N-Formyl Peptide Receptor Phosphorylation Domains Differentially Regulate Arrestin and Agonist Affinity*

Arrestins regulate the signaling and endocytosis of many G protein-coupled receptors (GPCRs). It has been suggested that the functions of arrestins are dependent upon both the number and pattern of phosphorylation sites present in an activated GPCR. However, little is currently known about the relationships between the sites of receptor phosphorylation, the resulting affinities of arrestin binding, and the ensuing mechanisms of receptor regulation for any given GPCR. To investigate these interactions, we used an active truncated mutant of arrestin (amino acids 1–382) and phosphorylation-deficient mutants of the N-formyl peptide receptor (FPR). In contrast to results with wild type arrestins, the truncated arrestin-2 protein bound to the unphosphorylated wild type FPR, although with lower affinity and a low affinity for the agonist as revealed by competition studies with heterotrimeric G proteins. Using FPR mutants, we further demonstrated that the phosphorylation status of serines and threonines between residues 328–332 is a key determinant that regulates the affinity of the FPR for arrestins. Furthermore, we found that the phosphorylation status of serine and threonine residues between amino acids 334 and 339 regulates the affinity of the receptor for agonist when arrestin is bound. These results suggest that the agonist affinity state of the receptor is principally regulated by phosphorylation at specific sites and is not simply a consequence of arrestin binding as has previously been proposed. Furthermore, this is the first demonstration that agonist affinity of a GPCR and the affinity of arrestin binding to the phosphorylated receptor are regulated by distinct receptor phosphodomains.

The regulation of G protein-coupled receptors (GPCRs) 1 is a high fidelity mechanism that involves, at a minimum, receptor phosphorylation, arrestin binding, and internalization. Agonist activation leads to the rapid phosphorylation of serines and threonines in the GPCR carboxyl terminus and/or intracellular loops through the activity of a G protein-coupled receptor kinase (GRK) (1)(2)(3)(4)(5). Phosphorylation partially quenches heptahelical signaling, as heterotrimeric G proteins display lower affinity for certain phosphorylated receptors (6,7). Subsequent binding of arrestins to activated phosphorylated GPCRs further desensitizes them as a result of the steric preclusion of G protein binding. In addition, arrestin binding can direct the translocation of phosphorylated GPCRs to the endocytic machinery (8 -10) as well as target numerous kinases to GPCR-based scaffolds (9,(11)(12)(13)(14)(15). Many GPCRs resensitize over time in a process presumably dependent upon ligand and arrestin dissociation and the actions of phosphatase(s) (16 -18). Notwithstanding this basic model, there is great diversity in the processing of GPCRs with respect to sites of phosphorylation and the ensuing mechanisms of desensitization and internalization (8). Thus, it is likely that multiple biological models are necessary to account for the complex mechanisms of GPCR processing.
Surprisingly little is known regarding arrestin's mechanism of activation through its interactions with GPCRs and the reciprocal effects of arrestin binding on receptor properties. It has been suggested that the ionic interactions of polar residues in the "phosphate-sensing" core of arrestin with phosphoamino acids of an activated G protein-coupled receptor are required for stable complex formation (18 -22). Presumably, phosphoreceptor binding disrupts a delicate charge balance and causes reorientation of the inactive arrestin molecule around a "fulcrum" in the molecule (20,23). It has been put forward that the binding of activated arrestins to phosphorylated receptors preferentially stabilizes a receptor conformation that displays enhanced ligand affinity (7,24). Despite these advances, a single model that accounts for the coordinated actions of multiple receptor and arrestin domains in terms of complex formation and agonist affinity regulation has yet to be described. Furthermore, the precise correlation between the type, number, and location of receptor phosphorylation sites and their role in determining arrestin binding affinity as well as the regulation of receptor ligand binding affinity has yet to be elucidated.
Carboxyl-terminal truncation of arrestin-2 has been characterized as an "activating" mutation (19,(25)(26)(27). It is believed that the tail of arrestin stabilizes the basal inactive state by intramolecular interactions with the polar core domain. Truncation reduces the activation energy of visual arrestin by ϳ50%, in theory, by the removal of stabilizing interactions that maintain the molecule in an inactive conformation (28,29). Arrestin truncation has been demonstrated to lead to partial * This work was supported by funds from the New Mexico Cancer Research Fund (to L. A. S.) and National Institutes of Health Grants AI36357 and AI43932 (to E. R. P.), GM60799, EB00265, and RR01315 (to L. A. S.), and EY11500 and GM63097 (to V. V. G). Confocal images and flow cytometry data in this paper were generated in the Fluorescence Microscopy and Flow Cytometry Data facilities at the University of New Mexico Health Sciences Center, which received support from National Center for Research Resources Grants 1 S10 RR14668 and P20 RR11830, National Science Foundation Grant MCB9982161, National Cancer Institute Grant R24 CA88339, the University of New Mexico Health Sciences Center, and the University of New Mexico Cancer Center. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
** To whom correspondence should be addressed. Tel.: 505-272-5647; Fax: 505-272-1421; E-mail: eprossnitz@salud.unm.edu. 1 The abbreviations used are: GPCR, G protein-coupled receptor; FPR, N-formyl peptide receptor; GTP␥S, guanosine 5Ј-3-O-(thio)triphosphate; GFP, green fluorescent protein; FITC, fluorescein 5-isothio-phosphorylation-and activation-independence with a number of GPCRs, including receptors from which the carboxyl terminus, containing critical phosphorylation sites, has been removed (25). We have recently demonstrated that the affinity of truncated arrestin-2 for a carboxyl-terminal peptide of the Nformyl peptide receptor (FPR) is significantly higher than that of wild type arrestin-2 (30). Thus, it is likely there is a complex interplay of multiple receptor and arrestin domains. However, it has yet to be fully determined how multiple, distinct mutations can render arrestins phosphorylation-independent.
We have previously demonstrated that FPR internalization, mitogen-activated protein kinase (MAPK) signaling, and chemotaxis can proceed through arrestin-independent mechanisms (31)(32)(33). We have also described the characteristics of G protein-and arrestin-FPR complexes in both cell-based and cell-free systems using flow cytometry, spectrofluorometry, and confocal microscopy (7,34,35). Surprisingly, our work has suggested that phosphorylation of the wild type FPR is sufficient to prevent G protein binding in vitro (7). Recently, we also demonstrated that two partial phosphorylation-deficient mutants of the FPR are unable to interact with wild type arrestins (36). Presumably, the absence of key phosphates in either of two distinct serine and threonine clusters in the carboxyl terminus of these receptors completely inhibited their ability to bind arrestins. Despite this inability, these receptors were fully proficient in terms of their signaling and internalization.
In the current report, we examined the interactions of a truncated arrestin-2 mutant (amino acids 1-382) with both wild type and phosphorylation-deficient forms of the FPR. We initially demonstrated both high and low affinity binding of the mutant arrestin to the phosphorylated and nonphosphorylated wild type receptor, respectively. Furthermore, we showed that the relative agonist affinity of such complexes was entirely dependent upon the phosphorylation status of the receptor. Through subsequent reconstitutions with phosphorylation-deficient FPRs, we identified regions of receptor phosphorylation regulating the affinity of arrestins for the FPR as well as sites critical for high agonist affinity complex formation. To our knowledge, these studies have allowed, for the first time, conclusive discrimination of phosphorylation sites regulating arrestin binding from sites regulating agonist affinity of the receptor.

EXPERIMENTAL PROCEDURES
Reagents-fNle-Leu-Phe-Nle-Tyr-Lys-Alexa546 was synthesized as follows. Alexa Fluor 546 carboxylic acid succinimidyl ester (Molecular Probes) and fNle-Leu-Phe-Nle-Tyr-Lys (Sigma) were each dissolved in anhydrous dimethyl sulfoxide (Me 2 SO) to 2 mM. Equal volumes were incubated with 100 mM triethylamine at room temperature overnight, and the product was used directly. All other chemicals were from Sigma unless otherwise indicated.
Cell Stimulation and Membrane Preparation-As described previously, spinner flasks containing near confluent FPR-expressing U937 cells were stimulated for 8 min at 37°C with 10 M fMLF, which results in ϳ90% maximal receptor phosphorylation (7). Following stimulation, flasks were placed on ice, and an equal volume of ice-cold PBS was added.
For membrane preparation, cells were harvested by centrifugation and resuspended in cavitation buffer (10 mM HEPES, 100 mM KCl, 30 mM NaCl, 3.5 mM MgCl 2 , 600 g/ml ATP, pH 7.3) at 4°C. The cell suspension was then placed in a nitrogen bomb for 15-20 min at 500 p.s.i. Following cavitation, nuclei were separated by centrifugation. The membrane fraction was resuspended in HEPES sucrose buffer (200 mM sucrose, 25 mM HEPES, pH 7.0) with protease inhibitor and phosphatase inhibitor cocktails (Calbiochem) prior to flash freezing. Aliquots were stored until use at Ϫ80°C. Detergent Solubilization-Membranes were thawed, diluted in an intracellular binding buffer (30 mM HEPES, 100 mM KCl, 20 mM NaCl, 1 mM EGTA, 0.1% w/v bovine serum albumin, 0.5 mM MgCl 2 ), and isolated by centrifugation. The membrane pellet was resuspended in binding buffer containing protease inhibitor mixture set I, phosphatase inhibitor mixture, and 1% n-dodecyl ␤-D-maltoside (Calbiochem). Suspensions were solubilized on a nutator (Clay Adams) for 90 min. The soluble fraction was collected by centrifugation and placed on ice for immediate experimentation.
Receptor Reconstitution-Detergent-solubilized FPR (8 -12 l of receptor preparation) was incubated with either a mixture of bovine brain G i /G o heterotrimer, purified arrestins, or buffer alone. A fluorescent agonist, N-formyl-Met-Leu-Phe-Lys-fluorescein 5-isothiocyanate (10 nM), was added, and the samples were gently mixed at 4°C for up to 120 min. Blocked samples received a large excess of an unlabeled peptide, N-formyl-Met-Leu-Phe-Phe (fMLFF), prior to fluorescent ligand addition.
Spectrofluorometric Analysis-As described previously, fluorescence was measured by an SLM 8000 spectrofluorometer (Spectronics) using the photon counting mode in time acquisition mode (7). Excitation was fixed at 490 nm with a 490-nm, 10-nm band pass filter (Corion) and an excitation monochrometer. Emission was monitored using a 520-nm, 10-nm band pass interference filter (Corion) and a 500-nm long pass filter (Kopp). Antibody and nucleotide additions were made through a microinjection port above the sample holder.
Following protein reconstitution and ligand incubation at 4°C, samples were diluted with a room temperature binding buffer containing 0.1% n-dodecyl ␤-D-maltoside and inhibitors. Samples were placed into the spectrofluorometer with gentle stirring, and data were acquired for 70 -100 s with a 1.0-s integration time. For the first 10 s, equilibrium levels were obtained. At 10 s, excess anti-fluorescein antibodies, prepared as described previously, were added to the sample. The antibodies rapidly quenched the fluorescence of unbound, labeled ligand. In some assays, excess GTP␥S (guanosine 5Ј-3-O-(thio)triphosphate) (Sigma) was added at 40 s to assess receptor-G protein coupling.
Confocal Microscopy-U937 cells (8 ϫ 10 6 ) that stably express the wild type FPR were electroporated with 25 g of DNA for wild type or mutant arrestin-GFP (pEGFP-N1) constructs. The wild type arrestin-GFP construct was a generous gift from Dr. Jeffrey Benovic (Thomas Jefferson University). The truncated arrestin-GFP was generated through PCR amplification of amino acids 1-382. After 24 -48 h, cells were incubated with 10 nM fNle-Leu-Phe-Nle-Tyr-Lys-Alexa546 for 10 min at either 0°C or 37°C. Stimulated samples were immediately fixed with ice-cold 2% paraformaldehyde for 30 min. Cells were washed, resuspended in Vectashield (Vector Laboratories), and mounted onto glass slides. Fluorescence images were acquired on a Zeiss LSM 510 confocal microscope to localize both the FPR (red) and arrestin-2 (green).
Data Analysis-Data were analyzed using GraphPad Prism software. In general, data were plotted as normalized FITC intensity as a function of time. Non-linear regression analyses were performed to generate dose response curves and normalized to maximal values.

Effects of Truncated Arrestin-2 on the Phosphorylated, Wild
Type FPR-Carboxyl-terminal truncation of arrestin-2 has been shown to result in phosphorylation-independent binding to GPCRs as well as confer enhanced affinity for an agonist (24). In the current report, we examine the effects of carboxylterminal truncation of arrestin-2 on full-length FPR assemblies, employing a non-cellular, spectrofluorimetric assay. The system is predicated on three facts. First, the anti-fluorescein antibody specifically quenches the fluorescence of unbound ligand (37). The assay therefore provides a direct measurement of the FITC-ligand dissociation rate. Second, ligand dissociation rates vary on the basis of receptor assemblies. Ternary complexes of both arrestins and G proteins with the phosphorylated and nonphosphorylated FPR, respectively, display high affinity for the agonist (7,24,34,35,38). Finally, the addition of guanine nucleotides causes the rapid dissociation of G proteins, but not arrestins, from receptors (7,24,34). This dissociation, in turn, results in the rapid dissociation of ligand, because the uncoupled receptor exhibits low affinity for the agonist.
We initially sought to examine the effects of arrestin truncation on the ligand dynamics of the phosphorylated, wild type FPR. As shown in Fig. 1A, incubation of the phosphorylated wild type receptor with 10 M wild type arrestin-2 and fluorescent ligand for 90 min leads to the formation of a slowly dissociating complex, as described previously (7). Truncated arrestin-2 reconstitution, in contrast, at an equivalent concentration results in a high fraction of slowly dissociating species (Fig.  1A). Titration studies of truncated arrestin-2 with the phosphorylated FPR demonstrate an EC 50 of 220 Ϯ 50 nM (Fig. 1B) in contrast to results with wild type arrestin-2, which display an EC 50 of ϳ600 nM (7). The truncated arrestin-2 therefore exhibits higher affinity for the phosphorylated FPR. Moreover, the magnitude of the ligand affinity shift at a theoretical maximum (B max ) is greater in the case of the truncated mutant (see Ref. 7). These effects were entirely dependent upon incubation time with the ligand, because pre-incubation with arrestins alone in the absence of ligand had no demonstrable effects on reconstitution. The t1 ⁄2 for FPR-truncated arrestin-2 complex formation was 5 Ϯ 2 min (Fig. 1C), which was similar to results with wild type arrestins (t1 ⁄2 , ϳ9 min). Thus, we see no evidence of activation-independent reconstitution in the case of the soluble FPR, in contrast to studies with other GPCRs (25).
Effects of Truncated Arrestin-2 on the Nonphosphorylated, Wild Type FPR-Our prior work has suggested that a truncated protein of arrestin-3 can bind to the nonphosphorylated, fulllength FPR without inducing a ligand affinity change (34). However, we have also shown that neither truncated arrestin-2 nor arrestin-3 exhibits specific binding to the nonphosphorylated FPR carboxyl terminus alone (30). Together, these results support a role for additional receptor domains in arrestin binding.
Incubation of the nonphosphorylated FPR with truncated arrestin-2 had no discernible effects on ligand affinity, in contrast to results with the phosphorylated FPR (Fig. 2, cf. Fig. 1). Therefore, to discriminate between a lack of binding and a lack  1. Effects of exogenous arrestins on the agonist affinity characteristics of wild type, phosphorylated FPR. A, phosphorylated, wild type FPRs were incubated with 10 nM FITC-fMLFK, a fluorescent agonist, on ice for 90 min with binding buffer alone (O), 10 M exogenous wild type (wt) arrestin-2 (q), or 10 M truncated arrestin-2 (छ). B, to assess the relative affinity of the truncated protein for the phosphorylated receptor, dose response curves were generated by titration of the mutant arrestin protein from 35 nM to 9 M, indicating an EC 50 of 220 Ϯ 50 nM. C, the time dependence of the arrestin-mediated agonist affinity changes was assessed by incubating 10 M truncated arrestin-2 with a phosphorylated receptor and fluorescent ligand for 0 to 90 min, revealing a t1 ⁄2 of 5 Ϯ 1 min. In general, samples were prepped immediately prior to spectrofluorimetric analysis by dilution in room temperature binding buffer and transfer to glass cuvettes. At 10 s, 60 nM anti-fluorescein antibody was added through a micro-injection port, which quenches the fluorescence of unbound ligand. At 40 s, excess GTP␥S was added, which disrupts coupling of receptor and G proteins. Data are plotted as peptide intensity versus time. Normalized means are representative of at least three independent experiments conducted in duplicate.

FIG. 2.
Competition studies of truncated arrestin-2 and heterotrimeric G proteins for the nonphosphorylated, wild type FPR. Solubilized samples of the nonphosphorylated, wild type FPR were incubated with fluorescent ligand (10 nM) and either buffer alone, 10 M wild type arrestin-2 (Arr-2), or 10 M truncated arrestin-2 (Arr-2 ) for 90 min on ice. Given the lack of discernible reconstitution, competition studies were undertaken between heterotrimeric G proteins and truncated arrestin-2 to assess coupling. Nonphosphorylated FPR were reconstituted with FITC-ligand, 1.5 M bovine brain G protein and either buffer alone or 10 M truncated arrestin-2 for 120 min prior to analysis. Addition of the truncated protein at high concentrations partially inhibited the formation of the high agonist affinity, nucleotide-sensitive G protein-coupled complex. Thus, similar to results with uncoupled receptors, assemblies of the nonphosphorylated receptor with arrestins display low agonist affinity. Normalized means are representative of at least three independent experiments conducted in duplicate.
of a ligand affinity shift, we undertook competition studies between heterotrimeric G proteins and truncated arrestin-2. Incubation of the nonphosphorylated FPR with G proteins results in the formation of a high affinity, nucleotide-sensitive complex (7,34). We therefore incubated the receptor with high concentrations of truncated arrestin-2 and 1.5 M G protein to assess competition. As shown in Fig. 2, partial inhibition of high agonist affinity complex formation was observed in the presence of excess truncated arrestin-2. The truncated mutant can therefore compete for the nonphosphorylated FPR with an estimated EC 50 of 15-25 M. Moreover, FPR-arrestin complexes display low agonist affinity in comparison to phosphoreceptor-arrestin assemblies.
Effects of Truncated Arrestin-2 on the Phosphorylated ⌬A FPR Mutant-We have previously demonstrated that two phosphorylation-deficient mutants of the FPR cannot bind to wild type arrestins either in vivo or in vitro despite full competency in terms of signaling and internalization (33,36). These ⌬A and ⌬B receptors contain targeted substitutions of groups of serines and threonines in the carboxyl terminus of the FPR with non-phosphorylatable alanines and glycines (Table I). Because the truncated arrestin displayed binding toward both the fully phosphorylated and unphosphorylated forms of the FPR, we sought to investigate the binding properties and effects of reconstitution on the ligand dynamics of the phosphorylated ⌬A and ⌬B receptors.
Incubation of the phosphorylated ⌬A receptor with 10 M truncated arrestin-2 results in the formation of a high affinity complex, in contrast to an equivalent concentration of wild type arrestin-2 (Fig. 3A). This result demonstrates at a minimum that the truncated arrestin-2 acts on a greater subset of the receptor population than wild type arrestins and not simply with greater effect on the same receptors, because wild type arrestins display no detectable binding to this phosphorylationdeficient receptor. Titration studies reveal that the affinity of the truncated arrestin-2 for the phosphorylated ⌬A is significantly lower than for the phosphorylated, wild type receptor, with an EC 50 of 2.2 Ϯ 0.3 M (Fig. 3, A and C). However, the magnitude of the affinity shift at a maximum is approximately equivalent to that seen with the phosphorylated, wild type FPR (data not shown). Thus, the truncated arrestin is able to induce a ligand affinity change in a submaximally phosphorylated receptor population to the same extent as with a fully phospho- rylated form of the FPR, despite its weaker binding affinity to the ⌬A receptor. These results demonstrate that phosphorylation of the A site is critical in determining the affinity of the receptor for arrestin, but it is not responsible for arrestin-dependent agonist affinity changes.
Because phosphorylation of the ⌬A receptor does not prohibit G protein coupling, we could also demonstrate the successful competition of heterotrimeric G protein binding with the truncated arrestin-2, as suggested by the loss of nucleotide-sensitivity (Fig. 3B). The EC 50 for competition of G protein binding (used at its K d of ϳ1 M) was ϳ1.7 Ϯ 0.5 M (Fig. 3C), similar to that observed for the ligand affinity shift dose response data (i.e. 2.2 M; see Fig. 3, A and C).
Effects of Truncated Arrestin-2 on the Phosphorylated ⌬B FPR-The ⌬B receptor contains targeted substitution of a cluster of the four serines and threonines between amino acids 334 -339 of the FPR. Incubation of the phosphorylated ⌬B receptor with truncated arrestin-2 had no discernible effects on agonist affinity (Fig. 4A). We therefore sought to discriminate between binding without an accompanying agonist affinity shift and a complete lack of binding by carrying out competition studies with heterotrimeric G proteins. Reconstitution of the phosphorylated ⌬B receptor with 1 M exogenous G proteins results in the formation of a high agonist affinity, nucleotideinsensitive complex (36). The addition of truncated arrestin-2 results in the conversion of the receptor from a high affinity to low agonist affinity state, as expected from the lack of a ligand affinity change in the direct binding assay above. This competition occurred with an EC 50 of 340 Ϯ 60 nM, a similar affinity to the interaction of wild type arrestin-2 with the phosphorylated, wild type FPR. Thus, although phosphorylation in the A site is not sufficient for arrestin-dependent agonist affinity changes, it can provide for a high affinity interaction of arrestin with the FPR.
In Vivo Co-localization of Arrestins and the FPR-To confirm the binding characteristics of the truncated arrestin with the FPR, we electroporated U937 cells stably expressing FPR constructs with the DNA for either arrestin-2-GFP or truncated arrestin-2-GFP. After 24 h of recovery, cells were treated with an Alexa546-labeled formyl hexapeptide and analyzed by confocal microscopy. We initially verified the proper trafficking of the arrestin-GFPs using wild type arrestin-2 and the wild type FPR. As shown in Fig. 5, stimulation for 10 min at 37°C results in the co-localization of both wild type and truncated arrestins with wild type receptors into punctate, internal structures, which is consistent with prior results (31). Stimulation of cells on ice did not result in either receptor clustering or arrestin co-localization. In agreement with in vitro results, both the ⌬A and ⌬B receptors co-localized with the truncated arrestin-2 into punctate structures after 10 min of stimulation (Fig. 6), unlike results with wild type arrestin-2. These results confirm the interaction of the truncated arrestin mutant with the phosphorylation-deficient receptors in a native environment. DISCUSSION In this study, we initially demonstrated that carboxyl-terminal truncation of arrestin-2 resulted in activation-dependent reconstitution with the wild type FPR. The truncated mutant displayed enhanced affinity for the phosphorylated receptor as well as an enhanced agonist affinity shift in comparison to wild type arrestins. In agreement with prior studies, the lack of receptor phosphorylation did not prohibit binding to the FPR (36), but such interactions were of markedly lower affinity. Furthermore, receptor-arrestin assemblies of nonphosphorylated receptors displayed low agonist affinity, in contrast to results with phosphorylated receptors. Together, these results suggest that truncation of arrestin-2 reduces its selectivity for the phosphorylated FPR, which may permit its association with a greater subset of a heterogeneously phosphorylated receptor population. It also implies that the phosphorylation status of the receptor is a critical determinant underlying the agonist affinity of receptor-arrestin assemblies.
Because the truncated arrestin binds to both the phosphorylated and unphosphorylated wild type FPR but only induces an affinity change for the ligand in the former case, we investigated the individual contributions of distinct phosphorylation sites. We have previously determined that wild type arrestin binding to the FPR is contingent upon phosphorylation in both the A and B domains (36). Binding of the truncated arrestin to the phosphorylated ⌬A FPR mutant resulted in the formation of a high agonist affinity complex, although with an order of magnitude decrease in the arrestin binding constant. This suggests that phosphorylation of the A sites plays a pivotal role in the regulation of the affinity of arrestins for the receptor but not in the formation of a high agonist affinity complex. In contrast, the phosphorylated ⌬B receptor was determined to bind to truncated arrestin with a similar affinity as that of the wild type receptor. However, analogous to results with the nonphosphorylated receptor, such complexes displayed low agonist affinity. This implies that the phosphorylation status of the B sites is a critical determinant in the formation of a high agonist affinity complex. Furthermore, in comparison to the A sites, the B sites appear to play a less significant role in determining the affinity of arrestins for receptors.
The presence of two essential arrestin-binding motifs in the FPR is not unique. Studies with the m2 muscarinic receptor have demonstrated the importance of two distinct phosphorylation domains in the third intracellular loop of the receptor (39,40). It has been shown that targeted mutation of either of these motifs disrupts stable arrestin interactions. Furthermore, work on other GPCRs, such as the neurotensin-1 receptor, oxytocin receptor, angiotensin II type 1A receptor, and substance P receptor, has revealed the importance of multiple phosphoamino acid clusters in regulating arrestin interactions (27). However, the differential effects of distinct clusters or sites of phosphorylation have not previously been described. Furthermore, this is the first definitive identification of specific sites mediating the high agonist affinity of arrestin-GPCR assemblies. This is an entirely novel finding in that it has been widely assumed that the regulation of receptor agonist affinity by arrestin was an all or nothing phenomenon, dependent only on the actual binding of arrestin to the receptor and not the specific sites of phosphorylation (24). It stands to reason, however, that a discrete subset of the phosphorylation sites within the A and/or B domains of the FPR are responsible for the phosphorylation-dependent differences. We are therefore continuing studies to examine the role(s) of individual phosphorylation sites in the regulation of arrestin and receptor function.
One puzzling observation is the apparent inverse relation- FIG. 7. Model of FPR ternary complex formation with arrestin. Ternary complex assembly with arrestins involves the stepwise interactions of multiple phosphorylation-dependent and -independent domains in both arrestin and the FPR. Phosphorylated, wild type receptor binds to wild type arrestin in an activation-and phosphorylation-dependent manner to yield a high agonist affinity complex. Carboxyl-terminal truncation of arrestin obviates the need for arrestin tail release and allows for binding to phosphorylation-deficient receptors. Following interactions between the phosphoreceptor and the polar core of arrestin, the latter induces and/or stabilizes a high agonist affinity state of the receptor. The "ligand affinity" associated with a given ternary complex is inferred from both the direction and magnitude of a shift in the ligand dissociation curve. It is therefore a function of both the number of receptors accessible to a given arrestin protein and the extent to which a liganded receptor is activated by arrestin binding.
b Given the lack of discernible binding within the range of protein concentrations tested, estimates of arrestin affinity are necessarily speculative. Consequently, the corresponding ligand affinity of such complexes is indeterminate.
ship between arrestin affinity and agonist affinity in terms of the phosphorylation status of the A and B clusters of the FPR (Table II). One possible explanation emerges from the recent discovery of an additional phosphate-binding element in arrestin. It has been suggested that lysines 14 and 15 in visual arrestin (equivalent to lysines 10 and 11 in arrestin-2) act as a phosphate-sensitive "guide" for GPCRs, presumably by optimally orienting the receptors for polar core interactions (41). The engagement of lysines 10 and 11 by GPCR phosphodomains most likely also results in arrestin activation, due to disruption of a hydrophobic anchor in the carboxyl terminus (Fig. 7). This activation may be similar to what is achieved by truncation of the carboxyl terminus of the protein. However, mutation of these residues to neutral amino acids results in significantly diminished binding to receptors, even in the face of additional activating arrestin mutations, indicating that these residues are also critical in establishing stable receptorarrestin complexes (41).
Our results suggest that there are likely four functional sites of interaction between GPCRs and arrestins (Fig. 7). The important and regulatable sites on the FPR include the ligand binding site, an activation-dependent binding site for G proteins and arrestins (consisting in part of the "DRY" sequence of the second intracellular loop), and at least two distinct sites of phosphorylation that we define here as regulators of arrestin binding affinity and receptor affinity for agonist. Arrestins possess at least three primary sites of interaction with activated, phosphorylated GPCRs. The first site recognizes the activation state of the receptor, likely through interactions with the second intracellular loop of the receptor. Two additional sites, represented by lysines 10/11 and the polar core, interact with phosphate moieties on the receptor. Initial interactions with lysine 10/11 (Fig. 7, Step 1) result in the partial activation of arrestin, leading to alterations of the carboxyl terminus of arrestin and exposure of the polar core (Fig. 7, Step 2). In the case of the truncated form of arrestin this activation step is not required, and the truncated arrestin can bind to forms of the FPR with lower levels of phosphorylation (mutants ⌬A and ⌬B). Following binding of phosphoresidues to the polar core (Fig. 7, Step 3), the final interaction between arrestins and GPCRs results in the stabilization of a receptor conformation with a high affinity for the ligand (Fig. 7, Step 4).
In conclusion, this work describes completely unanticipated results that are essential in unraveling the complex interactions between an activated phosphorylated GPCR and arrestins. Because arrestins bind to virtually all GPCRs and yet elicit receptor-specific effects including important differences in desensitization, internalization, and G protein-independent kinase activation, it is likely that the pattern of receptor phosphorylation specifies to a large extent the subsequent functions of the bound arrestin molecule. Additional work with the FPR and other GPCRs will be required to resolve these complexities.