Identification of an Active Sequence within the First Immunoglobulin Domain of Intercellular Cell Adhesion Molecule-1 (ICAM-1) That Interacts with Fibrinogen*

Monocytic cells bind fibrinogen (fg) through integrin (cid:97) M (cid:98) 2 . fg-bound monocytic cells demonstrate an en- hanced adhesion to endothelial cells, which is dependent on intercellular adhesion molecule-1 (ICAM-1). Our studies differentiate fg interactions with stimulated and resting endothelial cells, which are ICAM-1 dependent and independent, respectively. This report documents a direct interaction between fg and intact ICAM-1 and with a two-Ig domain form of ICAM-1. A small region within the first Ig domain of ICAM-1, ICAM-1-(8–21) (KVILPRGGSVLVTC), was identified to interact with fg in a specific and selective manner. ICAM-1-(8–21) bound to plasmin-derived fg fragments X , D100, and D80 but not to fragment E. Consistent with this finding, fg (cid:103) -chain peptide, fg- (cid:103) -117–133, blocked fg interaction with ICAM-1-(8–21). ICAM-1-(8–21) peptide and antibod- ies directed against ICAM-1-(8–21) also blocked the adhesion and binding of ICAM-1-bearing Raji cells with fg. ICAM-1-(8–21) and fg- (cid:103) -117–133 are likely to be one of the contact pairs mediating fg-ICAM-1 interactions. conformation using Homology. conformation-gener-ating displayed give preliminary model. carried The model was subjected to 200-steps steepest decent optimization. The final structure reached using the conjugate gradient method until the maxi- deviations were (cid:44) 0.1 kcal/mol Å.

While fibrinogen (fg) 1 is a plasma protein and intracellular adhesion molecule 1 (ICAM-1) is primarily a cell surface protein, both play central roles in cell-cell interactions (1,2). fg, a dimeric 340-kDa molecule, circulates in blood at 2-3 mg/ml. It is composed of three pairs of nonidentical polypeptide chains, organized into a central E and two peripheral D domains (3). A cell-cell interaction that is critically dependent on fg is platelet aggregation (4), which involves the binding of fg to platelet integrin, ␣ IIb ␤ 3 (5). The extreme COOH-terminal aspect of the ␥ chain, ␥-406 -411, is involved in the recognition of fg by ␣ IIb ␤ 3 (6). fg also interacts with other integrins, including ␣ v ␤ 3 and ␣ M ␤ 2 (7,8). ␣ v ␤ 3 is expressed on many cell types, including endothelial cells (EC). fg recognition by this receptor is inhib-ited by Arg-Gly-Asp (RGD)-containing peptides, and fg has two RGD sequences within its A␣ chain (3). Evidence is emerging to indicate that the interaction of fg with ␣ M ␤ 2 (Mac-1) may also be important for cell-cell interactions. fg bound to ␣ M ␤ 2 on leukocytes can facilitate the bridging of these cells to EC (1), thereby potentially contributing to inflammatory response. Recognition of fg by ␣ M ␤ 2 involves a ␥-chain sequence within the D domain, ␥-191-202 (9).
Recently, a direct interaction between fg and ICAM-1 has been demonstrated (1). A mechanism can be envisioned in which fg bound to ␣ M ␤ 2 on leukocytes bridges to ICAM-1 on EC, thus mediating adhesion between the two cell types. This interaction has been implicated in leukocyte transmigration through endothelium (19). As fg-ICAM-1 interactions may have important pathophysiological ramifications, we have sought to define the molecular basis for their recognition. In this study, we describe the identification of a small sequence within ICAM-1 that is critical for its interaction with fg, specifically with ␥-117-133 of fg. Furthermore, we demonstrate that ICAM-1 can mediate the adhesion of EC to fg, suggesting that this interaction not only may influence leukocyte transmigration but also may directly affect EC function.
Cells-EC were harvested from human umbilical cords (25). EC were plated on tissue culture-treated polystyrene (Costar Corp., Cambridge, MA) coated with 1.0 g/cm 2 human fibronectin (Boehringer Mannheim, Indianapolis, IN) and grown in Dulbecco's modified Eagle's medium F-12 (BioWhittaker, Walkersville, MD) containing 15% fetal calf serum and 180 g/ml EC growth supplements (Clonetics, San Diego, CA). Cells were initially grown in T75 culture flasks until confluent and subcultured at a 1:3 density. Only EC from passages 2-4 were used in this study. Twenty-four hours prior to assay, EC were stimulated by addition of 10 ng/ml TNF-␣. Raji cells were obtained from ATCC (Rockville, MD) and grown in RPMI 1640 (BioWhittaker, Walkersville, MD) containing 7.5% fetal calf serum and 1.0 mM glutamine. Raji cells grow in suspension and were split (1:3) every 3-4 days.
Synthetic Peptides-Peptides were synthesized by solid phase synthesis on an Applied Biosystems model 430A peptide synthesizer (Foster City, CA), using N-(9-fluorenyl)methoxycarbonyl chemistry. Peptides were cleaved from the resin and deprotected using crystalline phenol and thioanisole. The cleaved peptides were purified on high pressure liquid chromatography (HPLC), and purity was confirmed by either amino acid composition or mass spectrometry (26,27).
Recombinant Proteins-The full-length five-domain form of ICAM-1 and truncated first two domains of ICAM-1 (D 1 D 2 ICAM-1) were expressed in the baculovirus expression system using the viral vector Pvl1392 (InVitrogen, San Diego, CA) into which the full-length or truncated human cDNA encoding for ICAM-1 was ligated (28). Virus containing vector DNA was plaque purified and then transfected into Spodoptera frugiperda (SF9) insect cells. Cells were grown in Grace's insect cell media (Life Technologies, Grand Island, NY) for 4 days, after which they were washed with Tris-buffered saline (TBS) and lysed in 1% octylglucoside. The lysates were tested by enzyme-linked immunosorbent assay for ICAM-1 protein using a specific polyclonal antiserum. Full-length ICAM-1 and D 1 D 2 ICAM-1 migrated as single species at 60 and 31 kDa, respectively (29). The apparent molecular mass of the engineered proteins were smaller than expected due to underglycosylation. Because D 1 D 2 ICAM-1 contains four of the eight potential glycosylation sites in ICAM-1, the observed molecular mass of D 1 D 2 ICAM-1 (31 kDa) is as predicted. The expressed proteins were recognized by ICAM-1 antibodies. In Western blot analyses, these antibodies interacted strongly with a 95-kDa protein (ICAM-1) expressed on TNF-␣-stimulated EC, characteristic of ICAM-1 expression on EC (2). Full-length ICAM-1 was purified by preparative isoelectric focusing (Bio-Rad) using a pH gradient of 6 -8. The purity of ICAM-1 was Ͼ80%, by silver staining of SDS-gels.
Fg Binding to ICAM-1 Peptides-Peptides (100 l) at 0.5 mM in TBS were coated onto Falcon 3911 microtiter plates (Becton Dickinson Labware, Oxnard, CA) for 16 h at 4°C. The peptide solution was removed, and the plates were washed and blocked with 1% bovine serum albumin (BSA). 125 I-labeled fg (20 nM) in TBS, containing 1.0 mM CaCl 2 , was applied to the wells for 4 h at 37°C. Plates were washed three times with TBS, and the individual wells were counted in a gamma counter (Iso-Data 20/20, San Marcos, CA).
Ligand Blotting-Detergent lysates of SF9 cells transfected with viral vector encoding full-length five-domain and truncated two-domain forms of ICAM-1 were separated by SDS-polyacrylamide gel electrophoresis (34) and electrophoretically transferred to polyvinylidene difluoride Immobilon membranes (Millipore, Bedford, MA) using 10 mM CAPS in 10% methanol at pH 11. The membranes were probed with 125 I-fg (20 nM) in TBS containing 0.1% BSA and 1 mM MgCl 2 , washed once with TBS containing 0.05% Tween 20 and four times with TBS alone, and then dried and autoradiographed using X-Omat films (Eastman Kodak Company).
Adhesion Assays-Adhesion of EC or Raji cells to immobilized fg was performed as described previously (27). Briefly, cells were washed three times gently with Hanks' balanced salt solution (HBSS) without divalent ions and labeled with 0.5 ml of 51 Cr (1 mCi/ml) at 22°C for 30 min. The radiolabeled cells were washed in HBSS, and the final cell pellet was suspended in HBSS containing 0.1% BSA and 2 mM CaCl 2 to give 7.5 ϫ 10 5 cells/ml. Cells were allowed to adhere to fg-coated wells (1.0 g/well) for 30 min at 37°C. The nonadherent cells were aspirated, and the adherent population was lysed in a solution containing 2% SDS and 0.2 M NaOH for 30 min, transferred to vials containing 3 ml of BioSafe scintillation liquid (Research Products International, IL), and counted in a beta counter (LS 3801, Beckman Instruments, Fullerton, CA).
Fg Binding to Raji Cells-Fg binding was performed as described previously (31,35). Briefly, cells were washed three times in HBSS. 125 I-fg (300 nM) was allowed to bind 5 ϫ 10 6 /ml Raji cells in a total volume of 200 l in HBSS containing 0.1% BSA for 30 min at 22°C. Following incubation, the cell-bound radioactivity was separated from unbound material by passing through a cushion of 20% sucrose by centrifugation at 16,000 rpm for 2.5 min in an Eppendorf centrifuge. The cell pellet was counted on a gamma counter, and molecules of fg were bound per cell was calculated as described (31,35).
Fluorescence-activated Cell Sorting (FACS)-Resting and TNF-␣stimulated EC were isolated from culture flasks by brief trypsin treatment and washed twice in Dulbecco's phosphate-buffered saline. Cells were resuspended in a staining medium of HBSS containing 2.0 mM CaCl 2 , 2.0 mM MgCl 2 , 10 mM HEPES (pH 7.4), and 0.1% BSA and incubated at 4°C for 30 min with 5.0 g/ml of control mouse IgG, anti-␣ v ␤ 3 mAb LM 609, anti-ICAM-1 mAbs QE2, or LB-2. Cells were centrifuged through a cushion of fetal calf serum and resuspended in staining medium containing 50 g/ml fluorescein isothiocyanate-conjugated goat anti-mouse IgG antibodies (Zymed Laboratories, South San Francisco, CA). Cells were incubated for 30 min at 4°C, then centrifuged, and resuspended in staining medium. Cell-bound antibodies were detected by FACScan using the LYSIS program (Becton Dickinson).
Molecular Modeling of the First Ig Domain of ICAM-1-Software InsightII (Biosym Technologies Inc., San Diego, CA) and Homology were used to carry out the modeling. The first Ig domain of VCAM-1 was taken as a template to construct the backbone conformation of ICAM-1. The structure of VCAM-1 was taken from the Protein Data Bank (36). Sequence alignment of ICAM-1 and VCAM-1 was performed to align the two molecules (37). The conformation of the ICAM-1 first-Ig domain backbone was determined in two steps using Homology. Initially, the structurally conserved regions were built. Because ICAM-1 is thought to have the same secondary structures as VCAM-1, we took the ␤ strands and the integrin-binding motif as structurally conserved regions. The conformation of structurally conserved regions were determined by copying the coordinates of the corresponding backbone atoms in VCAM-1. The other parts of the molecule were treated as loops. The conformations of these loops were determined by searching the protein conformation data base in the Biosym software. A conformation-generating procedure was used that fit the loops into the structurally conserved regions. The best ten conformations for each loop were displayed and selected manually to find the most suitable ones. The side chains were adjusted to remove bumps to give a preliminary model. Energy minimization was carried out to refine this structure. The model was subjected to 200-steps steepest decent optimization. The final structure was reached by using the conjugate gradient method until the maximum deviations were Ͻ0.1 kcal/mol Å.

ICAM-1-Fibrinogen Interactions in Endothelial Cell
Adhesion-Previous studies have yielded conflicting results regarding the roles of the ␣ v ␤ 3 integrin and ICAM-1 in mediating fg binding with EC (1, 38). The described interactions have ranged from being entirely blocked by RGD peptides and ␣ v ␤ 3 mAbs, characteristics of ␣ v ␤ 3 -mediated interactions, to being entirely insensitive to these reagents (7,39,40). We hypothesized that these differences might reflect the extent of ICAM-1 expression by EC. Accordingly, the adhesion of EC, either stimulated with TNF-␣ (10 ng/ml) to up-regulate ICAM-1 expression or in a basal state, to fg was compared. FACS analysis verified that ICAM-1 expression was low on the nonstimulated EC and considerably higher on TNF-␣-stimulated EC, (Fig. 1) as previously reported (2). The mean fluorescence intensity with an ICAM-1 mAb was 35.6 for unstimulated EC and 425.5 for TNF-␣-stimulated EC. In contrast, ␣ v ␤ 3 expression was unchanged by TNF-␣ stimulation. The mean fluorescence intensities with an ␣ V ␤ 3 specific mAb (LM 609) were 43.7 and 48.7 for unstimulated and stimulated EC, respectively.
In adhesion assays, the proportion of TNF-␣-stimulated EC adherent to fg was consistently 20 -30% greater than that observed with nonstimulated EC. The ␣ V ␤ 3 RGD ligand peptide, GRGDSP, inhibited the adhesion of both stimulated and nonstimulated EC; however, the profile of inhibition was significantly different (Fig. 2). At 12.5 M RGD, the inhibition observed was 70% for nonstimulated EC ( Fig. 2A) but only 32% for TNF-␣-stimulated EC (Fig. 2B), yet at similar concentrations the control GRGESP peptide had a negligible effect in each case. Similar differences in the inhibition profiles were noted with 7E3, a mAb that blocks ␣ v ␤ 3 -mediated function (41). At 2 g/ml, 7E3 inhibited the adhesion of nonstimulated EC to fg by 71%, whereas the adhesion of stimulated EC to fg was only 46% inhibited.
The differences in these inhibition profiles suggest that a component of stimulated EC adhesion to fg may be ␣ v ␤ 3 independent. The contribution of ICAM-1 to this ␣ v ␤ 3 -independent adhesion was verified using an anti-ICAM-1 mAb (LB-2) at 20 g/ml in adhesion assays. The ICAM-1 mAb decreased the adhesion of stimulated EC to fg by 53%, whereas a control mAb had no effect (Fig. 2B). In contrast, the anti-ICAM-1 mAb had only a slight effect on the adhesion of nonstimulated EC to fg, producing 15% inhibition ( Fig. 2A).
A direct interaction between fg and ICAM-1, both the fulllength and truncated form consisting of domains 1 and 2 (D 1 D 2 ICAM-1), was detected in a ligand blot analysis. ICAM-1-expressing SF9 cell lysates were separated by SDS-polyacrylamide gel electrophoresis, transferred onto Immobilon filters, and probed with 125 I-fg. Autoradiograms of filters (Fig. 2, C and  D) indicate that fg specifically bound to ICAM-1 (29). As determined by densitometric scanning, fg binding to ICAM-1 was abrogated Ͼ90% by preincubating the blots with R803, an anti-ICAM-1 polyclonal antibody (Fig. 2C, lane 2), whereas a control antiserum (R155, directed to ␣ L integrin subunit) had a minimal effect (Fig. 2C, lane 1). Similarly, 125 I-fg bound D 1 D 2 ICAM-1. This binding was blocked by ICAM-1 antibody (Fig.  2D, lane 2) but not by an irrelevant antibody (Fig. 2, lane 1). These results indicate that the fg-interactive site resides within the first two Ig domains of ICAM-1.
These results indicate that the fg-binding activity of ICAM-1-(8 -21) was not due to dimerization and that the cysteine residue was itself important in the formation of a stable peptide conformation able to interact with fg.
Specificity of fg-ICAM-1-(8 -21) Interaction-The binding of fg to ICAM-1-(8 -21) was highly specific. First, under conditions in which 125 I-fg bound avidly to ICAM-1-(8 -21), other radiolabeled proteins (fibronectin, low density lipoprotein, lipoprotein(a), and IgG) applied at 20 -50 nM concentrations did not bind or bound minimally to ICAM-1-(8 -21)-coated wells (Fig. 4A). Second, as shown in Fig. 4B, unlabeled fg inhibited the binding of 125 I-fg to the peptide, whereas other unlabeled proteins (fibronectin, transferrin, and IgG) were ineffective. For example, unlabeled fg (2.0 M) inhibited binding of the radiolabeled fg by 86% while fibronectin, at the similar concentration, produced only 12% inhibition. The concentration of fg producing 50% inhibition was in the 0.1-0.2 M range. Third, R152, a rabbit polyclonal antipeptide antibody directed against ICAM-1-(8 -21), blocked the interaction of 125 I-fg with immobilized ICAM-1-(8 -21). At 10 g/ml, R152 blocked the binding by 82%, whereas two other unrelated antipeptide antibodies had  Table I) in TBS were immobilized on microtiter plate wells. Radiolabeled fg (20 nM, 680 ng) were then added as described under "Experimental Procedures." Amino acid sequences corresponding to each of the peptides are designated with the first and the terminal residue numbers. B, maximal fg binding (f) to ICAM-1-(8 -21) peptide was determined using increasing concentration of coating peptide. A representative experimental result is shown from a total of six individual experiments performed for each set.    Fibrinogen Interactions with ICAM-1 no effect (Fig. 4C). R152 was prepared by immunizing rabbits with ICAM-1-(8 -21) peptide. The IgG-purified fraction of the antisera reacted specifically with ICAM-1-(8 -21) and not with several other ICAM-1 peptides as judged by enzyme-linked immunosorbent assay analysis. The interaction of fg with ICAM-1-(8 -21) was divalent cation independent. No differences in the extent of 125 I-fg binding were detected using CaCl 2 or MgCl 2 at concentrations of 0 -4 mM. Furthermore, addition of 5 mM EDTA did not affect fg binding to ICAM-1-(8 -21) (data not shown).
Fibrinogen Recognition of ICAM-1-(8 -21)-The RGD sequences in the A␣ chain and the dodecapeptide (H12) at the extreme COOH terminus of the ␥ chain are the recognition sequences for the binding of fg to the ␤ 3 integrins, ␣ IIb ␤ 3 and ␣ v ␤ 3 (4,6). To verify that these ␤ 3 integrin recognition sequences within fg have a role in ICAM-1 binding, representative RGD and ␥-chain peptides were tested for their effect on 125 I-fg binding to ICAM-1-(8 -21). GRGDSP and H12 had no FIG. 4. Specificity of fg binding to ICAM-1-(8 -21). A, binding of fg and other radiolabeled proteins to ICAM- (8 -21) peptide. 20 nM of fg and 20 -50 nM of lipoprotein(a) (Lp[a]), IgG, low density lipoprotein (LDL), and fibronectin (Fn) were allowed to bind ICAM-1- (8 -21). Bound proteins are reported in ng amounts. B, binding of radiolabeled fg (20 nM) to ICAM-1- (8 -21) in the presence of excess nonlabeled proteins. C, binding of 125 I-fg to ICAM-1- (8 -21) in the presence of increasing concentrations of an antipeptide antibody (R152) directed against ICAM-1-(8 -21) (E) or two control antibodies directed against an integrin ␣ L peptide (ϫ, å). Bound fg is reported as the percentage of fg bound to ICAM-1- (8 -21) in the absence of competing nonlabeled proteins. The above results are from a single representative experiment. Each set of experiments was performed on three separate occasions. Fibrinogen Interactions with ICAM-1 effect on fg binding to ICAM-1-(8 -21), even when used up to 400 M concentration (Fig. 5). At a substantially lower concentration of 50 M, these peptides had a profound effect on fg binding to ␣ IIb ␤ 3 and ␣ v ␤ 3 (data not shown), thus verifying their activity (7).
While these studies were in progress, a peptide from within the D80 region was shown to block ICAM-1 expressing cells from binding to fg. To determine whether this sequence is involved in ICAM-1-(8 -21) recognition, peptides corresponding to the ␥-117-133 region and two other peptides from the gamma chain of fg were tested (Fig. 6B). The peptide corresponding to the ␥-117-133 sequence (NNQKIVNLKEKVA-QLEA) blocked fg binding to ICAM-1-(8 -21) by 82% when used at 400 M concentration, whereas three other peptides, including one overlapping the COOH-terminal end of ␥-117-133 (␥-124 -133) showed no activity. These results indicate that ICAM-1-(8 -21) interaction with fg may involve the NH 2 -termi-nal sequence within ␥-117-133 of fg.
Role of ICAM-1- (8 -21) Sequence in Cell Adhesion-Assays were performed with Raji cells, a lymphoblastoid B cell line that constitutively expresses levels of ICAM-1 comparable to those of TNF-␣-stimulated EC (Fig. 1). Importantly, the expression of ␣ v ␤ 3 on these cells is negligible. Because Raji cells grow readily in suspension and do not require cytokines to stimulate ICAM-1 expression, these cells offer an advantage for measuring ICAM-1-related functions. fg binding to Raji was specific and ICAM-1 dependent. Excess unlabeled fg competed 78% of the labeled fg bound, and an anti-ICAM-1 antibody (20 g/ml) blocked fg binding to Raji by 72% (Fig. 7A). As previously noted, it was not possible to utilize ICAM-1-(8 -21) peptide to block soluble fg binding due to precipitation of fg in the presence of peptide; consequently, we sought the use of antipeptide ICAM-1-(8 -21) antibody (R152). This antibody, when used at 40 g/ml, blocked fg binding to Raji by 69%, and at 20 g/ml it was still capable of inhibiting fg binding by 55%. Normal rabbit IgG had no effect on fg binding.
The effect of ICAM-1-(8 -21) peptide on the adhesion of Raji cells to immobilized fg was assessed (Fig. 7B). A large component of Raji cell adhesion to fg was mediated through ICAM-1 given that 69% of the adhesion was inhibited by the ICAM-1specific mAb, LB-2. The expression of other fg-binding integrins such as ␣ v ␤ 3 and ␣ M ␤ 2 on Raji was low by FACS analyses; thus, a small portion of Raji adhesion to fg may be mediated by other unidentified mechanisms. ICAM-1- M blocked Raji cells adherence to fg by 82% and ICAM-1- (8 -20) blocked adhesion by 48%, whereas at the same concentration three control peptides, including VCAM-1- (34 -46), had minimal effects. Data for the control peptide ICAM-1-(130 -145) are indicated because this peptide is structurally similar to ICAM-1- (8 -21); both peptides contain a charged and hydrophilic NH 2 -terminal portion and a COOH-terminal end which is hydrophobic. Taken together, these results implicate ICAM-1- (8 -21) in disrupting fg-ICAM-1 interactions in cellular adhesive processes. DISCUSSION An interaction involving fg and ICAM-1 results in cellular bridging of monocytic cells and ICAM-1-expressing cells (1). We have demonstrated a direct interaction between fg and intact ICAM-1 and with a truncated two-domain form of ICAM-1 (Fig.  2). Thus, the first two Ig domains of ICAM-1 contain the sites necessary for fg binding. The blockage of fg-mediated Raji adhesion by LB-2, a mAb reported to bind to the first two domains of ICAM-1 (16,43), further verified that fg interacted with the first two domains of ICAM-1. A sequence within the first Ig domain, ICAM-1-(8 -21) (KVILPRGGSVLVTC) was identified that specifically bound fg. Excess unlabeled fg blocked the binding of radiolabeled fg to ICAM-1-(8 -21) (Fig.  4B); and peptides containing sequences within ICAM-1- (8 -21) were effective in competing fg binding to ICAM-1-(8 -21) (Fig.  5). Comparison of ICAM-1-(8 -21) sequence in human and murine ICAM-1 reveals that 8 of the 14 residues are identical, while another 4 have conservative substitution. This high degree of homology at ICAM-1-(8 -21) is interesting because murine ICAM-1 binds human fg (1). Furthermore, comparison of the ICAM-1-(8 -21) sequence within other Ig-like adhesion receptors, such as ICAM-2, ICAM-3, and VCAM-1, reveals amino acid identity only at glycine (Gly-15) and cysteine (Cys-21). However, the COOH end of the sequence is hydrophobic in ICAM-1, -2, and -3 and VCAM-1.
Deletion analysis indicated that the intact ICAM-1-(8 -21) sequence is required for optimal binding to fg. The COOHterminal hydrophobic 6-residue peptide (SVLVTC) demonstrated low fg-binding activity, whereas sequential deletion of the first three NH 2 -terminal residues retained 60% of fg-binding activity (Table I). Furthermore, a substitution of Asp for Arg at position 13 rendered a 39% loss in fg-binding activity. Arg-13 also is predicted to be an important residue from the structural model of ICAM-1-(8 -21) (Fig. 8), and point mutation of this and other residues within the ICAM-1-(8 -21) will establish the importance of these residues. These studies, however, further validate the specificity of fg binding to ICAM-1-(8 -21) and indicate that fg binding to ICAM-1 is not simply a hydrophobic interaction.
The fg-binding activity of ICAM-1-(8 -21) was not due to dimerization, because HPLC analysis of ICAM-1- (8 -21) in the presence and absence of the reducing agent (dithiothreitol) indicated that ICAM-1-(8 -21) did not exist in a dimeric state. In addition, peptide ICAM-1-(8 -20) lacking the cysteine residue also bound fg, but less avidly than ICAM-1-(8 -21). These results indicate that the cysteine residue is important in perhaps stabilization of the overall ICAM-1-(8 -21) structure, but not for dimerization. Secondary structure analyses of ICAM-1-(8 -21) indicated that the amino terminus is hydrophilic, and the COOH end is predicted to be hydrophobic. The NH 2 -terminal region of ICAM-1-(8 -21) lies in ␤-strand A and forms a loop involving the two glycine and single proline residues (12). The COOH half, including the hydrophobic residue from the ␤-strand B, is predicted to be partially buried and offering rigidity to the intact molecule. We modeled the first Ig domain of ICAM-1 using coordinates reported for the crystal structure of VCAM-1 (36). The NH 2 -terminal residues are solvent exposed, and peptide deletion experiments (Table I) confirm the importance of these residues in fg binding. The ICAM-1-(8 -21) sequence may not be fully exposed in the intact protein (Fig. 8) which may explain why the binding of fg to intact ICAM-1 appeared to be of lower affinity compared with the binding of fg to ICAM-1- (8 -21). It is likely that exposure of the ICAM-1-(8 -21) region may be regulated by either ICAM-1 activation or receptor rearrangement for optimal binding of fg to ICAM-1-expressing cells.
ICAM-1- (8 -21) bound to the D80 fragment of fg. This result is consistent with our observation that the GRGDSP peptide was ineffective in blocking fg binding to ICAM-1-(8 -21). Hatzfeld et al. (49,50) noted fragment D but not an RGDcontaining peptide to compete for fg binding to a mitogenic fg receptor on Raji. The fg-␥ chain sequence (117-133) which was recently shown to block fg binding to ICAM-1-expressing cells (51) also blocked fg binding ICAM-1-(8 -21), indicating that Fibrinogen Interactions with ICAM-1 these two sequences may be the reactive pairs in fg-ICAM-1 interaction.
Cellular assembly involving fg is reported to be vital in the recruitment of inflammatory cells onto biomaterial implants in animal models (52,53). fg-ICAM-1 association is implicated in cellular transmigration (19), and recently, contraction of endothelium was reported to be in part mediated by fg interaction with ICAM-1 (54). However, whether the ICAM-1-(8 -21) peptide is capable of blocking leukocyte transmigration and endothelium contraction remains to be investigated. We have noted that fg-bound platelets bridge via ICAM-1 on EC. The participation of ICAM-1-(8 -21) and fg-␥-(117-133) in platelet-EC interaction is under investigation in an effort to understand the role of platelets in inflammatory processes, such as atherosclerosis.