Phosphorylation-independent association of CXCR2 with the protein phosphatase 2A core enzyme.

Protein phosphatase 2A (PP2A) is postulated to be involved in the dephosphorylation of G protein-coupled receptors. In the present study, we demonstrate that the carboxyl terminus of CXCR2 physically interacts with the PP2A core enzyme, a dimer formed by PP2Ac and PR65, but not with the PP2Ac monomer, suggesting direct interaction of the receptor with PR65. The integrity of a sequence motif in the C terminus of CXCR2, KFRHGL, which is conserved in all CC and CXC chemokine receptors, is required for the receptor binding to the PP2A core enzyme. CXCR2 co-immunoprecipitates with the PP2A core enzyme in HEK293 cells and in human neutrophils. Overexpression of dominant negative dynamin 1 (dynamin 1 K44A) in CXCR2-expressing cells blocks the receptor association with the PP2A core enzyme, and an internalization-deficient mutant form of CXCR2 (I323A,L324A) also exhibits impaired association with the PP2A core enzyme, suggesting that the receptor internalization is required for the receptor binding to PP2A. A phosphorylation-deficient mutant of CXCR2 (331T), which has previously been shown to undergo internalization in HEK293 cells, binds to an almost equal amount of the PP2A core enzyme in comparison with the wild-type CXCR2, suggesting that the interaction of the receptor with PP2A is phosphorylation-independent. The dephosphorylation of CXCR2 is reversed by treatment of the cells with okadaic acid. Moreover, pretreatment of the cells with okadaic acid increases basal phosphorylation of CXCR2 and attenuates CXCR2-mediated calcium mobilization and chemotaxis. Taken together, these data indicate that PP2A is involved in the dephosphorylation of CXCR2. We postulate that this interaction results from direct binding of the regulatory subunit A (PR65) of PP2A to the carboxyl terminus of CXCR2 after receptor sequestration and internalization.

kines are mediated through interaction with their cognate receptors, which are members of the G protein-coupled receptor (GPCR) 1 superfamily. Like other members of the GPCR superfamily, the functional status of many chemokine receptors is determined largely by the phosphorylation state (4 -6). Agonist treatment enhances phosphorylation of the receptors by protein kinases, presumably G protein-coupled receptor kinases and protein kinase C, which results in desensitization of the receptors (4,7,8). This phenomenon is common to many hormonal and neurotransmitter signaling systems (9), but the underlying mechanisms are still only partially understood, especially in the case of chemokine receptors. Based on work on several chemokine receptors, the phosphorylated receptor is then internalized via clathrin-coated pits into early endosomes (6, 10 -14) and subsequently dephosphorylated by intracellular protein phosphatases (10). The dephosphorylated receptors might be either recycled through sorting endosomes back to the plasma membrane or transported to the lysosomes for degradation. The recycling and degradation rate might vary among different chemokine receptors. For example, after down-modulation by interleukin-8 (CXCL8), the expression of CXCR1 fully recovers within 1.5 h, while the recovery rate of CXCR2 expression is very slow and never reaches 40% of the control level during a 3-h culture period (15).
Several investigations have demonstrated that neutrophil chemotactic responses require chemokine receptor internalization and recycling (16). Recycling of chemokine receptors might be very important to maintain the directional migration of cells toward a chemokine concentration gradient. Moreover, dephosphorylation of the receptors appears to play a key role in the recycling and resensitization of chemokine receptors. Therefore, regulation of the receptor dephosphorylation by protein phosphatases represents an important mechanism for modulating the function of chemokine receptors.
Four major classes of serine/threonine-specific protein phosphatases (PPs) have been described. These include PP1, PP2A, PP2B (calcineurin), and PP2C. PP2B and PP2C are calciumdependent, whereas PP1 and PP2A are not. PP1 and PP2A are widely expressed in the cytoplasm of mammalian cells and have been reported to be involved in signal transduction, proliferation, and metabolic events (17). Studies on the ␤ 2 -adrenergic receptor (␤ 2 -AR) suggest that PP2A is involved in the dephosphorylation of the receptor. Pitcher et al. (18) have identified a plasma and vesicular membrane-associated phosphatase that dephosphorylates the ␤ 2 -AR phosphorylated by G protein-coupled receptor kinases. This phosphatase, referred to as G protein-coupled receptor phosphatase, is a subclass of PP2A. The ␤ 2 -AR interacts with a G protein-coupled receptor phosphatase in an acidic pH condition in endosomal vesicles (19). Increasing the endosomal pH not only blocks the association of the receptor with the phosphatase but also prevents the receptor dephosphorylation. In addition, PP2A has been indicated to be the chief enzyme acting on the cholecystokinin receptor from pancreatic acinar cells (20), rhodopsin (21), and C5a receptor in HL60 cells (22).
The holoenzyme of PP2A contains a dimer (core enzyme) composed of a 36-kDa catalytic subunit (PP2Ac) and a 65-kDa regulatory subunit A (PR65). In addition, there are several associated variable regulatory subunits (B) that bind to PR65 and confer substrate specificity to the dephosphorylating activity (17). Although ␤ 2 -AR co-immunoprecipitates with PP2Ac in response to agonist stimulation (19), it has not been determined whether there is a direct binding event and, if so, to which subunit of PP2A the receptor directly binds. For chemokine receptors, the association between receptor and protein phosphatase(s) has not been characterized.
In the process of investigating proteins that interact with chemokine receptors using the yeast two-hybrid system and in vitro binding assay, we discovered that the CXC chemokine receptor, CXCR2, binds to the PR65/PP2Ac dimer but not to the PP2Ac monomer, suggesting direct interaction of the receptor with PR65. The binding requires the integrity of a sequence motif that is conserved in all CC and CXC chemokine receptors and in many other GPCRs. Moreover, CXCR2 associates with the PP2A core enzyme in a phosphorylation-independent and internalization-dependent manner.

EXPERIMENTAL PROCEDURES
Plasmid Construction-Wild-type and truncated mutant (331T) of CXCR2 were constructed previously (4). For the construction of PAS2/ CXCR2 tail for the two hybrid screen, the CXCR2 carboxyl-terminal domain was cut from PRc/CMV-CXCR2 with NcoI and HindIII, blunted with T4-DNA polymerase, and inserted into the PAS2 bait vector that had been digested with NcoI and blunted with T4-DNA polymerase. The correct orientation and in frame fusion of the insert were determined by DNA sequencing. Constructs for glutathione S-transferase (GST) fusion proteins of the C-terminal tails of wild-type or mutant CXCR2 were generated using PCR-amplified fragments. A BamHI site was included in the 5Ј primer, and a HindIII site was included in the 3Ј primer. pGEX-KG and the PCR-generated fragments were cut with BamHI and HindIII, ligated, and used to transform Escherichia coli DH5␣.
Yeast Two-hybrid Assay-Yeast two-hybrid techniques were performed as described (23,24). For screening cDNA libraries, the bait plasmid PAS2/CXCR2 tail was transformed into yeast strain Y190 (CLONTECH) using a lithium acetate protocol (CLONTECH manual). After confirming expression of the bait protein, a human B lymphocyte library in the vector PACT2 was transformed into the strain harboring the bait plasmid. The transformants expressing both the bait and the prey proteins were selected on medium lacking leucine, tryptophan, and histidine (SD/ϪLeu/ϪTrp/ϪHis). Colonies capable of growing on the SD/ϪLeu/ϪTrp/ϪHis medium were then tested for ␤-galactosidase activity (LacZ ϩ ) using the filter lift assay. Clones that were consistently phenotypically His ϩ and LacZ ϩ were further characterized. Approximately 2.6 ϫ 10 6 transformants were screened, and several of them were His ϩ and LacZ ϩ . A single clone was chosen for further pursuit based on its strong His ϩ /LacZ ϩ phenotype.
Filter Lift Assay-A dry nitrocellulose filter was placed on the yeast colonies. The filter was then carefully lifted, transferred (colonies facing up) into liquid nitrogen, completely submerged for 10 s, and then allowed to thaw at room temperature. The filter was then placed on a Whatman filter paper soaked in Z buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 1 mM Mg 2 SO 4 , pH 7.0) containing 50 mg/ml 5-bromo-4-chloro-3-indolyl-galactopyranoside, with colonies face up until blue color appeared.
In Vitro Binding Assay-Bacteria encoding GST or GST fusion proteins were cultured overnight at 37°C, and then isopropyl-1-thio-␤-Dgalactopyranoside was added, and incubation was continued for another 3 h to induce protein expression. The bacteria were lysed in RIPA buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 10 mM NaF, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, and 10 g each of leupeptin and aprotinin) and then sonicated on ice for 10 s. The supernatant of the bacterial lysate was incubated with glutathione-Sepharose at 4°C for 30 min. After washing three times with RIPA buffer, the beads were resuspended in RIPA buffer. Purified PP2A subunits (a generous gift from Dr. Brian Wadzinski) or HEK293 cell lysates were incubated with the GST or GST fusion proteins bound to glutathione-Sepharose for 2 h at 4°C with rotation. Beads were pelleted by centrifugation (12,000 rpm) for 2 min and washed four times with RIPA buffer. Bound proteins were released by boiling in SDS-PAGE sample buffer for 5 min and detected by SDS-PAGE and Western blot.
Co-immunoprecipitation and Western Blot-Human neutrophils were isolated from fresh heparinized peripheral blood from single human donors as described previously (25). HEK293 cells stably expressing CXCR2 were serum-starved overnight in DMEM containing 0.5% fetal bovine serum before the experiment. The cells were treated with or without agonists, and then the cells were washed three times with ice-cold PBS and lysed in 1 ml of RIPA buffer. The cell debris was removed by centrifugation for 4 min at 13,000 rpm in an Eppendorf microcentrifuge. The supernatant was precleared for 1 h to reduce nonspecific binding by the addition of 40 l of protein A/G-agarose (Pierce). After removal of the protein A/G-agarose by centrifugation in an Eppendorf microcentrifuge at 3000 rpm for 1 min, the cleared supernatant was collected, and 10 l of affinity-purified anti-CXCR2 antibody (prepared in our laboratory) was added for overnight precipitation at 4°C. 40 l of protein A/G was then added and incubated at 4°C for 2 h. The protein A/G-antibody-antigen complex was then collected by washing three times with ice-cold RIPA buffer. The final pellet was resuspended in 50 l of SDS sample buffer containing 5% ␤-mercaptoethanol and heated to 50°C for 10 min. 20 l of this preparation was electrophoresed on a 10% SDS-polyacrylamide gel, and the proteins on the gel were transferred to nitrocellulose membranes (Bio-Rad) as previously described (10). Co-precipitated PR65 or PP2Ac was detected using a goat polyclonal antibody (catalog no. SC6112; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and a mouse monoclonal antibody (catalog no. P47720; Transduction Laboratories), respectively.
In Vitro Phosphorylation and Dephosphorylation-CXCR2 was immunoprecipitated with a rabbit anti-CXCR2 antibody and protein A/G beads as described above. The immunoprecipitates were then washed four times with kinase assay buffer containing 50 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 500 M CaCl 2 . The immunoprecipitates were incubated with 100 g/ml phosphatidylserine, 2.5 Ci of [␥-32 P]ATP, and 50 microunits of purified protein kinase C (Pierce) at 30°C for 30 min in kinase assay buffer in a final volume of 20 l. 1 unit of protein kinase C activity is defined as the amount of enzyme required to transfer 1 mol of phosphate from ATP to histone H1 per min at 30°C. The immunoprecipitates were then washed with phosphatase assay buffer containing 5 mM Tris-HCl, pH 7.0, 0.1 mM EDTA, 0.1% 2-mercaptoethanol, and 1 mg/ml bovine serum albumin. Dephosphorylation was carried out by incubation of the immunoprecipitates with different concentrations of purified PP2A core enzyme in the phosphatase assay buffer in a volume of 30 l at 30°C for 30 min. The reaction was terminated by adding Laemmli sample buffer and heating at 50°C for 10 min. Samples were subjected to 10% SDS-PAGE, and phosphorylated CXCR2 was detected by autoradiography.
In Vivo Phosphorylation and Dephosphorylation-Receptor phosphorylation assay was performed as described previously (10). In brief, the transfected cells were replated on six-well plates 1 day after the transfection. On the following day, after incubating in serum-and phosphatefree medium for 1 h, cells were labeled by incubating in [ 32 P]orthophosphate (100 Ci/ml) (PerkinElmer Life Sciences) in the same medium at 37°C for 2 h. Cells were then stimulated with or without agonists. Dephosphorylation was performed by allowing cells to recover in fresh serum-free media at 37°C for 1 h. The cells were then lysed in RIPA buffer. CXCR2 was immunoprecipitated as described above with a specific antibody. The immunoprecipitates were electrophoresed through a 10% SDS-polyacrylamide gel and transferred to a nitrocellulose membrane (Bio-Rad). The phosphorylated receptors were then detected by autoradiography. The amount of receptor immunoprecipitated was determined by blotting the membrane with a monoclonal antibody against CXCR2 to evaluate relative receptor phosphorylation.
Chemotaxis Assay-A 96-well chemotaxis chamber (Neuroprobe Inc.) was used for chemotaxis assays, and the lower compartment of the chamber was loaded with 400-l aliquots of 1 mg/ml ovalbumin/DMEM (chemotaxis buffer) or CXCL8 diluted in the chemotaxis buffer (1-200 ng/ml). Polycarbonate membranes (10-m pore size) were coated on both sides with 20 g/l human collagen type IV, incubated for 2 h at 37°C, and then stored at 4°C overnight. To prepare cells for chemotaxis assay, they were removed by trypsinization, washed with Hanks' solution, and incubated in 10% fetal bovine serum/DMEM for 2 h at 37°C to allow time for restoration of receptors. The cells were washed with chemotaxis buffer and then loaded into the upper chamber in the chemotaxis buffer. The chamber was incubated for 4 h at 37°C in humidified air with 5% CO 2 , and then the membrane was removed, washed, fixed, and stained with a Diff-Quik kit. Cell chemotaxis was quantified by counting the number of migrating cells present in 10 microscope fields (ϫ 20 objective).
Calcium Fluorimetry-HEK293 cells stably expressing CXCR2 were grown until confluent. Cells were released by shaking, collected by centrifugation at 300 ϫ g for 6 min, and washed with Hanks' buffer containing 5 mM HEPES. Cells were resuspended at 2 ϫ 10 6 cells/ml and incubated with 2.5 M Fluo-3 (Molecular Probes, Inc., Eugene, OR) for 30 min at 37°C. After incubation, the cells were washed once with Hanks' buffer containing 5 mM HEPES and 2 mM Ca 2ϩ . The cells were finally adjusted to 2 ϫ 10 6 cells/ml. Ca 2ϩ mobilization experiments were performed as described previously (26).

RESULTS
In an attempt to isolate chemokine receptor-associated proteins, we used the yeast two-hybrid system to identify proteins that interact with the carboxyl terminus of the chemokine receptor, CXCR2 (Fig. 1A). Screening of a human B lymphocyte library fused to the GAL4 transactivation domain (generous gift from Dr. Stephen J. Elledge) yielded several potential candidate genes that were both His ϩ and LacZ ϩ . The prey cDNAs were recovered from yeast and transformed into bacteria. The cDNAs were then sequenced using primers complementary to 5Ј or 3Ј ends of the inserts. Among them, one (clone 91) encoding PR65 was chosen for further study based on its moderately strong LacZ ϩ . The specificity of the interaction in yeast was tested by retransforming PACT2/clone 91 along with the original bait PAS2/CXCR2 tail or PAS2 alone back into yeast strain Y190. The interaction between the receptor C terminus and PR65 specifically allowed growth on SD medium lacking leucine, tryptophan, and histidine (SD/ϪLeu/ϪTrp/ ϪHis) ( Fig. 2A, left panel). Neither the bait protein, PAS2/ CXCR2 tail, nor the prey was able to activate transcription of the reporter genes in the presence of only empty prey or bait vectors, respectively ( Fig. 2A, left panel). Using the ␤-galactosidase assay, we found that only the yeast co-transformed with PACT2/clone 91 and the PAS2/CXCR2 tail displayed LacZ ϩ ( Fig. 2A, right panel).
To confirm the specific biochemical interaction between the C terminus of CXCR2 and PR65, we used an in vitro binding assay to test for direct interaction. A GST fusion protein containing the C terminus of CXCR2 was made (Fig. 1A) and tested for binding to the purified PP2Ac monomer or PP2Ac/ PR65 dimer. We observed that the GST-CXCR2 tail fusion protein specifically bound the purified PP2Ac/PR65 dimer but not the PP2Ac monomer (Fig. 2B), suggesting that the C terminus of CXCR2 only binds to PR65. Several GST fusion proteins show a consistent pattern of lower molecular weight bands in the purified sample. These are believed to be either unstable degradation products or incompletely translated products. We next identified the region of CXCR2 involved in the PP2A binding by using GST fusion proteins encoding various fragments of the C terminus of CXCR2 (Fig. 1A). We found that the GST-CXCR2-(311-330) bound to an equal amount of the PP2A core enzyme as compared with GST-CXCR2 tail, whereas the GST-CXCR2-(331-355) did not bind to the PP2A core enzyme, implying that the minimal CXCR2 binding region resides in residues 311-330 of the C-terminal domain of CXCR2 (Fig. 3A). Interestingly, this domain is immediately upstream of the phosphorylation sites, suggesting that PP2A does not bind to the C-terminal phosphorylation sites of the receptor.
To identify which residues are required for the binding of the PP2A core enzyme, in vitro binding assays using GST fusion proteins with group or single mutations in the C terminus of CXCR2 were performed (Fig. 1A). We found that the GST-CXCR2 carboxyl-terminal mutants GQK313-315A, FRH316 -318A, and GLL319 -321A bound less PP2A core enzyme than the wild-type GST-CXCR2 carboxyl terminus fusion protein (Fig. 3B). A GST pull-down assay using single mutants revealed that residues Lys 315 , Arg 317 , His 318 , and Leu 320 were important for the receptor binding to the PP2A core enzyme (Fig. 3C). These data indicate that the integrity of the sequence motif, KFRHGL, is required for the interaction of CXCR2 with PP2A. This motif is conserved in all CC and CXC chemokine receptors (Fig. 1B) and in some other GPCRs (not shown).
We next examined whether a functional complex consisting of CXCR2 and the PR65/PP2Ac dimer could be detected in HEK293 cells overexpressing the receptors using an immunoprecipitation assay. Immunoprecipitation of CXCR2 from HEK293 cells revealed a weak basal association of the receptors with the PP2A core enzyme, and CXCL1 (100 ng/ml) treatment time-dependently increased the association between the immunoreactive PP2A core enzyme and the receptor, which peaked at 10 min (Fig. 4, A and B). We also tested the interaction of CXCR2 with PP2A in human neutrophils. Treatment of the cells with CXCL8 (100 ng/ml) for 10 min significantly increased the association of CXCR2 with the PP2A core enzyme (Fig. 4C).
Our previous study has shown that inhibition of CXCR2 internalization blocks the receptor dephosphorylation (10), suggesting that receptor internalization is required for the association of the receptors with PP2A. The internalization of CXCR2 and other chemokine receptors can be blocked by mutation of the carboxyl-terminal dileucine motifs (26) or cotransfection of dominant negative dynamin 1 (dynamin 1 K44A) (10,27). As shown in Fig. 5, A and B, mutation of Ile 323 -Leu 324 greatly impaired the receptor binding to the PP2A core enzyme. Overexpression of dynamin 1 K44A in HEK293 cells stably transfected with CXCR2 also significantly decreased the interaction of the receptors with the PP2A core enzyme (Fig. 5, C and D). These data indicate that the receptor internalization is required for the agonist-dependent association of the receptors with PP2A.
To test whether agonist-induced phosphorylation is required for the association of the receptors with the PP2A core enzyme, the carboxyl-terminal truncated mutant of CXCR2 (331T) was used to co-immunoprecipitate with the PP2A core enzyme. Because 331T, which no longer undergoes agonist-induced phosphorylation, still transduces downstream signaling and undergoes agonist-induced internalization in HEK293 cells (26), it can be used as a model to investigate the potential role of phosphorylation in the association of the receptor with PP2A. Interestingly, compared with wild-type CXCR2, 331T coimmunoprecipitated an almost equal amount of the PP2A core enzyme in response to agonist treatment (Fig. 6, A and B). These data support the hypothesis that agonist-induced phosphorylation of the receptor is not required for the receptor binding to PP2A.
To assess the potential involvement of PP2A in the dephosphorylation of CXCR2, we used okadaic acid (OA), a potent cell-permeant inhibitor of PP1 and PP2A (28,29). As shown in Fig. 7A, exposure of HEK293 cells expressing CXCR2 to CXCL8 resulted in a robust phosphorylation of the receptors (Fig. 7A,  lane 1). The phosphorylation was reversed after withdrawal of the agonist followed by continued incubation for 1 h at 37°C (Fig. 7A, lane 2). Treatment of the cells with concentrations of OA ranging from 0.1 nM to 1 M inhibited the dephosphorylation of the receptors in a concentration-dependent fashion (Fig. 7A, lanes 3-7). A maximal effect was obtained for concentrations of OA equal to or higher than 100 nM. Quantification of the phosphorylation of CXCR2 indicated that the IC 50 for OA to inhibit the dephosphorylation of CXCR2 was about 10 nM (Fig.  7B). In addition, OA was also found to increase the basal level of phosphorylation in the absence of CXCL8 stimulation (Fig.  7C, compare lane 2 with lane 1). Thus, the results indicate that a protein phosphatase(s) sensitive to OA, presumably PP2A and/or PP1, is involved in regulating the phosphorylation state of CXCR2. Although the inhibition of PP2A has been reported to occur in vitro at subnanomolar concentrations of OA, whereas that of PP1 requires 100-fold higher concentration (30), the pharmacological sensitivity of the inhibitor is likely to be quite different in intact cells. The relatively high concentra-FIG. 2. Interaction of CXCR2 carboxyl terminus with PP2A. A, yeast strain Y190 containing the PAS2/CXCR2 tail, the PACT2/PR65, or both plasmids were replica-plated on SD medium without tryptophan, leucine, and histidine. The cotransformation of p53 and SV40 provided by the manufacturer was performed as a positive control (left). A ␤-galactosidase assay was carried out using a YPD plate replicate (right) as described under "Experimental Procedures." B, the indicated GST or GST-CXCR2 carboxyl tail fusion protein was incubated with the purified PP2Ac monomer (C monomer) or PR65/PP2Ac dimer (AC dimer) and then absorbed onto glutathione-Sepharose beads. After washing, the beads were resuspended in loading buffer. Proteins were separated using 10% SDS-PAGE and transferred to a nitrocellulose membrane. Coprecipitated PR65 and PP2Ac were analyzed by immunoblotting with anti-PR65 and anti-PP2Ac. The membrane was stripped and reblotted with a mouse monoclonal anti-GST antibody to confirm the protein expression and equal loading. tion of OA (equal to or higher than 100 nM) that is required for a maximal effect might reflect the abundance of PP2A present in our experimental conditions, i.e. at high cell density, rather than a preference for PP1, since treatment of the cells with the predominant PP1 inhibitor, tautomycin (500 nM), did not affect the receptor dephosphorylation (data not shown).
To confirm that CXCR2 is dephosphorylated by PP2A, an in vitro phosphorylation and dephosphorylation experiment was performed. HEK293 cells stably expressing CXCR2 were immunoprecipitated with an anti-CXCR2 antibody. The immunoprecipitates were phosphorylated by purified protein kinase C, which has been reported to be capable of phosphorylating CXCR2 (4). The phosphorylated receptors were then incubated with purified PP2A core enzyme. As expected, PP2A dephosphorylated the receptors in a concentration-dependent manner (Fig. 7D).

FIG. 3. Identification of the PP2A binding domain in the carboxyl terminus of CXCR2.
A, the indicated GST-CXCR2 carboxyl tail, GST-CXCR2-(311-330), and GST-CXCR2-(331-355) fusion proteins were incubated with HEK293 cell lysate. GST pull-down assays were performed as described in the legend to Fig. 2B. B, a series of the indicated GST fusion proteins of the wild-type or group mutants of the carboxyl-terminal domain of CXCR2 (shown schematically in Fig.  1A) were incubated with HEK293 cell lysate. The pull-down assays were performed as described in the legend to Fig.  2B. C, a series of the indicated GST fusion proteins of the wild-type or single mutant carboxyl-terminal domain of CXCR2 shown schematically (Fig. 1A) were incubated with HEK293 cell lysate. The pulldown assay was performed as described in the legend to Fig. 2B. Data shown are representative of three independent experiments.

FIG. 4. CXCR2 associates with the PP2A core enzyme in HEK293 cells and in neutrophils.
A, HEK293 cells stably expressing CXCR2 were exposed to CXCL1 (100 ng/ml) for the indicated times, and the cell lysates were immunoprecipitated (IP) with a rabbit polyclonal anti-CXCR2 antibody. A preimmune rabbit serum (mock) was used in a parallel experiment to confirm the specificity of the immunoprecipitation. Proteins were separated using 10% SDS-PAGE and transferred to a nitrocellulose membrane. Co-precipitated PP2Ac and PR65 were blotted using specific anti-PP2Ac and anti-PR65. The membrane was stripped and reblotted with a mouse monoclonal anti-CXCR2 antibody to confirm equal loading. The results represent one of three independent experiments. B, quantification of the density of bands (mean Ϯ S.E.) representing PR65 (black bars) and PP2Ac (striped bars) was determined by densitometric scanning. C, human neutrophils prepared from single donors were treated without (lane 1) or with (lane 2) CXCL8 (100 ng/ml) for 10 min. The cells were lysed, and immunoprecipitation was performed as described above. Co-precipitated PP2Ac and PR65 were blotted using specific anti-PP2Ac and anti-PR65. The membrane was stripped and reblotted with a mouse monoclonal anti-CXCR2 antibody (E2; Santa Cruz Biotechnology) to confirm equal loading.
To assess the functional role of PP2A in the signaling of CXCR2, OA was used to pretreat HEK293 cells stably expressing CXCR2, and CXCL8-induced calcium mobilization and chemotaxis were observed. Pretreatment of the cells with OA (100 nM) for 1 h significantly attenuated CXCR2-mediated calcium mibilization (Fig. 8A) and chemotaxis (Fig. 8B).

DISCUSSION
One of the most important functions of chemokine receptors is to mediate chemotaxis of neutrophils and lymphocytes. The receptor desensitization and resensitization processes are postulated to provide an on-off mechanism for the receptor-medi-ated chemotaxis (10). Compared with the well established mechanisms for the receptor desensitization (5,(31)(32)(33)(34), the mechanisms underlying the receptor resensitization are poorly understood. Dephosphorylation appears to play a key role in the recycling and the subsequent resensitization of the receptors (10). However, little is known about the mechanisms underlying the receptor dephosphorylation. We present evidence in this study that PP2A is involved in the dephosphorylation of the chemokine receptor, CXCR2, by physically interacting with the receptors. It is surprising that CXCR2 only binds to the purified PP2A core enzyme, a dimer composed of the regulatory subunit A (PR65) and the catalytic subunit (PP2Ac), but not the PP2Ac monomer, suggesting that the receptor interacts exclusively with PR65. More interestingly, a binding domain in the C terminus of CXCR2, which is localized upstream of the potential phosphorylation sites and is conserved in all CC and CXC chemokine receptors as well as many other kinds of GPCRs, was identified. Several charged residues in this domain are required for the receptor binding to PP2A. We were unable to test the significance of this binding domain in an in vivo study, since mutation of any residue in this domain seriously impaired the receptor localization on the cell membrane (data not shown). In future experiments, it will be of interest to investigate the binding domain in the sequence of PR65 that is involved in the interaction between PP2A and CXCR2 or other types of GPCRs.
An in vivo interaction of CXCR2 with PP2A was demonstrated in this study. Agonist treatment induced an association of CXCR2 with the PP2A core enzyme in a time-dependent manner. The association peaked at about 10 min and lasted for at least 30 min, whereas, within the same time frame, most portions of CXCR2 and other GPCRs become sequestered into endosomes (26,27,31,35,36). We previously demonstrated that inhibition of CXCR2 internalization impairs the receptor dephosphorylation (10). In addition, studies of ␤ 2 -AR indicate that the receptor is dephosphorylated in endosomal vesicles. It has been shown that ␤ 2 -ARs in vesicular fractions are in a less phosphorylated state than are receptors in the plasma membrane (37). Suppressing the receptor internalization by internalization-blocking reagents or mutation of the internalization  1 and 2) or I323A,L324A (IL/AA) mutant (lanes 3 and 4) CXCR2 were treated without (lanes 1 and 3) or with (lanes 2 and 4) CXCL1 (100 ng/ml) for 10 min, and the cell lysates were immunoprecipitated (IP) with a rabbit polyclonal anti-CXCR2 antibody. Proteins were separated using 10% SDS-PAGE and transferred to a nitrocellulose membrane. Co-precipitated PP2Ac and PR65 were blotted using specific anti-PP2Ac and anti-PR65. The membrane was stripped and reblotted with a mouse monoclonal anti-CXCR2 antibody to confirm equal loading. Data shown represent one of three independent experiments. B, quantification of the density of bands (mean Ϯ S.E.) representing PR65 was determined by densitometric scanning without CXCL1 treatment (black bars) or with CXCL1 treatment (striped bars). The data were analyzed using Student's t test. *, p Ͻ 0.05, compared with the control cells transfected with wild-type CXCR2. C, cells stably expressing CXCR2 were transiently transfected with vector (lanes 1 and 2) or dynamin 1 K44A (lanes 3 and 4) and then treated without (lanes 1 and 3) or with (lanes 2 and 4) CXCL1 (100 ng/ml) for 10 min. The cell lysates were immunoprecipitated with a rabbit polyclonal anti-CXCR2 antibody. Proteins were separated using 10% SDS-PAGE and transferred to a nitrocellulose membrane. Co-precipitated PP2Ac and PR65 were blotted using specific anti-PP2Ac and anti-PR65. The membrane was stripped and reblotted with a mouse monoclonal anti-CXCR2 antibody to confirm equal loading. Expression of dynamin 1 K44A was confirmed by Western blot using whole cell lysate with a monoclonal anti-dynamin 1 antibody. Data shown represent one of three independent experiments. D, quantification of the density of bands (mean Ϯ S.E.) representing PR65 was determined by densitometric scanning without CXCL1 treatment (black bars) or with CXCL1 treatment (striped bars). *, p Ͻ 0.05, for CXCL1-treated as compared with the control cells transfected with vector.  1 and 2) or the truncated mutant (331T) (lanes 3 and 4) of CXCR2 were treated without (lanes 1 and 3) or with (lanes 2 and 4) CXCL1 (100 ng/ml) for 10 min, and the cell lysates were immunoprecipitated (IP) with a rabbit polyclonal anti-CXCR2 antibody. Proteins were separated using 10% SDS-PAGE and transferred to a nitrocellulose membrane. Co-precipitated PP2Ac and PR65 were blotted using anti-PP2Ac and anti-PR65. The membrane was stripped and reblotted with a mouse monoclonal anti-CXCR2 antibody to confirm equal loading. Data shown represent one of three independent experiments. B, quantification of the density of bands (mean Ϯ S.E.) representing PR65 was determined by densitometric scanning without CXCL1 treatment (black bars) or with CXCL1 treatment (striped bars). No significant difference was found in the amount of PR65 associated with the wild-type and the mutant (331T) form of CXCR2 (p Ͼ 0.05). motifs impairs the receptor dephosphorylation and the subsequent resensitization (38,39). Moreover, an increase in endosomal pH suppresses the receptor dephosphorylation and coimmunoprecipitation with PP2Ac (19). Based on these data, we sought to investigate the potential requirement of the internalization of CXCR2 for its association with PP2A.
CXCR2 internalization is regulated differentially in a variety of cell types. The carboxyl-terminal domain of CXCR2, which includes the phosphorylation domain appears to be required for the receptor internalization in 3ASubE cells but is not required for internalization in HEK293 cells (4,26). Although the underlying mechanisms are not yet fully understood, it is postulated that different adaptor proteins present in these cell lines might be responsible for the difference in CXCR2 internalization (26). Previous studies have demonstrated that CXCR2 binds to AP-2 and ␤-arrestins, two adaptor proteins involved in the internalization of CXCR2, through its C-terminal dileucine motifs and the downstream phosphorylation sites, respectively (26). It has been suggested that in HEK293 cells, the internalization of the truncated mutant CXCR2 (331T), which loses all of the C-terminal phosphorylation sites and fails to bind to ␤-arrestins, is mediated, at least in part, by interaction with AP-2 (26). We postulate that the level of AP-2 in 3ASubE cells is not sufficient to mediate the internalization of 331T in the absence of ␤-arrestin association with the carboxyl-terminal domain of the receptor. To investigate the potential role of CXCR2 internalization in the interaction of the receptor with PP2A, the receptor internalization was blocked by overexpression of a dominant negative mutant (K44A) of dynamin 1, a major component of clathrin-coated pits (10), or by mutation of the dileucine motifs in the receptor C terminus (26). The results demonstrate that blocking the receptor internalization impaired the receptor binding to the PP2A core enzyme. These data strongly support the hypothesis that the internalized CXCR2 associates with PP2A in the endosomes.
We purpose that PP2A binding to the receptors is to reverse the phosphorylated state of the receptors and that agonistinduced phosphorylation of the receptors facilitates the receptor association with PP2A. CXCR2 provides an ideal model to investigate the potential role of phosphorylation in the interaction between GPCRs and PP2A, because truncation of the C-terminal domain containing all of the potential phosphorylation sites totally abolishes agonist-induced phosphorylation of the receptor yet does not affect the receptor/G protein coupling and internalization in HEK293 cells (4,26,31). Interestingly, the present data clearly demonstrate that the agonist-induced phosphorylation of CXCR2 is not required for the receptor binding to the PP2A core enzyme, since the C-terminal truncated mutant (331T) of CXCR2 associated with an almost equal amount of PP2A, as compared with the wild-type CXCR2. These data, taken together with the evidence from the GST pull-down assay, indicating that truncation of the C-terminal domain containing the serine/threonine residues did not affect the receptor C terminus binding to the PP2A core enzyme,  7)). The cell lysates were immunoprecipitated with a rabbit polyclonal anti-CXCR2. Proteins were separated using 10% SDS-PAGE and transferred to a nitrocellulose membrane. Phosphorylated proteins were detected by autoradiography. The membrane was blotted (IB) with a mouse monoclonal anti-CXCR2 antibody (E2; Santa Cruz Biotechnology) to confirm equal loading. Data shown represent one of three independent experiments. B, quantification of the density of bands (mean Ϯ S.E.) representing phosphorylated CXCR2 was determined by densitometric scanning. C, HEK293 cells stably expressing CXCR2 were pretreated without (lane 1) or with (lane 2) OA (100 nM) for 1 h at 37°C before being metabolically labeled with [ 32 P]orthophosphate for 1 h. The cell lysates were immunoprecipitated with a rabbit polyclonal anti-CXCR2. Proteins were separated using 10% SDS-PAGE and transferred to a nitrocellulose membrane. Phosphorylated proteins were detected by autoradiography. The membrane was blotted with a mouse monoclonal anti-CXCR2 antibody (E2; Santa Cruz Biotechnology) to confirm equal loading. D, cell lysates of HEK293 cells stably expressing CXCR2 were immunoprecipitated with rabbit polyclonal anti-CXCR2. The immunoprecipitates were then phosphorylated by purified protein kinase C, followed by incubation of the immunoprecipitates with the indicated concentrations of purified PP2A core enzyme as described under "Experimental Procedures." Proteins were separated using 10% SDS-PAGE and transferred to a nitrocellulose membrane. Phosphorylated proteins were detected by autoradiography. The membrane was blotted with a mouse monoclonal anti-CXCR2 antibody (E2; Santa Cruz Biotechnology) to confirm equal loading. Data shown represent one of three independent experiments. strongly suggest a modest effect of the receptor phosphorylation in the direct interaction with PP2A. These data support the hypothesis that it is the agonist-induced conformational change and not the phosphorylation of the receptors that facilitates the association of the receptors with PP2A.
The involvement of PP2A in the dephosphorylation of CXCR2 is demonstrated by in vivo and in vitro experiments. OA has been demonstrated to enter cells and inhibit PP2A at low concentrations without affecting PP1, another phosphatase that is only inhibited by higher concentrations of OA (28,29). In the present study, OA inhibited the dephosphorylation of CXCR2 in a concentration-dependent manner with an IC 50 value of about 10 nM. Penetration of OA through the cell membrane is time-and concentration-dependent and is affected by temperature and pH (40). Higher concentrations of OA are needed in some cell types to overcome the high concentration of PP2A in intact cells (41). This may account for the 10 times higher value of the IC 50 of OA to inhibit the intracellular PP2A than the IC 50 of OA to inhibit purified PP2A activity (42). A maximal inhibitory effect on CXCR2 dephosphorylation was obtained for concentrations of OA equal to or higher than 100 nM, and this concentration of OA has been reported to specifically inhibit intracellular PP2A without affecting PP1 activity in intact cells (43,44). Moreover, the phosphorylated state of CXCR2 was reversed by purified PP2A core enzyme in a concentration-dependent manner. These data strongly suggest the critical role of PP2A in regulating the phosphorylation of CXCR2.
The functional role of phosphorylation/dephosphorylation in the signaling of CXCR2 and other GPCRs is still not fully understood. Phosphorylation of the receptors in response to agonist stimulation appears to be a key step in termination of the receptor/G protein coupling, which results in desensitization of the receptors (4,7,8). Receptor phosphorylation may also play a role in facilitating certain chemokine receptor internalization (4). It is increasingly clear that dephosphorylation is an important step in reestablishing a normal responsiveness after agonist removal. Studies of the C5a receptor have pointed out that internalized C5a receptors are recycled to the plasma membrane with a time course consistent with the kinetics of dephosphorylation, suggesting that dephosphorylation of the C5a receptor might be essential to receptor recycling and resensitization during chemotaxis (22). In view of the basal association of CXCR2 with the PP2A core enzyme in the absence of agonist stimulation and of the higher incorporation of phosphate in the unoccupied CXCR2 after OA treatment, it appears that the basal level of serine/theonine dephosphorylation has to be relatively active to maintain a low state of phosphorylation of unoccupied CXCR2, which is necessary for the normal responsiveness of the receptors. Thus, a fraction, if not all, of the unoccupied CXCR2 appears to undergo a constitutive phosphorylation-dephosphorylation cycle, presumably controlled by protein kinases such as G protein-coupled receptor kinases and/or protein kinase C and by protein phosphatases such as PP2A. Disrupting the phosphorylation-dephosphorylation cycle would affect the receptor signaling. Studies on 3ASubE cells and on RBL-2H3 cells have demonstrated that truncation of the cytoplasmic tail of CXCR2 (331T) prolonged its signaling relative to wild-type CXCR2, increased its resistance to internalization, and induced phospholipase D activation (4,33). Although 331T undergoes internalization in HEK293 cells, the calcium mobilization is significantly prolonged in response to activation of the mutant receptor (26). On the other hand, pretreatment of the cells with OA results in an increase in the basal level of CXCR2 phosphorylation and significantly attenuates the CXCR2-mediated signaling such as calcium mobilization and chemotaxis. However, we cannot rule out the possibility that other intracellular molecules required for calcium mobilization and chemotaxis might also be affected by OA treatment.
In conclusion, the present study demonstrates for the first time that the G protein-coupled chemokine receptor, CXCR2, interacts with the PP2A core enzyme, probably through its C-terminal association with the regulatory subunit A (PR65) of PP2A. A conserved sequence motif in the carboxyl-terminal domain, upstream of the serine/threonine residues, is the potential PP2A binding domain. Agonist-induced increase in the interaction of the receptors with the PP2A core enzyme is phosphorylation-independent but internalization-dependent. PP2A is involved in the dephosphorylation of CXCR2 and plays an important role in regulating the receptor signaling.