Activation-dependent Conformational Changes in β-Arrestin 2*

β-Arrestins are multifunctional adaptor proteins, which mediate desensitization, endocytosis, and alternate signaling pathways of seven membrane-spanning receptors (7MSRs). Crystal structures of the basal inactive state of visual arrestin (arrestin 1) and β-arrestin 1 (arrestin 2) have been resolved. However, little is known about the conformational changes that occur in β-arrestins upon binding to the activated phosphorylated receptor. Here we characterize the conformational changes in β-arrestin 2 (arrestin 3) by comparing the limited tryptic proteolysis patterns and matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) profiles of β-arrestin 2 in the presence of a phosphopeptide (V2R-pp) derived from the C terminus of the vasopressin type II receptor (V2R) or the corresponding nonphosphopeptide (V2R-np). V2R-pp binds to β-arrestin 2 specifically, whereas V2R-np does not. Activation of β-arrestin 2 upon V2R-pp binding involves the release of its C terminus, as indicated by exposure of a previously inaccessible cleavage site, one of the polar core residues Arg394, and rearrangement of its N terminus, as indicated by the shielding of a previously accessible cleavage site, residue Arg8. Interestingly, binding of the polyanion heparin also leads to release of the C terminus of β-arrestin 2; however, heparin and V2R-pp have different binding site(s) and/or induce different conformational changes in β-arrestin 2. Release of the C terminus from the rest of β-arrestin 2 has functional consequences in that it increases the accessibility of a clathrin binding site (previously demonstrated to lie between residues 371 and 379) thereby enhancing clathrin binding to β-arrestin 2 by 10-fold. Thus, the V2R-pp can activate β-arrestin 2 in vitro, most likely mimicking the effects of an activated phosphorylated 7MSR. These results provide the first direct evidence of conformational changes associated with the transition of β-arrestin 2 from its basal inactive conformation to its biologically active conformation and establish a system in which receptor-β-arrestin interactions can be modeled in vitro.

␤-Arrestins are multifunctional adaptor proteins, which mediate desensitization, endocytosis, and alternate signaling pathways of seven membrane-spanning receptors (7MSRs). Crystal structures of the basal inactive state of visual arrestin (arrestin 1) and ␤-arrestin 1 (arrestin 2) have been resolved. However, little is known about the conformational changes that occur in ␤-arrestins upon binding to the activated phosphorylated receptor. Here we characterize the conformational changes in ␤-arrestin 2 (arrestin 3) by comparing the limited tryptic proteolysis patterns and matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) profiles of ␤-arrestin 2 in the presence of a phosphopeptide (V 2 R-pp) derived from the C terminus of the vasopressin type II receptor (V 2 R) or the corresponding nonphosphopeptide (V 2 R-np). V 2 R-pp binds to ␤-arrestin 2 specifically, whereas V 2 R-np does not. Activation of ␤-arrestin 2 upon V 2 R-pp binding involves the release of its C terminus, as indicated by exposure of a previously inaccessible cleavage site, one of the polar core residues Arg 394 , and rearrangement of its N terminus, as indicated by the shielding of a previously accessible cleavage site, residue Arg 8 . Interestingly, binding of the polyanion heparin also leads to release of the C terminus of ␤-arrestin 2; however, heparin and V 2 R-pp have different binding site(s) and/or induce different conformational changes in ␤-arrestin 2. Release of the C terminus from the rest of ␤-arrestin 2 has functional consequences in that it increases the accessibility of a clathrin binding site (previously demonstrated to lie between residues 371 and 379) thereby enhancing clathrin binding to ␤-arrestin 2 by 10-fold. Thus, the V 2 R-pp can activate ␤-arrestin 2 in vitro, most likely mimicking the effects of an activated phosphorylated 7MSR. These results provide the first direct evidence of conformational changes associated with the transition of ␤-arrestin 2 from its basal inactive conformation to its biologically active conformation and establish a system in which receptor-␤-arrestin interactions can be modeled in vitro.
Seven membrane-spanning receptors (7MSRs), 1 also referred to as G protein-coupled receptors (GPCRs), constitute the largest known family of cell surface receptors (1,2). The human genome encodes ϳ1,000 7MSRs, which function primarily in the transmission of diverse signals (including light, odorants, chemoattractants, neurotransmitters, and hormones) from the extracellular environment to the interior of the cell (1,2). The dynamic sensitivity of 7MSR function is in large part a function of their regulation by the G proteincoupled receptor kinase (GRK)/␤-arrestin system (1,3). This regulation is accomplished by a two-step process involving the phosphorylation of the receptor, usually at its C terminus, by GRKs, and the subsequent binding of ␤-arrestins, which prevents further receptor activation of G proteins (desensitization) (1,4). ␤-Arrestin binding to the receptor also facilitates clathrin-mediated endocytosis (internalization) of the receptor (5)(6)(7).
In stark contrast to the multiplicity of 7MSRs, there are only four known isoforms of arrestin: visual arrestin (arrestin 1), cone arrestin, ␤-arrestin 1 (arrestin 2), and ␤-arrestin 2 (arrestin 3) (3,8). Visual arrestin and cone arrestin are expressed primarily in the eye and are almost exclusively involved in visual signaling processes. ␤-Arrestins 1 and 2 are ubiquitously expressed and are fundamental to the regulation of 7MSR-mediated signaling throughout the body. In addition to their well characterized role in 7MSR desensitization and internalization, recent evidence suggests that the ␤-arrestins also function as signaling transducers, which interact directly with a variety of effectors, such as non-receptor tyrosine kinases of the c-Src family and extracellular-regulated kinases (ERK1/2) and c-Jun N-terminal kinase (JNK3), in a receptor activation-dependent manner (3, 5-7, 9 -11). These receptor activation-dependent interactions of ␤-arrestins suggest that the conformations of free (basal or inactive) and receptor-bound (active) ␤-arrestins are different, and that conformational changes in ␤-arrestin occur upon binding to activated phosphorylated receptors.
Multiple lines of evidence, including mutagenesis and biochemical and biophysical studies, suggest that visual arrestin undergoes substantial conformational changes upon binding to light-activated phosphorylated rhodopsin (12,13). However, to date, there has been no direct evidence of conformational changes in ␤-arrestins. For example, the solved crystal structure of bovine ␤-arrestin 1 is of the molecule in its basal inactive conformation (14,15). Thus, the structural and molecular basis for ␤-arrestin activation as well as the associated conformational changes remains largely unknown.
7MSRs can be divided into two broad classes, "Class A" and "Class B," based on the nature of their interaction with ␤-arrestin (16). Class A receptors, such as the ␤ 2 -adrenergic receptor, interact with ␤-arrestin transiently, whereas Class B receptors, such as the V 2 R, form relatively stable receptor/␤-arrestin complexes. This increased stability of the interaction between the receptor and ␤-arrestin makes Class B 7MSRs, like V 2 R, ideal for studying the ability of a receptor to induce conformational changes in ␤-arrestin. Accordingly, in an effort to investigate the activation mechanism of ␤-arrestins, we used limited proteolysis and MALDI-TOF MS to characterize conformational changes in ␤-arrestin 2 upon binding a synthetic phosphopeptide (V 2 R-pp) derived from the C terminus of the Class B V 2 R to mimic receptor binding to ␤-arrestin 2.

EXPERIMENTAL PROCEDURES
Peptide Synthesis-The phosphopeptide (V 2 R-pp) and the corresponding nonphosphopeptide (V 2 R-np) derived from the C terminus of the human V 2 R were synthesized by the Protein Chemistry Core Laboratory of Baylor College of Medicine. The C-terminal sequence of V 2 R is ARGRT-PPSLGPQDESCTTASSSLAKDTSS, with the phosphorylated residues in V 2 R-pp underlined. Both phosphopeptide and nonphosphopeptide have been subjected to analytical HPLC, mass spectrometry, and amino acid analysis before their use in our in vitro binding experiments. The other two nonspecific peptides, a 28-mer WKKELTETFMEAQRLLRRAPK-FLNKSRS (termed 28-mer peptide) and a 30-mer NVWRPDGQMPD-DMKGVSGQEAAPSSKSGMC (termed 30-mer peptide), were synthesized previously by our laboratory.
To overexpress GST-␤-arrestin 2 in E. coli, 800 ml of LB medium were inoculated with 10 ml of overnight cell cultures and grown initially at 37°C until the OD 600 reached 0.6 -0.8. The temperature was then lowered to 25°C, and the expression of GST-␤-arrestin 2 was induced with 100 M isopropyl-1-thio-␤-D-galactopyranoside. After overnight incubation, cells were harvested by centrifugation at 4,500 ϫ g, and lysed by freeze-thaw followed by sonication in binding buffer (25 mM Tris-HCl, pH 8.5, 150 mM NaCl, 2 mM DTT, 2 mM EDTA, 2 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, and 0.2 mg/ml benzamidine). The lysate was centrifuged at 18,000 ϫ g for 30 min and the supernatant filtered through a 0.8-m membrane. The clarified supernatant was then loaded onto a glutathione-Sepharose column (Amersham Biosciences) by gravity and washed with 20 column volumes of binding buffer. The GST-␤-arrestin 2-bound glutathione Sepharose was resuspended in two column volumes of 25 mM Tris-HCl, pH 8.5, 150 mM NaCl, 2.5 mM CaCl 2 , 2 mM DTT. Thrombin protease (Sigma) (10 unit/ liter of cell culture) was added to cleave the GST protein tag (seven extra amino acids (GSPNSRV) remain after thrombin cleavage (Fig. 1)). The thrombin digestion was performed at 4°C overnight, and flow through was collected followed by two additional washes with the digestion buffer. Washes were pooled with the initial flow through and diluted to a final NaCl concentration of 50 mM. The sample was then loaded on a heparin-Sepharose column and eluted with a 50 -350 mM linear NaCl gradient. Fractions containing ␤-arrestin 2 were pooled and dialyzed overnight against 25 mM Tris-HCl, pH 8.5, 50 mM NaCl, 2 mM DTT, 2 mM EDTA, 2 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, and 0.2 mg/ml benzamidine. After dialysis, the sample was loaded onto a SP-Sepharose high performance column, and ␤-arrestin 2 was eluted with a 50 -350 mM linear NaCl gradient in the dialysis buffer. Fractions were analyzed by SDS-PAGE and pooled based on purity, concentrated to 5-10 mg/ml, flash frozen in liquid nitrogen, and stored at Ϫ80°C. Protein purity was more than 95% based on SDS-PAGE analysis, and typical yield was 1-2 mg of purified ␤-arrestin 2/liter of cell culture.
Cell growth conditions for the S-␤-arrestin 2 was the same as that for the GST-␤-arrestin 2. An S-Tag TM Thrombin purification kit (Novagen) was used to purify the S-␤-arrestin 2. Cells were lysed under the same conditions as that for the GST-␤-arrestin 2, and the cell lysate was incubated with S-protein beads for 2 h with agitation at 4°C, followed by washing with 20 column volumes of binding buffer. The ␤-arrestin 2-bound S-protein beads were then divided into two portions. One portion of the beads was digested with thrombin (10 unit/liter of cell culture), and the ␤-arrestin 2 was eluted and subjected to limited tryptic proteolysis. The other portion of the beads was used directly for in vitro clathrin binding assays.
Limited Tryptic Proteolysis-A 5:1 molar ratio of ligand (V 2 R-pp, V 2 R-np, the other two nonspecific peptides or heparin) to ␤-arrestin 2 (0.5-1 mg/ml) was used to reveal the effects of ligand on limited tryptic proteolysis of ␤-arrestin 2 in all experiments except where otherwise indicated. An average molecular mass of 12,000 Da was used to calculate the concentration of heparin (Sigma). Prior to proteolysis, ␤-arrestin 2, in the absence or presence of ligand, was incubated at room temperature for 30 min. An appropriate amount of TPCK-treated trypsin (Sigma) was added to the mixture for limited proteolysis. The samples were incubated at 37°C for indicated time points. At each time point, 5 l (2.5-5 g of ␤-arrestin 2) was removed from each reaction. For SDS-PAGE analysis, each sample was transferred to a new microcentrifuge tube containing 5 l of 2ϫ SDS-PAGE buffer, and boiled for 5 min to quench the tryptic digestion. The samples were run on 4 -20% SDS-PAGE gels (Invitrogen) to determine the effects of ligands on the digestion pattern of ␤-arrestin 2. For MALDI-TOF MS analysis, the samples were transferred to new empty microcentrifuge tubes and flash frozen for MALDI-TOF analysis to measure the molecular masses for the proteolytic fragments.
MALDI-TOF MS-MALDI-TOF mass spectra were acquired on a Voyager DE Biospectrometry Work station (Applied Biosystems) in the linear mode using a nitrogen laser (337 nm). Mass spectra were collected in the positive ion mode using an acceleration voltage of 25 kV and a delay of 900 ns. The grid voltage, guide wire voltage, low mass gate, and laser intensity were set to 92.5%, 0.15%, 10,000.0 m/z, and 2580, respectively. Each mass spectrum collected represents the sum of the data from 75 laser shots.
Sinapinic acid (SA) (Sigma) was used as the matrix and the SA matrix solution was prepared as a saturated, aqueous solution that contained 45% acetonitrile and 0.1% trifluoroacetic acid. During MALDI-TOF sample preparation, 1 l of flash frozen limited tryptic proteolysis products was mixed with 9 l of SA matrix solution before depositing 1 l (1-2 M final protein concentration) of the sample matrix mixture on the MALDI-TOF sample plate. The sample was allowed to air-dry at room temperature and then subjected to MALDI-TOF analysis. Cytochrome c and carbonic anhydrase were used as internal calibrants for data calibration. For each limited tryptic proteolytic fragment, the mean of the experimental molecular mass (m/z) was determined, and a standard deviation was calculated from six independent experiments. ␤-Arrestin 2 was also subjected to theoretical MALDI-TOF MS analysis of the molecular masses for all potential tryptic proteolytic fragments using a ProteinProspector program (prospector.ucsf.edu/). The experimental molecular mass (m/z) of each limited proteolytic fragment was compared with the theoretical masses (m/z) predicted by ProteinProspector program, and all the candidate fragments with theoretical masses (m/z) within the standard deviations were selected (Table I). The limited proteolytic fragments with only one possible theoretical mass (m/z) within standard deviation were assigned to individual proteolytic fragments directly, with or without further confirmation of the assignments by N-terminal sequencing or by other methods. The fragments with more than one possible theoretical mass (m/z), which could not be distinguished by MALDI-TOF MS, were designated by coupling the MS data with N-terminal sequencing, Western blot analysis with antibodies that recognize different domains of ␤-arrestin 2, and limited tryptic proteolysis of truncated ␤-arrestin 2 mutants.
N-terminal Sequencing-The limited proteolytic products were subjected to SDS-PAGE analysis. The proteolytic fragments of ␤-arrestin 2 were then electrotransferred to a polyvinylidene difluoride membrane. Proteins bound to polyvinylidene difluoride membrane were stained by Coomassie Blue R-250, and the membrane was washed with Millipore water for 5 min and air-dried. The stained protein bands were excised with a clean razor blade, placed in a 1.5-ml conical centrifuge tube and subjected to N-terminal sequencing. The N-terminal sequencing was performed by the Molecular Biology Resource Facility of University of Oklahoma Health Science Center.
Clathrin Binding-To measure clathrin binding to ␤-arrestin 2, 25 l of ␤-arrestin 2 S-protein beads (containing 20 g of S-␤-arrestin 2), in the absence or presence of 5:1 (ligand:␤-arrestin 2) molar ratio of ligand (V 2 R-pp, V 2 R-np, 28-mer peptide or heparin) were incubated in binding buffer (25 mM Tris-HCl, pH 8.5, 150 mM NaCl, 2 mM EDTA, 1 mM DTT) at room temperature for 30 min. After the incubation, the volume of each mixture was brought up to 500 l with binding buffer, followed by the addition of 5 g of clathrin heavy chain (Sigma) and agitated for 2 h at 4°C. The ␤-arrestin 2 bound S-protein beads were then centrifuged at 20,000 ϫ g in a benchtop microcentrifuge, washed with 1 ml of binding buffer 5 times, and incubated with 25 l of 2ϫ SDS-PAGE buffer. Clathrin binding to ␤-arrestin 2 was measured by Western blot analysis with an anti-clathrin antibody (Transduction Laboratories). Blots were re-probed with an anti-␤-arrestin 2 antibody (A2CT) to ensure equal loading of ␤-arrestin 2 for each reaction.

RESULTS AND DISCUSSION
Conformational Changes in ␤-Arrestin 2 in Vitro upon Its Specific Association with Phosphopeptide V 2 R-pp-Rat ␤-arrestin 2 was expressed and purified from E. coli strain BL21 (DE3) pLysS as described under "Experimental Procedures." The primary sequence of the purified recombinant rat ␤-arrestin 2 cleaved from GST-␤-arrestin 2 is shown in Fig. 1. To investigate the ␤-arrestin 2 activation mechanism by 7MSR and the associated conformational changes, the purified rat ␤-arrestin 2 was subjected to limited proteolysis in the absence or presence of a synthetic phosphopeptide, V 2 R-pp, or the corresponding nonphosphopeptide, V 2 R-np, derived from the C terminus of the Class B receptor V 2 R. Among several proteases tested in this study, TPCK-treated trypsin yields the best resolution of the digestion fragments. The optimal w/w ratio of trypsin to ␤-arrestin 2 for resolution of digestion fragments by SDS-PAGE was determined to range from 1:2,000 to 1:5,000 at 37°C.
The limited proteolysis of ␤-arrestin 2 in the absence of peptide and in the presence of 5:1 (peptide/␤-arrestin 2) molar ratio of V 2 R-np results in identical digestion patterns on SDS-PAGE (panels I and II in Fig. 2A), which we term the "control pattern." As illustrated in the control pattern panel of Fig. 2B, full-length ␤-arrestin 2 (residues Gly Ϫ7 -His 416 ) appears at 48 kDa and digestion with trypsin results in fragments with the apparent molecular masses of 42, 33, and 32 kDa (this study focused only on these major fragments). However, in the presence of 5:1 (peptide/␤-arrestin 2) molar ratio of V 2 R-pp, the limited proteolysis pattern of ␤-arrestin 2 is altered and represented as "V 2 R-pp pattern" (panel III of Fig. 2A). Within 15 min of proteolysis in the presence of V 2 R-pp, a new fragment with an apparent molecular mass of 45 kDa is generated, whereas the prominent 32-kDa fragment seen in the control pattern is not generated (V 2 R-pp pattern of Fig. 2B). An identical V 2 R-pp pattern is also obtained when ␤-arrestin 2 cleaved from S-␤arrestin 2 is used (data not shown). Limited tryptic proteolysis of ␤-arrestin 2 was also conducted in the presence of two other nonspecific peptides, a 28-mer peptide and a 30-mer peptide (the sequences are shown under "Experimental Procedures"). Neither peptide had any effect on the limited proteolysis pattern of ␤-arrestin 2.
Trypsin is a pancreatic serine protease that cleaves peptide bonds in proteins that have carbonyl groups donated by arginine or lysine. A change in protein conformation can mask or unmask such cleavage sites, and an alteration in the limited proteolysis pattern of a protein can be indicative of conformational changes (17)(18)(19)(20)(21)(22)(23). Thus, the change in digestion pattern of ␤-arrestin 2 in the presence of V 2 R-pp implies that V 2 R-pp binds to ␤-arrestin 2 specifically in vitro and induces conformational changes. V 2 R-np does not alter the digestion pattern of ␤-arrestin 2, suggesting that V 2 R-np either cannot induce conformational changes upon binding or cannot bind ␤-arrestin 2 (see below).
Requirement for Phosphorylation on Residues in Synthetic Peptide for Inducing Conformational Changes in ␤-Arrestin 2 in Vitro-We have shown that V 2 R-np cannot induce changes in the pattern of limited tryptic proteolysis of ␤-arrestin 2 at a 5:1 (peptide/␤-arrestin 2) molar ratio. This could be caused by a lower affinity of V 2 R-np for ␤-arrestin 2. We therefore examined the effect of different peptide to ␤-arrestin 2 molar ratios on the limited tryptic proteolysis pattern of ␤-arrestin 2. As shown in the upper panel of We next performed a competition binding experiment of V 2 R-pp and V 2 R-np to ␤-arrestin 2 (Fig. 3B). ␤-Arrestin 2 was incubated, simultaneously, with a 1:1 (peptide/␤-arrestin 2) molar ratio of V 2 R-pp and with different ratios of excess V 2 R-np prior to limited tryptic proteolysis. A 1:1 molar ratio of V 2 R-pp induces a V 2 R-pp pattern (lane 4), whereas even a 100-fold excess of V 2 R-np cannot compete with the binding of V 2 R-pp to ␤-arrestin 2 (lane 6). This further supports the hypothesis that V 2 R-pp binds specifically to ␤-arrestin 2 and that the V 2 R-np binds either very weakly or not at all.
We also tested the effects of salt concentration ([NaCl]) on limited tryptic proteolysis of ␤-arrestin 2 in the presence of V 2 R-pp (data not shown). Our data indicate that the association of V 2 R-pp with ␤-arrestin 2 is salt concentration-dependent. At low NaCl concentration, V 2 R-pp binds to ␤-arrestin 2 and causes a V 2 R-pp pattern of limited proteolysis. When the NaCl concentration increases to 550 mM, the association of V 2 R-pp with ␤-arrestin 2 is disrupted and the proteolysis pattern returns to a control pattern, indicating that electrostatic interactions may play a crucial role in the association of V 2 R-pp with the positively charged residues in ␤-arrestin 2. Thus phosphorylation of the synthetic peptide is required for inducing conformational changes in ␤-arrestin 2 in vitro, consistent with the notion that ␤-arrestin 2 binds to 7MSRs in a phosphorylationdependent fashion (3,8,11,24,25).
Involvement of Polar Core Residue Arg 394 and Both N Terminus and C Terminus of ␤-Arrestin 2 in the Conformational Changes Induced by V 2 R-pp-The above SDS-PAGE analysis of limited tryptic proteolysis products clearly indicates that V 2 R-pp induces conformational changes in ␤-arrestin 2 upon its FIG. 3. The effects of different peptide to ␤-arrestin 2 molar ratios on the limited tryptic proteolysis of rat ␤-arrestin 2 (A) and competitive binding of V 2 R-pp and V 2 R-np to ␤-arrestin 2 (B). A, ␤-arrestin 2 was incubated without (labeled 0) or with different peptide to ␤-arrestin 2 molar ratios (labeled 1:1 to 100:1) of V 2 R-pp (upper panel) or V 2 R-np (lower panel) for 30 min at room temperature before limited proteolysis. STD is the MagicMark TM XP Molecular Weight Standards. B, ␤-arrestin 2 was simultaneously incubated with 1:1 (peptide:␤-arrestin 2) molar ratio of V 2 R-pp and different molar ratios of V 2 R-np (lanes 4 -6) at room temperature for 30 min prior to addition of trypsin. STD is the rainbow standard (Amersham Biosciences). In both A and B, a 1:5,000 (trypsin/␤-arrestin 2) w/w ratio of TPCK-treated trypsin was added to the sample, and the limited proteolysis was performed at 37°C for 15 min. The apparent molecular masses of the tryptic digestion fragments are labeled on the left of each panel.
binding. However, SDS-PAGE analysis can only provide a molecular mass estimate of the proteolytic fragments. To map the domains in ␤-arrestin 2 involved in conformational changes induced by V 2 R-pp, the proteolytic fragments were analyzed by MALDI-TOF MS. The spectra of ␤-arrestin 2 with V 2 R-np are identical to the spectra of ␤-arrestin 2 without any added peptide. Fig. 4 shows the MALDI-TOF spectra of the proteolytic fragments of ␤-arrestin 2 between m/z 38,000 and 52,000 at time points of 15 and 60 min. In the presence of V 2 R-np, at 15 min, there are two peaks with m/z values of 47857 Ϯ 23 and 41646 Ϯ 18 (Table I and (Table I and the V 2 R-np panels of Figs. 4 and 5). In addition to the two peaks seen with V 2 R-np, in the presence of V 2 R-pp, a peak with m/z value of 45,158 Ϯ 30, corresponding to the 45-kDa fragment on SDS-PAGE, is detected at 15 min. However, at 60 min a peak with an m/z value of 40,102 Ϯ 21 is not observed (Table I and  Peaks of the proteolytic fragments in MALDI-TOF spectra between m/z 30,000 and 38,000 are not apparent at time points of 15 and 60 min. However, they are apparent at time points of 30 min (Fig. 5). In the presence of V 2 R-np, two peaks with m/z values of 33,318 Ϯ 22 and 31,759 Ϯ 24, corresponding to the 33and 32 kDa-fragments on SDS-PAGE, are present in the spectrum (Fig. 5). Surprisingly, in the presence of V 2 R-pp, the m/z 31,759 Ϯ 24 peak is not detected (Fig. 5 and Table I).
In order to assign each proteolytic fragment of ␤-arrestin 2, the full-length recombinant ␤-arrestin 2 (residues Gly Ϫ7 -His 416 ) was subjected to a theoretical limited tryptic proteolysis using a ProteinProspector program as described under "Experimental Procedures." The theoretical masses (m/z) of the proteolytic fragment candidates within the standard deviation of the experimental mass (m/z) of each fragment were selected and listed in Table I. The fragments (residues Gly Ϫ7 -His 416 , Gly Ϫ7 -Arg 364 , Val 9 -Arg 364 , and Val 9 -Arg 287 ) with only one possible theoretical mass (m/z) were assigned directly, with or without further confirmation of the assignments by N-terminal sequencing or by other methods (Table I and Fig. 5). The fragments with more than one possible theoretical mass (m/z), which could not be distinguished by MALDI-TOF MS, were designated by coupling the MS data with N-terminal sequencing, Western blot analysis with antibodies that recognize different domains of ␤-arrestin 2, and limited tryptic proteolysis of truncated ␤-arrestin 2 mutants. The fragment with m/z value of 45,158 Ϯ 30, generated in the presence of V 2 R-pp, could be assigned to two possible theoretical proteolytic fragments, residues Gly Ϫ7 -Arg 394 (C-terminal cleavage) or Asp 19 -His 416 (N-terminal cleavage) ( Table I). The five N-terminal amino acids of this proteolytic fragment were sequenced as GSPNS, indicating that this fragment is the C-terminal cleavage fragment, residues Gly Ϫ7 -Arg 394 , from full-length ␤-arrestin 2 (residues Gly Ϫ7 -His 416 ). This assignment was further confirmed by the limited proteolysis of two C-terminal-truncated mutants, ␤-arrestin 2 (residues Gly Ϫ7 -Thr 383 ) and ␤-arrestin 2 (residues Gly Ϫ7 -Arg 394 ) (Fig. 6). In the presence of V 2 R-pp, no protein band, with an apparent molecular mass 3 kDa less than those of the truncated ␤-arrestin 2 bands, was detected on SDS-PAGE. This result supports that the fragment with m/z value of 45,158 Ϯ 30 (corresponding to the 45-kDa band on SDS-PAGE) from the limited proteolysis of wild-type ␤-arrestin 2 is a proteolytic fragment Gly Ϫ7 -Arg 394 . Appar- ently, in the presence of V 2 R-pp, a previously hidden tryptic cleavage site at residue Arg 394 is exposed and becomes accessible to trypsin (Fig. 5). The fragment with m/z 41,039 Ϯ 27 (with m/z 41,042 Ϯ 23 without peptide and 41,062 Ϯ 15 in the presence of V 2 R-pp, respectively) has two possible assignments, amino acid Val Ϫ1 -Arg 364 and Val 35 -Lys 398 . With our experimental conditions, we were not able to distinguish these two possibilities. However, this fragment is most likely a derivative from fragment Gly Ϫ7 -Arg 364 because of a possible cleavage site at position Arg Ϫ2 . The fragment with m/z 33,318 Ϯ 22 (with m/z 33,315 Ϯ 12 without peptide and 33,317 Ϯ 15 in the presence of V 2 R-pp, respectively) has two possible assignments, amino acid Gly Ϫ7 -Arg 287 and Leu 101 -Arg 396 , which have very close molecular masses (m/z) and cannot be differentiated by MALDI-TOF MS. However, this fragment was recognized by a mouse monoclonal antibody (26) (F4C1) directed against the fragment Asp 39 -Asp 45 (sequence DGVVLVD) of ␤-arrestin 2 (data not shown), indicating that this fragment is residues Gly Ϫ7 -Arg 287 .
The two peaks with m/z values of 40,102 Ϯ 21 and 31,759 Ϯ 24, which are not generated in the presence of V 2 R-pp, were assigned directly as proteolytic fragments residues Val 9 -Arg 364 and Val 9 -Arg 287 , respectively (Table I and Fig. 5). These data suggest that a previously accessible cleavage site, residue Arg 8 , is protected from trypsin digestion upon V 2 R-pp binding (Fig. 5). Protection of residue Arg 8 could result either from direct masking by the bound V 2 R-pp or from ligand-induced changes in conformation of the ␤-arrestin 2 N terminus and subsequent change in solvent accessibility of residue Arg 8 . The latter explanation seems more rational for two reasons. First, in the crystal structures of visual arrestin and ␤-arrestin 1 and the structural model of ␤-arrestin 2 (Fig. 5), Arg 8 (or the corresponding residue in visual arrestin and ␤-arrestin 1) is located on the convex side, away from the concave surface, which is believed to be the docking site for receptor. This location is not favorable for the interaction of Arg 8 with V 2 R-pp. Second, evidence suggests that Lys 14 and Lys 15 of visual arrestin (equivalent to Lys 11 and Lys 12 of rat ␤-arrestin 2, which are only 2 and 3 residues away from Arg 8 ) are the major possible "phosphate sensors" and directly interact with receptor-attached phosphate moieties. Charge reversal and elimination of residues Lys 14 and Lys 15 of visual arrestin dramatically reduces arrestin binding to light-activated phosphorylated rhodopsin (27). Thus, it is most likely that V 2 R-pp directly interacts with Lys 11 and Lys 12 in ␤-arrestin 2, forcing a rearrangement of its N terminus. This rearrangement moves Arg 8 away from its original solvent accessible position to a solvent inaccessible position.
Arrestin family members are comprised of two domains (Ndomain and C-domain) and an extended C-terminal tail (14,15,28,29). In a recently proposed model, the inactive conformation of arrestins is stabilized by an intact polar core, which includes residues Asp 26 , Arg 169 , Lys 170 , Asp 290 , Asp 297 , and Arg 393 of bovine ␤-arrestin 1 (Asp 27 , Arg 170 , Lys 171 , Asp 292 , Asp 299 , and Arg 394 of rat ␤-arrestin 2), as well as a three element hydrophobic interaction involving ␤-strand I of the N terminus, the last ␤-strand XX of the C terminus, and ␣-helix I (15). Upon activation-dependent phosphorylation of a 7MSR, multiple phosphate moieties at the C terminus of the receptor would disrupt the delicate charge network within the polar core, thus releasing the constraints that keep arrestin in the basal state. Our data indicate that, without V 2 R-pp binding, the ␤-arrestin 2 molecule is in an inactive conformation in which the residue Arg 394 is an integral part of the polar core and therefore inaccessible to trypsin. Binding of V 2 R-pp induces conformational changes in ␤-arrestin 2 and the previously hidden tryptic cleavage site Arg 394 is exposed and becomes accessible to trypsin (Fig. 5). Because the polar core residue Arg 394 is located in the C terminus of ␤-arrestin 2, the exposure of residue Arg 394 prompts us to hypothesize that the C terminus is released from the rest of the protein molecule during the transition of ␤-arrestin 2 from its inactive conformation to the active conformation. The findings of: 1) exposure of the polar core residue Arg 394 and 2) involvement of conformational changes in both N terminus and C terminus of ␤-arrestin 2 during its activation by V 2 R-pp provide the first direct evidence for the conformational changes associated with ␤-arrestin 2 activation.
Comparison of V 2 R-pp-induced Conformation versus Heparin-induced Conformation of ␤-Arrestin 2-Polyanions, like heparin, have previously been shown to induce conformational changes in visual arrestin similar to those induced by the photoactivated phosphorylated rhodopsin as well as by a synthetic phosphopeptide comprising the fully phosphorylated Cterminal region of rhodopsin (20,30,31). This tempted us to examine the effect of heparin on the conformational changes in ␤-arrestin 2. Limited proteolysis of ␤-arrestin 2 in the absence or presence of V 2 R-np, V 2 R-pp or heparin was performed in parallel, and the digestion patterns were compared (Fig. 7). Heparin binding changes the digestion pattern of ␤-arrestin 2, inducing conformational changes in ␤-arrestin 2. Addition of heparin, leads to release of the C terminus of ␤-arrestin 2, as indicated by the generation of the 45-kDa fragment. In this aspect, heparin acts similar to V 2 R-pp. However, striking differences are apparent in the overall digestion patterns induced by V 2 R-pp and heparin (Fig. 7). The 42 kDa protein is dramatically protected upon V 2 R-pp binding, whereas the digestion of the 42 kDa protein as well as the full-length ␤-arrestin 2 are accelerated in the presence of heparin. Possible explanations for this difference are that V 2 R-pp and heparin may have strated that ␤-arrestins function as adaptor proteins in clathrin-mediated endocytosis to promote agonist-induced internalization of 7MSRs (5). ␤-Arrestins have been reported to directly interact with clathrin heavy chain with high affinity via in vitro binding methods (5). The clathrin binding domain was localized to the C terminus of the ␤-arrestins, and the predominant binding sites were further demonstrated to lie between residues 371 and 379 of ␤-arrestin 2 (a clathrin binding motif, On the left is the SDS-PAGE gel of the limited proteolysis of ␤-arrestin 2 in the presence of V 2 R-np or V 2 R-pp. The amino acid assignment of each band was labeled on the right of the gel. The structural model of ␤-arrestin 2 was generated by HyperChem software using structures of bovine visual arrestin and ␤-arrestin 1 as templates (PDB files: 1G4R.pdb, 1G4M.pdb, 1CF1.pdb, and 1JSY.pdb). The structural figure was made from ViewerLite4.2 (www.accelrys.com/) and POV-Ray v3.6 (www.povray.org/). residues DTNLIEFDT) (32). To test the functional consequences and biological implications of the conformational changes which occur upon ␤-arrestin 2 induced by the V 2 R-pp, in vitro clathrin binding to S-␤-arrestin 2 (S-tagged ␤-arrestin 2) was examined (Fig. 8). Limited tryptic proteolysis of purified ␤-arrestin 2 cleaved from S-␤-arrestin 2 in the presence of V 2 R-np, V 2 R-pp or heparin shows that it behaves similarly to ␤-arrestin 2 purified from GST-␤-arrestin 2 and that the phosphopeptide induces the same conformational changes, as demonstrated by SDS-PAGE and MALDI-TOF MS analysis of the limited proteolysis products (data not shown). Prior to the addition of clathrin heavy chain, S-␤-arrestin 2 was incubated in the presence or absence of 5:1 (ligand/␤-arrestin 2) molar ratio of 28-mer control peptide, V 2 R-np, V 2 R-pp or heparin. We found that V 2 R-pp binding enhanced clathrin binding to S-␤arrestin 2 by 10-fold (Fig. 8), whereas 28-mer control peptide and V 2 R-np had no effect. Interestingly, heparin binding also enhanced clathrin binding to S-␤-arrestin 2 to 89 Ϯ 15.5% of that observed with V 2 R-pp. This result not only supports our hypothesis that the C terminus is released from the rest of the ␤-arrestin 2 molecule upon its activation by V 2 R-pp and heparin, but also implies that the release of the C terminus of ␤-arrestin 2 is sufficient to enhance clathrin binding.
Previous studies indicate that although heparin and the phosphopeptide from the C terminus of rhodopsin can induce similar conformational changes in visual arrestin, they have distinct effects on the interactions between arrestin and rhodopsin (23,30). Heparin inhibits visual arrestin binding to rhodopsin (30), whereas the phosphopeptide from rhodopsin enhances light-activated binding of arrestin to both unphosphorylated rhodopsin in disk membranes as well as to endoproteinase Asp-N-treated rhodopsin (deletion of 330 -348) (23). Our data reveal that both V 2 R-pp and heparin-induced conformations of ␤-arrestin 2 are functionally active in terms of clathrin binding. However, based on the proteolysis data (Fig. 7), we believe that the conformations induced by heparin and V 2 R-pp are not identical. The V 2 Rpp-induced conformation is more likely to mimic the conformation induced by the activated receptor.
The findings in our study on ␤-arrestin 2 activation suggest that a similar activation mechanism is likely conserved among different members of the arrestin family. Based on our data and other mutagenesis, biochemical studies, and the crystal structures of the basal inactive state of bovine visual arrestin and ␤-arrestin 1, we propose a possible ␤-arrestin 2 activation model (Fig. 9), which is similar to the model proposed for visual arrestin and ␤-arrestin 1 (14,15). Upon binding to V 2 R-pp or agonist-occupied phosphorylated receptor, the multiple phosphate moieties of V 2 R-pp or the C terminus of the receptor would disrupt the delicate charge network within the polar core, thus releasing the constraints that keep ␤-arrestin 2 in the basal inactive state. This induces the dissociation of the N terminus and C terminus away from their original positions and dislodges Arg 394 from the polar core. Additional global rearrangements of both N-domain and C-domain are most likely involved in this activation process. The release of the C terminus of ␤-arrestin 2 upon its activation exposes the clathrin binding site, and thus promotes clathrin binding to the active ␤-arrestin 2, favoring clathrin-mediated endocytosis of 7MSRs (Fig. 9).  7. Comparison of the effects on ␤-arrestin 2-limited proteolysis of heparin and that of V 2 R-pp. The limited proteolytic products of rat ␤-arrestin 2 without peptide (A), with 5:1 molar ratio (ligand/␤-arrestin 2) of V 2 R-np (B), V 2 R-pp (C), or heparin (D) at different time points were analyzed by SDS-PAGE. A 1:5,000 (w/w) ratio of TPCK-treated trypsin to ␤-arrestin 2 was used, and the tryptic digestion was performed at 37°C. The apparent molecular masses of the tryptic digestion fragments are labeled on the right of each panel.
In summary, our study reports the activation of ␤-arrestin 2 by the binding of a phosphopeptide (V 2 R-pp) derived from the C terminus of V 2 R. This active conformation is different from the basal conformation and heparin-induced conformation, as indicated by the different limited tryptic proteolysis patterns. It is also functional, indicated by the enhancement of clathrin binding. The conformational changes of ␤-arrestin 2 upon activation by V 2 R-pp are consistent with the predicted changes in conformation of visual arrestin and ␤-arrestin 1 upon binding to receptor as suggested from previous mutagenesis, biochemical, and biophysical studies. This, in conjunction with the clathrin binding data, implies that the active conformation of ␤-arrestin 2 induced by V 2 R-pp binding may closely mimic the functionally and physiologically active conformation of ␤-arrestin 2 which is induced upon binding to activated phosphorylated 7MSRs. It seems possible that ␤-arrestin 1 might display FIG. 8. Enhancement of clathrin binding to ␤-arrestin 2 upon binding V 2 R-pp. The experiments were performed as described under "Experimental Procedures." Panel A, the reaction components; panel B, the immunoblot for clathrin bound to S-␤-arrestin 2; panel C, the immunoblot for S-␤-arrestin 2 to measure for equal sample loading in each reaction; panel D, the quantification of clathrin bound to S-␤-arrestin 2 by densitometry. The nonspecific binding of clathrin to the S-protein beads in the absence of S-␤-arrestin 2 was subtracted from specific binding. The clathrin binding data were normalized to the binding for S-␤-arrestin 2 in the presence of V 2 R-pp (100%), and the relative percentages of clathrin binding were indicated above the bars, representing the mean Ϯ S.E. of at least three independent experiments. 28-mer is the 28-mer nonspecific control peptide.
FIG. 9. Schematic model of activation-dependent conformational changes of ␤-arrestin 2 upon V 2 R-pp binding and subsequent binding of clathrin. The N-domain (N) of ␤-arrestin 2 is in blue and the C-domain (C) is in green. The N terminus and C terminus of ␤-arrestin 2 is shown in yellow and black, respectively. The clathrin binding site in the C terminus of ␤-arrestin 2 is marked as a red square. Three trypsin cleavage sites (Arg 8 , Arg 364 , and Arg 394 ) are also indicated (1). Binding of V 2 R-pp induces conformational changes in ␤-arrestin 2. The conformational changes involve the release of the C terminus from the rest of the ␤-arrestin 2, increasing the accessibility of a clathrin binding site (residues Asp 371 -Thr 379 ). (2) ␤-arrestin 2 in the activated conformation displays enhanced clathrin binding. similar conformational changes. Recent research has suggested that the two ␤-arrestin isoforms are functionally non-redundant. Our study provides a way to delineate the conformational changes of different isoforms of arrestin induced by the C termini of different 7MSRs. Future studies using this in vitro approach should provide valuable information regarding the activation mechanism of ␤-arrestins and shed new light on their shared as well as distinctive molecular features.