The Viral CC Chemokine-binding Protein vCCI Inhibits Monocyte Chemoattractant Protein-1 Activity by Masking Its CCR2B-binding Site*

Monocyte chemoattractant protein-1 (MCP-1) is a chemotactic cytokine mainly acting on monocytes and T cells that elicits its biological effects by interacting with the seven-transmembrane helix receptor CCR2B. The vaccinia virus strain Lister and many other poxviruses express soluble proteins (vCCI) that bind MCP-1 and other CC chemokines and inhibit their function. In order to define the interaction site of MCP-1 with vCCI from vaccinia, surface exposed residues of MCP-1 were identified and mutated to alanine. The MCP-1 variants were expressed, purified, and their interaction with vCCI was characterized. The site on MCP-1 for vCCI binding is dominated by arginine 18 with important additional contributions from tyrosine 13 and arginine 24. These residues define a binding site that largely overlaps with the CCR2B receptor interaction site. The viral chemokine-binding protein vCCI thus inhibits the biological function of MCP-1 by directly masking its CCR2B receptor-binding site.

Chemokines are small (8 -14 kDa) structurally related proteins that regulate cell trafficking of various leukocyte subtypes through interaction with a set of G protein-coupled receptors. The two major subfamilies are the CXC chemokines which act on neutrophils and non-hemopoietic cells and the CC chemokines which bind to receptors mainly expressed on monocytes, T cells, eosinophils, and basophils. Additional members of the chemokine family are the C chemokine lymphotactin and the CX3C chemokine fraktalkine. Chemokines are important for the development, homeostasis, and function of the immune system and play a pivotal role in host defense (for reviews, see Refs. [1][2][3]. Given the importance of chemokines for defense against pathogens it is no surprise that viruses themselves developed mechanisms to neutralize the function of chemokines to further their own propagation or evade host defense (4). Poxviruses express secreted chemokine neutralizing proteins termed virus-encoded chemokine-binding proteins (vCKBP). 1 Two fami-lies of vCKBPs have thus far been identified that use distinct mechanisms of chemokine neutralization (5). The vCKBP-1 family binds C, CC, and CXC chemokines with low affinity, possibly by interaction with the proteoglycan-binding site on chemokines, thereby interfering with the proper localization and presentation of chemokines in inflamed tissues (6,7). The vCKBP-2 proteins, however, bind preferably CC chemokines with high affinity and inhibit the interaction of chemokines with their cellular receptors (8 -10). vCKBP-2 proteins have been identified in strains of vaccinia (11), cowpox (12), rabbitpox (13), myxomavirus (14), and variola virus (15). Prototypic members of this protein family are the M-T1 gene product from myxoma virus and the rabbitpox virus major secreted 35-kDa protein (10). In rabbits infected with a rabbitpox virus which had the gene for the secreted 35-kDa protein deleted, an increased leukocyte influx into virus-infected tissues was observed, confirming the role of this protein in inhibition of chemokine-mediated leukocyte infiltration (14).
The vCKBP-2 from vaccinia virus strain Lister (vCCI, viral CC chemokine inhibitor) (11) has been cloned and characterized in vitro (9). It binds CC chemokines with high affinity (K D ϭ 100 -150 pM) and blocks their interaction with cellular receptors. Moreover, the three-dimensional structure of the highly related vCKBP-2 from cowpox was determined by x-ray crystallography and revealed a novel ␤-sandwich topology not found in any other protein (16).
Although it has been suggested that vCCI binds to CC chemokines through residues distinct from the glycosaminoglycan (GAG)-binding site, the exact nature of the vCCI interaction site on CC chemokines is unknown. In this study, the entire surface of a well characterized member of the CC chemokine family, human monocyte chemoattractant protein-1 (MCP-1) (17,18) was probed by alanine-scanning mutagenesis to identify residues important for the interaction with vCCI. The results define the surface on MCP-1 that is used to bind vCCI and show that vCCI partially masks the CCR2B receptor interaction site on MCP-1. Moreover, insights on how vCCI can interact with a large number of structurally diverse CC chemokines are provided.

EXPERIMENTAL PROCEDURES
Cloning and Mutagenesis of Human MCP-1-The human MCP-1 gene was cloned from a yeast two-hybrid cDNA library which was synthesized from activated human leukocytes (CLONTECH). The oligonucleotides used to amplify the human MCP-1 gene were caggtcaat-gtcgacTCGCCTCCAGCATGAAAGTCTCTGCC and caggtcaattctagaT-CAGTGATGGTGATGGTGATGAGTCTTCGGAGTTTGGGTTTG (the start and the stop codon are shown in bold, lowercase nucleotides indicate restriction sites and cleavable extensions). The oligonucleotides incorporated SalI and XbaI sites (underlined), respectively, for cloning into the mammalian expression vector pRS5a (a modification of pRK5 (Pharmingen) provided by S. Geisse, Novartis Pharma AG). * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Moreover, the 3Ј-amplification oligonucleotide incorporated a C-terminal (His) 6 tag for detection and purification purposes.
In the polymerase chain reaction, 0.4 g of the cDNA library were amplified with Pfu polymerase (Promega) using the following protocol: 2 min at 95°C, 30 cycles: 30 s at 94°C, 30 s at 55°C, 1 min at 72°C, 5 min at 72°C. The amplified fragments were isolated, digested with SalI/XbaI, and cloned into pRS5a. Mutations were introduced into the MCP-1 cDNA by using the QuickChange Site-directed Mutagenesis kit (Stratagene) and high performance liquid chromatography purified oligonucleotides carrying the desired mutations.
Expression and Purification of MCP-1 Mutants-The day prior to the transfection, 2 ϫ 10 5 /well HEK.EBNA cells were seeded into 6-well plates. Two micrograms of pRS5a expression plasmids carrying the MCP-1 sequences and 10 l of Geneporter transfection reagent (Gene Therapy Systems) were each separately diluted in 500 l of serum-free Dulbecco's modified Eagle's medium. The diluted DNA was added to the diluted transfection reagent and incubated at room temperature for 45 min. The culture medium was aspirated from the cells and the DNA/ transfection reagent mixture was carefully added to the cells and incubated for 5 h at 37°C. After the incubation, 2 ml of Dulbecco's modified Eagle's medium, 10% fetal bovine serum were added and the cells were incubated for 3 days at 37°C. The supernatants were then collected, centrifuged for 5 min, and subjected to purification prior to storage at Ϫ20°C.
To the 3 ml of culture supernatant containing the secreted MCP-1 protein, imidazole was added to a final concentration of 10 mM. To this solution 100 l of Ni-NTA Magnetic Agarose Beads (Qiagen) were added and the suspension was incubated for 2 h at 4°C on a spinning wheel. The beads were separated from the solution with a 12-tube magnet (Qiagen) and washed three times with wash buffer (50 mM Na 2 HPO 4 , 300 mM NaCl, 10 mM imidazole, 0.05% Tween 20, pH 8.0). The bound MCP-1 was eluted with 200 l of elution buffer (50 mM Na 2 HPO 4 , 300 mM NaCl, 250 mM imidazole, 0.05% Tween 20, pH 8.0) by resuspending the beads and incubation at room temperature for 1 min. The concentration of purified MCP-1 was measured with a Protein Assay (Bio-Rad) or by anti-MCP-1 ELISA (Biomol). Mutant MCP-1 concentrations were determined using the Protein Assay (Bio-Rad) with purified MCP-1 as standard. Typically 40 g of purified human MCP-1 or MCP-1 mutants were obtained from 3 ml of culture.
The purity of expressed MCP-1 proteins was assessed by SDS-PAGE on 18% pre-cast gels (Novagen). The gels were stained with GelCode Blue Stain (Pierce).
Binding of MCP-1 to vCCI by ELISA-An ELISA was developed to measure binding affinities of MCP-1 and mutant MCP-1 to the vCCI protein which was fused to the Fc portion of human IgG (R ϩ D Systems). 96-well Maxisorp plates (Nunc) were coated with 100 l of goat F(abЈ) 2 anti-human IgG Fc (Cappel) at a concentration of 3.2 g/ml and incubated at 4°C overnight. The plates were then washed with phosphate-buffered saline, 0.05% Tween 20, azide using a plate washer (Tecan). Excess binding sites on the plates were blocked by adding 300 l/well of blocking buffer (phosphate-buffered saline, 0.5% bovine serum albumin, 0.05% Tween 20) for 90 min. The plates were washed and 100 l/well vCCI was added at a concentration of 100 ng/ml and incubated for 2 h at room temperature with constant agitation. The plates were washed and serial dilutions of MCP-1 (0 -2700 ng/ml) or MCP-1 mutants (0 -8100 ng/ml) in blocking buffer were added and incubated for 2 h at room temperature with constant agitation. After washing, 100 l/well of a biotinylated polyclonal anti-human MCP-1 antibody (RϩD Systems) in blocking buffer was added at a concentration of 100 ng/ml, incubated for 2 h at room temperature and washed. Bound antibody was detected by incubation with alkaline phosphatase conjugated to streptavidin (1:3000, Immuno Research) for 30 min at room temperature. The color was developed for 30 min with 100 l/well p-nitrophenyl phosphate (0.14 mg/ml) in diethanolamine buffer (0.97% in water, pH 9.8). The enzymatic reaction was stopped with 50 l/well of 1.5 M NaOH and the absorption was measured at 405 and 490 nm with a Spectra Max 250 reader (Molecular Devices).
Dose-response curves were fitted to a four-parameter logistic function. EC 50 values of MCP-1 mutants were normalized to the EC 50 observed with MCP-1 (24.9 ng/ml).
Kinetic Characterization of the vCCI/MCP-1 Interaction by Surface Plasmon Resonance (SPR)-Certified BIAcore sensorchips CM-5 (BIAcore, Uppsala, Sweden) were utilized throughout this study. For activation of the carboxymethylated dextran layer, equal volumes of 0.2 M N-ethyl-NЈ-(3-diethylaminopropyl)-carbodiimide and 0.05 M N-hydroxysuccinimide (BIAcore "Amine coupling kit") were premixed by the autosampler of the instrument, and 35 l of the mixture were injected at a flow rate of 5 l/min. A 30 g/ml solution of anti-human Fc␥ antibody (Jackson Immunochemicals), prepared by diluting the stock solution into 10 mM acetic acid, pH 4.5, was then injected at 5 l/min for 3 min to yield an increase in resonance units of about 10,000. Remaining activated N-hydroxysuccinimide groups were quenched during a pulse of 1 M ethanolamine, pH 8.5, and the surface was exposed to a regeneration procedure of two injections of 20 mM HCl. Next, vCCI was accumulated on the surface by injecting 5 l of a 5 g/ml solution, at a flow rate of 5 l/min. This procedure yielded about 200 to 300 resonance units (statistics of a representative experiment: 268 Ϯ 2 resonance units; mean Ϯ S.E.; n ϭ 35), and was repeated each time after a regeneration and before the injection of chemokine. Dilutions of MCP-1 or MCP-1 mutants were prepared in BIAbuffer (20 mM HEPES, pH 7.4, 150 mM NaCl, 0.05% (v/v) Tween 20) to yield final concentrations of 2 to 20 nM. Association was followed for 12.5 min, dissociation for 5 min, and the flow rate was kept at at 20 l/min. Each cycle was finished by two 30-s pulses of 100 mM HCl at 20 l/min. After subtraction of the sensorgram obtained by injecting BIAbuffer, the titration series was analyzed using the BIAevaluation 3.0 software (BIAcore). The K D values given for the mutants are the average of K D values determined from individual injections and are not calculated from the average k on and k off values shown in Table III.
Intracellular Ca 2ϩ Mobilization Assay-Chinese hamster ovary (CHO) cells, stable transfected with the human CCR2B variant V64I (19) (CHO/CCR2B), were trypsinized, washed, and resuspended in Hanks' balanced salt solution buffer (Life Technologies) containing 20 mM HEPES (Life Technologies) and 0.5% bovine serum albumin (Sigma). The cells were loaded with 3 M Fura Red (Molecular Probes, emission at 660 nm) and 1.5 M Fluo3 (Molecular Probes, emission at 530 nm) in the presence of 0.04% pluronic acid (Molecular Probes) for 1 h at room temperature. After two washes with Hanks' balanced salt solution, HEPES, 0.5% bovine serum albumin, the cells were ready for stimulation with MCP-1 or MCP-1 mutants. Fluorescence emission in response to MCP-1 was measured at room temperature with a FACScan (Becton Dickinson) and the ratio of green and red fluorescence was calculated (20). Data were analyzed with the FlowJo software (TreeStar Inc.). The values measured with MCP-1 were normalized to 100% and the signals obtained with MCP-1 mutants were expressed relative to this. Results shown are the mean of two independent experiments with triplicate values in each experiment.
Binding of MCP-1 to Monoclonal Anti-MCP-1 Antibodies-In order to assess the structural integrity of MCP-1 mutants, their ability to interact with a panel of five monoclonal antibodies directed against human MCP-1 (Anogen: S8, S14, S101, S382, and 9G10) was measured in ELISAs using Ni-NTA HisSorp strips (Qiagen). MCP-1 mutants were bound via their C-terminal (His) 6 tag to the Ni-NTA on the solid support. Bound MCP-1 mutants were detected with the monoclonal antibodies and the relative binding compared with MCP-1 was calculated.
MCP-1 or mutants were added to the wells at 2700 ng/ml in 50 l of blocking buffer and incubated for 2 h at room temperature. The plates were washed and monoclonal antibodies S8 or S101 were added at 500 ng/l. For the monoclonal antibodies S14, S382, and 9G10 a concentration of 2000 ng/l was used. The plates were kept for 2 h at room temperature and then washed. Bound monoclonal antibodies were detected with 1:7500 diluted horseradish peroxidase-coupled goat antimouse antibodies (Santa Cruz Biotechnologies, 30 min at room temperature). The color was developed by addition of 50 l/well of BM blue horseradish peroxidase substrate (Roche Molecular Biochemicals) and incubation for 15 min. The enzymatic reaction was stopped by addition of 50 l/well of 1 M H 2 SO 4 and the absorption was measured at 450 and 690 nm on a Spectra Max 250 (Molecular Devices). Mutant MCP-1 binding to the five monoclonal antibodies was determined twice independently with triplicate values in each experiment and normalized to the values obtained for MCP-1 binding. Prior to normalization, the values were corrected for the absorption measured with buffer alone which accounted always for less than 5% of total binding.

RESULTS
Design of MCP-1 Mutants-The three-dimensional structure of MCP-1 was determined by NMR (21) and x-ray crystallography (22). We identified surface-exposed residues by visual inspection of the MCP-1 structure and probed a total of 57 residues that could potentially interact with vCCI by alaninescanning mutagenesis (23). This approach has been successfully applied to define protein-protein interfaces of growth hormone/growth hormone receptor (24), neurotrophin/receptor interactions (25) and other ligand/receptor pairs (26). Alanine can adopt the dihedral angles of all secondary structural elements in proteins and mutation to alanine is thus not expected to perturb the structure of the mutated MCP-1 variants. Residues that are likely involved in maintaining the overall integrity of the MCP-1 structure were not analyzed. These residues included amino acids forming disulfides, Cys-11, Cys-12, Cys-36, and Cys-52 as well as amino acids involved in hydrophobic core formation, Leu-25, Val-41, Phe-43, Trp-59, Val-60, and Leu-67. Finally, proline and alanine residues were generally not probed (Pro-2, Ala-4, Ala-26, Pro-37, Ala-40, Ala-53, Pro-55, and Pro-74).
Expression, Purification, and Analysis of MCP-1 Variants-MCP-1 and MCP-1 mutants were transiently expressed in HEK.EBNA cells and proteins were purified from conditioned medium using Ni-NTA magnetic beads. Peptide mapping of the purified protein followed by N-terminal sequencing of the fragments yielded the expected sequence (data not shown). Expressed and purified MCP-1 induced a similar dose response as Escherichia coli-derived commercially available MCP-1 (RϩD Systems) in Ca 2ϩ mobilization assays using CHO/CCR2B cells (data not shown). The affinity of MCP-1 to vCCI was determined by SPR. The K D for this interaction was 294 Ϯ 22 pM with k on ϭ 2.47 Ϯ 0.14 M Ϫ1 s Ϫ1 ϫ 10 6 and k off ϭ 6.79 Ϯ 0.33 s Ϫ1 ϫ 10 Ϫ4 .
The Main Determinants for MCP-1 Binding to vCCI Are Arg-18, Tyr-13, and Arg-24 -The most important residue for the interaction of MCP-1 with vCCI was discovered by analysis of the mutation R18A. This mutation caused a more than 10-fold increase in the EC 50 value in the ELISA (Table I). This result was confirmed by SPR measurements using the BIAcore instrument where no specific binding could be detected for this mutant (Table II). The mutation R18A thus strongly affects binding to vCCI. In order to ensure that the overall structure of this mutation was not perturbed, we determined the functional response mediated through the human CCR2B receptor by measuring the Ca 2ϩ mobilization in CHO/CCR2B cells. The CCR2B response was first titrated with MCP-1 and a halfmaximal response was observed at 1 nM (data not shown). Subsequently, all mutants were assessed at 1 nM in order to ensure maximal sensitivity of the functional assay. R18A activates CCR2B with an efficacy of 76 Ϯ 3% of the MCP-1 response at 1 nM, indicating that Arg-18 is not crucially important for the MCP-1 interaction with CCR2B (Table III). This mutant has also been shown to have full binding affinity to THP-1 cells and to a recombinant cell line expressing the CCR2B receptor (28). A loss in activity of a protein upon mutation of a side chain can be attributed either to the importance of this side chain to the function of the protein or to an involvement in maintaining the structural integrity of the protein. An additional method to verify the structural integrity of mutants with reduced activity is to determine their reactivities against a panel of monoclonal antibodies (26). The interaction of the R18A mutant with a panel of five monoclonal anti-MCP-1 antibodies was assessed and it was found that the residue Arg-18 is important for the interaction with S8 but not S14 and S382 (Table III). Therefore, the strong activity of R18A on the CCR2B receptor and its interaction with two of the monoclonal antibodies suggests that the overall structure of R18A is intact and that the large decrease of affinity to vCCI is due to the removal of a functional group crucial for this particular interaction. Hence, Arg-18 is the major determinant on MCP-1 for binding to vCCI.
Two additional residues, Arg-24 and Tyr-13, displayed strongly reduced affinity to vCCI after mutation to alanine ( Table I). Determination of the kinetic constants by SPR showed that the mutation Y13A affected mainly the off-rate of the kinetics (6-fold increase) with a small effect on the on-rate (2-fold decrease) (Table II). Both effects together led to an almost 13-fold decrease of affinity to vCCI. The mutation R24A affected both the on-and off-rate component of the kinetics and led to a 12-fold decrease in affinity to vCCI (Table II). The two mutations Y13A and R24A resulted in an almost complete loss of functional activity on the CCR2B receptor (Table III). An earlier study in which the binding of these mutations to CCR2B was analyzed came to similar conclusions (28). However, Y13A bound equally well as MCP-1 or even slightly better to all monoclonal antibodies in the panel suggesting that its overall structure remained intact. R24A strongly affected binding to two (S14 and 9G10) of the five antibodies tested, but displayed a strong interaction with the remaining three (S101, S8, and S382). Moreover, the structure of the R24A mutant was probed by NMR and a similar pattern of cross-peaks was found as in MCP-1, reflecting the structural integrity of this MCP-1 variant (28). Hence, Y13A and R24A are likely folded correctly. These observations, taken together with the strong decrease of affinity to vCCI, let us conclude that Tyr-13 and Arg-24 are important vCCI binding determinants on MCP-1. Hence, the residues Arg-18, Tyr-13, and Arg-24 define the surface on MCP-1 required for its interaction with vCCI.
Additional Residues of Importance for MCP-1 Binding to vCCI-In addition to the three main determinants, residues of lower importance for the MCP-1 interaction with vCCI were identified. The mutations P8A, N14A, F15A, N17A, K19A, I20A, and K38A all led to a 3-fold or larger decrease of affinity to vCCI (Table II). All mutants either displayed strong activity on CCR2B (P8A, F15A, K19A, I20A, and K38A) and/or bound well to all (P8A, N14A, F15A, N17A, I20A, and K38A) or a subset (K19A) of monoclonal antibodies (Table III), suggesting that none of these mutations led to important structural changes. These residues thus constitute the binding site on MCP-1 in addition to the three key residues.
Two point mutations (V9A and K49A) and a variant, ⌬MCP-1, where the first eight amino acids were removed (29), displayed an increased affinity to vCCI. The SPR experiments yielded K D values of 119 Ϯ 20, 102 Ϯ 26, and 42 Ϯ 6 pM for V9A, K49A, and ⌬MCP-1, respectively. Val-9 is located in direct contact to Tyr-13 and removal of this side chain may either change the conformation of Tyr-13 or lead to a better access of vCCI to this key residue on MCP-1. Removal of the first eight amino acids may similarly facilitate access to Tyr-13 or interfere with the monomer/dimer equilibrium and thereby modulate the interaction with vCCI. Finally, Lys-49 is located close to Arg-24 and could increase the affinity of vCCI to MCP-1 by analogous mechanisms as suggested for the V9A mutant.
Two additional residues, Tyr-28 and Met-64 displayed small but significant effects upon mutation to alanine. The interactions of these mutants with monoclonal antibodies and their activity on the CCR2B receptor suggest that they are correctly folded. These two residues form a small surface-exposed patch that is located opposite to the main vCCI interaction site formed by Tyr-13, Arg-18, and Arg-24.
The mutations K56A, Q57A, K58A, and Q61A affected the signal in the ELISA strongly (Table I) but when these mutations were analyzed with SPR their affinity to vCCI was found to be unchanged (Table II). These four residues form a contiguous patch on the surface of MCP-1 which is separate from the vCCI-binding site and it is likely that these mutations affected the epitope of the detection antibody and thereby led to the strong reduction of the signal observed in ELISA. In this respect it is interesting to note that the mutations of Lys-58 and Gln-61 to alanine resulted in a complete loss or strong reduction, respectively, of binding to three of the five antibodies in TABLE II Surface plasmon resonance analysis of the interaction of vCCI with MCP-1 variants Affinities of MCP-1 and mutants were determined by surface plasmon resonance as described under "Experimental Procedures." Values for k on , k off , and K D are given as mean Ϯ S.E. of at least two independent dose-response experiments. At least five measurements of all the kinetic constants per experiment were performed. K D was derived by fitting of experimental data to a 1:1 Langmuir association/dissociation model. Kinetic constants of mutants probing the main vCCI binding determinants Tyr-13, Arg-18, and Arg-24 as well as ⌬MCP-1 are marked bold. the panel (S101, S8, and S382) suggesting the presence of an immunodominant site at this position. The interaction with the polyclonal detection antibody could thus be affected by these mutations.
Residues Involved in Signaling through CCR2B-All mutants for which a decrease in the affinity to vCCI was demonstrated were also assessed for their ability to induce Ca 2ϩ mobilization in CHO/CCR2B cells. Among this set of residues, the main determinants for MCP-1 signaling activity are Tyr-13, Arg-24, Ile-31, Ile-42, and Lys-49. Mutation of these residues to alanine reduced the functional signal to less than 5% of the MCP-1 response (Table III). Moreover, ⌬MCP-1 (29) was inactive in the Ca 2ϩ mobilization assay (Table III) but was a potent antagonist of MCP-1 activity (data not shown) as described earlier (29,30). Previous studies using site-directed mutagenesis defined a similar surface of interaction of MCP-1 with its receptor CCR2B (27, 28, 30 -32).
The residue Arg-18 which is key for vCCI binding is not of great importance for CCR2B binding while the residues Tyr-13 and Arg-24 are important for both interactions. Moreover, mutations of the residues Ile-31, Ile-42, and Lys-49 to alanine strongly decreased the signaling capacity of the resulting mutant proteins but had only modest effects on the vCCI interaction. vCCI thus uses an overlapping but non-identical site on MCP-1 compared with CCR2B. Finally, the ⌬MCP-1 mutant had opposing effects on the interactions of MCP-1 with CCR2B and vCCI, respectively. While this variant completely lost its agonistic activity on the CCR2B receptor, its affinity to vCCI was strongly increased.

Residues Important for GAG Binding Are Not Involved in the Interaction of MCP-1 with vCCI-Chemokines interact with
GAGs such as heparin and heparane sulfate. These highly charged polysaccharide chains are thought to tether MCP-1 secreted from endothelial cells. Substitutions of Lys-58 and His-66 in MCP-1 by alanine resulted in a loss of GAG binding (33) and it was concluded that these two residues are key determinants for the MCP-1/GAG interaction. In this study we demonstrated that the mutations K58A and H66A did not affect binding to vCCI (Table I). The K58A mutant displays an affinity of 495 Ϯ 18 pM to vCCI which is only 1.7-fold lower than the K D of MCP-1 (Table II). Hence, residues involved in GAG binding are not important for the interaction of MCP-1 with vCCI. This finding is in agreement with the observation that heparin and heparin sulfate does not interfere with MIP-1␣ binding to vCCI (9).

DISCUSSION
The vCCI-binding Site on MCP-1-In the present study, site-directed mutagenesis and molecular modeling was employed to determine the interaction site of the CC chemokine MCP-1 with the secreted chemokine-binding protein vCCI from vaccina virus (strain Lister). Probing of all surface-exposed residues on MCP-1 for their importance to interact with vCCI revealed a functional binding site dominated by the three residues Tyr-13, Arg-18, and Arg-24 (Fig. 1A). Located between these three key amino acids are residues of significant but lower importance for the interaction with vCCI. These residues are Pro-8, Asn-14, Phe-15, Asn-17, Lys-19, Ile-20, Lys-38 and  1 mutants and binding to a panel of monoclonal anti-MCP-1 antibodies Ca 2ϩ mobilization in CHO cells expressing CCR2B was measured at 1 nM MCP-1 or mutant MCP-1. Binding of MCP-1 variants to monoclonal antibodies S101, S8, S14, S382, and 9G10 was measured by ELISA. All activities are expressed relative to the MCP-1 response and are shown as mean Ϯ S.D. of two independently performed experiments with triplicate determinations. Values lower than 30% are marked bold. Dominant residues for the Ca 2ϩ mobilization response through CCR2B and monoclonal antibody binding, respectively, are highlighted in the bottom row. Mutants probing the main vCCI binding determinants Tyr-13, Arg-18, and Arg-24 as well as ⌬MCP-1 are marked bold. Asp-54. Residues that display an increased affinity to vCCI when mutated to alanine are Lys-49 and Val-9. The former is located adjacent to Arg-24 and removal of this side chain may change the conformation of Arg-24 to facilitate binding of vCCI to MCP-1. Finally, removal of the first eight amino acids in the ⌬MCP-1 mutant increased binding to vCCI by 7-fold. The structural basis for this effect is unclear but it has been shown that ⌬MCP-1 has lost its ability to homodimerize. Therefore, access to Tyr-13, which is located close to the dimer interface, could be facilitated in the ⌬MCP-1 variant. It is well established that the functional binding sites of protein-protein interactions are dominated by only few residues located in the interface (24). This has been demonstrated by structural and functional analyses for the interactions of human growth hormone (26,34), vascular endothelial growth factor (35,36), and neurotrophins (25,37) with their respective receptors. However, these residues are usually localized close to each other forming a hydrophobic central patch (34) or providing a shielded hydrogen bond (37). The key residues on MCP-1 for its interaction with vCCI, however, are more than 10 Å apart. Although in their composition they consist of a hydrophobic residue (Tyr-13) and arginines (Arg-18 and Arg-24) which are both frequently found within protein-protein interfaces. While the presence of arginines and a tyrosine may be driven by the need to provide binding energy, their unusual arrangement may be a consequence of the fundamentally different evolutionary pressure applied on this viral chemokine inhibitor and on extracellular domains of growth factor receptors, respectively. Alternatively, vCCI may bind to the two sites sequentially by first docking to the Arg-18/Arg-24 area and only in a second step get tight binding by interacting with the more hydrophobic surface around Tyr-13.
Comparison of the CCR2B-and vCCI-binding Sites and the Mechanism of Action of vCCI-Various structure/function analyses of MCP-1 have been performed to determine its interaction site with the CCR2B receptor or to identify residues important for chemotaxis (27, 28, 30 -32). The key amino acids Arg-24 and Tyr-13 are each located at the center of two clusters of important residues which are separated by a hydrophobic groove (28). The side chains of the N-terminal residues do not contribute significantly to binding but are crucial for transmitting the signal. In the present study we could confirm these findings (Fig. 1B, Table III). Although the most important residue for binding to vCCI, Arg-18, is not of crucial importance for the recognition of CCR2B, the vCCI interaction site on MCP-1 otherwise largely overlaps with the CCR2B-binding region (Fig. 1). The residues Tyr-13 and Arg-24 are used by both MCP-1 interacting proteins and seem to provide hot-spots of binding energy (24) that are recognized by completely unrelated proteins (i.e. vCCI and CCR2B). The intrinsic physicochemical properties of these two surfaces may facilitate these interactions. In addition to these dominant residues, amino acids located in the hydrophobic groove between Arg-24 and Tyr-13 make various contacts with the CCR2B receptor and vCCI. Finally, deletion of the N-terminal eight residues resulted in an enhanced affinity to vCCI mainly through a decrease in the off-rate while the same residues were shown to interact with CCR2B to induce signaling.
Taken these results together, we suggest that vCCI inhibits MCP-1 activity by effectively masking important residues required for its interaction with the CCR2B receptor. Complexed soluble MCP-1 will thus not be able to elicit a signal through CCR2B. Moreover, MCP-1 bound to GAGs can likely still interact with vCCI since the GAG-binding residues are not involved in vCCI binding. Therefore, even MCP-1 displayed on GAGs can potentially be inhibited by vCCI, making this viral antagonist of CC chemokines even more effective.
The vCCI-binding Site on MCP-1 in Relation to the Structure of vCCI-The three-dimensional structure of vCCI from cowpox, which is highly homologous to the vaccinia protein, was for the viral binding protein vCCI and the cellular receptor CCR2B. The MCP-1 structure is shown as a monomer (gray). A, the interaction site for vCCI is dominated by Arg-18, Arg-24, and Tyr-13 (red: Ͼ10-fold effect) with smaller contributions of additional residues (magenta, 3-to 10-fold effect; pink, Ͻ3 and Ͼ2-fold effect). Residues with increased affinity to vCCI upon mutation to alanine are Val-9 and Lys-49 (orange). B, the site for CCR2B binding is formed by residues Tyr-13, Arg-24, Ile-31 (not visible), Ile-42, and Lys-49 (red). An MCP-1 variant with the first eight amino acids deleted (⌬MCP-1, red) cannot signal through CCR2B. The sites on MCP-1 for vCCI and CCR2B, respectively, partially overlap. vCCI thus masks the CCR2B interaction site on MCP-1 and thereby prevents binding to and signaling through this receptor.
determined at a resolution of 1.85 Å (16). The vCCI protein adopts a ␤-sandwich topology not previously described which is very likely distinct from the CCR2B structure. A patch of conserved residues on the exposed face of a ␤-sheet that is strongly negatively charged was suggested to interact with CC chemokines. We have identified the MCP-1 residues Arg-18 and Arg-24 as important contributors to binding energy and suggest that these residues could indeed be the counterparts of the negatively charged amino acids on vCCI.
vCCI interacts with most if not all CC chemokines with remarkably high affinity (9). CC chemokines are related in their protein sequences but nevertheless display a large degree of sequence variation. The question arises thus, how vCCI can recognize such a diverse set of surfaces. It is possible, that there are physicochemical features on the surface of chemokines that are conserved in the absence of a strict amino acid identity and that vCCI evolved to retain a certain flexibility to accommodate slightly different arrangements of key determinants in different ligands. In an analogous case, the hinge region of the Fc fragment of human immunoglobulin G has been shown to interact with completely unrelated proteins with high affinity using a common binding site (38). This site was highly accessible, adaptive, and hydrophobic. In contrast to the presumed vCCI surface it also contains few sites for polar interactions.
vCCI Recognizes a Conserved Surface on Various CC Chemokines-The interaction of five CC chemokines MCP-1, Eotaxin-1, RANTES, MIP-1␣, and I-309 with vCCI was characterized by displacement binding assays with radiolabeled MIP-1␣ (9). The strength of the interaction of these chemokines with vCCI was determined and the observed rank order was MIP-1␣ Ͼ eotaxin-1 Ͼ RANTES Ͼ MCP-1 Ͼ Ͼ I-309 (no binding). The amino acids at the key positions 13, 18, and 24 were compared in an alignment of the protein sequences of these chemokines (Fig. 2). Arg-18 was identified as the key MCP-1 residue for binding to vCCI and consequently the multiple sequence alignment reveals that all vCCI-binding CC chemokines carry an arginine at this position. The only non-binding protein, I-309, carries a glutamine. Therefore, an arginine at position 18 (or equivalent) seems to be a precondition for binding to vCCI. Interestingly, this residue is not directly involved in receptor binding, yet it is conserved in the four CC chemokines. An other important residue in MCP-1, Tyr-13, is a phenylalanine in all other sequences suggesting a requirement for a large hydrophobic amino acid at this position. The requirements for the position 24 are less stringent since basic and hydrophobic amino acids are allowed. It is possible that these amino acid differences account for the different affinities of the CC chemokines for vCCI.
It was suggested that vCCI could be useful for the treatment of inflammatory conditions due to its broad neutralizing activity against CC chemokines (8). Indeed, administration of vCCI in a model of allergic airway hyperreactivity (39) has proven effective in amelioration of some aspects of the disease. A small molecule mimetic of vCCI could be a therapeutically more amenable form of this effective chemokine inhibitor. Mutational analyses of vCCI and the CC chemokines in combination with structural studies of chemokine/vCCI complexes could facilitate the design of such molecules.