Endothelial cell surface F1-F0 ATP synthase is active in ATP synthesis and is inhibited by angiostatin.

Angiostatin blocks tumor angiogenesis in vivo, almost certainly through its demonstrated ability to block endothelial cell migration and proliferation. Although the mechanism of angiostatin action remains unknown, identification of F(1)-F(O) ATP synthase as the major angiostatin-binding site on the endothelial cell surface suggests that ATP metabolism may play a role in the angiostatin response. Previous studies noting the presence of F(1) ATP synthase subunits on endothelial cells and certain cancer cells did not determine whether this enzyme was functional in ATP synthesis. We now demonstrate that all components of the F(1) ATP synthase catalytic core are present on the endothelial cell surface, where they colocalize into discrete punctate structures. The surface-associated enzyme is active in ATP synthesis as shown by dual-label TLC and bioluminescence assays. Both ATP synthase and ATPase activities of the enzyme are inhibited by angiostatin as well as by antibodies directed against the alpha- and beta-subunits of ATP synthase in cell-based and biochemical assays. Our data suggest that angiostatin inhibits vascularization by suppression of endothelial-surface ATP metabolism, which, in turn, may regulate vascular physiology by established mechanisms. We now have shown that antibodies directed against subunits of ATP synthase exhibit endothelial cell-inhibitory activities comparable to that of angiostatin, indicating that these antibodies function as angiostatin mimetics.

Regulation of invasion of epithelial ovarian cancer by transforming growth factor-beta.

OBJECTIVE: The metastatic process in epithelial ovarian cancer is thought to involve surface shedding and subsequent dissemination of ovarian cancer cells, facilitated by localized proteolysis at the interface between ovarian cancer cells and peritoneal surfaces. The factors regulating the metastatic process, however, are not well understood. Transforming growth factor-beta (TGF-beta) is a multifunctional peptide that elicits numerous cellular effects pertinent to the metastatic process. The purpose of this study was to evaluate the regulatory role of TGF-beta on metastasis in ovarian cancer. METHOD: We evaluated the effect of TGF-beta on the metastatic characteristics (adhesion, invasion, motility, proteolysis) of five ovarian cancer cell lines (DOV-13 and OVCA 420, 429, 432, and 433), two short-term primary ovarian cancer cell cultures (OVCA 10 and OVCA 208), and five normal ovarian surface epithelial (NOSE) cell cultures (OSE 133, 185, 186, 188, and 189). The effect of TGF-beta on invasion and proteolysis was quantified using a modified Boyden chamber invasion assay, zymography, a coupled colorimetric activity assay, and an HPLC-based quantitation of synthetic substrate cleavage. RESULTS: TGF-beta significantly increased invasion in five of seven ovarian cancer cell lines in amounts ranging from 2- to 20-fold. In contrast, TGF-beta significantly decreased invasion in two of five NOSE isolates by 50 to 80% and had no significant effect on invasion in three. TGF-beta treatment increased matrix metalloproteinase (MMP) expression in OVCA 420 and 433 and DOV-13, resulting in MMP-dependent collagen cleavage and invasive activity. Addition of the MMP inhibitor GI12947 neutralized the enhancing effect of TGF-beta on invasion. TGF-beta had no effect on ovarian cancer cell motility and only increased adhesion in DOV-13. CONCLUSIONS: These data suggest that TGF-beta may enhance the invasiveness of ovarian cancers through induction of MMP activity.

Angiostatin binds ATP synthase on the surface of human endothelial cells.

Angiostatin, a proteolytic fragment of plasminogen, is a potent antagonist of angiogenesis and an inhibitor of endothelial cell migration and proliferation. To determine whether the mechanism by which angiostatin inhibits endothelial cell migration and/or proliferation involves binding to cell surface plasminogen receptors, we isolated the binding proteins for plasminogen and angiostatin from human umbilical vein endothelial cells. Binding studies demonstrated that plasminogen and angiostatin bound in a concentration-dependent, saturable manner. Plasminogen binding was unaffected by a 100-fold molar excess of angiostatin, indicating the presence of a distinct angiostatin binding site. This finding was confirmed by ligand blot analysis of isolated human umbilical vein endothelial cell plasma membrane fractions, which demonstrated that plasminogen bound to a 44-kDa protein, whereas angiostatin bound to a 55-kDa species. Amino-terminal sequencing coupled with peptide mass fingerprinting and immunologic analyses identified the plasminogen binding protein as annexin II and the angiostatin binding protein as the alpha/beta-subunits of ATP synthase. The presence of this protein on the cell surface was confirmed by flow cytometry and immunofluorescence analysis. Angiostatin also bound to the recombinant alpha-subunit of human ATP synthase, and this binding was not inhibited by a 2,500-fold molar excess of plasminogen. Angiostatin's antiproliferative effect on endothelial cells was inhibited by as much as 90% in the presence of anti-alpha-subunit ATP synthase antibody. Binding of angiostatin to the alpha/beta-subunits of ATP synthase on the cell surface may mediate its antiangiogenic effects and the down-regulation of endothelial cell proliferation and migration.

Evidence for preferential adhesion of ovarian epithelial carcinoma cells to type I collagen mediated by the alpha2beta1 integrin.

Epithelial ovarian carcinoma, the leading cause of gynecologic cancer death, is characterized by widespread intra-abdominal metastases mediated primarily by surface shedding of tumor cells and peritoneal implantation. Whereas hematogenous metastasis is known to involve cellular adhesion, extracellular matrix proteolysis and cell migration, the role of these processes in the intraperitoneal dissemination of ovarian cancer remains unclear. To analyze further the role of adhesion and proteolysis in ovarian carcinoma dissemination, we have characterized the adhesive profiles of 4 primary cultures of ovarian carcinoma cells and 5 ovarian carcinoma cell lines. Our data demonstrate preferential adhesion of ovarian carcinoma cells to interstitial type I collagen. Analysis of adhesion molecule expression demonstrated the presence of the alpha2 and beta1 integrin subunits by cell surface ELISA, immunoprecipitation and immunohistochemistry. Furthermore, antibodies directed against the alpha2 and beta1 subunits inhibited adhesion of ovarian carcinoma cells to type I collagen by 56% and 95%, respectively. Plasminogen activator and matrix metalloproteinase production by adherent cells was not altered as a consequence of adhesion to individual extracellular matrix proteins; however, adhesion to an extracellular matrix comprised primarily of interstitial collagen increased plasminogen activator activity in 5 of 5 cell lines. Since the ovarian carcinoma micro-environment is rich in type I collagen, our data suggest that preferential adhesion to type I collagen followed by secretion of serine and metalloproteinases may represent a biochemical mechanism by which the intraperitoneal dissemination of ovarian carcinoma is mediated.

Specific binding of urinary-type plasminogen activator (u-PA) to vitronectin and its role in mediating u-PA-dependent adhesion of U937 cells.

The present paper described interactions of urinary-type plasminogen activator (u-PA) with isolated protein components of the extracellular matrix (ECM) using kinetic and ligand-blotting analyses, as well as adhesion studies with u-PA-saturated U937 monocytic cells. Kinetic analyses showed that fibronectin and laminin were moderately effective at decreasing activation of plasminogen by u-PA (3-4-fold decrease in kcat/Km), while activation was stimulated slightly by collagen types I and IV (2-4-fold increase in kcat/Km). Ligand-blotting experiments using intact immobilized ECM proteins demonstrated that u-PA binds predominantly to vitronectin. This was supported by ELISA studies, which showed concentration dependent, saturable, reversible binding of u-PA to vitronectin (Kd,app. of 97 nM). Limited proteolysis of vitronectin followed by ligand-blotting analysis demonstrated u-PA binding to a specific vitronectin fragment (M(r) 49,000), and binding was shown to occur through the N-terminal fragment of u-PA. N-terminal sequence analysis indicated that this binding fragment of vitronectin originates with Thr-122 and comprises the hemopexin domain, including the heparin-binding region of the vitronectin molecule. Plasminogen activator inhibitor type I did not compete with u-PA for binding to vitronectin, suggesting both molecules may co-localize on vitronectin. In contrast, binding of u-PA to vitronectin was significantly inhibited by plasminogen, suggesting these molecules share a common binding site on vitronectin. In addition to in vitro studies, experiments were performed to assess the contribution of direct binding of u-PA to vitronectin on the adhesive behaviour of U937 cells. Binding of u-PA-saturated U937 cells to vitronectin was inhibited 66% by excess vitronectin, suggesting that direct binding of u-PA to vitronectin is the mechanism by which u-PA-dependent adhesion of U937 cells to vitronectin is mediated.

Coordinate expression of urinary-type plasminogen activator and its receptor accompanies malignant transformation of the ovarian surface epithelium.

OBJECTIVE: Because elevated expression and cell surface association of urinary-type plasminogen activator have been linked to invasive potential in certain tumor types, we examined the expression of urinary-type plasminogen activator and urinary-type plasminogen activator receptor in ovarian epithelial carcinoma tissues and cells as compared with normal ovarian epithelium. STUDY DESIGN: Monoclonal antibodies specific for urinary-type plasminogen activator and urinary-type plasminogen activator receptor were used for immunohistochemical staining of tissues and cells to assess expression of these antigens in frozen sections of normal and tumor tissue. Substrate zymography was used to detect plasminogen activator activity in ovarian carcinoma ascites and in conditioned media of cultured cells, whereas a Western blot assay was used to identify urinary-type plasminogen activator receptor in cultured cells. RESULTS: Normal ovarian epithelium expressed urinary-type plasminogen activator receptor (4/4 positive) but little or no urinary-type plasminogen activator (0/4 positive), whereas epithelial ovarian carcinomas frequently expressed urinary-type plasminogen activator (4/8 positive) in conjunction with urinary-type plasminogen activator receptor (7/9 positive). High levels of urinary-type plasminogen activator were detected in 15 of 19 samples of ascites. DOV 13, OVCA 420, OVCA 429, OVCA 432, and OVCA 433 cell lines secreted urinary-type plasminogen activator in variable quantities, whereas normal ovarian epithelial cells did not secrete any detectable plasminogen activator. Urinary-type plasminogen activator receptor had similar levels of expression in all cancer cell lines and normal ovarian epithelium. CONCLUSION: Overexpression of urinary-type plasminogen activator is associated with malignant transformation of the ovarian epithelium. Increased cell surface proteolysis mediated by urinary-type plasminogen activator bound to cell surface urinary-type plasminogen activator receptor may contribute to metastatic behavior in ovarian carcinoma.

Secretion of extracellular matrix-degrading proteinases is increased in epithelial ovarian carcinoma.

The biochemical events associated with tumor invasion involve localized degradation of the basement membrane by tumor-associated proteinases. In this study, we have characterized the proteinase secretion profiles of 5 ovarian epithelial carcinoma cell lines (DOV 13, OVCA 420, OVCA 429, OVCA 432, OVCA 433) as well as normal ovarian epithelial cells. Immunocapture assays demonstrated that all 5 carcinoma cell lines produce both secreted and surface-associated plasminogen activator. Urinary-type plasminogen activator (u-PA) production was one order of magnitude greater than production of tissue-type plasminogen activator (t-PA). Furthermore, t-PA secretion by normal ovarian epithelial cells was not detectable, whereas u-PA production was 17- to 38-fold lower than in ovarian carcinoma cells. Western-blotting analysis demonstrated that u-PA was secreted as the single chain form (scu-PA) when cells were cultured in serum-free medium. Incubation of plasminogen with ovarian carcinoma cell-conditioned medium resulted in direct activation of the zymogen to plasmin. Furthermore, following incubation of cells with plasminogen, plasmin was eluted from the cell surface, indicating that ovarian carcinoma cells contain binding sites for plasminogen/plasmin which are accessible to surface-associated plasminogen activators. In addition to plasminogen activators, metalloproteinases were also produced by DOV 13, OVCA 429 and OVCA 433 cells. DOV 13 cells produce a 68-kDa metalloproteinase similar to matrix metalloproteinase 2 (MMP-2) whereas a 92-kDa enzyme similar to MMP-9 is secreted by OVCA 429 and 433. Together, ovarian carcinoma-associated plasminogen activators and metalloproteinases catalyze the hydrolysis of the major basement membrane protein components, type-IV collagen, type-IV gelatin, laminin and fibronectin. The enhanced proteolytic capability of ovarian carcinoma cells relative to normal ovarian epithelium suggests a biochemical mechanism by which invasion and spread of ovarian epithelial carcinoma may be mediated.

The extracellular matrix proteins laminin and fibronectin contain binding domains for human plasminogen and tissue plasminogen activator.

This study describes the binding of plasminogen and tissue-type plasminogen activator (t-PA) to the extracellular matrix proteins fibronectin and laminin. Plasminogen bound specifically and saturably to both fibronectin and laminin immobilized on microtiter wells, with Kd(app) values of 115 and 18 nM, respectively. Limited proteolysis by endoproteinase V8 coupled with ligand blotting analysis showed that both plasminogen and t-PA preferentially bind to a 55-kDa fibronectin fragment and a 38-kDa laminin fragment. Amino acid sequence analysis demonstrated that the 5-kDa fragment originates with the fibronectin amino terminus whereas the laminin fragment was derived from the carboxyl-terminal globular domain of the laminin A chain. Ligand blotting experiments using isolated plasminogen domains were also used to identify distinct regions of the plasminogen molecule involved in fibronectin and laminin binding. Solution phase fibronectin binding to immobilized plasminogen was mediated primarily via lysine binding site-dependent interactions with plasminogen kringles 1-4. Lysine binding site-dependent binding of soluble laminin to immobilized plasminogen kringles 1-5 as well as an additional lysine binding site-independent interaction between mini-plasminogen and the 38-kDa laminin A chain fragment were also observed. These studies demonstrate binding of plasminogen and tissue-type plasminogen activator to specific regions of the extracellular matrix glycoproteins laminin and fibronectin and provide further insight into the mechanism of regulation of plasminogen activation by components of the extracellular matrix.

Binding of human plasminogen to basement-membrane (type IV) collagen.

Plasminogen, the zymogen form of the serine proteinase plasmin, has been implicated in numerous physiological and pathological processes involving extracellular-matrix remodelling. We have previously demonstrated that the activation of plasminogen catalysed by tissue plasminogen activator is dramatically stimulated in the presence of basement-membrane-specific type IV collagen [Stack, Gonzalez-Gronow & Pizzo (1990) Biochemistry 29, 4966-4970]. The present paper describes the binding of plasminogen to type IV collagen. Plasminogen binds to both the alpha 1(IV) and alpha 2(IV) chains of basement-membrane collagen, with binding to the alpha 2(IV) chain preferentially inhibited by 6-aminohexanoic acid. This binding is specific and saturable, with Kd,app. values of 11.5 and 12.7 nM for collagen and gelatin respectively. Although collagen also binds to immobilized plasminogen, this interaction is unaffected by 6-aminohexanoic acid. Limited elastase proteolysis of plasminogen generated distinct collagen-binding fragments, which were identified as the kringle 1-3 and kringle 4 domains. No binding of collagen to mini-plasminogen was observed. These studies demonstrate a specific interaction between plasminogen and type IV collagen and provide further evidence for regulation of plasminogen activation by protein components of the extracellular matrix.

Limited proteolysis of vitronectin by plasmin destroys heparin binding activity.

Vitronectin (VN) stabilizes plasminogen activator inhibitor type 1 (PAI-1) activity and prevents the fibrin(ogen)-induced acceleration of plasminogen activation by t-PA. These antifibrinolytic activities as well as other functions are mediated by the glycosaminoglycan (GAG) binding domain of VN. Since the GAG binding region is rich in arginyl and lysyl residues, it is a potential target for enzymes such as plasmin. In this paper, the dose and time-dependent proteolysis of VN by plasmin is demonstrated. The addition of urokinase or streptokinase (200 units/ml) to plasma also produced proteolysis of VN. With minimal proteolysis, the 75 kDa band was degraded to a 62-65 kDa form of VN. This minimal proteolysis destroyed the binding of [3H]-heparin to VN and reversed the neutralization of heparin by VN. Thus, the plasmin-mediated proteolysis of the GAG binding activity of VN could destroy the antifibrinolytic activity of VN during physiologic conditions and during thrombolytic therapy. Furthermore, other functions of VN in complement and coagulation systems that are mediated by the GAG binding domain may be destroyed by plasmin proteolysis.

Vitronectin in the substratum of endothelial cells is cross-linked and phosphorylated.

Vitronectin (VN), previously shown to be a substrate for purified transglutaminases, was demonstrated in this study to be cross-linked when incubated with HUVEC and EAhy926 cells. The cross-linking was calcium-dependent and required that VN be plated at the substratum of the cells. These cells also phosphorylated VN, but in contrast to a previous study demonstrating a cAMP-dependent protein kinase in platelets, the phosphorylation of VN by was decreased with the addition of 1mM cAMP. The cross-linking of VN by endothelial cells demonstrates that the adhesion of these cells to VN is a dynamic process in which the substratum may be enzymatically altered. Furthermore, the modifications of VN by cross-linking and phosphorylation could modulate the functions of VN and influence events such as endothelial cell proliferation and angiogenesis.

Highly sulfated glycosaminoglycans augment the cross-linking of vitronectin by guinea pig liver transglutaminase. Functional studies of the cross-linked vitronectin multimers.

Vitronectin (VN) is an adhesive glycoprotein with roles in the complement, coagulation, and immune systems. Many of the functions of VN are mediated by a glycosaminoglycan binding site, near its carboxyl-terminal end. In this paper, we show that the highly sulfated glycosaminoglycans (GAGs), dextran sulfate, pentosan polysulfate, and fucoidan effectively augment [14C]putrescine incorporation into VN and cross-linking of VN into high molecular multimers by guinea pig liver transglutaminase (TG). Other GAGs including heparin, low molecular weight heparin, dermatan sulfate, keratan sulfate, and the nonsulfated dextrans were ineffective in accelerating these reactions. Dextran sulfate of average molecular mass 500 kDa was more effective than dextran sulfate of average molecular mass 5 kDa, supporting a template mechanism of action of the GAGs, in which VN molecules align on the GAG in a conformation suitable for cross-linking. The VN multimers catalyzed by TG retained functional activity in binding [3H]heparin, platelets, and plasminogen activator inhibitor type-1 (PAI-1). [3H]Heparin bound selectively to the 65-kDa monomeric band of VN and to the multimers derived from this band. PAI-1, however, bound equally to both the 75- and 65-kDa monomeric forms of VN, suggesting that the PAI-1 binding site on VN is distinct from the GAG binding site. The interaction of GAGs with the TG-catalyzed cross-linking of VN may facilitate studies of VN structure-function relationships.

Vitronectin is a substrate for transglutaminases.

Vitronectin (VN) was found to be a substrate for both plasma transglutaminase (Factor XIIIa) and guinea pig liver transglutaminase (TG). Incorporation of [3H]-putrescine indicated the presence of reactive glutaminyl residues in VN. When VN was incubated with TG or Factor XIIIa, in the absence of putrescine, multimeric covalent complexes were identified, indicating that VN can also contribute lysyl residues to the bond catalyzed by transglutaminases. Cross-linking of VN by TG and Factor XIIIa may modulate the effects of VN on the complement and coagulation systems in hemostatic plugs and extracellular matrix.