestin1 or GST alone were mixed before His-MEK was pulled down using anti-His-agarose beads. Immunopurified samples of His-MEK1 were subjected to Western blot analysis for the presence of both GST and His tags. b, HEK293 cells were transfected with FLAG-␤arrestin1 or a mutant form of ␤arrestin1 where Asp 26 and Asp 29 had been dually mutated to alanine. Both forms of ␤arrestin1 were immunopurified using the FLAG epitope, and preparations were immunoblotted for MEK, ERK, and PDE4D. The lower panel shows transfection efficiency of the ␤arrestin1 constructs. Ab, antibody. (27). These analyses confirm the peptide array studies in indicating that both Asp 26 and Asp 29 play a key role in underpinning the interaction of MEK1 with ␤arrestin1.
␤Arrestin has also been shown to interact with ERK (10). Since MEK1 can interact with ERK, it has been postulated that ERK may act as an adaptor that, in binding to ␤arrestin, sequesters MEK1 there. We thus set out to determine if loss of MEK1 association with FLAG-tagged D26A/D29A-␤arrestin1 also led to loss of association of ERK with ␤arrestin1 (Fig. 2). Doing this, we see quite clearly that this is not the case (Fig. 2). Indeed, if anything, we noted an increase in association of ERK with the D26A/D29A-␤arrestin1 mutant that cannot bind MEK1, compared with wild-type ␤arrestin1 (Fig. 2).
Probing a MEK1 Peptide Array with ␤Arrestin1-In order to gain insight into how MEK1 may bind to ␤arrestin1, we probed a MEK1 peptide array with a GST-␤arrestin1 fusion protein.
The ␤arrestin1 fusion protein, but not GST alone, appeared to bind to a single MEK1 peptide (spot 10), which included amino acids Gln 46 -Lys 70 (Fig. 3a). Interestingly, this region contained a cluster of positively charged amino acids, namely Arg 47 , Lys 48 , and Arg 49 . Substitution of Lys 48 and Arg 49 for alanine attenuated ␤arrestin1 binding, whereas alanine substitution of either Arg 47 and Lys 48 or all three basic residues (Arg 47 , Lys 48 , and Arg 49 ) for alanine ablated the interaction of this mutant form of ␤arrestin1 with MEK1. In contrast, substitution of an acidic cluster, Asp 65 , Asp 66 , and Asp 67 , had little effect (Fig. 3b).
In order to explore whether this cluster of positively charged amino acids in MEK1 is involved in the binding of ␤arrestin1, we generated a triple mutation of HA epitope-tagged MEK1, namely R47A/K48A/R49A-MEK1. In cells transfected with both FLAG-tagged ␤arrestin1 and HA-tagged MEK1, the kinase clearly co-immunoprecipitated with ␤arrestin1, as seen using anti-FLAG antiserum (Fig. 3c). In marked contrast to this, the triple mutant form, R47A/K48A/R49A-MEK1, failed to coimmunoprecipitate with ␤arrestin1 (Fig. 3c).
Use of a 25-Mer ␤Arrestin Peptide to Disrupt MEK1-␤Arrestin Complexes in Cells-The "native" 25-mer peptide, 6 TRVFKKASPNGKLTVYLGKRDFVD 29 (Thr 6 -Asp 26 -Phe 27 -Val 28 -Asp 29 ) derived from ␤arrestin1 includes the Asp 26 -XX-Asp 29 motif crucial for MEK1 interaction (see above). This peptide was modified by N-terminal stearoylation, which has been shown to allow entry of a variety of peptides of this size into cells so as to disrupt complexes (26,33,34). We also constructed a stearylated "mutant" peptide that had Asp 26 and Asp 29 each substituted with alanine, 6 TRVFKKASPNGKLTVYLGKR-AFVA 29 (Thr 6 -Ala 26 -Phe 27 -Val 28 -Ala 29 ). Two issues influenced our selection of this peptide. First, the fact that it contained only one DXXD motif, namely the critical one that when mutated causes loss of MEK binding, allowed us to undertake the "cleanest" experiment, where we focus on the key region, and also the "cleanest" control peptide, where this one motif FIGURE 3. ␤arrestin1 binds to the N-terminal of MEK1 at residues Arg 47 -Lys 48 -Arg 49 . a, immobilized peptide spots of overlapping 25-mer peptides, each shifted by 5 amino acids, covering the entire sequence of MEK1 were probed for interaction with either GST-␤arrestin1 or GST alone by immunoblotting. Positively interacting peptides are represented by dark spots. Spot 10, covering the sequence Gln 46 -Glu 50 , produced the greatest interaction with GST-␤arrestin1 and was selected for alanine-scanning analysis. b, multiple residues, indicated as boldface and underlined, within spot 10 (Gln 46 -Glu 50 MEK1) were substituted with alanines and monitored for interaction with GST-␤arrestin1 as in a. c, HEK293 cells were dually transfected with FLAG-␤arrestin1 and either HA-MEK1 or a mutant of HA-MEK1 in which Arg 47 -Lys 48 -Arg 49 had been mutated to AAA. ␤Arrestin1 was immunopurified using the FLAG epitope, and preparations were immunoblotted for the presence of HA-tagged MEK1. Ab, antibody.

JOURNAL OF BIOLOGICAL CHEMISTRY 11429
(DXXD) could be changed, in this instance to AXXA. Also, the stearate group, which allows cell insertion, is added to the N-terminal end of the peptide, and we wished this to be as far as possible from the critical DXXD motif so as not to cause any possible steric interference.
The Importance of MEK1 Binding to ␤Arrestin in Regulating the ERK Phosphorylation Status of ␤Arrestin1-In cells that have not been stimulated by G-protein receptor agonists, ␤arrestin resides primarily in the cytoplasm, where it is constitutively phosphorylated at Ser 412 through the action of ERK (9). Expression of our FLAG-tagged wild-type ␤arrestin1 showed clear evidence of phosphorylation at such a site through analysis with a specific phosphoantiserum (Fig. 5). This event was mediated by ERK, since it was attenuated using a 4-h pretreatment of the MEK-selective inhibitors UO126 and PD98059 and a combination of both (Fig.  5a). However, no such phosphorylation was evident for FLAG-tagged D26A/D29A-␤arrestin1, which does not bind MEK1 (Fig. 5a).
We then set out to explore whether the cell-permeable, stearoylated 25-mer ␤arrestin peptide (Thr 6 -Asp 26 -Phe 27 -Val 28 -Asp 29 ) that disrupts ␤arrestin-MEK1 complexes affected the ERK phosphorylation status of ␤arrestin1 in cells (Fig. 5b). Indeed, treatment of HEK293 cells with this peptide, but not with the control D26A/D29A-substituted one, led to a marked reduction in the Ser 412 phosphorylation status of ␤arrestin1 (Fig. 5b). To counteract the notion that mutation of charged residues Asp 26 /Asp 29 on the control peptide may differentially affect its entry into cells, fluorescein-labeled versions of both control and mutant peptides were compared for cell distribution (supplemental Fig. 1). From this we see (supplemental Fig. 1) that both peptides seemingly had the ability to accumulate in the membrane and cytosol of HEK cells to a similar degree. To further test the efficiency of the MEK disruptor peptide, we treated HEK cells with EGF to markedly increase the pool of phosphorylated, activated ERK available within the cells. Interestingly, the amount of Ser 412 phosphorylation of ␤arrestin1 remained unchanged compared with control during peak EGF stimulation (5 min of EGF, vehicle control, second lane), indicating that ␤arrestin1 was maximally phosphorylated by ERK under basal conditions. Additionally, the efficiency of the disruptor peptide to facilitate the dephosphorylation of ␤arrestin1 on Ser 412 was unaffected by EGF treatment (Fig. 5c, third panel), suggesting that disruption of the MEK-␤arrestin1 complex induced by the displacement peptide is dominant even under conditions of elevated global ERK activity.
The Importance of MEK1 Binding to ␤Arrestin in Regulating the Association of Clathrin and Src to ␤Arrestin-It is well established that the dephosphorylation of ␤arrestin1 at Ser 412  26 and Asp 29 mutated to alanine (Control peptide). Cell lysates were immunoblotted for expression of FLAG-␤arrestin1 using the FLAG epitope or immunoblotted for phosphorylation at serine 412 on ␤arrestin1 using a site directed phosphoantibody. **, statistical significance (p Ͻ 0.01 using Student's t test). c, HEK293 cells were transfected with FLAG-␤arrestin1 before treatment (2 h) with cell-permeable peptides (10 M) that encompassed amino acids 6 -29 (Thr 6 -Asp 26 -Phe 27 -Val 28 -Asp 29 ) of ␤arrestin, one of which had Asp 26 and Asp 29 mutated to alanine (Cont peptide). Cells were then treated with EGF (1 M) for the indicated times. Cell lysates were immunoblotted for expression of FLAG-␤arrestin1 and phosphorylation at serine 412 on ␤arrestin1, ERK, and phospho-ERK using appropriate antibodies. Wt, wild type. occurs when ␤arrestin translocates to the agonist-occupied ␤ 2 -adrenergic receptor (11). This is believed to act as a trigger for clathrin-mediated receptor endocytosis, since dephosphorylation at Ser 412 increases the affinity of ␤arrestin for binding to clathrin and also to c-Src (9,12). Here we show that upon isoprenaline challenge of HEK293 cells transiently overexpressing the ␤ 2 -adrenergic receptor and FLAGtagged ␤arrestin1, there is an increase in the association of both clathrin and c-Src with ␤arrestin1 (Fig. 6). When these cells were treated with the MEK displacer peptide (Thr 6 -Asp 26 -Phe 27 -Val 28 -Asp 29 ), there was a marked increase in isoprenaline-induced association of ␤arrestin1 with both clathrin and c-Src (Fig. 6). These findings are consistent with the notion that the peptide promotes dephosphorylation of ␤arres-tin1 at Ser 412 . Indeed, treatment with this MEK-displacing peptide seemed to promote association of c-Src and clathrin with ␤arrestin1 in the basal, unstimulated cells. In contrast, the D26A/D29A-substituted "mutant" peptide failed to influence the binding of either clathrin or c-Src to ␤arrestin1 under either basal or stimulated conditions (Fig. 6). Control vesicular stomatitis virus immunoprecipitations showed that neither FLAGtagged ␤arrestin1, c-Src, clathrin, nor ERK were pulled down in a nonspecific manner (supplemental Fig.  2) by the agarose beads.
Displacement of the MEK1-␤Ar-restin1 Complex Promotes ␤ 2 -AR Endocytosis-Plasma membrane-recruited ␤arrestin, when dephosphorylated at Ser 412 , undergoes receptormediated endocytosis through its increased association with clathrin and c-Src. We followed isoprenaline-triggered ␤ 2 -AR endocytosis in HEK-B2 cells, which constitutively express ␤ 2 -ARs tagged with both FLAG and GFP (35). We did this by monitoring the internalized population of receptors in a biochemical assay and by confocal microscopy utilizing the GFP tag on the stably expressed receptor. First, we show, using an assay where the receptors are biotinylated, that treatment of the cells with the 25-mer ␤-arrestin peptide (Thr 6 -Asp 26 -Phe 27 -Val 28 -Asp 29 ) that disrupts MEK1 binding acts to attenuate ERK-mediated Ser 412 phosphorylation of ␤arrestin1 and markedly facilitates the rate of isoprenaline-mediated internalization of the ␤ 2 -AR (Fig. 7a). In contrast to this, treatment with the mutant D26A/D29A-substituted peptide did not.
When we visualized the isoprenaline elicited internalization of the ␤ 2 -AR in these cells using confocal microscopy, an obvious (Fig. 7b) and significant (Fig. 7c) potentiation of receptor internalization was observed in the MEK1 binding disruptor peptide-treated but not the D26A/D29A-substituted mutant peptide-treated cells. Together, these findings support the notion that the displacement of the MEK1-␤arrestin1 complex attenuates the phosphorylation of Ser 412 by ERK MAPKs and promotes ␤arrestin1 association with clathrin and c-Src to enhance ␤ 2 -AR internalization.

DISCUSSION
The phosphorylation of ␤arrestin1 at Ser 412 by ERK MAPK has been shown to regulate its endocytotic properties but is not connected to the ability of ␤arrestin to desensitize the ␤ 2 -AR (9). Upon receptor activation, ␤arrestin translocates to the plasma membrane, binds the phosphorylated receptor, and is dephosphorylated, a process that allows clathrin/c-Src binding followed by c-Src-mediated phosphorylation of dynamin that is essential for G-protein-coupled receptor endocytosis to ensue (36,37). Interestingly, the control of the ␤arrestin locality via phosphorylation/dephosphorylation is also seen in ␤arrestin2, where phosphorylation occurs at different sites (Ser 361 and Thr 383 ) and by a different kinase (casein kinase II) (38,39). Phosphorylation of ␤arrestin1 at Ser 412 can also be triggered by insulin, and this can block isoprenaline-induced dephosphorylation and subsequent ␤ 2 -AR receptor internalization and downstream ERK signaling (40). In all examples, the phosphorylation of ␤arrestin1 seems to be dependent on signaling via MEK, the amplicon for ERK, since either MEK inhibitors or dominant negative inactive MEK attenuated modification at Ser 412 (10,31,40). In this study, we demonstrate that ␤arrestin1 is largely phosphorylated by ERK at Ser 412 and that this pool of ERK is activated by ␤arrestin1-associated MEK.
Here, we show that MEK1 can bind directly to ␤arrestin1 and that this association does not require any complexing to ERK in order to occur. Previous studies had indicated that MEK, ERK, and c-RAF1 can simultaneously associate with ␤arrestin, and, on this basis, it was tentatively suggested that c-RAF1 might act as a scaffold for MEK and ERK to facilitate their association with ␤arrestin (10,31). Furthermore, during revision of this paper, utilizing transfected truncates of ␤arrestin1 and ␤arres-tin2, Song et al. (24) suggested that MEK 1 may bind to both the N and C domains of ␤arrestins. In agreement, we demonstrate that MEK can directly bind to ␤arrestin1 at a docking site situated within the N-domain of ␤arrestin1. Interestingly, a recent report has identified a docking site for ERK on ␤arrestin that is clearly distinct from that identified here for MEK1, since it resides in the ␤arrestin1 C-domain (32). In light of all of these reports, it seems likely that ␤arrestin may directly scaffold ERK (MAPK), MEK (MAPK kinase), and c-RAF1 (MAPK kinase kinase) in order to orient each of them correctly for efficient downstream signaling. Our studies and those of Song et al. (24) may imply that there is be a predominant direct interaction of MEK with the N-domain of ␤arrestin together with a subsidiary indirect interaction with the ␤arrestin C-domain achieved via an interaction of MEK with ERK. The binding sites for c-RAF on ␤arrestin, however, have yet to be elucidated.
The MEK1-␤arrestin heterodimer is known to regulate the endocytotic properties of ␤arrestin via ERK phosphorylation, which has downstream implications on the rate and amount of internalized ␤ 2 -AR within the cell. Peptide array technology has allowed us to characterize specific ␤arrestin residues that play a role in the docking of MEK1, and scanning substitution arrays have identified a stretch of residues that contain a MEK1 binding motif (14), including amino acids Asp 26 and Asp 29 (Fig.  8). Interestingly, Asp 26 and Asp 29 have partial surface exposure in the basal arrestin conformation (Protein Data Bank codes 1G4M and 1G4R (41) and 1ZSH (42)), but they are essentially sequestered by interactions with adjacent residues. Asp 26 forms salt bridges to Lys 355 and Arg 393 in the C-terminal sequence and also contacts the "phosphate sensor" residue, Arg 169 , whereas Asp 29 forms a salt bridge to Lys 170 and hydrogenbonds to the side chain of Gln 172 (Fig. 8). This makes it rather unlikely that MEK binds to these residues in ␤arrestin1 in the basal conformation identified by crystallography for ␤arrestin without any liganded partner proteins (Fig. 8). The involvement of Asp 26 in binding of MEK is only possible if these residues are freed from their sequestered engagement by the arrestin C-terminal sequence. Displacement of the C-terminal tail sequence leaves Asp 26 more exposed, but further conformational change must occur in order for Asp 26 and Asp 29 to be available to bind MEK. There is a paradigm for this in that arrestins can undergo a profound conformational change upon binding to their activated phospho-GPCR partners. This involves GPCR-phosphate interaction with the arrestin phosphate sensor and unlatching of the C-terminal tail sequence from the position it occupies in the basal conformation, folded across the N-terminal domain. It is known that ␤arrestin1 can bind to a myriad of partner proteins (21,43), some of which are highly likely to cause structural alterations in this signaling scaffold protein.  29) . The ␤ 2 -ARs expressed in HEKB2 cells were biotinylated before cells were treated with isoprenaline (10 M) for 5 min. Receptors at the surface were stripped before cells were lysed and lysates were immunoblotted for internalized receptor using an antibody specific for the ␤ 2 -Ar. *, statistical significance (p Ͻ 0.05 using Student's t test). b, HEKB2 cells were similarly treated with peptide and control peptide as for a before being stimulated with isoprenaline (10 M) for 5 min. Cells were then fixed and subjected to analysis by confocal microscopy visualizing the GFP tag on the stably transfected ␤ 2 -ARs in these cells. c, quantification of internalized GFP-␤ 2 -AR done as described under "Experimental Procedures." n ϭ number of cells analyzed/treatment. ***, p Ͻ 0.001.
Indeed, it is well appreciated that ␤arrestin1 can undergo structural changes upon post-translational modification and binding to receptors (44,45). Thus, we would suggest that MEK probably binds to either a modified or complexed subpopulation of ␤arrestin1 that allows exposure of Asp 26 and Asp 29 in its N-terminal region. Another possibility is that ␤arrestin1 may exist in equilibrium between its basal state and a low abundance "open conformation" resembling that to which MEK, and perhaps the phospho-GPCR, can bind. If so, MEK might bind to the open conformation, stabilizing this and shifting the equilibrium through to form a ␤arrestin1 pool complexed by MEK. It is noteworthy that the Asp 26 -and Asp 29 -containing sequence will be prominently surface-exposed within the isolated ␤arrestin N-domain. Therefore, the ability of this isolated domain to bind MEK1 (24) would be consistent with an interaction mediated, at least in part, by these two aspartate residues. Furthermore, if, as Song et al. (24) suggest, there is an interaction of MEK1, either directly or indirectly, with the C-domain, then this may act as the trigger to expose Asp 26 and Asp 29 at the N-domain of ␤arrestin for MEK1 binding.
We have also used peptide array analysis to determine the sites on MEK that associate with ␤arrestin1. The accessibility of the identified ␤arrestin-binding arginines, Arg 47 and Arg 49 , of MEK1 is not known, because they are absent in available crystal structures (46). However, their localization in the N terminus upstream of the kinase domain would be consistent with the notion that they may be surface-exposed and thereby available for interaction.
The identification of the MEKbinding motif on ␤arrestin1 has allowed the generation of doublesubstituted alanine mutants of ␤arrestin1 that cannot bind with MEK1. We have also used a novel small molecule technique in order to develop a cell-permeable peptide inhibitor of MEK1 binding to ␤arrestin1 and used this to show that MEK1 binding to ␤arrestin1 regulates the phosphorylation of ␤arrestin1 at the established ERK phosphorylation site at Ser 412 even in times of heightened global ERK activity. Lack of direct MEK1 binding to the ␤arrestin1-ERK complex thus decreases the phosphorylation levels of ␤arrestin1, which allows ␤arrestin1 to bind more readily to clathrin and c-Src, integral parts of the endocytotic machinery. This augmented association of ␤arres-tin1 with clathrin/c-Src induces an increase in ␤ 2 -AR internalization, a process known to be initiated by dephosphorylated ␤arrestin at Ser 412 (11). Indeed, the fact that the MEK disruptor peptide induces ␤arrestin1 dephosphorylation, heightens the interaction of ␤arrestin1 with c-Src/clathrin, and promotes receptor internalization strongly suggests that the dephosphorylation of ␤arres-tin1 at Ser 412 is a rate-limiting step for the receptor internalization process.
We have presented data that represent the first identification of the MEK1-binding site on ␤arrestin1. In mapping the binding sites on both MEK1 for ␤arrestin1 and on ␤arrestin1 for MEK1, we have generated mutants that disrupt the interaction of these two components while not affecting ERK association. Additionally, we have been able to design and apply a small peptide that disrupts the MEK1/␤arrestin1 interaction in cells to facilitate study of the functional consequences of this interaction. This disruption of MEK1 association has clear implications for ERK action on ␤arrestin1, since it ablates ERK-mediated phosphorylation of ␤arrestin1 and alters ERK-mediated regulation of the internalization and recycling of the ␤ 2 -AR. Such a peptide has thus potential for manipulation of the internalization and recycling of the ␤ 2 -AR.