The C-terminal Tail of CRTH2 Is a Key Molecular Determinant That Constrains Gαi and Downstream Signaling Cascade Activation*

Prostaglandin D2 activation of the seven-transmembrane receptor CRTH2 regulates numerous cell functions that are important in inflammatory diseases, such as asthma. Despite its disease implication, no studies to date aimed at identifying receptor domains governing signaling and surface expression of human CRTH2. We tested the hypothesis that CRTH2 may take advantage of its C-tail to silence its own signaling and that this mechanism may explain the poor functional responses observed with CRTH2 in heterologous expression systems. Although the C terminus is a critical determinant for retention of CRTH2 at the plasma membrane, the presence of this domain confers a signaling-compromised conformation onto the receptor. Indeed, a mutant receptor lacking the major portion of its C-terminal tail displays paradoxically enhanced Gαi and ERK1/2 activation despite enhanced constitutive and agonist-mediated internalization. Enhanced activation of Gαi proteins and downstream signaling cascades is probably due to the inability of the tail-truncated receptor to recruit β-arrestin2 and undergo homologous desensitization. Unexpectedly, CRTH2 is not phosphorylated upon agonist-stimulation, a primary mechanism by which GPCR activity is regulated. Dynamic mass redistribution assays, which allow label-free monitoring of all major G protein pathways in real time, confirm that the C terminus inhibits Gαi signaling of CRTH2 but does not encode G protein specificity determinants. We propose that intrinsic CRTH2 inhibition by its C terminus may represent a rather unappreciated strategy employed by a GPCR to specify the extent of G protein activation and that this mechanism may compensate for the absence of the classical phosphorylation-dependent signal attenuation.

Prostaglandin D 2 activation of the seven-transmembrane receptor CRTH2 regulates numerous cell functions that are important in inflammatory diseases, such as asthma. Despite its disease implication, no studies to date aimed at identifying receptor domains governing signaling and surface expression of human CRTH2. We tested the hypothesis that CRTH2 may take advantage of its C-tail to silence its own signaling and that this mechanism may explain the poor functional responses observed with CRTH2 in heterologous expression systems. Although the C terminus is a critical determinant for retention of CRTH2 at the plasma membrane, the presence of this domain confers a signaling-compromised conformation onto the receptor. Indeed, a mutant receptor lacking the major portion of its C-terminal tail displays paradoxically enhanced G␣ i and ERK1/2 activation despite enhanced constitutive and agonist-mediated internalization. Enhanced activation of G␣ i proteins and downstream signaling cascades is probably due to the inability of the tail-truncated receptor to recruit ␤-arrestin2 and undergo homologous desensitization. Unexpectedly, CRTH2 is not phosphorylated upon agonist-stimulation, a primary mechanism by which GPCR activity is regulated. Dynamic mass redistribution assays, which allow label-free monitoring of all major G protein pathways in real time, confirm that the C terminus inhibits G␣ i signaling of CRTH2 but does not encode G protein specificity determinants. We propose that intrinsic CRTH2 inhibition by its C terminus may represent a rather unappreciated strategy employed by a GPCR to specify the extent of G protein activation and that this mechanism may compensate for the absence of the classical phosphorylation-dependent signal attenuation.
Prostaglandin D 2 (PGD 2 ) 2 is a lipid mediator that has been considered essential in the development of inflammatory diseases such as asthma and atopic dermatitis (1)(2)(3). It is the major cyclooxygenase metabolite synthesized in allergen-activated mast cells and is released upon their immunological activation (4). The biological effects of PGD 2 are mediated by two G protein-coupled receptors, DP1 and DP2/CRTH2 (chemoattractant receptor homologous molecule expressed on T helper type 2 cells), respectively (5,6). DP1 activation leads to G␣ s -mediated elevation of intracellular cyclic AMP, whereas activation of CRTH2 results in an increase in intracellular Ca 2ϩ levels via the G␣ i pathway and a decrease in cAMP, but also G protein-independent, arrestin-mediated cellular responses have been observed (5)(6)(7).
CRTH2 in particular is expressed on eosinophils, basophils, and T helper type 2 lymphocytes. Activation by PGD 2 or its active metabolites transduces the chemokinetic activity on these immune cells and, by doing so, mediates their recruitment to sites of inflammation (2, 3, 6, 8 -13). In mouse models of allergic asthma or atopic dermatitis, CRTH2 activation promotes eosinophilia and exacerbates pathology (14 -17). In humans, the proinflammatory role of CRTH2 is underscored by the finding that sequence variants conferring enhanced mRNA stability onto the receptor are associated with a higher degree of bronchial hyperreactivity and the occurrence of fatal asthma (16).
Notably, since deorphanization of CRTH2 in 2001 (6), quite a number of reports became available highlighting a proinflammatory role for this receptor in native cells and animal models as well as in humans (2, 3, 6, 8 -13, 16, 18 -22). In contrast, no study has yet addressed structure function relationships of CRTH2 in recombinant cells, and only a single report addresses this topic for the mouse CRTH2 receptor (23). In fact, molecu-
DNA Constructs-The coding sequence of human CRTH2 (GenBank TM accession number NM_004778) was amplified by PCR from a human hippocampus cDNA library and inserted into the pcDNA3.1(ϩ) expression vector (Invitrogen) via 5Ј HindIII and 3Ј EcoRI. To create a mutant CRTH2 receptor lacking the entire C terminus except for the putative helix 8 (herein referred to as CRTH2 ⌬Ctail), a STOP codon was introduced by PCR mutagenesis after amino acid Arg 317 , and the truncated receptor was cloned into pcDNA3.1(ϩ) via 5Ј HindIII and 3Ј EcoRI. Construction of the chimeric G protein G␣ q G66Di5 in the pcDNA3.1(ϩ) ZEO expression vector was reported previously (25). The high affinity nicotinic acid receptor HM74a (26) was cloned from adipose tissue and inserted into pcDNA3.1(ϩ) via HindIII/EcoRI sites. The OXE receptor (27) and the chemerin receptor ChemR23 (28) were cloned from human leukocyte cDNA and inserted via 5Ј HindIII, 3Ј EcoRI and 5Ј BamHI, 3Ј EcoRI, respectively, into pcDNA3.1(ϩ). Correctness of the constructs was verified by restriction endonuclease digestion and sequencing in both directions (MWG Biotech, Ebersberg, Germany).
Cell Culture and Transfection-COS-7 and HEK293 cells were grown in Dulbecco's modified Eagle's medium supple-mented with 10% (v/v) heat-inactivated fetal calf serum, 1% sodium pyruvate, 100 units/ml penicillin, and 100 g/ml streptomycin and kept at 37°C in a 5% CO 2 atmosphere. To facilitate functional analysis of CRTH2 wild type (WT) and CRTH2 ⌬Ctail, respectively, stable HEK293 cell clones were established using media containing 500 g/ml G418 and the FLAG-tagged versions of the receptors. For transient transfections, the calcium phosphate DNA precipitation method was used as previously described (7). For functional inositol phosphate assays, HEK293 cells were transiently cotransfected with CRTH2 WT or CRTH2 ⌬Ctail and a promiscuous G␣ protein facilitating inositol phosphate production by a G i -selective CRTH2 receptor (25).
Membrane Preparation-48 h after transfection, HEK293 cells were harvested, and cell pellets were resuspended in an ice-cold buffer containing 20 mM HEPES and 10 mM EDTA. Then cells were ruptured using Dounce homogenization. Nuclei were pelleted (800 ϫ g, 10 min, 4°C), and the postnuclear supernatant was then fractionated (30,000 ϫ g, 30 min, 4°C) into membrane pellets and supernatants. Pellets were resuspended in buffer containing 20 mM HEPES, 0.1 mM EDTA, and a protease inhibitor mixture (Roche Applied Science) and stored at Ϫ80°C. Membrane protein concentrations were quantified with protein assay kit (Pierce) using bovine serum albumin as a standard.
Whole Cell Binding Experiments-24 h after transfection, HEK293 cells were seeded into poly-D-lysine-coated 96-well plates at a density of 30,000 cells/well. Competition binding experiments on whole cells were then performed ϳ18 -24 h later using 1.0 nM [ 3 H]PGD 2 (172 Ci/mmol; PerkinElmer Life Sciences) in a binding buffer consisting of HBSS and 10 mM HEPES, pH 7.5. Total and nonspecific binding were determined in the absence and presence of 10 M PGD 2 , respectively. Binding reactions were conducted for 3 h at 4°C and were terminated by two washes (100 l each) with ice-cold binding buffer. Radioactivity was determined by liquid scintillation counting in a TopCount liquid scintillation counter (PerkinElmer Life Sciences) (27% counting efficiency) after overnight incubation in MicroScint 20. Binding assays using stable HEK293 cell clones were performed as described above. Saturation binding analysis on stable HEK293 cell clones was performed essentially as described previously (7). Briefly, increasing concentrations of [ 3 H]PGD 2 (specific activity 172 Ci/mmol) were incubated with cells for 3 h at 4°C in the absence or presence of 10 M unlabeled PGD 2 .
Membrane Binding Experiments-Cell membranes from transiently transfected HEK293 cells (15 g of protein) were incubated with 1.0 nM [ 3 H]PGD 2 (172 Ci/mmol) in a binding buffer consisting of HBSS and 100 mM HEPES (pH 7.4) under continuous shaking at 4°C for 3 h. Total and nonspecific binding were determined in the absence and presence of 10 M PGD 2 , respectively. For inhibition of binding by GTP, varying concentrations of GTP were added to the binding mixture. The receptor-bound radioligand was filtered on a Tomtech 96-well Mach III Harvester (PerkinElmer Life Sciences) using filters presoaked with 0.1% polyethyleneimine (Filtermat A; PerkinElmer Life Sciences). Filtration was immediately followed by three rinses with ice-cold 100 mM NaCl. Thereafter, scintillation wax (Meltilex A; PerkinElmer Life Sciences) was melted onto the dried Filtermat. The filters were placed in sample bags (PerkinElmer Life Sciences), and filter-bound radioactivity was measured using a Microbeta Trilux-1450 scintillation counter (PerkinElmer Life Sciences). Determinations were made in triplicates in two independent experiments.
Inositol Phosphate (IP) Accumulation Assays-24 h after transfection, cells were seeded in poly-D-lysine-coated 96-well tissue culture plates and loaded with 0.5 Ci of [2-3 H]myoinositol (TRK911; Amersham Biosciences). The next day, cells were washed twice in HBSS buffer (including CaCl 2 and MgCl 2 ) and stimulated with the respective agonists in HBSS buffer supplemented with 10 mM LiCl for 45 min at 37°C. The reactions were terminated by aspiration and the addition of 50 l of 10 mM ice-cold formic acid/well. After a 90-min incubation on ice, 20 l of the resulting cell extract was transferred to 80 l of yttrium silicate scintillation proximity assay beads (12.5 mg/ml; Amersham Biosciences), and shaken for 60 min at 4°C. Yttrium silicate beads were centrifuged to settle and incubated overnight at 4°C before counting on a TopCount microplate scintillation counter.
cAMP Accumulation Assays-Inhibition of forskolin-stimulated cAMP accumulation in HEK293 cells stably expressing either CRTH2 WT or CRTH2 ⌬Ctail was performed using the HTRF-cAMP dynamic kit (CIS Bio International, Gif-sur-Yvette cedex, France). In brief, cells were resuspended in assay buffer (HBSS, 20 mM HEPES, 1 mM 3-isobutyl-1-methylxanthine) and dispensed in 384-well microplates at a density of 50,000 cells/well. After preincubation in assay buffer for 30 min, cells were stimulated with PGD 2 in the presence of 5 M forskolin for 30 min at room temperature. The reactions were stopped by the addition of 50 mM phosphate buffer (pH 7.0), 1 M KF, and 1.25% Triton X-100 containing HTRF assay reagents. The assay was incubated 60 min at room temperature, and time-resolved FRET signals were measured after excitation at 320 nm using the Mithras LB 940 multimode reader (Berthold Technologies, Bad Wildbad, Germany). Data analysis was made based on the fluorescence ratio emitted by labeled cAMP (665 nm) over the light emitted by the europium cryptate-labeled anti-cAMP (620 nm). Levels of cAMP were normalized to the amount of cAMP elevated by 5 M forskolin alone.
Bioluminescence Resonance Energy Transfer (BRET) Assay-BRET assays were performed on HEK293 cells transiently transfected to co-express human CRTH2-Rluc or CRTH2 ⌬Ctail-Rluc and ␤-arrestin2-GFP2 (green fluorescent protein 2) fusion proteins, using the Mithras LB 940 microplate reader (Berthold Technologies), as described previously (29). In brief, 5 s after the simultaneous addition of agonist and Deep Blue C, light output was measured sequentially at 400 and 515 nm, and the BRET signal (mBRET ratio) was calculated as the ratio of the fluorescence emitted by ␤-arrestin2-GFP2 (515 nm) over the light emitted by the receptor-Rluc (400 nm).
Biotin Protection Degradation Assay-HEK293 cells stably expressing CRTH2 WT or CRTH2 ⌬Ctail were grown to confluence in poly-D-lysine-pretreated 10-cm plates. Cells were treated with 0.3 mg/ml disulfide-cleavable biotin (Pierce) at 4°C for 30 min and washed in TBSC (137 mM NaCl, 25 mM Tris-base, 3 mM KCl, 1 mM CaCl 2 ). Cells were then incubated at 37°C in Dulbecco's modified Eagle's medium for 20 min before adding 10 M PGD 2 or left untreated for another 30 min. The two control plates, 100% and strip, remained at 4°C in TBSC. Except for the 100% plate, all plates were washed in phosphatebuffered saline, and remaining cell surface bound biotin was removed in strip buffer (50 mM glutathione, 75 mM NaCl, 75 mM NaOH, 1% fetal bovine serum) at 4°C for 30 min. All plates were then quenched in buffer containing 9 mg/ml iodoacetamide and 10 mg/ml bovine serum albumin at 4°C for 20 min, followed by cell lysis in IPB (150 mM NaCl, 25 mM KCl, 10 mM Tris-HCl, 1 mM CaCl 2 , 0.1% Triton X-100, pH 7.4, with added protease inhibitors (Complete; Roche Applied Science) and 1 mg/ml iodoacetamide). Cellular debris was removed by centrifugation at 10,000 ϫ g at 4°C for 10 min, and lysates were immunoprecipitated with anti-FLAG M2 antibodies linked with rabbit anti-mouse linker antibodies to Protein A-Sepharose beads at 4°C, washed extensively, and treated with peptide:N-glycanase F at 37°C for 1 h. Samples were denatured in nonreducing SDS sample buffer, resolved by SDS-PAGE using 4 -20% Tris-glycine precast gels (Invitrogen), transferred to nitrocellulose membrane, overlaid with streptavidin (Vectastain ABC immunoperoxidase reagent; Vector Laboratories), and developed with ECL plus reagents (Amersham Biosciences).
Dynamic Mass Redistribution (DMR) Assays (Corning Epic Biosensor Measurements)-A beta version of the Corning Epic system was used, consisting of a temperature control unit, an optical detection unit, and an on-board robotic liquid handling device. Briefly, each well in the 384-well Epic microplate contains a resonant wave guide grating biosensor. The system measures changes in the local index of refraction upon mass redistribution within the cell monolayer grown on the biosensor. Ligand-induced DMR in living cells is manifest as a shift in the wavelength of light that is reflected from the sensor. The magnitude of this wavelength shift is proportional to the amount of DMR. Increase of mass contributes positively and decrease contributes negatively to the overall response. For the Epic system, the penetration depth is 150 nm (i.e. DMR that takes place within penetration depth can be detected) (30,31). 24 h before the assay, HEK293 cells were seeded onto fibronectin-coated 384-well Epic sensor microplates at a density of 15,000 cells/well and cultured for 20 -24 h (37°C, 5% CO 2 ) to obtain confluent monolayers. After the removal of medium, cells were washed with HBSS containing 20 mM HEPES and kept for 1 h in the Epic reader at a constant temperature of 28°C. Hereafter, the sensor plate was scanned, and a base-line optical signature was recorded. Then compound solutions were transferred into the sensor plate, and DMR was monitored for at least 3000 s.
ERK/MAPK Activation-HEK293 cells stably expressing CRTH2 WT or CRTH2 ⌬Ctail were cultured to confluence and then starved in serum-free Dulbecco's modified Eagle's medium overnight with one change of starvation medium after 1.5 h. Cells were treated for 16 h with pertussis toxin (PTX) (Calbiochem) or for 50 min with different inhibitors (bisindolylmaleimide I (Calbiochem), PP2 (Calbiochem), AG 1478 (Calbiochem), Iressa (Astra Zeneca), or the CRTH2-specific arrestin translocation inhibitor (compound 1 in Ref. 7, herein referred to as 27868)) and stimulated for 10 min with 10 M PGD 2 . Following aspiration, cells were frozen in liquid nitrogen and lysed, and lysates were subjected to SDS-PAGE and Western blot analysis as described previously (32).
Receptor Phosphorylation-Receptor phosphorylation was analyzed as previously described (33,34). Briefly, confluent HEK293 cells grown in 6-well plates were transfected with 1.5 g/well plasmid DNA of empty expression vector, CRTH2 WT, CRTH2 ⌬Ctail, or control protein that is known to be highly phosphorylated (ASAP1) (32); all proteins were FLAG-tagged. 24 h after transfection, cells were washed twice with phosphatefree HEPES-buffered Dulbecco's modified Eagle's medium and labeled in the same medium with 0.5 mCi/ml [ 33 P]orthophosphate for 6 h. Following a 5-min stimulation with increasing concentrations of PGD 2 , cells were lysed in 1 ml of ice-cold radioimmune precipitation buffer containing protease and phosphatase inhibitor mixtures (Calbiochem), and FLAG-tagged proteins were isolated with an anti-FLAG monoclonal antibody precoupled to agarose beads (Sigma) under gentle shaking for 3 h at 4°C. Beads were washed three times with lysis buffer, and proteins were released with SDS sample buffer and a 5-min incubation at 98°C. Samples were subjected to SDS-PAGE, transferred to nitrocellulose membranes, and analyzed using a phosphor imaging system (BAS-2000; Fuji). To control expression of CRTH2 proteins, the same nitrocellulose membranes were probed with an anti-FLAG monoclonal antibody (Sigma).
Calculations and Data Analysis-IC 50 and EC 50 values were determined by nonlinear regression using Prism 4.02 (Graph-Pad Software, Inc.). Values of the dissociation and inhibition constants (K d and K i ) were estimated from competition binding experiments using the equations K d ϭ IC 50 Ϫ L and K i ϭ where L is the concentration of radioactive ligand and K d is its dissociation constant.

RESULTS
The C-terminal Tail of CRTH2 Is Important for Cell Surface Expression-Analysis of CRTH2 receptor function in transiently transfected cells is hampered by the poor responses generated upon PGD 2 stimulation. In fact, comparison of the signaling capabilities of a set of bona fide G i/o -coupled receptors in second messenger assays reveals that CRTH2 is functionally expressed but that the efficacy of activation is significantly smaller as compared with other G i -selective receptors, such as the Chemerin receptor ChemR23, the nicotinic acid receptor HM74A, the 5-oxo-eicosatetraenoic acid receptor OXE, or the chemokine receptor CXCR2 (Fig. 1). C termini have emerged as a crucial region of GPCRs governing various aspects of their function, such as expression (35,36), dimerization (37), signaling (36, 38 -40), trafficking (35,36,41), and binding to regula-  (25) that funnels G i -selective receptors to the G q pathway. Cells were treated with increasing agonist concentrations for 45 min and assayed for total IP accumulation, as described in detail under "Experimental Procedures." The following agonists were used: chemerin for ChemR23, nicotinic acid for HM74A, 5-oxoeicostetraenoic acid for the OXE receptor, interleukin-8 for the chemokine receptor CXCR2, and PGD 2 for CRTH2. Data (mean Ϯ S.E., n ϭ 3) are representative of experiments repeated on three separate occasions. tory proteins of the arrestin family (40 -43). Herein, we tested the hypothesis that the C terminus of CRTH2 may serve to constrain receptor signaling. To this end, a CRTH2 truncation mutant was created (hereafter referred to as ⌬Ctail) by cDNA deletion. The mutant protein was terminated at position 317 to remove the major portion of the C-terminal tail except for the domain that contains functionally important residues of the putative cytoplasmic helical structure termed helix 8 (Hx8) ( Fig. 2A). Hx8 is adjacent to transmembrane domain VII and thought to be present in most if not all family A GPCRs (44 -47). Sequence alignments of ⌬Ctail with selected GPCRs, crystal structures of which are available (Fig. 2B), and a molecular model of the ⌬Ctail region forming the end of TM7 (transmembrane 7) and the predicted Hx8 (Fig. 2C) show that this construct retains all of the important elements of Hx8 that permit interaction with TM7, the adjacent intracellular loop 1 as well as the C-terminal domain of G␣ i . The presence of these structural features is important, since their integrity has been shown to be crucial for proper receptor function (38, 39, 41, 45, 48 -53). The effect of C-terminal truncation on CRTH2 surface expression was assessed in whole cell ligand binding assays utilizing [ 3 H]PGD 2 . Although surface expression was significantly reduced for the tail-truncated receptor (Fig. 2D), C-tail deletion was accompanied by an increased affinity of the receptor for its agonist PGD 2 , which in turn might be indicative of an intrinsic inhibitory role of the C-tail. To verify that the observed reduction in binding sites reflected an actual decrease of total receptor number at the cell surface, WT and mutant receptor expression was analyzed by an ELISA (Fig. 2E). ELISA analysis confirmed a reduction in total receptor numbers for ⌬Ctail. To facilitate analysis of receptor behavior in subsequent studies, stable cell lines expressing FLAG-tagged WT and ⌬Ctail receptors were generated, and isolated clones were selected to resemble receptor expression under transient conditions (Fig. 2, F and G). Introduction of N-terminal FLAG tags did not affect [ 3 H]PGD 2 pharmacology of CRTH2 WT and ⌬Ctail, respectively (not shown). Importantly, decrease in cell surface expression of ⌬Ctail was not due to deficient cellular expression, because similar total cellular levels of WT and ⌬Ctail were detected in confocal images and radioligand binding assays on membrane preparations (Fig. 3, A and B).
The CRTH2 C Terminus Is a Negative Regulator of G Proteindependent Signaling-To assess the effect of C-tail truncation on receptor-G protein coupling, CRTH2 WT and ⌬Ctail were tested for their ability to stimulate IP production in HEK293 cells transiently transfected with the receptor constructs and a chimeric G protein linking G i -selective receptors to the G qphospholipase C␤ pathway (Fig. 4A). Interestingly, IP production of the tail-truncated receptor was significantly increased upon PGD 2 stimulation as compared with the WT receptor. In fact, deletion of the tail appears to paradoxically enhance receptor signaling and suggests that the tail may act to constrain maximum receptor function. Functional superiority of the ⌬Ctail is not due to altered affinity of PGD 2 for the receptor in the presence of the chimeric G protein (supplemental Fig. 1A) or to altered surface expression due to G protein cotransfection (supplemental Fig. 1B). Enhancement by tail deletion of receptor signaling was also observed in COS-7 cells, excluding the possibility that functional superiority of ⌬Ctail reflects an artifact of a particular cell line (Fig. 4B). Furthermore, comparable signaling properties of WT and ⌬Ctail were also observed in functional cAMP assays, which do not require the presence of a chimeric G␣ protein (Fig. 4C). To verify differential signaling capability of the respective receptor constructs with yet another method, functional GTP␥S binding assays were performed using membranes from HEK293 cells, stably expressing CRTH2 WT and ⌬Ctail, and transiently cotransfected with the G protein ␣ subunit G␣ i2 . Evidently, this biochemical signaling assay also reveals the functional superiority of ⌬Ctail, since fewer receptors are sufficient to evoke the same extent of G protein activation, as observed for WT receptors (Fig. 4D).
Does ⌬Ctail Adopt a High Affinity Conformation Even after G Protein Activation?-Exchange of GTP for GDP by the receptor-associated G protein is known to lower the receptor affinity for agonists. To test whether the functional superiority of ⌬Ctail was due to its inability to switch to a low affinity conformation subsequent to G protein activation, we examined the effects of GTP on [ 3 H]PGD 2 binding in membrane preparations from HEK293 cells transiently transfected with CRTH2 WT and ⌬Ctail. GTP led to a dose-dependent decrease in high affinity agonist sites in membranes expressing CRTH2 WT (Fig. 4E). However, a comparable decrease in high affinity agonist sites was also observed for ⌬Ctail. Hence, the functional superiority of ⌬Ctail is not due to its inability to undergo a conformational change to the low affinity state after G protein activation.

Real Time Recording of CRTH2 Function Is in Agreement with GTP␥S Binding Data but Distinct from IP and cAMP Assay
Outcomes-Since assays monitoring the accumulation of intracellular second messengers or membrane-based GTP␥S binding assays do not permit real time analysis of receptor signaling, we employed the novel resonant wave guide grating biosensor  . The receptor C terminus is dispensable for G protein-dependent signaling and conformational change to the low affinity state after G protein activation. A, HEK293 cells transiently transfected with the indicated CRTH2 receptor constructs were exposed to increasing concentrations of PGD 2 for 45 min, and the resulting increases in intracellular inositol phosphates were measured using yttrium-coated beads as described under "Experimental Procedures." B, IP hydrolysis assay as described in A in COS-7 cells transiently transfected with CRTH2 WT and ⌬Ctail. C, inhibition of forskolin-stimulated cAMP accumulation by PGD 2 . HEK293 cells stably transfected to express CRTH2 WT or ⌬Ctail were stimulated with varying concentrations of PGD 2 in the presence of 5 M forskolin, and the resulting decrease in cAMP was measured using the HTRFா-cAMP dynamic kit as described under "Experimental Procedures." Data points are shown as mean values Ϯ S.E. of three independent experiments. D, [ 35 S]GTP␥S binding assay on membranes prepared from HEK293 cells stably expressing CRTH2 WT or ⌬Ctail, respectively, and the G protein ␣ subunit G␣ i2 . E, competition for [ 3 H]PGD 2 binding by GTP. HEK293 cells transiently transfected with CRTH2 WT and CRTH2 ⌬Ctail, respectively, were harvested 48 h after transfection for membrane preparation. Fifteen g of membrane protein were incubated in PGD 2 binding buffer containing 1 nM technology (Corning Epic) to resolve signaling capability of CRTH2 WT and ⌬Ctail in real time. In contrast to all other optical studies involving FRET or BRET approaches, this novel biosensor does not require introduction of any label into proteins. As such, this method is noninvasive and label-free and minimizes artifacts due to protein tagging. Activation of GPCRs is known to cause translocation of multiple signaling molecules upon receptor stimulation, and this DMR of cellular matters is captured as an optical signature. Of note, activation of the major G protein signaling pathways engaged by receptors from different coupling classes (G i/o , G s , and G q/11 ) yield pathway-specific optical signatures (54,55). HEK293 cells transiently transfected with CRTH2 WT or ⌬Ctail, respectively, responded with specific optical signatures when exposed to increasing concentrations of PGD 2 . These signatures are unequivocally due to activation of the G i -sensitive CRTH2 receptor, since (i) they are completely abrogated when cells are pretreated with the G␣ i/o inhibitor PTX (Fig. 5A); (ii) they are sensitive to inhibition with the selective CRTH2 antagonist TRQ11238, which when given alone, does not induce any change in DMR (Fig. 5, B and C); and (iii) they are absent in cells transfected with empty pcDNA3.1 vector and challenged with PGD 2 or TRQ11238 (Fig. 5C). Real time DMR recordings show that ⌬Ctail optical signatures were virtually superimposable to those of the WT receptor despite significantly lower cell surface expression (Fig. 5, D-F, and Fig. 2, F and G, for comparison). These results are in agreement with the GTP␥S binding data but in apparent contrast to those obtained in second messenger IP and cAMP assays. Thus, monitoring receptor function in real time may imply that signaling superiority of ⌬Ctail in IP and cAMP assays could be due to a lack of receptor desensitization, which would become apparent under those conditions that allow attenuation of a biological signal to occur in response to sustained agonist stimulation. In agreement with this notion, temporal resolution of IP production reveals that CRTH2 WT but not ⌬Ctail is exposed to a molecular mechanism limiting further stimulation of downstream signaling and suggests that the C terminus plays a crucial role in determining the balance between ligand-stimulated activity and negative feedback inhibition (supplemental Fig. 2). Nevertheless, DMR data do confirm that the C terminus constrains maximum receptor activation since the magnitude of receptor signaling is (i) identical despite significantly lower expression of ⌬Ctail (Fig. 5, D-F) and (ii) significantly increased for ⌬Ctail when cells expressed approximately equal levels of cell surface CRTH2 WT and ⌬Ctail, respectively (Fig. 6, A and B).
The CRTH2 C Terminus Is Required for PGD 2 -mediated Recruitment of ␤-Arrestin2-It is well established that ␤-arrestin family members mediate desensitization of many GPCRs by uncoupling the stimulated receptors from their cognate G proteins (56,57). To test whether CRTH2 WT and ⌬Ctail differ in their ability to recruit arrestin proteins, both receptors were tested for physical association with ␤-arrestin2 using BRET assays. HEK293 cells transiently transfected with either CRTH2 WT or ⌬Ctail fused in frame to Renilla luciferase (energy donor) and ␤-arrestin2-GFP2 (energy acceptor) were stimulated with increasing concentrations of PGD 2 , and BRET was monitored. PGD 2 induced robust arrestin recruitment by CRTH2 WT, whereas the ability to recruit arrestin was lost by C-terminal truncation (Fig. 7). These data define the C terminus as the major site for physical association with arrestin and imply that lack of arrestin recruitment by the ⌬Ctail receptor might account for its inability to limit G protein signaling upon sustained agonist exposure (compare Fig. 4, A and B, and supplemental Fig. 2).
CRTH2 Is Not Phosphorylated upon Agonist Exposure-We have previously reported that CRTH2 upon stimulation with PGD 2 recruits ␤-arrestin2 in a predominantly G protein-independent manner (7). Since CRTH2 recruits arrestin independently of G protein activation, we investigated whether phospho-rylation is a prerequisite for arrestin binding. HEK293 cells transiently expressing FLAG-tagged versions of the receptors were metabolically labeled with 33 P i and stimulated with either PGD 2 or the synthetic agonist Indomethacin. Interestingly, neither basal nor agonist-mediated phosphorylation was detected with a highly sensitive phosphor imager, although both proteins could be visualized with the expected apparent molecular mass bands (ϳ40 versus ϳ35 kDa) (Fig. 8). This finding was rather surprising, but the same technologies and similar experimental conditions have previously been applied successfully to detect phosphorylation of other proteins, including GPCRs (32, 34, 58 -60). This suggests that CRTH2 phosphorylation should have been detected if it had occurred. In agreement with this notion, a positive control for phosphorylation, ASAP1 (32), was successfully phosphorylated in the same experiment. Thus, CRTH2 appears to represent a receptor that utilizes arrestin recruitment but not receptor phosphorylation as a negative feedback mechanism to limit downstream signaling.
Despite Its Inability to Recruit ␤-Arrestin2, CRTH2 ⌬Ctail Remains Competent to Internalize upon Agonist Exposure-Since CRTH2 WT but not ⌬Ctail is capable of recruiting ␤-ar-restin2 to the plasma membrane and many receptors internalize in an arrestin-dependent manner, we sought to determine whether the tail-deleted receptor had lost the ability to internalize in response to agonist stimulation. Importantly, lack of internalization of ⌬Ctail could also contribute to increased functional responsiveness upon agonist challenge and thus explain the divergent signaling properties of both receptors. HEK293 cells stably expressing CRTH2 WT or ⌬Ctail were incubated with anti-FLAG M1 antibody to label surface receptors only. The cells were then stimulated with 1 M PGD 2 for 30 min, and cellular distribution of receptors was visualized by confocal imaging. Unexpectedly, upon stimulation with PGD 2 , immunofluorescence was observed within the cells for both CRTH2 WT and the tail-truncated mutant (Fig. 9A). In fact, FIGURE 6. CRTH2 ⌬Ctail is functionally superior in real time DMR assays. HEK293 cells were transiently transfected with 12 g of CRTH2 WT or 16 g of CRTH2 ⌬Ctail plasmid DNA to express approximately equal amounts of both receptor constructs on the cell surface. Cells were analyzed in parallel for receptor functionality using DMR assays (A) and cell surface expression using ELISA assays (B). DMR assay details are identical to those in the legend of  ⌬Ctail even appeared to exhibit enhanced receptor internalization as compared with the wild type receptor. Similar results were obtained when internalization was quantified using Biotin protection assays (Fig. 9B). Thus, lack of arrestin recruitment, but not lack of agonist-mediated internalization, may explain the functional superiority of ⌬Ctail.

PGD 2 -mediated ERK1/2 Activation by CRTH2 Requires Coupling to G␣ i Proteins and EGF Receptor Transactivation but Does Not Involve Arrestin Proteins or Protein
Kinase C-It is well established that GPCRs are connected to MAPK signaling pathways through G␣ i protein stimulation but in parallel also through a nonclassical, G protein-independent, arrestindependent way (61)(62)(63)(64). Since CRTH2 is coupled to G␣ i -type G proteins and capable of recruiting arrestin without prior G␣ i activation (7), we tested both mechanisms of CRTH2-dependent MAPK activation and whether the C-terminal tail also constrains this downstream signaling event similar to its impact on G␣ i activation. HEK293 cells stably expressing CRTH2 WT or ⌬Ctail were treated with 10 M PGD 2 in the absence and presence of various pharmacological inhibitors, and ERK1/2 phosphorylation was measured using Western blot analysis of whole cell lysates with phospho-ERK-specific antibodies (Fig. 10). We found CRTH2-stimulated ERK1/2 activation to be (i) G␣ i -dependent, since it is sensitive to PTX treatment; (ii) arrestinindependent, since ⌬Ctail remains competent to induce ERK1/2 phosphorylation and since the CRTH2-specific arrestin translocation inhibitor 27868 (compound 1 in Ref. 7) does not impair ERK activation; and (iii) protein kinase C-and Srcindependent, since the inhibitors GF109203X and PP2, respectively, are without effect on ERK phosphorylation but (iv) partly dependent on EGF receptor transactivation due to decreased ERK1/2 phosphorylation in the presence of the EGF receptor tyrosine kinase inhibitor Iressa. Despite lower surface expression of ⌬Ctail (Fig. 2, F and G), similar intensity of ERK activation was observed for both receptors, suggesting that the C terminus also constrains engagement of the ERK signaling cascade similar to constraining G␣ i signaling.
The C Terminus of CRTH2 Does Not Determine G Protein Signaling Specificity-To investigate if the tail of CRTH2 also acts as a G protein signaling specificity filter and whether signaling to alternative G␣ proteins may be similarly enhanced as compared with G␣ i signaling, HEK293 cells transiently transfected with CRTH2 WT or ⌬Ctail, respectively, were tested for their ability to engage the three major G protein signaling pathways: G␣ i , G␣ s , and G␣ q . To analyze and compare signaling capabilities of the wild type and mutant receptor, we took advantage of the novel DMR assays (Corning Epic Biosensor), which allow monitoring of all three major G protein pathways in real time within a single assay platform (54,55). In addition, classical second messenger and biochemical GTP␥S assays were performed to support and validate the optical signatures obtained in the DMR assays ( Fig. 4 and data not shown). It was apparent from DMR assays that deletion of the receptor C terminus yields optical signatures similar to those obtained for CRTH2 WT, suggesting that the engaged signaling pathways for WT and ⌬Ctail are identical and reflect stimulation of G␣ i/o proteins ( Fig. 11A; compare also optical recordings in Fig. 5, D and E). In agreement with this notion, both signatures were abrogated in the presence of PTX, indicating that they result from G␣ i/o activation (compare Figs. 5A and 11A). Furthermore, neither CRTH2 WT nor ⌬Ctail were competent to engage with G␣ s or G␣ q signaling cascades, since no apparent activation of the latter pathways can be detected in the optical signatures. For means of comparison, optical signatures obtained upon stimulating the G␣ s and G␣ q signaling cascades are also depicted (Fig. 11, B and C). Our data clearly show that the receptor C terminus attenuates maximum receptor activation but does not carry any information to activate G proteins selectively.

DISCUSSION
G protein-coupled receptors are endowed with C termini that vary greatly in length and sequence. In many cases, C termini serve as docking sites for regulatory proteins, such as those of the arrestin family (65,66). However, in many instances, tails appear to be dispensable, or of only modest relevance for G protein coupling (42,67,68). The most intriguing finding revealed by our study therefore is that the C terminus of CRTH2 is utilized as a molecular brake to limit the extent of CRTH2-triggered cellular responses. In fact, it is conceivable that the C-terminal tail prevents the heptahelical core of CRTH2 from engagement with G␣ i and its downstream signaling cascade.
Recently, the second extracellular loop domain of the C5a receptor, the closest phylogenetic neighbor of CRTH2, was discovered as a negative regulator of receptor function, most likely FIGURE 8. CRTH2 WT and CRTH2 ⌬Ctail are not phosphorylated upon agonist stimulation. HEK293T cells grown on 6-well plates were transfected with 1.5 g/well of FLAG-tagged CRTH2 WT, ⌬Ctail, or a FLAG-tagged control protein known to be phosphorylated (FLAG-ASAP1). After labeling with 0.625 mCi/ml 33 P-phosphate for 6 h, cells were stimulated with the indicated PGD 2 (A) and indomethacin (B) concentrations for 5 min and lysed, and FLAG-tagged proteins were isolated by immunoprecipitation using anti-FLAG-Sepharose beads. Following SDS-PAGE and transfer onto nitrocellulose membranes, radiolabeled proteins were analyzed using a phosphor imager (top). To control for proper immunoprecipitation, membranes were subsequently probed with an anti-FLAG antibody (bottom), and the expected molecular weight bands for WT and ⌬Ctail were detected.
by stabilizing the inactive receptor conformation (69). Our study identified the CRTH2 C terminus as a key determinant that constrains receptor activation. This notion is supported by the fact that the tail-deleted receptor shows higher affinity toward the agonist PGD 2 (Fig. 2, D and F) and that it paradoxically displays enhanced signaling in a variety of functional assays monitoring activation of G protein-dependent pathways (Figs. 4 -6 and 10). As such, our results highlight the role of the C terminus in allosterically regulating CRTH2 and thus provide evidence that negative regulation of GPCR function is not only brought about by kinase-mediated phosphorylation and arrestin recruitment but also by the receptor itself. Hence, a receptor can take advantage of its C-terminal tail to sterically interdict coupling to G proteins.
C-terminal serines and threonines have been implicated in regulating GPCR desensitization and internalization. Princi-pally, upon receptor activation, these residues are phosphorylated by G protein-coupled receptor kinases and/or second messenger-dependent kinases, which facilitate ␤-arrestin binding and can lead to both G protein uncoupling and internalization by targeting receptors to clathrin-coated pits (70 -72). Although the C terminus of CRTH2 is spiked with potential phosphorylation sites, neither agonist-dependent nor constitutive phosphorylation could be detected (Fig. 8). Thus, CRTH2 lacks the classical mechanism for negative feedback regulation that is operative for the majority of GPCRs (66, 70 -72). It is therefore tempting to propose that lack of negative regulation of CRTH2 signaling through phosphorylation may be compensated by the ability of the C terminus to constrain maximum receptor activation. As such, CRTH2 appears to use an up to now rather unappreciated strategy to limit its own signaling. At present, however, we cannot rule out the possibility that the C terminus is directly involved in interaction with an unknown regulatory protein and that this interaction could impose a similar limitation onto receptor signaling.
Another plausible explanation accounting for the enhanced G protein signaling capacity of the tail-deleted CRTH2 receptor could be its inability to undergo ligand-mediated receptor internalization. Unexpectedly, however, CRTH2 ⌬Ctail remained competent to undergo both constitutive and agonistmediated internalization (Fig. 9). In fact, lack of the C-terminal domain rather facilitated both internalization phenomena. Hence, this internalization phenotype would be counterproductive to efficient signal transduction as observed for ⌬Ctail and cannot explain its functional superiority as compared with the WT receptor. Interestingly, CRTH2 ⌬Ctail internalizes FIGURE 9. CRTH2 ⌬Ctail is incompetent to recruit ␤-arrestin2 but remains competent to internalize. A, HEK293 cells stably expressing either CRTH2 WT (top) or CRTH2 ⌬C-tail (bottom) fused to an N-terminal FLAG tag, were "antibody-fed" for 30 min, followed by 30 min of either no treatment (left) or incubation with 10 M PDG 2 (right). All cells were then permeabilized, immunostained with a fluorescent secondary antibody, and imaged by confocal microscopy. Scale bar, 10 m. B, biotinylation protection assay performed on the same clones as used in A. Cells were treated with membrane-impermeable biotin for 30 min at 4°C labeling only cell surface receptors (100%). The cells were subsequently incubated for 20 min at 37°C before adding 10 M PGD 2 or left untreated (NT) for another 30 min. Noninternalized, surfacebound biotin was stripped away (Strip), and receptors were immunoprecipitated, subjected to Western blot, and visualized using a streptavidin overlay. The blots are representative of two independent experiments. FIGURE 10. The CRTH2 C terminus also constrains receptor signaling to the ERK1/2 cascade. HEK293 cells expressing CRTH2 WT or ⌬Ctail, were grown on 10-cm tissue culture dishes and incubated in the absence of serum for 16 h. Cells were then treated with the indicated inhibitors for 1 h (except for PTX that was present for 16 h) and then challenged with 10 M PGD 2 for 10 min. Cells were lysed, and proteins were separated by SDS-PAGE and transferred onto nitrocellulose membranes. Subsequently, the activation of p42/44 MAPK (ERK1/2) was assessed using a phospho-specific antibody. Parental, nontransfected HEK293 cells served as control for endogenous PGD 2 receptors (mock) and stimulation with 1 M phorbol 12-myristate 13-acetate and 10% fetal calf serum (FCS) was used to obtain maximal p42/44 MAPK activation. WB, Western blot.

Silencing of CRTH2 Signaling by Its C-terminal Tail
without being competent to physically associate with ␤-arrestin2 (Fig. 7). CRTH2 therefore represents a receptor, which is not only competent to internalize without prior phosphorylation but also without prior arrestin recruitment. A similar observation has also been made for the muscarinic M2 receptor (73,74) and for selected phosphorylation-deficient mutants of the somatostatin sst2A receptor, which remain competent to internalize despite their inability to undergo ligand-mediated phosphorylation and arrestin recruitment, respectively (75).
Although ␤-arrestin2 recruitment is not necessary for CRTH2 internalization, PGD 2 stimulation induces arrestin translocation. However, this process does not require prior receptor phosphorylation. Phosphorylation-independent arrestin recruitment has recently also been observed for ␤ 2 -adrenergic, angiotensin II AT1, and parathyroid hormone receptors (76). Our data therefore suggest that arrestin translocation and receptor phosphorylation may be completely separable molecular events and provide additional support for a model in which multiple receptor conformations possess distinct signaling properties that are differentially regulated.
Binding of arrestin proteins to phosphorylated receptors has long been considered as a means by which G protein activation is turned off. Recent evidence, however, indicates that arrestin binding may also serve as a parallel pathway for signal transduction, particularly for initiation of the MAPK signaling cascade (56). We have recently demonstrated that CRTH2 is also competent to recruit arrestin proteins but that this process does not require G protein activation (7). Herein, we demonstrate that CRTH2 is competent to engage with MAPK ERK1/2 signaling but that activation of this pathway is accomplished through G␣ i but not arrestin proteins. Hence, CRTH2 appears to recruit arrestin solely for desensitization purposes but not for propagating downstream signaling to the ERK1/2 cascade. We cannot rule out, however, the possibility that any yet unknown GPCR-mediated signaling events occur as a consequence of arrestin recruitment and that arrestin recruitment does not only serve the purpose of turning off receptor signaling. Nevertheless, the inability of the ⌬Ctail to physically interact with ␤-arrestin2 may likely contribute to the enhanced signaling capacity of this receptor variant.
Recently, a new cardioprotective signaling pathway has been identified for the ␤ 1 -adrenergic receptor utilizing arrestin proteins for transactivation of the EGF receptor that in turn induces MAPK activation (77). Our study revealed that CRTH2-mediated ERK1/2 stimulation resulted, at least in part, from EGF receptor transactivation, as evidenced by using the EGF receptor tyrosine kinase-specific inhibitor Iressa (Fig. 10). However, we can rule out involvement of arrestin proteins in EGF receptor transactivation, since both CRTH2 WT and ⌬Ctail, the latter incompetent to recruit arrestin, remain competent to engage the ERK1/2 pathway. Furthermore, the presence of PTX, an inhibitor of G␣ i/o protein function, abrogated ERK phosphorylation, whereas Iressa only partly diminished it. Since EGF receptor transactivation is a downstream event of G␣ i signaling, intrinsic inhibition of CRTH2 function is also operative for this branch of the kinase signaling network.
In summary, the experiments described herein provide the first detailed investigation of the role of the CRTH2 C terminus in receptor localization and signaling. Our data show that the C terminus is critically important for membrane localization and that it drives recruitment of ␤-arrestin2. Concurrently, the tail acts as an inhibitor of G␣ i and its downstream signaling cascade. CRTH2 is not detectably phosphorylated; nor does it require phosphorylation for arrestin recruitment or arrestin recruitment for internalization. As such, our study reveals that the molecular mechanisms governing CRTH2 receptor signaling and function are distinct from those characteristic for many members of the rhodopsin family of GPCRs. FIGURE 11. The CRTH2 C terminus does not control G protein coupling specificity of the receptor. A, HEK293 cells stably expressing CRTH2⌬Ctail were challenged with 100 nM PGD 2 , and receptor signaling was monitored in real time using DMR on the Corning Epicா biosensor. Overnight treatment (18 h) of cells with 100 ng/ml PTX abrogated the specific PGD 2 optical signature, indicating its G␣ i/o nature. B, optical signature obtained upon stimulation of the G␣ s cascade using 100 M forskolin in naive HEK293 cells. C, optical signature obtained upon stimulation of the G␣ q cascade with a 1 M concentration of the thromboxane agonist U46619 in naive HEK293 cells. U46619 activation was completely abolished when cells were preincubated with a 10 M concentration of the thromboxane/CRTH2 receptor antagonist ramatroban (Ram.). U46619 activation was insensitive to PTX pretreatment (data not shown).