Identification of a novel integrin alphaMbeta2 binding site in CCN1 (CYR61), a matricellular protein expressed in healing wounds and atherosclerotic lesions.

CCN1 (cysteine-rich 61) and CCN2 (connective tissue growth factor) are growth factor-inducible immediate-early gene products found in atherosclerotic lesions, restenosed blood vessels, and healing cutaneous wounds. Both CCN proteins have been shown to support cell adhesion and induce cell migration through interaction with integrin receptors. Recently, we have identified integrin alphaMbeta2 as the major adhesion receptor mediating monocyte adhesion to CCN1 and CCN2 and have shown that the alphaMI domain binds specifically to both proteins. In the present study, we demonstrated that activated monocytes adhered to a synthetic peptide (CCN1-H2, SSVKKYRPKYCGS) derived from a conserved region within the CCN1 C-terminal domain, and this process was blocked by the anti-alphaM monoclonal antibody 2LPM19c. Consistently, a glutathione S-transferase (GST) fusion protein containing the alphaMI domain (GST-alphaMI) bound to immobilized CCN1-H2 as well as to the corresponding H2 sequence in CCN2 (CCN2-H2, TSVKTYRAKFCGV). By contrast, a scrambled CCN1-H2 peptide and an 18-residue peptide derived from an adjacent sequence of CCN1-H2 failed to support monocyte adhesion or alphaMI domain binding. To confirm that the CCN1-H2 sequence within the CCN1 protein mediates alphaMbeta2 interaction, we developed an anti-peptide antibody against CCN1-H2 and showed that it specifically blocked GST-alphaMI binding to intact CCN1. Collectively, these results identify the H2 sequence in CCN1 and CCN2 as a novel integrin alphaMbeta2 binding motif that bears no apparent homology to any alphaMbeta2 binding sequence reported to date.

The expression of CCN1 is essential for normal embryonic development inasmuch as targeted disruption of the CCN1 gene in mouse results in embryonic lethality (15). Interestingly, the majority of the CCN1-null embryos exhibit vascular defects, consistent with an essential role of CCN1 in normal embryonic angiogenesis. In adults, CCN1 is present at low levels in the cardiovascular system (16,17); however, its expression is up-regulated in a number of vascular diseases including atherosclerosis and proliferative restenosis (12,13,17,18). High expression of CCN2 in human advanced atherosclerotic lesions has also been observed (19,20). Moreover, both CCN1 and CCN2 are expressed during cutaneous wound healing (21)(22)(23). It is well established that leukocyte adhesion and emigration play an important role in inflammation, wound healing, and atherosclerosis (24). In a recent report, we showed that human peripheral blood monocytes adhere to CCN1 and CCN2 in an activation-dependent manner through integrin ␣ M ␤ 2 (12), underscoring the importance of these proteins in the pathophysiologic function of leukocytes. To establish these CCN proteins as novel ligands of ␣ M ␤ 2 , we demonstrated direct binding of the I domain of the integrin ␣ M subunit (␣ M I) to immobilized CCN1 and CCN2 (12). It is noteworthy that monocyte adhesion and ␣ M I domain binding to CCN1 are blocked by anti-␣ M monoclonal antibodies as well as by soluble heparin. These findings suggest that the ␣ M ␤ 2 binding site may lie in close proximity to the heparin binding motifs within the Cterminal domain of CCN1.
To gain further insight into the interaction of ␣ M ␤ 2 with CCN proteins, we sought to identify the ␣ M ␤ 2 binding site in CCN1. In the present study, we demonstrated ␣ M ␤ 2 -dependent adhesion of monocytes to SSVKKYRPKYCGS present in the C-terminal domain of CCN1. Furthermore, the ␣ M I domain binds specifically to this 13-residue sequence of CCN1 with a similar affinity as to the P2 (YSMKKTTMKIIPFNRLTIG) sequence, an ␣ M ␤ 2 binding site in the fibrinogen ␥ chain (25). Our newly identified ␣ M ␤ 2 binding sequence in CCN1 bears no sequence homology to any known ␣ M ␤ 2 binding motif reported to date and may provide a target for blocking CCN1-␣ M ␤ 2 interaction.

MATERIALS AND METHODS
Antibodies, Peptides, and Reagents-The anti-␣ M monoclonal antibody 2LPM19c was purchased from Dako Corp. MOPC 21, an isotypematched control mouse IgG, was obtained from Sigma. For solid phase binding assays, anti-GST goat polyclonal antibody was from Amersham Biosciences, and horseradish peroxidase (HRP)-conjugated rabbit antigoat IgG was from Sigma. For ELISA, HRP-conjugated donkey antirabbit antibody was from Pierce. An anti-peptide antibody (anti-CCN1 367-381 ) against a peptide sequence corresponding to the extreme C terminus of CCN1 (F 367 PFYRLFNDIHKFRD 381 ) was raised in rabbits and affinity-purified as previously described (13). Anti-domain I and anti-domain II polyclonal antibodies, raised in rabbits using GST fusion proteins linked to domain I or domain II of CCN1 as immunogens, were pre-absorbed with GST and then affinity-purified. In ELISA, both anti-domain I and anti-domain II antibodies reacted with fulllength CCN1 as expected. Furthermore, these antibodies reacted specifically with their cognate CCN1 domains, and no cross-reactivity was observed (Table I).
Protein Purification-Recombinant mouse CCN1, synthesized in a baculovirus expression system using Sf9 insect cells, was purified from serum-free conditioned media by Sepharose S chromatography as previously described (27). The CCN1 truncation mutant lacking the Cterminal domain (CCN1⌬CT) was produced as a hexahistidine-tagged fusion protein and purified by nickel-agarose chromatography as described (10). Purified CCN1 and CCN1⌬CT were analyzed by SDSpolyacrylamide gel electrophoresis followed by Coomassie Blue staining and immunoblotting.
Recombinant I domain of the integrin ␣ M subunit was expressed as a fusion protein with GST (GST-␣ M I) and purified as described previously (28). Briefly, the coding region for human ␣ M I domain sequence Asp 132 -Ala 318 was amplified and inserted into the expression vector pGEX-4T-1 (Amersham Biosciences). Escherichia coli cells were transformed with the above recombinant vector, and protein expression was induced with 1 mM isopropyl-␤-D-thiogalactopyranoside for 4 h at 37°C. The GST-␣ M I fusion protein was affinity-purified from cell lysates using glutathioneagarose (Sigma).
Isolation of Peripheral Blood Monocytes-Peripheral blood monocytes were isolated from human blood as previously described (12). Acid-citrate-dextrose anticoagulated blood was collected from healthy donors and centrifuged to remove platelet-rich plasma. For isolation of mononuclear cells, the remaining packed red blood cells and buffy coat was diluted with phosphate-buffered saline (10 mM sodium phosphate, pH 7.35, 0.15 M NaCl) and centrifuged through a layer of Ficoll-Paque (Amersham Biosciences) at 400 ϫ g for 60 min at 4°C. The mononuclear cell layer was diluted in an equal volume of phosphate-buffered saline containing 2 mM EDTA, sedimented, and washed twice with modified Tyrode's buffer (10 mM Hepes, pH 7.35, 135 mM NaCl, 2.9 mM KCl, 12 mM NaHCO 2 , 1 mM MgCl 2 , 1 mM CaCl 2 , 0.1% dextrose, and 0.2% bovine serum albumin (BSA)) by centrifugation at 130 ϫ g for 10 min. To further separate monocytes from lymphocytes, the mononuclear cells were suspended in modified Tyrode's buffer and subjected to centrifugation through a discontinuous density gradient of Percoll (Amersham Biosciences). Monocytes were isolated between a Percoll density of 1.047 and 1.050 g/ml, washed with modified Tyrode's buffer, and resuspended to a final concentration of 2-3 ϫ 10 6 cells/ml. The purity of the monocyte preparation was greater than 80%, as measured by anti-CD14 staining in flow cytometry, and cell viability was more than 95% as judged by trypan blue exclusion.
Cell Adhesion Assay-Microtiter wells (Immulon 2 Removawell strips, Dynex Technologies, Inc.) were coated with 10 g/ml CCN1 or CCN1⌬CT protein or with 4 mg/ml peptide for 20 h at 4°C. After protein or peptide coating, the wells were blocked with 0.2% polyvinyl alcohol (PVA) for 30 min at 37°C. The amounts of immobilized CCN1 and CCN1⌬CT protein were quantified by an ELISA using anti-domain I and anti-CCN1 367-381 antibodies as indicated. The amounts of immobilized peptides were measured by incubation for 2 h at 22°C with 10 M PEO-maleimide-activated biotin (Pierce), which reacts with the free sulfhydryl group in the cysteine residue of the peptides. After washing, the amounts of coupled biotin were quantified by incubation with 0.5 M 125 I-labeled streptavidin, and bound radioactivity was measured by ␥-counting.
In cell adhesion experiments, isolated monocytes were activated with 20 M ADP, added to the wells (100 l/well), and incubated for 20 min at 37°C. After washing to remove non-adherent cells, adherent cells were quantified using the acid phosphatase assay by incubation with the substrate solution (0.1 M sodium acetate, pH 5.5, 10 mM p-nitrophenylphosphate, and 0.1% Triton X-100; 100 l/well) for 2 h at 37°C (29). The reaction was stopped by the addition of 15 l of 1 N NaOH/well, and A 450 was measured. In inhibition studies, monocytes were preincubated with antibodies for 30 min at 37°C before addition to microtiter wells.
Binding of GST-␣ M I to Immobilized CCN1 and CCN-derived Peptides-Microtiter wells were coated with 30 g/ml wild type or truncated CCN1 or 4 mg/ml synthetic peptides for 20 h at 4°C and blocked with 0.2% PVA for 30 min at 37°C. GST-␣ M I fusion protein was added to the wells and incubated for 1-2 h at 22°C. Unbound GST-␣ M I was removed by washing with 30 mM Tris-HCl, pH 7.4, 0.2 M NaCl, 1 mM MgCl 2 , and 0.02% PVA. Bound GST-␣ M I was detected by an ELISA using anti-GST followed by an HRP-conjugated secondary antibody. Bound antibodies were detected using o-phenylenediamine dihydrochloride (Sigma) as the substrate. The reaction was stopped by the addition of 25 l of 6 N HCl, and A 490 was measured. In inhibition experiments using an anti-␣ M monoclonal antibody, GST-␣ M I was preincubated with the antibody for 30 min at 37°C before addition to the wells. In inhibition studies using anti-CCN1-H2 polyclonal antibodies, CCN1-coated wells were blocked with 1% BSA and preincubated with the antibodies for 30 min at 37°C before the addition of the GST-␣ M I fusion protein. Binding proceeded for 30 min at 37°C, and bound GST-␣ M I was detected as described above. In binding isotherm studies, input GST-␣ M I concentrations required for half-maximal binding of the fusion protein to immobilized peptides were estimated using the Graph-Pad Prism software in which the data was fitted to the one-site binding equation.

The C-terminal Domain of CCN1 Is Required for Monocyte
Adhesion and ␣ M I Domain Binding-We previously identified ␣ M ␤ 2 as the major adhesion receptor on monocytes for CCN1 and CCN2 and showed that the I domain of the ␣ M subunit bound specifically to both proteins (12). Moreover, soluble heparin inhibited monocyte adhesion and ␣ M I domain binding by interacting with the heparin binding motifs at the C-terminal domain of CCN1. The inhibitory effect of heparin suggests that the ␣ M ␤ 2 recognition site in CCN1 may be located in its Cterminal domain. To investigate this possibility, we performed monocyte adhesion and ␣ M I domain binding experiments on a truncated CCN1 mutant that lacks the C-terminal domain (CCN1⌬CT) (10). Fig. 1A shows that monocytes adhered to wells coated with wild type CCN1 protein. By contrast, no cell adhesion to CCN1⌬CT-coated wells was observed. In agreement with the cell adhesion results, a GST fusion protein containing the ␣ M I domain (GST-␣ M I) bound only to intact CCN1 but not to CCN1⌬CT (Fig. 1B). To determine whether similar amounts of CCN1 and CCN1⌬CT protein were immobilized onto microtiter wells, we performed an ELISA using two polyclonal antibodies raised against the entire domain I of CCN1 (anti-domain I) or a synthetic peptide corresponding to residue Phe 367 -Asp 381 at the C terminus of CCN1 (anti-CCN1 367-381 ). As shown in Fig. 1C, almost equal signals were obtained using the anti-domain I antibodies, indicating that similar amounts of CCN1 and CCN1⌬CT were immobilized. As expected, anti-CCN1 367-381 , directed against the extreme C terminus of CCN1, reacted with intact CCN1 but not with CCN1⌬CT, which lacks the C-terminal domain. Together, these results indicate that the C-terminal domain of CCN1 is required for both monocyte adhesion and ␣ M I domain binding. Thus, the ␣ M ␤ 2 binding site likely resides within the C-terminal region of CCN proteins.
The H2 Sequence in the CCN1 C-terminal Domain Supports Monocyte Adhesion-The inability of CCN1⌬CT to support monocyte adhesion and ␣ M I domain binding suggests that the recognition site for ␣ M ␤ 2 is located in the C-terminal domain of CCN1. We focused on the first half of the C-terminal domain, because this region is highly homologous between CCN1 and CCN2, and both proteins support monocyte adhesion and ␣ M I domain binding. Also, heparin has been shown to inhibit monocyte adhesion and ␣ M I domain binding to CCN1 (12), and therefore, we examined the possibility that ␣ M ␤ 2 binds directly to the heparin binding motifs in CCN1. In these studies we synthesized two peptides, CCN1-H1 and CCN1-H2, that correspond to the two heparin binding motifs of CCN1 ( Fig. 2A). The peptides were immobilized onto microtiter wells, and their ability to support monocyte adhesion was examined. To determine the coating efficiency of CCN1-H1 and CCN1-H2 onto the wells, PEO-maleimide-activated biotin was allowed to react with the free sulfhydryl group in the single cysteine residue present in each peptide followed by detection with 125 I-labeled streptavidin. Quantitation of bound radioactivity indicates that similar amounts of both peptides were immobilized onto the wells (Fig. 2B). In cell adhesion studies we found that monocytes adhered to CCN1-H2, but not to CCN1-H1 (Fig. 2C). Furthermore, monocyte adhesion to CCN1-H2 was completely blocked by 2LPM19c, an anti-␣ M monoclonal antibody, whereas control mouse IgG had no effect. Microscopic examination of the monocytes adherent to CCN1 protein and CCN1-H2 peptide revealed that the cells were well spread on both substrates (data not shown). Together, these results indicate that the H2 sequence in the C-terminal domain of CCN1 specifically mediates monocyte adhesion in an ␣ M ␤ 2 -dependent manner.
We previously reported that monocyte adhesion to CCN1 is markedly enhanced after cellular activation with ADP and formylmethionylleucylphenylalanine (12). Enhanced adhesion of activated monocytes to CCN1 is likely due to inside-out signaling, resulting in increased ␣ M ␤ 2 affinity for CCN1 and/or mobilization of internal ␣ M ␤ 2 to the cell surface. In addition to inside-out signaling, the activation states of integrins can be modulated by extracellular divalent cations inasmuch as Mn 2ϩ has been shown to increase the apparent affinity/avidity of multiple integrins including ␣ M ␤ 2 (30). We, therefore, examined the effect of extracellular Mn 2ϩ on monocyte adhesion to intact CCN1 protein and to the CCN1-H2 peptide. As expected, the addition of 1 mM Mn 2ϩ to the cell suspension resulted in enhanced monocyte adhesion to CCN1 (Fig. 3). Likewise, monocyte adhesion to the CCN1-H2 peptide was also significantly enhanced in the presence of Mn 2ϩ . In a specificity control, monocytes were allowed to adhere to wells coated with a scrambled CCN1-H2 sequence (CCN1-scrH2). To confirm that CCN1-scrH2 was adsorbed onto microtiter wells, we performed binding of PEO-maleimide-activated biotin and 125 I-labeled streptavidin as described above. Even though a higher amount of CCN1-scrH2 (95.6 Ϯ 6.1 fmol/well) was coated onto the wells as compared with CCN1-H2 (43.6 Ϯ 0.9 fmol/well), no signifi- cant cell adhesion to CCN1-scrH2-coated wells was observed in the absence and presence of Mn 2ϩ . The ability of Mn 2ϩ to enhance monocyte adhesion to CCN1 and CCN1-H2, but not to CCN1-scrH2, indicates that activated ␣ M ␤ 2 binds with a higher affinity to intact CCN1 and to the CCN1-H2 sequence.
Binding of ␣ M I Domain to the H2 Sequences in CCN proteins-The above cell adhesion data suggest that the CCN1-H2 sequence, but not CCN1-H1, serves as a binding site for integrin ␣ M ␤ 2 . To corroborate cell adhesion studies with receptor binding, we performed ␣ M I domain binding experiments on immobilized CCN1-H1 and CCN1-H2. Fig. 4A shows that GST-␣ M I bound directly to wells coated with the CCN1 protein or with the CCN1-H2 peptide, but not to wells coated with the CCN1-H1 peptide (filled bars). As controls, GST itself did not bind to CCN1, CCN1-H1, or CCN1-H2 (open bars). To demonstrate binding specificity, we tested the effect of 2LPM19c on ␣ M I domain binding. As shown in Fig. 4B, inhibition of GST-␣ M I binding to CCN1-H2 was observed with 2LPM19c (anti-␣ M ) but not with an isotype-matched mouse IgG. These results indicate that integrin ␣ M ␤ 2 binds directly to the CCN1-H2 sequence to mediate monocyte adhesion to CCN1 protein.
In addition to CCN1, we previously reported that the ␣ M I domain also binds to CCN2, another CCN family member highly homologous to CCN1 (12). Therefore, we performed solid phase binding studies to examine whether GST-␣ M I also binds to the corresponding H2 sequence in CCN2. Fig. 5A shows the H2 sequences in CCN1 and CCN2 with the non-conserved residues underlined. The ␣ M I domain was found to bind dose dependently to both CCN1-H2 and CCN2-H2 peptides, whereas no binding was observed with wells coated with the CCN1-scrH2 sequence (Fig. 5B). Using the GraphPad Prism software, we estimated that half-maximal binding of GST-␣ M I to immobilized CCN1-H2 and CCN2-H2 occurred at input GST-␣ M I concentrations of 47.2 Ϯ 6.4 and 128.0 Ϯ 26.3 nM (means Ϯ S.E.; n ϭ 5 for CCN1-H2 and n ϭ 3 for CCN2-H2), respectively. Thus, the results of the binding isotherms indicate that the ␣ M I domain binds with a higher affinity to CCN1-H2 than to CCN2-H2 (Student's t test, p Ͻ 0.01). Consistently, we previously showed that monocytic THP-1 cells adhere more strongly to CCN1 than to CCN2 (12).
It has been reported that integrin ␣ M ␤ 2 binds to fibrinogen in part through interaction with the P2 sequence (YSMKKTT-MKIIPFNRLTIG) at the C-terminal region of the fibrinogen ␥ chain (25). In the present study, we compared GST-␣ M I binding to the H2 sequence of CCN1 and the P2 sequence of fibrinogen (Fbg-P2). As shown in Fig. 6A, GST-␣ M I bound saturably to both CCN1-H2 and Fbg-P2. The difference in maximal binding to the two peptides may be due to different coating efficiencies of these peptides onto microtiter wells. Nonetheless, half-maximal binding of GST-␣ M I to CCN1-H2 and Fbg-P2 occurred at 42.6 Ϯ 7. with similar affinities to both peptide sequences. To further investigate the relationship of the CCN1-H2 and Fbg-P2 binding site(s) in the ␣ M I domain, we examined the effect of soluble Fbg-P2 peptide on GST-␣ M I binding to intact CCN1 protein and CCN1-H2 peptide. Fig. 6B shows that Fbg-P2 dose-dependently inhibited ␣ M I domain binding to CCN1-and CCN1-H2coated wells, indicating mutual exclusive binding of these two peptide sets to the ␣ M I domain. We then compared the potency of soluble CCN1-H2 and Fbg-P2 peptides in blocking GST-␣ M I binding to immobilized CCN1 protein. Although CCN1-H2 also significantly inhibited GST-␣ M I binding to CCN1-coated wells, much weaker inhibition was attained with CCN1-H2 (ϳ25%) as compared with Fbg-P2 (ϳ65%) at a high concentration (1 mM) of these peptides (Fig. 6C). As a specificity control, soluble CCN1-scrH2 had no inhibitory effect.
Effect of Anti-CCN1 Antibodies on ␣ M I Domain Binding to CCN1-The above experiments show that a synthetic peptide corresponding to the CCN1-H2 sequence supports monocyte adhesion and ␣ M I domain binding. To demonstrate the importance of the H2 sequence in intact CCN1 protein for interaction with ␣ M ␤ 2 , we produced an anti-peptide antibody against CCN1-H2 and tested its effect on GST-␣ M I binding to CCN1. Anti-CCN1-H2 antibodies were affinity-purified by sequential chromatography on protein A-Sepharose and CCN1-H2-coupled Sepharose. In ELISA, anti-CCN1-H2 antibodies bound with a higher affinity to CCN1 when protein coating was performed in the presence of 1% ␤-mercaptoethanol than in the absence of the reducing agent. Also, we found that GST-␣ M I bound equally well to non-reduced and reduced CCN1 (data not shown). Thus, in inhibition studies, CCN1 was immobilized onto microtiter wells under reducing conditions and preincubated with the indicated antibodies before the addition of GST-␣ M I. As shown in Fig. 7, the anti-CCN1-H2 antibody effectively inhibited ␣ M I domain binding to CCN1 by ϳ75%. By contrast, no inhibition was observed with preimmune rabbit IgG or with two polyclonal antibodies directed against the C-terminal sequence (Phe 367 -Asp 381 ) or domain II of CCN1. The lack of inhibition by anti-CCN1 367-381 and anti-domain II antibodies was not due to the failure of these antibodies to bind to CCN1 because they reacted strongly with immobilized CCN1 in ELISA (see Fig. 1C and Table I). Thus, these findings show that anti-CCN1-H2 specifically blocked GST-␣ M I binding to intact CCN1 protein, indicating that the H2 sequence is a major binding site for integrin ␣ M ␤ 2 in CCN1. DISCUSSION In a previous study we established CCN1 and CCN2 as novel ligands of integrin ␣ M ␤ 2 on activated monocytes (12). The major finding of this study is that CCN1-H2, a 13-residue sequence in the C-terminal domain of CCN1, is sufficient to mediate ␣ M ␤ 2 -dependent monocyte adhesion. Consistent with the cell adhesion results, the ␣ M I domain binds specifically to the CCN1-H2 peptide, and an anti-CCN1-H2 antibody effectively blocked ␣ M I domain binding to intact CCN1 protein.
Likewise, the ␣ M I domain also binds to the corresponding H2 Microtiter wells were coated with 30 g/ml CCN1 protein or with 4 mg/ml CCN1-H1 or CCN1-H2 peptide and blocked with 0.2% PVA. A, GST-␣ M I fusion protein or GST (1 M each) was added to the wells and allowed to bind for 2 h at 22°C. After washing, bound GST-␣ M I or GST was detected by ELISA as described in the legend of Fig. 1 sequence of CCN2, another CCN protein that has been shown to support ␣ M ␤ 2 -dependent monocyte adhesion. These findings elucidate a structure-function relation of CCN proteins and define the H2 sequences in their C-terminal domains as a novel recognition site for integrin ␣ M ␤ 2 .
Both H1 and H2 sequences in the CCN1 C-terminal domain contain the BBXB (B represents a basic amino acid) consensus glycosaminoglycan binding sequence that interacts with heparin or cell surface HSPGs (31). Thus, alanine substitutions of some of the lysine and arginine residues in either H1 or H2 in CCN1 markedly diminish its binding affinity to an heparin-Sepharose column (7). Moreover, human skin fibroblasts fail to adhere to a CCN1 mutant (CCN1-DM) with alanine substitutions of the basic residues in both H1 and H2, which is due to the inability of CCN1-DM to interact with cell surface HSPGs on fibroblasts (7). With respect to ␣ M ␤ 2 -dependent monocyte adhesion to CCN1, we reported earlier that cell surface HSPGs are involved but not absolutely required for this adhesion process (12). Nonetheless, soluble heparin has been shown to block monocyte adhesion as well as ␣ M I domain binding to wild type CCN1, suggesting that the ␣ M ␤ 2 binding site in CCN1 may lie in close proximity to the H1 and H2 sequences in the C-terminal domain. Consistent with this speculation, a CCN1 truncation mutant lacking the C-terminal domain does not support monocyte adhesion and ␣ M I domain binding (Fig. 1). Using synthetic peptides that encompass the H1 and H2 sequences of CCN1, we demonstrated that monocytes adhere and ␣ M I domain binds to immobilized CCN1-H2 but not to CCN1-H1 (Fig.  2). The specificity of ␣ M ␤ 2 interaction with CCN1-H2 has been further confirmed by the lack of reactivity of a scrambled CCN1-H2 sequence in both monocyte adhesion and ␣ M I domain binding assays. In subsequent studies, we showed that an  anti-peptide antibody against CCN1-H2 specifically blocks ␣ M I domain binding to CCN1, thus affirming the importance of the H2 sequence in intact CCN1 protein in mediating interaction with ␣ M ␤ 2 . Based on these results, we suggest that CCN proteins interact with monocytes through the cooperative binding of the H1 sequence to cell surface HSPGs and the H2 sequence to integrin ␣ M ␤ 2 . In support of this model, there is increasing evidence that certain cell surface HSPGs (e.g. syndecans) and chondroitin sulfate proteoglycans act as co-receptors with integrins to mediate cell adhesion and signaling (32, 33). Moreover, it is interesting to note that the integrin and proteoglycan binding sites in the fibronectin heparin III domain and in the laminin ␣ 5 N-terminal domain are also spatially close together as in CCN1 (33,34).
THP-1 cells also adhere to CCN2, and the ␣ M I domain also binds to CCN2 (12); however, we consistently observed weaker THP-1 cell adhesion to CCN2 as compared with CCN1. Thus, it is not surprising that the ␣ M I domain binds to the CCN2-H2 peptide with a lower affinity than to the CCN1-H2 peptide, which is likely due to the non-conserved Lys to Thr and Pro to Ala substitutions in the H2 sequences of these two CCN proteins (see Fig. 5A). Nonetheless, the ability of the ␣ M I domain to bind to both CCN1-H2 and CCN2-H2 suggests that the conserved residues in these H2 sequences may provide coordination sites for interaction with integrin ␣ M ␤ 2 . The H2 sequence is highly conserved among other CCN proteins but is missing in CCN5, and therefore, it is tempting to speculate that ␣ M ␤ 2 may also bind to CCN3, CCN4, and CCN6 through this sequence. Thus, further studies of the H2 sequences in different CCN proteins may help to define the binding specificity of this novel ␣ M ␤ 2 recognition motif.
Among integrin family members, ␣ M ␤ 2 has the broadest ligand binding specificity that is not shared by the closely related ␣ L ␤ 2 leukocyte integrin. Recently, we demonstrated that insertion of the ␣ M -(Lys 245 -Arg 261 ) sequence into ␣ L ␤ 2 imparts to the chimeric integrin the ability to recognize multiple ␣ M ␤ 2 ligands, including CCN1 (35). The broad ligand binding specificity of ␣ M ␤ 2 suggests that it is capable of interacting with many different recognition sequences derived from diverse ␣ M ␤ 2 ligands. Indeed, several ␣ M ␤ 2 binding sequences have been identified (Table II), including the P1 and P2 sequences in the fibrinogen ␥ chain (25, 36) and a nine-residue sequence in the coagulation factor X (37). In addition, a 22-residue sequence in intercellular adhesion molecule-2 binds to both ␣ M ␤ 2 and ␣ L ␤ 2 (38,39), and a LLG-C4 biscyclic nonapeptide derived from phage display screening interacts with ␣ M ␤ 2 , ␣ L ␤ 2 , and ␣ X ␤ 2 (40). Our newly identified CCN1-H2 sequence exhibits no apparent homology with these ␣ M ␤ 2 binding motifs other than a pair of basic amino acid residues is present in both P1 and P2 of fibrinogen, the factor X sequence, and CCN1-H2. However, it is not likely that the basic amino acid pair is the sole ␣ M ␤ 2 recognition motif because CCN1-H1, which contains three basic amino acid pairs, does not support monocyte adhesion and ␣ M I domain binding. Moreover, the active CCN2-H2 and P2-C ( 383 TMKIIPFNRLTIG 395 from the C-terminal portion of fibrin-ogen P2) peptides, do not contain a pair of basic residues. At present, the critical determinants in these ␣ M ␤ 2 binding sequences that mediate receptor-ligand interaction remain to be defined.
In the solid phase binding assay, we found that the ␣ M I domain binds with a similar affinity to immobilized CCN1-H2 peptide as to the fibrinogen-P2 peptide. Soluble P2 effectively cross-competes ␣ M I domain binding to CCN1 protein and CCN1-H2 peptides, suggesting that both peptide sets interact with the same or overlapping binding sites within the ␣ M I domain. As mentioned earlier, cells expressing a chimeric receptor in which the ␣ M -(Lys 245 -Arg 261 ) sequence has been grafted into ␣ L ␤ 2 adhere more avidly to fibrinogen and CCN1 as compared with cells expressing wild type ␣ L ␤ 2 (35). Thus, ␣ M -(Lys 245 -Arg 261 ) would likely play an important role for ␣ M ␤ 2 interaction with both fibrinogen-P2 and CCN1-H2 sequences. Although ␣ M I domain binds strongly to immobilized CCN1-H2 peptide, soluble CCN1-H2 is a weak inhibitor for blocking ␣ M I domain binding to CCN1. One possible explanation is that the ␣ M I domain may bind to additional sites on CCN1 other than the H2 sequence. Alternatively, soluble CCN1-H2 peptide may not be able to assume the optimal conformation for high affinity interaction with the ␣ M I domain. It is noteworthy that both CCN1-H1 and CCN1-H2 sequences contain a cysteine residue, which may pair with other cysteines in the CCN1 C-terminal domain (4), thereby affecting the affinity of ␣ M ␤ 2 interaction with CCN1-H2. At present, the disulfide bond pattern of CCN1 has not been resolved experimentally. However, GST-␣ M I binds similarly to both non-reduced and reduced CCN1 (results not shown), suggesting that disulfide bond formation in CCN1 does not play a major role in ␣ M ␤ 2 binding. Whether the neighboring regions of CCN1-H2 would affect its binding affinity to the ␣ M I domain is under investigation.
Although the pathophysiologic function of CCN proteins remains to be established, the increased expression of CCN1 and CCN2 in healing wounds, restenosed blood vessels, and atherosclerotic lesions coupled with their ability to interact with monocytes suggests that these CCN proteins may play an important role in inflammatory responses. Thus, both CCN1 and CCN2 are ligands of integrin ␣ M ␤ 2 and may induce integrindependent outside-in signaling and gene expression in monocytes. In this regard, we recently observed that adhesion of peripheral blood monocytes to CCN1 induces the expression and secretion of proinflammatory mediators including interleukin 1␤ and monocyte chemotactic protein-1. 2 Identification of the ␣ M ␤ 2 binding site in CCN proteins would greatly facilitate further mutational studies to demonstrate the biologic significance of ␣ M ␤ 2 -CCN protein interaction on monocytes. Furthermore, it may lead to the development of novel therapeutic agents targeting the ␣ M ␤ 2 binding site on this new family of matricellular signaling molecules.