Targeting Platelet–Leukocyte Interactions : Identification of the Integrin Mac-1 Binding Site for the Platelet Counter Receptor Glycoprotein Ibα

Adhesive interactions between vascular cells play important roles in orchestrating the inflammatory response. Recruitment of circulating leukocytes to vascular endothelium requires multistep adhesive and signaling events, including selectin-mediated attachment and rolling, leukocyte activation, and integrin-mediated firm adhesion and diapedesis that result in the infiltration of inflammatory cells into the blood vessel wall (1). Firm attachment is mediated by members of the β₂ integrin family, LFA-1 (α_Lβ ₂, CD11a/CD18), Mac-1 (α_Mβ₂, CD11b/CD18), and p150,95 (α_Mβ₂, CD11c/CD18), which bind to endothelial counter ligands (e.g., intercellular adhesion molecule [ICAM]-1; 2), endothelial-associated extracellular matrix proteins (e.g., fibrinogen; 3), or glycosaminoglycans (4).

Leukocyte recruitment and infiltration also occur at sites of vascular injury where the lining endothelial cells have been denuded and platelets and fibrin have been deposited. A similar sequential adhesion model of leukocyte attachment to and transmigration across surface-adherent platelets has been proposed (5). The initial tethering and rolling of leukocytes on platelet P-selectin (6) are followed by their firm adhesion and transplatelet migration, processes that are dependent on α_Mβ₂ (5).

Our laboratory has focused on identifying the platelet counter receptor for α_Mβ₂. Evaluation of the structural features of integrins provides insight into candidate platelet counter receptors for α_Mβ₂. Integrins are heterodimeric proteins composed of one α and one β subunit. A subset of integrin α subunits, including α_M, contains an inserted domain (I domain) of ∼200 amino acids that is implicated in ligand binding (7–9) and strikingly similar to the A domains of von Willebrand factor (vWf; 10), one of which, A1, mediates the interaction of vWf with its platelet receptor, the glycoprotein (GP) Ib-IX-V complex. Because of the similarity of the vWf A1 domain and the α_MI domain, we hypothesized that GP Ibα might also be able to bind α_Mβ₂ and reported that GP Ibα is indeed a constitutively expressed counter receptor for α_Mβ₂ (11). Furthermore, under the conditions used in these studies, the predominant interaction between neutrophils and platelets appeared to be between α_Mβ₂ and GP Ibα (11).

The α_MI domain contributes broadly to the recognition of ligands by α_Mβ₂ (12) and specifically to the binding of GP Ibα (11). This region has been implicated in the binding of ICAM-1 (13), iC3b (14), fibrinogen (12, 13), and neutrophil inhibitory factor (NIF; 15), as well as GP Ibα. Previous studies suggested that overlapping, but not identical, sites are involved in the recognition of iC3b, fibrinogen, and NIF (16, 17). Although the binding sites for iC3b, NIF, and fibrinogen in the α_MI domain have been mapped extensively (18–23), the recognition site for GP Ibα is unknown.

In this study, we have localized the binding site for GP Ibα within the α_MI domain. The strategy developed was based on the differences in the binding of GP Ibα to the α_MI and α_LI domains and involved several independent approaches, including screening of mutant cells, synthetic peptides, site-directed mutagenesis, and gain in function analyses. The binding site for GP Ibα was localized within the segment α_M(P²⁰¹–K²¹⁷). The grafting of two amino acids within this segment into the α_LI domain converted it to a GP Ibα–binding protein. Thus, a small segment that has a defined structure within the α_MI domain is necessary and sufficient for GP Ibα binding.

The soluble extracellular region of GP Ibα (sGP Ibα; i.e., glycocalicin) was purified as we previously reported (11). Human fibrinogen depleted of plasminogen, vWf, and fibronectin was purchased from Enzyme Research Laboratories.

The CD11/CD18 mAbs used included the following: LPM19c, directed to the α_MI domain (provided by K. Pulford, Radcliffe, Oxford, United Kingdom; 12); OKM1, directed to the α_M subunit of human Mac-1 (American Type Culture Collection [ATCC]); M1/70, directed to the α_M subunit mouse α_Mβ₂ (ATCC; 24); CBRM1/5, an activation-specific α_M reporter antibody (provided by T. Springer, Harvard Medical School, Boston, MA; 25); TS1/22, directed to the α_L subunit of α_Lβ₂ and capable of blocking ICAM-1 binding (ATCC); MEM-83, an activation-specific α_L reporter antibody (Caltag); and IB4, a blocking mAb directed to the β₂ subunit (ATCC). The stimulating CD18 mAb KIM127 (26) was provided by M. Robinson (Celltech Ltd., Slough, United Kingdom). mAb24, a β₂ activation reporter antibody (27), was provided by N. Hogg (Imperial Cancer Research Fund, Lincoln's Inn Fields, London, United Kingdom).

Peptides were obtained from the W.M. Keck Biotechnology Resource Center at Yale University. The peptides were diluted in DMSO and stored at −80°C.

THP-1 monocytic cells (ATCC) were maintained and differentiated with 1 ng/ml TGF-β1 and 50 nM 1,25-(OH)₂ vitamin D3, as previously described (28). 293 cells expressing human α_Lβ₂, α_Mβ₂, or mutant α_Mβ₂ receptors were established and maintained as previously described (16, 20, 23, 29).

To systematically define the GP Ibα–binding site in α_M, a homologue-scanning mutagenesis strategy was implemented (20). Accordingly, guided by the crystal structure (8, 9), the hydrated surface of the α_MI domain was replaced with sequences of the α_LI domain in segments of 7–11 amino acids. To apply this approach to the α_MI domain (∼200 amino acids), 16 segments were switched (20, 23). Site-directed mutagenesis of the α_MI domain was performed using QuikChange Site-Directed Mutagenesis Kit (Stratagene). The mutations introduced and the mutagenic primers used have been reported (23). The appropriate DNA sequence of the entire I domain (from I¹³⁹ to A³³²) was confirmed for each mutant before transferring back into the α_M subunit cDNA.

The expression vector pcDNA3.1 (Invitrogen) was used for cloning α_M, α_L, and β₂ from human leukocyte cDNA library. 293 cells were transfected using the Lipofectamine 2000 reagent (Invitrogen) with 24 μg DNA/vessel for 4 h, according to the manufacturer's instructions. After transfection, the medium was replaced with full growth medium. The functional assays were prepared 48 h after transfection.

FACS^® analyses were performed to assess the expression of wild-type and mutated forms of α_Mβ₂ and α_Lβ₂ on the surface of transfected 293 cells, as previously described (11). Platelet P-selectin expression was assessed using FITC-conjugated AK-4 or isotype control (BD Biosciences).

Neutrophils from wild-type (Mac-1^+/+) and Mac-1–deficient (Mac-1^−/−; 30) C57Bl/J6 mice were harvested and purified from the peritoneal cavity after the intraperitoneal injection of 1 ml sterile 3% thioglycollate broth, as previously described (11).

Venous blood was obtained from volunteers who had not consumed aspirin or other nonsteroidal antiinflammatory drugs for at least 10 d and was anticoagulated with 13 mM trisodium citrate, which also contained 100 nM prostaglandin E₁. Platelet-rich plasma was prepared by centrifugation at 150 g for 10 min. Gel-filtered platelets were obtained by passage of platelet-rich plasma over a Sepharose-2B column in calcium-free Tyrode's-Hepes buffer containing 100 nM prostaglandin E₁, as previously described (11).

Adherent cells were assayed by loading 293 cells and thioglycollate-elicited murine neutrophils with 1 μM 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein, acetoxymethyl ester (BCECF AM), according to the manufacturer's instructions (Molecular Probes). 10⁵ cells/well were placed in 96-well microtiter plates coated with 10 μg/ml sGP Ibα or 10 μg/ml fibrinogen and blocked with 0.2% gelatin. Adhesion was stimulated with 20 ng/ml PMA or 5 μg/ml of the β2-stimulating mAb KIM 127. Plates were washed and adhesion was quantified by measuring the fluorescence of BCECF AM–loaded cells using a Cytofluor II fluorescence microplate reader (PerSeptive Biosystems). The effect of anti-α_M mAb on adhesion was assessed by preincubating cells with 10 μg/ml LPM19c. The effect of peptides M1–M8 on adhesion was investigated by incubating the indicated peptide with sGP Ibα–coated wells for 30 min at 37°C before the addition of cells. Data are expressed as percent adhesion of control treatment.

Neutrophil adhesion to surface-adherent platelets was investigated as previously described (11). The effect of α_M peptides on leukocyte adhesion to platelets was examined by preincubating surface-adherent platelets with peptide (1–1,000 nM) or vehicle for 30 min at 37°C. Data are expressed as percent adhesion of control treatment.

The cDNAs of α_MI domain (675 nucleotides, R¹¹⁵–S³⁴⁰), α_LI domain (630 nucleotides, P¹²⁰–S³³⁰), and mouse α_MI domain (675 nucleotides, L¹¹⁵–S³⁴⁰) were cloned and inserted into the pGEX-5X-3 expression vector (22). All wild-type and mutant I domains were expressed as glutathione S-transferase (GST) fusion proteins. Mutations were created in these I domains and intact α_L by oligonucleotide-directed mutagenesis. The selective introduction of the desired mutations into the I domains was confirmed by DNA sequence analyses. The GST-I domain fusion proteins were purified by adsorption onto glutathione-Sepharose 4B (Amersham Biosciences) and eluted with buffer containing 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM glutathione, 2.5 mM CaCl₂. Such preparations of the fusion proteins were ∼90% pure as assessed by SDS-PAGE.

To test the interaction of the wild-type α_M, α_L, and chimeric I domains with biotinylated GP Ibα, 96-well plates (Immulon 4BX; Dynex Technologies Inc.) were coated with the I domains at 50 μg/ml and blocked with 2% BSA. sGP Ibα in 20 mM Tris-HCl, pH 7.6, containing 100 mM NaCl and 2 mM CaCl₂, was added to the wells and incubated for 1 h at 37°C. After washing, bound GP Ibα was detected using avidin-alkaline phosphatase and p-nitrophenyl phosphate. Background reaction on BSA-coated wells was subtracted.

Real-time protein–protein interactions were examined using surface plasmon resonance on a BIAcore 1000 (BIAcore AB). Immobilization of peptides was performed via thiol coupling onto a sensor chip C1 using 20 mM Tris, 150 mM NaCl, 2 mM MgCl₂, pH 7.4, as running buffer at a flow rate of 5 μl min⁻¹. Modified versions of peptides containing an amino-terminal cysteine residue were synthesized and then diluted in10 nM acetate buffer, pH 5.0 (coupling buffer). The peptide was loaded on the chip for 16 min and immobilization occurred via thiol disulfide exchange. Identical treatment was applied to the reference flow cell without peptide (control surface). For analysis, sGP Ibα diluted into running buffer was injected at a flow rate of 10 μl/min. Binding (resonance unit [RU]) was measured as a function of time(s). Binding data (after subtracting nonspecific background binding to the control surface) are presented as sensorgrams, binding curves, and Scatchard plots (31).

The laminar flow chamber used in this assay has been described (32). 25 mm² diameter glass coverslips (Assistent; Carolina Biological Supply Company) were coated with a solution of 20 μg/ml soluble P-selectin (provided by R. Camphausen, Genetics Institute, Cambridge, MA) and 40 μg/ml sGP Ibα for 2 h at room temperature. Nonspecific interactions were blocked with PBS containing 1% human serum albumin for 1 h at room temperature and with 0.1% Tween 20 immediately before use. The effect of M2 and scM2 peptides on THP-1 cell rolling and firm arrest was examined by incubating 50 μM of the peptide with the ligand-coated glass coverslip for 1 h at room temperature. To block PSGL-1 and α_Mβ₂ interactions, THP-1 cells were incubated with 20 μg/ml KPL-1 (BD Biosciences) or 20 μg/ml LPM19c mAb, respectively, for 20 min before perfusion. α_Mβ₂ was activated by treating cells with 10 μg/ml KIM127 for 5 min before perfusion. The rolling and firm adhesive events scored were ligand specific as confirmed in parallel determinations on control substrates coated with human serum albumin. Coverslips were inserted in the flow chamber and 0.6 × 10⁶/ml THP-1 cells were drawn across at an estimated shear stress of 0.75 dynes/cm² using a syringe pump (Harvard Apparatus). After 3 min of perfusion, the number of rolling and arrested cells was quantified in each of five random 10x fields (area ∼0.3 mm²) by an investigator blinded to treatment.

Data are presented as the mean ± SD or SEM. Groups were compared using the nonpaired t test. Rolling and arrest data were analyzed using repeated measure ANOVA with a Bonferroni corrective for multiple comparisons. P values <0.05 were considered significant.

Our previous report indicated that the α_MI domain serves as a recognition site for GP Ibα (11). The adhesion of α_Mβ₂-bearing cells to sGP Ibα was inhibited by LPM19c, an mAb that binds to the α_MI domain. To establish definitively that the α_MI domain serves as a recognition site for GP Ibα, we transfected 293 cells with wild-type α_Mβ₂, α_Lβ₂, or a chimeric α_Lβ₂ receptor that contained the I domain of α_Mβ₂ (i.e., α_L(Iα_M)β₂-transfected 293 cells). α_Lβ₂-transfected 293 cells did not adhere to sGP Ibα. In contrast, α_L(Iα_M)β₂-transfected 293 cells adhered robustly to sGP Ibα in a manner similar to α_Mβ₂-transfected 293 cells, indicating that the α_MI domain is required for adhesion to sGP Ibα. These data suggest that although α_Mβ₂ and α_Lβ₂ are highly homologous integrins, the lack of binding of sGP Ibα to α_Lβ₂ might be the result of sequence and/or structural differences.

As the first step to define the binding site for GP Ibα within the α_MI domain, mutant cell lines, each expressing a mutant α_Mβ₂ in which a short α_MI domain sequence, corresponding to a structural unit in the crystal structure (8, 9), was replaced for the corresponding region of the α_LI domain, were tested for their adhesion to immobilized sGP Ibα. 16 segments were switched from their original α_MI domain to their counterparts in α_L (16, 20, 23). These segment swaps are placed throughout the P¹⁴⁷–Q³¹⁴ region of the α_MI domain, D¹³²–A³¹⁸, and are intended to cover its entire hydrated surface (8, 9). For such experiments to be readily interpretable, cell lines expressing high and similar levels of wild-type and mutant receptors were selected by cell sorting using mAbs OKM1 and IB4 followed by cloning by limiting dilution. The expression of the receptors as assessed by the mean fluorescence intensity (MFI) in FACS^® analyses differed by less than twofold from the internal wild-type control (not depicted).

The results of adhesion of 16 mutants to sGP Ibα are summarized in Fig. 1. The data are expressed as the percent adhesion of the wild-type α_Mβ₂-expressing cells to sGP Ibα. Substitutions for the following regions of α_MI domain abrogated adhesion: α_M(P¹⁴⁷–R¹⁵²), α_M(M¹⁵³–T¹⁵⁹), α_M(E¹⁶²–L¹⁷⁰), α_M(P²⁰¹–G²⁰⁷), α_M(R²⁰⁸–K²¹⁷), α_M(K²⁴⁵FG²⁴⁷), α_M (ΔD²⁴⁸–Y²⁵²), α_M(E²⁵³–R²⁶¹), α_M(D²⁷³–K²⁷⁹), α_M(R²⁸¹–I²⁸⁷), and α_M(F²⁹⁷–T³⁰⁷). These cell lines form a group of negative mutants. As the essential control, the α_Lβ₂-expressing cells adhered poorly to sGP Ibα, consistent with lack of interaction of α_Lβ₂ with sGP Ibα.

The lack of adhesion of these mutants was not the result of decreased surface expression of the receptor because there was no correlation between the level of adhesion and expression. Specifically, surface expression of the α_M(P²⁰¹–G²⁰⁷) mutant was 1.5-fold higher than that of the cells expressing the wild-type α_Mβ₂ cells, but adhesion was abrogated completely. Surface expression of negative mutants, α_M(P¹⁴⁷–R¹⁵²), α_M(R²⁰⁸–K²¹⁷), α_M(E²⁵³–R²⁶¹), and α_M(D²⁷³–K²⁷⁹) was very similar to that of the wild-type receptor. Adhesion of α_M(E¹⁷⁸–T¹⁸⁵) and ΔE²⁶²G²⁶³ was partially affected. The maximal level of adhesion to sGP Ibα reached 69 and 64%, respectively, of wild-type α_Mβ₂ cells. These two receptors were classified as intermediate mutants.

The following mutants supported the same or higher levels of adhesion to sGP Ibα compared with the cells expressing the wild-type α_Mβ₂ receptor: α_M(Q¹⁹⁰–S¹⁹⁷), α_M(K²³¹NAF²³⁴), α_M(E²⁶²–G²⁶³), and α_M(Q³⁰⁹–E³¹⁴). The cell lines exhibiting adhesion similar to or greater than the wild-type α_Mβ₂ cells were identified as positive mutants. Two of these positive mutants, α_M(Q¹⁹⁰–S¹⁹⁷) and α_M(K²³¹NAF²³⁴), bound sGP Ibα better than wild-type α_Mβ₂. Previous analyses using these mutants showed that they bound neither an α_Mβ₂ activation-independent ligand, NIF (20), nor α_Mβ₂ activation-dependent ligands, C3bi (17) and fibrinogen (23), to a greater extent than wild-type receptor. Thus, the enhanced binding of these mutants appears to be selective for sGP Ibα recognition. In further studies, we found that binding of the β₂ activation–specific reporter mAb 24 was similar for wild-type and the α_M (Q¹⁹⁰–S¹⁹⁷) and α_M(K²³¹NAF²³⁴) mutants. In contrast, increased binding of the α_M activation–specific reporter CBRM1/5 to α_M(K²³¹NAF²³⁴) compared with wild-type α_Mβ₂ and α_M(Q¹⁹⁰–S¹⁹⁷) suggests possible conformation change of the α_M(K²³¹NAF²³⁴) mutant (not depicted). Together, these observations suggest that the activation states of α_Mβ₂ may reflect the reporter ligand or mAb used, and these two mutants might be functionally activated with respect to sGP Ibα recognition.

In subsequent analyses, we focused on the 11 negative mutants. A series of peptides (M1–M8) corresponding to the wild-type α_MI domain sequences were synthesized and tested for their ability to interact with sGP Ibα. Because the sequences within several of the negative mutants were contiguous, the following linear peptides spanning these sequences were synthesized: P¹⁴⁷–T¹⁵⁹ (designated M1), P²⁰¹–K²¹⁷ (M2), K²⁴⁵–R²⁶¹ (M3), D²⁷³–I²⁸⁷ (M4), F²⁹⁷–T³⁰⁷ (M5), Q¹⁹⁰–S¹⁹⁷ (M6), E¹⁶²–L¹⁷⁰ (M7), and E¹⁷⁸–T¹⁸⁵ (M8). All peptides correspond to negative mutant sequences except M6 and M8, which correspond to intermediate and positive mutant sequences, respectively.

Next, we examined the effect of these α_MI domain peptides on α_Mβ₂-dependent adhesion to sGP Ibα or fibrinogen (Table I). 50 μM peptides were preincubated in ligand-coated wells, α_Mβ₂-expressing 293 cells were added, and adhesion was stimulated with KIM 127. Mac-1 293 cells adhered to sGP Ibα and fibrinogen, and this adhesion was blocked by the anti-CD11b mAb LPM19c, indicating that adhesion is predominantly α_Mβ₂ dependent (Table I). Peptide M2 inhibited adhesion to sGP Ibα (percent inhibition = 74 ± 13; P < 0.01). Interestingly, M2 had no effect on cellular adhesion to fibrinogen. All other peptides had minimal or no effect on α_Mβ₂-expressing 293 cell adhesion to sGP Ibα or fibrinogen. M2, but not a scrambled version (scM2), blocked α_Mβ₂-dependent adhesion to sGP Ibα in a dose-dependent manner (IC₅₀ = 20 μM; Fig. 2). Taken together, these observations suggest that the α_MI domain sequence corresponding to M2 (P²⁰¹–K²¹⁷), which spans an exposed loop and amphipathic α4 helix in the three-dimensional structure of the α_MI domain, contributes to α_Mβ₂ binding of GP Ibα.

To examine direct interactions between GP Ibα and M2, we used an in vitro real-time binding assay using surface plasmon resonance (BIAcore). M2 and scM2 peptides were modified by addition of an amino-terminal cysteine and immobilized via thiol coupling on a C1 sensor chip. Cys-M2, but not Cys-scM2, inhibited α_Mβ₂-dependent adhesion to sGP Ibα, verifying that cysteine addition did not adversely affect peptide structure (Table I). Increasing concentrations of purified sGP Ibα in running buffer containing 2 mM MgCl₂ were injected over the chip surface and the sensorgrams were recorded. As shown in Fig. 3, the sensorgrams detected little interaction between sGP Ibα and control chips or chips containing immobilized scM2. Significant interaction responses were detected for sGP Ibα with M2. Apparent equilibrium dissociation constant was estimated from the equilibrium resonance signal as a function of analyte (GP Ibα) concentration (kD ∼20 μM) and calculated by Scatchard analysis (kD = 17 μM). Similar results were obtained when purified sGP Ibα in running buffer containing 2 mM CaCl₂ were injected over the chip surface (not depicted). Thus, BIAcore analysis revealed real-time, direct binding for bimolecular interactions between GP Ibα and M2.

To begin localization of specific amino acid residues within the α_MI domain involved in GP Ibα recognition, recombinant fragments containing wild-type α_MI domain, R¹¹⁵–S³⁴⁰, wild-type α_LI domain, P¹²⁰–S³³⁰, and a mutant α_MI domain were expressed as GST fusion proteins in Escherichia coli. The mutant α_MI domain contained a swap of the P²⁰¹–K²¹⁷ segment, which we implicated in sGP Ibα binding to the corresponding residues of the α_LI domain. The recombinant proteins were purified from the bacterial lysates on glutathione-Sepharose. A facile binding assay for quantifying sGP Ibα binding to the recombinant I domains was developed by measuring the binding of biotinylated sGP Ibα to the recombinant I domains immobilized onto 96-well plastic plates. As shown in Fig. 4, GP Ibα exhibited minimal reaction with α_LI domain. Binding was observed and was similar with both human and mouse α_MI domains. With the α_MI domains, binding was dependent on the sGP Ibα concentration with 50% maximal binding observed at 50 nM sGP Ibα, which was used in subsequent experiments. As shown in Fig. 5 A, the interaction of the α_MI domain “swap” mutant with biotinylated sGP Ibα was greatly diminished compared with wild-type α_MI domain and similar to that of the α_LI domain. Essentially, no specific binding of sGP Ibα to this mutant I domain was detected. This result is consistent with our BIAcore binding experiments showing a direct interaction between sGP Ibα and immobilized M2 peptide corresponding to this P²⁰¹–K²¹⁷ segment.

To begin to identify the individual residues within the P²⁰¹–K²¹⁷ segment that mediated sGP Ibα recognition, a series of six triple mutants were created. In each of these triple mutants, a set of three consecutive amino acids within the α_MI domain was changed to the corresponding α_L residues. If the α_M and α_L residues were the same, the amino acid was mutated to alanine. After the DNA sequence of each mutant I domain was confirmed, it was expressed in E. coli and purified on glutathione-Sepharose. When analyzed by SDS-PAGE, each mutant migrated as a single band of ∼52 kD (not depicted). sGP Ibα binding to each triple mutant was then assessed. The results in Fig. 5 A show the binding of each of the six triple mutants to 50 nM biotinylated sGP Ibα. Of the six triple mutants, three, H²¹⁰–A²¹², T²¹³–I²¹⁵, and R²¹⁶–K²¹⁷, showed a significant reduction in sGP Ibα binding. Three mutants, P²⁰¹–T²⁰³, Q²⁰⁴–L²⁰⁶, and G²⁰⁷–T²⁰⁹, did not impair binding to GP Ibα. These results, taken in light of the observation that mutating the entire region spanning P²⁰¹–G²⁰⁷ abolished cell adhesion to sGP Ibα (Fig. 1), suggest that the inactivation of sGP Ibα binding requires alteration of more than one residue within the P²⁰¹–G²⁰⁷ (i.e., no single residue within the triple mutants, P²⁰¹–T²⁰³, Q²⁰⁴–L²⁰⁶, and G²⁰⁷–T²⁰⁹, decreases affinity detectably), or the conformation of the P²⁰¹–G²⁰⁷ segment, which corresponds to a portion of a loop within the α_MI domain, is necessary for binding, and, again, no single substitution alters the conformation of this region sufficiently to prevent binding.

Next, within the three triple mutants with reduced sGP Ibα binding, each of the three amino acids was mutated individually to the corresponding residue in α_LI domain or in case of identical residues in two I domains, the amino acid was replaced with an alanine. After confirming the DNA sequences of these single mutants, each of the GST fusion proteins was purified. The mutant fusion protein carrying the K²¹⁷Y substitution was insoluble. Therefore, K²¹⁷ was mutated to Ala, and this α_MI domain with a K²¹⁷A substitution was readily purified, yielding a total of eight single mutants. The capacity of the eight single mutants to bind sGP Ibα is summarized in Fig. 5 B. Within each of the three triple mutants with reduced sGP Ibα recognition, only one of the three single mutants exhibited reduced sGP Ibα binding. These three single mutants showing reduced binding were T²¹¹A, T²¹³G, and R²¹⁶N.

Loss of GP Ibα binding function in these single mutants could reflect direct involvement of the specific residues in GP Ibα binding or conformational perturbation of the resulting α_MI domain due to substitutions at these positions. A gain in function approach was used to distinguish between these possibilities by introducing the identified point mutations into the α_LI domain. T²¹¹ is conserved in both α_MI and α_LI domains. Therefore, chimeric I domains containing either a single G²¹³T substitution or two substitutions, G²¹³T and N²¹⁶R, in the α_LI domain backbone were created. These mutant α_LI domains were expressed as GST fusion proteins, purified, and their sGP Ibα binding properties were evaluated. The chimeric I domain harboring both the G²¹³T and N²¹⁶R mutations bound sGP Ibα with an affinity substantially greater than wild-type α_LI domain and the I domain containing the single G²¹³T mutation (Fig. 6). Indeed, the binding capability of double substituted chimeric I domain was comparable to the wild-type α_MI domain. To confirm that these two amino acid residues are sufficient to impart sGP Ibα recognition to the mutated α_LI domain, we performed additional BIAcore binding assays with immobilized α_LI domain peptide (C²⁰¹HVKHMLLTNTFGAINY²¹⁷, termed C-L2) corresponding to the P²⁰¹–K²¹⁷ sequence within α_MI domain and the double substituted mutant C²⁰¹HVKHMLLTNTFTAIRY²¹⁷ (C-muL2). Binding assays were also performed with a mutant M2 peptide containing T²¹¹A, T²¹³G, and R²¹⁶N substitutions (C²⁰¹PITQLLGRTHAAGGINK²¹⁷, termed C-muM2) corresponding to the three single mutants showing reduced binding in the purified mutant I domain binding assays (Fig. 5 B). The sensorgrams detected little interaction between 50 μM sGP Ibα and chips containing immobilized C-L2 (RU = 10) or C-muM2 (RU = 25). Significant interaction responses were detected for sGP Ibα with C-muL2 (RU = 224).

The role of T²¹³ and R²¹⁶ in the GP Ibα binding function of the intact receptor was investigated. The T²¹³G and R²¹⁶N substitutions were introduced into the cDNA for the α_L subunit using site-directed mutagenesis and coexpressed with the cDNA for the β₂ subunit in 293 cells. Wild-type α_Lβ₂ and α_Mβ₂ were also transiently expressed in these cells as controls. 48 h after transfection, the cells were detached from tissue culture plates, and receptor expression levels were evaluated by FACS^®. α_M expression was evaluated with OKM1, β₂ expression with IB4, and α_L expression with TS1/22. The expression levels of both the α and β subunits of the integrins were comparable (not depicted). Next, the function of the receptors was assessed by evaluating their adhesion to immobilized sGP Ibα. PMA, Mn²⁺, or their combination were used to activate the integrins on the cells. As shown in Fig. 7, mock-transfected cells or cells expressing wild-type α_Lβ₂ showed little adhesion to sGP Ibα under all conditions. When α_Mβ₂-expressing cells were stimulated with PMA, Mn²⁺, or their combination, their adhesion to sGP Ibα increased markedly compared with the nonstimulated cells. The chimeric α_Lβ₂ cells adhered to sGP Ibα considerably better than the α_Lβ₂ cells or mock-transfected cells. The adhesion of the chimeric α_Lβ₂ cells approached that of the wild-type α_Mβ₂ cells. We considered whether inserting these amino acid residues into α_L might affect the activation state of chimeric α_Lβ₂ such that the increase in binding might reflect allosteric changes in the integrin. Accordingly, we assessed the activation states of the chimeric integrin using MEM-83, an mAb that reacts with activated α_Lβ₂ (33). In FACS^® analyses, MEM-83 failed to react with either the α_Lβ₂ or the chimeric α_Lβ₂ transfectants in the unstimulated state (MFI = 14 and 10, respectively), but reacted well and equivalently with both transfectants upon stimulation with the combination of Mn²⁺ and PMA (MFI = 79 and 83, respectively). Thus, the activation states of both the α_Lβ₂ and chimeric α_Lβ₂ transfectants, with and without stimulation, were similar.

Having previously reported that α_Mβ₂ and GP Ibα facilitate the heterotypic interaction between leukocytes and platelets, we turned to examining the effect of M2 on neutrophil adhesion to platelets. Thioglycollate-elicited neutrophils were added to wells containing surface adherent platelets and adhesion was stimulated by the addition of PMA to each well. We verified by FACS^® analysis that such PMA treatment up-regulated P-selectin expression 29.9-fold in platelets purified and prepared with PGE1 as described in Materials and Methods. Wild-type (Mac-1^+/+) neutrophils bound to adherent platelets (Fig. 8). In contrast, Mac-1–deficient (Mac-1^−/−) neutrophils demonstrated markedly reduced adhesion to platelets (percent wild-type adhesion = 18.2 ± 8.0) and adhesion of wild-type neutrophils was blocked by the rat anti–mouse α_Mβ₂ mAb M1/70 (percent wild-type adhesion = 29.9 ± 19.8), indicating that under these conditions, neutrophil adhesion to platelets is largely α_Mβ₂ dependent, although a complimentary role for P-selectin/PSGL-1 is likely operative. M2, but not a scrambled version, inhibited dose dependently (IC₅₀ = 30 nM) wild-type neutrophil adhesion to platelets. Taken together, these observations indicate that under these experimental conditions, neutrophil adhesion to platelets is mediated primarily by the α_MI domain sequence P²⁰¹–K²¹⁷.

To evaluate the potential for M2 to modulate the adhesion of blood cells under flow, we perfused THP-1 cells that express Mac-1 over coverslips coimmobilized with soluble P-selectin and GP Ibα using a parallel plate flow chamber system (0.75 dynes/cm²). The number of rolling and arrested cells was quantified on five random fields after 3 min of perfusion. THP-1 cells rolled and arrested on coverslips cocoated with P-selectin and sGP Ibα and the number of rolling (control vs. scM2, P > 0.05) or arrested (control vs. scM2, P > 0.05) cells was unaffected by scM2 (Fig. 9 A). In contrast, M2 peptide inhibited THP-1 cell arrest, thereby increasing the number of rolling cells visualized. The effect of M2 on cell adhesion was similar to treatment with LPM19c, an anti-CD11b mAb that blocks α_Mβ₂-dependent binding to sGP Ibα, thereby confirming the involvement of α_Mβ₂ in cell adhesion. Both cell rolling and arrest were abolished by treating cells with KPL-1, an anti–PSGL-1 mAb that blocks P-selectin binding. The effect of M2 on cell arrest was also quantified. Thus, on coverslips incubated with vehicle or scM2, 64 and 66% of THP-1 cells arrested, respectively (Fig. 9 B). After treatment with M2 peptide only 34% of cells arrested (P < 0.05). Similarly, after preincubation of THP-1 cells with LPM19C, only 33% of cells arrested (P < 0.05). KPL-1 antibody treatment abrogated almost all rolling and subsequent arrest (<10%; P < 0.01).

In this study, we have identified the P²⁰¹–K²¹⁷ segment, which spans an exposed loop and amphipathic α4 helix in the three-dimensional structure of the α_MI domain, as the binding site for platelet GP Ibα. This conclusion is supported by the following data: (a) mutant cell lines in which the α_MI domain segments P²⁰¹–G²⁰⁷ and R²⁰⁸–K²¹⁷ were switched to the homologous α_LI domain segments failed to support adhesion to sGP Ibα, (b) mutation of amino acid residues within P²⁰¹–K²¹⁷, H²¹⁰–A²¹², T²¹³–I²¹⁵, and R²¹⁶–K²¹⁷ resulted in the loss of the binding function of the recombinant α_MI domains to biotinylated sGP Ibα, (c) synthetic peptides duplicating the P²⁰¹–K²¹⁷, but not scrambled versions, directly bound sGP Ibα and inhibited α_Mβ₂-dependent adhesion to sGP Ibα and adherent platelets, and (d) grafting key amino acids within the P²⁰¹–K²¹⁷ sequence onto α_L converted α_Lβ₂ into a GP Ibα binding integrin.

By virtue of binding diverse ligands including, among others, fibrin(ogen) (34, 35), ICAM-1 (36), factor X (37), C3bi (34), high molecular weight kininogen (38), and heparin (4), α_Mβ₂ regulates important leukocyte functions including adhesion, migration, coagulation, proteolysis, phagocytosis, oxidative burst, and signaling (30, 39–42). However, these ligands do not account for all of α_Mβ₂'s adhesive interactions. Although previous studies have shown that α_Mβ₂ directly facilitates the recruitment of leukocytes at sites of platelet and fibrin deposition (5), the precise platelet counter receptors, including GP Ibα (11) and JAM-3 (43), have been elucidated only recently.

In this study, we have identified key elements of the binding site for GP Ibα within the α_MI domain. The strategy to define the ligand binding site was based on the difference in the sGP Ibα binding properties of the α_MI and α_LI domains and entailed four complimentary approaches. In the first approach, a series of homologue-scanning mutants, used previously to map the binding regions for NIF, iC3b, and fibrinogen (16, 20, 23), were screened for adhesion to sGP Ibα. In these mutants, 16 segments at the hydrated surface of the α_MI domain were replaced with the corresponding segments from the homologous α_LI domain, which does not bind GP Ibα. 11 mutants lacked the ability to support adhesion sGP Ibα, and alteration of two other regions, α_M(E¹⁷⁸–T¹⁸⁵) and α_M(ΔE²⁶²G²⁶³), resulted in the partial loss of adhesive function. Thus, the initial insight provided by these mutant receptors indicated that the sGP Ibα binding interface within the α_MI domain was composed of several nonlinear sequences.

The second approach entailed the use of synthetic peptides duplicating the sequences of the critical segments in the α_M I domain. These analyses showed that two critical segments, α_M(P²⁰¹–G²⁰⁷) and α_M(R²⁰⁸–K²¹⁷), may contain amino acid residues that participate directly in binding GP Ibα because the peptide M2 that spanned P²⁰¹–K²¹⁷ bound sGP Ibα and inhibited α_Mβ₂-dependent adhesion to sGP Ibα and adherent platelets. The negative results for other peptides (M1, M3–M5, and M7) synthesized to correspond to other negative mutants (Fig. 1 and Table I) do not exclude a role for other α_MI domain segments in binding function. These segments may play an accessory role in ligand binding or the short peptides may simply not assume the appropriate conformation for recognition by the ligand. Finally, segments implicated in the binding of sGP Ibα by the negative mutants, other than P²⁰¹–K²¹⁷, may reflect interference by α_LI domain residues rather than residues that actually participate in binding. It is also possible that the activation of α_Mβ₂, rather than ligand binding per se, was affected adversely by the introduction of these α_L segments within α_MI domain. Because these same mutants have been used in previous studies (17, 20, 23) to analyze interaction of other activation-dependent ligands with α_Mβ₂, such effects on activation would need to specifically perturb GP Ibα recognition.

To obtain direct evidence that the α_M(P²⁰¹–K²¹⁷) sequence constitutes the functional binding site for the GP Ibα, we turned to a third approach using site-directed mutagenesis of the α_MI domain. Binding experiments with purified sGP Ibα and GST-I domain mutants provided the independent confirmation that the P²⁰¹–K²¹⁷ segment is important for sGP Ibα binding because mutations of these residues resulted in significant loss of sGP Ibα binding. These experiments also served to narrow further the binding region to H²¹⁰–K²¹⁷ and subsequently identified three single mutants showing reduced binding to sGP Ibα (T²¹¹A, T²¹³G, and R²¹⁶N). Of these, T²¹¹ in α_MI domain is conserved in the α_LI domain. Thus, it was the conversion of the residue to A that perturbed GP Ibα binding, suggesting that this substitution exerts a negative influence on the conformation of the α_MI domain that is required for GP Ibα recognition, rather than participating directly in ligand contact. The positioning of this residue on the interior of the α4-helix and facing the central core of the α_MI domain, according to the crystal structures (8, 9), supports this interpretation. In contrast, T²¹³ and R²¹⁶ are appropriately positioned, including when the recent crystal structure of the GP Ibα/vWF A domain (44) is used as a template.

In the fourth approach, the two amino acids (T²¹³G and R²¹⁶N substitutions) within the α_M(P²⁰¹–K²¹⁷) were grafted into the corresponding positions of the α_LI domain in the context of intact α_Lβ₂. As demonstrated in Figs. 6 and 7, this manipulation imparted GP Ibα binding capacity to the chimeric molecule. Thus, the role of amino acids within the α_M(P²⁰¹–K²¹⁷) sequence in GP Ibα binding, which initially was inferred from the loss in function experiments, was verified by the gain in function approach. Taken together, the four approaches substantiated independently the role of α_M(P²⁰¹–K²¹⁷) in GP Ibα binding and provided evidence that two to three critical residues participate in ligand docking.

To date, several lines of evidence have emphasized that multiple ligands share overlapping binding sites within α_Mβ₂, including the fact that one ligand (e.g., NIF) is capable of blocking the interaction of multiple ligands (C3bi, ICAM-1, fibrinogen) with the receptor (15, 45). However, although the binding sites for these ligands might be overlapping, they need not be identical (16, 17). Ustinov and Plow (22) have proposed a mosaic model in which many of the same loops and helices of the α_MI domain, on or near its MIDAS face, may engage ligands, but different amino acid residues within these structures may contact the ligands. Direct support for this mosaic model is seen within the K²⁴⁵–R²⁶¹ segment of the α_MI domain, which has been implicated in both NIF and fibrinogen recognition (23). Key contact amino acids in this loop for fibrinogen binding are F²⁴⁶, D²⁵⁴, and P²⁵⁷, whereas Y²⁵² and E²⁵⁸ are involved in NIF binding. This same α_MI domain segment is also involved in C3bi recognition by α_Mβ₂ (17).

Under the experimental conditions used in this study, which assayed the adhesion of activated neutrophils to surface-adherent platelets after vigorous washing, the predominant interaction between neutrophils and platelets appeared to be mediated by α_Mβ₂ binding to sGP Ibα, based on the ability of M2 to inhibit >80% of neutrophil adhesion. The enhanced potency of M2 with respect to inhibiting neutrophil adhesion to platelets (IC₅₀ = 30 nM) compared with BIAcore (kD = 17 μM) is possibly secondary to the fact that leukocytes express α_Mβ₂ and platelets express GP Ibα in their native conformations. In contrast, BIAcore experiments used sGPIbα with immobilized M2 peptide.

Our data do not rule out the possibility of additional platelet surface receptors for α_Mβ₂. Other potential α_Mβ₂ ligands present on the platelet membrane include fibrinogen (bound to GP IIb-IIIa; 34, 35), ICAM-2 (46), high molecular weight kininogen (38), and JAM-3 (43). However, a leukocyte–platelet interaction mediated by fibrinogen bridging between α_Mβ₂ and GP IIb/IIIa has been discounted by Ostrovsky et al. (47), who found that neither RGDS peptides nor the replacement of normal platelets with thrombasthenic platelets (i.e., lacking GP IIb/IIIa) affected the accumulation of the leukocytes on platelets. Although α_Mβ₂ binds ICAM-1, this receptor is not found on platelets. Platelets express a related receptor, ICAM-2 (48), but Diacovo et al. (5) have shown that ICAM-2 blockade has no effect on the firm adhesion of neutrophils on monolayers of activated platelets under flow. Santoso et al. (43) have reported recently that α_Mβ₂ may also bind to platelet JAM-3, cooperating with GP Ibα to mediate neutrophil–platelet adhesive contacts (43).

The present observations also suggest a possible target for therapeutic intervention. In particular, the specificity of M2 inhibitory action toward GP Ibα (i.e., noninhibitory toward fibrinogen) suggests that it might be possible to prevent leukocyte attachment to platelets by targeting GP Ibα without inhibiting other α_Mβ₂ functions. Our recent observations have identified α_Mβ₂ as a molecular determinant of neointimal thickening after experimental arterial injury that produces endothelial denudation and platelet deposition. We found that antibody-mediated blockade (49) or selective absence (50) of α_Mβ₂ impaired transplatelet leukocyte migration into the vessel wall, diminishing medial leukocyte accumulation and neointimal thickening after experimental angioplasty or endovascular stent implantation. Therefore, this study identifying the precise binding site responsible for α_Mβ₂–GP Ibα interaction might provide a molecular strategy for disrupting leukocyte–platelet complexes that promote vascular inflammation in thrombosis, atherosclerosis, and angioplasty-related restenosis.

This work was supported in part by grants from the National Institutes of Health (HL53993, HL36028, and HL65090 to F.W. Luscinskas, HL65967 and HL64796 to J.A. Lopez, HL66197 to E.F. Plow, and HL57506 and HL60942 to D.I. Simon). R. Ehlers received a post-doctoral research fellowship from Society for Thrombosis and Hemostasis Research, Germany. V. Ustinov received a post-doctoral research fellowship from The Ohio Valley Affiliate, American Heart Association (0120394B). R.M. Rao is an ARC Copeman Traveling Research Fellow.

Note added in proof. Hoffmeister et al. (Hoffmeister, K.M., E.C. Josefsson, N.A. Isaac, H. Clausen, J.H. Hartwig, and T.P. Stossel. 2003. Glycosylation restores survival of chilled blood platelets. Science. 301:1531–1534.) have recently reported that chilled platelets are cleared by hepatic macrophage α_Mβ₂ receptors via an interaction between platelet GP Iβα and a non–I-domain, lectin-binding site within a_Mβ₂.