β-Arrestin Recruitment and G Protein Signaling by the Atypical Human Chemokine Decoy Receptor CCX-CKR*

Background: CCX-CKR is considered to be a chemokine decoy receptor that is unable to signal. Results: Chemokines induce β-arrestin recruitment to CCX-CKR and pertussis toxin (PTX)-dependent CRE activity. Conclusion: PTX-sensitive G proteins hinder CCX-CKR coupling to other G proteins and consequently keep receptors silent. Significance: Recruitment of β-arrestin to CCX-CKR requests re-evaluation of the signaling capacity of this atypical receptor. Chemokine receptors form a large subfamily of G protein-coupled receptors that predominantly activate heterotrimeric Gi proteins and are involved in immune cell migration. CCX-CKR is an atypical chemokine receptor with high affinity for CCL19, CCL21, and CCL25 chemokines, but is not known to activate intracellular signaling pathways. However, CCX-CKR acts as decoy receptor and efficiently internalizes these chemokines, thereby preventing their interaction with other chemokine receptors, like CCR7 and CCR9. Internalization of fluorescently labeled CCL19 correlated with β-arrestin2-GFP translocation. Moreover, recruitment of β-arrestins to CCX-CKR in response to CCL19, CCL21, and CCL25 was demonstrated using enzyme-fragment complementation and bioluminescence resonance energy transfer methods. To unravel why CCX-CKR is unable to activate Gi signaling, CCX-CKR chimeras were constructed by substituting its intracellular loops with the corresponding CCR7 or CCR9 domains. The signaling properties of chimeric CCX-CKR receptors were characterized using a cAMP-responsive element (CRE)-driven reporter gene assay. Unexpectedly, wild type CCX-CKR and a subset of the chimeras induced an increase in CRE activity in response to CCL19, CCL21, and CCL25 in the presence of the Gi inhibitor pertussis toxin. CCX-CKR signaling to CRE required an intact DRY motif. These data suggest that inactive Gi proteins impair CCX-CKR signaling most likely by hindering the interaction of this receptor with pertussis toxin-insensitive G proteins that transduce signaling to CRE. On the other hand, recruitment of the putative signaling scaffold β-arrestin to CCX-CKR in response to chemokines might allow activation of yet to be identified signal transduction pathways.

Chemokine receptors form a large family of G protein-coupled receptors (GPCRs) 3 involved in migration, activation, and differentiation of immune cells. The 19 known human chemokine receptors and their ϳ50 endogenous peptide ligands form a complex system in which many chemokine receptors can bind multiple chemokines, and many chemokines can bind more than one receptor (1). Chemokine receptors signal predominantly via heterotrimeric G i proteins, resulting in an inhibition of cAMP production by adenylyl cyclases and induction of intracellular calcium mobilization (2). Interestingly, the chemokine receptors D6, DARC, CCX-CKR, CXCR7, and CCRL2 bind chemokines with high affinity, but ligand binding does not result in G protein-mediated intracellular calcium mobilization or chemotaxis (3). Rather, these atypical chemokine receptors act as decoy receptors to regulate chemokine availability. Upon internalization, the receptor-bound chemokines are either targeted for lysosomal degradation or transported across the cell to be subsequently exposed or released on the other side of the cell (i.e. transcytosis) (3). CCX-CKR is a high affinity receptor for the chemokines CCL19/ELC, CCL21/ SLC, and CCL25/TECK (4). These chemokines are important for the development of acquired immunity by activating CCR7 or CCR9. CCL19 and CCL21 recruit CCR7-expressing dendritic and T cells into the T cell compartments of secondary lymphoid organs (5). CCL25, on the other hand, recruits antigen-experienced lymphocytes to the small intestine by activating CCR9 (6). Mouse CCX-CKR knock-out models demonstrated that CCX-CKR is important for steady-state homing of dendritic cells to skin-draining lymph nodes, T cell differentiation, and immune response kinetics in an experimental autoimmune encephalomyelitis model, by regulating local chemokine levels (7,8). Moreover, CCR7 and CCR9 have been shown to play a role in various cancers (9). CCX-CKR scavenges CCL19 and CCL21 both in vitro and in vivo, thereby decreasing the free concentration of these chemokines (7,8,10). Indeed, low CCX-CKR expression levels in breast cancer tumor cells were correlated with an increase in lymph node metastasis and consequently poor survival rate of patients (11). Additionally, CCX-CKR has been suggested to mediate CCL19 and CCL21 transcytosis across lymphatic endothelium, although direct evidence is still lacking (12). CCX-CKR is expressed in many tissues, including heart, lung, and intestine, as well as by stromal cells of skin-draining lymph nodes, thymic epithelial cells, and a number of hematopoietic cell types (4,7,8,(13)(14)(15)(16). Transgenic overexpression of CCX-CKR decreased hematopoietic precursor cell numbers in the thymic anlage at embryonic stages, whereas cell numbers returned to normal levels in newborn and adult mice (7).
Following or alternative to G protein coupling, activated GPCRs can also recruit ␤-arrestin upon phosphorylation of their intracellular domains. ␤-Arrestin bound to the GPCR may then act as a scaffold protein for receptor internalization and G protein-independent signaling to, for example, extracellular stimulus-regulated kinase 1/2 (ERK1/2) and Akt (protein kinase B) (17). The atypical chemokine receptor CXCR7 is internalized and signals exclusively through ␤-arrestin in a chemokine-dependent manner (18,19). D6 constitutively recruits ␤-arrestin, which is essential for the observed continuous internalization of D6 (20). However, a more recent study provided evidence that ␤-arrestin was not necessarily involved in D6 internalization but increased receptor stability (21). CCX-CKR has been previously suggested to internalize CCL19 in a ␤-arrestin-independent manner (10). In the present study, however, we demonstrate for the first time the concentrationdependent recruitment of ␤-arrestins to the atypical chemokine receptor CCX-CKR upon stimulation with CCL19, CCL21, or CCL25 using three different methodologies in various transfected cell lines. Moreover, we provide evidence that G i proteins impair CCX-CKR-mediated signaling to CRE. These new aspects of CCX-CKR signaling provide additional avenues through which the role of CCX-CKR may be further explored.
High Content Analysis of ␤-Arrestin-2 Redistribution-U2OS-␤-arr2-GFP, U2OS-PTHR1, or U2OS-CCX-CKR cells were seeded in clear-bottom 96-well plates (PerkinElmer Life Sciences) with 15,000 cells/well in 90 l of assay medium (DMEM/F-12 supplemented with 2% (v/v) fetal calf serum, 100 units/ml of penicillin, and 100 g/ml of streptomycin) and cultured overnight. The following day, medium was replaced with 45 l of assay medium. Five microliters of assay medium containing chemokines was added and cells were incubated at 37°C for 45 min. Cells were fixed in 4% (v/v) paraformaldehyde (BioConnect, Huissen, The Netherlands) and incubated with 1 M Hoechst (Invitrogen) in PBS for 30 min. Plates were analyzed on an Operetta automated fluorescence microscope (PerkinElmer Life Sciences).
Enzyme Fragment Complementation-based ␤-Arrestin Recruitment Assays-Cells were seeded in 384-well Cultur-Plates (PerkinElmer) at 10,000 cells/well in 15 l of Opti-MEM (Invitrogen) containing 1% (v/v) bovine calf serum (assay medium). The next day, cells were stimulated with 10 l of chemokine in assay medium and then returned to the incubator for 90 min. Cells were disrupted using 12 l of substrate-containing lysis buffer from the PathHunter Detection Kit in the formulation specified by the supplier (DiscoveRx). Plates were incubated in the dark for 90 min at room temperature before measurement of ␤-galactosidase activity (luminescence) on an Envision multilabel plate reader (PerkinElmer).
BRET-based ␤-Arrestin Recruitment Assay-One day after transfection, HEK293T cells were transferred to poly-L-lysinecoated white 96-well plates. Growth medium contained 100 ng/ml of PTX (Sigma) where applicable. The next day, medium was replaced with Hanks' balanced salt solution containing 100 ng/ml of PTX where applicable and fluorescence was measured on a Victor 3 multilabel plate reader (excitation 485 nm; emission 535 nm; PerkinElmer Life Sciences). Ten minutes after addition of coelenterazine-h (5 M final concentration; Promega, Madison, WI), ligand solutions in Hanks' balanced salt solution supplemented with 0.05% BSA were added in the stated concentrations and incubated for an additional 5 min. BRET (emission 535 nm) and Rluc expression (emission 460 nm) were measured with a Victor 3 multilabel plate reader. Baseline-corrected BRET ratios were calculated by first dividing BRET by Rluc emission values, followed by subtraction of the BRET ratio of cells expressing CCX-CKR-Rluc alone.
cAMP Accumulation Assay-cAMP measurements in CHO-CCX-CKR cells were performed with a homogenous time-resolved fluorescence kit from Cisbio (Gif-sur-Yvette, France) essentially as described previously (27). Briefly, chemokines diluted in 10 l of dilution buffer either containing forskolin (0.5 M final concentration) or DMSO control were dispensed in 384-well OptiPlates (PerkinElmer Life Sciences). Subsequently, 7,500 CHO-CCX-CKR cells in 10 l of assay medium (DMEM/F-12 containing 5 g/ml of apo-transferrin (Sigma), 1 g/ml of insulin, 100 units/ml of penicillin, and 100 g/ml of streptomycin) were added and plates were incubated at 37°C for 60 min. Cells were disrupted by addition of 10 l of cAMP-XL665 conjugate and 10 l of Europium-labeled anti-cAMP antibody diluted in lysis buffer (all provided with the kit). Plates were incubated in the dark at room temperature for 60 min before measurement of time-resolved fluorescence at 615 and 665 nm on an Envision multilabel plate reader (PerkinElmer Life Sciences). Absolute cAMP concentrations were calculated from a cAMP standard.
Inositol Phosphate Accumulation Assay-Inositol phosphate measurements in CHO-CCX-CKR cells were performed using a homogenous time-resolved fluorescence kit from Cisbio (Gifsur-Yvette, France). CHO-CCX-CKR cells were plated in 384well proxiPlates with 10,000 cells/well in 20 l of assay medium (DMEM/F-12 containing 10% (v/v) fetal calf serum, 100 units/ml of penicillin, and 100 g/ml of streptomycin) and were allowed to adhere overnight. Medium was aspirated and cells were stimulated with 14 l of chemokine in stimulation buffer (provided in the kit) containing 0.1% (w/v) bovine serum albumin (BSA; Sigma). Plates were incubated at 37°C for 90 min before the addition of 3 l of IPone-XL665 conjugate and 3 l of Europium-labeled anti-IPone antibody diluted in lysis buffer (all provided with the kit). Plates were incubated in the dark at room temperature for 60 min before measurement of timeresolved fluorescence at 615 and 665 nm on an Envision multilabel plate reader. Absolute IPone concentrations were calculated from an IPone standard.
Cyclic AMP-responsive Element (CRE)-driven Reporter Gene-One day after transfection, HEK293T cells were transferred to poly-L-lysine-coated white 96-well plates. Where required, growth medium contained 100 ng/ml of PTX. The following day, growth medium was replaced with serum-free medium supplemented with 0.05% BSA and 100 ng/ml of PTX, 1 M forskolin (FSK), and/or ligands, as indicated. After 6 to 8 h incubation at 37°C, stimulation medium was removed and cells were incubated for 5 min with 25 l of substrate solution (39 mM Tris⅐H 3 PO 4 , pH 7.8, 39% glycerol, 2.6% Triton X-100, 860 M dithiothreitol, 18 mM MgCl 2 , 825 M ATP, 77 M disodium pyrophosphate, 230 g/ml of beetle luciferin) (Promega). Luminescence was measured using a Victor 3 multilabel plate reader.
BRET-based cAMP Biosensor Assay-Changes in cAMP levels were detected using a BRET-based cAMP sensor in a similar manner as the BRET-based ␤-arrestin recruitment assay, except that cells were rinsed once with Hanks' balanced salt solution, and incubated with fresh Hanks' balanced salt solution for 30 min before being stimulated. In addition, the nonspecific phosphodiesterase 3-isobutyl-1-methylxanthine was added simultaneously with coelenterazine-h to a final concentration of 40 M (23).
Radioligand Binding and Internalization-One day after transfection, HEK293T cells were transferred to poly-L-lysine (Sigma)-coated 96-well plates. The next day, whole cells were incubated for 4 h at 4°C with ϳ0.75 nM 125 I-CCL19 in binding buffer (50 mM HEPES, 100 mM NaCl, 1 mM CaCl 2 , 5 mM MgCl 2 , pH 7.4, 0.5% BSA) containing the indicated concentrations of unlabeled displacer. Incubations were terminated by washing the cells with ice-cold wash buffer (50 mM HEPES, 0.5 M NaCl, 1 mM CaCl 2 , 5 mM MgCl 2 , pH 7.4) followed by lysis in RIPA buffer (0.5% Nonidet P-40, 0.5% sodium deoxycholate, and 0.1% sodium dodecyl sulfate). Cell lysates were transferred to vials and counted in a Wallac Compugamma counter (PerkinElmer Life Sciences). 125 I-CCL19 internalization assays were performed in a similar manner as binding assays on intact cells, except that incubations were performed in a 37°C waterbath (28). Incubations were terminated at the indicated time intervals by placing the 48-well plates on ice and rapid removal of free 125 I-CCL19 by washing the cells three times with ice-cold wash buffer. The fraction 125 I-CCL19 bound to receptors at the cell surface was removed using ice-cold acidified DMEM (pH 2.0). The cells were then solubilized in RIPA buffer to collect the acidresistant (i.e. internalized) 125 I-CCL19 fraction, which was quantified using a Wallac Compugamma counter. Control experiments confirmed that acid treatment removed all surface-bound chemokine.
ELISA-One day after transfection, HEK293T cells were seeded in 96-well plates and cultured overnight. Next, the cells were fixed for 30 min with 4% formaldehyde in PBS and washed twice with Tris-buffered saline (TBS; 150 mM NaCl, 10 mM Tris⅐HCl, pH 7.5). Half of the samples were permeabilized by incubating the cells for 30 min with 0.5% Nonidet P-40 in TBS after fixation. Samples were then blocked for 4 h at room temperature with 1% fat-free milk in 0.1 M NaHCO 3 (pH 8.6). Samples were incubated overnight at 4°C with the mouse anti-hCCX-CKR clone 2F11 (a kind gift from Dr. Chiba, Tokyo University of Science, Japan) diluted 1:1000 in TBS containing 1% fat-free milk powder and subsequently washed with TBS. Samples were incubated for 3 h at room temperature with goat anti-mouse HRP-conjugated antibody (Bio-Rad) diluted 1:2500 in 0. Data Analysis-Sigmoidal concentration-response curves were plotted using GraphPad Prism 5.0 software. All data are presented as averages of mean Ϯ S.E. Statistical analyses were performed using GraphPad Prism 5.0 software.
BRET technology was then used to confirm this close proximity between CCX-CKR and ␤-arrestin2 upon stimulation with CCL19, CCL21, and CCL25 ( Fig. 3B; Table 1). Moreover, a similar concentration-dependent increase in BRET ratio was observed between CCX-CKR-Rluc and ␤-arrestin1-EFYP in response to these chemokines ( Fig. 3C; Table 1). CCL19 was the most potent chemokine in inducing ␤-arrestin1/2 recruitment to CCX-CKR, whereas CCL21 and CCL25 had comparable potency. To evaluate whether ␤-arrestin1/2 recruitment to CCX-CKR required phosphorylation of the CCX-CKR C-terminal tail, all serine and threonine residues in this domain were alanine substituted (i.e. 323 SWRRQRQSVEEFPFDSEGPTEPTSTFS 349 into 323 AWRRQRQAVEEFPFDAEGPAEPAAAFA 349 ). The expression level of this CCX-CKR-S(T/A)-Rluc mutant was ϳ60% of that of non-mutant CCX-CKR-Rluc, as determined by Renilla luciferase activity and ELISA on intact cells (data not shown). However, binding assays showed that CCX-CKR-S(T/A)-Rluc had no affinity for 125 I-CCL19 and 125 I-CCL25, and consequently was not able to recruit ␤-arrestin2-EYFP in response to chemokines (data not shown).
CCX-CKR Does Not Activate Typical Chemokine Receptorinduced ␤-Arrestinor G Protein-mediated Signaling-To evaluate the ability of CCX-CKR to activate downstream effectors of ␤-arrestin signaling, CHO-CCX-CKR cells were stimulated with vehicle or 100 nM CCL19 for 2 to 90 min and lysates were probed for phosphorylation of ERK1/2 (Thr 202 /Tyr 204 ) and Akt (Ser 473 ). CCL19-induced phosphorylation of ERK1/2 was not observed, whereas treatment with the positive reference PMA for 10 min elicited robust ERK1/2 phosphorylation (Fig. 4A). Furthermore, CCL19 did not affect Akt phosphorylation in CHO-CCX-CKR cells, whereas treatment with insulin caused a marked increase in Akt phosphorylation (Fig. 4B). In summary, no CCX-CKR-induced signaling to ERK1/2 or Akt was observed.
Because most chemokine receptors signal through G i proteins upon receptor activation, CHO-CCX-CKR cells were stimulated with 100 nM CCL19, CCL21, or CCL25 for 60 min and cAMP levels were subsequently determined using an homogenous time-resolved fluorescence-based assay. These chemokines did not affect basal or FSK-induced cAMP accumulation, suggesting that CCX-CKR is not coupling to G s or G i   proteins, respectively (Fig. 4C). This was not due to an inability of the chemokines to induce such responses, as CCL19 and CCL25 decreased FSK-induced cAMP levels in CHO-CCR7 and CHO-CCR9 cells, respectively (data not shown). The ability of CCX-CKR to activate G q proteins was also investigated. Intracellular inositol phosphate concentrations in CHO-CCX-CKR cells were not affected by treatment with CCL19, CCL21, and CCL25 (Fig. 4D). In contrast, inositol phosphate formation was observed in CHO-CCX-CKR cells after activation of endogenous purinoceptors by 100 M ATP (Fig. 4D).
CCX-CKR IL2/3 Chimeras Stimulate cAMP Signaling-In contrast to CCR7 and CCR9, the atypical chemokine receptor CCX-CKR did not decrease FSK-induced cAMP levels in response to CCL19 and CCL25 stimulation. CCX-CKR chimeras were constructed by substitution of intracellular loop (IL) 2 and/or 3 with corresponding sequences of CCR7 and CCR9 to investigate whether intracellular domains of CCX-CKR might impede G protein coupling (Fig. 5a). These CCR7 and CCR9 ILs were hypothesized to confer G i -coupling capacity to the CCX-CKR chimeras, resulting in decreased adenylyl cyclase activity. A CRE-driven luciferase reporter gene assay was used to quantify chemokine-stimulated changes in cAMP levels in HEK293T cells transiently expressing wild type CCX-CKR or CCX-CKR chimeras. Cell surface expression of all CCX-CKR chimeras was comparable (p Ͼ 0.05) to wild type CCX-CKR as measured by ELISA (Fig. 5B), whereas 125 I-CCL19 binding to CCX-CKR chimeras was significantly increased (p Ͻ 0.05) in comparison to wild type CCK-CKR (Fig. 5C). Except CCX/ R9IL3 bound less 125 I-CCL19 than wild type CCX-CKR. As expected, FSK-induced CRE activity was decreased by treat-ment with 100 nM CCL19 in cells expressing CCR7 but not in cells transfected with wild type or chimeric CCX-CKR constructs (Fig. 5, D and E). CCL19 does not interact with CCR9 and consequently had no effect on FSK-induced CRE activity in CCR9-expressing cells. As a control for G i -mediated signaling responses, cells were pre-treated overnight with 100 ng/ml of the inhibitor Bordetella pertussis toxin (PTX). As anticipated, PTX abolished the G i -mediated decrease in CRE activity in CCR7-expressing HEK293T cells in response to CCL19. On the other hand, PTX significantly potentiated the CCL19-induced increase in CRE activity in cells expressing wild type CCX-CKR, CCX/R7IL3, CCX/R7IL2IL3, and CCX/R9IL2IL3 (Fig. 5D). Pre-treatment with PTX did not affect the binding affinities of CCL19, CCL21, and CCL25 for CCX-CKR (Fig. 6, A-C). Likewise, the potencies of these chemokines to recruit ␤-arrestin1/2 to CCX-CKR were not affected by PTX (Fig. 6, D-I). Next, the effect of CCL19 treatment on basal adenylyl cyclase activity was measured in the absence and presence of PTX pre-treatment. Only cells expressing CCX/R9IL2 showed a significant increase in CRE activity upon stimulation with CCL19 in the absence of FSK and PTX as compared with vehicle-treated cells (Fig. 5E). Interestingly, PTX pre-treatment increased CCL19-induced CRE activity in cells expressing wild type CCX-CKR, CCX/ R7IL2, CCX/R7IL2IL3, and CCX/R9IL2IL3 (Fig. 5E). The absence of CCL19 responsiveness of CCX/R9IL3 with and without PTX pre-treatment might be related to the reduction in CCL19 binding to this mutant receptor as compared with wild type CCX-CKR (Fig. 5C). The CCR9 chemokine CCL25 showed a comparable CRE-inducing activity profile on CCX-CKR and CCX-CKR chimeras as CCL19 (Fig. 5, F and G). As expected CCL25 activated CCR9 but not CCR7, resulting in an inhibition of forskolin-induced CRE activity (Fig. 5F). Surprisingly, however, PTX pretreatment significantly increased CCR9-mediated CRE activity upon CCL25 stimulation in a forskolin-dependent manner (Fig. 5F).
G i Proteins Impair CCX-CKR Signaling to CRE-The apparent ability of wild type CCX-CKR to stimulate CRE-mediated luciferase activity in response to CCL19 upon PTX pre-treatment, prompted us to focus on wild type CCX-CKR signaling to CRE in more detail. First, concentration-response curves of CCL19, CCL21, and CCL25 were measured in the absence and presence of PTX (100 ng/ml) and/or FSK (1 M). In the absence of PTX pre-treatment, these chemokines did not affect CRE activity in the absence (control) or presence of FSK, when tested at concentrations up to 316 nM (Fig. 7, A-C). Pre-treatment overnight with PTX resulted in an increased CRE activity upon stimulation with Ͼ100 nM CCL19 or Ͼ316 nM CCL25 (Fig. 7,  A-C). Interestingly, pre-treatment with PTX and chemokine stimulation in the presence of FSK showed that PTX and FSK synergistically increased CRE activity upon stimulation with CCL19, CCL21, or CCL25 as compared with vehicle-stimulated cells (Fig. 7, A-C). Moreover, this co-treatment with PTX and FSK significantly increased cAMP levels in HEK293T cells coexpressing CCX-CKR and a BRET-based cAMP biosensor upon stimulation with 100 nM CCL19 as compared with cells that were not pre-treated with PTX and/or stimulated with vehicle (Fig. 7D). Binding of cAMP to the EPAC domain of the biosensor results in a decrease of intramolecular BRET (23,24).
Co-transfection of PTX-insensitive G␣ i1 -C351I, G␣ i2 -C352I, or G␣ i3 -C351I mutants (29) in CCX-CKR-expressing cells abolished the CCL19-induced increase in CRE activity upon PTX pre-treatment (Fig. 7E), which suggests that interaction of G i proteins with CCX-CKR impairs signaling of this receptor to CRE. Co-expression of the PTX-resistant G i proteins did not affect CCX-CKR surface levels or limit expression of the luciferase reporter gene in response to 10 M FSK (data FIGURE 5. IL2 and IL3 of functional chemokine receptors do not convey G␣ i -activating properties to CCX-CKR. A, CCX-CKR, CCR7, and CCR9 amino acid sequences of TM3-IL2-TM4 and TM5-IL3-TM6 are aligned on the basis of highly conserved residues. The IL2 and IL3 sequences that have been exchanged in the CCX-CKR chimeras are indicated with white boxes. HEK293T cells were co-transfected with a CRE-driven reporter gene and CCX-CKR, CCX/R7IL2, CCX/R7IL3, CCX/R7IL2IL3, CCX/R9IL2, CCX/R9IL3, or CCX/R9IL2IL3 as indicated. B, cell surface expression was measured by ELISA. C, 125 I-CCL19 binding was performed on intact cells. Cells were pre-treated with PTX (100 ng/ml) overnight (filled bars). Cells were treated with vehicle, 100 nM CCL19 (D and E), or 100 nM CCL25 (F and G) for 5 h in the presence (D and F) or absemce (E and G) of 1 M FSK. Statistical comparisons were performed using one-way analysis of variance followed by the Bonferroni test. Significant difference in cell surface expression and 125 I-CCL19 binding as compared with wild type CCX-CKR is indicated by §. B and C, significant difference (p Ͻ 0.05) in CRE activity between vehicle and corresponding chemokine-treated cells is indicated by *, whereas # indicates significant difference in chemokine-induced CRE activity between cells pretreated with or without PTX (D-G). Data are shown as averages Ϯ S.E. of fold over vehicle data from 4 independent experiments performed in duplicate (B-E) or 2 independent experiments performed in triplicate (F and G). not shown). Functional expression of the G␣ i1 -C351I, G␣ i2 -C352I, or G␣ i3 -C351I mutants was confirmed by the inability of PTX to inhibit FSK-induced CRE activity by CCR7 upon stimulation with CCL19 (Fig. 7E).
CCX-CKR-mediated CRE Activation Requires an Intact DRY Motif-The conserved DRY motif of GPCRs at the border of transmembrane domain 3 and IL2 is known to be essential for controlling G protein activation, and mutation of Arg 3.50 to alanine in this motif generally results in a receptor that is unable to activate G proteins (30). Wild type CCX-CKR and CCX/ R3.50A were expressed at similar levels (t test p Ͼ 0.05) at the cell surface (Fig. 8A). Similar to wild type CCX-CKR (Fig. 6A) and had comparable affinities for CCL19 with or without PTX pre-treatment (Fig. 8B). In contrast to wild type CCX-CKR, CCX/R3.50A was unable to activate CRE upon CCL19 stimulation in cells pre-treated with PTX (Fig. 8C), which supports the hypothesis that activation of CRE by wild type CCX-CKR is G protein-dependent.
PTX Pre-treatment Did Not Unmask CRE Activation by CXCR7-Several studies indicated that CXCR7 is seen as a decoy receptor due to its incapability to induce G protein signaling (18,31). On the other hand, CXCR7 signals to ERK1/2 and Akt in a ␤-arrestin-biased manner upon stimulation with CXCL11/ITAC or CXCL12/SDF1␣ (18,31). CXCR7-expressing HEK293T cells were pre-treated overnight with PTX to evaluate whether G i proteins might hamper CXCR7-induced G protein signaling via a similar mechanism as observed for CCX-CKR. The CXCR7 chemokines CXCL11 and CXCL12 (100 nM) were not able to increase FSK-induced CRE activity in the absence or presence of PTX in cells transfected with CXCR7 ( Fig. 9). CXCL12 decreased FSK-induced CRE activity in mock and CXCR7-expressing HEK293T cells that were not pretreated with PTX, by interacting with the chemokine receptor CXCR4 (Fig. 9). CXCR4 is endogenously expressed on HEK293(T) cells (32).

DISCUSSION
CCX-CKR plays an important role in controlling the migration of immune and cancer cells that express chemokine receptors CCR7 and CCR9, by reducing the availability of CCL19, CCL21, and CCL25 through internalization (8,10,11,15). CCX-CKR internalization was previously found to be independent of ␤-arrestins and clathrin-coated pits, as dominantnegative ␤-arrestin, Eps15 and Rab5 mutants did not impair CCL19 internalization in CCX-CKR-expressing HEK293 cells (10). In addition, CCL19 was internalized in ␤-arrestin1/2-null mouse embryo fibroblasts expressing CCX-CKR but not in cells expressing CCR7 (10). On the other hand, dominant-negative caveolin isoforms and cholesterol depletion diminished CCL19 sequestering by CCX-CKR, which indicated that CCX-CKR is internalized via caveolae (10). However, inspection of the intracellular C-terminal tail of CCX-CKR reveals the presence of multiple serine and threonine residues that might function as substrate for GPCR kinases and promote interaction with ␤-ar- restins upon phosphorylation (33). In this study, we demonstrate for the first time that stimulation of CCX-CKR with CCL19, CCL21, and CCL25 indeed induced ␤-arrestin2-GFP translocation. CCX-CKR did not translocate ␤-arrestin2-GFP in the absence of chemokines, which is in contrast to the constitutively phosphorylated chemokine decoy receptor D6 (21). Moreover, fluorescently labeled CCL19 co-localized in the same intracellular vesicles as ␤-arrestin2-GFP, suggesting recruitment of ␤-arrestin to CCL19-bound CCX-CKR. Enzyme fragment complementation and BRET, which are two different close proximity-based approaches, then demonstrated actual recruitment of ␤-arrestin to CCX-CKR in response to chemokines. The recruitment of ␤-arrestin to CCX-CKR and co-localization with internalized CCL19 suggests that a possible involvement of ␤-arrestins in CCX-CKR endocytosis cannot be excluded. The occurrence of parallel internalization pathways for one receptor is not unprecedented. For instance, the dopamine D 1 receptor internalizes through both caveolae and clathrin-coated pits (34,35). Interestingly, inhibition of clathrin-coated pit-dependent endocytosis by various methods did . D, HEK293T cells were transiently transfected with the CCX-CKRcoding plasmid along with a plasmid encoding the CAMYEL cAMP biosensor. Cells were pre-treated or not with 100 ng/ml of PTX overnight. Cells were stimulated with 100 nM CCL19 in the presence of 1 M FSK for 10 min prior to measurement of the intramolecular BRET signal. Elevation of cAMP levels is detected as decreased BRET ratio. Data are shown as averages Ϯ S.E. of normalized and baseline corrected BRET ratios from three independent experiments performed in duplicate. E, HEK293T cells were co-transfected with a CRE-driven reporter gene, CCX-CKR, or CCR7 and the PTX-insensitive mutants of G␣ i subunit G␣ i1 /C351I, G␣ i2 /C352I, and G␣ i3 /C351I as indicated. Cells were pre-treated (black bars) or not (empty bars) with 100 ng/ml of PTX overnight. Cells were incubated with vehicle or 100 nM CCL19 in the presence of 1 M FSK for 6 to 8 h. Statistical comparisons to identify significant differences between vehicle and chemokine-stimulated cells (*) as well as between control and PTX-pretreated cells (#) were performed using one-way analysis of variance followed by the Bonferroni test. p value Ͻ 0.05 indicate significant difference. Graphs (A-C and E) show averages Ϯ S.E. of fold over vehicle data from three to six independent experiments performed in duplicate.

FIGURE 8. Potentiation of CCX-CKR-mediated CRE activation is G protein-dependent.
A, cell surface expression of wild type CCX-CKR and the DRY mutant CCX/R3.50A in the reporter gene assay was determined using ELISA with anti-CCX-CKR antibody. Graph shows averages Ϯ S.E. of fold over mock data from two independent experiments performed in duplicate. B, 125 I-CCL19 displacement curves with CCL19 were obtained in the absence (empty symbols) or presence (filled symbols) of 100 ng/ml of PTX in HEK293T cells transfected with CCX/R3.50A. Data are shown as averages Ϯ S.E. of normalized specific binding of one independent experiment performed in duplicate. C, HEK293T cells were co-transfected with a CRE-driven reporter gene and wild type CCX-CKR or CCX/R3.50A as indicated. Cells were pre-treated with 100 ng/ml of PTX (empty and black bars) overnight. Cells were incubated with vehicle (empty bars) or 100 nM CCL19 (filled bars) in the presence of 1 M FSK for 5 h. Statistical comparisons to identify significant differences between vehicle and chemokine-stimulated cells (*) as well as between control and PTX-pretreated cells (#) were performed using one-way analysis of variance followed by the Bonferroni test. p value Ͻ 0.05 indicate significant difference. Graph show averages Ϯ S.E. of fold over vehicle data from two independent experiments performed in duplicate.
not affect D 1 receptor internalization, suggesting that ␤-arrestin and clathrin-coated pits might not be the dominant pathway for this receptor (34).
Previous studies showed that CCX-CKR is unable to induce G protein-dependent intracellular Ca 2ϩ mobilization (7,14). In line with these observations and in contrast to CCR7 and CCR9, CCX-CKR was unable to modulate intracellular cAMP and ino-sitol phosphate levels upon stimulation with 100 nM CCL19, CCL21, and CCL25. Nevertheless, analysis of the CCX-CKR amino acid sequence reveals the conservation of typical GPCR signaling motifs at the bottom of TM helix 3 (i.e. DRY) and TM7 helix 8 (i.e. NPXXY(X) 7 F) (40 -42), as well as a reasonable sequence identity with CCR7 and CCR9 in the TM helices that flank intracellular loops 2 and 3 (Fig. 6A). Interestingly, substitution of CCX-CKR IL2 and/or IL3 with corresponding CCR7 and CCR9 sequences conferred a small but consistent CRE stimulatory activity to CCX-CKR mutants in response to chemokines, rather than the anticipated G i -mediated decrease in CRE signaling. Preventing G i protein interaction with GPCRs by pre-treating the cells with PTX resulted in a potentiation of this signaling by the CCX-CKR mutants. In the absence of forskolin, PTX treatment resulted in a significant chemokine-induced CRE activation by CCX/R7IL2, whereas CCX/R9IL2 activated CRE both in the absence and presence of PTX. These data for the two CCX-CKR chimeras suggest that the IL2 of CCX-CKR decreases the propensity to interact with G proteins.
Moreover, this PTX pre-treatment unmasked signaling by wild type CCX-CKR, resulting in a significant increase in CRE activity in response to CCL19, CCL21, and CCL25. The chemokine concentration-response curves did not allow proper determination of potency and efficacy. However, CCL19 seemed slightly more potent to activate CRE than CCL21 and CCL25, which corresponds to their rank order of CCX-CKR affinities and the ␤-arrestin recruitment potencies in the different assay formats (Table 1). Interestingly, all ligands are ϳ30 -100-fold less potent to stimulate CRE activity in comparison to recruiting ␤-arrestins, which might suggest that ␤-arrestin is not involved in CRE activation in PTX-treated CCX-CKR-expressing cells by interacting with the transcription factor cAMPresponse element-binding protein and the histone acetyltransferase p300 as previously observed for the ␦-opioid receptor (43). Moreover, the absence of CCL19-induced ERK1/2 phosphorylation in CCX-CKR-expressing cells indicates that CRE activation is also not induced via a ␤-arrestinand ERK1/2-dependent pathway, which has been recently described in tracheal epithelial cells (44). . CXCR7 does not mediate CRE activation. HEK293T cells were co-transfected with a CRE-driven reporter gene and empty vector (Ϫ), CXCR7, or CCX-CKR as indicated. Cells were pre-treated (black bars) or not (empty bars) with 100 ng/ml of PTX overnight. Cells were incubated with vehicle, CXCL11, CXCL12, or CCL19 (100 nM chemokine) as indicated in the presence of 1 M FSK. Data are shown as averages Ϯ S.E. of fold over vehicle data from two independent experiments performed in duplicate. Statistical comparisons to identify significant differences between vehicle and chemokine-stimulated cells (*) as well as between control and PTX-pretreated cells (#) were performed using one-way analysis of variance followed by the Bonferroni test. p value Ͻ 0.05 indicate significant difference. FIGURE 10. Proposed model of CCX-CKR interaction with ␤-arrestins and G proteins. Chemokine binding to CCX-CKR (a) recruits G i proteins and ␤-arrestin (␤-arr) with high affinity (b), consequently hindering the low affinity interaction between CCX-CKR and G s proteins. Inactive G i protein may stay bound to CCX-CKR, whereas the chemokine-bound CCX-CKR internalizes with ␤-arrestin (c). Inhibition of G i coupling to CCX-CKR by PTX pre-treatment did not affect ␤-arrestin recruitment but allows G s to interact with CCX-CKR (d), which results in stimulation of adenylyl cyclase (AC) and CRE activity (e).
The recent crystal structure of the ␤2-adrenoreceptor in complex with the G␣ s protein showed that the conserved Arg 3.50 in the DRY motif interacts directly with the G protein (42), which was earlier observed as well in the structure of opsin in complex with a C-terminal peptide of the G␣ protein (45). Ala substitution of Arg 3.50 in CCX-CKR completely abolished CCL19-induced CRE activation in PTX-treated cells, suggesting that the interaction of PTX-insensitive G proteins with CCX-CKR is crucial to signal to CRE. Besides G s proteins, G q proteins have also been shown to stimulate several adenylyl cyclase subtypes via their ␤␥-subunits (46,47). The fact that CCX-CKR-mediated CRE activation is only observed after disabling G i proteins to interact with the receptor suggests that G i proteins might impair the interaction of PTX-insensitive G proteins to couple to CCX-CKR upon chemokine stimulation. The atypical chemokine receptor CXCR7 signals exclusively through ␤-arrestins (18,31). Interestingly, however, BRET assays showed that CXCR7 interacts constitutively with G i proteins without activating them (48). A similar scenario of constitutive or chemokine-induced interaction of inactive G i proteins with CCX-CKR is hypothesized to maintain this decoy receptor "silenced" for typical chemokine receptor-induced G protein signaling by impeding subsequent and apparently less favorable coupling of PTX-insensitive G proteins to this receptor (Fig.  10). On the other hand, pre-treatment of CXCR7-expressing cells with PTX did not unmask CRE activation in response to CXCL11 and CXCL12, suggesting that G i -mediated silencing of CCX-CKR-induced CRE activation is at least specific for this receptor and not the consequence of other cellular changes. Detection of G protein interactions with CCX-CKR using BRET-based methods might unravel the identity and dynamics of G protein subtypes that are recruited to the receptor.
In conclusion, the atypical chemokine receptor CCX-CKR is capable of increasing cAMP levels and CRE activity in response to chemokine stimulation, however, PTX-sensitive G proteins normally prevent this signaling. On the other hand, the chemokine-induced recruitment of ␤-arrestins to CCX-CKR might open avenues to activate yet to be identified G protein-independent signaling pathways.