Kininostatin Associates With Membrane Rafts and Inhibits vβ3 Integrin Activation in Human Umbilical Vein Endothelial Cells

【摘要】  Objective— The cleaved form of high molecular weight kininogen (HKa) is a potent inhibitor of angiogenesis and tumor growth in vivo; the functional domain has been identified as domain 5 (D5, named as kininostatin). We now identify the subcellular targeting site for D5 on endothelial cells (ECs), and investigate D5 inhibition of integrin functions.

Methods and Results— Endothelial membrane rafts were isolated using sucrose density gradient centrifugation. D5, bound to ECs, was predominantly associated with membrane rafts, in which uPAR, a HKa receptor, was also localized. In contrast, other HKa receptors, cytokeratin-1 and gC1q receptor, were not detected in membrane rafts. Colocalization of D5 with caveolin-1 was demonstrated on ECs by confocal microscopy. Disruption of membrane rafts by cholesterol removal decreased D5 binding to ECs. On stimulation with vascular endothelial growth factor, vβ3 integrin formed a complex with uPAR and caveolin-1, which was accompanied by an increase in ligand binding affinity of vβ3 integrin. These events were inhibited by D5. Consistently, D5 suppressed specific vβ3 integrin-mediated EC adhesion and spreading as well as small guanosine triphosphatase Rac1 activation.

Conclusions— D5 binds to ECs via membrane rafts and downregulates vβ3 integrin bidirectional signaling and the downstream Rac1 activation pathway.

The cleaved form of high molecular weight kininogen and its functional domain (D5) are potent inhibitors of angiogenesis. This study indicates that D5 binding to endothelial cells depends on membrane rafts, by which it inhibits vβ3 integrin bi-directional signaling and downstream Rac1 activation.

【关键词】  kininogen membrane rafts integrin uPAR angiogenesis

Introduction

Caveolae are specialized components of membrane rafts, 1 which are plasmalemmal invaginations formed by the sequestration of cholesterol and glycosphingolipids with self-associating molecules named caveolins. The major structural protein of endothelial caveolae is caveolin-1. Because the definitive identification of caveolae in experimental protocols can be only made by electron microscopy, we use the term membrane rafts in most of this paper and indicate when they contain caveolin. Caveolae are abundant on the surface of endothelial cells (ECs). 2 Caveolae and caveolin-1 are widely involved in the regulation of EC function, including EC proliferation, migration, and angiogenesis. Angiogenesis is the process of formation of new blood vessels from the existing microvessels, initiated by stimulation of vascular ECs with growth factors and cytokines. The functional significance of caveolae in angiogenesis has been suggested by defective nitric oxide–dependent angiogenesis in caveolin-1–null mice, 3 and by reduced tube formation in the absence of caveolin-1 in vivo and in vitro. 4 Integrins, particularly the v integrin family, are major players in each step of angiogenesis. 5 Interaction of integrins with extracellular matrix (ECM) is required for EC function and maturation of newly formed blood vessels. v integrins, highly expressed on activated ECs during angiogenesis, colocalize with caveolin-1 in membrane rafts. 6 Downstream effectors of integrins such as Src family kinases, and small guanosine triphosphatases are localized in membrane rafts and are involved in regulation of integrin activation as well as integrin-mediated cytoskeletal rearrangement. 7

Human plasma high molecular weight kininogen (HK) is a major component (660 nmol/L) of the plasma kallikrein-kinin contact system, 8 and serves as precursor for both bradykinin (BK) and the cleaved form of HK (HKa). The EC plasma membrane is an important site for the generation of BK and HKa, both of which in turn affect EC function. 9 HKa exhibits a potent antiangiogenic activity, which is mediated through its Domain 5 (D5), known as kininostatin. 10 HKa and D5 inhibition of angiogenesis is associated with their antiadhesive activity, which inhibits EC proliferation and induces EC apoptosis. 11 D5 binds to the somatomedin B domain of vitronectin (Vn, aa 1 to 43), thereby masking the RGD region (aa 45 to 47) and preventing integrin ligation. 12 D5 also directly binds to domains 2/3 of uPAR, which serves as a binding site for Vn. 13 HKa thus competes for Vn interaction with both vβ3 integrin and uPAR. However, considering the potent inhibition of HKa and D5 on in vivo angiogenesis, the selective antiadhesive activity, which is tightly regulated by Vn, is not solely responsible for HKa antiangiogenic effect.

Our current study demonstrates that D5 binding to ECs depends on membrane rafts, by which it exerts inhibitory effect on vβ3 integrin functions, which represents a novel mechanism of D5 antiangiogenic activity.

Materials and Methods

HUVECs (Clonetics, Wakersville, MD) were cultured in endothelial growth culture medium (EGM, Clonetics) with additives (Bullet Kit, Clonetics) provided by the manufacturer and 10% fetal bovine serum (FBS). HUVECs from 3 to 6 passages were used in individual experiments. To establish a stable cell line expressing vβ3 integrin, CS-1 cells, provided by Dr. D.A. Cheresh (The Scripps Research Institute, La Jolla, CA), were transfected with pcDNA3 expressing human integrin β3 subunit (a gift from Dr. S. Santoso, Justus-Liebig University, Germany) using lipofectamine 2000. The sequence of human uPAR cDNA (5'-GGTGAAGAAGGGCGTCCAA-3') was selected as the targeting region, and a nonsilencing siRNA, 5'-AACCTGCGGGAAGAAGTGG-3', was used as a control. Control siRNA or uPAR siRNA (Qiagen) was transfected into HUVECs with HiPerfect (Qiagen) following the manufacturers? instructions. Immunoprecipitation and Western blotting were performed as described previously. 11 GST-free D5 (aa 420-513) was prepared 10 and binding to endothelial cells performed as described. 13 Caveolin-1-enriched membrane rafts were isolated and purified as described previously. 16 Evaluation of vβ3 integrin binding affinity (WOW-1 binding) using HUVECs cultured on gelatin-coated 96-well plates after incubation with or without 600 nmol/L D5 for 30 minutes followed by stimulated with 20 ng/mL of VEGF for 5 minutes. WOW-1 Fab (30 µg/mL) was then added, and followed by addition of 10 µg/mL of Alexa Fluor 488-F(ab?)2 fragment of anti-mouse IgG (Molecular Probes, Eugene, Oregon). After 30 minutes, the cells were washed and the fluorescence intensity was immediately measured by 1420 Multilabel Counter (Wallac-Victor 2). The data were calculated as average±SEM from experiments done at least 3 times, and statistically analyzed by Students? t test (two groups only) or One Way Analysis of variance (ANOVA) and Student-Newman-Keuls test (multiple groups). The materials and methods used in this study are fully described in the supplemental material (available online at http://atvb.ahajournals.org).

Results

D5 Binds to Caveolin-1–Enriched Membrane Rafts via Association With uPAR

In this study, we focus on the function of D5, which is the functional region of HKa in antiangiogenesis and contains an EC binding site. Because uPAR, which is a HKa receptor, is localized in membrane rafts in tumor cells, and colocalized with caveolin-1, 14 we tested whether the targeting site of D5 on ECs is associated with membrane rafts. To monitor D5 binding, we labeled recombinant human D5 with biotin. Biotin-D5, bound to human umbilical vein endothelial cells (HUVECs), was cross-linked to the receptor by 3, 3'-dithiobis (DTSSP), a cleavable disulfide cross-linking reagent. The concentration of D5 chosen (200 nmol/L), 30% of the plasma HK concentration (660 nmol/L), has been observed in human sepsis syndrome at the molecular weight of HKa (100 kDa) compared with HK (120 kDa). 15 Membrane raft fractions were separated by sucrose density gradient centrifugation. Figure 1 A shows that fractions 3 and 4 represented the caveolin-1–enriched membrane rafts, as evidenced by the presence of caveolin-1. The cytoskeletal protein, β-tubulin, was not detected in this fraction and served as a control for nonraft proteins ( Figure 1 A). The majority of Biotin-D5 bound to HUVECs was recovered in the membrane rafts, which also contained uPAR ( Figure 1 A). Moreover, D5 was detected in uPAR immunoprecipitates from membrane rafts ( Figure 1 B), suggesting D5 association with uPAR in the membrane rafts. In contrast, neither gC1q receptor (gC1qR) nor cytokeratin 1 (CK1), 2 other receptors for HKa, were identified in the membrane rafts ( Figure 1 A). Compared with the control, the association of D5 with uPAR in membrane rafts was enhanced by DTSSP treatment (supplemental Figure IA). This treatment seemed to prevent dissociation of D5 from the receptor during subsequent ultracentrifugation and solubilization, whereas it did not affect total D5 binding to ECs, as well as D5 association with membrane raft fraction and nonmembrane raft fraction (supplemental Figure IB and IC). We cannot exclude additional binding sites for D5 in the membrane rafts to explain the larger amount of D5 in fraction 3 than in fraction 4 ( Figure 1 A).

Figure 1. D5 binds to caveolin-1–enriched membrane rafts via association with uPAR. A, HUVECs were incubated with 200 nmol/L Biotin-D5. Membrane rafts were isolated and analyzed as described in the Materials and Methods. B, Raft fractions (fraction 3 and 4) were immunoprecipitated with anti-uPAR mAb. C, Raft fractions were pooled (starting material, SM) and incubated with magnetic beads preabsorbed with caveolin-1 mAb. The bound materials on the beads (caveolin-1-enriched membrane rafts, Cav) were separated magnetically from nonbound material (NBD). Western blots (WB) were performed with antibodies as indicated and Avidin-HRP to detect Biotin-D5.

D5 Association With Caveolin-1

The ultracentrifugation method yields a heterogenous population of membrane rafts. Membrane rafts (fractions 3 and 4 shown in Figure 1 A) were pooled (designated as starting materials, SM), and incubated with Dynal magnetic beads preabsorbed with anti–caveolin-1 mAb. This method allowed successful separation of caveolae from other membrane rafts ( Figure 1 C). The majority of D5 bound to the membrane rafts was retained in caveolar rafts ( Figure 1 C).

To confirm the physical association of D5 with caveolin-1–enriched membrane rafts, HUVECs were incubated with 200 nmol/L GST or GST-D5. In the cell lysates, caveolin-1 was precipitated with GST-D5, but not GST ( Figure 2 A). The colocalization of D5 with caveolin-1 was demonstrated in a single ECs by confocal microscopy. Figure 2 B shows that when HUVECs were labeled with fluorescein isothiocyanate (FITC)-D5 and stained with a fluorescent antibody to caveolin-1, an intense immunofluorescence signal was obtained in certain stretches of the plasma membrane, overlay of the immunofluorescent images for each label showed a high degree of colocalization. The binding of D5 to HUVECs and its colocalization with caveolin-1 were inhibited in the absence of Zn 2+ ( Figure 2 C, i and data not shown), consistent with previous observations that D5 binding to HUVECs is specifically Zn 2+ -dependent. 8 The integrity of membrane rafts depends on plasma membrane cholesterol. Cholesterol depletion by methyl-β-cyclodextrin (MβCD) results in their disappearance. D5 binding to ECs was inhibited by MβCD in a concentration-dependent manner ( Figure 2 C, ii), although MβCD did not affect cell viability, as determined by trypan blue dye exclusion (dead cells/live cells ratio, data not shown). MβCD altered membrane localization of uPAR and caveolin-1 (supplemental Figure II; see also Yang and Rizzo 16 ), suggesting that the depletion of cholesterol reduces D5 binding to ECs because of the disturbed uPAR and caveolin-1 localization on membrane. Because uPAR is a GPI-anchored protein, MβCD depletion of membrane cholesterol prevents uPAR localization on the membrane surface.

Figure 2. D5 colocalization with caveolin-1. A, HUVECs were incubated with 200 nmol/L GST or GST-D5. Cell lysate was incubated with gluthathione-Sepharose 4B beads. Lane1, cell lysate; lane 2, GST precipitates; lane 3, GST-D5 precipitates. B, HUVECs were labeled with FITC-D5 and caveolin-1 mAb plus Alexa Fluor 594-(Fab')2 of goat anti-mouse IgG as described in the Materials and Methods. Subcellular localization of FITC-D5, caveolin-1, as well as an overlay of the images are shown. C, Binding of FITC-D5 to HUVECs was measured in the presence or absence of Zn 2+ (50 µmol/L, i). In ii, binding of FITC-D5 to HUVECs was measured after treatment with 0, 0.3, 1.0 mmol/L MβCD, the fluorescence intensity at 0 mmol/L MβCD was set at 100% (* P <0.01 compared with 0 mmol/L MβCD, n=4).

D5 Dissociates uPAR- vβ3 Integrin Complex in the Membrane Rafts and Downregulates vβ3 Integrin Ligand Binding Affinity

The association of D5 with membrane rafts stimulated us to hypothesize that D5 exerts its inhibitory effect via raft localization. In HUVECs, vβ3 integrin formed a complex with uPAR and caveolin-1, which was enhanced by VEGF stimulation ( Figure 3 A). However, VEGF did not change the expression of each molecule during the period of stimulation (data not shown). MβCD treatment markedly dissociated vβ3 integrin from caveolin-1, but not uPAR ( Figure 3 B), suggesting that vβ3 integrin association with uPAR is direct, and its binding to caveolin-1 is dependent on intact structure of the rafts/caveolae. D5, like HKa, 11 inhibited the formation of vβ3 integrin–uPAR–caveolin-1 complex ( Figure 3 C), implying that D5 is the functional moiety. The concentration (600 nmol/L) used is lower than the dose necessary for the inhibition of EC migration. 17 As indicated in Figure 3 D, vβ3 integrin formed a complex with uPAR and caveolin-1 only in the membrane rafts, but not in the nonraft fractions. D5 dissociated uPAR and caveolin-1 from vβ3 integrin in the membrane rafts ( Figure 3 D), suggesting D5 disrupts this complex formation via binding to membrane rafts.

Figure 3. D5 dissociates uPAR– vβ3–caveolin-1 complex in membrane rafts and decreases vβ3 ligand binding affinity. A, After treatment with (+) or without (–) 50 ng/mL VEGF for 2 hours, HUVECs were lysed, and cell lysates were immunoprecipitated with LM609. The immunoprecipitates were analyzed by WB. B, Before VEGF stimulation, HUVECs were pretreated with (+) or without (–) 10 mmol/L MβCD. C, Before VEGF stimulation, HUVECs were pretreated with (+) or without (–) 600 nmol/L D5. D, Alternatively, pooled membrane rafts or nonrafts fractions were immunoprecipitated from equal amount of proteins using LM609. E, HUVECs cultured in 96-well plate were incubated with EBM containing 0.5% FBS for 4 hours. After pretreatment with or without 600 nmol/L D5, HUVECs were stimulated with 20 ng/mL VEGF for 5 minutes. Binding of WOW-1 Fab (open column), LM609 (striped column), and P1F6 (closed column) were measured as described in the Materials and Methods (n=3).

uPAR and caveolin-1 are required for integrin activation. 18 We further found that D5 disruption of this complex was accompanied by its downregulation of vβ3 integrin binding affinity. WOW-1 is a Fab fragment which specifically recognizes activated vβ3 and vβ5 integrins. 19 VEGF stimulated WOW-1 binding to ECs ( Figure 3 E), supporting that VEGF increases active integrin recycling. D5 significantly inhibited the increase in WOW-1 binding to ECs stimulated by VEGF, but did not affect the overall expression of vβ3 or vβ5 integrin on membrane surface ( Figure 3 E). To confirm that the complex formation of vβ3 integrin–uPAR–caveolin-1 is required for v integrin function, uPAR was depleted by silencing RNA (siRNA) approach, in which case, the 3-molecule complex no longer existed. Nonsilencing siRNA served as control for nonspecific effects of double-strand RNA. Transfection of HUVECs with uPAR-siRNA reduced uPAR expression by 91.3% versus 8.74±4.58% (control, n=3) and did not affect FAK and β3 integrin expression ( Figure 4 A). In the absence of uPAR, VEGF-mediated HUVEC migration on Vn was dramatically suppressed ( Figure 4 B, compare b and d). The reduction of migration on Vn was about 70%, whereas there was less than 30% decrease in migration on collagen ( Figure 4 C), suggesting that uPAR is necessary for v integrin-mediated EC migration, and the physical association of uPAR with vβ3 integrin is functional. The decreased EC migration on collagen in the absence of uPAR supports that uPAR is also required for collagen-mediated migration, 20 although Vn-mediated EC migration seems more dependent on uPAR.

Figure 4. uPAR is required for Vn-mediated HUVEC migration. A, Expression of uPAR in HUVECs transfected with control siRNA (con) or uPAR siRNA (uPAR). B, Vn-mediated migration of HUVECs which were transfected with control siRNA or uPAR siRNA. C, Migration of HUVECs transfected with con-siRNA or uPAR-siRNA on Vn or collagen-coated 24-transwell on VEGF stimulation. The number of migrated HUVECs transfected with con-siRNA was considered 100%. (* P <0.05, *** P <0.001).

D5 Inhibits vβ3 Integrin Ligation-Dependent Cytoskeletal Reorganization and Rac1 Activation

D5 downregulation of v integrin ligand binding affinity raises a possibility that D5 inhibits vβ3 integrin outside-in signaling. We measured EC spreading on fibrinogen (Fbg)-coated surface, as Fbg does not bind to either D5 or uPAR, but is a ligand for vβ3 integrin. D5 blocked Fbg-mediated cell spreading, particularly lamellipodial formation ( Figure 5 A). In contrast, D5 did not inhibit EC spreading on collagen surface (data not shown). The inhibitory pattern was similar to that with cRGDfK, an inhibitory peptide specific for vβ3 integrin ( Figure 5 A), and both D5 and cRGDfK lowered EC spreading surface area by 60% to 70% ( Figure 5 B), suggesting that D5 suppresses actin cytoskeletal reorganization of ECs, possibly by inhibiting vβ3 integrin activation. Consistently, D5 abrogated Fbg-mediated Rac1 activation ( Figure 5 C). Supplemental Figure III further indicates that HUVECs rapidly underwent spreading on Fbg surface, which was accompanied by Rac1 activation. D5 significantly inhibited these 2 activation processes, especially at early time points (0.5, 1, and 2 hours). The inhibitory effect decreased at 4 hours. The inhibitory effect of D5 on spreading is more striking than on EC adhesion, and the inhibitory pattern is different. Because human ECs express 2 receptors for Fbg, vβ3 and 5β1 integrins, 21 the delay induced by D5 in Fbg-mediated cell spreading and Rac1 activation seems attributable to its inhibition of vβ3 but not 5β1 integrin. The interaction of 5β1 integrin with Fbg may reverse D5 inhibition of vβ3 integrin-mediated outside-in signaling.

Figure 5. D5 inhibits Fbg-mediated HUVEC spreading and Rac1 activation. A, After pretreatment with PBS (control), 600 nmol/L D5 or 10 µmol/L cRGDfK for 30 minutes, HUVECs were incubated on Fbg-coated culture plates. The actin cytoskeleton was visualized by staining with Rhodamine-Phalloidin. Bar, 50 µm. B, The cell surface area of spreading cells was determined. C, After pretreatment with (+) or without (–) 600 nmol/L D5, HUVECs were incubated on 50 µg/mL poly- L -lysine (PLL) or 2 µg/mL Fbg. The cell lysates were precipitated with Glutathione Sepharose-4B beads coated with GST-PAK CRIB.

D5 Inhibits Specific vβ3 Integrin Adhesive Function

To confirm that D5 indeed regulates specific vβ3 integrin function, we generated a cell line expressing vβ3 integrin using CS-1 cells (murine melanoma cell line). CS-1 cells maintain an intracellular pool of v subunit, they do not express β3 or β5 integrin, and thus cannot adhere on culture dishes ( Figure 6 A, i). Once CS-1 cells were transfected with human β3 subunit cDNA, vβ3 heterodimer was expressed on the cell membrane ( Figure 6 A, ii), which allowed the CS-1 cells to acquire adhesion capacity ( Figure 6 A, i). LM609, a mAb directed to vβ3 integrin, inhibited adhesion of vβ3CS-1 cells to culture dishes as well as Fbg and Vn (data not shown). Thus, the acquired adhesion property indicates specific vβ3 integrin function. In a concentration-dependent manner, D5 inhibited adhesive property of vβ3CS-1 cells on culture plate ( Figure 6 B). The inhibition was also effective on Fbg or Fn coated surface, both of them are ligands for vβ3 integrin ( Figure 6 C). The detachment is not the result of anoikis because tumor cells (CS-1) are resistant to that form of apoptosis.

Figure 6. D5 blocks specific vβ3 integrin adhesive function. A, CS-1 or vβ3CS-1 cells cultured on culture dishes were visualized by microscopy (i). In ii, flow cytometric measurement of vβ3 integrin complex on control CS-1 cells (open) and vβ3CS-1 (solid). B, vβ3CS-1 cells were incubated with PBS (control), 200, or 600 nmol/L D5. After removal of nonadherent cells by washing, the remaining adherent cells were fixed and photographed by microscopy. C, Adhesion of vβ3CS-1 cells pretreated with (white column) or without (gray column) D5 at 600 nmol/L was evaluated on Fbg- or Fn-coated plates. OD, optical density.

Discussion

In this study we demonstrate that the subcellular binding site of D5 on ECs is localized in membrane rafts. D5 is associated with caveolin-1–enriched membrane rafts. In addition, colocalization of D5 with caveolin-1 was observed in a live ECs using confocal microscopy. Disruption of membrane rafts decreases D5 binding to HUVECs. Caveolin-1 is coprecipitated with D5 in HUVECs. Although there are several surface proteins on ECs known to bind D5 such as gC1qR, cytokeratin-1, and uPAR, the relative contribution of each binding moiety has been uncertain. We found that gC1qR is not localized in membrane rafts, which is consistent with a previous finding that gC1qR is prominently localized in vesicular fractions associated with mitochondria, microsomal fractions, and cytoplasm in ECs. 22 The failure to detect CK1 in membrane rafts might be attributable to any of the following reasons: CK1 is part of a large family of cytoskeletal proteins, most of which are not released from cytoskeletal constraints during ultracentrifugation. 23 Only a small fraction of total CK1 forms a complex with uPAR. This complex is too low to be detected by the membrane raft isolation procedures. uPAR is more highly expressed on HUVEC membrane than CK1 (about 250 000 molecules per cell for uPAR versus 72 000 molecules per cell for CK1). 24 However, further studies are required to determine how uPAR and other receptors are assembled in the D5-receptor complex in membrane rafts.

Caveolae and their principal protein caveolin-1 are closely involved in the various processes of EC activation. Caveolin-1 is not only a structural protein, but also serves as a scaffold to assemble many signaling molecules and generate preassembled signaling complexes. In addition to concentrating these signal transducers within a distinct region of the plasma membrane, caveolin-1 binding may functionally regulate the activation state of caveolae-associated signaling molecules. 2 The identification of membrane rafts/caveolae as a major membrane subdomain for D5 binding is intriguing. These membrane microdomains are responsible for subcellular localization and regulation of integrins in angiogenesis. 18 Integrins interact closely with caveolae/caveolin-1 in the context of EC proliferation, migration, and tube formation. 25 We now show that D5, whose binding to ECs depends on membrane rafts, is able to inhibit vβ3 integrin activation in both outside-in and inside-out signaling. This inhibitory effect is independent of D5 antiadhesive activity, which is regulated by D5 interaction with ECM components. D5 suppresses Fbg-induced Rac1 activation and EC spreading, similarly to cRGDfK, an inhibitory peptide of vβ3 integrin. D5 inhibition of specific vβ3 integrin function is confirmed by the finding in vβ3CS-1 transfection model ( Figure 6 ). As demonstrated by WOW-1 binding assay, D5 inhibition of vβ3 is attributable to downregulation of its ligand binding affinity ( Figure 3 E). WOW-1 Fab is monomeric, and thus insensitive to changes in integrin clustering and indicates only changes in affinity. WOW-1 binds only to unoccupied, high-affinity integrins. Like other integrins, vβ3 can exist in different functional states with respect to ligand binding, which is modulated by growth factors such as VEGF, through inside-out signaling. 26 Although D5 did not change the total expression of vβ3 and vβ5 integrins on the membrane, D5 inhibition of WOW-1 binding to ECs presumably results from blocking activated integrin recycling. 27

In the context of the crucial role of integrins in angiogenesis, understanding the connection between D5 targeting site and its inhibition of integrin function becomes important. The interaction between integrins and membrane rafts is necessary for integrin-mediated EC adhesion, spreading, and migration to the ECM on growth factor stimulation. The association of v integrins with caveolin-1 in membrane rafts is disrupted by MβCD, thereby suppressing vβ3 integrin signaling. 28 Given that vβ3 integrin association with caveolin-1 is sensitive to cholesterol depletion, 27 and that caveolin-1 does not extend outside the cell, this association must be indirect, and mediated by receptors in membrane rafts. VEGF enhances vβ3 integrin association with uPAR and caveolin-1. Recruitment of vβ3 integrin into membrane rafts allows its proximal localization to downstream effectors including Src and Rac1 GTPases, as well as its partners such as VEGF receptor. The dynamic redistribution of vβ3 integrin is required for its functional crosstalk with growth factor receptors and EC function. We found that D5 blockade of VEGF-stimulated recruitment of vβ3 integrin to membrane rafts is accompanied by a decrease in vβ3 ligand binding affinity, which may account for D5 inhibition downstream Rac1 activation and EC spreading. D5 disrupts the formation of vβ3 integrin–caveolin-1–uPAR complex in membrane rafts, suggesting that D5 exerts its inhibitory effect on vβ3 integrin through membrane rafts.

Functional interaction between uPAR and vβ3 integrins has become an important issue in tumor-induced angiogenesis. Both proteins are localized in membrane rafts of several cell types and form a complex. 28–31 uPAR has no cytoplasmic domain and is bound to the plasma membrane by a glycosylphosphatidylinositol anchor. Therefore, its signaling is mediated by integrins, in particular vβ3. 32 On the other hand, uPAR behaves as an important modulator of integrin function. This concept is supported by our finding that uPAR is required for v integrin-mediated EC migration ( Figure 4B and 4 C). Because Vn-mediated migratory signal does not depend on uPAR interaction with Vn, 29 the decrease of EC migration represents inhibition of v integrin. VEGF activates vβ3 integrin forming a complex with uPAR and caveolin-1 in ECs. Because the association of uPAR with vβ3 integrin is resistant to removal of cholesterol, the interaction between uPAR and vβ3 integrin seems direct, in agreement with the finding of Degryse et al that direct binding of uPAR to vβ3 integrin occurs a region of uPAR domain 2. 29 Because D5 binds to the D2 domain of uPAR, and an antibody against uPAR restores HKa inhibition of EC migration on Vn, 17 uPAR is a primary binding site for HKa inhibition of vβ3 integrin function. D5 occupancy of uPAR may not only block internalization of both receptors and the turnover of focal adhesion plaques during EC migration, but also prevent the assembly of reexpressed uPAR and integrin on the membrane surface during receptor cycling. Thus, D5 blockage of vβ3 integrin-uPAR interaction inhibits the signaling of each receptor as well as their functional crosstalk.

During the dynamic process of angiogenesis, vβ3 integrin acts as a biosensor and modulates cell survival and death as a function of ECM composition. 5,33 Ligation of the upregulated vβ3 integrin during neovascularization represents a potential mechanism to promote angiogenesis. Our current studies support a model in which D5 downregulates vβ3 integrin function by targeting membrane rafts, which represents a new mechanism underlying D5 inhibition of EC function.

Acknowledgments

Sources of Funding

This work was supported by the following grants: R01CA083121, R01AR051713, and T32HL07777 from National Institute of Health (to R.W.C.).

Disclosures

None.

【参考文献】
  Pike LJ. Rafts defined: a report on the Keystone Symposium on Lipid Rafts and Cell Function. J Lipid Res. 2006; 47: 1597–1598.

Frank PG, Woodman SE, Park DS, Lisanti MP. Caveolin, caveolae, and endothelial cell function. Arterioscler Thromb Vasc Biol. 2003; 23: 1161–1168.

Sonveaux P, Martinive P, DeWever J, Batova Z, Daneau G. Caveolin-1 expression is critical for vascular endothelial growth factor-induced ischemic hindlimb collateralization and nitric oxide-mediated angiogenesis. Circ Res. 2004; 95: 154–161.

Griffoni CSE, Santi S, Riccio M, Guarnieri T, Tomasi V. Knockdown of caveolin-1 by antisense oligonucleotides impairs angiogenesis in vitro and in vivo. Biochem Biophys Res Commun. 2000; 276: 756–761.

Stupack DG, Cheresh DA. Get a ligand, get a life: integrins, signaling and cell survival. J Cell Sci. 2002; 115: 3729–3738.

Alain P RF, Monique L, Arbeillec B, and Menashia S. MMP-2 colocalizes with caveolae on the surface of endothelial cells. Exp Cell Res. 2001; 262: 28–36.

Ikeda S, Ushio-Fukai M, Zuo L, Tojo T, Dikalov S, Patrushev NA, Alexander RW. Novel role of ARF6 in vascular endothelial growth factor-induced signaling and angiogenesis. Circ Res. 2005; 96: 467–475.

Colman RW, Schmaier AH. Contact System: A vascular biology modulator with anticoagulant, profibrinolytic, antiadhesive, and proinflammatory attributes. Blood. 1997; 90: 3819–3843.

Zhao Y, Qiu Q, Mahdi F, Shariat-Madar Z, Rojkjaer R, Schmaier AH. Assembly and activation of HK-PK complex on endothelial cells results in bradykinin liberation and NO formation. Am J Physiol Heart Circ Physiol. 2001; 280: H1821–H1829.

Colman RW, Jameson BA, Lin Y, Johnson D, Mousa SA. Domain 5 of high molecular weight kininogen (kininostatin) down-regulates endothelial cell proliferation and migration and inhibits angiogenesis. Blood. 2000; 95: 543–550.

Cao DJ, Guo Y-L, Colman RW. Urokinase-type plasminogen activator receptor is involved in mediating the apoptotic effect of cleaved high molecular weight kininogen in human endothelial cells. Circ Res. 2004; 94: 1227–1234.

Asakura SYW, Sottile J, Zhang Q, Jin Y, Ohkubo I, Sasaki M, Matsuda M, Hirata H, Mosher DF. Opposing effects of low and high molecular weight kininogens on cell adhesion. J Biochem (Tokyo). 1998; 124: 473–484.

Colman RW PR, Najamunnisa S, Yan W, Wang J, Mazar A, and McCrae KR. Binding of high molecular weight kininogen to human endothelial cells is mediated via a site within domains 2 and 3 of the urokinase receptor. J Clin Invest. 1997; 100: 1481–1487.

Stahl A, Mueller B. The urokinase-type plasminogen activator receptor, a GPI-linked protein, is localized in caveolae. J Cell Biol. 1995; 129: 335–344.

Schmaier AH, Farber A, Schein R, Sprung C. Structural changes of plasma high molecular weight kininogen after in vitro activation and in sepsis. J Lab Clin Med. 1988; 112: 182–192.

Yang B, Rizzo V. TNF-alpha potentiates protein-tyrosine nitration through activation of NADPH oxidase and eNOS localized in membrane rafts and caveolae of bovine aortic endothelial cells. Am J Physiol Heart Circ Physiol. 2007; 292: H954–H962.

Katkade VSA, Isordia-Salas I, Bradford HN, Gaughan JP, Colman RW, Panetti TS. Domain 5 of cleaved high molecular weight kininogen inhibits endothelial cell migration through Akt. Thromb Haemost. 2005; 94: 606–614.

Chapman HA, Wei Y, Simon DI, Waltz DA. Role of urokinase receptor and caveolin in regulation of integrin signaling. Thromb Haemost. 1999; 82: 291–297.

Pampori N, Hato T, Stupack DG, Aidoudi S, Cheresh DA, Nemerow GR, Shattil SJ. Mechanisms and consequences of affinity modulation of integrin alpha Vbeta 3 detected with a novel patch-engineered monovalent ligand. J Biol Chem. 1999; 274: 21609–21616.

Schwab W, Gavlik JM, Beichler T, Funk RH, Albrecht S, Magdolen V, Luther T, Kasper M, Shakibaei M. Expression of the urokinase-type plasminogen activator receptor in human articular chondrocytes: association with caveolin and beta 1-integrin. Histochem Cell Biol. 2001; 115: 317–323.

Suehiro K, Gailit J, Plow EF. Fibrinogen is a ligand for integrin alpha5beta1 on endothelial cells. J Biol Chem. 1997; 272: 5360–5366.

Dedio J, Jahnen-Dechent W, Bachmann M, Muller-Esterl W. The multiligand-binding protein gC1qR, putative C1q receptor, is a mitochondrial protein. J Immunol. 1998; 160: 3534–3542.

Steinert PM, Roop DR. Molecular and cellular biology of intermediate filaments. Ann Rev Biochem. 1988; 57: 593–625.

Barnathan E, Kuo A, Rosenfeld L, Kariko K, Leski M, Robbiati F, Nolli M, Henkin J, Cines D. Interaction of single-chain urokinase-type plasminogen activator with human endothelial cells. J Biol Chem. 1990; 265: 2865–2872.

Beardsley A, Fang K, Mertz H, Castranova V, Friend S, Liu J. Loss of caveolin-1 polarity impedes endothelial cell polarization and directional movement. J Biol Chem. 2005; 280: 3541–3547.

Byzova TV, Goldman CK, Pampori N, Thomas KA, Bett A, Shattil SJ, Plow EF. A mechanism for modulation of cellular responses to VEGF: activation of the integrins. Mol Cell. 2000; 6: 851–860.

Schwartz M. Integrin signaling revisited. Trends Cell Biol. 2001; 11: 466–470.

Green JM, Zhelesnyak A, Chung J, Lindberg FP, Sarfati M, Frazier WA, Brown EJ. Role of cholesterol in formation and function of a signaling complex involving {alpha}v{beta}3, integrin-associated protein (CD47), and heterotrimeric G proteins. J Cell Biol. 1999; 146: 673–682.

Degryse B, Resnati M, Czekay R-P, Loskutoff DJ, Blasi F. Domain 2 of the urokinase receptor contains an integrin-interacting epitope with intrinsic signaling activity: generation of a new integrin inhibitor. J Biol Chem. 2005; 280: 24792–24803.

Degryse BOS, Resnati M, Rabbani SA, Blasi F. Urokinase/urokinase receptor and vitronectin/alpha(v) beta(3) integrin induce chemotaxis and cytoskeleton reorganization through different signaling pathways. Oncogene. 2001; 20: 2032–2043.

Gellert GC GR, Kitson RP. Physical association of uPAR with the alphaV integrin on the surface of human NK cells. Biochem Biophys Res Commun. 2004; 315: 1025–1032.

Simon DI, Wei Y, Zhang L, Rao NK, Xu H, Chen Z, Liu Q, Rosenberg S, Chapman HA. Identification of a urokinase receptor-integrin interaction site. promiscuous regulator of integrin function. J Biol Chem. 2000; 275: 10228–10234.

Cheresh DA, Stupack DG. Integrin-mediated death: An explanation of the integrin-knockout phenotype? Nat Med. 2002; 8: 193–194.

作者单位：From The Sol Sherry Thrombosis Research Center (Y.W., Y.L., I.M.S., N.G.S., R.W.C.) and the Cardiovascular Research Center and Department of Anatomy and Cell Biology (V.R.), Temple University School of Medicine, Philadelphia, Pa.