Determinants of the trans-dominant negative effect of truncated forms of the CCR5 chemokine receptor.

The human immunodeficiency virus, type 1 (HIV-1) entry process is triggered by interaction between the viral envelope and a seven membrane-spanning domain receptor at the cell surface, usually the CCR5 chemokine receptor. Different naturally occurring mutations in the CCR5 gene abolish receptor function, the most frequent being a 32-nucleotide deletion resulting in a truncated protein (Delta32) lacking the last three transmembrane domains (TM5-7). This mutant is retained in the endoplasmic reticulum and exerts a trans-dominant negative (TDN) effect on the wild type, preventing its exit from this compartment. This TDN effect is often considered as evidence for the oligomerization of CCR5 during transport to the cell surface. Here we use a genetic approach to define the structural determinants of the TDN effect of the Delta32 mutant. It was abolished by certain deletions and by mutations of cysteine residues preventing formation of a disulfide link between the first and second extracellular loops, suggesting that conformation of Delta32 is important for its interaction with CCR5. To circumvent this problem, we used chimeric forms of the Delta32 and wild type CCR5, consisting in substitutions with homologous domains from the mouse CCR5. All chimeric full-length receptors were expressed at the cell surface and were functional for interaction with HIV-1 or with a chemokine ligand, when assayed. The TDN effect was only observed if both the TM3 domain in CCR5 and the TM4 domain in Delta32 were from human origin, whereas the rest of the proteins could be from either origin. This suggests that the TDN effect involves some form of interaction between these transmembrane domains. Alternatively, but less likely to us, substitutions in TM4 could affect the conformation of CCR5 in the endoplasmic reticulum but not at the cell surface. However that may be, it seems that the TDN effect of the Delta32 mutant has no bearing to the issue of CCR5 dimerization and to its possible role in the processing of the receptor to the cell surface.

coupled to heterotrimeric G-proteins (reviewed in Refs. [1][2][3]. Most chemokines can be classified in the CC or CXC subgroups depending upon the relative position of two conserved cysteines in their amino-terminal region, and the same terminology (CCR or CXCR) is used for their receptors. The presence of chemokine receptors at the surface of cells allows their infection by the human immunodeficiency virus, type 1 (HIV-1) or by related retroviruses. At least 10 chemokine receptors can confer permissivity to HIV-1 under experimental conditions (4), but only two, CCR5 and CXCR4, seem to be used by HIV-1 in vivo (reviewed in Refs. [5][6][7]. The interaction of CCR5 or CXCR4 with the HIV-1 surface envelope glycoprotein gp120 usually occurs after a prior contact of gp120 with another cell surface protein, CD4, leading to a view of chemokine receptors as CD4-associated HIV-1 co-receptors. The selectivity of HIV-1 strains for CCR5 or for CXCR4 determines biological properties, in particular cell tropism, and could also influence the evolution of infection in vivo. Indeed, viral strains able to use CXCR4 (termed X4) or both CCR5 and CXCR4 (R5X4 strains) usually emerge at later stages of infection, whereas strictly CCR5-dependent strains (R5) are present throughout infection (7).
Genetic defects resulting in the absence of CCR5 expression are relatively frequent and apparently without pathological consequences (8,9). Infection by HIV-1 is usually not seen in CCR5-negative individuals, although exceptions have been reported (10 -12). The most frequent mutation in the CCR5 gene associated with a loss of function is a 32-nucleotide deletion (⌬32) in the open reading frame yielding a truncated protein with only four transmembrane domains (8,9,13). The level of CCR5 expression at the cell surface was found to be highly reduced in the blood lymphocytes of individuals heterozygous for the CCR5 ⌬32 allele (⌬32/ϩ) (14,15) and in cells co-transfected with expression vectors for the wild type (WT) CCR5 and the ⌬32 mutant (9,16). The ⌬32 mutant is not processed to the cell surface and accumulates with the WT receptor in a compartment likely to be the endoplasmic reticulum (16). The trans-inhibitory effect of the ⌬32 mutant on the expression of WT CCR5 was proposed to result from the formation of heterodimers unable to undergo normal processing to the cell surface. This could also represent an indirect argument for the ability of CCR5 to form homo-oligomers and for the role of this process in the routing of the receptor (16,17), because it is the case for a number of other cell surface proteins (18).
There is growing evidence that 7TM receptors can undergo oligomerization, but the relevance of this process to their natural function, i.e. ligand binding and signal transduction, is unclear (19 -21). Oligomers of 7TM receptor, usually dimers, have been directly detected by co-precipitation usually after stimulation of cells with the cognate ligand or indirectly for example by energy transfer techniques (22). The dimerization of 7TM receptors is also suggested by the trans-complementation of two defective mutant (23)(24)(25)(26) and by the trans-inhibitory effect of certain mutants (27)(28)(29). In the case of the chemokine receptors CCR2, CXCR4, and CCR5, dimerization has been directly observed in cells stimulated by specific ligands, either chemokines or antibodies (30 -35). Spontaneously formed CXCR4 dimers have also been observed in macrophages and proposed to play a role in the resistance of these cells to infection by X4 HIV-1 strains (36).
Here we have used a genetic approach to study the transinhibitory effect of the ⌬32 CCR5 mutant on the expression of the WT receptor. Our aim was to define the structural requirements for this effect and hence for the interaction of these forms of CCR5. Determinants for this interaction were located in different TM domains of the mutant and the WT receptor. A variant of CCR5 fully resistant to the trans-inhibitory effect was apparently processed normally to the cell surface and was functional for interaction with a chemokine ligand. This suggests that the interaction between the ⌬32 mutant and the WT CCR5 does not mimic the formation of CCR5 homo-oligomers or that CCR5 oligomerization is not required for processing to the cell surface.
Plasmid Vectors-All WT and mutant chemokine receptors cDNAs were subcloned downstream to the cytomegalovirus immediate-early promoter. Site-directed mutagenesis (42) was performed on a singlestranded template corresponding to the human or the mouse CCR5 cDNA using oligonucleotides containing distinctive restriction sites (sequences are available upon request). Mutants were checked by sequencing the CCR5 open reading frame. The ⌬32 deletion creates a frameshift after Tyr 187 in the second extracellular loop (ECL2) with addition of 31 amino acids irrelevant to the CCR5 sequence. Mutants 1-100, 1-127, 1-176, and 1-183 corresponding to the corresponding amino acids of human CCR5 were obtained by insertion of a termination codon. Mutants ⌬32⌬NT (amino-terminal deletion) and ⌬32⌬TM1 correspond to deletions engineered in the CCR5 ⌬32 context between Asp 2 and Lys 26 and between Gln 27 and Asn 56 , respectively. The ⌬32 mutants C101A and C178A correspond to alanine substitutions of the cysteine residues in ECL1(C101) and/or ECL2 (C178). Mutants 1-127/C101A and 1-176/ C101A were obtained by substituting a blunt-ended fragment from C101A into the 1-127 and 1-176 mutants. Chimeric receptors B, C, E, and F corresponding to reciprocal exchanges between the human and mouse CCR5 have been described (43). Other constructs (I, J, and K) were derived by ligating PCR fragments using a SacII restriction site created in ECL1. The M⌬32 mutant corresponds to residues 1-184 of the mouse CCR5, plus 31 irrelevant amino acids at the C terminus. The ⌬32.A-F constructs were obtained by ligation of PCR fragments from the 5Ј part of previously described chimeric CCR5s and from the 3Ј part of ⌬32 (junction at residue 184 in ECL2). The ⌬32.G and ⌬32.H constructs correspond to reciprocal substitutions between ⌬32 and M⌬32 involving residues Val 146 to Cys 178 . The FLAG epitope (DYKDDDDK) was inserted at the amino terminus of several mutants by PCR amplification using a 5Ј oligonucleotide encoding the FLAG sequence. The green fluorescent protein (GFP) was expressed from the EGFP-N1 vector (CLONTECH).
Detection of CCR5 Expression-Flow cytometric analysis of CCR5 expression was performed as described (44). Briefly, HeLa-P4 cells in 6-well trays (5 ϫ 10 5 cells/well) were transfected by a standard calcium phosphate method with plasmid vectors encoding GFP and WT or mutant CCR5 (the ratio of GFP vector to total transfected DNA was 1:6). The cells were detached with phosphate-buffered saline (PBS) containing 1 mM EDTA 36 h after transfection, stained for 1 h at 4°C with the 2D7 anti-CCR5 mAb (10 g/ml in PBS with 2% fetal calf serum), then stained with phycoerythrin-conjugated secondary Ab, fixed in 4% paraformaldehyde, and analyzed on an Epics Elite flow cytometer (Coultronics). Expression of CCR5 (red fluorescence) was assessed among GFP-positive cells. For confocal microscopy analysis, HeLa-P4 cells were grown and transfected on 1-cm glass coverslips. After 36 h, the cells were fixed in PBS, 1% paraformaldehyde (15 min), quenched in PBS, 0.1 M glycine (15 min) and permeabilized in PBS, 0.05% saponin, 0.2% bovine serum albumin (40 min) at room temperature. The cells were then incubated for 1 h at room temperature with the 2D7 mAb (10 g/ml), then incubated for 1 h with fluorescein isothiocyanate-coupled secondary Ab anti-mouse Ig, then mounted in 100 mg/ml mowiol, a powder used for the experiment, (Calbiochem), 30% (w/v) glycerol, 100 mM Tris-HCl (pH 8.5), and examined under a confocal microscope (model MRC-1024; Bio-Rad). Immunoprecipitations of epitope-tagged (FLAG) CCR5 mutants were performed in HEK 293T cells 48 h after transfection. Approximately 10 7 cells were lysed for 1 h at 4°C in 1 ml of 0.5% n-dodecyl-D-maltoside (Sigma), 25 mM Tris-HCl (pH 7.4), 140 mM NaCl, 2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride (Sigma), suppplemented with protease and phosphatase inhibitors. Lysates (1 ml) clarified by centrifugation at 12,000 ϫ g for 15 min were left in contact 4 h with 40 l of agarose-conjugated M2 (anti-FLAG) mAb (Sigma) at 4°C. Bound proteins were separated by SDS/ PAGE and transferred to a polyvinylidene difluoride membrane (Polyscreen; PerkinElmer Life Sciences) for Western blot analysis using the M2 mAb (1 g/ml) and a peroxidase-coupled anti-mouse antibody (0.5 g/ml). The reaction was revealed with the enhanced chemiluminescence detection system (Amersham Pharmacia Biotech).
CCR5 Endocytosis-Down-regulation of CCR5 cell surface expression after contact with chemokine ligand was assessed as described (46). Briefly, 10 6 transfected HEK 293T cells were detached in PBS-EDTA, pelleted, and resuspended in 100 l of PBS, 2% fetal calf serum with or without 10 nM MIP-1␤. After 30 min at 37°C, the cells were transferred to ice, washed with a buffer containing 50 mM glycine and 100 mM NaCl (pH 3.0), and then processed for analysis of CCR5 expression by flow cytometry, as previously described.

RESULTS AND DISCUSSION
trans-Inhibition of CCR5 Expression-The expression of CCR5 at the surface of transfected HeLa-P4 cells was analyzed by flow cytometry after staining with the 2D7 mAb directed against a conformational epitope formed at least in part by the second extracellular loop (ECL2) of the receptor (41). As expected, the fraction of stained cells and mean fluorescence intensity were markedly reduced when cells were co-transfected with expression vectors for the WT CCR5 and the ⌬32 mutant in equal amounts, relative to cells transfected with the WT CCR5 vector and a control plasmid (Fig. 1A). Similar results were obtained using other cell types (COS, HEK, and U373MG-CD4 cells) or when cells were stained with the 3A9 mAb, which detects a conformation-independent epitope in the amino-terminal (NT) extracellular domain of CCR5 (data not shown). When cells co-transfected with the WT CCR5 and ⌬32 vectors were stained with the 2D7 mAb, confocal microscopy revealed accumulation of CCR5 in a perinuclear compartment probably corresponding to the endoplasmic reticulum (Fig. 1B), in agreement with results obtained by Benkirane et al. (16).
The level of expression of CCR5 at the surface of cells can be indirectly assessed from their relative permissivity to HIV-1 infection or to formation of syncytia with cells expressing the HIV-1 envelope proteins (Env). Fusion of U373MG-CD4 human astroglioma cells transfected with CCR5 vectors and HeLa-Env/ADA cells (R5 strain) can be readily detected and quantitated by means of a simple ␤-galactosidase assay (40). The efficiency of fusion was markedly lower for U373MG-CD4 cells co-transfected with WT and ⌬32 CCR5 vectors (1:1) relative to cells transfected with the same amount of the WT vector (ϳ35%; Fig. 2A). There was no detectable fusion when cells were transfected only with the ⌬32 vector. Also, co-transfection with this vector had no apparent effect on the expression of the CXCR4 chemokine receptor measured by its HIV-1 co-receptor activity (Fig. 2B) or on the surface expression of a number of other markers, including CD4 (data not shown). These results seem to rule out an effect of the ⌬32 mutant on the machinery responsible for the routing of proteins to the cell surface.
The TDN effect of the ⌬32 mutant on the HIV-1 co-receptor activity of CCR5 could be observed using different ratios between transfected expression vectors (Fig. 2C), even when the WT CCR5 plasmid was in excess (5:1). Such an assay can be used to predict the stoichiometry of the interaction between a functional protein and a nonfunctional mutant, assuming a similar efficiency of expression for both plasmids, and assuming that the heterodimers are not functional (47,48). With these assumptions, the co-receptor activity curve was very close to the profile predicted in the case of a 1:1 interaction between the WT CCR5 and ⌬32 (Fig. 2C). Overall, these initial experiments confirmed the view that the ⌬32 mutant and the WT CCR5 form complexes unable to reach the plasma membrane.
Requirements in ⌬32 CCR5 for the TDN Effect-We next engineered a series of deletions in the ⌬32 mutant and assessed their effects on the surface expression of CCR5, judged from the ability of cells to engage fusion with Env ϩ cells (Fig. 3). There was a similar decrease in the efficiency of cell fusion when U373MG-CD4 cells were co-transfected with expression vectors for WT CCR5 and either the ⌬32 or the 1-184 mutant (1:3 ratio), indicating that the 31 carboxyl-terminal residues irrelevant to the CCR5 sequence do not contribute to the TDN effect of the ⌬32 mutant. Also, the TDN activity was not affected by the complete deletion of the NT domain (residues 2-26) and was therefore apparently dispensable to the interaction with WT CCR5. Using a yeast two-hybrid selection assay, Benkirane et al. (16) found that the complete CCR5 (residues 1-352) interacted with itself or with a ⌬32 mutant (residues 1-187) but not with an NT-deleted CCR5 (residues 58 -352) and concluded that the NT domain participated in the interaction. The apparent discrepancy between these results and ours likely results from the use of very different experimental settings. The type of interaction that occurs in yeast between forms of The in-frame deletion of the TM1 domain (deletion of residues 27-56) reduced but did not abolish the TDN effect of the ⌬32 mutant, which was the case upon truncation at the level of ECL1 (mutant 1-100), TM3 (mutant 1-127), or ECL2 (mutant 1-176). The result seen with the last mutant was unexpected, because it only differed from the 1-184 mutant by a few residues in an extracellular loop. To rule out the possibility that the lack of TDN effect was due to inefficient expression, HEK 293T cells were transfected with epitope-tagged forms of the ⌬32 mutant and derivatives, corresponding to in-frame insertion of a FLAG sequence at their N terminus. Immunoprecipitations with anti-FLAG antibodies showed a similar level of expression for the ⌬32, 1-178, and 1-100 CCR5 mutants in cell lysates (Fig. 4). We were therefore led to consider that the integrity of the ECL2, or in fact its initial nine residues, was required for the TDN effect of the ⌬32 mutant.
The first and second extracellular loops of 7TM receptors contain conserved cysteine residues (Cys 101 and Cys 178 in the case of CCR5) capable of engaging a disulfide bridge (49,50). We envisioned that the absence of this bridge could be detrimental to the TDN effect of the 1-178 mutant and other truncated forms of CCR5, for example by impairing their spatial structure. The mutation of either the ECL1 cysteine (C101A) or the ECL2 cysteine (C178A) into alanine was indeed sufficient to abolish the TDN activity of ⌬32 (Fig. 3). Interestingly, a TDN effect similar to that of the ⌬32 mutant was observed when both mutations were present simultaneously, indicating that neither the cysteine residues nor the putative disulfide link were directly involved. Likewise, the mutation of both cysteine residues in the CCR5 context was compatible with its expression at the cell surface and HIV-1 co-receptor activity (51). The C101A mutation was sufficient to restore the TDN activity of the 1-176 mutant (Fig. 3), indicating that the ECL2 residues present in the ⌬32 mutant (177-184) were not directly required for the TDN effect. These results suggest that the presence of a single cysteine, either in ECL1 or in ECL2, in the context of Human astroglioma cells U373MG-CD4 (LTR-lacZ) were co-transfected with CCR5 (A) or CXCR4 expression vectors (B) and either ⌬32 or control vectors (1:1 ratio), or with the ⌬32 vector alone, as indicated, and syncytia formation assays were performed with HeLa cells expressing envelope proteins from the R5 HIV-1 strain ADA (A) or the X4 HIV-1 strain LAI (B). The bars represent the mean number of ␤-galactosidase-positive foci, indicating cell fusion events, per well (6-well plates). C, same experiment as in A except that HeLa-P4 cells (LTR-lacZ) were co-transfected with CCR5 and either ⌬32 or pCDNA3 vectors in the indicated ratios. The total amount of transfected DNA was 3 g/well, and the amount of CCR5 plasmid varied from 0.5 to 3 g. The results (means of five independent experiments with standard error) are shown as percentages of fusion efficiency, relative to HeLa-P4 cells transfected with the CCR5 vector alone. The dotted line represents the predicted activity in case of a 1:1 stoichiometry of interaction between WT CCR5 and ⌬32 mutant, assuming that WT/⌬32 dimers are not functional (47) .   FIG. 3. Derivatives of the ⌬32 CCR5 and their effect on CCR5 activity. A, schematic representation of the ⌬32 and other CCR5 deletion mutants and derived mutants corresponding to alanine substitution of cysteines in ECL1 (C101A) or ECL2 (C178A). The lightly shaded box at the C terminus corresponds to 31 amino acids irrelevant to the CCR5 sequence. The disulfide bridge predicted to form between cysteine in the ECL-1 and ECL-2 is represented. SH indicates the ECL1 cysteine, and A indicates its mutation into alanine. B, trans-inhibitory effect on CCR5 expression was measured as described for Fig. 2C by testing fusion of HeLa-Env/ADA cells and HeLa-P4 cells co-transfected with expression vectors for WT CCR5 and indicated mutant (1:3 ratio). The fusion efficiency (means of three experiments) is shown as the percentage relative to cells transfected with WT CCR5 vector and pCDNA3.
⌬32 and other CCR5 mutants indirectly affects the TDN activity, most likely by modifying the conformation of the protein. A possible mechanism could be the engagement of Cys 101 (or Cys 178 ) into inadequate disulfide links, either intrachain with Cys 20 in the NT domain or with other protein chains. The C101A mutation did not restore TDN activity for the 1-126 mutant (Fig. 3). It shows that residues 127-176, comprising in particular the TM3 and TM4 domains, directly contribute to the TDN activity of the ⌬32 mutant and therefore to its interaction with CCR5.
We sought to further define the interface of interaction by assaying the TDN activity of chimeric proteins corresponding to reciprocal substitutions between the ⌬32 mutant and the mouse CCR5. We indeed observed that co-expression with a mouse CCR5 mutant equivalent to ⌬32 (frameshift in ECL2 downstream to the conserved cysteine) had no inhibitory effect on the HIV-1 co-receptor activity of human CCR5 (Fig. 5) or its cell surface expression measured by flow cytometry (data not shown). This mouse CCR5 mutant (M⌬32) seems therefore unable to interact with the WT human CCR5, despite a relatively high level of sequence identity (83%). An epitope-tagged form of M⌬32 was readily detected by immunoprecipitation (Fig. 4). Chimeric human/mouse forms of ⌬32 were co-expressed with WT human CCR5 in HeLa-P4 cells and fusion assays performed with Env ϩ cells (Fig. 5). There was no inhibitory effect for chimeras ⌬32.A, ⌬32.B, and ⌬32.C, and there was the same high efficiency of fusion with the latter two chimeras. In contrast, there was a reduction in fusion efficiency similar to that induced by the human ⌬32 for chimeras ⌬32.D, ⌬32.E, and ⌬32.F, indicating that sequence differences between human and mouse ⌬32 in the region corresponding to residues 1-127 (TM1-3) have no apparent role. Consequently, residues 128 -176 of human ⌬32, containing TM4 and the adjacent intracellular (i2) and extracellular loops (ECL2) seem to be involved in the interaction with CCR5. The analysis of this region was refined, and the minimal region of human CCR5 sufficient to confer TDN activity in the M⌬32 context was contained in 25 residues from TM4 and 4 residues from ECL2 (chimera ⌬32.H). The reciprocal substitution (chimera ⌬32.G) was sufficient to suppress the inhibitory activity of human ⌬32. Epitope-tagged forms of the ⌬32.G and ⌬32.H constructs were expressed at similar levels (Fig. 4). There are only four differences in TM4 and ECL2 supporting the phenotype difference between M⌬32 and the ⌬32.H chimera (Fig. 5). Replacing any of these residues in the M⌬32 context by the corresponding human CCR5 residue was not sufficient to restore a TDN effect (data not shown). Overall, these experiments suggest that the TM4 domain of human ⌬32 has a central role in the interaction with CCR5, explaining the lack of TDN activity of mutants 1-126 and 1-100.
Structural Requirements in CCR5-A similar strategy was used to define the interface of interaction in human CCR5. The mouse CCR5 cannot mediate fusion with Env ϩ cells, but chimeric human/mouse CCR5s bearing at least one extracellular domain from human CCR5 are endowed with such activity (43,52). The ability of CCR5 chimeras to interact with the ⌬32 mutant could therefore be functionally tested (Fig. 6). Chimera C in which ECL1 was the only extracellular domain derived from human CCR5 was less efficient at mediating cell fusion, but the number of syncytia was sufficient for a valid assay. Comparison of the results obtained with the B and F chimeras showed that the amino-terminal part of human CCR5, comprising in particular the TM1-3 domains, was required for a functional interaction with ⌬32, whereas the rest of the receptor could be either from human or from mouse CCR5. The analysis of results obtained with chimeras C, E, and K allowed us to rule out a role for differences between human and mouse CCR5 in the TM1, TM2, and adjacent domains (NT, ECL1, and i1). The sensitivity to the TDN effect observed for the I but not the F chimeric CCR5 indicated the importance of the TM3 domain. Conversely, the substitution of the mouse TM3 in the human CCR5 rendered chimera K resistant to the inhibitory effect of ⌬32 (fusion efficiency was 89%). The TM3 domain of CCR5 could therefore be part of the interface between this receptor and the ⌬32 mutant. This domain is relatively divergent, with FIG. 5. Domain substitutions in ⌬32 and their effect on the CCR5 activity. The truncated form of the mouse CCR5 (M⌬32) corresponds to residues 1-184 followed by 31 irrelevant residues (hatched box). Chimeric proteins ⌬32.A to ⌬32.H were obtained by substitution of homologous domains from mouse CCR5 in the ⌬32 context. The experiment was performed, and the results are presented as for Fig. 3. In the alignment of the CCR5 sequences, dashes represent identical residues. eight differences between the human and murine CCR5 sequences. In particular, we note the absence in the mouse CCR5 of the LXXXGXXXG motif proposed to play a role in the oligomerization of other receptors (53). We note that the divergence between human and mouse CCR5 in TM3 had no apparent role in the lack of interaction of M⌬32 with human CCR5, which confers further validity to our experiments with ⌬32 chimeras. Overall, these results are consistent with the view that the ⌬32/CCR5 interaction occurs through contact of nonhomologous transmembrane domains (TM3 in CCR5 and TM4 in ⌬32). Similar observations have been made for the oligomerization of other 7TM receptors (20).
An alternative possibility is that the conformational forms of chimeric CCR5 differ from that of the wild type receptor, thereby preventing interaction with the ⌬32 mutant in the endoplasmic reticulum but allowing exit from the compartment and transport to the cell surface. This possibility would be difficult to confirm, and we are not aware of comparable findings for other 7TM receptors. On the contrary, mutations that impair protein folding are well known to prevent their exit from the endoplasmic reticulum (54 -56), even if examples of pharmacological rescue have been (57,58). Thus, the conformational effects should selectively occur for chimeric CCR5 that contain the mouse TM3 domain, which as previously noted, has no apparent effect in the chimeric forms of the ⌬32 mutant.
Surface Expression and Function of a ⌬32-resistant CCR5 Mutant-The minimal mutant form of CCR5 fully resistant to the inhibitory activity of ⌬32 (chimera K) was studied in more detail for its processing and function. Flow cytometric analysis of transfected cells showed that it was expressed at the same level as the WT CCR5 at the cell surface (Fig. 7), which was consistent with the results obtained in cell fusion assays (Fig.   FIG. 6. Domain substitutions in CCR5 and sensitivity to ⌬32. A, schematic representation of chimeric receptors with dark and light shading for sequences derived from human CCR5 and mouse CCR5, respectively. When possible, the same nomenclature was used as in Fig. 5 (B, C, E, and F chimeras). B, fusion efficiency between HeLa-Env/ADA cells and HeLa-P4 cells transfected with the indicated CCR5 vector and either pCDNA3 (black boxes) or the ⌬32 vector (shaded boxes). The experiment was performed and presented as in Fig. 2A, except that the ratio of CCR5 to ⌬32 (or pCDNA3) was 1:3. 6). Treatment of cells expressing either the K chimera or the WT CCR5 with the MIP-1␤ chemokine resulted in similar down-regulation of cell surface expression (Fig. 7), indicating that both forms of CCR5 were functional in terms of ligand binding and interplay with the receptor endocytosis machinery (59). It seems, therefore, that the ability of CCR5 to interact with ⌬32 is independent of its ability to be transported to the cell surface in a functional form. Either CCR5 does not require oligomerization for normal processing or the formation of CCR5 oligomers involves a type of interaction different from the CCR5/⌬32 interaction. Our experiments do not allow us to distinguish between these possibilities. However, our results suggest that the interaction of the ⌬32 mutant with CCR5 does not relate to "natural" oligomerization of 7TM receptors but rather to the inhibition of their processing and/or function by peptides corresponding to their membrane-spanning domains (53,60,61). These observations strengthen the view that the tridimensional barrel-like structure of 7TM receptors, which must involve interactions between membrane-spanning domains, is relatively open and flexible (41). Disruption of this structure by agents mimicking the activity of peptides or truncated receptors could be a means of pharmacological intervention.