The Active Conformation of β-Arrestin1

β-Arrestins are multifunctional adaptor proteins that regulate seven transmembrane-spanning receptor (7TMR) desensitization and internalization and also initiate alternative signaling pathways. Studies have shown that β-arrestins undergo a conformational change upon interaction with agonist-occupied, phosphorylated 7TMRs. Although conformational changes have been reported for visual arrestin and β-arrestin2, these studies are not representative of conformational changes in β-arrestin1. Accordingly, in this study, we determine conformational changes in β-arrestin1 using limited tryptic proteolysis and matrix-assisted laser desorption ionization time-of-flight mass spectrometry analysis in the presence of a phosphopeptide derived from the C terminus of the V2 vasopressin receptor (V2Rpp) or the corresponding unphosphorylated peptide (V2Rnp). V2Rpp binds specifically to β-arrestin1 causing significant conformational changes, whereas V2Rnp does not alter the conformation of β-arrestin1. Upon V2Rpp binding, we show that the previously shielded Arg393 becomes accessible, which indicates release of the C terminus. Moreover, we show that Arg285 becomes more accessible, and this residue is located in a region of β-arrestin1 responsible for stabilization of its polar core. These two findings demonstrate “activation” of β-arrestin1, and we also show a functional consequence of the release of the C terminus of β-arrestin1 by enhanced clathrin binding. In addition, we show marked protection of the N-domain of β-arrestin1 in the presence of V2Rpp, which is consistent with previous studies suggesting the N-domain is responsible for recognizing phosphates in 7TMRs. A striking difference in conformational changes is observed in β-arrestin1 when compared with β-arrestin2, namely the flexibility of the interdomain hinge region. This study represents the first direct evidence that the “receptor-bound” conformations of β-arrestins1 and 2 are different.

␤-Arrestins are multifunctional adaptor proteins that regulate seven transmembrane-spanning receptor (7TMR) desensitization and internalization and also initiate alternative signaling pathways. Studies have shown that ␤-arrestins undergo a conformational change upon interaction with agonist-occupied, phosphorylated 7TMRs. Although conformational changes have been reported for visual arrestin and ␤-arrestin2, these studies are not representative of conformational changes in ␤-arrestin1. Accordingly, in this study, we determine conformational changes in ␤-arrestin1 using limited tryptic proteolysis and matrix-assisted laser desorption ionization time-of-flight mass spectrometry analysis in the presence of a phosphopeptide derived from the C terminus of the V 2 vasopressin receptor (V 2 Rpp) or the corresponding unphosphorylated peptide (V 2 Rnp). V 2 Rpp binds specifically to ␤-arrestin1 causing significant conformational changes, whereas V 2 Rnp does not alter the conformation of ␤-arrestin1. Upon V 2 Rpp binding, we show that the previously shielded Arg 393 becomes accessible, which indicates release of the C terminus. Moreover, we show that Arg 285 becomes more accessible, and this residue is located in a region of ␤-arrestin1 responsible for stabilization of its polar core. These two findings demonstrate "activation" of ␤-arrestin1, and we also show a functional consequence of the release of the C terminus of ␤-arrestin1 by enhanced clathrin binding. In addition, we show marked protection of the N-domain of ␤-arrestin1 in the presence of V 2 Rpp, which is consistent with previous studies suggesting the N-domain is responsible for recognizing phosphates in 7TMRs. A striking difference in conformational changes is observed in ␤-arrestin1 when compared with ␤-arrestin2, namely the flexibility of the interdomain hinge region. This study represents the first direct evidence that the "receptor-bound" conformations of ␤-arrestins1 and 2 are different.
cules, and these domains are connected through a 12-residue linker, or hinge, region ( Fig. 1A) (16). It has been suggested that upon "activation" of ␤-arrestins, the N-and C-domains move relative to one another via flexibility of the hinge region. Two major intramolecular interactions have been suggested to stabilize the basal conformation of arrestins. The first is a series of hydrophobic interactions between ␣-helix I, ␤-strand I, and the C terminus of the molecule, which folds back onto the N-domain (Fig. 1B). The second intramolecular interaction that "holds" arrestins in their inactive state is the polar core; a distinctive interaction of five charged residues that are shielded from water and embedded at the fulcrum of the molecule between the N-and C-domains (Fig. 1C). This polar core contains elements of both the N and C termini of arrestins and also a lariat loop region (Arg 282 -Gly 309 ). The lariat loop plays a central role in stabilizing the polar core because it contains the primary counterion, Asp 290 , for Arg 169 in the polar core. This loop apparently lacks any secondary structure, yet its tertiary structure is virtually identical in both visual arrestin and ␤-ar-restin1 suggesting that its conformation is essential in stabilizing the basal state of arrestins.
Studies have suggested that upon binding to 7TMRs, the polar core of arrestins is disrupted, and its C terminus is released (7,12,15,16,20,22,28,29). The C termini of ␤-ar-restins contain both clathrin-and AP2-binding sites, and exposure of this C terminus is essential for clathrin-mediated receptor internalization (5). Studies have demonstrated that visual arrestin binding to a phosphorylated peptide corresponding to the last 17 amino acids (C terminus) of rhodopsin can mimic visual arrestin binding to phosphorylated rhodopsin as assessed by limited trypsin proteolysis (20,22,26). Despite evidence for conformational changes in arrestins, no structure for the active, or receptor-bound, conformation has been determined for any arrestin family member. Moreover, conformational studies of arrestins are primarily limited to the visual arrestin system.
␤-Arrestin1 and -2 are ubiquitously expressed, and their universal regulation of 7TMRs attests to their importance in modulating cellular function. Despite the ubiquitous expression of both ␤-arrestin1 and -2, recent evidence has suggested that the two isoforms are in fact not functionally redundant. For example, ␤-arrestin1 is responsible for scaffolding RhoA activation in conjunction with G␣ q/11 and also IGF-1 activation of phosphatidylinositol 3-kinase (30,31). ␤-Arrestin2 scaffolds the mitogen-activated protein kinase (MAPK) cascade to activate ERK1/2, c-Jun N-terminal kinase (JNK3), and in some cases, p38 (4,6,(32)(33)(34)(35)(36). Our laboratory has previously reported conformational changes in ␤-ar-restin2 with a phosphorylated peptide corresponding to the C terminus of the human V 2 vasopressin receptor (V 2 Rpp) as assessed by limited tryptic proteolysis and MALDI-TOF analysis (29). Using this same approach, we now report that the conformation of ␤-arrestin1 is also altered upon binding V 2 Rpp, but in ways that suggest that its activated conformation is different from that of ␤-arrestin2.

EXPERIMENTAL PROCEDURES
Peptide Synthesis-The synthesis of both V 2 Rnp and V 2 Rpp have been described elsewhere, and the sequence of both peptides is as follows with phosphorylation sites boldface and underlined: ARGRTPPSLGPQDESCTTASSSLAKDTSS (29). Two other nonspecific synthetic peptides, a 28-and 30-mer, derived from GRK2 were used as controls and have been described previously (29).
Purification of Recombinant Rat ␤-Arrestin1-Wild type rat ␤-arrestin1 was subcloned into a pGEX4T1 vector and expressed as a GST fusion protein (GST-␤-arrestin1). This construct was confirmed by DNA sequencing and transformed into Escherichia coli strain BL21(DE3) pLysS. To overexpress GST-FIGURE 1. The basal conformation of bovine ␤-arrestin1. A, crystal structure of bovine ␤-arrestin1 is shown as a ribbon diagram with both the N and C termini labeled. The diagram is colored by domain with the N-domain in magenta, C-domain in blue, and the hinge region in orange. Circled regions of the structure in A are enlarged in B and C. B, hydrophobic interactions between ␣-helix I, ␤-strand I, and the C-tail, which stabilize the basal conformation of ␤-arrestin, are shown. ␣-Helix I and ␤-strand I are shown in magenta, and the C-tail is shown in blue. Hydrophobic contributions from ␣-helix I include three leucines, whereas phenylalanine residues from the C-tail are also shown. Conserved lysines in ␤-strand I (Lys 10 and Lys 11 ) are also shown as these residues guide receptor-attached phosphates to Arg 169 in the polar core. C, the local environment of the polar core (five interacting charged residues) is shown with sticks for residues Asp 26 , Arg 169 , Asp 290 , Asp 297 , and Arg 393 . The polar core is the main phosphate sensor in arrestins, and upon activation, this polar core is disrupted causing release of the C-tail (shown in blue).
␤-arrestin1, cultures were grown at 37°C to an A 600 of 0.8, and the cultures were then equilibrated to 17°C. GST-␤-arrestin1 expression was induced with 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside for 16 h, and cells were then harvested by centrifugation at 4,500 ϫ g. The bacterial pellet was resuspended in binding buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, 2 mM DTT, 1 mM phenylmethylsulfonyl fluoride, and 1 mM benzamidine) to give a 10% slurry (w/w). Cells were lysed with a cell cracker (Microfluidics) and then centrifuged at 18,000 ϫ g for 30 min. The clarified supernatant was loaded onto a glutathione-Sepharose (GS) column (GE Healthcare) by gravity and washed with 20 column volumes of column buffer 1 (CB1) (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, and 2 mM DTT). The GS resin with GST-␤-arrestin1 bound was resuspended in 2 column volumes of CB1, and the GST fusion protein was cleaved with thrombin protease (Hematologic Technologies Inc.) at a mg/mg ratio of 1:1000 of thrombin: GST-␤-arrestin1. The thrombin digestion was carried out at 4°C for 16 h with gentle agitation of the GS resin, and the supernatant was collected followed by two additional washes of the GS resin with CB1. The washes and supernatant were pooled, and the NaCl concentration was diluted dropwise to 50 mM NaCl by the addition of 20 mM Tris-HCl, pH 8.0, 1 mM EDTA, and 2 mM DTT. The cleaved ␤-arrestin1, which contains eight additional amino acids at the N terminus after thrombin cleavage (Fig. 2), was then loaded onto a 5-ml HiTrap Mono-Q column (GE Healthcare) and then eluted with a 50 -500 mM linear NaCl gradient. Fractions were analyzed by SDS-PAGE, and those containing ␤-arrestin1 were pooled and concentrated to 25 mg/ml and then loaded onto a Superdex 75 (GE Healthcare) gel filtration column. ␤-Arrestin1 was eluted in 20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, and 2 mM DTT. Fractions were analyzed by SDS-PAGE, and fractions containing ␤-arrestin1 were pooled and concentrated to 5-25 mg/ml, flash-frozen in liquid N 2, and stored at Ϫ80°C. Protein purity was Ͼ95% as assessed by SDS-PAGE, and the yield was 5 mg/liter of cell culture.
Limited Trypsin Proteolysis-In all experiments, except where noted, a 5:1 molar ratio of peptide:␤-arrestin was used to assess the effects of peptides on the limited tryptic digestion patterns of ␤-arrestin1 and -2. Prior to proteolysis, ␤-arrestins were incubated with ligand (V 2 Rpp, V 2 Rnp, 28-mer, 30-mer, or heparin) for 30 min at 4°C. An average molecular mass of 12,000 Da was used to determine the concentration of heparin (Sigma). L-1-Tosylamido-2-phenylethyl chloromethyl ketonetreated trypsin (Sigma) to ␤-arrestin1 ratio of 1:250 was used in all experiments shown; however, ratios of 1:100, 1:500, 1:1,000 and 1:2,000 were also used to assess the effects of the trypsin to ␤-arrestin1 ratio on the limited proteolysis pattern (data not shown). No proteolysis was seen in experiments containing trypsin:␤-arrestin1 ratios of 1:1,000 or 1:2,000. In the case of ␤-arrestin2, a trypsin to ␤-arrestin2 ratio of 1:2,000 was used. Higher trypsin concentrations (1:250, 1:500, and 1:1000) were also tested, but these concentrations of trypsin completely digested ␤-arrestin2. After incubation with ligand, trypsin was added to the ␤-arrestin:ligand mixture, and the samples were incubated at 25°C for the indicated time points. At each time point, 5 l (5 g) of ␤-arrestin1 or -2 were removed and trans-ferred to a new microcentrifuge tube containing 5 l of 2ϫ SDS-PAGE buffer, and samples were analyzed on 4 -20% SDS-PAGE (Invitrogen). For MALDI-TOF analysis, 5 l of proteolysis sample were transferred to an empty microcentrifuge tube and flash-frozen in liquid N 2 .
MALDI-TOF MS Analysis-Spectra were collected in positive-ion mode on a Voyager DE Biospectrometry workstation (Applied Biosystems Inc.) in linear mode using a N 2 laser (337 nm). The acceleration voltage, grid voltage, guide wire voltage, delay time, low mass gate, and laser intensity were set to 25 kV, 92.5%, 0.11%, 1200 ns, 10,000 m/z and 2500, respectively. Sixty laser shots were collected for each sample, and the spectra shown represent the sum of these 60 laser shots. Samples for MALDI-TOF MS analysis were thawed on ice and immediately diluted 25-50-fold in matrix solution (45% acetonitrile, 0.1% trifluoroacetic acid, and 5 mg/ml sinapinic acid) giving a final ␤-arrestin1 concentration of 2-4 M before depositing 1 l of the sample mixture onto the MALDI-TOF target plate (Applied Biosystems). Both internal and external standards were used to calibrate the data. For internal calibration, 1 l of carbonic anhydrase (Sigma) was deposited on the MALDI-TOF target after the ␤-arrestin1 sample had dried. Apomyoglobin and aldolase (Sigma) were used as external calibrants and were deposited on empty target spots. Samples were air-dried at room temperature prior to MALDI-TOF analysis. For each limited tryptic digestion, the mean mass and standard deviations were calculated from at least five independent experiments (Table 1). For low abundance peaks, samples were prepared by quenching the tryptic digest with 1 mM phenylmethylsulfonyl fluoride, and samples were then passed through a ZipTip (Millipore) according to the manufacturer's instructions. Samples were eluted from the ZipTip with matrix solution, and 1 l was directly deposited on the MALDI-TOF plate. Protein Prospector was used to determine all theoretical trypsin digestion fragments for rat ␤-arrestin1. The theoretical trypsin digest was compared with experimentally determined masses to assign candidate fragments for each ␤-arrestin1 fragment observed in the MALDI-TOF spectra ( Table 1). The limited proteolytic fragments with only one possible theoretical candidate fragment from Protein Prospector were assigned directly. Fragments with more than one theoretical candidate were assigned by Western blot analysis with antibodies that recognize either the N or C terminus of ␤-arrestin1 and by determining more accurate masses by liquid chromatography electrospray ionization MS (LC/ESI-MS).
LC/ESI-MS-␤-Arrestin1 samples were proteolyzed as described above either in the absence of ligand or in the presence of V 2 Rnp or V 2 Rpp and flash-frozen in liquid N 2 at various time points. Samples were analyzed on a Shimazu LC system (consisting of a solvent degasser, two LC-10A pumps, and an SCL-10A system controller) coupled to a QSTAR XL quadrupole time-of-flight tandem mass spectrometer (Applied Biosystems/MDS-Sciex, Toronto, Canada) equipped with an electrospray source. LC, with a Vydac C4 reverse phase column (2.1 ϫ 50 mm), was operated at a flow rate of 200 l/min with a linear gradient as follows: 100% A was held isocratically for 2 min and then linearly increased to 60% B over 18 min and then increased to 100% B over 5 min. Mobile phase A consists of water:aceto-nitrile (98:2 v/v) with 0.1% acetic acid. Mobile phase B consists of acetonitrile:water (90:10 v/v) with 0.1% acetic acid. The acquisition and deconvolution of ESI mass spectra were performed using the Analyst QS software.
Clathrin Binding-To determine the effects of ligand on ␤-arrestin1 binding to clathrin, 2.5 M ␤-arrestin1 was incu-bated at 4°C for 30 min in the absence or presence of a 5:1 molar ratio of ligand:␤-arrestin1 in binding buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl). The ␤-arrestin1:ligand mixture was diluted to 50 nM ␤-arrestin1 (500 l) with binding buffer, and clathrin was subsequently added at a 1:1 molar ratio to the reaction mixture. The reactions were tumbled at 4°C for 4 h, and then 7 l of a clathrin monoclonal antibody (BD Transduction Laboratories) was added to the mixture and tumbled for an additional 1 h at 4°C. 14 l of protein A-agarose beads (Roche Applied Science) was then added to the mixture and tumbled for 1 h at 4°C. The beads were centrifuged at 20,000 ϫ g in a benchtop microcentrifuge and washed with 1 ml of binding buffer five times and resuspended in 20 l of 2ϫ SDS-PAGE loading buffer. ␤-Arrestin1 binding to clathrin was measured by Western blot analysis with an anti-␤-arrestin1 antibody (A1CT). Samples were also subjected to SDS-PAGE and Western blot analysis with an anti-clathrin antibody (BD Transduction Laboratories) to normalize the amount of clathrin for each reaction.

RESULTS
Conformational Changes in ␤-Ar-restin1 upon V 2 Rpp Binding-Recombinant ␤-arrestin1 was purified as described under "Experimental Procedures," and the primary sequence is shown in Fig. 2. To study conformational changes in ␤-arrestin1 upon its association with a 7TMR, we employed an in vitro model system utilizing both a phosphopeptide (V 2 Rpp) and nonphosphopeptide (V 2 Rnp) corresponding to the C terminus of the human V2 vasopressin receptor (V 2 R peptide sequences are given under "Experimental Procedures"). ␤-Arrestin1 was incubated in the absence or presence of ligand (a 5:1 molar ratio of ligand:␤-arrestin1 was used in all experiments unless otherwise stated) and then subjected to limited tryptic proteolysis. The proteolysis patterns of ␤-arrestin1 alone or in the presence of V 2 Rnp are identical, and we will refer to this digestion pattern as "control pattern" (Fig. 3A, panels I and II). The control pattern in Fig. 3B illustrates that full-length ␤-ar-FIGURE 2. Sequence alignment of recombinant rat ␤-arrestin1 and -2. Sequence alignment of recombinant rat ␤-arrestins1 and -2 was performed using Geneious software. Both of these recombinant proteins contain additional amino acids at the N terminus because of thrombin cleavage of the GST tag during purification (number Ϫ8 to Ϫ1). The top sequence corresponds to ␤-arrestin1, and the bottom sequence corresponds to ␤-arrestin2. Residues are colored according to polarity, and the numbering is shown only for ␤-arrestin1. The start Met in the sequence of ␤-arrestin1 is replaced by a Leu and is followed by the wild type sequence with Gly at position 2. Important Arg residues to this study are indicated with arrows (Arg 188/189 and Arg 363/364 for ␤-arrestin1 and -2, respectively). The hinge region of ␤-arrestin, residues 173-184 of ␤-arrestin1, is indicated by a rectangle, and arginines 188 and 189, of ␤-arrestins 1 and 2, respectively, located just outside this region are indicated with an arrow. The Active Conformation of ␤-Arrestin1 restin1 (Gly Ϫ8 -Arg 418 ) has an apparent molecular mass of 47 kDa, and the addition of trypsin results in slow, continuous digestion generating fragments with apparent molecular masses of 47, 40, 32, 25, and 21 kDa. The addition of the V 2 Rpp resulted in a distinct digestion pattern with an accelerated proteolysis of the 40-and 32-kDa species as well as the full-length protein (Fig. 3A, panel III). This new digestion pattern also featured the appearance of new species with molecular weights of 44 and 45 kDa and the accumulation of the 21-kDa species over time. The V 2 Rpp pattern is depicted schematically in Fig. 3B. Proteolysis of ␤-arrestin1 was also conducted in the presence of two nonspecific peptides, a 28-and 30-mer, and ␤-arrestin1 digestion in the presence of these peptides resulted in the control pattern (data not shown).
Conformational Changes in ␤-Arrestin1 Require Phosphate Moieties-We have shown that the limited proteolytic digestion pattern of ␤-arrestin1 is unaltered by the addition of V 2 Rnp; the possibility exists that V 2 Rnp can indeed bind to ␤-arrestin1 with a lower affinity than V 2 Rpp, and binding of V 2 Rnp can also induce conformational changes in ␤-arrestin1 at a higher V 2 Rnp to ␤-arrestin1 ratio. We therefore titrated ␤-arrestin1 with increasing amounts of peptides and performed limited proteolysis to assess the effects of changes in peptide to ␤-arrestin1 ratios. V 2 Rnp does not alter the digestion pattern of ␤-arrestin1 even at a peptide to ␤-arrestin1 molar ratio of 20:1, whereas V 2 Rpp alters the conformation of ␤-arrestin1 even at a 1:1 molar ratio (Fig. 4, A and B). We also conducted a competition experiment to determine whether high concentrations of V 2 Rnp can compete with V 2 Rpp for binding to ␤-arrestin1. In the presence of a 1:1 molar ratio of V 2 Rpp, addition of a 100fold molar excess of V 2 Rnp did not convert the V 2 Rpp pattern to the control pattern, clearly demonstrating that the V 2 Rnp is incapable of competing with V 2 Rpp binding (Fig. 4C, lane 6).
MALDI-TOF MS Analysis of Limited Tryptic Proteolysis Fragments of ␤-Arrestin1-Limited proteolysis of ␤-arrestin1 clearly demonstrates a conformational change in the presence of V 2 Rpp. However, details of these conformational changes could not be obtained because of the low resolution of SDS-PAGE analysis of tryptic fragments. To map precisely the regions of ␤-arrestin1 involved in these conformational changes, we employed MALDI-TOF MS to determine sites of proteolysis by measuring the accurate masses of tryptic fragments. Taking advantage of the high mass accuracy of MALDI-TOF MS, most of the species observed by SDS-PAGE were precisely assigned by comparing the experimental mass value of a protein fragment to its theoretical value (Table 1). For those fragments that could not be definitively assigned by MALDI-TOF MS, we used the more accurate LC/ESI-MS with a higher mass accuracy (within less than 1.0 Da for a 5-kDa peptide) to help with assignments. Additionally, to assign some fragments, we also conducted Western blot analysis with antibodies that recognize different domains of ␤-arrestin1 (N-or C-domain). MALDI-TOF MS analysis from over the range of m/z 20,000 -50,000 was conducted on tryptic fragments from ␤-arrestin1 alone or in the presence of either V 2 Rnp or V 2 Rpp. At an early time point (5 min), the spectra for ␤-arrestin1 alone or in the presence of V 2 Rnp are identical, and we have therefore shown spectra of ␤-arrestin1 with V 2 Rnp only. The major peaks are 48 and 47 kDa and minor peaks at 25 and 21 kDa (Fig. 5A, left  panel). In the presence of V 2 Rpp, the major peaks are leftshifted with masses of 45 and 44 kDa, and a minor peak is also observed at 21 kDa (Fig. 5A, bottom of left panel). The most notable difference in spectra collected at an early time point is the accelerated proteolysis of ␤-arrestin1 in the presence of V 2 Rpp resulting in a left-shift of the major peaks.
MALDI-TOF MS of an early time point (5 min) of ␤-arres-tin1 tryptic fragments in the presence of V 2 Rnp indicates that the major species is full-length ␤-arrestin1 (Gly Ϫ8 -Arg 418 ), which has an experimental mass (m/z) of 47,840 Ϯ 41 Da (Fig.  5A). Full-length ␤-arrestin1 is then slowly proteolyzed to residues Leu 1 -Arg 418 because of an N-terminal clip at position Arg Ϫ1 (Fig. 5A). In the presence of V 2 Rpp, ␤-arrestin1 proteolysis is initially accelerated and then slow and continuous over In the 3rd and 4th lanes, ␤-arrestin1 was incubated with a 1:1 molar ratio of either V 2 Rnp or V 2 Rpp, respectively, prior to limited proteolysis. In the 5th and 6th lanes, ␤-arrestin1 was simultaneously incubated with a 1:1 molar ratio of V 2 Rpp and increasing concentrations of V 2 Rnp. All tryptic fragments shown in the 2nd to 6th lanes are of a 30-min time point. In all panels, the apparent molecular weights of the limited tryptic fragments are shown to the right.
time giving rise to a 45-kDa fragment (45,014 Ϯ 28 Da) corresponding to amino acids Gly Ϫ8 -Arg 393 as is evident in both the MALDI-TOF spectra and by SDS-PAGE (Fig. 5, A and B). The assignment for this fragment was confirmed by LC/ESI-MS, which gave a mass of 44,992 Da (Table 1). We further confirmed the assignment of this 45-kDa peak (Gly Ϫ8 -Arg 393 ) by the presence of the 25-amino acid C-terminal peptide, Arg 393 -Arg 417 , which had a mass of 2854.3 Da by LC/ESI-MS (data not shown). In addition, we also identified a peptide corresponding to amino acids Arg 395 -Arg 417 with an apparent mass of 2570.2 Da, which demonstrates that ␤-arrestin1 is actually proteolyzed at both Arg 393 and Arg 395 in the presence of V 2 Rpp, although we could not resolve these two species by MALDI-TOF MS or SDS-PAGE (data not shown). N-terminal proteolysis also occurs on this 45-kDa fragment at the Arg Ϫ1 position resulting in a peak at 44 kDa (44,181 Ϯ 30 Da) in the MALDI-TOF spectra, which corresponds to residues Leu 1 -Arg 393 (Fig. 5A). Fig.  5C is a ribbon diagram of the crystal structure of ␤-arrestin1 with Arg 393 shown, which displays increased accessibility in the presence of V 2 Rpp. Arg 395 is not present in the solved crystal structure of ␤-arrestin1 because this structure is of a truncated ␤-arrestin1 (Met 1 -Arg 393 ) and is therefore not shown in Fig. 5C. Spectra collected at a late time point (60 min) of ␤-arrestin1 alone or in the presence of V 2 Rnp are also identical, and we have again only shown a representative spectrum of ␤-arrestin1 in the presence of V 2 Rnp (Fig. 5A, right panel). The major peaks for these spectra are 47, 40, 32, 25, and 21 kDa, and the 47-kDa peak remained protected over time from the early (5-min) time point. In the spectra for ␤-arrestin1 with V2Rpp, the major peaks occur at 44 and 21 kDa, with minor peaks at 40, 32, and 31 kDa. The 44-kDa fragment that occurs from ␤-arrestin1 with V 2 Rpp remained protected over time and also occurs in the early (5-min) time point spectrum. A striking difference at the late time point (60 min) is that two peaks are observed at 32 and 31 kDa for ␤-arrestin1 with V 2 Rpp, whereas only one (32 kDa) peak is observed in the control situation.
The 32-kDa fragment of ␤-arrestin1 is present in all the samples that were analyzed (Fig. 5A), and the experimental masses for this species with both V 2 Rnp (top spectrum) and V 2 Rpp are 32,321 Ϯ 28 and 32,342 Ϯ 9 Da, respectively. There are two candidate fragments from the theoretical trypsin digest of ␤-ar-restin1 for the 32-kDa species that correspond to either residues Leu 1 -Arg 285 or Lys 107 -Arg 395 (Table 1). This tryptic fragment is consistent with Leu 1 -Arg 285 because it is recognized by an N-terminal antibody (F4C1) whose epitope is residues Asp 38 -Asp 44 (data not shown). Interestingly, fragment Leu 1 -Arg 285 is more readily proteolyzed in the presence of V 2 Rpp, as assessed by SDS-PAGE (Fig. 5D), and two species are observed by MALDI-TOF MS with masses of 31,920 Ϯ 41 and 32,342 Ϯ 9 Da, respectively (Fig. 5A, bottom spectrum). There are five candidate fragments for the 31-kDa species that occur in the presence of V 2 Rpp (Table 1), which cannot be assigned; however, the most likely assignment of this fragment is residues Gly 5 -Arg 285 because it represents further proteolysis of the already assigned 32-kDa species (Leu 1 -Arg 285 ). Fig.  5C shows the location of Arg 285 on the crystal structure of bovine ␤-arrestin1 (Protein Data Bank code 1G4R).
Perhaps the most striking feature of our ␤-arrestin1 study is the rapid appearance of a 21-kDa species in the presence of V 2 Rpp, which persists even up to 2 h after digestion (Fig. 3A, panel III) and the complete absence of a 25-kDa peak in the MALDI-TOF MS. MALDI-TOF analysis over the range m/z 20,000 -30,000 revealed that there are in fact two species, with masses of 25 and 21 kDa, for ␤-arrestin1 alone or in the presence of V 2 Rnp or V 2 Rpp (Fig. 5A, right panel). The 25-kDa species in all experiments is not visible by SDS-PAGE; however, this low abundance species is visible by MALDI-TOF analysis in the late time point (Fig. 5A, right-hand spectra). Experimental masses for the 25-kDa species were determined to be 25,742 Ϯ 17 and 25,736 Ϯ 41 for ␤-arrestin1 alone or in the presence of V 2 Rnp, respectively (Table 1). There are three candidate fragments (Asp 26 -Lys 250 , Val 171 -Lys 400 , and Gln 189 -Arg 418 ) for the 25-kDa species, and it is visible by Western blot analysis, and it is recognized by an N-terminal antibody, F4C1 (Table 1). Only one of the three candidate fragments is from the N-domain of ␤-arrestin1, and we have therefore assigned this fragment as residues Asp 26 -Lys 250 . Functional Consequences of Conformational Changes in ␤-Arrestin1-To assess the functionality of our ␤-arrestin1: V 2 Rpp system and the biological ramifications of conformational changes induced in ␤-arrestin1, we tested the in vitro binding of ␤-arrestin to clathrin (Fig. 6). Clathrin was bound to protein A beads through a monoclonal antibody that recognizes clathrin heavy chain, and ␤-arrestin1 was then incubated with the clathrin beads (clathrin-protein A beads) either in the absence or presence of V 2 R peptides. We assessed ␤-arrestin1 binding to clathrin by Western blot analysis. Fig. 6A indicates the input for each experiment. ␤-Arrestin1 interacts weakly and nonspecifically with empty protein A beads (Fig. 6B, 1st lane), and this low background binding is not altered by the addition of either clathrin-protein A beads or by the preincubation of ␤-arrestin1 with V 2 Rnp (Fig. 6B, 2nd and 3rd lanes). However, in the presence of V 2 Rpp, ␤-arrestin1 binding to clathrin-protein A beads is significantly enhanced (Fig. 6B, 4th  lane). Fig. 6C shows a Western blot for clathrin to ensure equal loading of clathrin for all experimental conditions tested. Quantitation of ␤-arrestin1 binding to either empty protein A beads or clathrin-protein A beads from five independent experiments is shown in Fig. 6D. ␤-Arrestin1 binding to clathrinprotein A beads is normalized to 100% and shows enhanced binding over ␤-arrestin1 in the presence of V 2 Rnp (29.7 Ϯ 4.1%) or absence of ligand (21.3 Ϯ 4.5% for ␤-arrestin1 alone and 29.1 Ϯ 1.5% for ␤-arrestin1 with clathrin).
The "Active" Conformations of ␤-Arrestin1 and -2 Are Different-A previous study from our laboratory reported conformational changes in ␤-arrestin2 upon its interaction with V 2 Rpp (29). However, this study did not include information on tryptic fragments below 30 kDa, which would exclude information on the individual N-and C-domains of ␤-arrestin2. Thus, we conducted limited tryptic proteolysis on ␤-arrestin2 as described under "Experimental Procedures" to directly compare conformational differences between the two ␤-arrestin isoforms (Fig. 7). Fig. 7A shows a limited tryptic digestion of FIGURE 5. MALDI-TOF spectra of ␤-arrestin1 tryptic fragments with either V 2 Rnp or V 2 Rpp. A, MALDI-TOF spectra from two different time points, one early (5 min) and one late (60 min), are shown for ␤-arrestin1 in the presence of V 2 Rnp or V 2 Rpp (spectra of ␤-arrestin1 alone are identical to those in the presence of V 2 Rnp and therefore not shown). Each peak is labeled with the mean molecular weight, standard deviation, and corresponding residues as determined from at least five independent experiments. An internal standard, carbonic anhydrase, was used to correct the molecular weights for each collected spectrum, and the peak for the standard is labeled STD. Peaks that are not labeled or indicated with a dashed line represent doubly charged species. B, SDS-PAGE analysis of tryptic fragments. High molecular weight (Ͼ40 kDa) fragments are labeled on the gel. All species labeled in B and D are highlighted in the MALDI-TOF spectra in A. C, structure of ␤-arrestin1 with important tryptic arginines to this study shown. The accessibility of residues 188, 285, and 393 change upon addition of V 2 Rpp. D, SDS-PAGE analysis of low molecular weight (Ͻ40 kDa) tryptic fragments important to this study. ␤-arrestin2 alone (panel I) or ␤-arrestin2 in the presence of either V 2 Rnp (panel II) or V 2 Rpp (panel III). In the control situation (␤-arrestin2 alone or in the presence of V 2 Rnp), we see both tryptic fragments with apparent molecular masses of 25 and 21 kDa identical to those seen for ␤-arrestin1. For ␤-ar-restin2, however, both the 25-and 21-kDa species are generated more slowly from the 42-kDa parent species and persist up to 2 h post-digestion in the presence of V 2 Rpp (Fig. 4B, panel  II). We have verified via Western blot analysis with F4C1 (epitope is residues Asp 38 -Asp 44 ) that both the 25-and 21-kDa species of ␤-arrestin2 are in fact N-domain fragments (data not shown). Fig. 7B shows a schematic representation of ␤-arres-tin2 tryptic digestions either alone or in the presence of V 2 Rpp (panels I and II) and the representation of a ␤-arrestin1 digestion in the presence of V 2 Rpp (panel III). The major species present in all ␤-arrestin2 digestions is a 42-kDa species, which was previously reported as Gly Ϫ7 -Arg 364 , and this species is significantly protected in the presence of V 2 Rpp (Fig. 7A, compare panels II and III). Residues Gly Ϫ7 -Arg 364 of ␤-arrestin2 include both the N-domain and most of the C-domain; thus, in the presence of V 2 Rpp, the majority of ␤-arrestin2 is protected over time with only the last 52 C-terminal residues missing from this fragment. This is, however, not the case for ␤-arres-tin1 as only the N-domain itself, Leu 1 -Arg 188 , is protected in the presence of V 2 Rpp (Fig. 7B, panel III). Taken together, these data suggest that the V 2 Rpp-bound, or active, conformations of ␤-arrestin1 and -2 are different.

DISCUSSION
␤-Arrestins, initially discovered for their role in terminating 7TMR signaling, have been shown more recently to interact with over a dozen nonreceptor partners and thereby serve as scaffolds for MAPK cascades initiating a second wave of signaling independently of G-proteins (37). The majority of all ␤-arrestin functions are receptor activation-dependent, and the obvious corollary for this is that ␤-arrestins undergo a conformational change when bound to agonist-occupied 7TMRs. The solved crystal structure of ␤-arres-tin1 shows that both the N and C termini are in close proximity in the overall fold of the molecule and that intramolecular interactions of these termini stabilize the basal conformation of the protein (16). The C terminus of ␤-arrestin1 contains both clathrin-and AP2-binding sites, and the release of this terminus from the fulcrum of the molecule is required to expose these sites (3). Clathrin and AP2 recruitment to 7TMRs via ␤-arrestin1 occurs only when 7TMRs are both active and phosphorylated, which demonstrates that ␤-arrestin1 must be in its active conformation for this now  The Active Conformation of ␤-Arrestin1 classical function. This study clearly demonstrates that, in the presence of V 2 Rpp, both the N and C termini are more flexible and, furthermore, that the C terminus is released as evidenced by accessibility of the previously shielded Arg 393 and enhanced clathrin binding. The mechanism by which visual arrestins and ␤-arrestins interact with activated, phosphorylated 7TMRs has been studied through a series of mutagenesis and biochemical studies in addition to solved crystal structures (7-10, 12, 16, 18 -19, 22-23, 25-29, 38 -40). One of the most striking features of all arrestin structures is the presence of a distinctive polar core, a series of five interacting charges completely shielded from water and embedded at the center of the molecule. Disruption of this polar core is necessary for arrestin activation, and we have demonstrated disruption of the polar core of ␤-arrestin1 in the presence of V 2 Rpp by increased accessibility of Arg 285 . This residue is actually located in what is termed the "lariat loop" (Arg 282 -Gly 309 ) region. The lariat loop, in part, maintains the basal conformation of ␤-arrestin because it contains Asp 290 , the primary counterion for Arg 169 in the polar core (Fig. 1C). The primary sequence of the lariat loop is not conserved among the four mammalian arrestins; however, the secondary structure appears to be conserved in all solved crystal structures of arrestin family members to date (9,16,18,21,25). Arg 169 of ␤-arrestin1 is the primary phosphate sensor because a charge reversal mutation (R169E) results in a phosphorylationindependent mutant. Our data represent the first direct biochemical evidence that "activation" of ␤-arrestin1 is dependent on both the disruption of the polar core of ␤-arrestin1 and thereby release of its C terminus to carry out functions such as clathrin binding.
Previous studies with both visual arrestin and ␤-arrestin1 have shown that the N-domain of the molecule contains the main phosphate sensor for phosphorylated 7TMRs (reviewed in Ref. 15). These studies have localized the main sensor to a single residue, Arg 169 . To date, only mutagenesis studies and structural activation models have been used to localize the phosphate sensor of ␤-arrestin. One of the most notable features of our ␤-arrestin1 limited tryptic proteolysis study is the rapid appearance and protection of a 21-kDa species in the presence of V 2 Rpp. MALDI-TOF MS analysis of this tryptic fragment confirms that it is indeed the N-domain of ␤-arrestin1 (Leu 1 -Arg 188 ), and this fragment persists even up to 2 h postdigestion, which indicates great stability of this domain when in complex with V 2 Rpp. Although our study does not show direct biophysical evidence of V 2 Rpp binding to the N-domain of ␤-arrestin1, our data are in accordance with previously published models based on crystal structures indicating that the more flexible N-domain of ␤-arrestin1 is responsible for recognizing the phosphate elements of 7TMRs, whereas the more rigid C-domain most likely serves as a structural scaffold for ␤-arrestin1 binding partners (16). The intact C-domain itself is never apparent as a separate entity once cleaved from the N-domain, which leads us to believe that a binding partner such as clathrin would be necessary to stabilize this domain and protect it from complete digestion. Our study represents the first direct evidence that the N-domain itself is in fact responsible for phosphate recognition.
One of the most valuable aspects of our in vitro model system is that it can be used to study conformational changes in ␤-arrestins with various phosphoreceptor peptides and, perhaps more interestingly, can also be used to study conformational differences between ␤-arrestin1 and -2. Numerous crystal structures have been solved for ␤-arrestin1, but the crystal structure of ␤-arrestin2, however, remains elusive, so any information on structural differences between the two isoforms will have to be garnered from direct biochemical assays. Conformational changes in ␤-arrestin2 with V 2 Rpp have been previously described from our laboratory (29). Briefly, this study similarly demonstrated the release of the C terminus of ␤-arrestin2 from the fulcrum of the molecule by accessibility of Arg 394 (homologous to Arg 393 of ␤-arrestin1) and disruption of the polar core of ␤-arrestin2 by increased accessibility of Arg 287 (homologous to Arg 285 of ␤-arrestin1). This previous study, however, contained no information on the tryptic fragments of ␤-arrestin2 below 30 kDa, which would preclude any information of individual N-and C-domains of ␤-arrestin2.
Accordingly, we directly compared tryptic fragments below 30 kDa for both ␤-arrestin1 and -2 (Fig. 7). The major protected species for ␤-arrestin1 is the N-domain (Leu 1 -Arg 188 ), and the major protected species for ␤-arrestin2 is both the N-and C-domains (a 42-kDa species) corresponding to residues Gly Ϫ7 -Arg 364 . Although similar N-domain fragments are observed for ␤-arrestin2 (25-and 21-kDa species), they are not more rapidly generated nor protected over time in the presence of V 2 Rpp. These data taken together suggest that although the overall activation mechanism is the same for ␤-arrestin1 and -2, the final conformations in the presence of V 2 Rpp are in fact different. We first need to exclude two obvious explanations for the observed differences in the active conformations, namely stoichiometric differences in peptide binding to the ␤-arrestins and primary sequence differences that would result in altered final conformations.
First, the stoichiometry of the binding of V 2 Rpp to ␤-arres-tin1 and -2 may be different. If, for example, ␤-arrestin2 were to bind two phosphopeptides, then it would make sense that both the N-and C-domains (Gly Ϫ7 -Arg 364 ) are protected, whereas ␤-arrestin1 binding a single peptide would explain protection of only the N-domain (Leu 1 -Arg 188 ). We have shown in this study, by limited proteolysis, that different molar ratios of V 2 Rpp:␤-arrestin1 do not alter the digestion pattern. A 1:1 molar ratio of V 2 Rpp:␤-arrestin1 is both necessary and sufficient to induce the observed conformational differences. Similarly, this same experiment in a previous study with ␤-arrestin2 demonstrated that increasing the molar ratio of V 2 Rpp:␤-arres-tin2 did not alter the digestion pattern (29). These data taken together suggest that a 1:1 molar ratio of V 2 Rpp to either ␤-ar-restin1 or -2 is sufficient to induce the active conformation. Moreover, we also performed native PAGE of ␤-arrestin1 and -2 in the presence of increasing amounts of V 2 Rpp and have determined via a mobility shift that a 1:1 molar ratio of V 2 Rpp: ␤-arrestin suffices for the same shift that we observe at a 5:1 molar ratio (data not shown). These data indicate that the stoichiometry of binding for V 2 Rpp to ␤-arrestin1 and -2 is in fact 1:1, and thus the conformational differences observed between the two isoforms is not because of a difference in the stoichiometry of peptide binding to the two ␤-arrestins.
Second, differences in the active conformation of ␤-arres-tin1 and -2 could simply be due to differences in their primary sequences. Fig. 2 shows the sequence alignment for recombinant rat ␤-arrestin1 and -2 used in this study. The two isoforms are 78% identical, and the two sequences are well conserved up to the residue numbered 340. We observed a 40-kDa fragment in all ␤-arrestin1 digests that corresponds to residues Leu 1 -Arg 363 . This fragment was protected in the early time points but was not the major protected species for ␤-arrestin1. In the case of ␤-arrestin2, cleavage was observed at Arg 364 producing a 42-kDa fragment that was significantly protected over time in the presence of V 2 Rpp. Although these two arginines in ␤-arrestin1 (Arg 363 ) and ␤-arrestin2 (Arg 364 ) do not align, we still see a somewhat similar pattern of digestion in this portion of their C termini. The striking difference, however, is that ␤-arres-tin2 digests show significant protection of the 42-kDa species, whereas ␤-arrestin1 digests do not show as dramatic a protection of the corresponding 40-kDa species (Leu 1 -Arg 363 ). We do not believe that differences seen in the digestion patterns of ␤-arrestin1 and -2 are because of primary sequence differences, and therefore, we conclude that the observed differences are in fact because of conformational differences in the final fold of the two proteins.
Because the major product in ␤-arrestin1 digestions in the presence of V 2 Rpp is Leu 1 -Arg 188 , we also inspected the sequence alignment of ␤-arrestin1 and -2 in this region (Fig.  2). The sequence of ␤-arrestin1 and -2 containing Arg 188 and Arg 189 , respectively, is just upstream of a flexible 12-residue interdomain hinge (boxed in Fig. 2) that connects the N-and C-domains. The primary sequence of the hinge region is very well conserved, and Arg 188 of ␤-arrestin1 aligns perfectly with Arg 189 of ␤-arrestin2. We do see tryptic fragments in the digestion of ␤-arrestin2 with V 2 Rpp that correspond to the N-domain, but these fragments are generated more slowly from the 42-kDa parent fragment and are only slightly protected over time. Conversely, in the digestion of ␤-arres-tin1 with V 2 Rpp, the N-domain is dramatically protected over time and persists up to 2 h post-digestion. The protection of the hinge region tryptic residues in the final fold of ␤-arrestin2 makes this region, specifically Arg 189 , less susceptible to tryptic proteolysis. In summary, the major protected species for ␤-arrestin1 in the presence of V 2 Rpp is the N-domain (Leu 1 -Arg 188 ), whereas for ␤-arrestin2 it is both the N-and C-domains (Gly Ϫ7 -Arg 364 ). These data suggest that the conserved hinge region becomes more accessible in ␤-arrestin1 (as evidenced by proteolysis at Arg 188 ) than ␤-arrestin2 because proteolysis at the corresponding Arg 189 is less rapid in the presence of V 2 Rpp. This study shows, for the first time, that the flexibility of this hinge region is in fact different for the two ␤-arrestin isoforms. Moreover, other studies, including mutagenesis and solved crystal structures, have suggested that the two domains of ␤-arrestin move relative to one another upon binding 7TMRs, and this hinge region is responsible for the movement of the two domains (9, 15-16, 18, 21, 25-27, 40). Interestingly, successive short-ening of the hinge region of the visual arrestin abolishes its ability to bind light-activated, phosphorylated rhodopsin (26). We have evidenced here that this movement in ␤-ar-restin1 and -2 is in fact different when bound to V 2 Rpp.
Because structure dictates function, it follows that the two isoforms have different final conformations because evidence exists demonstrating that ␤-arrestins1 and -2 are functionally nonredundant, and such individual, nonoverlapping roles in 7TMR regulation are very compatible with distinct receptorbound conformations. The first evidence to suggest that ␤-ar-restins1 and -2 are in fact not functionally redundant lies in their differential binding to 7TMRs. 7TMRs are broadly broken down into two classes, class A and B (37). Class A receptors, such as the ␤ 2 AR, have a higher affinity for ␤-arrestin2 than ␤-arrestin1; however, the interaction of these 7TMRs with ␤-arrestin is transient in that internalization of class A receptors leads to dissociation of the ␤-arrestin-receptor complex. Class B receptors, such as the AT 1a R and V 2 vasopressin receptor, display equal affinity for both ␤-arrestin isoforms and form more stable ␤-arrestin-receptor complexes. This preference for one isoform versus the other in receptor binding was the first evidence to suggest that ␤-arrestins do not serve redundant roles, but rather that the two isoforms have nonoverlapping, distinct functions.
In addition to differential receptor regulation, the two ␤-arrestin isoforms also display functional differences with nonreceptor partners. ␤-Arrestins are multiadaptor scaffold proteins that regulate 7TMR signaling, and multiple lines of recent evidence have shown that ␤-arrestins1 and -2 are also nonredundant in this function (reviewed in Ref. 41,42). Over a dozen nonreceptor partners have been shown to interact with ␤-arrestins,andthemajorityoftheseinteractionsoccurinanagonistdependent manner indicating that ␤-arrestins must be in the active or receptor-bound conformation to elicit these functions. The first evidence that ␤-arrestins play a role in 7TMR signaling stemmed from the discovery that ␤-arrestin1 specifically, and not ␤-arrestin2, interacts with c-Src, a nonreceptor tyrosine kinase (43,44).
Studies following the discovery of the role of ␤-arrestin1 in c-Src recruitment and ERK activation led to what is now a newly appreciated paradigm, the idea that ␤-arrestins may in fact play a role in scaffolding MAPK cascades. In terms of ERK signaling, it appears as though different 7TMRs prefer one ␤-arrestin isoform over the other. In the case of the protease-activated 2 receptor, it has been demonstrated that agonist stimulation leads to complex formation containing receptor, ␤-arrestin1, Raf-1, and active ERK (44). However, for angiotensin 1A receptor (AT 1a R) stimulation of ERK activation via ␤-arrestins, ␤-arrestin2 carries the signal, whereas ␤-arrestin1 actually functions antagonistically (4,33). It has also been recently reported that in some receptor systems such as the parathyroid hormone receptor (PTH1R), agonist stimulation of ERK through ␤-arrestin requires both isoforms (32). Thus, in the case of ␤-arrestin scaffolding of ERK via 7TMR stimulation, it appears as though different receptor systems, in general, display differential preference for either ␤-arrestin1 or -2, and in some cases both isoforms may be necessary. Although the ERK MAPK cascade has The Active Conformation of ␤-Arrestin1 emerged as the prototypical ␤-arrestin scaffold function studied, it has also been demonstrated that ␤-arrestins directly scaffold the activation of JNK3 and possibly indirectly p38 activation, although these systems have not been studied to the same extent as the role of ␤-arrestins in ERK signaling (34,35,45,46). Nonetheless, the scaffolding of the JNK3 MAPK cascade has only been shown for ␤-arrestin2, yet again indicating isoform specificity and functional nonredundance for the ␤-arrestins. Additionally, ␤-arrestin1, but not ␤-arrestin2, has been shown to signal coordinately with G␣ q/11 to activate RhoA, a small G-protein, through AT 1a R activation (30).
The different active conformations documented here for ␤-arrestins1 and -2 are thus consistent with numerous functional differences demonstrated by a multitude of studies. Furthermore, the differences in the conformational changes observed in both ␤-arrestins1 and -2 may in fact be due to differences in the flexibility of their interdomain hinge regions. This hypothesis could be easily tested by simply swapping the hinge regions of the two isoforms and determining if their functions in various systems are reversed. The model system used in this study provides an excellent means for determining conformational differences not only between ␤-arrestin1 and -2 upon interaction with a given 7TMR, but also in comparing the conformational differences that may occur when ␤-arrestin1 and -2 are bound to phosphopeptides from various 7TMRs, which in turn may aid in delineating why some receptor systems (as in the case of ERK activation) prefer one of the ␤-arrestins or both. ␤-Arrestins1 and -2 are clearly not functionally redundant, and this study provides the first evidence to suggest that these physiological, functional differences may in fact be due differences in their active conformations.