Agkistin, a Snake Venom-derived Glycoprotein Ib Antagonist, Disrupts von Willebrand Factor-Endothelial Cell Interaction and Inhibits Angiogenesis*

Glycoprotein (GP) Ib, an adhesion receptor expressed on both platelets and endothelial cells, mediates the binding of von Willebrand factor (vWF). Platelet GPIb plays an important role in platelet adhesion and activa-tion, whereas the interaction of vWF and endothelial GPIb is not fully understood. We report here that agkistin, a snake venom protein, selectively blocks the interaction of vWF with human endothelial GPIb and inhibits angiogenesis in vivo . Agkistin specifically blocked human umbilical vein endothelial cell (HUVEC) adhesion to immobilized vWF in a concentration-dependent man-ner. Fluorescein isothiocyanate (FITC)-conjugated agkistin bound to HUVECs in a saturable manner. AP1, a monoclonal antibody (mAb) raised against GPIb, specifically inhibited the binding of FITC-conjugated agkistin to HUVECs in a dose-dependent manner, but other anti-integrin mAbs raised against a v b 3 , a 2 b 1 , and a 5 b 1 did not affect this binding reaction. However, neither agkistin (2 m g/ml) nor AP1 (40 m g/ml) apparently reduced HUVEC viability. Both agkistin and AP1 exhib-ited a profound anti-angiogenic

(ECMs) 1 including fibrinogen, vitronectin, collagen, laminin, and vWF, through cell surface adhesion receptors, are involved in the multiprocesses of neovascularization (2). For example, integrin ␣ v ␤ 3 is not readily detectable in quiescent vessels but becomes highly expressed in angiogenic vessels (3). The dependence of angiogenesis on vascular cell adhesion events in vivo is evidenced by the observation that antibody and snake venom proteins antagonizing integrin ␣ v ␤ 3 blocked angiogenesis in the chick chorioallantoic membrane (CAM) model (4,5).
GPIb is an adhesive receptor expressed both on platelets and endothelial cells. Binding of plasma vWF to platelet GPIb plays a critical role in the earliest phase of primary hemostasis, which consists of the anchoring of platelets to injury vessel walls (6). In contrast to the platelets, the functions of endothelial GPIb for vWF are not well understood. Several studies have demonstrated the presence of GPIb complex expressed on the endothelial cell surface and up-regulation of endothelial GPIb by cytokines, i.e. tumor necrosis factor and interferon ␥ (7-10). It has been shown that endothelial GPIb participates in HU-VEC binding to sickle cell erythrocytes (11) and also promotes platelet bridging to endothelial cells (12). Recently, Lian et al. (13) assumed that endothelial GPIb may contribute to mediate the process of endothelial cell migration during wound repair in vivo.
Snake venoms contain many unique components that affect cell-matrix interaction. Disintegrins represent a family of low molecular weight, cysteine-rich polypeptides that bind specifically to integrins ␣ IIb ␤ 3 , ␣ 5 ␤ 1 , and ␣ v ␤ 3 expressed on platelets and other cells including vascular endothelial cells and some tumor cells, leading to inhibition of platelet aggregation, inhibition of cell adhesion, migration, and angiogenesis (14). On the other hand, some GPIb-binding venom proteins modulate vWF-GPIb interaction, leading to platelet activation, inhibition of platelet aggregation (15), or mediating HUVEC adhesion (16). Agkistin (also named agkicetin (17)), belonging to C-type lectin polypeptide derived from the viper venom of Agkistrodon acutus, specifically inhibited human platelet aggregation and agglutination triggered by ristocetin in the presence of vWF in vitro by acting as the platelet GPIb antagonist (18). Furthermore, agkistin is a potent anti-thrombotic agent because it pronouncedly blocked platelet plug formation in an experimental model in vivo. 2 In this study, we showed that agkistin specifically inhibited HUVEC adhesion to immobilized vWF and it was shown to bind endothelial GPIb in a saturable manner. The role of endothelial GPIb in angiogenesis in vivo was thus evaluated.
Cell Culture-HUVECs were isolated from umbilical cord veins and maintained in Medium 199 containing 20% fetal bovine serum, 30 g/ml endothelial cell growth supplement, 4 mM L-glutamine, and antibiotics at 37°C with 5% CO 2 incubation. The cells were identified as positively immunofluorescent staining for human vWF antigen (DAKO) and used between the second and fourth passages.
Adhesion Assays-Synchronized HUVECs were harvested and labeled with BCECF/AM (10 g/ml) for 30 min at 37°C. After washing, cells were incubated with the indicated concentrations of agkistin or PBS for another 30 min. Control or the pretreated HUVECs were applied onto ECM-precoated plates at a density of 5 ϫ 10 4 cells/well and incubated for 90 min at 37°C allowing adhesion. After incubation, the non-adherent cells were removed by aspiration, and plates were subjected to a CytoFluor 2300 fluorescence plate reader (Minipore). The percent adherent HUVECs toward fibronectin (30 g/ml), vitronectin (10 g/ml), and vWF (10 g/ml) was calculated to be 49.2 Ϯ 2.2, 24.0 Ϯ 0.7, and 40.9 Ϯ 1.5%, respectively, based on the cell count by hemacytometer (Hausser Scientific) after trypsin deattachment. The amount of nonspecific adhesion was performed by using wells precoated with 1% BSA and estimated to be 3.26 Ϯ 0.8%.
HUVEC Viability-Cells were seeded at a density of 2 ϫ 10 4 cells/ well in 24-well plates (Costar) followed by incubation with agkistin, AP1, rhodostomin, or vehicle for 48 h in a humidified atmosphere (i.e. 37°C, 5% CO 2 ). For MTT assay, cells were incubated with MTT at a final concentration of 0.5 mg/ml for 4 h. After incubation, the medium was aspirated, and the cells were dissolved in dimethyl sulfoxide and then the developed color absorbance at 550 nm was measured.
Flow Cytometry-HUVECs (1 ϫ 10 6 ) were fixed with 1% paraformaldehyde and then incubated with FITC-conjugated agkistin or FITCconjugated BSA for 30 min. After incubation, cells were washed twice and resuspended in PBS and analyzed immediately by fluorescenceactivated cell sorter(Calibur, Becton Dickinson) using excitation and emission wavelength of 488 and 525 nm, respectively. In inhibitory studies, mAb 7E3 or AP1 was added to fixed HUVEC for 30 min prior to the incubation of FITC-conjugated proteins.
Chick CAM Angiogenesis Assay-Eggs of 10-day-old chick embryo were opened with a 1.0-cm square window that allowed direct access to underlying CAM by the method described previously (22). A filter paper disc (1.3 cm, Minipore) saturated with test compounds or an equal aliquot of PBS (final volume in 20 l) was applied to the top of CAM. The window was covered with sterile cellophane tape, and the embryos were incubated for a further 48 h at 37°C with 60% humidity to develop spontaneous angiogenesis.

Inhibition of HUVEC Adhesion to Immobilized vWF by
Agkistin-Multiple adhesive receptors expressed on endothelial cells mediate adhesion to extracellular matrices. First, we examined the effect of agkistin on HUVEC adhesion to a number of immobilized ECMs including collagen type I (80 g/ml), fibronectin (30 g/ml), vitronectin (10 g/ml), and vWF (10 g/ml). As shown in Fig. 1, agkistin specifically and dose dependently inhibited HUVEC adhesion to immobilized vWF (48.98 Ϯ 4.12% inhibition at 0.066 M) but apparently little affected those adhesive events toward other matrices used (less than 10% inhibition). Agkistin at a high concentration of 0.2 M did not exhibit a further inhibition on HUVEC adhesion to immobilized vWF (about 52% inhibition, data not shown). These results indicate that agkistin does not interfere with the adhesion of integrins ␣ v ␤ 3 , ␣ 2 ␤ 1 , and ␣ 5 ␤ 1 and their respective ligands, vitronectin, collagen, and fibronectin. The interaction of vWF with HUVECs is mediated by endothelial integrin ␣ v ␤ 3 and by GPIb complex (16). Therefore, agkistin may process a rather specific binding and high affinity toward GPIb on HU-VEC and functionally block vWF-endothelial GPIb interaction. A similar result (40 -50%) was reported with echicetin, another GPIb antagonist (23).
Inhibition of FITC-conjugated Agkistin Binding to HUVEC by AP1-To further explore the binding receptor of agkistin on HUVECs, we conjugated agkistin and BSA with FITC by a previously described method (24) and then examined the binding reaction of FITC-conjugated agkistin toward HUVEC by flow cytometry. Nonspecific binding was performed using FITC-BSA as a probe. After incubation of FITC-conjugated agkistin with HUVECs, the increment of relative fluorescence intensity of FITC-agkistin bound cells was concentration-dependent and reached a saturated binding at a concentration of 1 M (Fig. 2). The different concentration range of agkistin used in this binding assay and in the cell adhesive assay (Fig. 1) is attributable to the different number of cells used in the two independent systems (i.e. 1 ϫ 10 6 cells for agkistin binding and  (25), exerted a dose-dependent inhibition on the binding of FITC-agkistin to HU-VECs. However, all of anti-␣ v ␤ 3 mAb 7E3 and anti-␣ 2 ␤ 1 , ␣ 3 ␤ 1 , ␣ 4 ␤ 1 , and ␣ 5 ␤ 1 integrin mAbs (50 g/ml, Chemicon) did not affect this binding (Fig. 3 and data not shown). These data indicate that agkistin and AP1 might bind to a common epitope on endothelial GPIb␣ whereas agkistin did not bind to other endothelial integrins including ␣ v ␤ 3 , ␣ 2 ␤ 1 , and ␣ 5 ␤ 1 , consistent with the inference obtained from the above described adhesive experiment.
Neither Agkistin nor AP1 Affected HUVEC Viability-Primary endothelial cells are anchorage-dependent, and disruption of integrin-ECM interaction induces HUVEC death (26,27). In addition, ␣ v ␤ 3 receptor blockade resulted in unscheduled programmed cell death (apoptosis) of proliferating vascular cells (4). Recently, Feng et al. (28) indicated that GPIb complex expressed on Chinese hamster ovary cells regulates cell proliferation. Thus, it is interesting to investigate if the blockade of endothelial GPIb reduces the viability of primary endothelial cells. HUVEC viability assay was performed by determining the cell metabolic activity with MTT. As shown in Fig. 4, neither agkistin (at a high concentration of 2 g/ml, i.e. 0.066 M), which exerted a maximal effect in blocking vWF-HUVEC adhesion (Fig. 1), nor AP1 (40 g/ml) affected HUVEC viability. However, rhodostomin (2 g/ml), a member of disintegrins, significantly reduced HUVEC viability. These data suggest that blockade of the vWF-endothelial GPIb interaction is not sufficient to induce cell death and endothelial GPIb receptor is not essential for cell survival.
Inhibition of Spontaneous Angiogenesis by Agkistin and AP1 in CAM Model-Several useful models have been used to study the role of cell adhesion molecule in angiogenesis. One of the most common in vivo models is the chick embryo CAM assay (29). To evaluate the role of endothelial GPIb complex in neovascularization in vivo, the effects of agkistin and anti-GPIb mAb AP1 on spontaneous angiogenesis occurring in the chick CAM model were examined. Upon dissection of the CAM of 12-day-old chick embryo, both agkistin (Fig. 5, B and C) and g/embryo, respectively) in a total volume of 20 l were applied on 10-day-old CAMs. After a 48-h incubation, CAMs were resected, fixed, and photographed. This is a representative one of three similar experiments. AP1 (Fig. 5, D and E), which were topically applied on CAM for 48 h, dose dependently inhibited the spontaneous angiogenesis in the CAM model as examined under filter disc as compared with control CAMs (Fig. 5A, PBS). These data suggest that endothelial GPIb receptor may significantly contribute to spontaneous angiogenesis in vivo. Moreover, blockade of vWF-GPIb interaction either by anti-GPIb mAb (i.e. AP1) or by polypeptide (i.e. agkistin) markedly inhibited neovascularization. On the other hand, accutin, a disintegrin derived from the same viper venom of A. acutus, inhibited angiogenesis by acting as endothelial integrin ␣ v ␤ 3 antagonist and by inducing apoptosis (5). Taken together, snake venom constituents affect cell-matrix interaction via multiple mechanisms, resulting in modification of platelet and vascular cell behaviors, an important cellular process in thrombosis, hemostasis, and angiogenesis (30). Furthermore, at a higher dose (2 g/embryo), both agkistin and AP1 significantly disrupted preexisting blood vessels as evidenced by disappearance of large vessels (Fig. 5, C and E). This is different from the previous observation with integrin ␣ v ␤ 3 antagonists, which inhibited angiogenesis without affecting preexisting blood vessels (4,5). Therefore, the action mechanisms of anti-angiogenic activity elicited by endothelial GPIb antagonists are quietly different from those of integrin ␣ v ␤ 3 antagonists and remain to be further investigated. To our knowledge, it is the first report describing the functional role of endothelial GPIb in spontaneous angiogenesis, and GPIb antagonists can markedly inhibit this process. It remains to be elucidated regarding whether agkistin affects the multisteps in the angiogenic process such as tissue degradation, migration, and differentiation elicited by a specific angiogenic factor (e.g. vascular endothelial growth factor or basic fibroblast growth factor).
In conclusion, agkistin specifically inhibited HUVEC adhesion to immobilized vWF through selective binding of endothelial GPIb, resulting in a blockade of interaction between vWF and endothelial GPIb. Agkistin binds to endothelial GPIb in a specific and saturable manner. Furthermore, both agkistin and AP1 significantly inhibited spontaneous angiogenesis in the chick CAM model in vivo but apparently did not affect HUVEC viability. In addition to their anti-thrombotic activity of agkistin and crotalin, another venom platelet GPIb antagonist (31), these GPIb antagonists may be useful tools for developing a new therapeutic strategy in angiogenesis-related diseases, such as inflammation and tumor metastasis.