Determination of bradykinin B2 receptor in vivo phosphorylation sites and their role in receptor function.

Reversible phosphorylation plays important roles in G protein-coupled receptor signaling, desensitization, and endocytosis, yet the precise location and role of in vivo phosphorylation sites is unknown for most receptors. Using metabolic 32P labeling and phosphopeptide sequencing we provide a complete phosphorylation map of the human bradykinin B2 receptor in its native cellular environment. We identified three serine residues, Ser(339), Ser(346), and Ser(348), at the C-terminal tail as principal phosphorylation sites. Constitutive phosphorylation occurs at Ser(348), while ligand-induced phosphorylation is found at Ser(339) and Ser(346)/Ser(348) that could be executed by several G protein-coupled receptor kinases. In addition, we found a protein kinase C-dependent phosphorylation of Ser(346) that was mutually exclusive with the basal phosphorylation at Ser(348) and therefore may be implicated in differential regulation of B2 receptor activation. Functional analysis of receptor mutants revealed that a low phosphorylation stoichiometry is sufficient to initiate receptor sequestration while a clustered phosphorylation around Ser(346) is necessary for desensitization of the B2 receptor-induced phospholipase C activation. This was further supported by the specifically reduced Ser(346)/Ser(348) phosphorylation observed upon stimulation with a nondesensitizing B2 receptor agonist. The differential usage of clustered phosphoacceptor sites points to distinct roles of multiple kinases in controlling G protein-coupled receptor function.

activity. They share a common deduced structure comprising seven ␣-helical transmembrane domains connected by extraand intracellular loops. Through their intracellular domains they interact with heterotrimeric G proteins, which in turn modulate the activity of various effectors, such as adenylate cyclases, phospholipases, and ion channels. These effectors generate the intracellular second messengers, which ultimately evoke cellular responses (1). Signal transduction of GPCRs is carefully controlled: continuous or repeated agonist stimulation leads to an attenuation of the response, a phenomenon called desensitization. Although desensitization of receptor/G protein/effector systems generally involves perturbations of all three components, the impairment of the ability of receptors to activate G proteins appears to be the most important and seems to involve an agonist-induced receptor phosphorylation (2)(3)(4).
Much of the knowledge about the molecular mechanisms governing desensitization has come from studies of rhodopsin and ␤ 2 -adrenergic receptors (5)(6)(7). Rhodopsin, which is available in much greater quantities than any other GPCR, has been successfully subjected to mapping of in vitro (i.e. purified and reconstituted components) and in vivo (i.e. in cultured cells) phosphorylation sites in its C-terminal tail (8,9). More recently, the identification of phosphorylation sites in the reconstituted ␤ 2 -adrenergic receptor by G protein-coupled receptor kinase (GRK) 2 and GRK5 was reported (10). The functional relevance of these in vitro phosphorylation sites has subsequently been challenged by mutagenesis studies that failed to correlate the presence of the mapped residues with cellular receptor-mediated functions (11). This discrepancy could be explained by the rather poor substrate specificity of GRKs in reconstituted systems resulting in the phosphorylation of sites not used in vivo (2). For instance, rhodopsin kinase (GRK1) phosphorylates the ␤ 2 -adrenergic receptor in vitro, and rhodopsin is an excellent substrate for GRK2 in reconstituted systems (4,12). However, due to their differential tissue distribution both combinations are very unlikely to play a relevant role under physiological conditions. Thus, phosphorylation sites of GPCRs identified in vitro do not necessarily correlate with sites that tune receptor functions in a native cellular environment.
In many studies site-directed mutagenesis and total phosphorylation of GPCRs has been performed to screen for key amino acids involved in signal transduction. Using this approach, several mutant receptors have been generated that were useful to corroborate a role of GPCR phosphorylation in adaptation processes (5,13,14). For example, four consecutive serine residues in the third intracellular loop were identified as the major phosphorylation sites of the ␣ 2A -adrenergic receptor in Chinese hamster ovary cells, and their phosphorylation correlated in an additive manner with the desensitization of the * 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.
§ Supported by a postdoctoral fellowship from the Deutsche Forschungsgemeinschaft. Present address: Institute for Pharmacology, University of Heidelberg, Im Neuenheimer Feld 366, D-61290 Heidelberg, Germany.
ʈ Supported by the Fondation pour la Recherche Medicale and the Association de secours des amis des Sciences. receptor-mediated reduction of cellular cAMP levels (13). A general limitation of the mutagenesis approach is that truncations, deletions, or exchanges of amino acids often affect receptor structure, trafficking, localization, and stability or intervene with GRK recognition. Therefore, mutagenesis studies may allow valuable deductions about in vivo phosphorylation sites but the unequivocal identification of these sites requires receptor purification and biochemical analysis. Because of the hydrophobic nature and inherently low expression of GPCRs this particularly challenging task has not yet been accomplished for a single non-rhodopsin GPCR.
In an effort to identify the in vivo phosphorylation sites of a prototypical GPCR, we chose to study the human bradykinin B 2 receptor (B 2 R). Previous work has demonstrated that the B 2 R desensitizes upon prolonged or repeated agonist stimulation (15)(16)(17)(18)(19), and that the agonist-induced B 2 R phosphorylation and dephosphorylation correlate with its de-and resensitization (15). Furthermore, a cluster of serine and threonine residues located in the C-terminal tail of the B 2 R has been suggested to hold potential phosphorylation sites (15,17,18). Using two-dimensional phosphopeptide mapping and Edman sequencing we report the complete in vivo phosphorylation pattern of the human B 2 R and demonstrate that the differential usage of clustered phosphoacceptor sites contributes to the complex regulation of receptor sequestration and desensitization.  (Trasylol TM ) was from Bayer; AG-X8 anion exchanger resin was from Bio-Rad; GF109203X and PMA were from Calbiochem; bacitracin and Pefabloc TM were from FLUKA; FR190997 was a kind gift from Fujisawa Pharmaceutical Co., LipofectAMINE TM and protein markers were from Life Technologies; cellulose thin layer chromatography (TLC) plates were from Merck; sequencing grade trypsin was from Promega; leupeptin was from Roche Molecular Bioscience; nitrocellulose membranes were from Schleicher & Schuell; phosphoamino acid standards were from Sigma; and protein A-agarose was from Zymed Laboratories Inc. All tissue culture reagents were from Sigma and Life Technologies.
Cell Culture and Transfection-Human embryonic kidney cells HEK293T were grown to about 50 -70% confluence in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Transfections were done in serum-free medium with the indicated cDNAs using 0.2-0.4 g/well of a 24-well plate, 0.4 -1 g/well of a 12-well plate, 1-2 g/well of a 6-well plate or 10 g/10-cm dish by the LipofectAMINE TM method according to the suppliers manual. Cells were used for experiments 48 -60 h after transfection. HF-15 human foreskin fibroblasts were grown to confluence in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum.
Ligand Binding and Receptor Internalization-Ligand binding and internalization were measured according to established procedures using [ 3 H]bradykinin (20). HEK293T cells grown on 12-well plates were used 40 h after transfection with B 2 R constructs. Cells were washed twice with phosphate-buffered saline supplemented with 0.2% bovine serum albumin, 2 mM bacitracin, and incubated for 90 min at 4°C in the same solution containing 5 nM [ 3 H]bradykinin in the absence ("total binding") or presence ("nonspecific binding") of 5 M unlabeled bradykinin. Internalization of the receptor was initiated by incubating cells for different time periods at 37°C. Cells were washed twice to remove free ligand, and the cell-bound [ 3 H]bradykinin was extracted with 0.2 M acetic acid, pH 2.8, 0.5 M NaCl, 0.2% bovine serum albumin, and radioactivity of the extract was measured ("surface-associated bradykinin"). The acid-stripped cells were dissolved in 1 M NaOH and the radioactivity of the lysate was determined ("intracellular bradykinin").
Phospholipase C Assays-Phospholipase C activity was measured by analyzing inositol phosphate accumulation (21). Cells grown on 24-well plates were labeled with 1 Ci/ml myo-[ 3 H]inositol for 24 h in inositolfree Ham's F-12 including 0.1% (w/v) bovine serum albumin, treated for 5 min with 10 mM LiCl and then challenged with 0.01-1 M bradykinin for 10 min in the presence of 10 mM LiCl. Reactions were stopped by addition of 1 ml of ice-cold 10 mM formic acid. Water-soluble inositols were extracted for 2-12 h at 4°C and separated by anion exchange chromatography using AG-X8 as a resin. Inositol phosphates were eluted with 2 M ammonium formate and quantified by liquid scintillation counting. Results were normalized for total labeling of lipid pools, which was calculated from the radioactivity in water-soluble extracts and cells. To follow desensitization, cells were pretreated with 100 nM bradykinin for 5 min. Following removal of excess of ligand by washing cells three times with medium, 10 mM LiCl was added 10 or 15 min after the initial stimulation and inositol phosphate levels were measured as described above. 32 P Labeling and Two-dimensional Mapping of Phosphorylation Sites-32 P labeling of cells (1-2 mCi/ml for 6 -8 h), solubilization, and immunoprecipitation of B 2 R using an antiserum against the C-terminal receptor domain were done as detailed previously (15). For HF-15 cells with endogenous B 2 R, a 10-cm dish was used, whereas for transiently transfected HEK293T and COS-7 cells a well of a 6-well plate was sufficient for the following procedure. After 10% SDS-PAGE proteins were transferred onto nitrocellulose membranes using a semi-dry unit from Bio-Rad. Radiolabeled B 2 R was detected by PhosphorImager (BAS2000, Fuji) analysis and tryptic digests were performed as described by Boyle et al. (22) with minor modifications. Briefly, membrane pieces containing the 32 P-labeled B 2 R were cut out and blocked with 0.5% polyvinylpyrrolidon 40 in 0.6% acetic acid for 30 min at 37°C. Following extensive washes with water, membrane bound B 2 R was cleaved in situ with 1 g of modified sequencing grade trypsin in 200 l of 50 mM (NH 4 )HCO 3 for 12 h at 37°C. Released tryptic peptides were vacuum-dried and oxidized with 50 l of performic acid for 1 h on ice. Reactions were stopped by dilution with 500 l of 20% (v/v) ammonia solution. Thereafter samples were frozen, vacuum-dried, and a second digest was performed with 1 g of trypsin in 50 l of 50 mM (NH 4 )HCO 3 for 12 h at 37°C. Following vacuum drying, samples were dissolved in 5 l of electrophoresis buffer (formic acid:acetic acid:water, 46:156:1790 (v/v/v)) and phosphopeptides were separated by electrophoresis on cellulose thin layer plates in a first dimension (2000 V, 40 min, electrophoresis buffer) and ascending chromatography in a second dimension (15 h, isobutyric acid, 1-butanol, pyridine, acetic acid, water, 1250:38: 96:58:558 (v/v/v/v/v)). Phosphopeptides were detected by PhosphorImager analysis and eluted from the cellulose matrix with 2 ϫ 250 l of 20% (v/v) acetonitrile in a sonicated water bath for 15 min. Part of the extract (25-100 cpm) was hydrolyzed with 6 M HCl for 1 h at 110°C and subjected to a phosphoamino acid analysis (15,22). The second fraction (50 -500 cpm) was sequenced by Edman degradation using a solid phase sequencer (ABI 477). Twenty sequencing cycles were collected, dried, and analyzed for their content of 32 P radioactivity using a Phosphor-Imager. Data obtained from Edman degradation (position of phosphoamino acids in tryptic peptides) and phosphoamino acid analysis (phosphorylated amino acid), the known sequence of tryptic peptides (Table I) and the anticipated transmembrane topography of the B 2 R (23) were used to predict phosphorylation sites. The prediction was verified by in vitro mutagenesis of corresponding phosphoacceptor sites followed by two-dimensional phosphopeptide mapping.

Mapping of Serine 339, 346, and 348 as Major in Vivo Phosphorylation Sites of B 2
Receptor-To molecularly characterize the importance of receptor phosphorylation in signal transduction of the B 2 R, we applied an analytical two-dimensional phosphopeptide mapping strategy to identify in vivo phosphorylation sites of the receptor (22). HF-15 human fibroblasts expressing 400 -750 fmol of endogenous B 2 R/mg of protein were labeled with [ 32 P]orthophosphate, stimulated with bradykinin, lysed, and B 2 R was immunoprecipitated with a specific anti-peptide antibody (15). Following SDS-PAGE, proteins were transferred onto nitrocellulose membranes and the radio-labeled B 2 R was detected as a 55-68-kDa band by Phospho-rImager analysis (Fig. 1A). Notably the broadness of the band and an occasionally observed doublet are likely due to heterogeneous glycosylation (24). In average a 4.0 Ϯ 0.5-fold increase of the basal B 2 R phosphorylation was observed in bradykininstimulated cells. In situ digestion with trypsin-generated phosphopeptides that were separated on thin layer plates by high voltage electrophoresis and ascending chromatography. Phos-phorImager analysis revealed a major phosphopeptide in untreated cells that contained more than 80% of the total radioactivity (Fig. 1A, peptide 1). The same phosphopeptide was present in bradykinin-stimulated cells as well as three additional 32 P-labeled peptides (Fig. 1A, peptides 2-4). Total hydrolysis and phosphoamino acid analysis performed with a fraction of the material isolated from the spots revealed that peptides 1, 2, and 3 contained exclusively phosphoserine whereas peptide 4 carried the 32 P label both on serine and threonine (Fig. 1B). The major fraction of the peptides was used for solid phase Edman sequencing. Because peptide quantities were too low for chemical conversion and identification of re-leased amino acids, cleavage products of 20 sequencing cycles were collected and analyzed for their 32 P content using a Phos-phorImager. The majority of radioactivity from peptide 1 eluted in cycle 4 (Fig. 1C). Together with the finding that phosphate in peptide 1 is exclusively attached to serine (cf. Fig. 1B), we conclude that the fragment covering amino acid positions 345-351 of the human B 2 R sequence (Ref. 23, Table I) is the only one present in the tryptic digest that can be serine phosphorylated at position 4, namely at Ser 348 (pS348). Peptide 2 generated a peak of radioactivity in sequencing cycle 9 thus identifying Ser 339 as the phosphorylation site (Fig. 1C, Table I). In the case of peptide 3, peaks of radioactivity were detected in cycles 2 and 4, whereas peptide 4 released 32 P in cycles 9 and 12 (Fig. 1C). Alignment with the sequence of tryptic peptides (Table I) and consideration of the data from the phosphoamino acid analysis (Fig. 1B) allowed the identification of Ser 346 and Ser 348 (peptide 3) and Ser 339 and Thr 342 (peptide-4) as B 2 R phosphorylation sites.
For a more quantitative comparison, two-dimensional maps from several experiments were analyzed using a PhosphorIm- Following immunoprecipitation and 10% SDS-PAGE, isolated proteins were transferred onto nitrocellulose membranes and analyzed using a Phosphor-Imager (BAS2000, Fuji). The 32 P-labeled B 2 R was in situ digested with trypsin and resulting peptides were separated on thin layer chromatography (TLC) plates by high voltage electrophoresis and ascending chromatography. A cross illustrates where samples were applied; ϩ and Ϫ indicate the polarity during electrophoresis. Thereafter, phosphopeptides were localized by PhosphorImager analysis. B, peptides were eluted and a fraction was hydrolyzed, subjected to phosphoamino acid analysis followed by two-dimensional electrophoretic separation on TLC plates and PhosphorImager analysis. Phosphorylated amino acids were identified by commercial standards (locations indicated by dashed circles). C, major fractions of phosphopeptides were subjected to 20 cycles of Edman degradation and cleaved amino acids were collected and analyzed using a PhosphorImager to locate the position of the phosphorylation site(s) as exemplified for peptide 1. The content of 32 P radioactivity of each sequencing cycle was quantified and expressed in arbitrary units (AU). D, phosphopeptides on TLC plates were quantified using a PhosphorImager and the relative contributions of individual phosphopeptides to total basal (light) as well as bradykinin-mediated (dark) phosphorylation were determined. Mean Ϯ S.D. from eight independent experiments are shown.
ager. The contribution of the individual 32 P-labeled peptides to total phosphorylation was calculated for untreated and bradykinin-stimulated cells and corrected for the increase in total phosphorylation. The contribution of peptide 4 containing pS339/T342 to total receptor phosphorylation was always very low (Ͻ5%), and therefore this peptide was excluded from quantitative studies. As shown in Fig. 1D, phosphorylation of Ser 348 in peptide 1 did not significantly change during agonist challenge on this relative scale. In contrast, phosphorylation of Ser 339 and the double-labeled peptide 3 containing pS346/S348 increased about 10-and Ͼ40-fold, respectively, in the presence of bradykinin (Fig. 1, A and D).
Phosphorylation of B 2 Receptor Mutants Lacking Identified Phosphorylation Sites-To validate the assignment of B 2 R phosphorylation sites we used receptor mutants in which Ser 339 , Ser 346 , and Ser 348 individually or in combinations were replaced by alanine residues. Comparable expression levels of all mutants were confirmed by binding assays and immunoprecipitation of 35 S-labeled proteins (not shown). Immunoprecipitation of the 32 P-labeled B 2 R mutants revealed that the ligandinduced phosphorylation of S339A and S346A was slightly reduced whereas the basal phosphorylation in the absence of a ligand was essentially unchanged over the wild-type receptor ( Fig. 2A). In contrast, we observed a complete lack of basal phosphorylation in S348A and S346A/S348A mutants, together with a significantly decreased bradykinin-induced phosphorylation. Basal phosphorylation of the double mutant S339A/ S346A was slightly increased, and only a minor increment was observed upon bradykinin stimulation. Finally, the ⌬S (S339A/ S346A/S348A) and ⌬ST (S339A/S346A/S348A-T342A/T345A) mutants failed to produce any significant phosphorylation above background (Fig. 2A).
Mutant B 2 R were further subjected to two-dimensional phosphopeptide mapping and the resultant phosphopeptides were characterized by phosphoamino acid analysis and Edman degradation. Surprisingly, a major phosphorylated peptide was seen in the S348A mutant in a similar location as peptide 1 in wild-type B 2 R (Fig. 2B, arrow). Sequence and phosphoamino acid analysis revealed that the S348A mutant had a compensatory phosphorylation at Ser 346 in position 2 of the corresponding peptide (Fig. 2B). In the single mutants S339A and S346A as well as in the double mutants S339A/S346A and S346A/S348A, phosphopeptides corresponding to peptides 2 and 3 of wild-type B 2 R (cf. Fig. 1A) were absent from the two-dimensional maps (Fig. 2C) thus confirming our identification of vivo B 2 R phosphorylation sites. Under these conditions, no phosphopeptide(s) of significant quantity was detected in the two-dimensional maps of the ⌬S and ⌬ST variants. Taken together, we have identified (i) graded phosphorylation of four closely spaced residues, Ser 348 Ͼ Ser 346 Ͼ Ser 339 Ͼ Ͼ Thr 342 , (ii) basal phosphorylation at a single site, Ser 348 , (iii) bradykinin-induced phosphorylation at two major sites, Ser 346 and Ser 339 , and (iv) combined phosphorylation at Ser 346 and Ser 348 .
Identification of Kinases That Can Phosphorylate the B 2 Receptor-Co-expression of receptors with GRKs is commonly used to obtain information about the nature of kinases involved in GPCR phosphorylation (25)(26)(27)(28)(29). To identify potential kinase(s) executing B 2 R phosphorylation and to locate their corresponding substrate sites we analyzed the phosphopeptide patterns of B 2 R co-transfected with human GRK2-6 in HEK293T cells. Immunoprecipitates from 32 P-labeled cells indicated that total basal and bradykinin-mediated B 2 R phosphorylation did not significantly change upon co-expression with GRK2, GRK3, GRK5, or GRK6 (Fig. 3A). In contrast, expression of GRK4␣ drastically increased the basal level of 32 P incorporation into B 2 R. However, phosphopeptide maps revealed quantitative changes in the distribution of phosphopeptides for the various GRKs. For example, 32 P labeling of peptide 3 containing pS346/pS348 was enhanced 1.5-3-fold as compared with mock-transfected cells in the order GRK6 Ͻ GRK5 Ͻ GRK2 Ͻ GRK4␣ Ͻ GRK3. Most prominently, GRK4␣ elevated the basal phosphorylation of Ser 339 and Ser 346 /Ser 348 15-and 24-fold, respectively. These results suggest that several endogenous GRKs may phosphorylate the B 2 R and that the various GRKs, even without apparent effect on total GPCR phosphorylation levels, may induce distinct phosphorylation patterns with possible functional consequences for receptor desensitization and sequestration.
We have previously observed that activators of protein kinase C may induce a ligand-independent phosphorylation of the B 2 R (15). Under identical conditions PMA pretreatment significantly reduced the bradykinin-induced PLC stimulation (not shown), indicating that an agonist-independent ("heterologous") receptor phosphorylation may negatively affect signal transduction of the B 2 R.
To identify residue(s) in the B 2 R sequence targeted by PKC we analyzed two-dimensional phosphopeptide maps from 32 Plabeled HF-15 cells treated with PMA. PhosphorImager analysis revealed a new spot (peptide 5) that partially overlapped with the major phosphopeptide 1 containing pS348 (Fig. 3B). Peptides 1 and 5 were isolated avoiding cross-contamination, and sequence and phosphoamino acid analysis confirmed the identity of pS348 in peptide 1 (Fig. 3C). Peptide 5 showed 32 P-labeled serine in position 2 suggesting residue Ser 346 in peptide 345-351 as potential PKC phosphorylation site (Table  3B). Together these findings point to Ser 346 as the major PKC target site in the B 2 R. This notion was confirmed by the repression of PMA-induced Ser 346 phosphorylation in the presence of PKC inhibitors (not shown). Quantitative evaluation of two-dimensional maps revealed a 3-fold increase in Ser 346 phosphorylation upon PMA treatment, and a 40% decrease in 32 P incorporation in the major B 2 R phosphorylation site Ser 348 (Fig. 3D). PKC stimulation did not produce any double phosphorylated peptide 4 bearing pS346/S348 (Fig. 3, B and D 2 Receptor Phosphorylation-Having identified the principal phosphoacceptor sites in intact cells we tested whether the phosphopeptide pattern of the B 2 R changes during the time course of stimulation. We followed the kinetics of phosphorylation of the endogenous B 2 R in HF-15 fibroblasts over a period of 60 min and found that bradykinin-induced B 2 R phosphorylation is a fast process reaching a maximum after 5 min and decreasing almost to basal levels after 60 min (Fig. 4A, bottom  panel). The appearance of pS339 and pS346/S348 strictly followed this time course, whereas pS346 or pS348 remained constant over the entire period of the experiment (Fig. 4A, top  panel).
Analyzing the dose-dependence of B 2 R phosphorylation, a half-maximum effect was observed with ϳ10 nM bradykinin, and Ն100 nM of the ligand was sufficient to trigger full receptor phosphorylation (Fig. 4B, bottom panel). Whereas pS348 was constant and independent of the applied bradykinin concentration, single phosphorylation at Ser 346 increased up to 2.5-fold during stimulation with low ligand concentrations (Յ1 nM Bk), remained constant at intermediate bradykinin concentrations (1-100 nM), and decreased at the highest agonist concentrations (Ն100 nM). Phosphorylation of Ser 339 and Ser 346 /Ser 348 increased markedly in correlation with the dose of the ligand (Fig. 4B, top panel)

Two-dimensional Mapping of B 2 Receptor Phosphorylation Sites
as a probe. The internalization process of wild-type B 2 R was fast with a half-maximal effect after 5-10 min and a maximum of almost 70% internalized receptors after 60 min (Fig. 5A). Single or double mutations of serine residues in B 2 R caused only minor effects on the internalization capacity of the corresponding constructs. The time course and the extent of ⌬S sequestration were initially similar to wild-type B 2 R, but diverged beyond 20 min of incubation such that only 40% of receptors were internalized after 60 min. In contrast, internalization of the ⌬ST mutant was clearly reduced at all time points tested, and the majority of mutant receptors (Ͼ80%) remained surface exposed during the whole experiment. These results suggest that the initiation of B 2 R sequestration requires only a low stoichiometry of phosphorylation without any obvious prevalence for specific residue(s) and that additional, probably phosphorylation-independent processes may be involved in the relocation of the receptor (17).
We also determined desensitization of the bradykinin-mediated PLC activation of wild-type B 2 R and various phosphorylation-deficient mutants. To circumvent the problem that bradykinin, which has a high affinity to the B 2 R cannot be properly washed out after receptor stimulation, we adapted an alternative protocol that monitors signal duration as a measure of receptor desensitization (21). In transfected HEK293T cells inositol phosphate levels triggered by wild-type B 2 R were reduced by ϳ20% after 10 or 15 min of delayed accumulation as compared with control (Fig. 5B). Minor variations were seen with S339A and S348A mutants although difference to wildtype B 2 R did not reach statistical significance. Under the same conditions the S346A mutant showed a slight increase of inositol phosphates, while an augmented second messenger accumulation was particularly evident and significant for the S339A/S346A, S346A/S348A, ⌬S, and ⌬ST mutants. We conclude that Ser 346 is a critical residue for desensitization of the B 2 R, and that a clustered phosphorylation of Ser 346 and at least one additional serine residues seems to be necessary for full desensitization of B 2 R-mediated PLC activation.
To further confirm the role of receptor phosphorylation in desensitization we analyzed the pattern of B 2 R phosphorylation upon stimulation of cells with FR190997. This synthetic non-peptidic agonist that has been reported to mediate a sustained activation of B 2 R indicative of reduced desensitization (30,31). Stimulation of B 2 R-expressing HEK293T cells with increasing concentrations of FR190997 led to a dose-dependent rise in receptor phosphorylation (Fig. 5C) comparable to that observed with bradykinin (cf. Fig. 4B). However, analysis of the FIG. 3. Identification of kinases that can phosphorylate the B 2 receptor. A, HEK293T cells grown on 6-well plates were co-transfected with the human B 2 R and human GRK2, GRK3, GRK4␣, GRK5, and GRK6. Cells were labeled with [ 32 P]orthophosphate, left untreated (Ϫ) or stimulated with 1 M bradykinin (Bk) for 5 min, lysed, and B 2 R was isolated by immunoprecipitation. Following 10% SDS-PAGE and transfer onto nitrocellulose, B 2 R was cleaved with trypsin and the resulting phosphopeptides were separated. Spots were quantified using a PhosphorImager and relative contributions of individual phosphopeptides (white, pS339; gray, pS348; and black, pS346/S348) to total receptor phosphorylation were calculated and ex- phosphopeptide pattern revealed a significantly reduced 32 P incorporation in peptide 3 containing pS346/S348 as well as in peptide 2 representing pS339 (Fig. 5D). These findings are in accord with the observation of a sustained B 2 R signaling upon FR190997 stimulation (30,31), and they lend further support to our hypothesis that phosphorylation of Ser 346 in tandem with Ser 348 is an important event during desensitization of B 2 R-mediated signal transduction. DISCUSSION The implication of receptor phosphorylation in regulation of GPCR functions has been studied for more than a decade (2)(3)(4)32). Most of the knowledge has come from in vitro studies using purified components for reconstitution and/or from mutagenesis approaches targeting anticipated phosphoacceptor sites (2). To date the biochemical identification of in vivo phosphorylation sites of a GPCR and characterization of their biological role(s) has not been successful for any GPCR but rhodopsin (2,8,33,36). In this report we present the precise mapping of phosphorylation sites of the human bradykinin B 2 receptor in its native cellular environment. This approach has allowed us to discriminate between (i) constitutive phosphorylation of human B 2 R on Ser 348 , (ii) homologous phosphorylation at Ser 339 and/or Ser 346 in tandem with Ser 348 , (iii) and heterologous phosphorylation of Ser 346 (Fig. 6). We were also able to follow discrete, but important changes in the phosphopeptide pattern of the B 2 R upon GRK co-expression, during the kinetics of agonist stimulation and over a broad range of ligand concentrations. At last, the phosphorylation of specific residues was correlated with the initiation of receptor internalization and the regulation of its desensitization. This is thus the first report about two-dimensional mapping of in vivo phosphorylation sites of a non-rhodopsin GPCR with a detailed analysis of the importance of specific phosphoacceptor sites in controlling GPCR functions.
A highly sensitive strategy combining two-dimensional phos-phopeptide mapping together with Edman degradation was employed to directly identify basal and agonist-induced phosphorylation sites of the human B 2 R. A recent mass spectrometry study described the constitutive phosphorylation of six out of seven possible serine/threonine residues in the C terminus of rat B 2 R that had been purified from transfected CHO cells (35). However, this report did not provide any quantitative information about the relative frequency of the identified phosphopeptides and did not distinguish between basal, heterologous, and agonist-induced phosphorylation. The present study was performed in an analytical scale, which enabled us to follow B 2 R phosphorylation with temporal resolution in different physiological situations. We could clearly discriminate a basal phosphorylation at Ser 348 and an agonist-mediated phosphorylation of Ser 339 and/or Ser 346 in tandem with Ser 348 in native and recombinant cells. These markedly distinct phosphorylation patterns were not expected, since the two-state model of receptor activation would suggest rather quantitative differences that should reflect the equilibrium between inactive and active, subsequently phosphorylated receptors. The finding that various GRKs affect the relative abundance of phosphorylation of specific phosphoacceptor sites in the B 2 R underlines the power of our approach in identifying subtle positional changes that would escape total phosphorylation studies but may affect receptor fine-tuning. An unanticipated result of our experiments is the functionally compensatory phosphorylation of Ser 346 that was found when the major phosphorylation site Ser 348 had been mutated to alanine. This alternative phosphorylation points to a rather relaxed substrate specificity of the B 2 R kinase(s) that could scan from the receptor C terminus toward the membrane-inserted region for appropriate phosphoacceptor sites. Such a "sliding kinase" mechanism is also supported by the observed quantitative differences in the contribution of identified serine residues to total B 2 R phosphorylation, i.e. Ser 348 Ͼ Ser 346 Ͼ Ser 339 (Fig. 6). Based on mass spectrometry studies, a similar GRK-driven sequential modification has been suggested for the C-terminal portion of rhodopsin (36). However, in other GPCRs such as the ␣ 2A -adrenergic receptor the phosphorylation of multiple serines in the third intracellular loop seems to occur independently of any directional order (13).
Data from GRK co-expression experiments and the finding that 32 P labeling of Ser 348 remained largely invariable during agonist challenge suggests that either a non-GRK activity constantly provides new pS348 for subsequent GRK-mediated Ser 346 phosphorylation or that synchronous de novo phosphorylation of Ser 346 /Ser 348 occurs (Fig. 6B). The former hypothesis would support a priming function of pS348 as it has been proposed for other GPCRs such as rhodopsin, the -opioid and the A 3 adenosine receptor (36 -38). Such a priming phosphorylation of B 2 R at Ser 348 would also convert Thr 345 and Ser 346 to consensus sites for further GRK-mediated phosphorylation (29).
In contrast to the bradykinin-induced dual phosphorylation of Ser 346 and Ser 348 , PKC selectively triggered phosphorylation of Ser 346 that appeared to block rather than promote subsequent Ser 348 phosphorylation. The PKC-induced phosphorylation of Ser 346 in the B 2 R could be involved in agonist-independent, heterologous desensitization as it has been suggested for other GPCRs (2,28,33,34,37,39). This hypothesis is supported by our finding that carbachol stimulation of co-expressed G␣ q -coupled m 1 and m 3 but not of G␣ i -coupled m 2 muscarinic receptors resulted in a moderate but significant increase of B 2 R single Ser 346 phosphorylation (data not shown). However, results from our analysis of the dose dependence of B 2 R phosphorylation also suggest that PKC-mediated phosphorylation contributes to homologous B 2 R desensitization upon stimulation with low doses of bradykinin. Based on inhibitor studies, such a scenario was earlier proposed for protein kinase A in regulating ␤ 2 -adrenergic receptor signaling (40).
Receptor internalization has been implicated in desensitization of GPCRs, although it is often too slow (t1 ⁄2 Ϸ 5-20 min) for a significant contribution to acute desensitization that usually occurs within the first few minutes of agonist challenge (2,4). The sequestration of ␤ 2 -adrenergic receptors and the acidification of the corresponding intracellular compartments have been suggested to constitute the initial steps of resensitization, because both processes were found to be essential for GPCR dephosphorylation (41,42). Indeed B 2 R internalization seems to be necessary for its full dephosphorylation and subsequent resensitization (15, 43) but in addition, receptor phosphorylation was shown to initiate internalization of the B 2 R (17). The FIG. 5. Role of specific phosphorylation sites in the regulation of receptor sequestration and desensitization of the bradykinin-induced phospholipase C activation. A, HEK293T cells were transfected with the wild-type human B 2 R (q) or the following mutants: S339A (‚), S346A (छ), S348A (Ⅺ), S339A/S346A (‰), S346A/ S348A (shaded diamond), ⌬S (ϫ), and ⌬ST (shaded cross). Internalization was studied after binding of 5 nM [ 3 H]bradykinin at 4°C by shifting the temperature to 37°C for the indicated time periods. Extracellular and internalized ligand was separated and quantified by liquid scintillation counting. Means from a typical experiment performed in triplicates are shown. B, desensitization of the B 2 R-induced PLC activation was studied by following intracellular inositol phosphate accumulation in HEK293T cells transfected with the indicated B 2 R mutants and pretreated with 100 nM bradykinin for 5 min. After removal of excess of ligand and a lag phase of 10 (gray bars) or 15 min (black bars) 10 mM LiCl was added and inositol phosphate (IP n ) accumulation was measured for 10 min. Mean Ϯ S.D. from a typical experiment performed in triplicates are shown. C, 32 P-labeled HEK293T cells were stimulated with increasing doses of the non-peptidic B 2 R agonist FR190997, B 2 R was immunoprecipitated, resolved by 10% SDS-PAGE, and visualized using a PhosphorImager. D, typical two-dimensional maps of tryptic phosphopeptides of B 2 R from cells treated either with 10 nM bradykinin (left panel) or 10 nM FR190997 (right panel) are shown.
fact that receptor mutants with deletions of two principal phosphorylation sites (S339A/S346A or S346A/S348A) do not display significant changes in their sequestration kinetics demonstrates that a low stoichiometry of phosphorylation is sufficient to trigger B 2 R internalization. A relaxed phosphorylation requirement with respect to the location of phosphoacceptor sites and stoichiometry has been proposed for internalization of m 2 muscarinic and N-formylpeptide receptors (14,44). Even a B 2 R mutant with all three major phosphorylation sites replaced (⌬S) allowed internalization of a sizable receptor fraction (Ϸ40%). The finding that this fraction was further reduced in the ⌬ST variant, where five potential phosphorylation sites (3 serines and 2 threonines) have been replaced, could be explained by the minute levels of Thr 342 and Thr 345 phosphorylation in the ⌬S mutant that became obvious after pretreatment of cells with serine/threonine phosphatase inhibitors (data not shown).
Unlike the low stoichiometry phosphorylation requirement for receptor internalization, we found that tandem phosphorylation of Ser 346 with Ser 339 or Ser 348 is necessary and sufficient to desensitize the B 2 R-mediated PLC activation. These data correlate well with the findings of Leeb-Lundberg and co-work-ers (18) who described an increased spontaneous activity of a B 2 R mutant replacing, among other residues, Ser 346 and Ser 348 . However, single phosphorylation of the major acceptor site Ser 348 that was also found in the absence of ligand does not affect receptor signaling by itself, but may prime the B 2 R for desensitization. Other examples for the critical role of clustered phosphoserine and phosphothreonine residues in desensitization have been reported for the m 2 muscarinic and N-formylpeptide receptors (14,44). Furthermore, the observation that FR190997, an agonist capable of sustained signaling, is a weak inducer of Ser 346 /Ser 348 phosphorylation points out the importance of this tandem phosphorylation for B 2 R desensitization and provides an intuitive explanation for the delayed B 2 R desensitization upon FR190997 stimulation (30,31).
The results from this comprehensive two-dimensional mapping study of in vivo phosphorylation sites demonstrate the power of this analytical method to reveal subtle temporal and positional changes in the phosphorylation pattern that translate into substantial alterations in the functional capacity of a prototypic GPCR. Future studies will unravel whether the molecular insights into differential phosphorylation requirements for internalization and desensitization of B 2 R hold for GPCRs in general.
FIG. 6. In vivo phosphorylation sites present in the B 2 R tail domain. A, schematic representation of the amino acid sequence (single letter code) of the B 2 R C terminus starting from the potential palmitoylation sites (indicated by serpentine lines). Identified phosphoacceptor sites are numbered and highlighted, and candidate kinases are indicated. The thickness of the arrows point to the relative quantity of phosphate incorporation. B, phosphorylation reactions occurring upon stimulation of B 2 R with bradykinin and PMA and S348A mutant with bradykinin.