Two faces of high-molecular-weight kininogen (HK) in angiogenesis: bradykinin turns it on and cleaved HK (HKa) turns it off
Department of Biological Sciences, The University of Southern Mississippi, Hattiesburg, MS 39406, USA; and The Sol Sherry Thrombosis Research Center, Temple University School of Medicine, Philadelphia, PA 19140, USA
Department of Biological Sciences, The University of Southern Mississippi, Hattiesburg, MS 39406, USA; and The Sol Sherry Thrombosis Research Center, Temple University School of Medicine, Philadelphia, PA 19140, USA
R. W. Colman, The Sol Sherry Thrombosis Research Center, Temple University School of Medicine, 3400 N. Broad Street, Room 300 OMS, Philadelphia, PA 19140, USA. Tel.: +1 215 707 4665; fax: +1 215 707 2783; e-mail: firstname.lastname@example.org
Summary. High-molecular-weight kininogen (HK) is a plasma protein that possesses multiple physiological functions. Originally identified as a precursor of bradykinin, a bioactive peptide that regulates many cardiovascular processes, it is now recognized that HK plays important roles in fibrinolysis, thrombosis, and inflammation. HK binds to endothelial cells where it can be cleaved by plasma kallikrein to release bradykinin (BK). The remaining portion of the molecule, cleaved HK, is designated cleaved high-molecular-weight kininogen or HKa. While BK has been intensively studied, the physiological implication of the generation of HKa is not clear. Recent studies have revealed that HKa inhibits angiogenesis while BK promotes angiogenesis. These findings represent novel functions of the kallikrein–kinin system that have not yet been fully appreciated. In this review, we will briefly discuss the recent progress in the studies of the molecular mechanisms that mediate the antiangiogenic effect of HKa and the proangiogenic activity of BK.
High-molecular-weight kininogen (HK) is a glycoprotein with a plasma concentration about 80 µg mL−1 (670 nm). HK is a 120-kDa polypeptide consisting of six domains (designated D1 to D6, respectively) in which each may have distinct functions. HK is divided into a heavy chain (D1 through D3) and a light chain (D5 and D6). The heavy chain and light chain are linked by D4, which contains the sequence of bradykinin (BK) (Fig. 1). After releasing BK by proteolytic cleavage, the cleaved HK (HKa) contains a heavy chain and a light chain that remain connected by a single disulfide bond (Fig. 1). HKa undergoes dramatic conformational changes as detected by electron microscopy . As a result of domain rearrangement, HKa acquires new properties, such as the ability to bind to an anionic charged surface [2,3]. In comparison with HK, HKa has an increased antiadhesive effect, perhaps due to the exposure of D5 on the surface of the molecule. BK regulates a variety of physiological processes, such as inflammation and changes in local blood pressure, resulting in vasodilation and increased microvessel permeability . The gene, protein structures, and domain functions of HK have been extensively studied and reviewed elsewhere [4,5].
Generation of HKa and BK at the endothelial cell surface
HK interacts with blood and vascular cells, including platelets, neutrophils, monocytes, and endothelial cells. On each cell type, HK may exert discrete physiological effects . HK can specifically and reversibly bind to endothelial cells in a Zn2+-dependent manner  through both D3 and D5. HK bound to endothelial cells is a substrate of plasma kallikrein which releases BK, a nonapeptide sequence from D4 through proteolytic cleavage. Therefore, the endothelial cell is an important site for the generation of HKa and BK, which in turn may affect the physiology of endothelial cells. Several endothelial cell membrane proteins that interact with HK/HKa have been identified, including urokinase plasminogen activator receptor (uPAR) , globular C1q receptor (gC1qR) [8,9] and cytokeratin-1 (CK1) . Antibodies to any of these proteins inhibit the binding of HK to endothelial cell membrane. A recent study  revealed that uPAR, gC1qR, and CK1 colocalize on the surface of endothelial cells. The relationship among these proteins has been further defined  by the report that CK1 interacts with either uPAR or gC1qR but uPAR and gC1qR do not directly bind to each other.
Among these proteins, uPAR is particularly relevant to angiogenesis and tumor metastasis . uPAR is a glycophosphatidylinositol-anchored cell surface protein that is expressed in monocytes, endothelial cells and most tumor cells. Upon the binding of prourokinase to uPAR, prourokinase is converted to urokinase, which plays a critical role in initiating cell migration by degrading the extracellular components , a process that is essential for cell penetration through the extracellular matrix (ECM) during angiogenesis and tumor metastasis [15,16]. In addition to its role in the activation of the uPA–plasmin system, uPAR also interacts with integrins at the cell membrane and binds to vitronectin with high affinity through its extracellular portion . As a result of these interactions, uPAR acts as an adhesion receptor for vitronectin and transduces signals through integrin pathways. A previous study  has shown that binding of HKa to endothelial cells was completely blocked by an anti-uPAR antibody as well as by soluble recombinant uPAR (suPAR). Furthermore, in vitro binding assays revealed that HKa and suPAR form a complex and that interaction between HKa and uPAR was through D5 of HKa and D2 and D3 of uPAR. The amino acid sequences responsible for the interaction between HKa and uPAR have been further mapped out in a recent study . In contrast, the binding of HK preferentially utilizes CK1 and gC1qR as receptors. Therefore, it is likely that uPAR–CK1–gC1qR complex is responsible for mediating HK and prekallikrein binding to endothelial cells and subsequent release of BK, a proangiogenic molecule, and HKa, an angiogenic inhibitor.
It is likely that HKa is present in circulation after its formation from cleavage of HK, and therefore may serve as an endogenous angiogenic inhibitor. Whether HKa or HK are further proteolyzed in vivo to release D5 is not known. Existing evidence in vitro supports this possibility. In addition to kallikrein, HK is also subject to proteolytic hydrolysis by enzymes such as plasmin  and neutrophil elastase . A particularly interesting finding recently made by Mauron and Wuillemin  is that HK can be cleaved by factor XIa into three peptides, one of which contains the amino acid sequence between Arg410-Thr503, which is similar to the D5 domain . This result suggests that D5 could circulate in plasma under conditions of increased proteolysis such as cancer.
Evidence for HKa as an angiogenic inhibitor
Angiogenesis is the process of formation of new capillaries from existing blood vessels. It involves several steps, beginning with localized degradation of the basement membrane of the existing vessels. This process is followed by detachment of endothelial cells and migration into the perivascular space where endothelial cells proliferate. The new endothelial cells then form tube-like structures that eventually join and form new capillaries [22,23]. This process is highly regulated by both positive and negative polypeptides [24,25]. Many growth factors, such as bFGF (basic fibroblast growth factor) and VEGF (vascular endothelial growth factor), stimulate angiogenesis. An emerging paradigm has been developed that proteolytic fragments of plasma or ECM proangiogenic proteins are potent inhibitors of angiogenesis [25,26]. Angiostatin, derived from plasminogen, and endostatin (a fragment of collagen XVIII), respectively, are prototypes of this group of polypeptides [27,28]. The current hypothesis is that a balance in favor of angiogenic factors could lead to new vessel formation, whereas the prevalence of angiogenic inhibitors would switch the equilibrium to vessel quiescence or vessel regression.
The first evidence showing that HKa is an angiogenic inhibitor came from a study  which showed that HKa or its recombinant D5 (GST-G5) inhibited endothelial cell proliferation and migration, two important steps required for angiogenesis. Further mapping studies indicated that a peptide (G486-K502) in the D5 region inhibited cell migration, while the peptide G440-H455 was responsible for the inhibition of cell proliferation. The antiangiogenic activities of HKa and D5 were further demonstrated in an in vivo model using chicken chorioallantoic membrane assay. Both polypeptides markedly inhibited bFGF-induced new blood vessel formation . The major findings described in this study were subsequently confirmed  when the antiangiogenic effect of HKa was further demonstrated in two additional in vivo models: matrigel plug assay and corneal micropocket angiogenesis analysis. Recombinant D5 (GST-D5) mimicked the antiangiogenic activity of HKa, indicating that D5 is most likely to be the active region responsible for the antiangiogenic activity of HKa.
Another member of the kininogen family, low-molecular-weight kininogen (LK), a 68-kDa β-globulin with a plasma concentration of 220 µg mL−1, has identical domains D1 through D4 of HK. However, LK domain 5 is completely different from HK and D6 is lacking. LK does not display any antiangiogenic activities exhibited by HKa [29,30,32,33], supporting the hypothesis that D5 is the effective region for the antiangiogenic activity of HKa [30,34]. Therefore, D5 was named kininostatin by Colman et al. .
HKa and D5 inhibit proliferation and modulate the cell cycle
The initial studies showed HKa and D5 inhibited endothelial cell proliferation [29,30]. Further investigation revealed that HKa and D5 inhibited the de novo synthesized DNA in proliferating cells stimulated by bFGF, as demonstrated by bromodeoxyuridine incorporation analysis . D5-inhibited cell proliferation is associated with a decreased expression of cyclin D1 , an important regulator for the G1/S phase transition during cell cycle. HKa also blocked proliferation induced by VEGF, HGF (hepatocyte growth factor) and PDGF (platelet-derived growth factor). D5-treated cells showed the typical morphology of apoptotic cells, which was further confirmed by DNA and nuclear fragmentation assays . Similarly, HKa caused cell death in proliferating cells by apoptosis . To clarify the mechanisms by which HKa and D5 inhibit cell proliferation, we studied additional cell cycle regulators. Cdc2, a cell cycle-dependent protein kinase whose activity is regulated by its association with cyclin A or cyclin B, is required for G2/M transition during cell cycle . Surprisingly, the expression of Cdc2 kinase and cyclin A was significantly increased during HKa- and D5-induced apoptosis of endothelial cells, concurrent with a marked increase of Cdc2 activity . A similar pattern has been reported in apoptotic cells induced by several distinctive stimuli, including tumor necrosis factor-α, radiation , Fas-ligand  and microtubule damaging agents . The increased expression of Cdc2 and cyclin A by HKa was not associated with changes in cell cycle profiles of proliferating cells, but closely correlated with a marked increase of apoptosis. These findings are in line with an emerging hypothesis that Cdc2 and cyclin A are important regulators for the cell cycle, as well as for apoptosis .
Antiadhesive effect of HKa and D5 and its implication in angiogenesis
Vroman and Adams  found that within seconds after normal human plasma contacts an anionic surface, fibrinogen can be detected immunochemically, but within minutes is no longer available to the antibodies. This phenomenon is due to the displacement of fibrinogen by HKa following surface-dependent autoactivation of factor (F)XII to FXIIa [42,43]. FXIIa, both directly and by forming plasma kallikrein, cleaves HK to HKa. The generation of HKa via contact activation of plasma accounts for the ‘Vroman effect’ or physical displacement of adherent fibrinogen from the surface by HKa . In 1992, Asakura et al.  purified a plasma protein that inhibited adhesion and spreading of osteosarcoma. The protein was subsequently identified as HKa. In comparison with HK, HKa has a significantly increased antiadhesion activity. Further studies have revealed that amino acids in a histidine-glycine-rich region (residues 441–457) within D5 are responsible for its antiadhesion effect. Using deletion mutagenesis, a second subdomain has been further defined: a histidine-glycine-lysine-rich region (residues 475–502), which also supports binding to an anionic surface , and zymogen activation of contact proteins is also involved. Under in vivo conditions, the cell membrane may actually serve as a negatively charged surface to assemble the components of the contact system.
Cell adhesion to ECM proteins is crucial for angiogenesis. The ECM not only provides a mechanical support for endothelial cells but also initiates discrete intracellular signaling through integrins, dictating each step of angiogenesis. Without proper attachment to the ECM, many types of cells rapidly undergo apoptosis, a phenomenon known as ‘anoikis’. Endothelial cells are among the cells that are very sensitive to detachment-induced apoptosis. Our recent observations  provide substantial evidence that the antiadhesive activities of HKa and D5 are involved in their apoptotic effect and consequent antiangiogenic activity. The antiadhesive effect depends on which of the ECM proteins the endothelial cells were seeded; HKa inhibited cell adhesion to vitronectin and gelatin, but it had no apparent effect on cell adhesion to fibronectin. HKa selectively induced apoptosis of endothelial cells grown on vitronectin or gelatin but not of the cells grown on fibronectin, which correlates well with its antiadhesive potency . Similar observations were also made by Al-Fakhri et al. . Further results revealed that the antiadhesive and apoptotic effects of HKa are associated with its ability to inhibit phosphorylation of focal adhesion kinase and paxillin , two important signal molecules required for cell adhesion and maintenance of cell viability . These studies suggest that the effects of HKa and D5 in inhibition of cell proliferation, migration and induction of apoptosis are related to their antiadhesive property and regulated by the composition of the ECM. Vitronectin, an abundant provisional ECM component that plays a critical role in angiogenesis during vascular remodeling and wound healing , may be involved in mediating the effect of HKa in vivo. In support of this view, HK (or HKa) has been observed to colocalize with vitronectin at the site of atherosclerotic plaques  where angiogenesis is a prominent event.
The role of uPAR–gC1qR–CK1–uPAR complex in mediating the effect of HKa at the cell surface
The initial hypothesis that uPAR may mediate the apoptotic effect of HKa derived from our study showed that HKa binds to endothelial cells through D2 and D3 of uPAR . The binding of HKa to endothelial cells can be completely blocked by an anti-uPAR D2 and D3 antibody or by soluble recombinant uPAR. Therefore, uPAR could be a logical target of HKa action. The uPAR–gC1qR bimolecular complex may provide a cell surface structure for the assembly of HK/prekallikrein and subsequent generation of HKa and BK [11,12]. This complex may also mediate the effect of HKa in inhibiting cell proliferation and induction of apoptosis. However, this concept was challenged by one study  because none of three antibodies against gC1qR, CK1 or uPAR, which have been previously shown to inhibit HK or HKa binding to endothelial cells [7,8,10], affected the ability of HKa to induce apoptosis of endothelial cells when fibronectin was used to coat the plates. However, in a recent investigation , we were able to show that HKa or D5 completely inhibited uPAR-mediated cell adhesion to vitronectin but not to fibronectin in U937- and uPAR-transfected BAF-3 cells and thereby promoted cell detachment. The antiadhesive effects of HKa and D5 in these cell systems are also critically dependent on the ECM proteins on which the cells were plated . Taken together, these studies suggested that uPAR could be the target of HKa but may require the presence of vitronectin. Unique functions of uPAR specifically associated with vitronectin have been well documented in other types of cells. For example, uPAR expression induces multiple rapidly advancing protrusions that resemble the leading edge of migrating cells. The cytoskeletal changes are independent of uPA but require uPAR binding to vitronectin . In fibrosarcoma (HT-1080) cells, uPA activity is significantly higher in the cells grown on vitronectin than in those grown on fibronectin. Under these conditions, uPAR were detected as clusters in the focal adhesion contacts in the cells grown on vitronectin, but were evenly distributed in the cells grown on fibronectin , indicating that the expression as well as the distribution of uPAR, thus its function, is affected by the nature of ECM proteins. Similar patterns of uPAR distribution as found in fibrosarcoma were observed in endothelial cells (unpublished data). These results may explain why HKa inhibited cell adhesion to vitronectin but not to fibronectin. Vitronectin–uPAR interactions, which are concentrated in the focal contact areas (the most critical part for cell adhesion), play an important role in cell adhesion to vitronectin. On the other hand, uPAR may play a less important role in cell adhesion to fibronectin because of their uniform distribution throughout cells grown on fibronectin, especially in the absence of its ligand, vitronectin. In addition, fibronectin is the ligand for at least 12 integrins, while vitronectin and gelatin (denatured collagens) are the ligands for only four (αvβ, αvβ3, αvβ5, αIIbβ3) and three integrins (αvβ3, α5β1, αIIbβ3), respectively . The larger number of integrins binding to fibronectin probably contributes to the resistance of some cells to the antiadhesive effect of HKa.
Our most recent study  provided more convincing evidence to support the role of uPAR in mediating the effect of HKa. We showed that the apoptotic effect of HKa was blocked by three different anti-uPAR antibodies in cells grown on vitronectin. Further results revealed that uPAR formed a signaling complex containing integrins αvβ3 or α5β1, caveolin, and Src kinase Yes in endothelial cells. HKa physically disrupted the formation of this complex in a manner that paralleled its apoptotic effect. Together with a previous report , we can now establish a signaling cascade in endothelial cells, vitronectin–uPAR–αvβ3–caveolin–Src–FAK–paxillin, which mediates adhesion, proliferation, and survival of many types of cells. We proposed an action model of HKa (Fig. 2). HKa disrupts this signaling cascade by at least two mechanisms. First, HKa directly binds to the D2 and D3 domains of uPAR through its D5 region, thus preventing the binding of vitronectin to uPAR and its subsequent interaction with integrins as well as other signaling molecules. Second, through its direct binding to the amino terminal region of vitronectin , proximal to the RGD region (integrin binding site), HKa can potentially disrupt the binding of vitronectin to integrins that utilize vitronectin as ligand, such as αvβ3. As a result of either of these mechanisms, cells will not be able to adhere to vitronectin properly and will eventually undergo apoptosis. It should be pointed out that this model is a simplified version of the complex interactions among signaling molecules. Many other molecules, such as α5β1, gC1qR, and CK1, are probably involved but not included in the model for simplicity. We also emphasize that this model does not exclude the possibility that other targets and mechanisms of HKa action may exist in endothelial cells in addition to uPAR and vitronectin. For example, HKa can directly bind to Mac1 integrin found in leukocytes, thus blocking adhesion of HEK293 cells transfected with Mac1 to fibrinogen and intercellular adhesion molecule-1 . Therefore, HKa may potentially disrupt integrin-mediated adhesion in a similar manner through its direct association with certain integrins in endothelial cells. It was recently reported that the apoptotic activity of HKa may be mediated through its interaction with tropomyosin . In addition, HK has been shown to bind heparin sulfate proteoglycans at the endothelial cell surface , which may also be involved in mediating the effect of HK/HKa. However, other studies demonstrate that complete removal of sulfated mucopolysaccharides by heparinases did not alter HK binding . Apparently, the interaction between HKa and cell surface proteins is highly complex and requires further investigation.
Proangiogenic activity of HK
Inhibitors of angiogenesis are often proteolytic fragments of plasma or ECM proteins that are proangiogenic. Angiostatin, derived from plasminogen, and endostatin, a fragment of collagen XVIII, are prototypes of this group of polypeptides [27,28]. Tumstatin [58,59], derived from type IV collagen, is another example. The generation of HKa from HK through proteolytic cleavage follows a similar pattern. Therefore, we postulated that intact HK would be proangiogenic. HK indeed stimulated angiogenesis, as demonstrated in the chicken chorioallantoic membrane (CAM) assay . Inhibition of kallikrein blocked the proangiogenic activity of HK, suggesting that BK was responsible for the stimulation of neovascularization. In supporting this conclusion, we further showed in a recent study  that a monoclonal antibody to HK (mAb C11C1), which blocks HK binding to endothelial cells, inhibited tumor growth in a nude mouse model in addition to significant reduction of microvascular density. This finding explains the observation that mAb C11C1 inhibits HK stimulation of angiogenesis as well as FGF2- or VEGF-induced neovascularization in the CAM . We show that mAb C11C1 may prevent the binding of HK to endothelial cells and thus subsequent generation of BK [60,61]. Recently, we showed that the same antibody is therapeutic in a rodent model of reactive arthritis , where angiogenesis may play a significant role in the pathogenesis of the joint disease. This mechanism is further supported by the finding that angiogenesis is suppressed in kininogen deficiency in rats .
The proangiogenic effect of BK has been demonstrated in both in vitro and in vivo studies under different experimental settings. BK promotes angiogenesis by upregulation of endogenous bFGF through the B1 receptor [64,65] or VEGF through the B2 receptor , by regulation of vascular permeability [67,68], or by stimulation of cell proliferation through the B2 receptor . In a recent study by Thuringer et al., it was found that BK transactivated VEGF receptor, KDR/Flk1, resulted in a further eNOS activation and other cellular activities in endothelial cells associated with angiogenesis . Using a B2 receptor-deficient mouse () model, it has been demonstrated that the proangiogenic effect of angiotensin-converting enzyme inhibition is mediated by the bradykinin B2 receptor . These studies suggest that BK may modulate several cell-signaling pathways through B1 and/or B2 receptors and thus play an important role in regulating angiogenesis. These combined studies emphasize the physiological and pathophysiological importance of the kinin-generating system in angiogenesis.
We thank V. Sheaffer for her help in editing and manuscript preparation. Research Grant Support: National Institutes of Health grants R01 AR051713 (R.W.C.), R01 CA63938 (R.W.C.), and a grant from the American Heart Association 0265404U (Y-L.G.).