Bioluminescence Resonance Energy Transfer Reveals Ligand-induced Conformational Changes in CXCR4 Homo- and Heterodimers*

Homo- and heterodimerization have emerged as prominent features of G-protein-coupled receptors with possible impact on the regulation of their activity. Using a sensitive bioluminescence resonance energy transfer system, we investigated the formation of CXCR4 and CCR2 chemokine receptor dimers. We found that both receptors exist as constitutive homo- and heterodimers and that ligands induce conformational changes within the pre-formed dimers without promoting receptor dimer formation or disassembly. Ligands with different intrinsic efficacies yielded distinct bioluminescence resonance energy transfer modulations, indicating the stabilization of distinct receptor conformations. We also found that peptides derived from the transmembrane domains of CXCR4 inhibited activation of this receptor by blocking the ligand-induced conformational transitions of the dimer. Taken together, our data support a model in which chemokine receptor homo- and heterodimers form spontaneously and respond to ligand binding as units that undergo conformational changes involving both protomers even when only one of the two ligand binding sites is occupied.

In recent years, the concept of GPCR 1 dimerization has raised questions about the molecular details and functional role of such oligomeric assembly (for a recent review, see Ref. 1). Given the clinical interest in GPCRs, insights into the structural and functional organization of the receptor complexes have the potential to facilitate the design of new drug candidates with increased efficacy and selectivity. Resonance energy transfer (RET) techniques have emerged as methods of choice to study receptor dimerization in living cells. Although most RET studies indicate that many if not all GPCRs exist as dimers or higher oligomers under basal conditions, apparent contradictions exist concerning their potential dynamic regulation upon ligand binding. Although numerous authors did not find any effects of ligands on constitutive RET signals in their systems (2)(3)(4)(5)(6)(7)(8), others observed ligand-promoted increases or decreases that were interpreted as either the formation (9 -11) or the dissociation (12)(13)(14)(15) of GPCR dimers in response to receptor activation. Conformational changes within pre-existing constitutive dimers have also been proposed as alternative explanations for agonist or antagonist-induced changes in RET (16 -18).
Chemokine receptors such as CCR2 and CXCR4 have been reported to form homo-and heterodimers (3, 4, 19 -24). In early co-immunoprecipitation studies, Vila-Coro et al. (21) proposed that the dimerization of CXCR4 is induced upon activation by its chemokine ligand SDF-1. In contrast, data obtained with RET techniques revealed that CXCR4 homo-dimers form spontaneously in the absence of ligand (3,4,24). In one study, no significant effect of SDF-1 was observed on the constitutive energy transfer (4), whereas a small but reproducible increase was detected by others (24). As for CXCR4, agonist stimulation of CCR2 was found to promote the formation of dimers as revealed by chemical cross-linking followed by immunoprecipitation, suggesting that receptor dimerization and activation are interconnected processes (19). Heterodimerization between CXCR4 and CCR2 has initially been proposed to occur only in the case of a frequent genetic variant of CCR2, termed CCR2V64I, but not for the wild-type form of CCR2 (20). Given that CCR2V64I was associated with delayed AIDS progression (25)(26)(27), such a specific heterodimerization pattern could have important pathophysiological consequences. Indeed, it has been speculated that the AIDS-protective phenotype of the variant could be the result of its block of HIV entry via CXCR4 (20). In line with this proposition, the same authors proposed that the antiviral properties of a monoclonal anti-CCR2 antiantibody resulted from its ability to force the heterodimerization of CCR2 with CXCR4 (23).
Taken together, the results summarized above raise a number of questions concerning the dynamic nature of the homoand heterodimerization processes regulating GPCR function.
Among these, the question of whether receptor ligands can induce homo-or heterodimer association or dissociation or promote conformational changes within preformed dimers that remain stable through the activation cycle is still highly debated. The potential role of CXCR4/CCR2 heterodimerization in inhibiting HIV entry makes it a particularly relevant model to study this question. In addition to their role in HIV infection, these chemokine receptors have been shown to be involved in cancer metastasis as well as in various aspects of inflammatory diseases including the directed migration of leukocytes during acute immune responses and homing (28 -34). Therefore, understanding the dynamics of CXCR4/CCR2 homo-and heterodimerization after ligand binding takes on a particular importance when considering their potential as drug targets for numerous disease states.
In the present study, we took advantage of bioluminescence resonance energy transfer (BRET) approaches to study CXCR4/ CCR2 homo-and heterodimers in the course of receptor activation. We found that CXCR4 and CCR2 exist as constitutive homo-and heterodimers and that different ligands promote distinct conformational rearrangements of preformed stable oligomers. Our data also indicate that peptides derived from CXCR4 transmembrane domains block receptor activation by preventing the agonist-promoted conformational rearrangement of both CXCR4 homo-and CXCR4/CCR2 heterodimers.

EXPERIMENTAL PROCEDURES
Plasmids-The cloning of CXCR4-YFP and CXCR4-RLuc have been described previously (3). CCR2-YFP and CCR2-RLuc were constructed by ligating the coding sequence of CCR2, after its amplification by PCR, into the pGFP-N1-Topaze (PerkinElmer Life and Analytical Sciences) and (hRLuc)-N3 (BioSignal) backbones of the CXCR4-YFP and -RLuc using the HinDIII and AgeI or HinDIII and BamHI sites, respectively. The constructs were sequenced to ensure the absence of unwanted mutations. The 64V and 64I variants were obtained by site-directed mutagenesis using the Kunkel method. The amino acid sequence of the fusion linker regions between the terminal receptor residue and the initiator methionine of either YFP or RLuc were as follows: CXCR4-YFP, FHSSKPVATMVSKG; CXCR4-RLuc, FHSSKPGDPPARAT-MTSKV; CCR2-YFP, SAGLGPVATMVSKG; and CCR2-RLuc, SAGL-GDPPARATMTSKV. Bold italic characters identify linker residues resulting from the cloning strategy and that derive neither from the receptor nor the fluorophore sequences. The sequences of the linker regions and of the mutated residues were verified by direct sequencing.
Cell Culture and Transfection-HEK293T cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin and streptomycin, 2 mM L-glutamine (all from Wisent). 24 h before transfection, cells were seeded at a density of 500,000 cells per well in 6-well dishes. Transient transfections were performed using the FuGENE-6 transfection reagent (Roche) in OptiMEM medium (Invitrogen) In general, 0.1 g of CXCR4-RLuc or CCR2-RLuc was transfected alone or with increasing quantities of YFP-tagged CXCR4 or CCR2. The total amount of DNA transfected in each well was completed to 2 g with empty vector. After overnight incubation, transfection medium was replaced with fresh Dulbecco's modified Eagle's medium for 3 h to allow cell recovery. Transfected cells were then seeded in 96-well white plates (with clear bottoms that had been pre-treated with poly-D-lysine) and left in culture for 24 h before being processed for BRET assay.
Flow Cytometry-Transfected HEK293T cells were detached with phosphate-buffered saline (PBS) containing 1 mM EDTA 24 h after cotransfection with CXCR4-RLUC and CXCR4-YFP as indicated. Peripheral blood mononuclear cells (PBMC) were isolated on a Ficoll (Amersham Biosciences) gradient from whole blood. The cells were stimulated with 10 g/ml phytohemagglutinin and cultured for 7 days in RPMI containing 10% fetal bovine serum and 50 IU/ml interleukin-2. For CXCR4 staining, the cells were incubated 45 min at 4°C in PBS containing 2% fetal bovine serum with anti-CXCR4 phycoerythrinconjugated 12G5 monoclonal antibody (Santa Cruz Biotechnology). They were then washed three times with PBS, and cell surface expression of CXCR4 was quantified by flow cytometry on a FACSCalibur flow cytometer (BD Biosciences).
BRET Assays-For routine BRET measurements, cells were washed once with PBS 36 to 48 h after transfection and coelenterazine H (Nanolight Technology) added to a final concentration of 5 M in PBS. Readings were then collected using a multidetector plate reader MITHRAS LB940 (Berthold Technologies, Bad Wildbad, Germany) allowing the sequential integration of the signals detected in the 480 Ϯ 20 nm and 530 Ϯ 20 nm windows for luciferase and YFP light emissions respectively. The BRET signal is determined by calculating the ratio of the light intensity emitted by the Receptor-YFP over the light intensity emitted by the Receptor-RLuc. The values were corrected by subtracting the background BRET signal detected when the Receptor-RLuc construct was expressed alone. To assess the effects of ligands, SDF-1, AMD3100, TC14012, and MCP-1 were added at the concentrations indicated in the text and in the figure legends and incubated at 37°C for 5 min before the addition of coelenterazine H and BRET reading. When indicated, ligands were added in presence of 0.1% BSA (Sigma). For experiments with transmembrane peptides, the cells were preincubated with the different peptides in PBS for 15 min at 37°C before agonist exposure.
For acquisition of full BRET spectra, cells were transfected as described above with different amounts of CXCR4-YFP for a given quantity of CXCR4-RLuc (0.1 g). Cells were detached and resuspended in PBS containing 0.1% (w/v) glucose. 200,000 cells were seeded in 100 l of PBS in a 96-well plate with a clear bottom (Corning), and BRET scan was performed in a Flex-station 2 (Molecular Devices) by reading luminescence between 400 and 600 nm immediately after the addition of coelenterazine for cells expressing different [acceptor]/[donor] ratios.
For BRET titration experiments, net BRET ratios were expressed as a function of the [acceptor]/[donor] ratio (39). Total fluorescence and luminescence were used as relative measures of total expression of the acceptor and donor proteins, respectively. Total fluorescence was determined with MITHRAS using an excitation filter at 485 nm and an emission filter at 535 nm. Total luminescence was measured in the MITHRAS, 10 min after the addition of coelenterazine and the reading taken in the absence of emission filter.
cAMP Production-To determine cAMP accumulation, HEK293T cells were seeded in 24-well microplates at 10 5 cells/well (coated with 0.1% poly-D-lysine) 24 h before the experiment and labeled for 2-3 h in Dulbecco's modified Eagle's medium without fetal bovine serum containing 2 Ci/ml [ 3 H]adenine (PerkinElmer Life and Analytical Sciences). Because CXCR4 is coupled to Gi proteins, the relative efficacy of SDF-1 to inhibit forskolin-induced cAMP production was monitored in different conditions. Cells were stimulated in presence of 20 M forskolin (Sigma) alone or 20 M forskolin and 1 nM SDF-1 for 30 min at 37°C in Dulbecco's modified Eagle's medium containing 50 mM HEPES, pH 7.4, 0.1% BSA, 1 mM 3-isobutyl-1-methylxanthine (Sigma) and supplemented or not with 10 M of each of the CXCR4 TM peptides. The reaction was terminated by removing the Dulbecco's modified Eagle's medium/3-isobutyl-1-methylxanthine/ligand solution and the addition of ice-cold 5% trichloroacetic acid. [ 3 H]cAMP was purified by sequential chromatography (Dowex resin/aluminum oxide).
Binding Assays-SDF-1 binding to CXCR4 was assessed indirectly by flow cytometry as described previously (40). In brief, the ability of SDF-1 to compete for the binding of the monoclonal anti-CXCR4 antibody 12G5 to CEM cells was used to determine SDF-1 binding. SDF-1 was co-incubated at 4°C for 30 min with the antibody and SDF binding determined by the loss of 12G5 labeling, as determined by flow cytometry. To test whether CXCR4 TM peptides interfered with SDF-1 binding to CXCR4, the peptides were incubated with the cells 15 min before addition of SDF-1 and 12G5. Control tubes were incubated with peptides and 12G5, without SDF-1.
For T14012 binding assay, CXCR4-expressing HEK293T cells were detached with 5 mM EDTA, washed twice in binding buffer (50 mmol/L HEPES, pH 7.4, 1 mmol/L CaCl 2 , 5 mmol/L MgCl 2 , and 0.5% BSA) and resuspended at final concentration of 5 ϫ 10 5 cells/ml. Total SDF-1 binding was measured with 0.1 nM 125 I-SDF-1 (2200 Ci/mmol; PerkinElmer Life and Analytical Sciences) as tracer, and TC14012 competition assays were performed with 100 nM TC14012 in the presence or absence of 10 M of each CXCR4 peptide. The samples were incubated for 90 min at 20°C, and binding was terminated by rapid filtration through glass fiber (GF/C) filters (Whatman) using ice-cold PBS containing 0.5 M NaCl. The retained radioactivity was counted in a ␥ counter (1271 RIAgamma Counter; PerkinElmer Life and Analytical Sciences).
Data Analysis-Data obtained in BRET assays were analyzed using Prism 3.0. Statistical significance of the differences between the different conditions were calculated using one-way analysis of variance with a Bonferroni post-test for p values less than 0.05.

Constitutive CXCR4 Dimers Revealed by BRET-
The existence of constitutive CXCR4 dimers was probed by monitoring the occurrence of intermolecular interactions among CXCR4 molecules under basal conditions using a proximity-based BRET assay. For this purpose, a constant amount of CXCR4-RLuc expression vector was cotransfected with increasing amounts of CXCR4-YFP encoding plasmid. The entire emission spectrum between 400 and 600 nm was then analyzed for various CXCR4-YFP/CXCR4-RLuc (energy acceptor/energy donor) ratios after the addition of the luciferase substrate coelenterazine. As shown in Fig. 1, increasing the expression level of CXCR4-YFP led to a progressive increase in the amount of light emitted in the 510 -550 nm region that resulted from the transfer of energy from the luciferase to the YFP with the ensuing emission of light by the latter (Fig. 1A). The occurrence of BRET between RLuc and YFP was further illustrated by the reduction in emission observed in the 450 -510 nm part of the spectrum that corresponds to the region of overlap between RLuc (energy donor) emission and YFP (energy acceptor) excitation wavelengths allowing the energy transfer. The basal BRET observed in the absence of any receptor ligand indicates, in agreement with previous reports (3,4,24), that CXCR4 exists as a constitutive homodimer. For all subsequent BRET experiments, the emission of light was measured only in the 460 -500 nm and 510 -550 nm windows, corresponding to the RLuc and YFP emission peaks, respectively, and the BRET defined as the ratio of light detected in these two channels after coelenterazine addition. As can be seen in Fig. 2A, the BRET signal increased as a hyperbolic function of the CXCR4-YFP/ CXCR4-RLuc ratio. The saturation of the BRET titration curve is indicative of a specific protein-protein interaction, because random molecular collisions that would give rise to bystander BRET would be expected to increase nearly linearly over a wide range of YFP/RLuc (39,41). The selectivity of the observed signal is further supported by the fact that co-expression of CXCR4-RLuc with an unrelated GPCR, GBR2-YFP, led to marginal signal that progressed linearly over the same range of energy acceptor/donor. The positive BRET signal did not result from a non-physiological overexpression of the receptors, because CXCR4 immunostaining followed by flow cytometry analysis revealed that the highest expression levels reached in transfected HEK293T cells (1 g of CXCR4-YFP ϩ 0.1 g of CXCR4-RLuc) was still lower than those observed in activated peripheral blood mononuclear cells (Fig. 1B).
Ligand-induced Modulation of the CXCR4 Homodimer BRET Signal-To assess the effect of ligand binding on the BRET signal observed for the CXCR4 homodimer, full BRET titration curves were obtained in the presence and absence of the CXCR4 agonist SDF-1 or the polyphemusin II-derived inverse agonists peptide analogue TC14012 (35). As can be seen in Fig. 2A, the addition of SDF-1 increased the maximal BRET signal observed, whereas TC14012 decreased it. It is interesting that neither compound affected the shape of the curve so that the concentration of CXCR4-YFP needed to reach 50% of the maximal BRET signal (BRET 50 ) remained unaffected by the treatments. Because the BRET 50 represents the propensity of the protomers to interact with one another (i.e. their relative affinity), our data indicate that the ligand treatments did not change the number of complexes. Rather, the maximal BRET signal increase most likely reflects conformational changes, within preformed receptor dimers, that affect the distance between the energy donor and acceptor. SDF-1 had no effect on the marginal signal observed between CXCR4-RLuc and the unrelated GBR2-YFP, confirming the selectivity of the effect.
The dose dependence of the ligand effect is illustrated in Fig.  2B. In cell expressing a given CXCR4-RLuc/CXCR4-YFP ratio, the agonist SDF-1 and inverse agonist TC14012 dose-dependently increase and decrease the basal BRET signal, respectively. It is interesting that the bicyclam weak partial agonist AMD3100 (42) increases the BRET signal, albeit to lower extent than the full agonist SDF-1. The unrelated CCR2-selective chemokine MCP-1 had no effect on the BRET signal, confirming that ligand interaction with CXCR4 is required to promote BRET changes. The dose-response curves were carried out in both the absence and the presence of the carrier protein BSA (0.1%). As can be seen in the Table inset, the presence of BSA significantly increased the apparent potency of the compounds, indicating the occurrence of nonspecific adsorption or inactivation of the diluted ligands in the absence of carrier. The EC 50 determined for SDF-1 in the presence of BSA was well within the range of K d values previously reported for SDF-1 binding to CXCR4 (4 -85 nM) (43)(44)(45)(46)(47), whereas the EC 50 for TC14012 and AMD3100 are comparable with the IC 50 values obtained in binding competition experiments using 125 I-SDF-1 as the radioligand (42).
The pharmacological selectivity of the ligand-promoted BRET changes was further demonstrated by the competitive nature of the effects. Indeed, as shown in Fig. 3, TC14012 dose-dependently blocked the ability of 100 nM SDF-1 to increase the BRET signal and eventually reverse this increase revealing the inhibitory action of the inverse agonist (Fig. 3A). Likewise, increasing concentration of the partial agonist AMD3100 progressively blocked the SDF-1-promoted BRET increase until it reached the modestly elevated level corresponding to the partial agonistic activity of AMD3100 (Fig. 3B). Taken together, these results indicate that three compounds with different intrinsic efficacies led to distinct conformational changes of the CXCR4 dimer.
CCR2 Homo-and Heterodimers-CXCR4 has been previously demonstrated to form heterodimers with CCR2, but it remained unclear whether these heterodimers are spontaneously formed or induced by the presence of one or both receptor ligands (20,23,48). To clarify this issue, we first investigated the formation of constitutive CCR2 homodimers and their modulation by the CCR2 selective agonist MCP-1. As can be seen in Fig. 4A, saturating hyperbolic BRET titration curves revealed the spontaneous formation of CCR2 homodimer. The selective agonist increased the maximal BRET signal without affecting the BRET 50 in a manner similar to that observed for the constitutive CXCR4 homodimer, indicating that the conformation and not the number of dimers was affected by the ligand binding. The pharmacological selectivity of the effects was again confirmed by the fact that the CXCR4-selective agonist SDF-1 had no effect on the CCR2 dimer BRET signal (Fig. 4B).
As was the case for each of the receptors expressed individually, coexpression of CCR2 and CXCR4 led to the formation of constitutive heterodimers revealed by specific basal BRET signals and hyperbolic saturating BRET titration curves between CCR2-RLuc and CXCR4-YFP as well as in the reverse orientation (between CXCR4-RLuc and CCR2-YFP) (Fig. 5A). It is interesting that the selective binding of ligands to a single protomer was sufficient to promote heterodimer BRET changes for the two BRET orientations. However, the pattern of ligand effects was different from that observed for each of the homodimers. In addition, the BRET pair orientation influenced the ligand response pattern. When considering the CXCR4-RLuc/CCR2-YFP orientation, the addition of the CCR2 agonist MCP-1 led, as was the case for the CCR2 homodimer, to an increase of the heterodimer BRET signal. In contrast, the CXCR4 agonist SDF-1, which promoted an increase of the CXCR4 homodimer BRET, decreased the BRET signal originating from the heterodimer. The CXCR4 inverse agonist TC14012 for its part had a similar effect on the homo-and heterodimer, leading to a decrease in the BRET signal (Fig. 5B,  left). When the reverse BRET partner orientation (CCR2-RLuc/ CXCR4-YFP) was investigated, the CXCR4 ligands SDF-1 and TC14012 retained their inhibitory effect on the heterodimer BRET signal. MCP-1, however, which increased the BRET observed between CXCR4-RLuc and CCR2-YFP, led to a dramatic reduction of the BRET signal obtained for the CCR2-RLuc/CXCR4-YFP pair (Fig. 5B, right). Taken together, these results clearly indicate that the orientation of the ligand-promoted BRET changes cannot be used as a direct reflection of the intrinsic ligand efficacy. Rather, it seems to be dependent on both the nature of the ligand and the BRET pairs considered; both the identity of the receptor protomers and the relative position of the energy donor and acceptor within the dimers influence the responses observed.
The observation that MCP-1 can either increase or decrease the BRET signal emanating from the CXCR4/CCR2 heterodimer depending on the relative RLuc/YFP orientation can hardly be reconciled with the hypothesis that BRET changes reflect ligand-induced dimer association or dissociation. Indeed, the same ligand should not lead to opposite effects on the same receptor heterodimer. Given that both the distance and the orientation between the energy donor and acceptor determine the BRET efficacy, the distinct BRET modulations observed most probably reflect conformational rearrangements that change either the distance or the relative orientation between the fluorophores. Because these parameters are affected by the initial relative position of the RLuc and YFP within the dimers, it is to be expected that the same conformational switch imposed by a ligand could result in very different BRET changes when different BRET configurations are considered.
A previous study suggested that CXCR4/CCR2 heterodimers could be formed only with the CCR2V64I variant form of the receptor and not with the wild-type CCR2 (20). Given that the CCR2 64I variant is associated with delayed AIDS onset in persons infected with HIV, the finding was suggestive that the phenotype of the 64I variant could be mediated by an inhibition of CXCR4 usage by HIV as a result of its heterodimerization (23,26). To reassess this possibility, we systematically used both CCR2 variants to measure both basal and ligand-modulated BRET signals generated by homo-and heterodimers but failed to detect any significant difference between them (data not shown). Therefore, the mechanism for the observed protective phenotype against AIDS progression of CCR2V64I is not related to its ability to heterodimerize with CXCR4.
Effects of Peptides Derived from the CXCR4 Transmembrane Domains on both CXCR4 Homo-and Heterodimers-Previous work had found that peptides derived from CXCR4 transmembrane domains are rapidly associating with the receptor, blocking its signaling as well as its HIV-1 coreceptor function (36). We asked whether these effects could result from the dissociation of constitutive CXCR4 dimers, a mechanism that had been suggested for the effect of a peptide derived from TMVI of the ␤-adrenergic receptor (38). For this purpose, the effect of four peptides derived from TMs II, IV, VI and VII was assessed on the basal CXCR4 homodimer BRET signal. As shown in Fig.  6A, none of the peptides affected the constitutive BRET signal as shown by the unaltered BRET titration curves, ruling out peptide-promoted dissociation as the basis of their functional inhibitory action. However, all peptides blocked the SDF-1- induced BRET increase, and TMs II and IV were the most efficacious (Fig. 6B), indicating that inhibition of the agonistpromoted conformational change could underlie the mechanism of action of the peptides. It is interesting that the efficacies of the peptides in the BRET assay were similar their relative ability to block SDF-1 promoted inhibition of adenylyl cyclase activity (Fig. 6D). Indeed, whereas TMs II and IV acted as complete inhibitors in both assays, TMs VI and VII acted as partial blockers at 10 M. When considering the inverse-agonist TC14012, only TMII significantly attenuated the ligand- promoted BRET reduction; TMs IV, VI, and VII led only to marginal inhibition that did not reach statistical significance (Fig. 6B). The inhibitory action of the peptides seems to be directly linked to the inhibition of the activation process and not to the ligand binding to the receptor, because neither SDF-1 nor TC14012 binding to CXCR4 was affected by the peptides (data not shown). The differential effect of the peptides on the agonist-and inverse agonist-promoted changes further confirmed that the two ligands promoted distinct conformational changes that are differentially affected by the peptides. The effect was specific for CXCR4 because the peptides did not interfere with the MCP-1-induced increase of the CCR2 homodimer BRET signal (Fig. 6C). In addition, two peptides derived from the ␤2-adrenergic receptor TM VI (38) were without effect on the SDF-1 promoted increase in BRET between CXCR4-RLuc and CXCR4-YFP (data not shown).
We then examined the effect of the CXCR4-derived peptides on the CXCR4/CCR2 heterodimer. In the absence of ligands, the four peptides had no effect on the basal heterodimer BRET signal obtained between CCR2-RLuc and CXCR4-YFP (Fig.  7A), similar to what was observed for the CXCR4 homodimer. However, all four peptides blocked, albeit to different extents, the SDF-1-promoted BRET change, although only TMII significantly affected the TC14012-induced BRET reduction (Fig.  7B). This pattern of inhibition, which is similar to that observed for the CXCR4 homodimer suggests that comparable CXCR4 conformational changes occur upon ligand binding whether the receptor is part of an homodimer or within a CXCR4/CCR2 heterodimer. It is interesting that a modest (ϳ25%) but statistically significant reduction of the heterodimer BRET response to the CCR2-selective agonist MCP-1 was also observed upon treatment with the CXCR4-derived TMII, TMIV, and TMVI peptides (Fig. 7B). This observation suggests that the heterodimer conformational changes induced by ligand binding to a single receptor protomer may not only involve changes within the ligand-bound receptor. Instead, trans-receptor conformational reorganization may be transmitted to the CXCR4 protomer upon ligand binding to the CCR2 protomer, and these transmitted conformational changes could be blocked by the CXCR4-derived peptides. On the other hand, the observed effects may be related to the capacity of the CXCR4-bound peptides to reduce CCR2 movements within the heterodimer. DISCUSSION Recent appreciation that GPCRs exist as dimers has raised a number of questions regarding the molecular dynamics and functional role of such oligomeric organization. Among these, whether dimerization is a constitutive phenomenon or is promoted by ligand binding remains highly debated. The role of dimerization in the receptor activation process also remains poorly understood. In this article, we present evidence indicating that, at least for CXCR4 and CCR2, receptors exist as constitutive homo-and heterodimers and that ligand binding induces conformational changes within pre-existing complexes without promoting the formation or dissociation of dimers. Taking advantage of CXCR4-derived peptides that can noncompetitively block receptor function, we also showed that ligand promoted changes in dimer conformations are intimately linked with receptor function.
Previous studies have led to conflicting interpretations concerning the dynamic nature of GPCR dimers. In particular, although many authors interpreted ligand-promoted changes in resonance energy transfer signals between GPCR protomers as evidence for receptor dimer formation or dissociation, (9,(12)(13)(14)48), others inferred conformational changes within constitutive receptor dimers (16,17,41).
When considering CXCR4 and CCR2, early co-immunoprecipitation studies suggested that dimers formed only after stimulation with chemokines (20 -22). Later work using either fluorescence or bioluminescence resonance energy transfer methods revealed the existence of spontaneous CXCR4 dimers (3,4,24). In two of these studies, the effect of the agonist SDF-1-was assessed on the basal RET signals. Whereas a modest increase that did not reach statistical significance was observed in one study (4), a reproducible increase was observed in the other (24). However, the data obtained did not allow determination of whether the RET augmentation represented an increase in dimer formation or a conformational change within the preformed dimers.
In the present study, in addition to confirming the existence of constitutive CXCR4 homodimers, our BRET results demonstrate that both CCR2 homodimers and CXCR4/CCR2 heterodimers can form spontaneously. Three lines of evidence supported that the basal BRET signals observed truly represent specific constitutive dimerization and are not simply nonspecific "bystander BRET" that would result from tight random-packing of monomeric receptors: 1) BRET titration experiments gave rise to saturating hyperbolic curves that are characteristic of specific protein-protein interactions rather than random molecular collisions (39), 2) no specific BRET signal could be detected between either CXCR4 or CCR2 and the unrelated GABAbR2, 3) receptor occupancy resulted in either ligand-specific increases or decreases of the BRET signal; such a result would not be expected if the basal BRET signal resulted from nonspecific random collisions. In addition, the receptor expression levels of the transfected cells used in this study are similar to those observed in primary lymphocytes, thereby excluding the idea that dimerization is a result of receptor overexpression.
Our study also provides a conclusive demonstration that ligand-promoted modulation of CXCR4 and CCR2 homo-and heterodimer BRET signals results not from changes in dimer numbers but from the rearrangements of preformed dimers. First, the BRET titration curves revealed that the BRET 50 values, which would be expected to change if the propensity to form dimers was affected (39,49), were unaltered in the presence of an agonist or an inverse agonist, suggesting that the ligands did not affect the apparent affinity of the receptor protomers for one another. Changes in maximal BRET signal in the absence of apparent altered affinity is best explained by conformational changes that change the distance and/or orientation between the energy donor and acceptor affecting the energy transfer efficacy. Second, the dependence of the MCP-1-promoted change in BRET signal on the orientation of the CXCR4/CCR2 heterodimer BRET partners (i.e. MCP-1 increased the BRET signal for the CXCR4-RLuc/CCR2-YFP pair but decreased it for CCR2-RLuc/CXCR4-YFP) is hardly compatible with ligand-promoted dimer formation or dissociation. In fact, it seems highly unlikely that the fluorophore fusion orientation alone would determine whether MCP-1 induces dimer association or dissociation. Rather, the dependence of the BRET changes (i.e. increase or decrease) on the BRET pair configuration probably reflects structural particularities of the respective pairs studied. Indeed, the specific initial structural state determined by the particular combination of a receptor C terminus fused to either the energy donor or the acceptor could greatly influence how the same conformational switch is sensed by the BRET partners.
Considering the extent of the BRET changes promoted by various ligands for the CXCR4 homodimer, our results seem to contradict the previously reported CXCR4 model proposed by Trent et al. (50). In this model, the inverse agonist T140 is predicted to induce only minor rearrangements, whereas more important changes are expected from the binding of the weak agonist AMD3100. It should be pointed out, however, that the model only considered the monomeric form of the receptor; it is perhaps more important that the initial state was assimilated to a fully inactive conformation (based on the available rhodopsin structure (51)). Taking into account the previous report that CXCR4 displays a level of spontaneous activity that can be inhibited by TC14012 but is almost not affected by AMD3100 in cells (42), one could propose that the average basal dimer conformation detected by BRET represents a partially activated conformation that resembles the one stabilized by AMD3100. It follows that AMD3100 would not promote important conformational changes thus only marginally affecting the basal BRET signal, whereas the stabilization of a fully inactive conformation by TC1402 would be translated in considerable BRET changes, which is what we observed in living cells.
In the case of the CXCR4 and CCR2 homodimers, ligandinduced changes in BRET signals nicely parallel the intrinsic efficacy of the ligands; that is, agonists promote signal increases (a full agonist yielding a greater response than a partial agonist) whereas inverse agonists decrease the signal. It would therefore be tempting to speculate that the direction of the BRET changes reflects specific conformational changes that can be directly linked to the signaling efficacy. However, the data obtained with the CXCR4/CCR2 heterodimer questions such a direct relationship. Indeed, the CCR2 agonist MCP-1 either increased or decreased the BRET signal depending on the BRET partner orientation used (see above), demonstrating that the direction of ligand-induced BRET signal inflection may depend on C-terminal structural constraints and thus differ in different systems. It would thus be prudent to conclude that different BRET changes reflect distinct ligandstabilized receptor conformations but cannot be used to predict intrinsic efficacies. Similar conclusions have already been suggested by other authors (16).
Although no direct relationship between the direction of the BRET changes induced by ligands and their intrinsic efficacies can be strictly established, our data with the CXCR4-derived TM peptides strongly suggest that the conformational changes detected are linked to receptor activity. The original underlying rationale for testing the effect of the inhibitory peptides developed by Tarasova et al. (36) was that TM-derived peptides could bind to the protomer interface within dimers, thereby interfering with dimerization and inhibiting receptor activities. Indeed, the activity of a similar peptide has been suggested to result from the disruption of the ␤2-adrenergic receptor dimers (38), an interpretation that is in line with the proposed roles of TM domains in other GPCR dimer interfaces (52,53). Hernanz-Falcon et al. recently reported that simultaneous mutation of two residues located in the first and fourth TM domains abolished CCR5 dimerization, and that small peptides corresponding to these regions had the same effect (48). On the other hand, it has been proposed that TM-peptides can act by interfering with intramolecular TM packing, thus inducing receptor distortion (54). Our finding that CXCR4 TM-derived peptides did not affect basal BRET levels indicate that they did not function by inhibiting dimerization or by causing major distortion in the initial conformation, because both of these should have caused detectable BRET changes. Our data suggest rather that peptide binding stabilizes the initial conformation hampering the occurrence of any ligand-promoted conformational changes. Given the anti-HIV activity of these peptides (Ref. 36 and data not shown), this interpretation has important implications for the role of CXCR4 in HIV entry, in that it suggests that the peptides could block envelope-induced conformational changes of CXCR4 that are mandatory for viral entry. It remains to be investigated whether the apparently different efficacy of the various peptides to inhibit both function and ligand-promoted conformational changes reflects the respective importance of specific TM domains in the dimerization process.
The observation that CXCR4-derived TM peptides could block the conformational rearrangement of the CXCR4/CCR2 heterodimer promoted by the selective CCR2 agonist MCP-1 provides some clues about the functional organization of chemokine receptor dimers. Indeed, these results demonstrate that the presence of only one selective chemokine is sufficient to change the conformation of the heterodimer. In addition, it is tempting to speculate that conformational changes could be transmitted in trans from one protomer to the other. Such interprotomer transmission of conformational changes would suggest that the activity of one receptor could be affected by ligand binding to the other. This type of transactivation has been shown previously for the metabotropic GABAb receptor, where agonist binding to the GABABR1 protomer led to the functional engagement of the heterotrimeric G protein by the GABAbR2 protomer (55,56). The occurrence of trans-dimer conformational rearrangements could have great impact on the therapeutic use of receptor ligands because they may impinge not only on the activity of the cognate receptor but also on that of the heterodimer partner. On the other hand, however, our results could also be explained by the capacity of the CXCR4derived peptides to interfere with CCR2 movements within the heterodimer.
The fact that CXCR4/CCR2 heterodimers are conformationally responsive to selective ligands of either receptor could have potentially important physiological implications. Cell migration frequently involves several chemokines, and the mechanisms leading to the integration of these multiple signals are poorly understood (57)(58)(59). Because many chemokine receptors are co-expressed in the same cell types and can be involved together in a single migration process, a role for heterodimers in integrating the multiple signals could be envisioned. For example, CCR2 and CXCR4 have been shown to both contribute to accumulation of activated/memory T-cells in lymph node (60). Whether they play sequential roles or are simultaneously activated, possibly by the intermediate of receptor heterodimers, will require further investigations.