Bradykinin is a potent inflammatory mediator that induces vasodilation, vascular leakage, and pain sensations. This short-lived peptide hormone is liberated from its large precursor protein high molecular weight kininogen (HK) through the contact system cascade involving coagulation factor XII and plasma kallikrein. Although bradykinin release is well established in vitro, the factors and mechanisms controlling bradykinin generation in vivo are still incompletely understood. In this study we demonstrate that binding of HK to glycosaminoglycans (GAGs) of the heparan and chondroitin sulfate type efficiently interferes with bradykinin release in plasma and on endothelial surfaces. Proteolytic bradykinin production on endothelial cells is restored following degradation of cell surface GAG through heparinase. Alternatively, application of HK fragments D3 or light chain, which compete with uncleaved HK for cell binding, promote kininogen proteolysis and bradykinin release. Intravital microscopy revealed that HK fragments increase bradykinin-mediated mesentery microvascular leakage. Topical application of D3 or light chain enhanced bradykinin generation and edema formation in the mouse skin. Our results demonstrate that bradykinin formation is controlled by HK binding to and detachment from GAGs. Separation of the precursor from cell surfaces is a prerequisite for its efficient proteolytic processing. By this means, fragments arising from HK processing propagate bradykinin generation, revealing a novel regulatory level for the kallikrein-kinin system.
Enhanced vascular permeability resulting in tissue edema formation is a hallmark of inflammatory responses to tissue destruction or infection. Following perturbation or damage both the endothelium and subendothelial surfaces serve as platforms for procoagulant and proinflammatory proteolytic cascades. A case in point is the plasma-born contact system linking inflammation and blood coagulation. The system encompasses three enzymatic factors, factor XII (FXII,3 Hageman factor), factor XI, and plasma kallikrein (PK), and the nonenzymatic cofactor high molecular weight kininogen (HK), which forms equimolar complexes with PK or factor XI, respectively (1, 2, 3). The cascade is initiated through binding of FXII to negatively charged biological or artificial surfaces such as subendothelial collagens, bacterial membranes, kaolin, and dextran sulfate collectively referred to as “contact phase” (4, 5). These surfaces induce autoactivation of the zymogen, FXII, to the active enzyme (FXIIa) that in turn converts HK-bound PK to its active form, active PK (PKa) (6). Reciprocal activation of endothelium-bound FXII and PK amplifies the initial signal through a positive feedback loop, and PKa efficiently splits HK to release the peptide hormone bradykinin. In this way, a circumscribed burst of bradykinin activity is produced at the site of injury (7, 8). The short-lived peptide hormone bradykinin acts in a paracrine mode and activates adjacent G protein-coupled receptors of the B2 type (B2R). Through the production of messengers and effectors such as NO, cGMP, and PGI2 (9, 10, 11), bradykinin causes vasodilation, enhanced endothelial permeability, and vascular leakage in microcapillaries (12, 13, 14). Excessive bradykinin generation is symptomatic of hereditary or acquired angioedema in patients (15, 16), and blockage of kinin action in mouse models of angioedema rescues this phenotype (13). Bradykinin-induced vascular leakage and edema formation are also critically involved in sepsis, vasculitis, allergic asthma, rhinitis, diabetes, and cancer (11, 14, 17, 18, 19). Importantly, bradykinin generation on bacterial surfaces is crucial for the massive infiltration and extravasion of plasma compounds into lung tissues seen in Salmonella infections (20). Thus, release of this potent peptide hormone needs to be tightly controlled in vivo.
It has long been known that HK, the 118-kDa precursor to the nonapeptide bradykinin, is associated with surfaces of endothelial cells and neutrophils (21, 22), and binds to heparin, i.e., the mast cell-derived glycosaminoglycan (GAG) (23, 24, 25). It has only recently been demonstrated that HK attaches to endothelial cell surfaces through domain 3 of the kininogen (D3) and domain 5 of the HK heavy chain (D5H) by docking to the heparan sulfate (HS) and chondroitin sulfate (CS) chains of proteoglycans (26, 27). However, functional consequences of HK binding to GAG on endothelial cells are unknown. In addition to their prominent role as structural components, GAGs have been shown to modulate the availability and activity of cytokines, chemokines, morphogens, and plasma proteins (reviewed in Ref.28). For example, binding to GAG has been shown to provide cellular reservoirs for cytokines such as fibroblast growth factors (29). Importantly, binding of such factors to heparin and HS transiently protects them from proteolytic and chemical inactivation (30, 31, 32). Therefore, association of hormonal factors with cell surfaces appears to promote paracrine signaling. Short-lived signaling molecules can be generated next to their cognate receptors in a protected microenvironment thereby escaping rapid degradation. Thus, GAG have the potential to regulate peptide hormone bioactivity both in vitro and in vivo.
In this study we show that binding of HK to GAG protects HK from cleavage. Degradation of surface-associated GAG through heparinases and chondroitinases resumes proteolytic bradykinin generation from HK on artificial and cellular surfaces. Furthermore, HK fragments arising from HK proteolytic processing efficiently detach the bradykinin precursor from GAG, and thereby catalyze hormone generation. We also demonstrate that in mice models of vascular permeability, the HK fragments of domain D3 and light chain boost hormone generation, causing edema formation in the skin. Intravital microscopy reveals that HK fragments promote bradykinin-induced microvascular leakage in mesenteric microvessels. We conclude that binding to GAG protects HK from proteolytic processing, and that detachment of HK from cell surface GAG is necessary for bradykinin generation in vivo. Therefore, GAG contributes to local bradykinin release revealing a novel regulation in the kallikrein-kinin system in vivo.
Materials and Methods
All experiments and animal care were approved by the local Animal Care and Use Committee. Wild-type C57BL/6J mice were purchased from Charles River Breeding Laboratories.
Materials and chemicals
HK was isolated and radiolabeled using Na125I and IODO-GEN (Pierce) as previously described (26). Following incubation with glass beads for 30 min at 37°C, human FXII (Enzyme Research Laboratories) was converted >98% into the active protease, FXIIa, and activation was monitored by Western blotting (33). Bradykinin concentrations were determined by MARKIT-M-Bradykinin ELISA according to the manufacturer’s instructions (Dainippon Pharmaceutical). Human HK-deficient plasma was obtained from George King Biomedical. Human citrate plasma was obtained from healthy volunteers (University Hospital of Wuerzburg, Bavaria, Germany). Production and characterization of mAbs to human HK (34, 35) and of polyclonal Abs to peptides bradykinin (anti-bradykinin; AS 348), LDC27 (anti-LDC27; AS 303), and HKH20 (anti-HKH20; AS 365) and to PK (PKH6) were described previously (21, 22, 26). HK light and heavy chains were kindly donated by Dr. R. Vogel (Technische Universität München, Munich, Germany). The PK substrate d-Pro-Phe-Arg-p-nitroanilide (S-2302) was obtained from Chromogenix and used as described previously (36). Ab F(ab′)2 were generated with ImmunoPure F(ab′)2 kit according to the manufacturer’s instructions (Pierce). HS, CS, dextran sulfate, bradykinin, and B2R antagonist HOE140 were from Sigma-Aldrich; elastase from human neutrophils (EC 126.96.36.199) was purchased from Serva; human mast cell tryptase (EC 188.8.131.52) was purchased from Promega; heparinase I (EC 184.108.40.206) and heparinase III (EC 220.127.116.11) were obtained from Seikagaku Kogyo; and chondroitinase ABC (EC 18.104.22.168) was purchased from Roche.
Endothelial cell lineages EA.hy926 (immortalized human endothelial cells), which was a generous gift from Dr. C. J. Edgell (University of North Carolina, Chapel Hill, NC), and MyEnd (37) were cultured in a humidified CO2 atmosphere at 37°C in gelatin-coated flasks with DMEM (Invitrogen Life Technologies) containing 4.5 g/L glucose, 10% FBS, and 0.01% penicillin/streptomycin. For EA.hy926 cells, medium was supplemented with 100 μM hypoxanthine, 0.4 μM aminopterin, and 16 μM thymidine.
HK cleavage was initiated in human citrate plasma using 0.8 nM FXIIa (molar ratio of FXII over FXIIa was 500:1) (33), 100 nM elastase, 100 nM tryptase (38), or 50 nM PK; in our hands, addition of a defined amount of preactivated FXIIa improved reproducibility over kaolin- or glass-induced FXII activation in the subsequent assays. Following incubation for 30 min at 37°C, the reaction was stopped by addition of sample buffer containing 4% SDS, and 0.3 μl of plasma samples was run under reducing conditions at 30 mA for 90 min in a linear 10% polyacrylamide gel with 0.1% SDS. Proteins were transferred to nitrocellulose, and the membranes were blocked with PBS containing 5% milk powder and 0.05% Tween 20. Monoclonal Ab to bradykinin (MBK)3 in PBS/milk (0.5 μg/ml) was used to detect uncleaved HK cleavage. Bound Abs were detected by HRP-coupled secondary Abs to mouse Ig (DakoCytomation) and the ECL detection kit (ECL; Amersham Biosciences). To test for the interference of Abs and GAG with HK processing in plasma, Abs or F(ab′)2 to domains D5H and D6H of HK were adjusted to 675 nM in 100 μl of PBS, and then 100 μl of plasma were added (molar ratio of Ab over HK = 1.5). In the case of Abs to domains D1–D4, which cross-react with low m.w. kininogen, we applied 2.6 μM of the Abs (1.5-fold molar excess over total kininogens). Similarly, serial dilutions (1/10) of HS or CS in plasma were prepared from 1 mg/ml to 1 ng/ml. Proteolytic processing was started after 5 min by the addition of 10 μg/ml kaolin or by FXIIa (0.8 nM final concentration), and HK processing was followed by Western blotting with MBK3. To test for HK binding and processing on cell surfaces, we applied 60 nM HK (including 20 nM 125I-HK) to the culture medium for 1 h at 37°C, followed by extensive washing and incubation with HK-deficient plasma containing heparinase or chondroitinase (final concentration 0.1 or 1 U/ml) and serial dilutions (1/2) of competitors D3 or D1 (down from 100 μM) (26). HK conversion was initiated by the addition of 0.8 nM FXIIa. For cleavage of GAG, 1 U/ml heparinase (mixture of heparinases I and III, 0.5 U/ml each) or chondroitinase was used (26). Pretreatment of HS-coated plates with heparinase completely abrogated HK binding, demonstrating the specificity of the interaction. Bound 125I-HK was quantified by a gamma counter.
Quantification of cell surface HS
Endothelial cell leakage
MyEnd and EA.hy926 cells were seeded at 105 cells/well onto polycarbonate inserts of a 12-well Transwell system (12 mm in diameter, 3-μm- diameter pores; Corning Glass), coated with 2% gelatin, and cultured for 2 days to reach complete confluency. Before seeding, using LipofectAMINE (Invitrogen Life Technologies) EA.hy926 cells were transiently transfected with cDNAs encoding the bradykinin B2R (40) and the endothelial NO synthase (41) in pcDNA3 vectors. The medium was replaced daily. Permeability increase stimulated by heparinase and FXII was examined using FITC-dextran (1 mg/ml, molecular mass 70 kDa; Sigma-Aldrich) as tracer loaded onto the upper chamber. After 20 min of stimulation the amount of FITC-dextran in the lower chamber was determined photometrically.
Cloning and recombinant expression of HK domains
Cloning, expression, and purification of maltose-binding protein fused to HK domains D1, D3, or D5H were conducted as previously published (42). SDS-PAGE and Western blotting using domain-specific Abs demonstrated that the purity of the fusion proteins was >95%.
Mapping of HK Abs
Epitopes of mouse monoclonal and rabbit polyclonal purified Abs (34, 43, 44) were further refined using recombinantly expressed domains of HK. The following assignments were made: Abs HKH4, HKH18, and HKH19 map to positions 1–123 of HK (domain D1); HKH6 and HKH8 to positions 130–244 (D2); HKH14 and HKH16 to 252–331 (D2); HKH10 and HKH13 to 331–357 (D3); HKH15 to 333–345 (D3); anti-LDC27 to 331–357 (D3); MBK1 and MBK3 to 363–371 (D4); anti-bradykinin to 363–371 (D4); HKL6 and HKL8 to 372–383 (D4); HKL10 and HKL13 to 402–419 (D5H); HKL12 and HKL14 to 420–492 (D5H); anti-HKH20 to 479–498 (D5H); HKL24 to 543–554 (D6H); HKL16, HKL18, and HKL20 to 569–596 (D6H); and HKL22 and HKL23 to 608–626 (D6H).
Analysis of microvascular leakage by intravital microscopy
Microscopic observations were made by confocal laser scanning microscopy (C1; Nikon) using EC-C1 2.10 software (Nikon). Images were analyzed by EC-C1 2.10 Viewer and ImageJ 1.34 NIH software. To prepare mesentery vessels for microscopy, male C57BL/6J mice were anesthetized by i.p. injection of 2,2,2-tribromoethanol and 2-methyl-2-butanol (Sigma-Aldrich; 0.15 ml/10 g of body weight using a 2.5% solution). The mesentery was externalized through a midline abdominal incision, placed on a heating pad (Linkam Scientific Instruments) and fixed plain, and microvessels (35–60 μm diameter) were visualized by a Nikon Eclipse E600 with a ×10 objective. Tissue scans were performed in 30-s intervals. To maintain physiological pH and moisture, exposed tissue was superfused with HEPES-buffered saline at 37°C. This procedure had no effect on rectal temperature and blood pressure. FITC-dextran (70 kDa; Sigma-Aldrich) was injected at 10 mg/kg body weight into the retro-orbital plexus as a plasma tracer to permit visualization in fluorescence light of changes in vascular permeability. After topical application of FXIIa (25 nM) and/or D3 (10 μM) in saline, macromolecular leakage was monitored for 20 min as previously described (45, 46).
Quantification of plasma leakage in mouse
Changes in permeability of microvessels of the skin were monitored by extravasation of intracardial administered Evans blue dye (13, 47). Briefly, anesthetized animals were injected into the left atrium with 30 mg/kg body weight of Evans blue dissolved in 0.9% saline. Five minutes after injection, FXIIa (25 nM in 0.1 ml of saline), D3, HK light chain (10 μM), heparinase (1 U/ml), HOE140 (50 or 500 nM), bradykinin (100 nM), or saline alone was administered intradermally into the dorsal skin of mice. After 15 min the animals were sacrificed, the skin at the injected site was excised, weighed, and minced. The homogenate was diluted with 2 volumes of formamide and incubated at 60°C for 24 h, followed by centrifugation at 5000 × g for 30 min. The supernatant was collected, and the concentration of the extracted dye was measured colorimetrically at 620 and 740 nm, and standardized to dry tissue weights.
GAG inhibits proteolytic processing of HK in plasma
Initially we assessed the role of GAG for kininogen processing in the fluid phase. To monitor proteolytic processing of HK we took advantage of a mAb that targets the bradykinin moiety in uncleaved (bradykinin-containing) HK but shows no cross-reactivity with the corresponding cleaved (bradykinin-free) form (35). Human plasma samples were supplemented with increasing concentrations of HS, ranging from 1 ng/ml to 1000 μg/ml (final concentrations), and HK cleavage was initiated with 0.8 nM FXIIa to activate PK. At concentrations of ≤1 μg/ml, HS failed to interfere with HK processing, as indicated by the complete loss of the bradykinin signal. Higher HS concentrations (1 μg/ml) partially inhibited HK cleavage, and HK conversion was blocked completely at concentrations ≥10 μg/ml (Fig. 1,A). To confirm that the disappearance of uncleaved HK corresponds to the release of the kinin peptide, we followed bradykinin concentrations in plasma by ELISA. We found that the increase in generated bradykinin (Fig. 1,C) correlated inversely with the disappearance of uncleaved HK (Fig. 1,B). To exclude an inhibitory effect of HS on the processing enzymes, we measured the enzymatic activity of PK by a chromogenic assay and found that HS at concentrations ≤100 μg/ml did not significantly interfere with the amidolytic activity of PKa (Fig. 1,C, inset). Only at the highest HS concentration (1000 μg/ml), a moderate inhibition of the amidolytic activity (<35%) was seen, indicating that the observed attenuation of HK processing was not due to a direct inhibition of PKa by HS. Next, we asked whether other GAG types such as CS would also attenuate HK cleavage. At a concentration of 1 μg/ml, CS partially blocked HK conversion and bradykinin liberation. At ≥10 μg/ml this GAG type completely inhibited HK processing (Fig. 1,D). As a control, we tested HS without addition of FXIIa and found no cleavage of HK (Fig. 1,E). Importantly, inhibition of HK processing by GAG was not restricted to PKa. Addition of HS (≥1 μg/ml) efficiently blocked HK cleavage by a mixture of neutrophil elastase and mast cell tryptase known to efficiently liberate bradykinin from HK (38) (Fig. 1 F). Next we investigated the molecular mechanisms underlying GAG-mediated protection from proteolytic processing.
HK cell binding sites are important for proteolytic processing
First we asked which portions of HK are critically involved in its proteolytic processing in the fluid phase. To this end we used a panel of monoclonal and polyclonal Abs that target 16 distinct epitopes (21, 22, 34, 35) of the entire human HK, comprising domains D1 through D6H (Fig. 2). Following incubation of human plasma with a 1.5-fold molar excess of Ab over endogenous HK, we initiated HK cleavage by FXIIa and monitored for uncleaved HK after 30 min. Of 27 distinct Abs, 15 significantly inhibited bradykinin liberation from HK, although to varying degrees. The inhibiting Abs mapped to four portions of HK, denoted as groups a–d (Fig. 2,A). Group a Abs (exemplified by HKH13, HKH15, anti-LDC27) were directed to the cell binding site located in domain D3 (residues 331–351) (22); group b Abs (MBK3, anti-bradykinin, HKL6) map to the kinin domain (residues 363–383); group c Abs (HKL14, anti-HKH20) target the cell binding site in D5H (420–498) (21); and group d Abs (HKL16) recognize the PK binding site in D6H (569–596) (48). Likewise, F(ab′)2 generated from nine representative Abs covering groups a–d inhibited HK processing, whereas F(ab′)2 from Abs to epitopes external to these regions (HKL10, HKL22, HKL24) failed to interfere (Fig. 2 B). Thus, at least four portions of HK must be accessible for bradykinin release to occur, and blocking of a single portion is sufficient to attenuate proteolytic processing of HK in plasma. As controls, we incubated FXIIa-containing plasma without addition of Abs and found complete HK cleavage. In contrast, the anti-PK Ab (PKH6), which blocks HK-PK complex formation (7), inhibited bradykinin liberation. Although the finding that regions exposing the kinin segment (group b) and the PK binding site (group d) prevent HK processing is not unexpected, the interference of groups a and c Abs with the proteolytic processing of HK was rather surprising. Importantly, the target domains of groups a and c Abs, i.e., D3 and D5H, respectively, have been previously mapped as the major GAG attachment sites of HK (26, 27).
HK binding to immobilized GAG on surfaces inhibits bradykinin generation
Given that GAGs inhibit HK cleavage in the fluid phase, we asked whether they might also attenuate bradykinin generation on artificial surfaces. Therefore, we immobilized purified uncleaved HK on HS or CS covalently bound to titer plates and applied HK-deficient plasma supplemented with 8 nM FXIIa. Due to the high affinity of HK for GAG, this experimental setting allowed us to test almost exclusively for GAG-bound HK. Following incubation for 30 min at 37°C, we probed for HK cleavage, bradykinin, and GAG-bound 125I-HK. We found that ≥97% of total HK remained in the uncleaved form (Fig. 3, A, left lanes, and B, left), and the bradykinin concentration was very low at <30 ng/ml (Fig. 3,C, left). In these samples 99% and 96% of HK was still bound to HS and CS, respectively (compared with coated HK, set 100%) (Fig. 3,D, left). Hence, GAG-bound HK appears to be protected from proteolytic cleavage by FXIIa-activated PK. To follow this hypothesis, we digested surface-bound HS and CS with heparinase and chondroitinase in the presence of HK-deficient plasma supplemented with FXIIa. Following GAG degradation, HK binding capacity decreased to 6% (HS) and 11% (CS) (Fig. 3,D, right). The loss of HK binding capacity was paralleled by a drastic increase in HK processing and bradykinin generation, i.e., only 3% (HS) and 13% (CS) of HK was still present in the uncleaved form (Fig. 3, A, right lanes, and B, right), whereas the bradykinin concentrations increased to >750 ng/ml (Fig. 3,C, right). Similarly we found GAG-mediated protection of HK from proteolytic processing by 50 nM PKa (Fig. 3,E, left lanes) or a mixture of 100 nM elastase and 100 nM tryptase (Fig. 3,F, left lanes), whereas prior degradation of GAG abolished this effect (Fig. 3, E and F, right lanes). Hence, the observed protection of GAG-bound HK from proteolytic procession is independent of the protease source, i.e., FXIIa-activated endogenous PK vs added PKa, and kininogenase type, i.e., PKa vs elastase/tryptase. To rule out the possibility that GAG degrading enzymes may contain traces of kininogenase activity we incubated plasma with heparinase or chondroitinase and found no processing of HK (Fig. 3 G). These findings indicate that binding of HK to GAG protects it from proteolytic cleavage on surfaces.
Cell-bound HK is protected from proteolysis
The dominant GAG type on endothelial cells is HS (28). To test our hypothesis in a cellular system, we incubated monolayers of EA.hy926 or MyEnd endothelial cells with 60 nM radiolabeled HK for 30 min, removed the supernatants and then added HK-deficient plasma including FXIIa and heparinase (0–1 U/ml). Following incubation, we monitored for HK cleavage (Fig. 4, A and B), bradykinin generation (Fig. 4,C), cell-bound HK (Fig. 4,D), cell-associated HS (Fig. 4,E) using radiolabeled 10E4 Ab specifically recognizing intact HS chains exposed by proteoglycan (39), and endothelial leakage (Fig. 4 F). In the absence of heparinase, 98% of HK was bound to cell surfaces as compared with controls (set to 100% before addition of FXIIa). Cell-bound HK was almost exclusively (>97%) in the uncleaved form, and the bradykinin concentration in the sample was very low (<25 ng/ml). Digestion of HS by heparinase lead to a dose-dependent decrease in the fraction of uncleaved HK to 48% (39%) at 0.1 U/ml and 6% (5%) at 1 U/ml for EA.hy926 (and MyEnd) cells. Enhanced processing of HK was paralleled by a decrease in HK cell-binding to 44% (40%) and 10% (13%) on EA.hy926 (MyEnd) cells. Cell surface-bound HS was reduced to 47% (46%) and 8% (9%) on EA.hy926 (MyEnd) cells with 0.1 and 1.0 U/ml heparinase, respectively. Additionally, we analyzed for macromolecular leakage representing a typical bradykinin-induced effect on endothelial cell monolayers. FXIIa/heparinase dose-dependently elevated the permeability of EA.hy926 (MyEnd) monolayers to 82% (119%) and 140% (221%) compared with heparinase-treated cells in the absence of FXIIa (control, 0%). Importantly, HK was not protected from proteolytic processing due to internalization because stripping of the cells with dextran sulfate (49) completely recovered cell-bound HK (data not shown). We conclude that binding to HS on endothelial cells protects HK from proteolytic processing and that detachment of HK from GAG on cell surfaces appears to be necessary for bradykinin liberation.
HK fragments promote proteolytic bradykinin generation on endothelial cells
Cleavage of cell surface GAG by GAG-digesting enzymes may contribute to bradykinin generation in disorder states such as cancer (50). However, it is likely that other mechanisms exist that may allow local propagation of HK detachment and thus bradykinin generation following an initial trigger of the contact system. We have previously reported that HK fragments containing domains D3 and D5H efficiently compete with HK for HS binding and cell surface association (26, 51). We therefore tested recombinant D3 and purified HK light chain (harboring D5H) for their effect on bradykinin generation. Uncleaved HK was bound to the surface of EA.hy926 cells and HK-deficient plasma, including 0.8 nM FXIIa, was added in the presence of increasing concentrations of D3 or L chain (0.8–100 μM). Recombinant HK domain D1 that lacks a GAG or cell binding site was used as a control. Following incubation the samples were analyzed for uncleaved HK, bradykinin concentration and HK cell binding. At a concentration of ≥3.2 μM, domain D3 significantly reduced the fraction of uncleaved HK (down to 74%), and at ≥26 μM the uncleaved HK was almost completely absent (<7%), whereas the bradykinin concentration increased concomitantly (Fig. 5,A, second and third panels). A dose-dependent decrease in the binding of radiolabeled HK to EA.hy926 cells was observed, thereby paralleling the decay of uncleaved HK (Fig. 5,A, bottom panel); essentially the same effects were found for the HK light chain. The fragment displaced HK from cells, and the loss of HK cell binding was accompanied by a dose-dependent decrease of uncleaved HK and an increase of bradykinin formation, respectively (Fig. 5,B). Similar results were found using the isolated HK heavy chain or recombinant D5H domain (data not shown). In contrast, domain D1 failed to induce HK displacement or processing and thus bradykinin generation, even at the highest concentrations applied (100 μM; Fig. 5 C). It appears that HK is protected from proteolytic processing by binding to cell surface GAG, and that this inhibition can be overcome by the displacement of HK from cell surface GAG through competitors that emerge from HK cleavage.
HK fragments increase microvascular leakage in vivo
If correct, our notion would predict that efficient HK processing, and thus bradykinin generation, is critically dependent on two stimuli: protease activity and detachment of bound HK from GAG. To test this hypothesis, in vivo intravital laser scanning confocal microscopy was used to follow a prototypical bradykinin-mediated effect, i.e., vascular leakage. To visualize the paracellular transition of macromolecules from plasma of mesentery microvessels into the surrounding mouse tissues, FITC-dextran was used as a plasma tracer. Administration of HK domain D3 (10 μM) in combination with FXIIa (25 nM) provoked prompt macromolecular leakage from postcapillary and small venules. Leaky spots appeared within 30 s and a maximum density of spots occurred after 10 min (Fig. 6, A and B). In contrast, individual application of D3 or FXIIa alone failed to significantly alter vascular permeability within 10 min (Fig. 6, C and D). These data support the idea that efficient bradykinin generation in biological processes depends on two signals: generation of PK activity and mobilization of HK from cell surface GAG.
HK fragments boost bradykinin-mediated edema formation in the skin
To further determine the prerequisites for bradykinin-mediated vascular leakage, edema formation in mouse skin was analyzed (Fig. 7). Subdermal application of 100 μl of 25 nM FXIIa increased plasma leakage of Evans blue by ∼10%, compared with control saline treatment. The application of 10 μM D3 (or HK light chain, related results shown in parenthesis) stimulated microvascular permeability by 3% (1%), whereas simultaneous application of D3 and FXIIa enhanced permeability by 51% (64%). This latter effect was inhibited by HOE140, a specific antagonist of the B2R: at concentrations of 50 and 500 nM it reduced vascular leakage to 22% (26%) and 10% (7%), respectively. Similarly, application of 1 U/ml heparinase together with FXIIa increased permeability to 50% (compared with 5% induced by heparinase alone) and the increase of permeability was reduced to 29 and 9% by 50 and 500 nM HOE140, respectively. For comparison, we directly applied 100 nM bradykinin and found a 44% increase of vascular leakage over the saline control, whereas HOE140 alone failed to induce edema formation. Collectively, our data indicate that HK processing, and thus bradykinin generation, are carefully controlled by HK attachment to and detachment from GAG exposed by endothelial surfaces. HK fragments resulting from HK proteolytic cleavage may play a critical role in propagating the initial stimulus of HK processing, thereby suggesting an obvious explanation for local bradykinin generation in vitro and in vivo.
Understanding the molecular mechanisms involved in the fine-tuning of hormonal release, signaling, and degradation has become a major issue in molecular endocrinology. In particular, the development of novel therapeutic strategies relies on the detailed knowledge of key players in this system and an in-depth understanding of their structural and functional interactions. In the cardiovascular system, the contact phase-driven kinin production linked to vascular inflammation, local blood flow regulation, and pain sensations provides a challenging example for such a signaling system.
Notably, the effects of bradykinin are regulated on at least three levels: first, bradykinin generation is controlled by the assembly, activation, and dissociation of binary complexes of HK with PK (7, 33); second, bradykinin signaling is regulated through specific B2Rs that rapidly desensitize and internalize upon bradykinin stimulation (40, 52, 53); and third, bradykinin degradation is efficiently executed by peptidases such as angiotensin-converting enzyme (54, 55, 56). Loss of control over any of these levels may have severe pathophysiological consequences. For example, deficiency in C1 inhibitor, the major PKa inactivator, is associated with hereditary angioedema, a life-threatening condition with recurrent attacks of edema (16, 57, 58). The central role of bradykinin in the pathology of this disease has recently been proven by an elegant mouse model, in which targeted deletion of the C1 inhibitor gene resulted in spontaneous vascular leakage. This phenotype was completely rescued by cross-breeding C1 inhibitor-deficient mice with mice lacking the B2R gene (13). It is not surprising that in vivo bradykinin release is carefully controlled by multiple mechanisms preventing unwanted liberation of this powerful effector. This report demonstrates a novel additional mechanism for control of local bradykinin generation. The liberation of the hormone precursor HK from GAG is crucial for bradykinin generation and therefore GAG on endothelial surfaces importantly contributes to the regulation of the kallikrein-kinin system.
The present study focuses on the molecular mechanisms involved in the regulation of circumscribed bradykinin release on endothelial surfaces through proteases such as PK that are activated by FXIIa, or elastase and tryptase secreted from activated neutrophils and mast cells. A key finding in our study is that attachment of HK to GAG provides protection from proteolytic cleavage (Figs. 3 and 4). Indeed, previous studies have shown that binding of HK to platelet cell surfaces provides (partial) protection from proteolytic cleavage and kinin release (59), although the underlying mechanisms have remained elusive. Our studies offer a rationale for the observed phenomenon. They also provide an intuitive solution to the fact that HK has two distinct cell binding sites in D3 (heavy chain portion of HK) and D5H (light chain) that differ by their affinity (21, 22). Competitors such as D3 may partially displace HK from cell surfaces, thereby exposing the kinin-bearing domain D4 for proteolytic attack, whereas the tight binding domain D5H may still anchor HK via its light chain to the cell surface or vice versa. This mechanism would ensure that short-lived kinins are released in proximity to their site of action. Common to the established scenarios of bradykinin generation is that minor protease activity is sufficient to trigger an amplifier cascade resulting in local bradykinin burst, with severe pathophysiological implications (60). Furthermore, our studies provide a model to explain local bursts of bradykinin activity showing that HK fragments have the capacity to overcome GAG-mediated protection, thereby making neighboring HK molecules susceptible to proteolytic cleavage. Thus, efficient bradykinin generation depends on two triggering factors: PK activation and HK detachment. We do not claim that proteolytic HK fragments are the sole candidates for HK detachment; however, they maintain hormone generation following an initial signal and thus boost bradykinin generation in vivo. As yet unidentified factors released, e.g., from activated neutrophils or from mast cells, may serve as competitors that displace HK from their tight GAG-binding. In cooperation with leukocyte-derived proteases, these factors may account for the initial trigger of bradykinin generation. Work is underway to scrutinize secretory components of leukocytes for their efficiency to detach HK from GAG and to promote bradykinin release.
It is well established that HK binds heparin via D5H, i.e., the same domain binds to HS and cell surfaces (23, 24, 25, 26, 61, 62). Notably, almost all known heparin-binding proteins readily bind to the HS-type of GAG (63); furthermore, and in support of our hypothesis, we note that histidine-rich glycoprotein, the closest structural relative and evolutionary predecessor of kininogens, binds to heparin and the HS chains of cell-associated proteoglycan (64, 65). Apart from GAG, other proteins have been identified such as urokinase receptor, cytokeratin-1, and gC1qR/p33 (66) that bind to HK independently of HS (67). To investigate whether these proteins contribute to bradykinin generation in vivo is an important task for the future.
In this context, it is noteworthy that bacteria and other parasites may provide contact phase-like platforms on their surface through which they bind HK and other contact factors. Indeed, Escherichia coli and Salmonella bind and activate contact phase factors, thereby promoting tissue invasion (60). Furthermore, pathogenic bacteria are known to express heparinase, which may not only serve to facilitate tissue destruction but may also promote cleavage of HK and bradykinin release through the digestion of endothelial GAG (68, 69). It is conceivable that bacteria “hijack” the host contact system to open the endothelial barriers. It is also known that many cancer cells express and secrete heparinase (70), which could promote bradykinin generation and thereby contribute to peritumorous edema formation (71).
In summary, we have demonstrated the importance of GAG for bradykinin generation on endothelial cells and in vivo in mice. The hormone precursor, HK, needs to be separated from GAG for efficient proteolysis. GAG-degrading enzymes and HK fragments resulting from proteolytic HK cleavage efficiently detach HK from GAG thereby promoting bradykinin generation. We therefore propose that GAGs are important gatekeepers in the regulation of vascular peptide hormone processing. These concepts provide novel insights into the regulation of local bradykinin generation and may enable the development of new strategies to modulate the biological and pathophysiological effects of the kallikrein-kinin system in health and disease.
We thank Dr. S. Rose-John (University of Kiel) for helpful discussions. The technical support of M. Kuhn and C. Dienesch (Julius-Maximilians-University Wuerzburg) is greatly appreciated. We also thank Dr. U. Walter (Julius-Maximilians-University Wuerzburg) for constant support throughout the project.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by grants from the Deutsche Forschungsgemeinschaft through the SFB 355 (to T.R.) and the SFB 628 and the Fonds der Chemischen Industrie (to W.M.-E.).
Abbreviations used in this paper: FXII, factor XII; PK, plasma kallikrein; HK, high molecular weight kininogen; FXIIa, active FXII; PKa, active PK; B2R, G protein-coupled receptor of the B2 type; GAG, glycosaminoglycan; D3, domain 3 of the kininogen; D5H, domain 5 of the HK heavy chain; HS, heparan sulfate; CS, chondroitin sulfate; MBK, mAb to bradykinin.