High Molecular Weight Kininogen Regulates Platelet-Leukocyte Interactions by Bridging Mac-1 and Glycoprotein Ib*

Leukocyte-platelet interaction is important in mediat-ing leukocyte adhesion to a thrombus and leukocyte recruitment to a site of vascular injury. This interaction is mediated at least in part by the (cid:1) 2 -integrin Mac-1 (CD11b/CD18) and its counter-receptor on platelets, glycoprotein Ib (cid:2) (GPIb (cid:2) ). High molecular weight kininogen (HK) was previously shown to interact with both GPIb (cid:2) and Mac-1 through its domains 3 and 5, respectively. In this study we investigated the ability of HK to interfere with the leukocyte-platelet interaction. In a purified system, HK binding to GPIb (cid:2) was inhibited by HK domain 3 and the monoclonal antibody (mAb) SZ2, directed against the epitope 269–282 of GPIb (cid:2) , whereas mAb AP1,

High molecular weight kininogen (HK) and especially twochain kinin-free kininogen (HKa) were previously reported to regulate adhesive events (13,14). Initially identified as a nonenzymatic cofactor in the initiation of the contact phase (15), HK appears to be associated with vascular injury, inflammation, or activation of complement in humoral immune defense. In particular, kallikrein can liberate the short lived vasodilator peptide bradykinin from HK thereby generating two-chain kinin-free HKa (15). Domains 3 and 5 in HK mediate the interactions with cell surfaces (15). Apart from heparan sulfate proteoglycans, the binding protein for globular C1q (denoted gC1qR), the urokinase receptor or cytokeratin-1 serve as HKbinding proteins on endothelial cells (16 -20). Moreover, do-main 5 of HK binds to the ␤ 2 -integrin Mac-1 on granulocytes and competes for fibrinogen and ICAM-1 binding in vitro and serves to regulate the Mac-1-dependent leukocyte recruitment in vivo (21). On platelets, cell-associated thrombospondin as well as GPIb mediate the binding of HK (22)(23)(24). Whereas binding to GPIb␣ has been mapped to the region Lys 270 -Gln 292 of HK domain 3, multiple binding sites for HK have been proposed within the GPIb␣ molecule, among others the region 269 -282 that contains three sulfated tyrosine residues, which is recognized by the mAb SZ2 (23). In contrast, the region 201-268 of GPIb␣ recognized by mAb AP1 mediates the Mac-1-GPIb␣ interaction (10).
These diverse observations prompted us to define the role and influence of HK on the Mac-1-GPIb␣ interaction. Our results demonstrate that HK enhances the binding between leukocytes and platelets by bridging Mac-1 and GPIb␣ and thereby provide new insights into the regulatory role of HK for leukocyte-platelet interactions.

EXPERIMENTAL PROCEDURES
Reagents-Single and two-chain HK and HKa were purchased from Enzyme Research Laboratories (South Bend, IN). The purified HK and HKa (Ͼ95%) appeared as a major band of 140 and 110 kDa, respectively, on non-reduced SDS gels. HK was digested with plasma kallikrein (HK to kallikrein ϭ 100:1, mol/mol) for 20 min at 37°C. The resulting HKa was composed of two bands of 62 and 46 kDa when analyzed by reduced SDS-gel electrophoresis. Biotin-labeled HK and HKa were produced as previously described (21). Glutathione S-transferase fused to domains 3, 5, and 6 of HK or to sequences derived from domain 5 were produced as previously described (24). Glutathione Stransferase was amino terminally attached to the following sequences of HK: Gly 235 -Met 357 (domain 3), Lys 420 -Ser 513 (domain 5), Thr 503 -Ser 626 (domain 6) as well as Lys 420 -Asp 474 and His 475 -Lys 502 (amino-and carboxyl-terminal domain 5 sequences). The mutants were purified on a glutathione column reaching more than 90% purity (25). Peptide synthesis and high performance liquid chromatography purification to more than 95% purity were performed by Dr. J. Lambris (University of Pennsylvania, Philadelphia, PA). Refolding of cysteine-containing peptides was carried out by air oxidation for 3 days at 4°C with continuous agitation in buffer containing 50 mM ammonium bicarbonate, pH 8.5, at a final concentration of 100 g/ml followed by freeze drying. In addition to peptides HK406 (Gly 406 -Asn 422 ), HK440 (Gly 440 -His 455 ), HK475 (His 475 -His 485 ), HK483 (His 483 -Gly 497 ), and HK486 (Gly 486 -Lys 502 ), the following scrambled peptides were employed: HK475M (GHHKKHGH-GHH), HK483M (GHHGHKKNGKKKGNK), and HK486M (KGH-KKNGKKNKGNHWGK). Properties of all peptides were described previously (21,26,27). ZnCl 2 was from Sigma. Purified Mac-1, LFA-1, and ICAM-1 were a generous gift from Dr. S. Bodary (Genentech, San Francisco, CA). Glycocalicin, the soluble extracellular domain of GPIb␣ was prepared from washed platelets as previously described (28) and biotinylated with biotinamidocaproate N-hydroxysuccinamide ester (Sigma) by standard methods. Peroxidase-conjugated streptavidin as well as secondary anti-mouse and anti-rabbit immunoglobulins were from Dako (Hamburg, Germany). Polyclonal antibody against GPIb␣ was previously described (28). The following mAb were provided from as Isolation of Cells from Peripheral Blood-Platelets were isolated from ACD anticoagulated blood and washed with Tyrode's buffer containing prostaglandin E 1 and apyrase as previously described (29). Granulocytes were isolated as previously described (9).
Cell Lines-Myelomonocytic cells (U937) were from American Type Culture Collection (ATCC) (Manassas, VA) and were cultured in RPMI 1640 medium containing 10% fetal calf serum. K562 cells (non-transfected, and transfected with LFA-1 or p150,95) were kindly provided by Dr. Y. van Kooyk (Amsterdam, The Netherlands) and were cultured in a mixture of 75% RPMI 1640 medium containing 10% fetal calf serum and 25% ISCOVE's medium supplemented with 5% fetal calf serum. K562 cells transfected with Mac-1 were kindly provided by Dr. M.
CHO cells stably expressing GPIb␤/GPIX were kindly provided by Dr. José A. López (Baylor College of Medicine, Houston, TX). GPIb␣ full-length cDNA was cloned into pZeoSV2ϩ expression vector (Invitrogen, Karlsruhe, Germany) and a stable cell line expressing GPIb␣/ Ib␤/IX was generated by selecting positive clones with Zeocin TM . All culture media were from Invitrogen.
Cell Adhesion Assay-Adhesion of neutrophils, U937 cells, or K562 cells to immobilized ligands, surface-adherent platelets, or CHO cells was carried out as previously described (9,21,30). Briefly, plates were coated with glycocalicin (10 g/ml) or non-transfected CHO cells, GPIb␤/GPIX-transfected CHO cells or GPIb␣/Ib␤/IX-transfected CHO cells were grown to confluency onto 96-well plates. Human platelets (1.5 ϫ 10 7 /well) were added to microtiter plates precoated with 0.2% gelatin. After 1 h at 37°C unbound platelets were removed. Fluorescence labeled neutrophils, U937 cells, or K562 cells (10 5 /well) were washed twice followed by no pretreatment or stimulation with phorbol myristate acetate (50 ng/ml) or the ␤ 2 -integrin stimulating mAb Kim185 (10 g/ml). Cells were washed and added to immobilized glycocalicin, platelets, or CHO cells at 37°C for 60 min in the absence or presence of inhibitors, as indicated in the figure legends. After washing, adhesion of neutrophils, U937, or K562 cells was quantified as the percentage of total cells added using a fluorescence microplate reader (Bio-Tek, Neufahrn, Germany).
Binding of biotin-glycocalicin (0 -500 nM) to immobilized Mac-1 was performed in 20 mM HEPES, 150 mM NaCl, 1 mM Mn 2ϩ , pH 7.2, supplemented with 0.05% Tween 20, 0.1% BSA without or together with 10 M Zn 2ϩ and in the absence or presence competitors. After the incubation period bound biotin-glycocalicin was detected by 1:2000 diluted peroxidase-conjugated streptavidin. Bound streptavidin or secondary antibody were quantified using 2,2Ј-azino-di-3-ethylbenzthiazoline 6-sulfonic acid as substrate (Roche Diagnostics) in a microplate reader (Molecular Devices, Menlo Park, CA) at 405 nm. Nonspecific binding to BSA-coated wells was used as blank in all cases and was subtracted to calculate specific binding.

Characterization of the Interactions between HK and Mac-1 or HK and GPIb␣-
We have previously demonstrated that binding of HK/HKa to Mac-1 is predominantly mediated by domain 5 (21) while binding of HK/HKa to GPIb␣ is mediated by domain 3 of HK (23). In the presence of Zn 2ϩ , HK and HKa bound to both immobilized glycocalicin and Mac-1 (Fig. 1). Binding of HK/HKa to glycocalicin was inhibited by HK domain 3 but not by HK domains 5 or 6. Moreover, mAb SZ2, directed against the 269 -282 region within GPIb␣ that contains three sulfated tyrosine residues, prevented the binding of HK/HKa to glycocalicin, whereas the antibody AP1 recognizing an epitope within the 201-268 region of GPIb␣, previously reported to mediate the binding of GPIb␣ to Mac-1 (10) was not effective in this respect (Fig. 1A). In contrast, the interaction between HK/HKa and immobilized Mac-1 was blocked by domain 5 but not domains 3 or 6 of HK. In addition, peptides from the 475-497 region (HK475, HK483) of HK domain 5 inhibited binding of HK/HKa to Mac-1, whereas scrambled peptides from the same region (HK475M, HK483M) or peptides from other regions of domain 5 (HK406, HK440; not shown) had no effect (Fig. 1B). Thus, binding of HK to Mac-1 is mediated by the 475-497 region of HK domain 5, whereas binding of HK to GPIb␣ is mediated by domain 3. In GPIb␣ the 269 -282 region participates in the binding to HK, whereas the epitope recognized by the mAb AP1 does not.
Effect of HK on the Interaction between Mac-1 and GPIb␣-Because Mac-1 directly interacts with GPIb␣ and HK binds to both adhesion receptors, we tested the effect of HK on the interaction between both molecules. As previously shown, an epitiope within the 201-268 region in GPIb␣ mediates binding to Mac-1 (10). Indeed, binding of GPIb␣ to Mac-1 was inhibited by mAb LPM19c against Mac-1 and mAb AP1 directed to the 201-268 region of GPIb␣ but not by mAb SZ2 directed to the 269 -282 region of GPIb␣. In the presence of HK or HKa (data with HKa not shown) binding of glycocalicin to immobilized Mac-1 was significantly increased about 2-fold. Whereas mAb LPM19c completely abolished the Mac-1-GPIb␣ interaction both in the absence or presence of HK, mAb AP1 provided a complete inhibition of the Mac-1-GPIb␣ interaction in the absence of HK but only a 50% inhibition in the presence of HK. In contrast, the effect of HK on the Mac-1-GPIb␣ interaction was prevented in the presence of mAb SZ2 (Fig. 2A). The enhancement of the Mac-1-GPIb␣ interaction by HK was prevented by both domains 3 and 5 but not by domain 6. Moreover, the peptides HK475 and HK483 also reversed the effect of HK on the Mac-1-GPIb␣ interaction. In contrast, domains 3 and 5 as well as the peptide HK475 and HK483 did not affect the Mac-1-GPIb␣ interaction in the absence of HK (Fig. 2B). Taken together, HK binds with its domains 3 and 5 to GPIb␣ and Mac-1, respectively, and thereby augments the Mac-1-GPIb␣ interaction.
To evaluate better the effect of HK on the interaction between Mac-1 and GPIb␣, increasing concentrations of biotinglycocalicin (0 -500 nM) were bound to immobilized Mac-1 in the absence or presence of increasing concentrations of HK (0 -500 nM). In the presence of HK the amount of bound glycocalicin was almost doubled (Fig. 2C) HK Enhances the Neutrophil-Platelet Binding via Mac-1-GPIb␣ Interactions-Adhesion of myelomonocytic U937 cells to immobilized glycocalicin is mediated by Mac-1, as it was abolished in the presence of mAb LPM19c (not shown). U937 cell adhesion to glycocalicin was stimulated about 2-fold in the presence of HK or HKa. The isolated domains 3 and 5 of HK but not domain 6 reversed the pro-adhesive effect of HK, whereas they did not affect adhesion in the absence of HK (Fig. 3).
To corroborate these results further, we investigated the adhesion of U937 cells to non-transfected CHO cells, to CHO cells transfected with GPIb␤/GPIX, or to GPIb␣/Ib␤/GPIXtransfected CHO cells. The expression level of GPIX was comparable in the CHO cells stably expressing GPIb␤/IX and the CHO cells expressing GPIb␣/Ib␤/IX. Also, the expression level of GPIb␣ was comparable with the level of GPIX in the cotransfected cells (Fig. 4A). No appreciable adhesion of U937 cells to untransfected CHO cells or CHO cells expressing GPIb␤/IX was noted both in the absence or presence of HK. In contrast, Mac-1-dependent adhesion to the CHO cells expressing GPIb␣/Ib␤/IX was detected. Here, cell adhesion was doubled in the presence of HK. Thus, HK enhances Mac-1-dependent adhesion to cells expressing the GPIb␣-Ib␤-IX complex only through its interaction with GPIb␣ (Fig. 4B). Moreover, HK and HKa dose dependently stimulated the adhesion of phorbol myristate acetate-stimulated U937 cells to GPIb␣-transfected CHO cells (Fig. 5A). Analogous to the results obtained with purified components, U937 cell adhesion to GPIb␣-transfected CHO cells was completely abolished by a blocking mAb against Mac-1 independent of the presence of HK/HKa. Moreover, anti-GPIb␣ mAb AP1 blocked U937 cell adhesion to GPIb␣-transfected CHO cells completely in the absence of HK but only partially (about 50 -60% inhibition) in the presence of HK or HKa, whereas anti-GPIb␣ mAb SZ2 only inhibited the stimulatory effect of HK on U937 cell adhesion. Strikingly, HKstimulated adhesion of U937 cells to GPIb␣-transfected CHO cells was almost completely abolished by combining the inhibitory action of both mAbs, AP1 and SZ2 (Fig. 5B). The isolated domains 3 and 5 of HK but not domain 6 reversed the proadhesive effect of HK, whereas they did not affect cell adhesion in the absence of HK (Fig. 5C). Similar results were obtained when we investigated the adhesion of Mac-1-transfected K562 (stimulated with the ␤ 2 -integrin activating mAb Kim185) to immobilized glycocalicin or to GPIb␣-transfected CHO cells (data not shown). In this system, LFA-1-transfected K562 cells did not adhere to CHO cells or GPIb␣-transfected CHO cells in the absence or presence of HK (data not shown).
As, the Mac-1-GPIb␣ interaction mediates the binding between neutrophils and platelets, we sought to determine the effect of HK on the adhesion of neutrophils isolated from human healthy donors to washed, surface-adherent platelets. Neutrophil adhesion to surface-adherent platelets was mediated by Mac-1 and was enhanced in the presence of HK (Fig. 6). Whereas domains 3 and 5 as well as antibody SZ2 reduced the increased adhesion of neutrophils to platelets in the presence of HK, they hardly affected adhesion in the absence of HK; in contrast, mAb AP1 reduced neutrophil adhesion to platelets more efficiently in the absence of HK (Fig. 6). Taken together, HK/HKa can provide an intercellular bridge between Mac-1 on leukocytes and GPIb␣ on platelets.

DISCUSSION
Leukocyte binding to platelets enables the recruitment of leukocytes to sites of vascular injury after denudation of the endothelial cell lining. Leukocyte-platelet interactions are largely controlled by the degree and duration of adhesive contacts between both cell types. A sequential adhesion model of leukocyte attachment to and transmigration across surfaceadherent platelets has been proposed, including P-selectinmediated initial rolling (8) and the subsequent Mac-1-dependent firm adhesion and transplatelet migration (32)(33)(34). Among others, GPIb␣ is an important platelet counter-receptor for Mac-1. Through its propensity to interact with both, Mac-1 and GPIb␣, HK was shown in the present study to augment leukocyte-platelet interactions by bridging the two adhesion receptors. These data provide new insights into the regulatory role of HK on adhesive events between vascular cells with potential implications in atherothrombotic disease.
The following features are consistent with a specific modulatory role of HK on the Mac-1-GPIb␣ interaction. (i) HK domain 5, and particularly the region 475-497, interacts with Mac-1, whereas HK domain 3 directly binds to GPIb␣. The binding site for GPIb␣ in HK was shown to be within the region 270 -292 of HK domain 3 (23). The binding sites for Mac-1 and HK in GPIb␣ are distinct, as binding GPIb␣ to Mac-1 and HK is inhibited by mAb AP1 (recognizing an epitope within the region 201-268 of GPIb␣) and mAb SZ2 (recognizing the region 269 -282 of GPIb␣), respectively (10,23). (ii) Consequently, HK promotes the interaction between isolated Mac-1 and GPIb␣ in a purified system, the Mac-1-dependent adhesion of myelomonocytic U937 cells to immobilized glycocalicin, and provides an intercellular bridge between Mac-1 on neutrophils and GPIb␣ on platelets. These results were corroborated using Mac-1-and GPIb␣-transfected cells. Whereas no significant difference in the affinity of the Mac-1-GPIb␣ interaction was observed in the absence or presence of HK, maximal binding of glycocalicin to immobilized Mac-1 doubles in the presence of HK. Thus, the number of GPIb␣ binding sites in the Mac-1-GPIb␣ interaction doubles in the presence of HK. Therefore, we conclude that Mac-1 simultaneously binds to HK and GPIb␣, with a bound HK molecule binding another molecule of GPIb␣. (iii) The stimulatory effect of HK on the Mac-1-GPIb␣ interaction was blocked in the presence of isolated domains 3 or 5 that compete for HK binding to GPIb␣ and Mac-1, respectively. In contrast, neither isolated domains of HK affected the Mac-1-GPIb␣ interaction in the absence of HK. (iv) Whereas mAb AP1 completely abolished the Mac-1-GPIb␣ interaction in the absence of HK, it only provided a partial inhibition in the presence of HK. In contrast, mAb SZ2 inhibited the stimulatory effect of HK on the Mac-1-GPIb␣ interaction, although it did not affect this interaction in the absence of HK. Strikingly, the combination of both antibodies completely abolished Mac-1-GPIb␣ interaction in the presence of HK. The results with the mAbs strengthen the hypothesis that HK doubles the binding sites of the Mac-1-GPIb␣ interaction. Thus, in the presence of HK additional binding sites on Mac-1 and GPIb␣ are engaged, augmenting cell-cell adhesion between leukocytes and platelets. Although it was previously shown that HK also binds to GPIX (23), this interaction is not sufficient to enhance the adhesion between Mac-1-bearing cells and platelets or cells that express GPIX. Thus, HK enhances Mac-1-dependent adhesion to platelets or cells expressing the GPIb␣-Ib␤-IX complex only through its interaction with GPIb␣. The hypothesis that HK dimers or oligomers may be formed that promote the bridging between the two adhesion receptors needs further experimental proof.
Besides HK, which regulates the Mac-1-GPIb␣ interaction and thereby leukocyte-platelet interactions, additional Mac-1dependent and -independent binding sites on platelets have been described previously. Other potential Mac-1 ligands present on the platelet membrane include junctional adhesion molecule-3 (9), fibrinogen (bound to ␣ IIb ␤ 3 -integrin) (11), ICAM-2 (35), or glycosaminoglycans (36). Further detailed functional analysis needs to be performed to clarify the contribution of each platelet-associated Mac-1 ligand in leukocyte-platelet interactions. Also Mac-1-independent interactions contribute to leukocyte-platelet aggregate formation including thrombospondin bridging between GPIV receptors on platelets and monocytes (37) or binding of P-selectin on activated platelets to leukocyte P-selectin glycoprotein ligand-1 (38, 39). Apart from its role as precursor of bradykinin, which serves important vasodilator functions in vascular homeostasis, HK exerts regulatory activities on hemostasis as well as on cell adhesion with several implications in inflammatory cell recruitment or angiogenesis (21,26). HK and especially HKa and domain 5 were previously described to exert anti-adhesive properties (14), which is not necessarily contradictory to the pro-adhesive effect of HK and HKa, described here. In fact, the data presented here help to explain the anti-adhesive and anti-inflammatory role of the isolated domain 5 observed in an acute peritonitis model in vivo (21). Namely, isolated domain 5 may be anti-inflammatory not only by interfering with the Mac-1-ICAM-1 interaction but also by abrogating the HK-stimulated Mac-1-dependent leukocyte binding to platelets. Furthermore, HK was previously described to regulate platelet aggregation. Domains 3 and 5, as well as fragments thereof inhibit thrombin-induced platelet aggregation and interfere with ligand binding to ␣ IIb ␤ 3 -integrin, respectively (24,40). However, the present work indicates that the antithrombotic properties of these domains might also be attributable to their ability to interfere with Mac-1-GPIb␣-dependent leukocyteplatelet interactions.
In the present work the whole molecule, HK or HKa, has proadhesive functions, whereas isolated domains 3 and 5 are anti-adhesive. This is in accordance with previous reports comparing the actions of the whole HK versus an isolated fragment or domain. In particular, isolated domain 5 inhibited platelet aggregation and prolonged bleeding time in vivo, whereas HK or HKa failed to do so (40). Moreover, HK is proangiogenic as opposed to the anti-angiogenic actions of the isolated domain 5 (26).
Taken together, HK interacts with platelet GPIb␣ and leukocyte Mac-1 via its domains 3 and 5, respectively, providing an intercellular bridge between both adhesion receptors. HK thereby potentially contributes to the accumulation of leukocytes to surface-adherent platelets at sites of vascular injury after denudation of the endothelial cell lining, which is relevant within atherosclerotic and restenotic lesions, or in areas of ischemia-reperfusion injury. HK may provide a novel molecular target for reducing inflammatory cell recruitment in atherothrombotic vascular pathologies.