INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

THE CONTACT SYSTEM, composed of coagulation factor XII, prekallikrein (PK), and high-molecular-weight kininogen (HK), is responsible for surface-mediated defense reactions (6) and has been immunocytochemically localized on the external surface of neutrophils (14). HK and PK circulate as a noncovalent complex in plasma (21). Cleavage of PK by activated factor XII results in the formation of plasma kallikrein, a serine protease that cleaves HK to yield bradykinin (26), a potent mediator of inflammation. In addition, plasma kallikrein serves as a naturally occurring agonist for neutrophils, stimulating chemotaxis (17), aggregation (24), and degranulation (29). Plasma kallikrein requires the presence of HK to stimulate neutrophils, either from cellular or plasma sources (12). HK binds saturably, specifically and reversibly to 40,000-70,000 sites per neutrophil with a dissociation constant of 9-18 nM (12). Using specific monoclonal antibodies, we have presented evidence that a receptor for HK is the leukocyte integrin CD11b/18 (Mac-1, _M2) (28) and that HK competes with fibrinogen for binding to that receptor (11).

HK and low-molecular-weight kininogen (LK) are coded for by a single-copy gene containing 11 exons (18). The first three domains (D1-D3) are identical in HK and LK, and each is composed of the products of three exons. HK (23) and LK (16) both inhibit thrombin binding to and activation of platelets. D3, composed of exon 7-9 products, expresses two distinct sites, one inhibiting thrombin activation of platelets (16) and the other serving as a cell binding site to platelets (Fig. 1) (16). Exon 10 codes for domain 4 (D4), which contains bradykinin and is present in both kininogens as well as domains 5 (D5) and 6 (D6), present uniquely in HK. D5, rich in histidine, glycine, and lysine, is responsible for binding to anionic surfaces (Fig. 1) (7), whereas D6 contains the amino acid sequence responsible for complex formation with PK (27).

View larger version (50K): [in this window] [in a new window]   Fig. 1.   Schematic structure of high-molecular-weight kininogen (HK) domains 3 (D3; A) and 5 (D5; B). D3 of HK or low-molecular-weight kininogen (LK): shaded areas indicate peptides used in this study; dark-shaded peptides Leu271-Ala277 and Cys333-Cys352 indicate peptides that blocked HK binding; light-shaded peptide Glu307-Glu323 was not inhibitory; arrows with letters indicate location of introns. Exon 7 product is F-G, exon 8 product is G-H, and exon 9 product is H-I. This figure was created by Dr. Robin Pixley and adapted from part of Fig. 11-3 (8). D5 (partial sequence): shaded areas indicate peptides used in this study; dark-shaded peptide Gly442-Lys458 inhibits HK binding; medium-shaded peptide Phe459-Lys478 also inhibits HK binding; light-shaded His479-His498 had little effect. Arrow is site of kallikrein cleavage in the light-chain D5 of HK.

We have recently localized the sites on HK responsible for binding neutrophils to D3 and D5 (28). Fine mapping of kininogen binding sites on endothelial cells has recently been performed (13, 15). The aim of the present study is to map the binding regions on kininogens for neutrophils. We demonstrate that two short sequences within D3, contained within exon 7 and 9 products, respectively, and a single amino acid sequence within D5 form part or all of the HK binding site to neutrophils.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Materials. 1,3,4,6-Tetrachloro-3,6-diphenylglycouril (Iodo-Gen) was obtained from Pierce Chemical. Na¹²⁵I (100 mCi/ml) was obtained from Du Pont NEN (Billerica, MA). Hanks' balanced salt solution (HBSS), free of calcium chloride, magnesium sulfate, and magnesium chloride, was obtained from Life Technologies. Histopaque-1083-1 was obtained from Sigma Diagnostics. All other reagents were of the highest purity available.

Isolation and purification of human neutrophils. Human neutrophils (polymorphonuclear leukocytes, PMN) were isolated and purified from whole blood by a modification (Majluf-Cruz, Khan, DeLa Cadena, and Colman, unpublished results) of a previously described method (3). Blood was drawn from normal volunteers after written informed consent on the morning of the experiment and collected in ACD-dextran (7.5%) to isolate PMN. After sedimentation of the whole blood for 30 min at 1 g, 15 ml of platelet-rich plasma were added to a 15-ml layer of Ficoll-Hypaque (1.119 g/ml), and the samples were centrifuged at 1,200 rpm for 45 min at 4°C. The supernatant was discarded, and the cell pellet at the bottom of the tube containing PMN was rapidly resuspended in 5 ml of a solution to lyse erythrocytes (0.15 M NH₄Cl and 0.01 M KHCO₃ in 100 ml of distilled water) and incubated for 3 min at 23°C. The tube containing the cell suspension was then gently vortexed and spun at 1,000 rpm for 5 min. This last step was repeated until the PMN pellet was free of erythrocytes. Then PMN were resuspended in a "blocking" buffer (to decrease nonspecific binding) containing either 1% BSA and 10 ml of avidin-conjugated alkaline phosphatase for fluorescent studies or 1% BSA in DMEM for cell binding studies employing radioactivity for 45 min at 4°C. After the blocking step, the PMN were then centrifuged at 1,000 rpm for 5 min and resuspended in binding buffer containing either 0.1% BSA in HBSS or 0.1% BSA in DMEM for fluorescent- and radioactive-based studies, respectively. For activated PMN, the cell pellets obtained after centrifugation of platelet-rich plasma with Ficoll-Hypaque were incubated directly with 5 mg/ml cytochalasin B for 3 min followed by the addition of 10⁷ M N-formyl-Met-Leu-Phe (FMLP) for 5 min at 37°C. After the activation step, the PMN were diluted 1:1 with either HBSS or DMEM at 4°C. The samples were then centrifuged at 1,000 rpm for 5 min, and the cell pellet was obtained and processed as described above.

Plasma proteins. Human purified HK cleaved by plasma kallikrein (specific clotting activity of 11.9 U/mg) was obtained from Enzyme Research Laboratories (South Bend, IN). Under nonreducing conditions, cleaved HK has two bands with molecular masses of 160 and 120 kDa on 10% polyacrylamide gels (Bio-Rad, Richmond, CA) with SDS, whereas under reducing conditions (3 mM dithiothreitol), it was primarily two bands with molecular masses of 62 and 45 kDa, respectively. HK was radiolabeled with Na¹²⁵I using Iodo-Gen (10). The specific radioactivity of the protein varied from 1 to 4.8 mCi/mg with >80% of the molecules of HK being iodinated. The radiolabeled protein (¹²⁵I-HK) retained >95% of its procoagulant activity as well as its antigenic properties, as previously reported (12).

Bacterial expression and purification of HK D3 (Gly²³⁵-Met³⁵⁷), D5 (Val³⁸⁴-Lys⁵⁰²), and D6 (Thr⁵⁰³-Ser⁶²⁶). The details of bacterial expression and purification of D3 (28) and D5 and D6 (9) have been described. (The NH₂ terminal and the COOH terminal are indicated by the three-letter code amino acid. The sequence of the peptide is designated in the one-letter code.) We have utilized similar methodology using LK cDNA (pHKG36) (18), kindly provided by Dr. S. Nakanishi, as a template to express various D3 fragments in bacteria. The DNAs coding for various fragments were separately amplified by PCR using specific sets of primers and inserted in-frame into pGEX2T vector (Pharmacia, Piscataway, NJ) by the procedures described earlier (20, 28). This construct expresses the protein products as fusion proteins with glutathione S-transferase (GST) as the NH₂-terminal region. All of the sense primers contained a BamH I recognition site, and the antisense primers contained EcoR I recognition site. Products of exons 7-9 (Gly²³⁵-Gly²⁹², Gly²⁹²-Gly³²⁸, and Gly³²⁹-Met³⁵⁷, respectively) of D3 were expressed using PCR with SPK-82A/B, SPK82C/D, and SPK-82E/F primers, respectively (Fig. 1) (19). All polypeptides were characterized by SDS-PAGE as previously described (19).

Peptide studies. Two peptides, Leu²⁷¹-Ala²⁷⁷ (LNAENNA) and Ile²⁶⁸-Ala²⁷⁷ (ITKLNAENNA), derived from exon 7, one peptide, Glu³⁰⁷-Glu³²³ (ETTCSKESNEELTESCE), derived from exon 8, and one peptide, Cys³³³-Cys³⁵² (CNAEVYVVPWEKKIYPTVNC), derived from exon 9 of D3, were obtained from Commonwealth Biotechnologies (Richmond, VA) (Fig. 1). Note that, in peptide Cys³³³-Cys³⁴³, GC was substituted for the naturally occurring WE, the cysteine to introduce a disulfide bond and the glycine to mimic the turn in the D3 model (4). The peptides containing two cysteines were each air oxidized to the disulfide form. Three peptides, Gly⁴⁴²-Lys⁴⁵⁸ (GLGHGHEQQHGLGHGHK), Phe⁴⁵⁹-Lys⁴⁷⁸ (FKLDDDLEHQGGHDLDHGHK), and His⁴⁷⁹-His⁴⁹⁸ (HKHGHGHGKHKNKGKKNGKH), derived from D5 (Fig. 1), were kindly provided by Dr. W. Müller-Esterl (Institute for Physiological Chemistry and Pathophysiology, Johannes Gutenberg University, Mainz, Germany). All peptides were synthesized by solid-state methods and purified on HPLC. Sequence was confirmed by NH₂-terminal analysis. Randomly scrambled peptides NEANANL-NH₂ · TFA for Leu²⁷¹-Ala²⁷⁷ (LNAENNA) and QGGLHGGLHHGGHQHE-NH₂ · TFA for Gly⁴⁴²-Lys⁴⁵⁸ (GLGHGHEQQHGLGHGHK) were synthesized and air oxidized to the respective cyclic peptide for 3 days at 100 µg/ml in 0.1 M NaHCO₂ (pH 7.0) at 23°C by Dr. Robert B. Harris (Virginia Commonwealth University, Richmond, VA). The oxidation was complete as judged by the lack of an increase in absorbance at 405 nm when reacted with DTNB. Purity was tested with analytical HPLC, and each peptide was a single peak, both by reverse HPLC and gel filtration HPLC (to detected polymers). Structure of the peptides was confirmed by amino acid composition and sequencing. All methods regarding peptides have previously been published (4).

Binding experiments: ¹²⁵I-HK binding to PMN. Binding experiments were performed as previously described (12), with modifications according to Majluf-Cruz et al. (unpublished results), using filters instead of gradient density centrifugation. In all binding experiments, PMN were used at a concentration of 10⁷/µl. In a typical binding experiment, 100 µl of washed PMN in HBSS without added Ca²⁺ and Mg²⁺ but with added ZnCl₂ (50 µM, pH 7.4) were incubated with recombinant polypeptide, synthetic peptide, or buffer at 23°C without stirring in a 1.5-ml conical polypropylene centrifuge tube for 45 min (Action Scientific, Carolina Beach, NC). ¹²⁵I-HK (60 nM) was then added to yield a total volume of 300 µl and incubated for another 45 min. Aliquots (50 µl) were removed (in quadruplicate) for each experimental point. Each sample was applied to multiscreen-GV N22 plates blocked overnight at 4°C by HBSS containing 1% BSA solution. Plates were set on a vacuum extractor to remove the soluble ligand, and the PMN remained attached to the membrane. Membranes with bound PMN were removed and counted for radiolabeling using a rack gamma counter. A 50-µl aliquot (with tip) for each corresponding experimental point was kept in separate 3-ml polypropylene tubes for counting radioactivity for total binding.

Calculation of binding experiments. Calculation of bound HK was based on the specific activities of the radiolabeled ligand, and the results were expressed as micrograms of HK bound per 10⁷ PMN. In a typical experiment, total binding is the amount of ¹²⁵I-HK bound in the absence of unlabeled ligand, whereas nonspecific binding is the amount of ¹²⁵I-HK bound in the presence of EDTA (0.2 mM). Binding in the absence of added Zn²⁺ or the presence of EDTA was previously shown to be the same as the binding in the presence of a 50-fold molar excess of unlabeled HK (12). Specific binding was obtained by subtracting the nonspecific binding from the total binding. The results with inhibitors, such as recombinant proteins and peptides, were normalized when the total binding was 100% without inhibitors. Competition studies were performed by incubating recombinant proteins or peptides with ¹²⁵I-HK before addition of PMN or with PMN before addition of ¹²⁵I-HK. Each point was determined in quadruplicate. Each experiment was repeated on three to five separate donor PMN, and binding was expressed as percent binding (mean ± SD) with 100% equal to the binding of ¹²⁵I-HK in the absence of any other ligand.

Statistics. The analysis of the data was performed with the program SigmaStat on an IBM platform. A one-way ANOVA was performed by use of nonparametric analysis and the Bonferroni correction for small numbers. Statistical significance was recorded if P < 0.05.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Influence of concentration of HK on binding to PMN. In preliminary studies, we used four different concentrations of ¹²⁵I-HK to establish an optimal concentration for competitive studies. Specific binding to activated PMN was found to saturate at 180 nM for ¹²⁵I-HK when it was determined. Accordingly, a concentration giving 60% of maximal binding (60 nM) for ¹²⁵I-HK was used.

Inhibition of ¹²⁵I-HK binding to PMN by exon 7-9 products of D3. We first used D3 to inhibit the binding of ¹²⁵I-HK to PMN. Unlabeled D3 in 100-fold molar excess (3 mM) inhibited the binding of ¹²⁵I-HK by 80.8 ± 1.5%. These results are similar to those reported from this laboratory (28) and confirm that one of the binding sites on HK is D3 on their common heavy chain. For further investigation of the cell binding epitopes of HK, three components of D3, exon 7-9 products, were used in 100-, 30-, 10-, 3-, and 1-fold molar excess to inhibit the binding of ¹²⁵I-HK (60 nM) to PMN. Binding of ¹²⁵I-HK (60 nM) to PMN (Fig. 2) was inhibited by exon 7 product as a function of concentration with 50% binding at 3,400 nM and by exon 9 product as a function of concentration with 50% binding at 1,650 nM. The exon 8 product showed only modest inhibition of binding. GST in 100-fold molar excess was used as control, since all recombinant polypeptides were fusion proteins with GST and showed no inhibition of binding of ¹²⁵I-HK to PMN. The inhibition by exon 7 and 9 products of HK binding (Fig. 2) was statistically significant. Exon 7 product inhibited less than D3 (P = 0.016), but exon 9 was not significantly different from exon 7 (P = 0.13). There were significant differences (P < 0.05) between the concentrations (6,000 and 1,800 nM) for exon 7 (P = 0.028) but not exon 9 products (P = 0.44), but there were no differences between the lower concentrations (600, 180, and 60 nM). Thus the results with ¹²⁵I-HK showed concentration-dependent inhibition of binding to PMN by exon 7 and 9 products. The inhibition by exon 8 products showed no significant difference in concentration, and was <50% at 100-fold molar excess.

View larger version (20K): [in this window] [in a new window]   Fig. 2.   Effect of exons 7-9 of D3 on inhibition of 125I-HK binding to polymorphonuclear neutrophils (PMN). Isolated PMN (107/ml) in HBSS without Ca2+ or Mg2+ (pH 7.4) were incubated with 125I-HK (60 nM) at 23°C for 45 min (see METHODS). Percent binding was calculated based on total binding of 125I-HK at 60 nM. , exon 7 product; , exon 8 product; , exon 9 product; , glutathione S-transferase; , D3. All determinations were performed in triplicate and expressed as means ± SD.

Inhibition of ¹²⁵I-HK binding to PMN by peptides derived from exons 7-9 of D3. Because exon 7 and 9 products of D3 showed concentration-dependent inhibition of binding to PMN (Fig. 2), we mapped the binding site on D3 by examining various peptides from exons 7-9 of D3 (Fig. 3). Leu²⁷¹-Ala²⁷⁷, derived from exon 7, significantly inhibited (P < 0.05) binding of ¹²⁵I-HK (60 nM) at 6,000, 1,800, and 600 nM but not at 180 or 60 nM. A significant difference (P < 0.05) was found between the 100- and 10-fold molar excess (6,000 and 600 nM, respectively). The 50% binding was found at 3,470 nM. Scrambled peptide, NEANANL-NH₂ · TFA for Leu²⁷¹-Ala²⁷⁷ (LNAENNA) did not inhibit the binding of ¹²⁵I-HK (60 nM) at 6,000 nM (Fig. 3). Ile²⁶⁸-Ala²⁷⁷ inhibited the binding of HK to PMN to the same extent as LNAENNA but was not more potent (data not shown). Cys³³³-Cys³⁵², derived from exon 9, inhibited the binding of ¹²⁵I-HK significantly (P < 0.05) at 6,000, 1,800, and 600 nM but not at 180 or 60 nM. There was a significant difference (P < 0.05) between the 10- and 100-fold molar excess in the degree of inhibition. The 50% binding point was 520 nM. Glu³⁰⁷-Glu³²³, derived from exon 8, showed no inhibition at 100-fold molar excess (Fig. 3).

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Effect of peptides derived from exon 7-9 products of D3 on inhibition of 125I-HK binding to PMN. Leu271-Ala277, derived from exon 7; Glu307-Glu323, derived from exon 8; and Cys333-Cys352 and Cys333-Cys343, derived from exon 9 of D3, were used as competitors for 125I-HK binding as described for Fig. 2. , Leu271-Ala277; , Glu307-Glu323; , Cys333-Cys352; , scrambled LNAENNA. All determinations were performed in triplicate and expressed as means ± SD. SDs of Glu307-Glu323 were 1.15, 0.57, 0.57, 0.57, and 0.57, respectively, too small to be seen.

Inhibition of ¹²⁵I-HK binding to PMN by D5 and its peptides. Unlabeled D5 in 100-fold molar excess (6,000 nM) inhibited 49.0 ± 1.5% binding of ¹²⁵I-HK (60 nM) to PMN and was less potent than D3, as previously observed (Fig. 4) (28). To map the binding site, we selected three peptides that together comprised the full sequence of the recombinant polypeptides which blocked binding of HK to anionic surfaces (Fig. 4) (7). Gly⁴⁴²-Lys⁴⁵⁸ significantly (P < 0.05) inhibited at 6,000, 1,800, and 600 nM, respectively, but no inhibition was observed at 180 or 60 nM. Each concentration inhibited significantly the binding of HK to PMN. However, a significant concentration dependence was not observed. Scrambled peptide, QGGLHGGLHHGGHQHE-NH₂ · TFA for Gly⁴⁴²-Lys⁴⁵⁸ (GLGHGHEQQHGLGHGHK) did not inhibit the binding of ¹²⁵I-HK (60 nM) at 60, 1,800, and 6,000 nM (Fig. 4). Phe⁴⁵⁹-Lys⁴⁷⁸ inhibited at 6,000, 1,800, and 600 nM, respectively, and inhibition was significant (P < 0.05), but no concentration dependence was found. D6, the PK binding site of HK, showed virtually no inhibition of binding of ¹²⁵I-HK (60 nM) to PMN at 100-fold molar excess (Fig. 4). His⁴⁷⁹-His⁴⁹⁸ did not inhibit binding of ¹²⁵I-HK at 100-fold molar excess (data not shown).

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Effect of peptides from D5 on inhibition of 125I-HK binding to PMN. Gly442-Lys458 and Phe459-Lys478 were derived from D5 and used as competitors for 125I-HK binding as described in Fig. 2. , Gly442-Lys458; , Phe459-Lys478; , D6; , D5; , scrambled Gly442-Lys458. All determinations were performed in triplicate and expressed as means ± SD. SD of D6 was 0.1, too small to be seen.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

This study deals with the binding of HK to a receptor on activated neutrophils. There may be a binding site on unactivated neutrophils, but it must be of very low copy number, below the level of detection of the assay we used. Standard neutrophils contain a considerable number of binding sites but were too variable. Therefore all neutrophils were maximally activated with FMLP for these studies. Because of the increase of binding sites, higher levels of ¹²⁵I-HK were required than previously reported (11). The increase in the number of sites with neutrophil activation is consistent with the identification of Mac-1 as a major binding site for HK (27).

¹²⁵I-HK was employed in these binding studies. As previously described (28), recombinant D3 inhibited the specific binding of HK to human neutrophils, and recombinant D5 displaced specifically bound HK. This inhibition was specific, since GST, the NH₂-terminal portion of the fusion protein containing D3 or D5, failed to inhibit HK or LK binding, and D6, the PK binding site of HK, also did not alter HK binding. To further map the sites on D3, we first expressed recombinant exons 7-9 as fusion proteins with GST. Although we used the GST-kininogen constructs, we have previously shown (19) that, after thrombin cleavage and removal of GST by HPLC purification, the resulting exon products of kininogen behaved similarly to the GST fusion proteins in functional assays in inhibiting thrombin activation of platelets. We demonstrated that exon 7 (Gly²³⁵-Gln²⁹²) and exon 9 (Gly³²⁹-Met³⁵⁷) showed a concentration-dependent inhibition of binding of LK to neutrophils. In contrast, exon 8 showed no concentration dependence, and the inhibition was modest.

To further map each site, we chose specific peptides contained within each of the exon products. We have recently demonstrated that a heptapeptide Leu²⁷¹-Ala²⁷⁷ within exon 7 product could inhibit thrombin activation of platelets with an IC₅₀ of 65 mM (19). Study of a homology model of D3 constructed as previously described (9), based on the crystalline structure of egg white cystatin (2), indicated that this peptide formed a subdomain on the surface. We reasoned that this might serve as a binding site on neutrophils. The results (Fig. 3) show that this peptide is one of the most effective inhibitors of kininogen binding to neutrophils. The specificity is supported by the failure of randomly scrambled peptides to inhibit HK binding. The peptide Glu³⁰⁷-Glu³²³, on the first hairpin loop and the surface of the domain, failed to produce inhibition, consistent with the lack of a concentration-dependent inhibition of binding by exon 8 product.

Herwald et al. (15), using a monoclonal antibody to D3 that blocked biotinylated HK binding to endothelial cells, identified its epitope using synthetic peptides. A peptide contained in the exon 9 product, Leu³³¹-Met³⁵⁷, inhibited HK binding with an IC₅₀ of 60 µM, whereas Cys³³³-Lys³⁴⁵ (unoxidized) showed an IC₅₀ of 113 µM. We synthesized a similar peptide, Cys³³³-Cys³⁵², to test the hypothesis that the region which is in the second hairpin loop of D3 would be a binding site to neutrophils. The oxidized form of the peptide inhibited binding of LK to neutrophils in a concentration-dependent manner.

To further map the site on the unique light chain of HK, we examined various peptides from D5. We had previously described a 17-member peptide, His⁴⁴⁵-His⁴⁵⁷, that functioned as the epitope of a monoclonal antibody, C11C1 (25), which inhibited the coagulant activity of HK and blocked the binding to anionic surfaces (7). Using deletion mutagenesis, we have previously defined two binding sites to anionic surfaces, one similar to His⁴⁴⁵-His⁴⁵⁷ (histidine-glycine-rich region) and a second, His⁴⁷⁵-Lys⁵⁰² (histidine-glycine-lysine-rich region), containing a consensus sequence for binding heparin (9). We therefore studied three peptides derived from these two sites for their ability to block HK binding to neutrophils. Peptides Gly⁴⁴²-Lys⁴⁵⁸ and Phe⁴⁵⁹-Lys⁴⁷⁸, both rich in histidine and glycine, inhibited HK binding significantly in a concentration-dependent fashion (Fig. 4). The randomly scrambled peptide from the sequence Gly⁴⁴²-Lys⁴⁵⁸ did not inhibit HK binding, confirming the specificity of that peptide. In contrast, His⁴⁷⁹-His⁴⁹⁸, rich in histidine, glycine, and lysine, failed to inhibit HK binding at 100-fold molar excess. These results on neutrophils are different from those of Hasan et al. (13), who found that overlapping peptides encompassing a sequence His⁴⁷¹-His⁴⁹⁸ had maximal inhibitory activity for binding of HK to endothelial cells and probably reflect differences in the identity of the receptor(s) between the two cells. It should be noted that D5 and selected peptides show modest inhibition and thus play a lesser role in binding to neutrophils than D3.

The use of three noncontiguous peptidyl sites for binding of a protein ligand to Mac-1 is not unique for HK. Such a situation has been reported for factor X (1). It should be noted that none of these three factor X peptides interferes with HK binding to neutrophils (28). Thus HK binding to neutrophils resembles HK binding to endothelial cells only with regard to the binding site on exon 9 product. The binding site in D5 for endothelial cells contains a sequence with high concentration of basic residues that could bind heparin present on endothelial cell surfaces (5, 22) similar to that in antithrombin. In contrast, the binding sequences in neutrophils are rich in histidine and glycine and contain little lysine. The third, in exon 7 product, the heptapeptide Leu²⁷¹-Ala²⁷⁷, which appears to be potent, was not described as an inhibitor of binding to endothelial cells. Results from our laboratory suggest that LNAENNA, which inhibited thrombin activation of platelets (19), inhibits thrombin binding to platelet GPIb but not HK binding to platelets. It appears that blood and vascular cells have different receptors, since the binding site on neutrophils, _M2, is not present on endothelial cells or platelets. It should be noted that although fibrinogen binding to platelets is Arg-Gly-Asp dependent, binding to neutrophils does not require Arg-Gly-Asp and thus uses different peptidyl motifs (11). However, the binding sites on HK show greater similarity, since the sites on D3 are limited to those exposed on the surface of the molecule, as supported by the homology model for D3 (4, 19). The three-dimensional structure of D3 is required for absolute confirmation of this supposition. Unfortunately, no model or tertiary structure is currently available for the site on D5, and its structure awaits definition by NMR or X-ray crystallography.

The binding of HK is required for kallikrein stimulation of neutrophils as well as the displacement of fibrinogen from the neutrophil surface. Thus peptides that inhibit binding can be expected to decrease neutrophil activation as well as adhesion to perturbed endothelial cells. Therefore such peptides could serve as templates to design peptidomimetic drugs that could decrease unwanted inflammatory responses in blood exposed to artificial surfaces or to cytokines activated in response to sepsis or trauma.