Identification of a Novel Recognition Sequence for Integrin αMβ2 within the γ-chain of Fibrinogen*

The interaction of leukocyte integrin αMβ2 (CD11b/CD18, Mac-1) with fibrinogen has been implicated in the inflammatory response by contributing to leukocyte adhesion to the endothelium and subsequent transmigration. Previously, it has been demonstrated that a peptide, P1, corresponding to residues 190–202 in the γ-chain of fibrinogen, binds to αMβ2 and blocks the interaction of fibrinogen with the receptor and that Asp199 within P1 is important to activity. We have demonstrated, however, that a double mutation of Asp199-Gly200 to Gly-Ala in the recombinant γ-module of fibrinogen, spanning region 148–411, did not abrogate αMβ2 recognition and considered that other binding sites in the γ-module may participate in the receptor recognition. We have found that synthetic peptide P2, duplicating γ377–395, inhibited adhesion of αMβ2-transfected cells to immobilized D100 fragment of fibrinogen in a dose-dependent manner. In addition, immobilized P2 directly supported efficient adhesion of the αMβ2-expressing cells, including activated and non-activated monocytoid cells. The I domain of αMβ2 was implicated in recognition of P2, as the biotinylated recombinant αMI domain specifically bound to both P2 and P1 peptides. Analysis of overlapping peptides spanning P2 demonstrated that it may contain two functional sequences: γ377–386 (P2-N) and γ383–395 (P2-C), with the latter sequence being more active. In the three-dimensional structure of the γ-module, γ190–202 and γ377–395 reside in close proximity, forming two antiparallel β strands. The juxtapositioning of these two sequences may form an unique and complex binding site for αMβ2.

sponses. The engagement of Fg by ␣ M ␤ 2 on activated leukocytes and by ICAM-1 on endothelial cells mediates leukocyte adhesion to the vessel wall and subsequent transmigration in vitro and in vivo (1)(2)(3). The binding of ␣ M ␤ 2 to fibrinogen/fibrin also may result in adhesion of monocytes and neutrophils at sites of vascular injury, such as at atherosclerotic plaques, which are rich in Fg derivatives (4,5). Furthermore, Fg and its derivatives directly promote the accumulation of inflammatory cells on biomaterial implants in animal models (6,7), and depletion of Fg ablates this response (6).
An ␣ M ␤ 2 -binding site has been previously localized within the peripheral D domain of the Fg molecule (8,9). A low molecular weight degradation product, D 30 fragment, generated by plasmin proteolysis of Fg was shown to produce dosedependent inhibition of Fg binding to stimulated monocytes and neutrophils (8). Furthermore, one of a series of overlapping peptides, spanning the constituent ␥-chain of D 30 (␥89 -206), inhibited 125 I-Fg binding to activated THP-1 cells (9). This peptide, designated P1, corresponded to ␥190 -202. A variant peptide in which Asp 199 was either deleted or mutated to Glu or Asn was substantially less potent in inhibiting Fg binding to ␣ M ␤ 2 -bearing cells (9), suggesting that this residue may play a role in the receptor recognition. In addition, when P1 was immobilized onto a surface and implanted into mice, it supported an inflammatory response (7). Nevertheless, the apparent affinity of P1 for ␣ M ␤ 2 was substantially lower than that of intact Fg (9) or the recombinant ␥-module, which encompasses residues 148 -411 of the ␥-chain of Fg (10). These observations suggest that either other Fg sequences, most probably present within the ␥-module, contribute to the recognition of Fg by ␣ M ␤ 2 or that P1 must adopt a conformation within Fg that is favorably constrained for ␣ M ␤ 2 recognition. Consistent with this possibility, synthetic peptides other than P1 inhibit the interaction of Fg with neutrophils (11)(12)(13). The activity of these latter peptides, however, may relate to blockade of other receptors on neutrophils (13)(14)(15), such as ␣ X ␤ 2 (CD11c/CD18), which recognize Fg or its derivatives generated by secreted proteolytic enzymes (14,15).
In this study, we have used 293 embryonic kidney cells stably transfected with ␣ M ␤ 2 (16) to analyze the molecular recognition of Fg by the receptor. These cells bind Fg exclusively via ␣ M ␤ 2 , thereby offering an advantage over isolated leukocytes and their derivative cell lines. We identify a novel peptide, designated P2, corresponding to ␥377-395 from the COOH-terminal part of the ␥-chain of Fg, which is a potent inhibitor of ␣ M ␤ 2 -mediated adhesion to Fg derivatives. In fact, on a molar basis P2 is more potent than the previously described P1 peptide. In the three-dimensional structure of the ␥-module of Fg, the P1 and P2 sequences reside adjacent to one another as part of two antiparallel ␤ strands and may form a complex and unique binding site for ␣ M ␤ 2 .
Expression of Recombinant ␥-Modules and ␣ M I Domain-A recombinant wild-type ␥-module corresponding to residues 148 -411 of the human Fg ␥-chain was produced in Escherichia coli using a pET-20b expression vector as described (10). A mutant ␥-module, in which Asp 199 -Gly 200 were mutated to Gly-Ala, was produced using Transformer TM site-directed mutagenesis kit (CLONTECH). The pET-20b construct containing DNA encoding the ␥-module was modified by sitedirected mutagenesis using two mutagenic primers. One primer, 5Ј-GTGTTTCAGAAGAGACTTGgcGcCAGTGTAGATTTCAAGAAAAAC, introduced the desired mutation (the lowercase letters indicate the mutagenic bases) and an unique restriction site for NarI (underlined) to facilitate further analysis. Another primer, 5Ј-GCGGCCGCACTgGAG-CACCACCAC, eliminated the unique XhoI restriction site in the polylinker region of the target plasmid to facilitate screening. The screening for the desired mutation was performed by Nar I digestion of the plasmid DNA isolated from selected colonies of DH5␣ cells. The mutation was also confirmed by sequencing of the entire DNA fragment encoding the ␥-module. The BL21 Lys S E. coli host cells were then transformed with the mutant plasmid, and the mutant ␥-module was prepared following the procedure previously described for the recombinant wild-type ␥-module (10) with some modifications. Specifically, as revealed by SDS-polyacrylamide gel electrophoresis and Western blot analysis of the bacterial lysates, the yield of the mutant ␥-module was substantially lower. As the mutant was not a dominant protein in the pellet of the lysates, two additional purification steps were introduced. First, the mutant was partially purified by size-exclusion chromatography on Superdex 75 equilibrated with 8 M urea and then refolded by the protocol described earlier (10). The remaining contaminants were subsequently removed by size-exclusion chromatography on Superdex 75 equilibrated with TBS and 1 mM Ca 2ϩ . The purity of the resulting mutant was verified by SDS-polyacrylamide gel electrophoresis and amino acid sequence analysis. The fluorescence-detected melting curve of the mutant (data not shown) was essentially the same as that of the wild-type protein reported previously (10), indicating that the mutant was properly folded.
Recombinant ␣ M I domain was produced and purified essentially as described earlier with minor modification (23). Briefly, the cDNA of the ␣ M I domain was amplified using the following primers: 5Ј-CTCCGAG-GATCCCCTCAAGAGGATAGTGACATT and 5Ј-TTGCTCGAGTCAG-CTGAAGCCTTCCTG, and inserted in the expression vector pET-32a (Novagen, Madison, WI) with restriction enzymes BamHI and XhoI. To express the ␣ M I domain as a fusion protein with thioredoxin, BL21-(DE3) cells were transformed with the above vector and grown to log phase in TB media. Protein expression was induced by addition of 0.3 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 37°C, and the fusion protein from the inclusion bodies was dissolved in 6 M guanidine HCl. After refolding in 20 mM TBS, the fusion protein was purified using ProBond resin (Invitrogen, Carlsbad, CA). To obtain the ␣ M I domain, the fusion partner (thioredoxin) was removed by thrombin cleavage, and the I domain was purified using Q-Sepharose resin (Amersham Pharmacia Biotech). The ␣ M I domain was conjugated with N-hydroxysuccinimide-biotin ester (Calbiochem) as described (24).
Peptides-Peptides corresponding to selected sequences in Fg were synthesized using an Applied Biosystems model 430 peptide synthesizer (Applied Biosystems, Foster City, CA) and purified by high performance liquid chromatography on a preparative C18 Vydac column using a 5-90% linear gradient of acetonitrile in 0.1% trifluoroacetic acid. Authenticity and purity of the peptides were verified by mass spectroscopy and by high performance liquid chromatography. Peptides H19 and H20, duplicating Fg sequences ␥340 -357 and ␥351-370, respectively, and peptide P1-Scr (FRLGWVQTSVDKG), were used as controls. Selected peptides were conjugated to ovalbumin using 1-ethyl-3-[3-(dimethylamino)propyl]carbodiimide as a cross-linker (Pierce) (25), and the conjugates were stored at Ϫ70°C until used.
For radiolabeling of peptides P2 (␥377-395) and P1 (␥190 -202), a variant P1 peptide with KY added to the NH 2 terminus of ␥190 -202 was synthesized, whereas the naturally occurring Tyr 377 at the NH 2 terminus in P2 served as the target for iodination. These peptides were radiolabeled with Na 125 I using a modified chloramine-T procedure (26). The radiolabeled peptides were separated from free 125 I on a Sep-Pak C18 column (Waters, Milford, MA). After extensive washing with sodium phosphate buffer, pH 7.0, 125 I-peptide was eluted with 60% acetonitrile in 0.1% trifluoroacetic solution, and the sample was lyophilized.
The amounts of peptides immobilized onto the microtiter plates used in cell adhesion assays were measured by two different approaches. In the first, peptides were immobilized on the plastic for 3 h at 37°C, washed with phosphate-buffered saline and the concentration was determined by BCA method according to manufacturer's protocol (Pierce). In the second approach, plates were coated with radiolabeled peptides, blocked with 0.5% PVD and then washed with phosphate-buffered saline. Bound peptides were solubilized in 2% SDS ϩ 0.5 N NaOH, and radioactivity was counted in a ␥ counter. The amount of peptide bound to plastic wells was calculated on the basis of the amount of radioactivity recovered and the specific activity of the peptide. The efficiency of peptide immobilization on the plastic as determined by both methods was similar. At an input concentration of 0.1-40 g/ml (0.2 ml), 10 and 8% of added P1 and P2, respectively, were immobilized.
Adhesion Assays-The wells of tissue culture plates (Costar, Cambridge, MA) were coated with varying concentrations of peptides, peptide-ovalbumin conjugates, recombinant ␥-modules, or D 100 fragment for 3 h at 37°C. The coated wells were then post-coated with 0.5% PVD for 1 h at 22°C. ␣ M ␤ 2 -and ␣ M -expressing cells, grown in 10-cm Petri dishes for 24 h as subconfluent monolayers, were harvested with celldissociating buffer (Life Technologies, Inc.) for 1 min at 22°C and washed twice in HBSS/HEPES solution containing 10 mg/ml BSA. These cells were labeled with Na 2 51 CrO 4 (0.5 mCi/ml) for 30 min at 22°C, washed twice and resuspended at 5 ϫ 10 5 /ml in HBSS/HEPES supplemented with 1 mM Ca 2ϩ , 1 mM Mg 2ϩ , and 10 mg/ml BSA. Aliquots (100 l) of the labeled cells were added to each well. For THP-1 cells, the cells were harvested by centrifugation, washed, radiolabeled, and resuspended as described for the ␣ M ␤ 2 -bearing cells. For inhibition experiments, the cells were mixed with selected concentrations of test peptides and incubated for 15 min at 22°C before adding to the coated wells. The cells were incubated on the adhesive substrates for 25 min at 37°C in a 5% CO 2 humidified atmosphere. The nonadherent cells were removed by three washes with HBSS containing 10 mg/ml BSA. The adherent cells were solubilized with 2% SDS, and 51 Cr was quantitated in a ␤ counter. In selected experiments, the cells were incubated with mAbs OKM1, 44a, 904, and IB4 directed against the ␣ and ␤ subunits of ␣ M ␤ 2 for 15 min at 22°C and then added to the wells coated with P1 or P2.
Solid Phase Binding Assay-96-well plates (Costar, Cambridge, MA) were coated with 25 pmol of P2, P1 and control H19 peptides for 3 h at 37°C and post-coated with 1% BSA for 1 h at 22°C. 10 g/ml biotinylated ␣ M I domain in 0.05 M TBS supplemented with 0.1% BSA, 1 mM Ca 2ϩ , and 1 mM Mg 2ϩ was added to the wells and incubated for 3 h at 22°C. In parallel, the same amount of the ␣ M I domain was incubated with 1 g/ml neutrophil inhibitory factor (NIF), 50 g/ml P1, and P2. After washing, streptavidin conjugated to alkaline phosphatase was added and incubated for 1 h at 37°C. The ␣ M I domain binding was detected by reaction with p-nitrophenyl phosphate, measuring the absorbance at 405 nm. Alternatively, the binding of a nonlabeled P2related peptide, ␥373-385, to the immobilized ␣ M I domain was detected with mAb 2G5, which recognizes an epitope within this peptide (18). The binding of this mAb to the peptide was measured by reaction with goat anti-mouse IgG, conjugated to alkaline-phosphatase and using p-nitrophenyl phosphate for detection.
Flow Cytometry-Fluorescence-activated cell sorting analyses were performed to assess the expression of ␣ M or ␣ M ␤ 2 on the surface of transfected cells or THP-1 cells. The cells were harvested as described above, and 10 6 cells were incubated with the selected mAbs in HBSS/ HEPES buffer. mAbs OKM1 and 44a were added at 15 g/200 l of cell suspension and incubated for 30 min at 4°C. The cells then were washed and incubated with fluorescein isothiocyanate-conjugated goat anti-mouse IgG (at 1: 100 dilution) for additional 30 min at 4°C. Finally, the cells were washed and analyzed on a FACStar (Beckton-Dickinson, Mountain View, CA).

Mutation of Asp 199 in the Recombinant ␥-Module Does Not
Abrogate Its Recognition by ␣ M ␤ 2 -Recently, we have shown that the recombinant ␥-module (␥148 -411), which preserves the conformational and functional properties of this region in native Fg and contains the previously identified ␣ M ␤ 2 binding site at ␥190 -202 (GWTVFQKRLDGSV), supports adhesion of the ␣ M ␤ 2 -bearing cells but not mock-transfected cells (10). We further observed that the ␥-module inhibited ␣ M ␤ 2 -mediated adhesion to the D 100 fragment of Fg, and its inhibitory potency was about 20-fold greater than that reported for the synthetic peptide P1 corresponding to the ␥190 -202. To consider whether other sequences within ␥148 -411 might contribute to the recognition of Fg by ␣ M ␤ 2 , we expressed a mutant ␥-module in which Asp 199 -Gly 200 within the P1 sequence were replaced by Gly-Ala. Previous studies have indicated (9) that Asp 199 is of major importance in the recognition of P1 by the receptor. As shown in Fig. 1, when the recombinant proteins were immobilized, the mutant ␥-module supported adhesion of the ␣ M ␤ 2expressing cells as efficiently as the wild-type ␥-module. Adhesion to both proteins was ␣ M ␤ 2 -dependent as NIF, a specific inhibitor of this integrin (28), effectively blocked binding (data not shown). These results prompted us to search for other sequences within the ␥-module that might be involved in ␣ M ␤ 2 recognition.
Effect of ␥377-395 on ␣ M ␤ 2 -mediated Adhesion-The crystal structure of the Fg ␥-module encompassing residues ␥144 -411 (29) revealed that a ␤ strand containing portions of P1 is adjacent to a second ␤ strand formed by residues ␥380 -390. This proximity led us to consider whether this second ␤ strand contributes to ␣ M ␤ 2 recognition. Accordingly, we synthesized a peptide, designated P2, corresponding to ␥377-395, which encompasses this ␤ strand and the flanking residues. As shown in Fig. 2, this peptide inhibited adhesion of the ␣ M ␤ 2 -expressing cells to the immobilized Fg derivative, D 100 , in a dose-dependent manner. On a molar basis, P2 was 12.5 Ϯ 3.4-fold (n ϭ 11) more potent than P1. Although a gradual drift in the ␣ M ␤ 2expressing cells toward a more adhesive phenotype was observed after several months in culture, which influenced the absolute concentrations of the peptides required for inhibition of adhesion, P2 was consistently 10 -15-fold more potent than P1 within the same experiment; 50% inhibition was attained at 0.4 -4.3 M P2 compared with 5-80 M P1. A second difference in the effect of two peptides was noted with respect to the maximal inhibition observed. While P2 was able to produce 80 -90% inhibition at concentrations of 3.5-17 M, maximal inhibition by P1 did not exceed 60 -65%, even at the highest concentration of P1 tested (130 M maximal testable concentration). Similar results were obtained when microtiter plates were coated with the recombinant ␥-module (data not shown). Specificity of the inhibitory effects of both P1 and P2 was apparent. Three control peptides: H19 (␥340 -357) and H20 (␥350 -375), originating from the neighboring region of the ␥-chain of fibrinogen, and a scrambled P1 peptide (P1-Scr), did not inhibit adhesion of the ␣ M ␤ 2 -expressing cells to D 100 (results for H19 are shown). Furthermore, neither P2 nor P1 affected adhesion of the cells to gelatin, a reaction not mediated by ␣ M ␤ 2 , and neither peptide affected cell viability as assessed by trypan blue exclusion.
P2 Directly Supports ␣ M ␤ 2 -mediated Adhesion-The capacity of P2 to directly support ␣ M ␤ 2 -mediated adhesion was tested. P2 was immobilized onto tissue culture wells in free form or as a conjugate with ovalbumin, and the ␣ M ␤ 2 -expressing cells were added. As shown quantitatively in Fig. 3A and visually in Fig. 3B, P2 supported efficient adhesion. Comparison of P2 and P1 demonstrated that both peptides were almost  equally efficient in the supporting of ␣ M ␤ 2 -mediated adhesion. Adhesion was concentration-dependent, and about 60 -65% of added cells adhered to each peptide (maximal adhesion). Three control peptides, P1-Scr, H19, and H20, did not support significant adhesion (results for P1-Scr are shown). Furthermore, mock-transfected cells adhered poorly to P2 as well as to P1 (data not shown). The molar concentrations of each immobilized P2 and P1 required for half-maximal adhesion were 2.4 Ϯ 0.8 and 7.2 Ϯ 1.5 pmol (Fig. 3A). These values are based upon direct quantitation of the amount of each immobilized peptide (see "Experimental Procedures"). When P2 and P1 were conjugated to ovalbumin, the maximal adhesion to the immobilized P2-ovalbumin conjugate was about 2-fold higher than to P1ovalbumin (data not shown). Although conjugation to protein carrier has been shown to improve the adhesive properties of some peptides (30), coupling did not significantly affect the activity of P1 and P2. It should be noted that, although P1 in solution was a relatively poor inhibitor of the ␣ M ␤ 2 -mediated adhesion (Fig. 2), upon immobilization, it supported adhesion almost as well as immobilized P2. The low inhibitory activity of soluble P1 and its propensity to aggregate has been reported previously (9). Thus, soluble P1 may not adopt a preferred conformation for interaction with ␣ M ␤ 2 .
We next tested the ability of P2 to support adhesion of THP-1 monocytoid cells that bear ␣ M ␤ 2 and are known to adhere to P1 (9). As shown in Fig. 4, P2 supported adhesion of the THP-1 cells in a concentration-dependent manner, and the extent of adhesion to P2 was greater than for P1. Adhesion to both immobilized peptides was enhanced 1.2-2-fold by PMA stimulation of the cells. Under the same conditions, with and without PMA stimulation, P1-Scr did not support cell adhesion (data not shown). The specificity of ␣ M ␤ 2 -P2 interaction was substantiated further using function-blocking mAbs, which recognize ␣ M and ␤ 2 subunits. Adhesion of ␣ M ␤ 2 -expressing cells to three adhesive substrates, P2, P1, and D 100 fragment, was reduced in a dosedependent manner by mAb 44a to the ␣ M subunit. At 5 g/ml, this mAb produced 71% inhibition of adhesion to P1, 55% to P2, and 87% to D 100 fragment (Table I). mAb IB4 to the ␤ 2 subunit was also inhibitory; at the 5 g/ml concentration, it reduced adhesion of the ␣ M ␤ 2 -expressing cells to P1, P2, and D 100 by 67, 50, and 77%, respectively. In contrast, another anti-␣ M mAb, 904, was moderately inhibitory, and mAb OKM1 was not effective. In parallel experiments, mAbs 44a and IB4 also inhibited adhesion of PMA-activated THP-1 cells to immobilized P2 and P1 (Table I). A control mAb w6/32 against the class I major histocompatibility complex was consistently non-inhibitory.
Cross-competition between P1 and P2 for ␣ M ␤ 2 -To investigate the relationship between P1 and P2 binding to ␣ M ␤ 2 , we have examined the ability of each peptide in soluble form to inhibit adhesion of the ␣ M ␤ 2 -bearing cells to immobilized P1 and P2. As shown in Fig. 5, both peptides were able to crossinhibit adhesion of the ␣ M ␤ 2 -expressing cells; i.e. P2 inhibited adhesion to immobilized P1, and P1 inhibited adhesion to immobilized P2. The concentrations of soluble peptides required for half-maximal inhibition (IC 50 ) of adhesion were 100 M P2 and 630 M P1 with immobilized P2 (3.5 pmol) as the substrate. The IC 50 values for P2 and P1 on immobilized P1 (5.8 pmol) were 250 and 890 M, respectively. Thus, as observed with D 100 as the immobilized substrate (Fig. 2), soluble P1 was a weaker inhibitor of cell adhesion than soluble P2. The concentrations of peptides required to obtain 50% inhibition were significantly higher on immobilized peptides than on D 100 . This difference was not due to variations in the amounts of immobilized peptides as the amounts of all three ligands adsorbed onto the surface were similar. The competition between peptides suggests that they can interact with sterically or allosterically overlapping sites within ␣ M ␤ 2 .
To explore further the contribution of the P2-C portion of P2 in ␣ M ␤ 2 recognition, we employed mAb 4-2, directed against the ␥392-406 (19). Preliminary enzyme-linked immunosorbent assay experiments (data not shown) indicated that mAb 4-2 interacted with immobilized P2 and P2-C, suggesting that at least part of its epitope resides in the P2-C. As shown in Fig. 7, mAb 4-2 inhibited adhesion of the ␣ M ␤ 2 -expressing cells to  Cr-labeled cells were preincubated with 5 g/ml of each mAb for 15 min at 22°C with constant agitation and then added to the wells coated with the peptides or D 100 . After 25 min at 37°C, nonadherent cells were removed and bound radioactivity was measured. Data are expressed as a percentage of inhibition of adhesion, with adhesion in the absence of mAbs defined as 0% inhibition. immobilized P2-C and the recombinant ␥-module. As a control, adhesion to P1 was not affected by mAb 4-2; and mAb 2G5, which recognizes an epitope in the vicinity of P2-C, ␥363-385 (18), did not affect adhesion.
Recognition Sites within ␣ M ␤ 2 for P2-Several independent studies and approaches have implicated the I domain of the ␣ M subunit in the recognition of fibrinogen by ␣ M ␤ 2 (23,31,32). Accordingly, we have tested whether P2 and P1 can directly bind the ␣ M I domain. Similar amounts of P2 and P1 were immobilized onto microtiter plate, and the binding of the biotinylated recombinant ␣ M I domain was assessed. As shown in Fig. 8A, both peptides bound the ␣ M I domain, and NIF and soluble P2 and P1 peptides inhibited the binding of the ␣ M I domain. In addition, we also were able to detect binding of a P2-derivative peptide to the ␣ M I domain using a different approach. mAb 2G5 recognizes an epitope within ␥363-385 but not in ␥377-395 (18). Accordingly, we used ␥373-385, which also supported ␣ M ␤ 2 -mediated adhesion (Fig. 6) and mAb 2G5 to detect its interaction with the immobilized ␣ M I domain. As shown in Fig. 8B, the immobilized ␣ M I domain bound soluble ␥373-385 in a concentration-dependent manner. Thus, with either component immobilized, ␣ M I domain or P2 peptide, a productive interaction was demonstrable.
Although the above data implicate the I domain of ␣ M in recognition of both P2 and P1, the ␤ 2 subunit does influence the interaction of these peptide ligands with the receptor. This conclusion was derived from experiments comparing the interaction of P2 and P1 with 293 cells expressing only ␣-subunit or the ␣ M ␤ 2 heterodimer. In these experiments, populations of ␣ M -and ␣ M ␤ 2 -expressing cells that reacted similarly with OKM1, a mAb specific for the ␣ M subunit, were selected by cell sorting. By fluorescence-activated cell sorting, the mean fluorescence intensity, expressed in arbitrary units, for OKM1 was 820 Ϯ 33 and 831 Ϯ 47 for ␣ M -and ␣ M ␤ 2 -expressing cells, respectively (n ϭ 5). As shown in Fig. 9, both P2 and P1 supported adhesion of ␣ M -transfectants in a dose-dependent manner. However, the maximal level of adhesion of the ␣ Mexpressing cells to the peptides was about 2-fold lower than that of the ␣ M ␤ 2 -bearing cells (33 Ϯ 3% and 33 Ϯ 13% maximal adhesion for the ␣ M cells to P2 and P1, respectively, versus 63 Ϯ 5% maximal adhesion to either peptide for the ␣ M ␤ 2 -bearing cells).

DISCUSSION
In this study, we identified a novel sequence within the COOH-terminal part of the Fg ␥-chain, ␥377-395, which is recognized by leukocyte integrin ␣ M ␤ 2 . A synthetic peptide corresponding to this sequence, designated P2, efficiently inhibited adhesion of the ␣ M ␤ 2 -transfected cells and THP-1 monocytoid cells to the D 100 fragment of Fg, directly supported adhesion of these cells, and interacted with the ␣ M I domain of ␣ M ␤ 2 . Thus, the P2 sequence defines a new recognition specificity for ␣ M ␤ 2 and may mediate the interaction of Fg derivatives with this receptor.
That recognition sites, in addition to the previously identified ␥190 -202 sequence, are involved in the interaction of Fg with ␣ M ␤ 2 is consistent with the lower potency of the P1 peptide and the D 30 fragment, which lacks the P2 sequence, to inhibit the binding of Fg to ␣ M ␤ 2 -bearing cells compared with Fg fragments D 100 and X (8,9) and the recombinant ␥-module (10), which all have intact COOH termini. Furthermore, as shown in this study, it is consistent with the capacity of a recombinant ␥-module, in which a key amino acid, Asp 199 , within the P1 sequence was mutated to Ala, to support efficient adhesion of the ␣ M ␤ 2 -expressing cells.
The three-dimensional structure of the Fg ␥-module, which became available recently (29), revealed that part of the P1 sequence forms a ␤ strand (residues 190 -198) and a short connecting loop (residues 199 -202) (Fig. 10A). This ␤ strand and loop are adjacent to a ␤ strand formed by residues 380 -390, portions of the P2 sequence. The unusual feature of this latter ␤ strand is that it originates from the COOH-terminal part of the ␥-chain but folds back and inserts into the middle subdomain of the ␥-module, next to the P1 ␤ strand. Thus, although ␥190 -202 and ␥380 -390 are separated by 178 residues in terms of linear amino acid sequence, the specific folding of the ␥-module brings these two ␣ M ␤ 2 recognition sequences into close proximity.
Analysis of partially overlapping peptides indicated that P2 may contain two adhesive sequences, one in its NH 2 -terminal part, ␥377-386 (designated P2-N), and a second in its COOHterminal region, ␥383-395 (designated P2-C). These two adhesive sites differed in their adhesion-promoting and inhibitory activities; P2-N is ϳ65-fold less active than parental P2, and P2-C is almost as active as P2. As the simplest interpretation of these results, P2-C represents a major ␣ M ␤ 2 recognition site and P2-N a minor one. Indeed, at only 13 amino acids, P2-C is a relatively short and potent peptide to begin detailed structure-function analyses. Key residues involved in P2-C recognition by ␣ M ␤ 2 should be exposed on the surface of the ␥-module of Fg. In the crystal structure of the ␥-module, the side chains of Thr 383 , Lys 385 and Phe 389 are partially exposed whereas the remaining residues within the ␥380 -390 ␤ strand are buried, and the extreme COOH terminus emerging from this ␤ strand is again well exposed, including the side chains of Asn 390 , Arg 391 , Leu 392 , Thr 393 , and Ile 394 (Fig. 10B). These exposed residues might well be critical to the activity of P2-C. Despite the weak adhesive activity of P2-N, it may be premature to dismiss its role in ␣ M ␤ 2 recognition. As shown in Fig. 10B, the exposed residues in the NH 2 -terminal part of P2 are Lys 380 and Lys 381 . These residues lie adjacent to the antiparallel ␤-strand containing the P1 region, which includes Arg 196 , Leu 198 , and Asp 199 as its only exposed residues. The distances between Asp 199 and Lys 380 and between Asp 199 and Lys 381 , as determined from the crystal structure, are 9.9 and 9.8 Å, respectively. This distance is considerably shorter than the 29 Å which separate Lys 381 in P2-N from Asn 390 in P2-C or 34Å between Lys 380 and Asn 390 .
What is the relationship among these three subsites (P1, P2-N, and P2-C)? Based on the dissection of the recognition sites for other integrin ligands, two models can be considered. In one model, the binding pocket for ␣ M ␤ 2 in fibrinogen may represent a wide surface that can engage multiple sequences, P1, P2-N, P2-C and/or others, simultaneously. This model would be analogous to the recognition of fibronectin by ␣ 5 ␤ 1 in , and 1 mM Ca 2ϩ , was added to the wells coated with 25 pmol of P2 and P1 in the absence (control) or in the presence of 1 g/ml NIF, 50 g/ml P2 or P1, and incubated for 3 h at 22°C. After washing, the bound ␣ M I domain was detected using streptavidin conjugated to alkaline phosphatase, and p-nitrophenyl phosphate for disclosure. Results are the mean Ϯ S.E. of three independent determinations. B, assessment of P2 binding to the ␣ M I domain with mAb 2G5. Different concentrations of the derivative P2 peptide, ␥373-385, were incubated with recombinant ␣ M I domain coated onto microtiter plates for 2 h at 22°C. After washing, mAb 2G5 (5 g/ml) was added to the wells and incubated for an additional 1.5 h. The binding of mAb 2G5 then was detected with a secondary goat anti-mouse IgG conjugated to alkaline phosphatase with subsequent development of the reaction with p-nitrophenyl phosphate. which the appropriate positioning of at least two distant short sequences, the RGDS sequence in the 10th type III repeat and the HPSRN synergistic sequence in the 9th type III repeat, are crucial for high affinity ligand binding of the receptor (33). In the second model, P1, P2-N, and P2-C may bind to the same or overlapping sites within the receptor. This model may apply to the binding of multiple fibronectin peptides to ␣ 4 ␤ 1 (30,34) or RGD and Fg ␥406 -411 to ␣ IIb ␤ 3 (35). The binding of both P1 and P2 to the ␣ M I domain and their cross-inhibition of cell adhesion seem to be consistent with the latter model.
An additional level of complexity in considering the recognition of P1 and P2 by ␣ M ␤ 2 is the exposure of these sequences on the surface of intact Fg. Although portions of these sequences are exposed in the ␥-module, our preliminary data indicate that residues in P1, involved in the receptor recognition, are only partially available on the surface of Fg and can be further exposed by proteolysis of the parent molecule. 2 Furthermore, the P2-C, ␥383-395, resides in immediate proximity and partially overlaps ␥392-406. This latter sequence is cryptic in intact soluble Fg based upon its lack of reactivity with mAb 4-2 (19). The epitope for this mAb becomes exposed when Fg is immobilized or proteolyzed to fragments D 100 or DD-dimer. The nature of the conformational alteration which occurs upon conversion of Fg to these fragments and exposes ␥392-406 is not known. A potential mechanism of this transition may be associated with the positioning of the COOH-terminal part of ␥-chain (␥392-411), which is a flexible appendage in intact Fg (29,36).
In summary, at least two sites within Fg can be recognized by ␣ M ␤ 2 . One site resides in ␥190 -202, and the second one is located in the ␥377-395. This latter site may be composed of two subsites. Although separated in linear amino acid sequence, ␥190 -202 and ␥377-395 are brought into close proximity by the folding of the ␥-module and could potentially form the recognition site for ␣ M ␤ 2 . Further studies involving mutation analyses should provide insight into the relationship between these recognition sequences and the ␣ M ␤ 2 binding pocket in Fg. Considering the complexity of the P2 site, such analyses will be complicated. Furthermore, our preliminary data indicate that the ␥146 -226 fragment with Asp 199 in the P1 region mutated to Ala still supported cell adhesion, 3 consistent with the previously described reduction but not the complete loss of the adhesive activity of P1 peptides with a single substitution at this position (9). Thus, multiple mutations in P1 and both P2 regions will be required to dissect the role of these as well as other regions in adhesive functions. Such systematic studies are in progress. FIG. 10. Spatial positioning of ␥190 -202 and ␥377-395 in the three-dimensional structure of the fibrinogen ␥-module. A, the ribbon diagram of the fibrinogen ␥-module (␥144 -402) is based upon its crystal structure (29). The ␥190 -202 and ␥380 -390 regions are colored in pink and blue, respectively. The stereo pair of the ␥-module is available from Yee et al. (29). B, space-filling model of the ␥-module with exposed P1 and P2 residues highlighted. Side chain groups of the solvent exposed residues in P1 and P2 regions are colored in pink and blue, respectively. The residues are numbered according to their position in the ␥-chain. The figure was constructed using the computer programs MolScript, BobScript, and Raster (37)(38)(39).