A β-Arrestin/Green Fluorescent Protein Biosensor for Detecting G Protein-coupled Receptor Activation*

G protein-coupled receptors (GPCR) represent the single most important drug targets for medical therapy, and information from genome sequencing and genomic data bases has substantially accelerated their discovery. The lack of a systematic approach either to identify the function of a new GPCR or to associate it with a cognate ligand has added to the growing number of orphan receptors. In this work we provide a novel approach to this problem using a β-arrestin2/green fluorescent protein conjugate (βarr2-GFP). It provides a real-time and single cell based assay to monitor GPCR activation and GPCR-G protein-coupled receptor kinase or GPCR-arrestin interactions. Confocal microscopy demonstrates the translocation of βarr2-GFP to more than 15 different ligand-activated GPCRs. These data clearly support the common hypothesis that the β-arrestin binding of an activated receptor is a convergent step of GPCR signaling, increase by 5-fold the number of GPCRs known to interact with β-arrestins, demonstrate that the cytosol is the predominant reservoir of biologically active β-arrestins, and provide the first direct demonstration of the critical importance of G protein-coupled receptor kinase phosphorylation to the biological regulation of β-arrestin activity and GPCR signal transduction in living cells. The use of βarr2-GFP as a biosensor to recognize the activation of pharmacologically distinct GPCRs should accelerate the identification of orphan receptors and permit the optical study of their signal transduction biology intractable to ordinary biochemical methods.

growing rapidly (1)(2)(3), creating many new orphan receptors whose properties remain undefined (4 -6). 2 Currently characterized GPCRs display many distinct pharmacologies. For example, they interact with a vast array of ligands and generate intracellular signals by multiple second messenger pathways (4,8,9). Based on work with rhodopsin and the ␤ 2 -adrenergic receptor (␤ 2 AR), it has been postulated that members of the GPCR superfamily desensitize via a common mechanism involving the arresting proteins visual arrestin, ␤-arrestin1 and ␤-arrestin2 (10 -13). However, mainly due to the inherent difficulties of examining the interaction of the components mediating desensitization in their native environment or the need for purified reconstituted systems, this has not been clearly established for many GPCRs. Biochemical studies indicate that arrestins regulate GPCR signal transduction (desensitization) by binding agonist-activated receptors that have been phosphorylated by G protein-coupled receptor kinases (GRKs) (12). While the functional source of arrestin molecules targeted to receptors remains unknown, it is apparent that arrestin binding terminates signaling by interdicting receptor interaction with G proteins (12).
To characterize the interaction between ␤-arrestin and different GPCRs and to assess the contribution of GRKs to this process, we examined using confocal microscopy how a green fluorescent protein/␤-arrestin2 conjugate responded to ligandmediated receptor activation. The results demonstrate a critical role for GRKs in the apparently universal regulation of GPCRs by ␤-arrestins. Moreover, they provide the first realtime, live-cell demonstration of a GPCR interacting with one of its regulatory proteins and demonstrate a practical role for ␤arr2-GFP in the study of GPCR activation.

EXPERIMENTAL PROCEDURES
Materials-Isoproterenol was obtained from Sigma and Research Biochemicals International. Anti-mouse antibody was obtained from Sigma and Molecular Probes. Mouse monoclonal antibody against the 12CA5 epitope was purchased from Boehringer Mannheim. Cell culture medium was obtained from Mediatech and fetal bovine serum from Atlanta Biologicals. Physiological buffers were from Life Technologies, Inc. Restriction enzymes were obtained from Promega or New England Biolabs, T4 ligase was from Promega, and Hot Tub DNA polymerase from Amersham. Plasmid containing variants of green fluorescent protein and anti-GFP antibodies were from CLONTECH.
Plasmid Construction, ␤arr2-GFP-Oligonucleotide primers surrounding the XhoI restriction site and C-terminal stop codon of ␤-arrestin2, in the expression vector pCMV5, were used to replace the stop codon with an in frame BamHI restriction site by directed mutagenesis (14). The product nucleotide was purified by electrophoresis on a 1.5% agarose gel and isolated. It was digested using XhoI and BamHI restriction enzymes, and the nucleotide fragment flanked by these sites was repurified. The N-terminal, proximal cDNA fragment of ␤-arrestin2 flanked by Sac/Xhol restriction sites was removed by digestion from pCMV5. It was ligated with the XhoI/BamHI fragment isolated above and with the purified expression vector (pS65T-GFP) that had been opened between the SacI/BamHI polylinker restriction sites using the respective enzymes (15). The resulting construct was grown in competent Escherichia coli, isolated, and verified by sequencing.
Confocal Microscopy-HEK-293 cells transfected as described above were plated onto 35-mm dishes containing a centered, 1-cm well formed from a glass coverslip sealed hole in the plastic. Primary and secondary antibody labeling of live cells was performed at 37°C for 30 min in media without serum in a 5% CO 2 incubator. Cells were washed three times between applications. Cells plated as above in minimal essential medium or Dulbecco's modified Eagle's medium buffered with 20 mM Hepes were viewed on a Zeiss laser scanning confocal microscope.
Sequestration-Flow cytometry analysis was performed as described (17).
Confocal microscopy of ␤arr2-GFP in an HEK-293 cell ( Fig.  2A) shows that in the absence of receptor activation ␤-arrestins are distributed throughout the cytosol and excluded from the nucleus. Moreover, the data demonstrate that ␤-arrestins are not predominantly compartmentalized at the plasma membrane prior to agonist stimulation. Upon the addition of saturating concentrations of the agonist isoproterenol to the cell medium, an enhancement of plasma membrane fluorescence and a concomitant loss of cytosolic fluorescence (Fig. 2B) can be readily observed and quantified (Figs. 2, C and D). This observation indicates that ␤-arrestins are not discretely compartmentalized and that the entire cytoplasmic content represents a functional ␤-arrestin reservoir.
HEK-293 cells overexpressing the ␤ 2 AR were used to investigate whether the main target of translocated ␤arr2-GFP was a plasma membrane site other than a GPCR. N-terminal epitope-tagged ␤ 2 ARs were cross-linked to one another prior to agonist exposure using a mouse monoclonal antibody against the epitope and a secondary goat anti-mouse antibody conjugated to the fluorophore Texas Red. Fig. 3 demonstrates that the geometry of the agonist-induced time-dependent translocation of ␤arr2-GFP to the plasma membrane mirrors the distribution of preaggregated ␤2ARs, strongly suggesting that the primary targeted site of ␤-arrestin is the ␤ 2 AR.
␤arr2-GFP translocation to ␤2ARs is not limited to HEK-293 FIG. 1. Linear model of the ␤-arrestin2/S65T-GFP conjugate and its characterization by SDS-polyacrylamide gel electrophoresis and ␤2AR sequestration. A, the model indicates the relative size of ␤-arrestin2 compared with GFP. In this construct the Cterminal stop codon of ␤-arrestin2 is replaced by a BamHI restriction site, and ␤-arrestin2 cDNA is inserted in frame to GFP between SacI and BamHI. B, shown are Western blots of homogenates from HEK-293 cells expressing ␤arr2-GFP. Equal amounts of material were loaded into each lane. The gel on the left was exposed to anti-␤-arrestin antibody (16), whereas the right gel was exposed to a mouse monoclonal antibody against GFP. The position of endogenous cellular ␤-arrestin2 is indicated by the lower bar on the left. The heavy band below 71,000 (upper left) corresponding to ␤arr2-GFP is mirrored by a similar band in the right gel. In contrast, no band corresponding to endogenous cellular ␤-arrestin2 is observed with anti-GFP antibody exposure. C, ␤-arrestin activity can indirectly be assessed by measuring its effect on receptor sequestration (16,22). The sequestration of ␤ 2 AR in COS cells with and without overexpressed ␤-arrestin2 (left bars) or with and without overexpressed ␤arr2-GFP (rightmost bars) is presented. Wild type ␤-arres-tin2 and ␤arr2-GFP enhance ␤2AR sequestration equally well, producing a 2. cells, but is also observable in COS-7 cells (Fig. 4). Consistent with their relatively larger surface area and lower efficiency of ␤ 2 AR sequestration compared with HEK-293 cells, agonistmediated ␤arr2-GFP translocation was less apparent in the COS cells. (Fig. 4, A and corresponding image C) (16). However, by coexpressing GRK2, the agonist-mediated ␤arr2-GFP translocation could be enhanced (Fig. 4, B and corresponding image D), suggesting that GRK phosphorylation increases the affinity of the receptor for ␤-arrestin (22). To further characterize the role of GRK phosphorylation in ␤-arrestin translocation, we examined the ability of a GRK phosphorylation-impaired mutant Y326A-␤ 2 AR to support ␤-arrestin redistribution (11,17,23,24). Consistent with its inability to be phosphorylated by endogenous GRKs, the Y326A-␤2AR mutant did not induce ␤arr2-GFP translocation with agonist exposure (Fig. 5A). However, with overexpression of GRK2 and agonist treatment, the Y326A mutant-mediated ␤arr2-GFP translocation (Fig. 5B) was indistinguishable from ␤ 2 AR-mediated translocation (7,(25)(26). These results indicate ␤arr2-GFP translocation not only accurately monitors the biology of the GPCR activation process but the GPCR phosphorylation state as well.
To establish that agonist-induced ␤arr2-GFP translocation represents a general property of GPCR activation and is not limited to the ␤2AR, other members of the GPCR superfamily were evaluated for their ability to mediate the movement of ␤arr2-GFP in HEK-293 cells. Shown in Fig. 6 are results with the dopamine D 1A receptor. Its behavior is representative of 16 different GPCRs that were tested belonging to the the angiotensin, ␣and ␤-adrenergic, dopamine, endothelin, intestinal peptide, chemokine, and opioid receptor subfamilies. Activation of the D 1A receptor with 20 M dopamine produced an increase in the amount of membrane-associated ␤arr2-GFP. Moreover, the increase in ␤arr2-GFP translocation was enhanced in the presence of overexpressed GRK2 (Fig. 6D). DISCUSSION In this work we demonstrate that ␤-arrestin interacts with GPCRs immediately following agonist stimulation and GRK phosphorylation. ␤arr2-GFP translocation was observed in response to more than 15 different GPCRs. Even though these GPCRs respond to a diverse array of ligands and different classes of G proteins, activation of each of the GPCRs elicits the agonist-dependent translocation of ␤arr2-GFP, with the magnitudes of plasma membrane fluorescence signals ranging up to 10 -20-fold above the intracellular background. While ␤arrestin behavior is regulated by multiple components of the signal transduction system, it is particularly sensitive to how well cellular GRKs are able to phosphorylate a particular GPCR. This was demonstrated with both GRK2 and GRK5 (data not shown). For instance, following overexpression of GRK2 to force phosphorylation of the Y326A-␤ 2 AR mutant, the mutant-mediated ␤arr2-GFP translocation is indistinguishable from the wild type ␤ 2 AR-mediated response. Consequently, with the appropriate cellular system, such as COS-7 cells in which endogenous GRKs and ␤-arrestins are relatively poorly expressed, the ␤-arrestin translocation paradigm could also be used to easily monitor the activity and specificity of each of the members of the GRK and arrestin families.
Biochemical measurements of GPCR properties, such as ligand binding, activation of G proteins or effectors, generation of second messengers, or extent of phosphorylation, assess functions that are receptor-specific and do not easily lend themselves to the development of rapid or convenient screening methods. However, since GPCR activation ultimately terminates with the association of ␤-arrestin and receptor, a convergent step of the GPCR signal transduction paradigm, the cellular visualization of the agonist-mediated translocation of ␤arr2-GFP provides a universal measure for detecting the activation of unknown GPCRs. Despite its present large size, the G protein-coupled receptor superfamily continues to expand rapidly as new receptors are discovered through automated sequencing of cDNA libraries. It is estimated that several thousand GPCRs may exist in the human genome, and at present with only a fraction of the genome sequenced, as many as 250 GPCRs have been cloned and only as few as 150 have been associated with ligands (4). 2 The means by which these or newly discovered orphan receptors will be associated with their cognate ligands and physiological functions represents a major challenge to biological and biomedical research. The identification of an orphan receptor generally requires an individualized assay and a guess as to its function. The interrogation of a GPCR's signaling behavior by monitoring ␤arr2-GFP translocation eliminates these prerequisites, since it can be performed with unlabeled ligands and without any prior knowledge of other signaling events. It is sensitive, rapid, easily performed, and should be potentially applicable to nearly all GPCRs, since the majority of these receptors should desensitize by a common mechanism, i.e. interaction with ␤-arrestins.
The visualization of ␤-arrestin2 translocation represents the first direct real-time assessment in a living cell of the interaction of a GPCR with one of its regulatory components. Moreover, the rapid and profound increases in the relative and absolute amounts of plasma membrane-bound ␤-arrestin provide an optical detection of GPCR signal transduction that is as sensitive as any chemical amplification normally produced by second messenger cascades. Therefore, ␤arr2-GFP is not only exquisitely adept as a biosensor for monitoring GPCR activation, but represents an excellent tool to study the kinetics and specificity of components involved in the regulation of GPCR activity. Furthermore, when used as an optical sensor, ␤arr2-GFP provides the unique potential to unite orphan GPCRs with their corresponding ligands.