The Amino-terminal Extracellular Domain of the MCP-1 Receptor, but Not the RANTES/MIP-1α Receptor, Confers Chemokine Selectivity

The chemoattractant cytokines, MCP-1 (monocyte chemoattractant protein) and MIP-1α (macrophage inflammatory protein), are recognized by highly homologous but distinct receptors. To identify receptor domains involved in determining ligand specificity, we created a series of chimeric MCP-1 and RANTES (regulated on activation, normal T cell expressed and secreted)/MIP-1α receptors that progressively interchanged the amino terminus and each of the three extracellular loops. Radiolabeled MCP-1 bound with high affinity to the wild-type MCP-1 receptor, but not to the RANTES/MIP-1α receptor (C-C CKR-1). Chimeras that retained the amino-terminal extension of the MCP-1 receptor bound MCP-1 with high affinity. In contrast, chimeric MCP-1 receptors, in which the wild-type amino terminus was replaced with the corresponding portion of the RANTES/MIP-1α receptor, bound MCP-1 with low affinity. These data indicate that the amino terminus of the MCP-1 receptor is necessary for high affinity binding of the ligand. Very different results were obtained using the RANTES/MIP-1α receptor. Radiolabeled MIP-1α bound with high affinity to chimeras that expressed the extracellular loops of the RANTES/MIP-1α receptor. In contrast to the MCP-1 receptor, substitution of the wild-type amino-terminal extension had little or no effect on MIP-1α binding. For the MCP-1, but not the RANTES/MIP-1α receptor, the presence of the wild-type amino terminus also significantly lowered the ligand concentration required for maximal signaling. We conclude that the amino-terminal extension of the MCP-1 receptor, but not the RANTES/MIP-1α receptor, is critically involved in ligand binding and signal transduction. These data reveal significant functional differences between the two C-C chemokine receptors and suggest a two-step mechanism for activation of the MCP-1 receptor.

Leukocyte chemotaxis to sites of inflammation and infection is initiated by the interaction of cytokines with G proteincoupled receptors (see Refs. 1-3 for recent reviews). A number of potent chemotactic cytokines have been identified including the activation peptide from the fifth component of complement (C5a) 1 (4), formyl peptides (5), and a family of chemotactic peptides known as the chemokines. Chemokines are 8 -10-kDa basic heparin-binding proteins that are related by both primary structure and the presence of 4 conserved cysteine residues. In addition to chemotaxis, chemokines mediate leukocyte degranulation (6) and the up-regulation of adhesion receptors (7), and have recently been implicated in the suppression of human immunodeficiency virus replication (8). MCP-1 is a member of the C-C branch of the chemokine family in which the first 2 cysteine residues are adjacent. Other C-C chemokine members include RANTES (regulated on activation, normal T cell expressed and secreted), macrophage inflammatory protein 1␣ and 1␤ (MIP-1␣, MIP-1␤), and the recently described eosinophil agonist, eotaxin (9). Interleukin-8 (IL-8) is a member of the C-X-C branch in which a single residue separates the first 2 cysteines, and lymphotaxin is a recently described chemokine in which only 2 cysteines are present (10) .
Two receptors for MCP-1 (MCP-1RA and MCP-1RB) have been cloned and are alternatively spliced variants of a single gene (11). A single receptor that binds both MIP-1␣ and RAN-TES (RANTES/MIP-1␣ receptor, also known as C-C CKR-1) has been cloned (12,13), and there are recent reports of two additional C-C chemokine receptors that are highly expressed in eosinophils (14) and basophils (15), respectively. Each of these receptors is a seven-transmembrane domain receptor (16). The degree of identity between members of this superfamily of G protein-coupled seven-transmembrane domain receptors is 45-70%, with the greatest similarity occurring in the transmembrane domains, and the greatest divergence in the amino-terminal extension and three extracellular loops.
The mechanism(s) of ligand binding and subsequent receptor activation of seven-transmembrane segment receptors are quite varied and appear to depend, in large part, on the size of the ligand (reviewed in Ref. 17). Small molecules such as retinal (18) and catecholamines (19) are believed to interact with several charged residues in a hydrophilic pocket formed by the transmembrane domains of the receptor. In contrast, large glycoprotein hormones, such as thyroid-stimulating hormone (20) and leutinizing hormone (21,22), bind with high affinity to long (350 -400 residues) amino-terminal extensions of their respective receptors. In the case of small peptides (opiates (23), endothelin (24), and the agonist peptide of the thrombin recep-tor (25), as well as medium-sized peptides (e.g. neuropeptide Y (26)), important interactions with the extracellular loops have been identified. The mechanism of receptor activation by the chemokines is not well understood, but studies of the binding of IL-8 (27)(28)(29) and C5a (30,31) have implicated the aminoterminal extension and one or more extracellular loops as directly interacting with the ligands.
We have investigated the sites of chemokine binding and activation of the MCP-1 and the RANTES/MIP-1␣ receptors. Taking advantage of the high degree of amino acid sequence conservation between these two receptors, we have created a series of chimeras in which the amino-terminal extension, and each of the three extracellular loops (including the transmembrane domains) of the RANTES/MIP-1␣ receptor, were progressively substituted for the corresponding region of the MCP-1 receptor, and vice versa. In this paper, we report that the amino terminus of the MCP-1 receptor is necessary for high affinity binding of 125 I-labeled MCP-1, but that other regions of the receptor are required to mediate signal transduction. In contrast, binding and signal transduction induced by MIP-1␣ do not appear to require the amino terminus, but do require the presence of the third extracellular loop of the RANTES/MIP-1␣ receptor. These data suggest a two-step model in which high affinity binding of MCP-1 by the receptor amino terminus allows subsequent low affinity interactions with the extracellular loops/transmembrane domain bundles to effect receptor activation and signaling.

Construction of Chimeric
Receptors-Receptor chimeras were constructed from cDNAs of MCP-1R (Type B; Ref. 11) and the RANTES/ MIP-1␣ receptor (12), and nucleotide (nt) numbers correspond to Gen-Bank accession numbers U03905 and L10918, respectively. We chose to study the Type B, rather than the Type A receptor, because the latter is poorly expressed in HEK-293 cells. 2 Each receptor contains the prolactin signal sequence, followed by the Flag epitope sequence (32) fused immediately upstream of the second translation codon (nt 84, MCP-1R; and nt 66, C-C CKR-1). The prolactin signal/Flag sequence was subcloned into pBluescript and the receptor cDNA fused to the Flag sequence at a SalI site. A HindIII/NotI fragment of the MCP-1RB and RANTES/MIP-1␣R constructs containing the leader sequence was then subcloned into the corresponding sites of the expression vector pcDNA3 (Invitrogen, San Diego, CA) to yield the tagged wild-type receptors. The designations MMMM (MCP-1RB) and RRRR (RANTES/MIP-1␣R) are used to represent the amino-terminal extension and each of the three extracellular loops of each receptor and chimera. Chimeras 1 and 2 were sequenced at the junctions used to create them by subcloning, and chimeras 3-6 were sequenced throughout the entire coding region to rule out mutations due to PCR amplification.
Chimeras 1 and 2-Receptors with the amino-terminal extension exchanged were created by using a conserved ApaI site in RRRR and MMMM. An ApaI digest in MMMM and a partial digest in RRRR each yielded approximately a 1.8-kb fragment that was then subcloned into the ApaI site in RRRR and MMMM, respectively to yield RMMM and MRRR. Thus, in RMMM amino acids 1-32 are from the RANTES/MIP-1␣R, while in MRRR amino acids 1-40 are from the MCP-1R. Amino acid numbers are from Ref. 11.
Chimeras 3 and 4 -Receptors that exchange the amino terminus and first extracellular loop was created by "overlapping PCR" (33). The complementary primers A (5Ј-agatacctggctattgtccacgc-3Ј) and B (5Јgcgtggacaatagccaggtatct-3Ј) correspond to the conserved sequence RY-LAIVHA, nt 492-514 and 453-476 of MCP-1RB and RANTES/MIP-1␣R, respectively. The amino-terminal and carboxyl-terminal halves of MMMM and RRRR were amplified in one round of PCR. A second round of PCR was performed using vector-specific primers that flank the cDNA insert, and purified fragments of the amino-terminal half of MMMM and the carboxyl-terminal half of RRRR as templates to create MMRR. To generate chimera RRMM, a second round of PCR was performed using purified fragments of the amino-terminal half of RRRR and the carboxyl-terminal half of MMMM as templates. The amplified products were purified and digested with HindIII/NotI and cloned into the respective sites in pcDNA3. Thus in MMRR, amino acids 1-127 are from the MCP-1 receptor, and amino acids 1-120 in RRMM are from the RANTES/MIP-1␣R.
Chimeras 5 and 6 -Receptors that exchange the amino terminus and first and second extracellular loops were also created by overlapping PCR. The complementary primers C (5Ј-tcatggtcatctgctac-3Ј) and D (5Ј-gtagcagatgaccatga-3Ј) correspond to the sequence MVICY, nt 730 -746 and 703-719 of MCP-1RB and RANTES/MIP-1␣R, respectively. The amino-terminal and carboxyl-terminal halves of MMMM and RRRR were amplified in one round of PCR. A second round of PCR, using vector-specific primers that flank the cDNA insert and purified fragments of the amino-terminal half of MMMM and carboxyl-terminal half of RRRR as templates, was performed to generate MMMR. RRRM was constructed in the same manner. Thus in MMMR, residues 1-209 are from the MCP-1R, and in RRRM, amino acids 1-206 are from the RANTES/MIP-1␣R. The amplified products were purified and digested with HindIII/NotI and cloned into the respective sites in pcDNA3.
Transfection and Fluorescence-activated Cell Sorting Analysis-HEK-293 cells (1573-CRL, American Type Culture Collection, Rockville, MD) were cultured in minimal essential medium with Earle's balanced salt solution supplemented with 10% fetal bovine serum, and 100 g/ml and 100 units/ml penicillin and streptomycin, respectively. Cells were transfected with DNA using Lipofectamine (Life Technologies, Inc.) according to the manufacturer's suggested protocol and placed under antibiotic selection, 800 g/ml G418 (Life Technologies, Inc.). Pools of G418-resistant cells were analyzed for cell surface expression of receptors by a fluorescence-activated cell sorter. Approximately 1 ϫ 10 6 harvested cells were incubated at room temperature for 1 h with culture medium containing a Flag epitope-specific antibody (M1; IBI, New Haven, CT) diluted 1:1000. Cells were washed three times with phosphate-buffered saline and resuspended in culture medium containing goat anti-mouse IgG-FITC (Zymed Laboratories, South San Francisco, CA) diluted 1:1000 and incubated at room temperature for 30 min. Unbound antibody was removed by washing with phosphate-buffered saline, and the cells were resuspended in phosphate-buffered saline plus 20 g/ml propidium iodide. Cell lines that expressed each of the constructs were selected for further study.
Cytokines and Binding Assays-MCP-1 and MIP-1␣ (R&D Systems, Minneapolis, MN) were labeled using the Bolton-Hunter reagent (diiodide, DuPont NEN), according to the manufacturer's instructions. Five micrograms of ligand in 10 l of 100 mM sodium borate, pH 8.5, was incubated with 1.0 mCi of Bolton-Hunter reagent for 15 min on ice, and the reaction terminated by the addition of 100 l of 0.5 M ethanolamine, 100 mM sodium borate, 10% glycerol, and 0.1% xylene cyanol. Unconjugated iodide was separated from labeled protein by elution using a PD-10 column (Pharmacia Biotech Inc.) equilibrated with phosphatebuffered saline and bovine serum albumin (1% w/v). Specific activity was determined by immunoassay (Quantikine; R&D) of the labeled protein and counting gamma emissions. The specific activity was typically 2200 Ci/mmol. Equilibrium binding was performed by adding 125 I-labeled ligand, with or without a 100-fold excess of unlabeled ligand, to 0.5 ϫ 10 6 cells in polypropylene tubes in a total volume of 300 l (50 mM Hepes, pH 7.4, 1.0 mM CaCl 2 , 5.0 mM MgCl 2 , 0.5% bovine serum albumin) and incubating for 90 min at 27°C on an orbital shaking set shaking at 150 rpm. The cells were collected, using a Skatron cell harvester (Skatron Instruments Inc., Sterling, VA), on glass-fiber filters presoaked in 0.3% polyethylimine and 0.2% bovine serum albumin. Unbound ligand was removed by washing with 4 ml of buffer (10 mM Hepes, 0.5 M NaCl, 0.5% bovine serum albumin) over a period of 10 s. After washing, the filters were removed, and bound ligand was quantitated by counting ␥ emissions. Ligand binding by competition with unlabeled ligand was determined by incubation of 0.5 ϫ 10 6 transfected cells (as above) with 1.5 nM radiolabeled ligand and addition of the indicated concentrations of unlabeled ligand to a final volume of 300 l. The samples were collected, washed, and counted as above. The data were analyzed using the curve-fitting program Prism (GraphPad Inc., San Diego, CA) and the iterative nonlinear regression program LIGAND (34).
Adenylyl Cyclase Assays-Inhibition adenylyl cyclase was assayed as described (35). Briefly, stably transfected HEK-93 cells were grown until confluent in 24-well tissue-culture dishes and labeled overnight with 2 Ci/ml [ 3 H]adenine (25-30 Ci/mmol) in minimal essential medium-Earle's balanced salt solution supplemented with 10% fetal calf serum. The cells were stimulated by addition of fresh medium containing either chemokine alone, forskolin alone (10 M, to activate adenylyl cyclase), or chemokine plus forskolin, all in the presence of 1 mM 3-isobutyl-1-methylxanthine, for 20 min at room temperature. The cAMP pool for each sample was normalized to its own ATP pool and 2 L.-M. Wong and I. F. Charo, unpublished data. expressed as a ratio by the equation (cAMP cpm/ATP cpm) ϫ 100. In each experiment, full dose-response curves were generated and expressed as a percent of the maximal stimulation achieved by forskolin alone. All data points were determined in duplicate.

RESULTS
Receptor Chimeras-To identify potential ligand-binding and signaling domains of the MCP-1 receptor, we utilized its high degree of relatedness to the RANTES/MIP-1␣ receptor. These two C-C chemokine receptors are 51% identical at the amino acid level but, with the exception of MCP-3 (36,37), have little overlap in ligand specificity. MCP-1RB and the RANTES/ MIP-1␣-R are most highly related in their transmembrane domains (greater than 70% identical), but diverge considerably in their extracellular domains. We therefore generated a series of chimeric receptors in which the amino-terminal extensions and each of the three extracellular loops were progressively exchanged between the two receptors ( Fig. 1). The wild-type MCP-1 receptor (Type B) is designated MMMM, to denote the amino terminus and each of its three extracellular loops. Similarly, the wild-type RANTES/MIP-1␣ receptor is designated RRRR.
Expression of Receptor Chimeras in HEK-293 Cells-HEK-293 cells were stably transfected with the wild-type and chimeric receptors. The Flag epitope (DYKDDDD) (32) was added at the amino terminus of each receptor to allow determination of receptor expression. The addition of this 8-amino acid sequence did not alter the binding or signaling properties of the MCP-1 or RANTES/MIP-1␣ receptors (data not shown). Surface expression was assayed by flow cytometry. Cell lines in which the chimeras were expressed at the cell surface were selected for further study (Fig. 2). Two chimeras in which the amino terminus was completely removed (⌬MMM and ⌬RRR) were not detected at the cell surface (data not shown).
Binding of MCP-1 and MIP-1␣ to Constructs with Exchanged Amino-terminal Extensions-We first examined the specific binding of 125 I-labeled MCP-1 and MIP-1␣ to constructs in which the amino-terminal extensions of the wild-type receptors were exchanged. Labeled MCP-1 bound well to the wild-type MCP-1 receptor (MMMM), as well as to the chimera MRRR, which contained only the amino terminus of the MCP-1 receptor (Fig. 3A). Analysis of these data by the method of Scatchard revealed essentially identical equilibrium dissociation constants (K d ) of 0.33 nM for MMMM and 0.27 nM for MRRR (Fig.  3B). In contrast, the wild-type RANTES/MIP-1␣ receptor (RRRR), as well as the chimera (RMMM), failed to bind MCP-1 with high affinity (Fig. 3A). To determine if RRRR or RMMM bound MCP-1 with a lower affinity, we examined the ability of higher concentrations of unlabeled MCP-1 to compete with a fixed concentration of 125 I-labeled MCP-1 for binding. As shown in Fig. 3C, MCP-1 did bind specifically to RMMM, but with a significantly lower affinity (K d ϭ 3.5 nM) as compared to chimeras with the amino terminus of the wild-type receptor. There was minimal binding to RRRR and no specific binding to untransfected HEK-293 cells. We conclude from these data that the amino-terminal extension of the MCP-1 receptor is necessary for high affinity binding of MCP-1 and that MCP-1 also interacts with a second site on the receptor with lower affinity. In contrast, the amino terminus of the RANTES/MIP-1␣R was not important for high affinity binding of MIP-1␣. Radiolabeled MIP-1␣ bound with high affinity to the wild-type receptor (RRRR, K d ϭ 0.58 nM), as well as to the construct that exchanged the receptor's amino terminus (MRRR, K d ϭ 0.69 nM) (Fig. 4). There was no detectable specific binding (in either direct or in competition binding assays) to the wild-type MCP-1 receptor (MMMM) or to the construct that substituted the amino terminus of the RANTES/MIP-1␣R onto the MCP-1 receptor (RMMM). We conclude from these data that one or more of the extracellular loops, but not the amino terminus, of the RANTES/MIP-1␣ receptor is necessary for high affinity binding.
As a further test of the hypothesis that MCP-1 and MIP-1␣ bind to different regions of their respective receptors, we took advantage of the fact that the chimera MRRR bound both ligands with high affinity (Figs. 3 and 4). In competition studies, unlabeled MCP-1, but not MIP-1␣, blocked the binding of 125 I-labeled MCP-1 to MRRR (Fig. 5). Similarly, unlabeled MIP-1␣, but not MCP-1, blocked the binding of radiolabeled MIP-1␣ to MRRR. These data strongly suggest that MCP-1 and MIP-1␣ bind to different regions of MRRR and are consistent with MCP-1 binding to the amino terminus and MIP-1␣ binding to one or more of the extracellular loops of their respective receptors.
Role of the Amino-terminal Extension in Receptor Signaling-We have shown previously that MCP-1RB and the RAN-TES/MIP-1␣ receptor couple via G␣i to inhibit adenylyl cyclase and lower intracellular levels of cAMP (35). In the next series of experiments, we examined the ability of the chimeric receptors to mediate signal transduction in response to MCP-1 and MIP1-␣. The wild-type MCP-1 receptor signaled well in response to MCP-1 (IC 50 ϭ 0.2 nM), as expected ( Fig. 6 and Ref. 35). There was little or no response of the RANTES/MIP-1␣ receptor (RRRR) to MCP-1, in agreement with published results (35). Substitution of the amino terminus of the MCP-1 receptor onto the RANTES/MIP-1␣ receptor, however, resulted in a dramatic increase in response to MCP-1 (compare MRRR, IC 50 ϭ 7.4 nM to RRRR, IC 50 Ͼ 100 nM) (Fig. 6). Similarly, substitution of the amino terminus of the MCP-1 receptor resulted in a greater than 30-fold loss in MCP-1 responsiveness (compare chimera RMMM, IC 50 ϭ 7.7 nM to MMMM, IC 50 ϭ 0.2 nM) (Fig. 6). These data are consistent with a critical role for the amino-terminal extension of the MCP-1 receptor in ligand binding and signal transduction and further suggest that the low affinity binding of MCP-1 to RMMM is sufficient to mediate signaling, albeit at higher ligand concentrations.
In contrast, the amino terminus of the RANTES/MIP-1␣ receptor was not essential for receptor signaling. In response to MIP-1␣, signaling of the chimera MRRR was reduced only 2-3-fold as compared to the wild-type receptor ( Fig. 7: MRRR, IC 50 ϭ 0.26 nM; RRRR, IC 50 ϭ 0.11 nM). Little or no inhibition of adenylyl cyclase was mediated by RMMM or MMMM in response to MIP-1␣ (Fig. 7). Similar results were obtained when mobilization of intracellular calcium was assayed as a measure of signaling, or when RANTES was used instead of MIP-1␣ (data not shown). We conclude that the amino-terminal Contribution of the Extracellular Loops to MCP-1 Binding and Receptor Activation-Four additional constructs that interchanged the amino terminus plus the first extracellular loop (RRMM and MMRR, see Fig. 1) or the amino terminus plus the first two extracellular loops (RRRM and MMMR) were examined for their ability to bind and signal in response to MCP-1 and MIP-1␣. Binding experiments with radiolabeled MCP-1 confirmed that the amino-terminal extension was responsible for virtually all of the high affinity binding (Table I). Furthermore, there was no detectable binding of MCP-1 to RRMM or RRRM in either direct or competition binding experiments ( Table I).
The results of signaling experiments suggested that each of the three extracellular loops contributed to the agonist-dependent inhibition of adenylyl cyclase by MCP-1. Thus, the chimera MMRR, which contained the amino terminus and first extracellular loop, had an IC 50 of 1.4 nM, whereas MMMR, which contained the amino terminus plus the first and second extracellular loops, had an IC 50 ϭ 0.4 nM (Fig. 7), despite its relatively low expression level at the cell surface (Fig. 2). As previously shown in Fig. 6, the chimera MRRR (with all three loops substituted) was the least active of this series of constructs, and the wild-type receptor (MMMM), which contained all three extracellular loops, had an IC 50 ϭ 0.2 nM. The chimeras RRMM and RRRM, which lacked the first and second extracellular loops, respectively (as well as the amino-terminal extension), failed to signal (Fig. 8), although each of these constructs was expressed at least as well as the wild-type receptors at the cell surface. These results are summarized in Table I.
Contribution of the Extracellular Loops to MIP-1␣ Binding and Receptor Activation-In contrast to the MCP-1 receptor in which each of the three loops as well as the amino-terminal extension contributed to signal transduction, the third extracellular loop of the RANTES/MIP-1␣ receptor was especially critical for signaling. Virtually no inhibition of adenylyl cyclase could be detected using the chimera RRRM, whereas MMMR, which contained only the third extracellular loop and cytoplasmic tail, did signal in response to MIP-1␣ (IC 50 ϭ 2.0 nM, Fig.  9). In the presence of the second and third extracellular loops (MMRR), signaling was enhanced 4-fold (IC 50 ϭ 0.46 nM). As shown earlier in Fig. 7, the addition of the first extracellular loop to this construct (MRRR, IC 50 ϭ 0.26 nM) resulted in a receptor that signaled almost as well as the wild-type C-C CKR-1. These results are summarized in Table I. We conclude from these data that in the presence of the third extracellular loop, each of the three loops of the RANTES/MIP-1␣ receptor contributes to signal transduction. In the absence of the third loop, however, no signaling was detected.
Binding studies confirmed the critical role of the third extracellular loop in the RANTES/MIP-1␣ receptor. Radiolabeled MIP-1␣ bound to the chimera MMMR (K d ϭ 24 nM), but not to RRRM (Fig. 10). Thus, the presence of the third extracellular loop alone was sufficient to support ligand binding. The presence of the second and third extracellular loops enhanced binding approximately 6-fold as compared to the third loop alone (MMRR, K d ϭ 4.1 nM versus MMMR, K d ϭ 24 nM), and the presence of all three loops (MRRR, Fig. 4A) resulted in high affinity binding (K d ϭ 0.69 nM) virtually equivalent to that of the wild-type receptor (RRRR, K d ϭ 0.58 nM). These data, summarized in Table I, are in sharp contrast to those obtained with the MCP-1 receptor, in which the amino terminus or all three of the extracellular loops were required for specific binding. transmembrane domain receptors. A number of receptors for chemokines and other chemotactic peptides have been cloned, and there has been considerable interest in identifying functional domains. To identify ligand binding and signaling do-mains of the MCP-1 receptor, we took advantage of the high degree of sequence identity between the MCP-1 receptor and the RANTES/MIP-1␣ receptor. These two C-C chemokine receptors are 51% identical overall, but are quite divergent in their extracellular domains. We constructed a series of chimeric receptors in which the amino-terminal extension and each of the three extracellular loops of the RANTES/MIP-1␣ R were progressively substituted into the MCP-1 receptor, and vice versa. Addition of the Flag epitope at the extreme amino terminus of each of these chimeras facilitated quantitation of receptor expression, and allowed selection of stable cell lines with comparable expression at the cell surface. In this paper, we report that the amino-terminal extension of the recently cloned MCP-1 receptor is necessary for high affinity binding of MCP-1. The amino terminus was not absolutely required for signaling, however, suggesting that low affinity interactions with one or more extracellular loops (or transmembrane domains) directly mediates receptor activation and signaling. In contrast, the third extracellular loop, but not the amino terminus of the closely related RANTES/MIP-1␣ receptor, was required for ligand binding and signal transduction. We conclude from these data that the amino terminus of the MCP-1 receptor, but not that of the RANTES/MIP-1␣ receptor, is critically involved in ligand binding and receptor activation.
The first set of chimeras (RMMM and MRRR) exchanged the amino-terminal extensions of the two wild-type receptors. Our  Table I 7. Signaling by the wild-type and chimeric receptors in response to MIP-1␣. Inhibition of adenylyl cyclase was assayed as described in Fig. 6. The data shown are the means of three independent experiments, and IC 50 values are shown in Table I initial studies with these constructs revealed that high affinity binding of MCP-1 correlated with the presence of the 40-residue amino terminus of the MCP-1 receptor. Signaling data obtained in transfected HEK-293 cells also supported a critical role for the amino terminus. Thus, substitution of the wild-type amino terminus of the MCP-1 receptor resulted in a greater than 30-fold increase in the IC 50 for inhibition of adenylyl cyclase (compare RMMM and MMMM, Fig. 6 and Table I).
That the chimera RMMM signaled in response to MCP-1 indicated that high affinity binding to the wild-type amino terminus was not absolutely required for receptor activation, and also suggested that MCP-1 bound to a second domain of the receptor to initiate signaling. Additional studies revealed low affinity binding of MCP-1 to RMMM, consistent with the higher IC 50 for signal transduction. To determine if this low affinity binding was to one or more of the extracellular loops, versus the substituted amino terminus (i.e. to the "MMM" versus the "R" portion of the RMMM chimera), we performed additional binding studies using the chimeras RRMM and RRRM. We were unable to demonstrate specific binding of MCP-1 to either of these chimeras, though both were expressed well at the cell surface. We conclude, therefore, that signaling of the MCP-1 receptor can be mediated by low affinity binding of MCP-1 to one or more of the three extracellular loops.
The presence of all three of the extracellular loops of the MCP-1 receptor appeared to be required for optimal signaling. Thus, loss of the third extracellular loop (MMMR) decreased signaling by 2-fold, loss of the second and third loops (MMRR) decreased signaling by an additional 3-fold, and loss of all three extracellular loops (MRRR) decreased signaling an additional 5-fold (Table I). Thus with regard to the MCP-1 receptor, the first extracellular loop may be particularly important for receptor activation.
It is possible that the receptor transmembrane domains contribute to the binding of MCP-1. However, there is greater than 70% identity at the amino acid level between the MCP-1 and the RANTES/MIP-1␣ receptor transmembrane domains, making it unlikely that these regions contribute to specificity. Furthermore, while agonist interactions with residues in the transmembrane domains are typical in the case of small molecules, these interactions are unusual for peptides exceeding 3 amino acids in length. Thus, the opiates (␦, , and ), tachykinins, endothelin, and the agonist peptide of the thrombin receptor have all been shown to interact with the extracellular loops of their respective seven-transmembrane domain receptors (reviewed in Ref. 17).
A model that incorporates the results of these experiments is shown in Fig. 11. In this model, the high affinity binding of MCP-1 (either in monomeric or dimeric form) to the aminoterminal extension serves to position MCP-1 for a lower affinity interaction with a second site on the receptor that ultimately initiates signaling. The model predicts that in the absence of high affinity binding to the amino terminus, signaling would still be possible, but would require higher MCP-1 concentrations. This is precisely what was found for the chimera RMMM; the dose-response curve for inhibition of adenylyl cyclase was shifted to the right. Signaling by the chimera MRRR is also consistent with this model. In this case, the parent receptor, the RANTES/MIP-1␣ receptor (RRRR), signals well in response to nanomolar concentrations of MIP-1␣, RANTES, and MCP-3 (35,37), but requires significantly higher concentrations of MCP-1 (IC 50 Ͼ 0.1 M). Substitution of the amino terminus of the MCP-1 receptor onto the RANTES/MIP-1␣ receptor to produce MRRR resulted in a chimera that bound MCP-1 with high affinity, and signaled in response to nanomolar concentrations of MCP-1 (i.e. this chimera exhibited a gain of function as compared to the parent receptor, RRRR). The most parsimonious explanation of these data is that the high "local" concentration of MCP-1 achieved by tethering it to the amino termi-nus allows a more efficient interaction with the extracellular loops, and hence causes activation of this chimera.
An analogous signaling mechanism has been described for the human thrombin receptor, whose amino-terminal extension is cleaved by thrombin to expose a truncated amino terminus, which then interacts with distal parts of the receptor as a "tethered ligand" (38). MCP-1 appears to use a modification of this mechanism in which the receptor amino terminus binds the peptide ligand with high affinity, effectively creating a "pseudo-tethered" ligand. It has been proposed, in this regard, that the high affinity binding of leutinizing hormone to the amino terminus allows interaction with a charged residue in the first extracellular loop of its receptor (22). Similarly, C5a appears to interact with the amino terminus, as well as a second site on the C5a receptor (30,31). A recent study with the IL-8R reports that multiple regions of the receptor are involved in binding IL-8 (29).
In sharp contrast to the MCP-1 receptor amino terminus, the amino-terminal 32 residues of the RANTES/MIP-1␣ receptor do not appear to be important in ligand binding or receptor activation. A chimera with a substituted amino terminus (MRRR) bound MIP-1␣ with virtually the same affinity as the wild-type receptor (RRRR). Similarly, signal transduction in response to MIP-1␣ was comparable in the presence of the wild-type (RRRR) or the substituted (MRRR) amino terminus. The presence of the third loop alone (chimera MMMR) was sufficient for binding and signaling in response to MIP-1␣, despite the relatively low surface expression of this chimera. These results are very different from those obtained with the MCP-1 receptor, in which the presence of the third extracellular loop alone (RRRM) did not result in detectable binding or signaling. Our data suggest that in the case of the RANTES/ MIP-1␣ receptor, the third extracellular loop is the critical domain of the receptor. Chimeras that lacked this third loop (e.g. RMMM, RRMM, or RRRM) neither bound nor signaled in response to MIP-1␣, despite their high level of surface expression as compared to MMMR. In the presence of the third loop, however, each of the three loops contributed to binding and signaling in a comparable manner. The requirement for multiple extracellular domains has also been observed with the formyl peptide receptor (fMLP) (39). Replacement of the first and third extracellular loops of the fMLP receptor, but not the amino terminus, significantly reduced the binding of fMLP (40,41).
In summary, we have created chimeric MCP-1 and RANTES/ MIP-1␣ receptors and correlated their ligand binding and signaling properties. Our data suggest a two-step model leading to activation of the MCP-1 receptor. In the first step, MCP-1 binds with high affinity to the amino-terminal extracellular domain (amino acids 1-40) of the receptor. The second step is a low  Table I. Binding is expressed as the percent observed in the absence of a 100-fold excess of unlabeled MIP-1␣. Shown is one of three similar experiments. å, RRRM; É, MMRR; ç, MMMR; q, RRRR.
FIG. 11. Hypothetical model for activation of the MCP-1 receptor. A twostep model is proposed for receptor activation. In the first step, MCP-1 binds with high affinity to the amino-terminal extension of the receptor. In the second step, the "pseudo-tethered" ligand binds with lower affinity to the extracellular loops to initiate signal transduction. In the absence of binding to the amino terminus, higher concentrations of MCP-1 are required to initiate signal transduction. Structures are not drawn to scale. affinity interaction between the "pseudo-tethered" MCP-1 and one or more of the receptor's extracellular loops, and serves to initiate signal transduction. The mechanism of ligand binding and activation of the MCP-1 receptor thus incorporates features of both large glycoprotein and small peptide interactions with seven-transmembrane domain receptors. In contrast, ligand binding and activation of the RANTES/MIP-1␣ receptor is dependent upon the third extracellular loop, and not the amino terminus. In comparison to other chemoattractant receptors, ligand interactions with the MCP-1 receptor are most similar to the IL-8 and C5a receptors. The RANTES/MIP-1␣ receptor, however, interacts with its ligands in a manner reminiscent of the binding of fMLP to the fMLP receptor in which multiple domains are required for high affinity binding. These data thus suggest significant functional differences between these two highly homologous C-C chemokine receptors. The identification of receptor binding and signaling domains provides an important starting point for the development of receptor agonists and antagonists. Such reagents are likely to prove useful in the treatment of a wide range of human diseases characterized by prominent monocyte/macrophage infiltrates.