Regulation of Human Chemokine Receptors CXCR4

Members of the chemokine receptor family CCR5 and CXCR4 have recently been shown to be involved in the entry of human immunodeficiency virus (HIV) into target cells. Here, we investigated the regulation of CXCR4 in rat basophilic leukemia cells (RBL-2H3) stably transfected with wild type (Wt CXCR4) or a cytoplasmic tail deletion mutant (ΔCyto CXCR4) of CXCR4. The ligand, stromal cell derived factor-1 (SDF-1) stimulated higher G-protein activation, inositol phosphate generation, and a more sustained calcium elevation in cells expressing ΔCyto CXCR4 relative to Wt CXCR4. SDF-1 and phorbol 12-myristate 13-acetate (PMA), but not a membrane permeable cAMP analog induced rapid phosphorylation as well as desensitization of Wt CXCR4. Phosphorylation of ΔCyto CXCR4 was not detected under any of these conditions. Despite lack of receptor phosphorylation, calcium mobilization by SDF-1 in ΔCyto CXCR4 cells was partially desensitized by prior treatment with SDF-1. Of interest, the rapid release of calcium was inhibited without affecting the sustained calcium elevation, indicating independent regulatory pathways for these processes. PMA completely inhibited phosphoinositide hydrolysis and calcium mobilization in Wt CXCR4 but only partially inhibited these responses in ΔCyto CXCR4. cAMP also partially inhibited these responses in both Wt CXCR4 and ΔCyto CXCR4. SDF-1, PMA, and cAMP caused phosphorylation of phospholipase Cβ3 in Wt and ΔCyto CXCR4 cells. Both SDF-1 as well as PMA induced rapid internalization of Wt CXCR4. SDF-1 but not PMA induced internalization of ΔCyto CXCR4 albeit at reduced levels relative to Wt CXCR4. These results indicate that signaling and internalization of CXCR4 are regulated by receptor phosphorylation dependent and independent mechanisms. Desensitization of CXCR4 signaling, independent of receptor phosphorylation, appears to be a consequence of the phosphorylation of phospholipase Cβ3.

quence (YPYDVPDYA) was inserted between the NH 2 -terminal initiator methionine and the second amino acid of human CXCR4 by polymerase chain reaction methods as described previously for other chemoattractant receptors (27,28). The same epitope tag was also placed at the COOH-terminal end before the stop codon in some constructs. A COOH terminally truncated CXCR4 at the amino acid 318 by altering the codon 319 into a stop codon was also made using standard polymerase chain reaction methods. The integrity of the epitope tag as well as the rest of the molecule was confirmed by dideoxy sequencing after cloning into eukaryotic expression vectors pcDNA3.1 and pRK-5.
Radioligand Binding Assays-For ligand binding assays, RBL cells (5 ϫ 10 5 ) were distributed in Eppendorf tubes in a final volume 100 l of binding medium (DMEM supplemented with 20 mM HEPES and 10 mg/ml bovine serum albumin) containing different amounts of unlabled SDF-1 and 0.2 nM 125 I-labeled SDF-1 (2200 Ci/mmol). After incubation at 4°C for 2 h, the cells were centrifuged through a 10% sucrose gradient. The tips of tubes containing the pellets were cut and the radioactivity determined in a ␥-counter. Nonspecific binding in the presence of 100 nM SDF-1 (25-40% of total binding) was subtracted to give specific binding (29).
GTPase Activity-RBL cells (5 ϫ 10 7 ) were washed in phosphatebuffered saline, membranes were prepared and GTPase activity was determined in the absence and presence of various concentrations of SDF-1 as described previously (28,30,31).
Phosphoinositide Hydrolysis and Ca 2ϩ Mobilization-RBL cells were subcultured overnight in 96-well culture plates (50,000 cells/well) in an inositol-free medium supplemented with 10% dialyzed fetal bovine serum and 1-2 Ci/ml [ 3 H]inositol. Cells were washed with HEPESbuffered saline containing 20 mM LiCl and 0.1% bovine serum albumin and incubated in the same buffer with and without SDF-1. Reactions were stopped by adding 200 l of chloroform, methanol, 4 N HCl (100: 200:2) and the generation of total [ 3 H]inositol phosphates was determined (27,28). For Ca 2ϩ mobilization, cells (3 ϫ 10 6 ) were washed in HEPES-buffered saline and loaded with 1 M Indo I-AM in the presence of 1 M pluronic acid for 30 min at room temperature. The cells were washed and resuspended in 1.5 ml of buffer and intracellular Ca 2ϩ mobilization was measured as described (30).
Immunoprecipitations-RBL-2H3 cells (5-10 ϫ 10 6 ) were surface labeled with 125 I, lysed, immunoprecipitated with 12CA5 antibody, resolved by SDS-PAGE, and visualized by autoradiography as described (27,28). Phosphorylation of CXCR4 was performed essentially as described for other chemoattractant receptors (27,28). Briefly, RBL-2H3 cells (2.5 ϫ 10 6 ) were subcultured overnight in 60-mm tissue culture dishes. The following day, cells were rinsed twice with 5 ml of phosphate-free DMEM and incubated in the same medium supplemented with [ 32 P]orthophosphate (150 Ci/dish) for 90 min to metabolically label the intracellular ATP pool. Labeled cells were stimulated and epitope-tagged CXCR4 were immunoprecipitated from lysates with 12CA5 (10 g) monoclonal antibody and analyzed by SDS-PAGE and visualized by autoradiography (27,28). Phosphorylation of PLC␤3 was performed essentially by the same method as described above except 1-1.5g of PLC␤3 antibody was used in place of 12CA5 antibody.
CXCR4 Internalization-Stably transfected RBL cells were incubated for 30 min at 37°C with 100 nM SDF-1 or 100 nM PMA in HEPES (20 mM)-buffered DMEM. Cells were washed with ice-cold medium and incubated with 12G5 antibody for 60 min at 4°C. The cells were again washed in cold medium and incubated with fluorescein isothiocyanatelabeled goat anti-mouse IgG. Single color immunofluorescence analysis of 10,000 cells was performed on a FACScan flow cytometer. Percent receptor internalization was calculated from the mean channel fluorescence values of cells treated with buffer versus ligand or PMA.

Functional Expression of CXCR4 in RBL-2H3 Cells-To
study the regulation of the chemokine receptor CXCR4, stably transfected clonal lines of RBL cells expressing epitope-tagged wild type (Wt CXCR4) or a COOH-terminal truncation mutant at amino acid 318 of CXCR4 (⌬Cyto CXCR4, Table I) were generated. Fig. 1A shows FACS analysis with the CXCR4specific 12G5 antibody (14). The epitope tag-specific 12CA5 antibody was used to immunoprecipitate the Wt CXCR4 (ϳ44 kDa) and ⌬Cyto CXCR4 (ϳ41 kDa) proteins from surfaceiodinated RBL cells (Fig. 1B, lanes 3 and 5). No 125 I-labeled proteins in this size range were detected in untransfected RBL cells. The epitope tag peptide (YPYDVPDYA) inhibited the immunoprecipitation of these proteins (Fig. 1B, lanes 4 and 6).

KFKTAAQHALA-COOH
The epitope-tagged Wt and ⌬Cyto CXCR4 expressed in RBL cells bound SDF-1 with similar apparent affinity (Wt CXCR4 K d ϭ 6.1 Ϯ 1.5 nM; ⌬Cyto CXCR4 K d ϭ 9.6 Ϯ 1.4 nM, Fig. 2A). The native receptors expressed in neuronal cells had a K d ϭ 54 Ϯ 6.4 nM. (17). Based on the maximum specific binding of SDF-1, a similar number of Wt CXCR4 (9800 Ϯ 5200) or ⌬Cyto CXCR4 (7200 Ϯ 1800) receptors were expressed in these cells. The receptors expressed in RBL cells were functionally active and induced SDF-1 dose-dependent GTPase activity in membranes, inositol phosphate production and calcium mobilization in whole cells (Fig. 2, B-D). The ⌬Cyto CXCR4 was more active relative to Wt CXCR4 in stimulating GTPase activity (Fig. 2B) and phosphoinositide hydrolysis (Fig. 2C). ⌬Cyto CXCR4 resulted in a ϳ4-fold increase in total inositol phosphates upon SDF-1 treatment as compared with the ϳ1.6-fold increase observed with the Wt CXCR4 (Fig. 2C). While the peak calcium mobilization by the Wt CXCR4 or ⌬Cyto CXCR4 were comparable (250 -300 nM) the calcium transients induced by the ⌬Cyto CXCR4 contained a prolonged phase of elevated intracellular calcium (Fig. 2D).
Desensitization of CXCR4 Signaling-Calcium mobilization in response to SDF-1 was completely desensitized in cells expressing Wt CXCR4 by prior treatment with SDF-1 or PMA both of which phosphorylate the receptor. Interestingly, cAMP which did not cause phosphorylation of the Wt CXCR4 also desensitized the calcium mobilization by ϳ70% (Fig. 4A), suggesting a distal site of inhibition. Desensitization of calcium responses in ⌬Cyto CXCR4 showed that prior treatment with SDF-1 resulted in almost complete inhibition of the initial calcium peak with substantial maintenance of the sustained response (Fig. 4A). The sustained response was not seen in the presence of EGTA, indicating the second phase results from calcium influx (data not shown). Treatment with pertussis toxin resulted in complete inhibition of both rapid and sustained increases in calcium in ⌬Cyto CXCR4 cells (data not shown). PMA and cAMP also resulted in inhibition of the calcium responses in ⌬Cyto CXCR4 (Fig. 4A). The effect of PKC and PKA activation on the generation of inositol phosphates by the Wt and ⌬Cyto CXCR4 cells was determined. PKC activation resulted in complete inhibition of inositol phosphate generation in Wt CXCR4 cells and partial inhibition in ⌬Cyto CXCR4 cells (Fig. 4B). cAMP resulted in partial inhibition of inositol phosphates both in Wt and ⌬Cyto CXCR4 cells (Fig.  4B). Among the known phospholipase C␤ isoforms, only PLC␤3 is expressed in RBL cells (25). The effects of SDF-1, PMA, and cAMP on the phosphorylation of PLC␤3 are shown in Fig. 5. cAMP resulted in ϳ2-fold increase in PLC␤3 basal phosphorylation where as SDF-1 and PMA resulted in ϳ3-fold increases in PLC␤3 phosphorylation.
Role of Phosphorylation in CXCR4 Internalization-Internalization was studied in RBL cells using the 12G5 antibody. and ⌬Cyto CXCR4 by SDF-1 or PMA treatment are shown in Fig. 6B. Both SDF-1 and PMA induced internalization of the Wt CXCR4. The ⌬Cyto CXCR4 was also internalized by SDF-1 treatment, although consistently lower than the Wt CXCR4. PMA did not induce internalization of ⌬Cyto CXCR4 (Fig. 6B).

DISCUSSION
In this study, a well established model system was utilized to investigate the regulation of CXCR4, a chemokine receptor of considerable biological interest because of its role as a coreceptor for HIV-1 infection. The results suggest at least two distinct mechanisms for regulation of signal transduction, one at the level of receptor phosphorylation and the other at the level of phospholipase C activation. Functional expression of the native and the cytoplasmic tail deletion mutant of CXCR4 allowed for clear distinctions to be made of relative contributions of multiple mechanisms in the overall regulation of these receptors.
The higher activity of ⌬Cyto CXCR4 in inositol phosphates formation and in the induction of sustained calcium fluxes relative to Wt CXCR4 receptors suggests loss of some downregulatory control of the function of the mutant receptors. While comparable number of receptors with similar affinity were expressed in Wt CXCR4 and ⌬Cyto CXCR4 transfectants, cells expressing the latter receptor displayed a more potent ligand-induced GTPase activity (Fig. 2B). Such enhanced Gprotein activation was previously noted for other phosphorylation-deficient mutants of G-protein-coupled receptors, including the chemoattractant receptors for platelet activating factor (32,33). This is likely due to the lack of desensitization in G-protein coupling (34). Of interest, activation of ⌬Cyto CXCR4 by SDF-1 resulted in a biphasic calcium transient (Figs. 2C and 4A). Desensitization of calcium mobilization revealed unexpected complexity in the regulation of these receptors. While the calcium mobilization in Wt CXCR4 was completely desensitized by prior treatment with SDF-1, only the rapid release but not the sustained increase was inhibited in ⌬Cyto CXCR4. Selective desensitization of the first calcium peak suggests that calcium influx, as opposed to mobilization of intracellular calcium, is independently regulated, perhaps by distinct pertussis toxin-sensitive G-proteins or G-protein independent effectors. This remains to be explored.
The cytoplasmic tail of CXCR4 contains 18 serine/threonine residues which are likely targets for phosphorylation by Gprotein-coupled receptor kinases and second messenger-activated protein kinases (35). Like other CXC chemokine receptors (28,36), CXCR4 is also phosphorylated by PKC activation by PMA (Fig. 3). The complete inhibition of PMA-induced phosphorylation and partial inhibition of SDF-1-induced phosphorylation by staurosporine, a PKC inhibitor, suggests that agonist-induced phosphorylation of CXCR4 has two components, one represented by the activation of PKC and another due to a staurosporine-insensitive G-protein-coupled receptor kinase (GRK) that phosphorylates the agonist occupied form of the receptors. Multiple G-protein-coupled receptor kinases were previously identified in phagocytic leukocytes (37).
Receptor phosphorylation is an important mechanism for the homologous and heterologous desensitization of chemoattractant receptors (27,28,38). However, signals other than receptor phosphorylation were reported to be responsible for a phe- FIG. 4. Desensitization of CXCR4. A, RBL cells expressing Wt CXCR4 or ⌬Cyto CXCR4 (3 ϫ 10 6 cells/assay) loaded with Indo-1 were stimulated with SDF-1 (10 nM) and calcium transients shown. For homologous desensitization, SDF-1 (10 nM) stimulated cells were washed and 3 min later were restimulated with SDF-1 (10 nM) and the calcium trace was recorded for 2-3 min. Whenever complete inhibition of calcium release was observed, thrombin (1.0 units/ml), which causes calcium mobilization in RBL cells, was used to verify normal Indo-1 loading of the cells (not shown). For determining the effects of PMA and cAMP, cells loaded with Indo-1 were treated with PMA (100 nM) or 8-(4-chlorophenylthio)-cAMP (1.0 mM) and 3 min later were stimulated with SDF-1 (10 nM). The data shown for each of the conditions was representative of a minimum of three to five independent measurements. B, effect of cAMP and PMA on generation of inositol phosphates. Cells were labeled with myo-[ 3 H]inositol and preincubated for 10 min in buffer (Control), PMA (100 nM), or 8-(4-chlorophenylthio)-cAMP (1.0 mM) and total inositol phosphates (IPs) released by stimulation with 100 nM SDF-1 was determined as described under "Experimental Procedures." Data are presented as fold stimulation over basal inositol phosphates from three independent experiments of the average of triplicate measurements in each experiment.  7), or 8-(4-chlorophenylthio)-cAMP (cpt-cAMP) (lanes 4 and 8) as indicated. Cells were lysed, immunoprecipitated with PLC␤3 antibody, and analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography. This experiment was repeated twice with similar results. nomenon termed class desensitization (38,39). Recently, it was shown that PLC␤3 was phosphorylated both by PKC and PKA mediated pathways (25) and PLC␤2 was phosphorylated by PKA (24). Previous Western blot experiments using isoformspecific antibodies have shown that PLC␤3 is the only known PLC␤ isoform expressed in RBL cells. Recent experiments using reverse transcriptase-polymerase chain reaction also indicate that PLC␤3 is the only PLC␤ isoform expressed in these cells. 2 The data presented herein with CXCR4 and ⌬Cyto CXCR4 indicates that both PKC and PKA activation negatively regulates the signaling of CXCR4 independent of receptor phosphorylation. In addition, PKC also has a receptor phosphorylation-dependent action on the function of CXCR4. Demonstration of PLC␤3 phosphorylation by SDF-1, PMA, and cAMP in cells expressing ⌬Cyto CXCR4 suggest that phosphorylation of this enzyme may be responsible for attenuation of signal transduction through phospholipase C (Fig. 5).
Ligand binding is known to cause rapid internalization of many G-protein-coupled receptors, including chemokine receptors CXCR1 and CXCR2 (36, 40 -42). SDF-1 binding results in the internalization of CXCR4 and phosphorylation facilitates this process but is not absolutely required. In the case of ␤2adrenergic receptors, phosphorylation by receptor kinases was shown to enhance the arrestin-dependent internalization (43). PKC activation by PMA also induced sequestration of CXCR4 which was completely dependent on the presence of cytoplasmic tail. Phosphorylation of the cytoplasmic tail of CXCR4 is likely responsible for the down-regulation of CXCR4, but this contention will have to be confirmed by substitution mutations of the PKC sites on the receptor. PMA-induced down-regulation of CD4 in the presence of GP120 required an accessory protein and it was recently shown that this accessory protein is likely CXCR4 (44,45). The results presented here suggest that this down-regulation may involve phosphorylation on the cytoplasmic tail of CXCR4.
The phosphorylation independent reduction in surface expression of ⌬Cyto CXCR4 suggests an additional motif regulating internalization. Multiple mechanisms for internalization of surface proteins were described (40,46,47). Recent studies have demonstrated that phosphorylation at either one of the two independent clusters of phosphorylation sites in m2-muscarinic receptors is sufficient for agonist-induced internalization of the receptor, whereas mutation of both clusters severely impaired internalization (48). Internalization of type A cholecystokinin receptor was unaffected by C-terminal truncation, whereas internalization of type B cholecystokinin receptor was significantly reduced in a C-terminal truncation mutant (49). While arrestins and dynamin appear to play an important role in phosphorylation-dependent sequestration of G-protein-coupled receptors (43,46), the molecular mechanisms of phosphorylation-independent sequestration observed here with CXCR4 and previously with other G-protein-coupled receptors (49) remain to be determined.
It is, at present, not known whether internalization of receptors has any direct relevance to HIV-1 infection. While signal transduction through G-proteins is not required for the usage of CCR5 as co-receptor for HIV-1L, it is not known whether internalization defective mutants will act as co-receptors (50,51). The rapid ligand-induced internalization will have the effect of reducing surface expression and as a consequence availability of the co-receptor. In addition, PMA induced desensitization as well as down-regulation of CXCR4 suggests that activation of other receptors that enhance PKC activity are likely to have an effect on signal transduction and surface expression of this receptor.
In summary, we have established a model for stable functional expression of CXCR4 and shown that receptor activity is regulated at multiple levels by receptor phosphorylation dependent and independent mechanisms. The ability to express mutant receptors should allow for the analysis and functional consequences of interactions of HIV-1 GP120 and CD4 with CXCR4. FIG. 6. Internalization of CXCR4. A, FACS analysis of surface expression of CXCR4. RBL cells (5 ϫ 10 5 /sample) expressing the Wt CXCR4 were incubated in HEPES (20 mM)-buffered DMEM for 30 min at 37°C in the presence and absence of SDF-1 (100 nM) and washed in ice-cold medium and incubated with 12G5 antibody for 1 h followed by the fluorescein isothiocyanate-labeled goat anti-mouse IgG for 30 min. Cells were washed and fixed in 2% formaldehyde in phosphate-buffered saline and analyzed in FACScan flow cytometer. The staining of cells in the absence of primary antibody 12G5 (thin line) or with 12G5 antibody in cells incubated with (broken line) and without SDF-1 (thick line) treatment are shown. B, effect of SDF-1 and PMA on internalization of Wt CXCR4 and ⌬Cyto CXCR4. The effect of PMA and SDF-1 treatment on the surface expression of Wt CXCR4 and ⌬Cyto CXCR4 was measured as described above. Mean fluorescence values without any treatment were taken as 100%, and based on the mean fluorescence after treatment, the percent receptors lost from the surface were calculated. Data shown are average of three independent experiments and in each experiment duplicate samples were measured.