Rescue of HIV-1 Receptor Function through Cooperation between Different Forms of the CCR5 Chemokine Receptor*

Interaction of the human immunodeficiency virus (HIV-1) envelope glycoproteins with the CCR5 chemokine receptor, a G-protein-coupled receptor, triggers a membrane fusion process and virus entry. Cooperation for HIV-1 receptor activity was observed when two forms of CCR5 were coexpressed, either the wild-type (WT) receptor and a defective mutant with deletion of the amino-terminal (NT) extracellular domain or the latter ΔNT mutant and a human-mouse CCR5 chimera bearing the NT domain from human CCR5. Cooperation was most efficient when the two forms of CCR5 were in a 1:1 ratio. It was not observed between the CCR5 ΔNT mutant and a chimeric receptor (5444) in which the NT domain of CCR5 was in the context of another G-protein-coupled receptor, the HIV-1 receptor CXCR4. These results suggested that physical association between two forms of CCR5 was required for their cooperation. Coimmunoprecipitation experiments in transfected cell lysates indeed showed that the ΔNT CCR5 mutant formed oligomeric complexes with the WT CCR5 or the HMMM chimera but not with the CXCR4-derived chimera 5444. These observations suggest that the formation of CCR5 oligomers is a constitutive process independent from activation by chemokine ligands. The interaction of HIV-1 with independent subunits of CCR5 oligomers could favor the local recruitment of fusiogenic proteins and the formation of a fusion pore.

Interaction of the human immunodeficiency virus (HIV-1) envelope glycoproteins with the CCR5 chemokine receptor, a G-protein-coupled receptor, triggers a membrane fusion process and virus entry. Cooperation for HIV-1 receptor activity was observed when two forms of CCR5 were coexpressed, either the wild-type (WT) receptor and a defective mutant with deletion of the amino-terminal (NT) extracellular domain or the latter ⌬NT mutant and a human-mouse CCR5 chimera bearing the NT domain from human CCR5. Cooperation was most efficient when the two forms of CCR5 were in a 1:1 ratio. It was not observed between the CCR5 ⌬NT mutant and a chimeric receptor (5444) in which the NT domain of CCR5 was in the context of another G-proteincoupled receptor, the HIV-1 receptor CXCR4. These results suggested that physical association between two forms of CCR5 was required for their cooperation. Coimmunoprecipitation experiments in transfected cell lysates indeed showed that the ⌬NT CCR5 mutant formed oligomeric complexes with the WT CCR5 or the HMMM chimera but not with the CXCR4-derived chimera 5444. These observations suggest that the formation of CCR5 oligomers is a constitutive process independent from activation by chemokine ligands. The interaction of HIV-1 with independent subunits of CCR5 oligomers could favor the local recruitment of fusiogenic proteins and the formation of a fusion pore.
Chemokines represent a family of at least 40 structurally related small proteins (8 -10 kDa) mediating chemotactic migration of leukocytes through binding and activating receptors with seven membrane-spanning domains coupled to heterotrimeric G proteins (GPCR) 1 (for review see Refs. [1][2][3][4]. Chemokines are classified according to the relative position of two conserved cysteine residues in their amino-terminal region, the major subgroups being termed CXC (or ␣) and CC (or ␤) and the same nomenclature (CXCR and CCR) used to designate their receptors (5). In addition to their role in cell signal transduction, chemokine receptors have attracted a lot of interest because they represent cell entry portals for the human immunodeficiency viruses (HIV-1 and HIV-2) and related simian or feline retroviruses. A list of chemokine receptors or related orphan GPCRs can mediate HIV-1 entry in certain experimental conditions, but only two of them, CCR5 and CXCR4, seem to be used in vivo. The predominant role of CCR5 is indicated by the resistance to HIV-1 infection of individuals genetically deficient for the expression of this receptor, the most frequent defect being a 32-nucleotide deletion (⌬32 allele) resulting in a translation frameshift after residue 187 (reviewed in Refs. 6 -9). Although HIV-1 strains infecting cells via CCR5 (termed R5 strains) can be isolated throughout infection, strains using CXCR4 (termed X4) or both receptors (R5X4) generally emerge at more advanced stages. The selectivity of HIV-1 for CCR5 or CXCR4 can explain differences in cell tropism such as the ability of R5 but not X4 strains to replicate in macrophages. The inhibition of HIV-1 infection by chemokines and other CCR5 or CXCR4 ligands raises hope for novel antiviral strategies (1,7,10) and stimulates investigations on the role of these receptors in the initial steps of the virus life cycle.
The cell entry of HIV-1 and other retroviruses is mediated by their envelope glycoproteins (Env), which consist in trimeric complexes of a surface subunit (gp120 in the case of HIV-1) responsible for contacts with target cells and a transmembrane subunit (gp41) mediating membrane fusion (reviewed in Refs. [11][12][13]. The conformation changes in the Env complex required to activate the fusiogenic properties of gp41 are considered to be triggered by the interaction of gp120 with CCR5 or CXCR4. This interaction has been observed by different techniques, usually in presence of soluble forms of the CD4 protein that led to consider CCR5 and CXCR4 as CD4-associated HIV-1 coreceptors (14 -16). However, the possibility of CD4-independent binding of gp120 to CCR5 and CXCR4 (17)(18)(19) as well as evolutionary and mechanistic considerations also allows us to view them as stricto sensu HIV-1 receptors. How CCR5 or CXCR4 interact with gp120 is not known at the molecular level. The current view is that the functional interaction with HIV-1 gp120, i.e. leading to infection, is mediated by the extracellular domains of chemokine receptors, in particular their amino-terminal (NT) domain and second extracellular loop, whereas their intracellular domains coupled to the cell signaling machinery are dispensable, at least in experimental situations (reviewed in Refs. 6, 20, and 21). The ability of CCR5 and CXCR4 to interact with gp120 is certainly a critical parameter, but it can be wondered if other features could also contribute to their HIV-1 receptor activity. Discrepancies have indeed been reported between gp120 binding and efficiency of HIV-1 infection in certain experiments (22,23). Among parameters proposed to influence the HIV-1 receptor activity of CCR5 or CXCR4 are their association with CD4 (14,24,25) and their density at the cell surface (26,27), which both could be linked to localization in particular domains of the plasma membrane such as glycolipid-rich "rafts" (28). A high local density of HIV-1 receptors could favor the clustering of activated viral fusiogenic proteins that seems important for the formation of fusion pores (29 -31). In line with this view, the oligomeric status of chemokine receptors could represent an important parameter for HIV-1 infection.
The possibility that GPCRs form oligomeric complexes has been the matter of a long debate but now seems to be an accepted fact supported by a list of coimmunoprecipitation and transcomplementation experiments and more recently evidenced by biophysical techniques based on light resonance energy transfer (for reviews see Refs. [32][33][34]. There is still no consensus on the possible function(s) of such dimers or higher order oligomers of GPCRs. In most instances, the formation of such structures appears to be a constitutive process, independent from the activation of these receptors by their ligands (32). Conflicting results have been obtained in the case of chemokine receptors. In a series of studies, the formation of CCR5, CCR2, or CXCR4 homodimers as well as CCR5/CCR2 heterodimers was found to occur upon treatment of cells with the cognate chemokine ligands (35)(36)(37)(38). On the contrary, Benkirane et al. found that wild-type CCR5 could be trapped in the endoplasmic reticulum of cells expressing the truncated form of CCR5 (residues 1-187) encoded by the ⌬32 allele and proposed that formation of dimers was a constitutive process necessary for efficient processing of the receptor to the cell surface (39). Although such an effect of the ⌬32 mutant on the cell surface expression of CCR5 could not be evidenced in a recent study (40), we have reported experiments confirming the finding of Benkirane et al. (39) and suggesting that the transdominant negative effect involved interactions between membranespanning domains of the two proteins, in particular TM3 of CCR5 and TM4 of ⌬32 (41). In similar experiments, we found that the HIV-1 receptor activity of CCR5 was enhanced when cells coexpressed a functionally defective CCR5 mutant obtained by complete deletion of the amino-terminal extracellular domain. Another example of cooperation between two forms of CCR5 was obtained by rescuing HIV-1 receptor activity upon expressing two defective CCR5 mutants in transfected cells. This functional transcomplementation apparently required physical association of the different forms of CCR5 detected in coimmunoprecipitation experiments. These results provide strong and direct evidence for the formation of constitutive CCR5 oligomers and shed new light on the molecular interplay between the HIV-1 envelope proteins and the target cell membrane.
Plasmid Vectors-All WT and mutant chemokine receptors cDNAs were expressed from the cytomegalovirus immediate-early promoter using a standard calcium phosphate technique. The expression vectors for WT and c-Myc-tagged CCR5 human CCR5 (44), the HMMM humanmouse chimeric CCR5 (51), truncated CCR5 1-100 (41), and WT CXCR4 (52), have been described previously. The CCR5 ⌬NT mutant corresponding to an in-frame deletion between Asp 2 and Lys 26 was obtained by site-directed mutagenesis on a single-stranded template. PCR amplification strategies were used to obtain the FLAG-tagged forms of CCR5 (in-frame insertion of the amino acid sequence DYKDDDDK after the initiation codon) and the 5444 chimera, which has the NT domain of human CCR5 (Met 1 to Lys 26 ) and the other domains of human CXCR4 (Cys 28 to Ser 352 ). The green fluorescent protein (GFP) was expressed from the enhanced GFP-N1 vector (Clontech).
Syncytia Formation Assays and HIV-1 Infections-Infection of HeLa-P4 cells expressing different forms of CCR5 or co-cultures with Env ϩ cells were performed essentially as described previously (44). Cells were transfected in 6-well trays (5 ϫ 10 5 cells/well) with a total amount of 3 g of DNA/well. Co-cultures or infections were initiated 24 h later by adding an equivalent number of freshly trypsinized HeLa-Env/ADA cells or 10 5 infectious units of each R5 HIV-1 strains. After 24 h (or 36 h for infections), the cells were washed, fixed in 0.5% glutaraldehyde, and stained for ␤-galactosidase activity with X-gal. Blue-stained foci were scored under ϫ20 magnification. Cell counts Ͼ200 were obtained by extrapolation from randomly selected fields.
Flow Cytometry Analysis-The surface expression of chemokine receptor was monitored as described previously (52). HEK293T cells were detached with phosphate-buffered saline, 1 mM EDTA 36 h after cotransfection with GPCR and GFP expression vectors (6:1 ratio). Cells were incubated 1 h at 4°C in phosphate-buffered saline, 2% fetal calf serum with 1 g/ml of anti-CCR5 (2D7 or 3A9) or anti-CXCR4 (12G5) mAb and then washed and stained for 1 h at 4°C with phycoerythrin (PE)-conjugated secondary antibody before fixation in 4% paraformaldehyde. The green (GFP) and red (CCR5 or CXCR4) fluorescence was analyzed on an Epics Elite flow cytometer (Coultronics). Results are shown as the percentage of PE-positive cells or mean PE intensity in the GFP-positive fraction as indicated.

Cooperation of Wild-type CCR5 and a Defective Mutant-The
HIV-1 receptor activity of CCR5 was assayed by transient transfection of CD4 ϩ HeLa-P4 cells followed by infection by R5 HIV-1 strains (ADA, JRCSF, or YU2) or co-culture with HeLa-Env/ADA cells (44). Infection by HIV-1 or fusion with Env ϩ cells (also expressing the HIV-1-transactivating protein, Tat) activates a ␤-galactosidase (lacZ) transgene in HeLa-P4 cells, allowing us to quantify these events by staining with X-gal. Syncytia formation and HIV-1 infection were not detected when HeLa-P4 cells expressed the ⌬NT CCR5 mutant corresponding to a 25-residue deletion in the amino-terminal extracellular domain (Fig. 1, A and B). Unexpectedly, the cotransfection of this defective mutant and the wild-type (WT) CCR5 resulted in markedly higher numbers of syncytia with Env ϩ cells and HIV-1 infection events by comparison with cells transfected with the WT CCR5 only (Fig. 1, A and B). The efficiency of cell fusion or infection was similar for cells transfected with equal amounts (1.5 g) of WT and ⌬NT CCR5 vectors or with 3 g of the WT CCR5 vector. The cooperation observed in this experiment could either be because of functional complemen-tation of the defective mutant or enhanced expression of the WT receptor.
The cell surface expression of the two forms of CCR5 was assayed by flow cytometry after staining with the 3A9 or 2D7 mAbs, which react with the NT domain and the second extracellular loop of CCR5, respectively. Similar flow cytometry profiles were obtained for cells transfected with the WT CCR5 or the ⌬NT mutant upon staining with the 2D7 mAb, indicating that the deletion did not affect cell surface expression, whereas the 3A9 mAb only stained cells expressing WT CCR5 as expected ( Fig. 2A; Table I). An analysis of HeLa-P4 cells cotransfected with equal amounts of the WT CCR5 vector and either a control plasmid or the ⌬NT CCR5 vector revealed similar staining efficiency with the 3A9 mAb (Fig. 2B). Therefore, the enhanced HIV-1 receptor activity of cells coexpressing WT and ⌬NT CCR5 seemed to be attributed to a complementation effect.
Coexpression of CCR5 Mutants Restores HIV-1 Receptor Ac-tivity-To obtain further evidence of cooperation between two forms of CCR5, we sought to rescue HIV-1 receptor function by coexpressing the ⌬NT mutant and another defective forms of CCR5. In our hands, point mutations in the extracellular loops of CCR5 had a very limited functional effect, whereas the deletions were not compatible with transport of CCR5 to the cell surface, probably because they caused improper folding and intracellular retention of the polypeptide chain. Therefore, our option was to use a CCR5 chimera (HMMM) in which the NT and TM1 domains derived from human CCR5 (H) and other domains from its mouse homologue (M). Although such a chimeric receptor has been reported to be a functional HIV-1 receptor (51, 53), its activity was very low in our assays. It was Ͻ10% relatively to WT CCR5 in fusion assays between transfected HeLa-P4 cells and HeLa-Env/ADA cells (Fig. 3A) and in infections with the YU-2 and JRCSF HIV-1 strains (Fig. 3B), whereas infection by HIV-1 ADA was somehow more efficient (ϳ30% relatively to WT CCR5). Flow cytometry analysis with the 3A9 mAb indicates that the lower HIV-1 receptor activity of the HMMM chimera does not result from reduced surface expression (Table I and Fig. 2A). Cotransfection of CCR5 ⌬NT and HMMM expression vectors (1.5 g each) in HeLa-P4 cells resulted in efficient fusion with HeLa-Env/ADA cells and infection by the HIV-1 JRCSF and YU-2 strains, whereas infection by HIV-1 ADA strain was markedly increased (Fig. 3, A and B). The number of syncytia or infection events was comparable with those seen when HeLa-P4 cells were transfected with 1.5 g of WT CCR5, again indicating that the functional complementation of these two CCR5 mutants is a relatively efficient process. Flow cytometry analysis confirmed that the ⌬NT mutant did not up-regulate the surface expression of the HMMM receptor (Fig. 2B). In similar experiments, the CCR5 ⌬NT mutant was not rescued by coexpressing the mouse CCR5 or other chimeric receptors not containing the amino-terminal domain of human CCR5 2 or by the CCR5 1-100 mutant corresponding to the NT and TM1 domains (Fig. 3A) and previously shown not to be processed to the cell surface (41).
To find out whether complementation was possible when the NT domain of CCR5 was in the context of a more distant GPCR, it was substituted to the homologous domain of CXCR4. Flow cytometry analysis showed that the resulting 5444 chimera and WT CXCR4 were expressed at a similar level in transfected cells, whereas the reactivity with the 3A9 mAb was ϳ50% relative to WT CCR5 (Table I), indicating that the aminoterminal domain of CCR5 was accessible. Upon expression of the 5444 chimera in HeLa-P4 cells we could detect fusion with HeLa-Env/ADA cells at a relatively low level (ϳ20% relative to WT CCR5) (Fig. 3A), which is in agreement with observations made with similar chimeric receptors (54). The efficiency of fusion was not enhanced when HeLa-P4 cells coexpressed the 5444 chimeric receptor and the ⌬NT mutant (Fig. 3A). These different results suggested that the cooperation of different forms of CCR5 for HIV-1 receptor activity required some form of interplay that cannot take place with a more distant GPCR such as CXCR4. It obviously led to an envision that functional complementation involved the formation of dimers or higher order oligomers.
In cells expressing two receptors capable to efficiently associate, the relative proportions of homodimers and heterodimers can be deduced from simple equations and follow theoretical curves shown in Fig. 4A. The efficiency of fusion with Env ϩ cells and the level of CCR5 at the cell surface were in approximately a linear relationship with the amount of CCR5 vector 2 M. Chelli and M. Alizon, unpublished data.

FIG. 1. Cooperation of wild-type and defective CCR5 mutants for HIV-1 receptor activity. A, syncytium formation assays between
HeLa-P4 cells (CD4 ϩ and LTR-lacZ) transfected with indicated amounts (g) of different expression vectors and HeLa-Env/ADA cells stably expressing R5 Env. Results represent numbers of syncytia detected by X-gal staining after 24-h co-culture. B, infections of transfected HeLa-P4 cells with three different R5 HIV-1 strains (10 3 units) scored at 36 h by X-gal staining. Results are shown relative to cells transfected with 3 g of WT CCR5. Data shown are the average from three independent transfections. transfected, at least for DNA amounts ranging between 0.5 and 3 g (Fig. 4B). 2 It allowed us to use this type of assay to estimate the stoichiometry of the different forms of CCR5 engaged in a functional complex. For cells cotransfected with the WT and ⌬NT CCR5 vectors in different ratios, the fusion efficiency closely followed the theoretical curve corresponding to the addition of the two curves corresponding to relative concentrations of the putative WT/WT and WT/⌬NT dimers (Fig.  4B). When cells coexpressed the ⌬NT CCR5 and HMMM chimera, the shape of the fusion efficiency curve was similar to that of the heterodimer concentration, although it was more compact (Fig. 4C), which could indicate that only a fraction of the two forms of CCR5 could associate or that the putative complex they form is a less efficient HIV-1 receptor than WT CCR5. Nevertheless, these experiments indicated that cooperation is most efficient when the different forms of CCR5 are in a 1:1 stoichiometry. Also, an important fraction of CCR5 or derived mutant seemed capable to form functional complexes.
Detection of CCR5 Oligomers-Immunoprecipitation of CCR5 in apparent correct conformation, i.e. capable to bind gp120, seems to be critically dependent on experimental conditions, in particular anti-CCR5 antibodies and reagents used for membrane solubilization (55). Based on these indications, cells were lysed with n-dodecyl-D-maltoside and a conformationdependent mAb (2D7) used for immunoprecipitations. Using these conditions, ϳ40-kDa species was detected by SDS-PAGE and Western blot, which corresponds to the expected size of a CCR5 monomer (38 kDa) in transfected HeLa-P4 or HEK293T cells (data not shown). Monomers were also the predominant CCR5 species in the stably transfected CEM-CCR5 cell line (42), although a ϳ80-kDa band, consistent with the size of homodimers, was detected upon longer exposure and when the sample was treated with 4 M urea (Fig. 5A). To address the effect of agonist binding on oligomerization, CEM-CCR5 cells were treated with the MIP-1␤ chemokine at a relatively high concentration (100 nM). This treatment resulted in almost a complete loss of CCR5 expression at the cell surface after 30 min because of internalization. 2 It indicates efficient binding of MIP-1␤ to most receptor sites in these conditions. The treatment of cells with MIP-1␤ did not seem to modify the relative abundance of the ϳ80-kDa CCR5 species but resulted in the apparition of heavier species that could correspond to the oligomers or aggregates of CCR5 (Fig. 5A). It also resulted in a slower migration of the different CCR5 species that was particularly evident after 15 and 30 min (Fig. 5A). Such an effect was previously reported for CCR5 monomers upon treatment with another ligand (AOP-RANTES) and seems to be the result of phosphorylation of serine residues in the COOH-terminal domain of the receptor (56). The results obtained in CEM-CCR5 cells indicate that CCR5 can form dimers in the absence of chemokine activation. How-  ever, such dimers represent a minor fraction of CCR5 in these cells and were not even detected in transiently transfected cells, which was in apparent contradiction with the efficiency of functional complementation. This finding could indicate that cooperation among the different forms of CCR5 mutants does not require their physical association. However, before accepting this conclusion, we had to envision the possibility that our experimental setting was not adapted to the detection of CCR5 oligomers, for example, because of their relative instability in SDS-PAGE conditions. To circumvent this problem, we sought to address the oligomerization of CCR5 in assays based on coimmunoprecipitation.
In a first series of experiments, HEK293T cells were cotransfected with two epitope-tagged forms of CCR5 obtained by insertion of a FLAG or a c-Myc sequence at their amino termini. Immunoprecipitation of cell lysates with anti-c-Myc mAb followed by SDS-PAGE and Western blot analysis with anti-FLAG mAb allowed us to detect a ϳ40-kDa band, consistent with the size of a CCR5 monomer (Fig. 5B). This finding indicates that the FLAG-and c-Myc-CCR5 formed a complex that was disrupted during SDS-PAGE. By the same technique, it was possible to immunoprecipitate a FLAG-tagged ⌬NT mu-tant with the c-Myc-CCR5, which indicates that the aminoterminal region of CCR5 is dispensable for this association (Fig. 5B). Bands with similar intensities were detected when the same Western blot was analyzed with the anti-c-Myc mAb (Fig. 5B). Therefore, an important fraction of the c-Myc-tagged and FLAG-tagged CCR5 seemed to form stable complexes.
Physical association of the ⌬NT CCR5 and HMMM chimera was also evidenced in transfected cells by immunoprecipitation with the 2D7 mAb (only reacting with the ⌬NT CCR5) followed by Western blot with the 3A9 mAb selectively detecting the HMMM chimera (Fig. 5C). The same approach yielded negative results when cells coexpressed the ⌬NT CCR5 and the 5444 chimera (data not shown) or when immunoprecipitation was performed after mixing lysates of cells, which are independently transfected, to express the ⌬NT CCR5 and the HMMM chimera (Fig. 5C), ruling out artifactual formation of the CCR5 complexes.

DISCUSSION
Our finding that different forms of the CCR5 chemokine receptor can cooperate to fulfil the HIV-1 receptor function and form complexes detected by coimmunoprecipitation experi- ments represents further evidence for the oligomerization of this family of G-protein-coupled receptors. It also suggests that the formation of CCR5 oligomers is independent from chemokine binding and from the resulting cell activation process. Before discussing the implications of these findings for the different functions of CCR5, we shall review the oligomerization of CCR5 and other chemokine receptors.
CCR5 Oligomerization-The first evidence that chemokine receptors can physically associate was the detection of CCR5 homodimers in cells transfected with an epitope-tagged receptor (39). This process was apparently constitutive and proposed to play a role in the routing of CCR5 to the cell surface. However, in other studies, dimers of the CCR2, CCR5, or CXCR4 were detected only if cells were treated with the cognate chemokine ligands (35)(36)(37) or with a monoclonal antibody in the case of CCR5 (36). In these different studies, dimers of chemokine receptors detected by SDS-PAGE and Western blot techniques seemed to represent a small fraction relative to monomers. The formation of CCR5/CCR2 heterodimers was observed by coimmunoprecipitation and also required the activation of cells by chemokine ligands (38).
We have used these two types of techniques to investigate the physical association of CCR5 monomers and different mutant forms of CCR5. Immunoprecipitation of cell lysates followed by SDS-PAGE and Western blot only allowed us to detect CCR5 monomers in transiently transfected cells. A CCR5 species with an apparent size of a dimer could be detected in a stably transfected T-cell line (CEM-CCR5) but represented a very minor fraction of total CCR5, which was mainly monomeric. The treatment of CEM-CCR5 cells with the MIP-1␤ chemokine did not enhance the relative abundance of the putative CCR5 dimers, although it resulted in the apparition of higher molecular weight species possibly representing tetramers or aggregates. According to these results, the formation of dimers seemed to be a marginally important process in the case of CCR5 and therefore unlikely to account for the apparent efficiency of functional complementation.
A different case could be made from results obtained in coimmunoprecipitation experiments. Indeed, the association of CCR5 monomers bearing different epitope tags and of the ⌬NT mutant with the HMMM chimera was readily detected in transiently transfected cells in the absence of chemokine activation or in contact with HIV-1 envelope proteins. Coimmunoprecipitation was not observed among CCR5 monomers that were independently expressed in different cells or between CCR5 and CXCR4, although these GPCRs were found to be clustered at the membrane in different cell types (57). This finding indicates that complexes detected in these experiments were not attributed to artifactual aggregation of chemokine receptors during cell lysis. Although these experiments did not allow precise quantification, they were consistent with the view that a high fraction of CCR5 was engaged in complexes, at least when expressed by transient transfection. The coimmunoprecipitation and functional complementation assays therefore were in complete agreement, because physical association was only detected between receptors cooperating for HIV-1 receptor activity and seemed to be an efficient process. This view is also in agreement with the recent detection of constitutive CCR5 oligomers by bioluminescence resonance energy transfer. By this approach, the bioluminescence resonance energy transfer signal was not enhanced when cells were treated with CCR5 chemokine ligands and CCR5 appeared unable to form heterodimers with CXCR4, even in cells expressing relatively high levels of both receptors (58,59).
The relative instability of CCR5 complexes in SDS-PAGE and other technical issues such as reagent used for membrane solubilization (55) probably account for the discrepancies between reports with regard to abundance of dimers and their constitutive or ligand-activated nature. If assays relying upon SDS-PAGE detect only the most stable fraction of dimers or oligomers, the apparent up-regulation of these structures may not allow us to infer that agonist binding induces dimerization. As recently discussed, chemokines or other GPCR ligands could indeed bind to preexisting dimers or oligomers and thereby modify their conformation, rendering them more stable in detergent or more accessible to antibodies and hence more easily detected in certain experimental conditions (32). In the case of the glutamate receptor, there is direct crystallographic evidence that the ligand binds to preexisting dimers and induces conformation changes in its receptor (60).
Interaction of CCR5 and HIV-1-The binding of the HIV-1 envelope glycoprotein gp120 to CCR5 or CXCR4 at the surface of target cells, usually after prior contact with CD4, is considered to trigger the virus entry process. How gp120 interacts with chemokine receptors is not known in molecular details, although indirect elements suggest that the amino-terminal domain and second loop of CCR5 and CXCR4 are involved as well as the third hypervariable domain (V3) and a pocket formed by more inner and conserved domains of gp120 (12,61). Our finding that a defective form of CCR5 attributed to the ⌬NT was rescued by a chimeric receptor bearing this domain suggests that gp120 must engage distinct contacts with two sites in CCR5, one being entirely located in the amino-terminal domain and the other formed by the extracellular loops. A similar model is usually envisioned for the interaction of chemokines and chemoattractants with their receptors (62).
Complementation for HIV-1 receptor activity only occurred between chemokine receptors capable to form complexes detected in coimmunoprecipitation experiments. This finding suggests that functional interaction of HIV-1 with the different domains of CCR5 has spatial constraints and requires the amino-terminal domain and loops to be in close vicinity. Whether these domains of CCR5 reconstitute a single gp120 binding site or engage contact with two gp120 subunits of a trimeric Env complex cannot be decided from our results. Assaying the binding of recombinant monomeric gp120 to cells coexpressing the ⌬NT CCR5 mutant and HMMM chimera could help clarify this issue. Of note, complementation experiments have shown the possibility of cooperative subunit interactions within HIV-1 Env (63).
The finding that complexes of defective CCR5 mutants mediate HIV-1 infection can appear to be contradictory with conclusions of a study linking the antiviral activity of an anti-CCR5 mAb (designated CCR5-02) to the dimerization of CCR5 that it apparently induced (36). Because the CCR5-02 mAb did not interfere with the surface expression and chemokine receptor activity of CCR5, the authors inferred that its conformation in the context of dimers does not permit a functional interaction with HIV-1. However, this mechanism seems difficult to reconcile with apparently normal gp120 binding in presence of CCR5-02 mAb. Also, the antiviral activity was monitored at day 7 after infection after several replicative cycles. Indirect effects of the mAb on other steps of the virus cycle cannot therefore be ruled out.
On the contrary, our results seem to be fully consistent with the observation that several CCR5 receptors, probably 4 -6, must cooperate to mediate HIV-1 infection (27). This conclusion was based on the shape of curves depicting efficiency of infection relative to the amount of CCR5 at the cell surface. An analysis of membrane fusion mediated by the hemagglutinin of influenza virus has shown that several hemagglutinin trimers must cooperate to form a fusion pore (29,30,64). In line with this view, the ability of HIV-1 to engage a functional interaction with different domains of a receptor prone to oligomerization can represent an advantage in terms of efficiency of infection. It can indeed favor the activation of a sufficient number of gp41 in the area of virus-cell contact to enable membrane fusion.