Lysine 58 and Histidine 66 at the C-terminal α-Helix of Monocyte Chemoattractant Protein-1 Are Essential for Glycosaminoglycan Binding*

Monocytes rolling on the endothelial cell layer interact with monocyte chemoattractant protein-1 (MCP-1) that is tethered to the proteoglycans on the luminal side of the endothelial cells and consequently initiate adhesion of monocytes in the early phase of immune response. The amino acid residues in MCP-1 involved in tethering to the proteoglycans have not been elucidated. MCP-1 showed binding to [3H]heparin with aK D of 1.5 μm. We substituted lysine or histidine residues at the C-terminal end of MCP-1 with alanine residues and tested these mutants for their ability to bind heparin, heparan sulfate, hyaluronic acid, and chondroitin sulfate-C. Substitution of Lys-58 or His-66 drastically reduced glycosaminoglycan binding. Substitution of Lys-56 or deletion of the five amino acid residues at the C terminus, including Lys-75, did not alter the heparin binding ability, suggesting that the other lysine residues at the C terminus are not involved in glycosaminoglycan binding. MCP-1 and its mutants did not bind hyaluronic acid as strongly as the other subunits of the GAGs. Substitution of Lys-58 or His-66 by alanine that prevented glycosaminoglycan binding did not affect Ca2+ influx, receptor binding, or chemotactic activity elicited by the chemokine on monocytic THP-1 cells. Therefore, we conclude that the Lys-58 and His-66 residues in the C-terminal α-helix of MCP-1 are essential for glycosaminoglycan binding and probably for the binding to the endothelial surface proteoglycans.

Chemokines are small chemotactic proteins, which mediate directional migration of leukocytes from the blood to the site of injury. Usually the chemokines, produced at the site of injury, cause leukocyte migration (1)(2)(3)(4)(5). Monocyte chemoattractant protein-1 (MCP-1), 1 a member of the CC chemokine family, promotes recruitment of monocytes and basophil in response to injury and infection signals in various inflammatory diseases (6 -8), different types of tumors (9 -13), cardiac allograft (14), AIDs (15), and tuberculosis (16). The monomeric structure of MCP-1 consists of a N-terminal loop and a ␤ sheet overlaid by an ␣-helix at the C-terminal end (17). Endothelial cells produce and secrete MCP-1 in response to tumor necrosis factor-␣, interferon-␥, 12-O-tetradecanoylphorbol 13-acetate, lipopolysaccharide, and lipoprotein (18 -22). Monocyte migration to the site of injury is most likely mediated by chemokine gradients (4,5). The seven-transmembrane receptor of MCP-1, present on the monocytes rolling on the endothelial layer, binds MCP-1 tethered to proteoglycan on the luminal surface of the endothelial cells (23)(24)(25). This receptor binding activates the monocyte adhesion to the endothelial cells, followed by diapedesis through interendothelial junction and migration of the monocytes through the MCP-1 gradient to the point of production of MCP-1 at the site of injury or infection.
Proteoglycans are proteins that are posttranslationally modified by the addition of glycosaminoglycan (GAG) side chains at serine residues. The common GAGs are heparin, chondroitin sulfate, heparan sulfate, dermatan sulfate, and hyaluronic acid characterized by the disaccharide repeats (26). These peripherally arranged, highly charged polysaccharide side chains are thought to tether MCP-1 secreted from endothelial cells. Certain basic residues in the C-terminal ␣-helix of PF4, a member of CXC chemokine family, were found to be responsible for heparin binding (27). Progressive C-terminal truncation of the ␣-helix of interleukin-8 inhibited heparin binding (28). Amino acid residues involved in the heparin binding of the members of CC chemokine family are not well understood. In MIP-1␣, one study reported that two basic residues were involved in binding heparin as well as the receptor (29). Another report indicated that three noncontiguous basic residues were responsible for binding GAG without affecting interaction with MIP-1␣ receptor (30). Residues involved in the proteoglycan binding of MCP-1 have not been elucidated. Here we report that substitution of Lys-58 and His-66 with alanine in the C-terminal ␣-helix of MCP-1, but not the other basic residues at the Cterminal end, prevents GAG binding without affecting receptor binding, Ca 2ϩ influx, or chemotactic activity. GGACGCCCTGGACAAG-3Ј, H66A (antisense), 5Ј-CTTGTCCAGGGC-GTCCATGG-3Ј. For deletion of five amino acids from C terminus designated C⌬5, a pair of primers incorporating EcoRI sites were used: C⌬5 (sense), 5Ј-CACCTGGACAAGCAAACCGAATTCGCGCCC-3Ј, C⌬5 (antisense), 5Ј-GGGCGCGAATTCGGTTTGCTTGTCCAGGTG-3Ј. Each substitution mutagenesis was carried out by a two-step polymerase chain reaction procedure (32). The products were purified from agarose gel through Ultrafree ® -MC centrifugal filter units (Millipore Corp.) and used as template for the second step of amplification using the primers designed from either end of the MCP-1 coding region. The polymerase chain reaction products were cloned into pFastBac1 vector (33). Sequencing was performed to confirm the targeted mutation and the absence of any other mutation. The recombinant baculoviral bacmid constructs were generated by following the protocol described in the BAC-TO-BAC baculovirus expression system instruction manual (Life Technologies, Inc.).
Cell Culture-Sf21 insect cells were maintained in serum-free Sf-900 II medium (Life Technologies, Inc.) at 28°C until the cell density was 9 ϫ 10 5 /ml. Monolayers were grown in 150-cm 2 flasks, and suspension cultures were maintained on a rotary shaker at 100 -120 rpm.
Expression and Purification of Recombinant MCP-1 and Its Mutants-The insect cell culture was transfected with the recombinant bacmids (BAC-TO-BAC baculovirus expression system instruction manual) in a monolayer and were incubated for 6 -7 days. For larger scale protein production, Sf21 cells were grown in 250 ml of Sf-900 II medium inoculated with 3 ϫ 10 5 cells/ml and incubated at 28°C until cell density was 9 ϫ 10 5 /ml. The cultures were transfected with recombinant virus stock and incubated for 3-4 days for protein expression. The supernatants were recovered from the cultures by centrifugation at 100,000 ϫ g for 45 min. The recombinant proteins were purified by a two-step FPLC protocol using Mono S HR 5/5 and Superdex-75 HR 10/30 columns (Amersham Pharmacia Biotech), as described earlier (31). The recombinant proteins were subjected to SDS-PAGE and Western blot analysis to verify their identity and purity.

Measurement of Heparin Binding to MCP-1 and Its Mutants-[ 3 H]Heparin (NEN Life Science Products) was purified by passing 150
l of solution through a Sephadex G100 column (0.8 ϫ 29 cm) in phosphate-buffered saline (pH 7.4). Heparin binding to MCP-1 and its mutants was measured using a procedure similar to that used to measure binding of heparin to N-CAM (34). [ 3 H]Heparin (10,000 cpm) and varying amounts of protein and PBS buffer (pH 7.4) were mixed in a total volume of 25 l in a 1.5-ml microcentrifuge tube, incubated at 37°C for 1 h, and then passed through a 0.45-m, Schleicher & Schuell NCTM nitrocellulose membrane. Unbound heparin passed through the filter, whereas heparin bound to MCP-1 was retained on the membrane. The membrane-bound [ 3 H]heparin was measured by liquid scintillation spectrometry in a Beckman LS-3801 scintillation counter.
In competition assays, microcentrifuge tubes containing [ 3 H]heparin, 3 g (14 M) of protein, and different concentrations of unlabeled GAGs (chondroitin sulfate C, heparan sulfate, hyaluronic acid, and heparin) in a total volume of 25 l of PBS buffer were incubated for 1 h at 37°C. The mixture was passed through nitrocellulose, and protein-bound [ 3 H]heparin on the nitrocellulose was measured by scintillation spectrometry. The K D values were determined using the Kaleidagraph program.
Measurement of Ca 2ϩ Influx Induced by MCP-1 and Its Mutants-Intracellular Ca 2ϩ influx was monitored in THP-1 monocytic leukemia cells as reported previously (35). THP-1 cells (10 7 /ml) were incubated with 2 M fura-2 acetoxymethyl ester, washed in PBS buffer, and resuspended in Hepes-Tyrode buffer in the presence of 1 mM CaCl 2 . Fluorescence was measured in the absence and presence of the chemokine in a Perkin-Elmer LS-3B fluorescence spectrometer with constant stirring. The samples were excited at 340 nm, and emission was recorded at 500 nm. The saturation fluorescence was measured after treating the cells with 50 M digitonin, and the values are expressed as percentage of saturation (maximum) fluorescence.
Monocyte Chemotaxis-Chemotaxis of THP-1 cells was measured as described previously (31). THP-1 cells (10 7 /ml) were suspended in Gey's balanced salt solution containing 0.2% bovine serum albumin, incubated with 2 M calcein acetoxymethyl ester (Molecular Probes, Eugene, OR) at 37°C for 30 min, washed with PBS buffer, and resuspended in Gey's bovine serum albumin at 1 ϫ 10 6 cells/ml. Different amounts of MCP-1 protein were placed in a 96-well Polytronics view plate (Neuroprobe, Cabin John, MD), and 200 l of cells were added to the top wells of the chamber. Following a 1-h incubation at 37°C, the number of migrating cells were determined by measuring the fluorescence with a Cytofluor 2300 plate reader (Millipore, Bedford, MA).
Ligand Binding Assay-MCP-1 (wild type) was iodinated by the chloramine-T method described previously (36) with minor modifications. MCP-1 (1.5 g) protein was mixed with 250 Ci of Na 125 I (NEN Life Science Products) in 100 l of 0.1 M sodium phosphate buffer (pH 7.2) and 26.4 g of chloramine-T. After a 30-s incubation at room temperature, the reaction was stopped by the addition of 100 g of sodium metabisulfite. Iodinated MCP-1 protein was isolated by elution through a QAE-Sephadex column and quantitated as described previously (31). MCP-1 binding to THP1 cells, which express CC-CKR2B receptor, was measured as described previously (31,37). The final assay volume (250 l) contained 5 ϫ 10 6 THP-1 cells, different amounts of mutant chemokine, and 0.02 pmol of iodinated MCP-1 protein. Following incubation, the mixture was passed through a Whatman GFC filter and radioactivity on the filters was measured in a Packard COBRA ␥-radiation counter. The dissociation constant (K D ) value for each ligand protein was evaluated using the LIGAND program (38).

RESULTS
Since positively charged residues in the C terminus, located away from the receptor-binding N-terminal region, appeared to be likely sites for binding negatively charged proteoglycan, the effect of removal of such charges on proteoglycan binding was tested. Constructs for deletion of five amino acids including Lys-75 (C⌬5) and substitution of several residues (K56A, K58A, V60K, H66A, K69A, K75A, and a double mutant K58A/ V60K) at the C terminus of MCP-1 were made to test for the effect of such modifications on GAG binding. Expression of the mutants in the insect cell system showed that C⌬5, K56A, K58A, and H66A recombinant constructs yielded proteins in the culture fluid; others were not secreted. SDS-PAGE of the resulting purified proteins showed single band at 8.66 kDa (Fig. 1). Immunoblot analysis with rabbit antibodies prepared against human MCP-1 showed a single cross reacting band that coincided with the protein band in each case (data not shown).
Interaction of MCP-1 with [ 3 H]heparin was measured. We analyzed the results by non-linear curve fitting and found that heparin bound to MCP-1 with a K D of 1.5 M (Fig. 2A). K58A and H66A mutants of MCP-1 showed drastically reduced binding compared with that of the wild type MCP-1 protein ( Fig. 2A and Table I). On the other hand, K56A and C⌬5 mutants bound [ 3 H]heparin just as well as the wild type MCP-1 with similar K D (Fig. 2, A and B, and Table I).
By competition assays, the binding affinities of the wild type MCP-1 was tested with various GAGs, such as heparan sulfate, chondroitin sulfate-C, and hyaluronic acid. Based on the satu- ration curve determined in the [ 3 H]heparin binding experiment ( Fig. 2A), 14 M chemokine was added for competition assays. Chondroitin sulfate-C, heparan sulfate, and heparin competed with similar affinities to bind MCP-1. The relative K D values for heparan sulfate, chondroitin sulfate-C, and heparin were close to each other (Table I). Since GAGs contain a range of molecular weights, the K D values for competition assays were calculated using the average molecular weights indicated in Table I. Binding affinity of wild type MCP-1 for hyaluronic acid was much lower compared with the affinity for heparin, heparan sulfate, and chondroitin sulfate-C (Fig. 3A).
K56A and C⌬5 mutants bound to [ 3 H]heparin just like wild type (Fig. 2, A and B). The experimentally determined K D value for both K56A and C⌬5 mutants was 1.55 M. Data from Fig. 2 and the K D values of MCP-1, K56A, and C⌬5 confirmed that the substitutions or deletion of those residues did not alter the binding of the ligand to heparin. In contrast, binding of K58A or H66K mutants was reduced drastically.
In the competition assays, chondroitin-C, heparin, or heparan sulfate competed with [ 3 H]heparin with similar affinities for binding the mutant K56A ( Fig. 3B and Table I). Similar to the wild type MCP-1, binding affinity of the mutant K56A for hyaluronic acid was lower compared with the affinity for the rest of the GAGs. The competition for mutants K58A and H66A could not be tested due to their poor heparin binding ability.
The involvement of histidine in GAG binding is unusual, as lysines and arginines are the commonly involved residues in other GAG-binding proteins. To further test for the role of histidine in GAG binding of MCP, pH dependence of heparin binding was determined (Fig. 4). Changes in GAG binding with pH showed the involvement of a residue with a pK a of 6.8, strongly supporting the proposed role of a histidine residue of MCP-1 in heparin binding. K D for heparin binding increased with pH from 5 nM at pH 5.8 to 16 nM at pH 6.8 and to 29 nM at pH 7.8.
Since substitution of amino acids might perturb the structure of the chemokine and thus affect its binding to GAGs and biological function, we tested whether the mutations in the C-terminal regions affected the receptor binding, elicitation of Ca 2ϩ influx, and chemotaxis, using the monocytic THP1 cell line, which expresses CCR-2B receptor of MCP-1. Receptor binding was tested in THP1 cells in presence of different amounts of wild type MCP-1. The mutants bound the receptor with the same K D as the wild type (Table I). Thus, the mutations in the C terminus did not affect receptor binding.
Elevation of intracellular calcium was measured with a range of concentrations of the wild type MCP-1 and four mutants with THP-1 cells (Fig. 5, A and B). At 2 nM concentration of the wild type MCP-1, the peak fluorescence was 49% of the saturation fluorescence (caused by digitonin). Similar pattern of fluorescence peaks were obtained by the mutants (Fig. 5A). At 4 nm concentration of MCP-1 and mutants, the fluorescence peak was similar to that observed at 2 nM concentration. Thus, the results indicated that the amino acid substitution at residues 58 and 66 did not affect the biological response as far as Ca 2ϩ influx was concerned.
Chemotaxis of monocytes was tested at various concentrations of MCP-1 and K58A and H66A mutants. The mutants  showed chemotactic activity virtually identical to that observed with wild type MCP-1. Induction of chemotactic activity was measured at various concentrations of the chemokine and maximum induction was observed at 10 Ϫ9 M and 10 Ϫ10 M concentrations of MCP-1 and its mutants (Fig. 6).

DISCUSSION
The proteoglycans on the luminal surface of endothelial cells are thought to tether MCP-1 to present it to circulating monocytes without being carried away by the blood flow. To identify the specific residues of MCP-1 involved in proteoglycan binding, we tested the effect of substitution of several basic residues at the C terminus of MCP-1 with alanine on GAG binding. The work presented here identifies two amino acids at the C-terminal ␣-helix of MCP-1 that probably play a critical role in the binding of MCP-1 with GAGs. The substitution of Lys-58 and His-66 with alanine abolished the heparin binding ability of MCP-1. However, substitution of a nearby lysine with alanine (K56A) did not affect binding. In addition, deletion of five amino acids including Lys-75 (C⌬5) from the C-terminal ␣-helix of MCP-1 also did not affect the heparin binding ability. The K D values of heparin binding to MCP-1 and its mutants were extremely similar at around 1.5 M. In the competition assays, the wild type MCP-1 or K56A bound to chondroitin sulfate-C and heparan sulfate much more strongly than to hyaluronic acid.
Affinity between the chemokine and GAG is most likely based on ionic interaction. The limited number of studies conducted on the binding of chemokines to GAGs have reached somewhat conflicting conclusions about the regions involved in proteoglycan binding. In CXC chemokine the C terminus is highly basic, and in the two cases of such chemokines that have been studied, this basic region was suggested to be involved in GAG association. Progressive reduction of heparin binding was caused by successive deletions of three to four residues from the C-terminal ␣-helix of interleukin-8 (28). In PF4, another member of CXC chemokine family, substitution of two pairs of lysines at 61/62 and at 65/66 with Gln and Glu in the ␣-helix abolished heparin binding (27). However, in this case, substitution of the positive charges with negative charges complicated the interpretation concerning the role of the lysine residues in binding the negatively charged heparin. Another study indicated that substitution of Arg-20, Arg-22, and Arg-49 simultaneously with Gln decreased heparin binding even more than that caused by substitution of Lys-61, Lys-62, Lys-65, and Lys-66 simultaneously with Ala (39). In certain members of the CC chemokine family such as RANTES (regulated on activation normal T cell expressed and secreted), MIP-1␣, MIP-1␤, and MCP-2, the conserved Arg residues found at positions 18, 46, and 48 are thought to be involved in heparin binding. Site-directed mutagenesis indicated that these residues were required for MIP-1␣ to bind heparin (29,30). Based on the limited amount of information available on chemokine binding to heparin, the need to review the proposed primary structure for heparin binding motif (40) has been pointed out (39). Crystal structures of different heparin-binding proteins other than chemokines (41)(42)(43) strongly suggest that spatial proximity of positively charged amino acids in the three-dimensional structure probably makes the heparin binding site. According to the reported results pertaining to the GAG binding of MIP-1␣ and the present results on the amino acid residues of MCP-1 involved in GAG binding, it is clear that even within the CC chemokine family different structural features may be involved in this binding. In the present study, substitution for Lys-58 and His-66 in the C-terminal ␣-helix of MCP-1 virtually eliminated heparin binding. In three-dimensional structure, these residues are located in spatial proximity to play a significant role in heparin binding. (Fig. 7). The unusual involvement of a histidine, rather than the arginines and lysines that are usually found to be involved in GAG binding, was strongly supported by the pH dependence of GAG binding that showed a critical role for a residue with a pK a of 6.8. As expected from the involvement of histidine in heparin binding, K D was pH-dependent, increasing from 5 nM at pH 5.8 to 29 nM at pH 7.8. Substitution at position 56 or deletion of residues 72-76 (C⌬5) from the C terminus did not affect heparin binding, suggesting that the positively charged residues at 56 and 75 had no role in GAG binding. According to the crystal structure, these residues do not reside within the ␣-helix region (17).
The relationship between the binding site of the chemokines for proteoglycan and the receptor is not clear. In MIP-1␣, substitution of amino acid residues suspected to be involved in GAG binding was reported to eliminate also receptor binding (29). However, in another study, the three noncontiguous amino acid residues that define GAG binding site were not found to be a prerequisite for receptor binding and signaling by MIP-1␣ in vitro (30). A triple mutant in which the basic amino acids at 45, 46, and 48 were replaced with neutral amino acids did bind CC-CKR1, albeit less well than the wild type, and showed undiminished chemotactic activity (44). Substitution of the basic amino acid residues in the putative GAG binding sites may affect the receptor binding indirectly and such affects may not have physiological relevance. In fact, the presence of GAG in the chemotaxis assay medium for MIP-1␣ showed no effect (30,44), indicating that the proteoglycan binding does not interfere with the receptor binding. In the case of MCP-1, our results show that mutations in the ␣-helix that drastically decreased heparin binding had no effect on receptor binding, calcium flux, and chemotaxis. These results clearly separate receptor binding of MCP-1 from proteoglycan binding and strongly suggest that the ␣-helix is involved only in the MCP-1 presentation to the monocyte. GAG binding can cause multimerization of chemokines (45). As expected from this hypothesis, the number of GAG bound per mole of chemokine is very small in all of the binding experiments. The "polymerization" of the chemokines increases the local concentration of the chemokine and thus enhances the binding to the high affinity receptors within the local environment (45).
The finding that the two positively charged residues in the C-terminal ␣-helix are involved in the tethering of the chemokine is consistent with a general model for the MCP-1 function (Fig. 7). According to the model, the C-terminal ␣-helix is used to tether the chemokine to the luminal side of the endothelial cells of the blood vessel via the ionic interaction between the two juxtapositioned basic amino acid residues (positions 58 and 66) (Fig. 7) with the negatively charged extracellular proteoglycans. This would leave the N-terminal extended rod and the base of this rod for binding to the receptor on the monocyte rolling on the endothelial layer. Based on the available information, it has been concluded that the residues Asp-3, Thr-10, Tyr-13, Ser-34, and Lys-35 at the N-terminal extended rod and Tyr-28 and Arg-30 at the globular region of MCP-1 interact with its receptor (31,46,47). Lys-58 and His-66 residues at the C-terminal ␣-helix of MCP-1 bind proteoglycans of endothelial layer without interfering with the receptor binding. This model will allow the interaction of N-terminal segment of the receptor with the globular body of the chemokine that might determine the chemokine selectivity of the receptor, as suggested by mutagenesis studies that indicated the importance of Tyr-28 in receptor selectivity (46,47) and by the binding of chemokines to the fusion protein receptor (48,49). The N-terminal rod of the chemokine would then interact with the membrane and trigger the G-protein-coupled signal transduction. FIG. 7. Proposed model for MCP-1 interaction with its receptor while tethered to the endothelial cells by the interaction of the C-terminal lysine 58 and histidine 66 (arrows) residues with the proteoglycan on the cell surface. The N-terminal rod inserted into the membrane is involved in interaction with G-protein, the red residues were shown to be required for receptor binding, and the green (Tyr) is involved in receptor selectivity. The amino acid sequence of the C-terminal region of MCP is shown at the bottom with the residues in the ␣-helix in bold.