Human ovarian cancer ascites fluid contains a mixture of incompletely degraded soluble products of fibrin that collectively possess an antiangiogenic property.

Ovarian cancer ascites fluid (OCAF) displayed an antiangiogenic property in a chick chorioallantoic membrane (CAM) assay. This property was attributed in part to angiostatin although angiostatin-free OCAF retained a net antiangiogenic property. Recently, immunopurified fibrin(ogen) degradation products (FDPs) from malignant effusions of VX2 tumor-burdened rabbits exhibited antiangiogenic activity on the CAM. We questioned whether the FDPs of OCAF were also antiangiogenic. FDPs were immunopurified from individual OCAF samples, characterized by sodium dodecyl sulfate-polyacrylamide gel electrophoresis /western blots, enzyme-linked immunosorbent assays, and CAM assays. FDPs of OCAFs consisted of soluble high molecular weight (MW) fragments (>200 kd; approximately 40% of total FDPs), D-dimer (approximately 180 kd; approximately 37%), fragment D (approximately 90 kd; approximately 15%), and fragment E (approximately 50 kd; approximately 8%); intact fibrinogen was absent. When applied to CAM surfaces (0.5-1.6 mg/10 mL), purified FDPs significantly reduced the area of chorionic capillaries from 90% (in controls) to 47% over a 48-h period; from CAM sections, capillary density was reduced from 60% (controls) to 26%. FDPs prepared from fibrinogen displayed a similar antiangiogenic effect. Further digestion of OCAF FDPs by human plasmin caused degradation of high MW fragments, releasing additional D-dimer, fragment D, and fragment E. Of the fibrinogen-related components, OCAF contained only soluble FDPs (including incompletely digested fibrin fragments). Collectively, these FDPs contributed to the net antiangiogenic property of ascites fluid.

Malignant ascites fluid (MAF), including ovarian-cancer-associated MAF, contains angiostatin and other factor(s) which inhibit angiogenesis.

OBJECTIVE: The aim was to determine whether human malignant ascites fluid (MAF) associated with abdominal cancer, including ovarian cancer, contained factors which inhibit angiogenesis as well as others which stimulate this process. METHODS: MAF was collected from six patients, four with ovarian cancer, one with gastric cancer, and one with liver metastases. Using the chick chorioallantoic membrane (CAM) the effect of MAF on 7-day-old CAM capillaries was examined for 48 h. Vascular endothelial growth factor (VEGF) was evaluated by ELISA. Five samples of MAF were fractionated by lysine-Sepharose chromatography and the lysine-bound and -unbound fractions were eluted by epsilon-amino-n-hexanoic acid. Whole MAF, the lysine-bound and -unbound fractions, and human angiostatin were subjected to SDS-PAGE/Western blot analysis and immunostained after exposure to anti-human plasminogen. Human plasminogen was exposed to conditioned medium from ovarian epithelial cancer (HEY) cells and subjected to similar Western blot analysis. RESULTS: Despite containing VEGF, each MAF sample examined caused a loss of capillaries from the CAM; a similar response was seen using purified human angiostatin. Whole MAF and the lysine-bound fraction contained plasminogen (90 kDa) and a 55-kDa protein which migrated in a similar manner to human angiostatin on Western blot. Both the lysine-bound and -unbound fractions caused a loss of capillaries in the CAM. Human plasminogen exposed to conditioned medium from HEY cells yielded a fragment which was similar in size to angiostatin. CONCLUSIONS: MAF from patients with various clinical presentations contains angiostatin and VEGF as well as other factors which are capable of inhibiting angiogenesis.

Metabolism of rabbit angiostatin glycoforms I and II in rabbits: angiostatin-I leaves the intravascular space faster and appears to have greater anti-angiogenic activity than angiostatin-II.

Plasminogen (PLG) exists in the circulation as two glycoforms, I and II. Angiostatin (AST) is a polypeptide that has been cleaved from the kringle region of PLG and has strong anti-angiogenic properties. AST-I and AST-II, which consisted only of kringles 1 through 3, were prepared by the action of urokinase on purified rabbit PLG-I and PLG-II, respectively, in the presence of N-acetyl cysteine, followed by affinity chromatography on lysine-Sepharose. Purified AST-I and AST-II were tested for functional activity with a chick chorioallantoic membrane (CAM) model; when similar amounts were applied to a 6-day CAM, AST-I was substantially more effective than AST-II in decreasing vascular supply to the CAM over a 72-hour period; this activity correlated with a loss of capillaries, probably through apoptosis of endothelial cells. Radiolabeled AST-I and AST-II (iodine 125 and iodine 131) were co-injected intravenously into healthy rabbits to determine their clearances from plasma measured over 3 days. Over a dose range of 0.08 to 2.7 microg/kg, the fractional catabolic rate within the intravascular space (j(3)) indicated that AST-I was cleared 3-fold to 4-fold more rapidly than AST-II (P < .001). The catabolic half-life of AST-I (2.01 +/- 0.19 days) was significantly less than that of AST-II (2.62 +/- 0.20 days). The faster clearance of AST-I from the intravascular space was matched by its more rapid passage than AST-II to the extravascular space of various organs over 60 minutes in vivo. This property of AST-I as compared with AST-II may partially explain its greater anti-angiogenic potential. From the plasma concentrations of PLG-I and PLG-II and their relative behaviors toward rabbit VX-2 lung tumors in vivo, we predict that substantially greater quantities of AST-II than AST-I may be released into the extravascular space of tumors.

Metabolism and distribution of the virus-encoded serine proteinase inhibitor SERP-1 in healthy rabbits.

SERP-1 is a secreted myxoma virus-encoded 55-kd protein of the serine proteinase inhibitor ("serpin") family that strongly inhibits the mitosis of medial arterial smooth muscle cells, thus preventing stenosis in injured rabbit and rat arteries. We have measured the fractional catabolic rate (FCR) and compartmental distribution of 1251-SERP-1 after injection of various doses into the circulation of healthy rabbits. The FCR within the intravascular space decreased from 2.99 d(-1) to 2.39 d(-1) and the whole-body FCR decreased from 0.66 d(-1) to 0.51 d(-1) as the dose was increased 35-fold from 0.11 microg/kg to 3.8 microg/kg. The fractional distribution of SERP-1 between the intravascular (0.21), noncirculating vascular wall (0.09), and extravascular compartments (0.70) at equilibrium did not change significantly over this dose range. SERP-1 did not appear to selectively accumulate in any organ in any of 11 rabbits studied over a 6-day interval. In comparison to other rabbit plasma serpins, the behavior of SERP-1 in vivo most closely resembled that of heparin cofactor II.

The procoagulant state of the VX-2 tumor in rabbit lung in vivo--relative accumulation of fibrinogen, prothrombin, plasminogen, antithrombin and heparin cofactor II within the tumor.

During their growth, many malignant solid tumors elicit a hemostatic response in the host. In this report, the fluxes of various rabbit plasma hemostatic proteins into pulmonary VX-2 tumors have been measured in vivo. VX-2 cells were contained within a small rabbit plasma clot which was injected intravenously into rabbits. At 28 d, each rabbit was injected intravenously with two radiolabeled (131I, 125I) proteins selected from fibrinogen, prothrombin, antithrombin-alpha, heparin cofactor II, or plasminogen-I or -II. After allowing the labeled proteins to circulate for 10-200 min, each rabbit was perfused with Krebs-Henseleit solution and the lungs excised. Solid tumors were isolated, weighed and measured for radioactivity content. A mean of 14 tumors/pair of lungs with a mean tumor weight of 0.3 g was obtained. Radioactivity per g of tumor was related to radioactivity/ml of blood at exsanguination for each protein at each time interval. Maximum flux rates were calculated (as pmol/g of tumor/min): Fibrinogen, 65.0; prothrombin, 53.0; antithrombin-alpha, 24.1; HCII, 17.2; plasminogen-II, 5.0; and plasminogen-I, 3.2. Using immunocytochemical staining, fibrin(ogen) was distributed heterogeneously within the VX-2 tumor, appearing largely in the perinodular region and in the necrotic core. From the net fluxes of these proteins, the profile of chronic hemostasis associated with the VX-2 tumor was shown to differ markedly from the hemostatic response that occurs after an acute vascular injury.

Cytokeratin 18 is expressed on the hepatocyte plasma membrane surface and interacts with thrombin-antithrombin complexes.

During experiments to identify putative hepatic receptors for thrombin-antithrombin (TAT) complexes, a 45-kDa protein was identified by ligand blotting. Following gel purification, amino acid sequencing revealed the 45-kDa TAT-binding polypeptide to be cytokeratin 18 (CK18). The presence of CK18 on the surface of intact rat hepatoma cells was demonstrated by binding of 125I-anti-CK18 antibodies. Anti-CK18 antibodies reduced the binding and internalization of 125I-TAT by rat hepatoma cells. Immunocytochemical analysis, to determine the location of CK18 in vivo, revealed a periportal gradient of CK18 staining; with hepatocytes around the portal triads demonstrating striking pericellular staining. In addition, anti-CK18 IgG associated with perfused livers to a significantly greater extent than preimmune IgG. Taken together, these data provide evidence that CK18 is found on the extracellular surface of hepatocytes and could play a role in TAT removal. Finally, these data, in conjunction with recent reports of CK8 (Hembrough, T. A., Li, L., and Gonias, S. L. (1996) J. Biol. Chem. 271, 25684-25691) and CK1 cell membrane surface expression (Schmaier, A. H. (1997) Thromb. Hemostasis 78, 101-107), indicate a novel role for these proteins as putative cellular receptors or cofactors to cellular receptors.

Catabolism of unglycated and naturally glycated forms of rabbit fibrinogen: their interaction with the healthy and deendothelialized aorta wall in normal and diabetic rabbits.

The incidence of nonenzymatic glycation of proteins within the hyperglycemic environment of diabetic plasma is increased compared with that in normal (i.e., nondiabetic) plasma. Whether glycation in vivo alters the behavior of proteins within the circulation is not well understood. Glycated fibrinogen, although not detected in normal rabbit plasma, was isolated from diabetic rabbit plasmas (glucose concentration 12 to 39 mmol/L) and separated from unglycated fibrinogen by boronate chromatography. The yield of glycated fibrinogen, which amounted to 3% to 6% of total fibrinogen, correlated with the content of plasma glucose. Glycated and unglycated fibrinogens facilitated aggregation of normal or diabetic platelets to a similar extent after adding adenosine-5'-diphosphate. Normal platelets stimulated by adenosine-5'-diphosphate bound more iodine 131-glycated fibrinogen than iodine 125-unglycated fibrinogen (p < 0.05), whereas the quantities of glycated and unglycated fibrinogens bound by diabetic platelets were not significantly different. When coinjected intravenously into normal or diabetic rabbits, 131I-glycated fibrinogen was cleared from the circulation faster than 125I-unglycated fibrinogen although the catabolic rates, measured as half-life, were not significantly different. At equilibrium, glycated fibrinogen was distributed significantly more in the extravascular and less in the vascular compartments of the normal and diabetic rabbit compared with the unglycated type. After balloon injury to the aorta in vivo, the unglycated/glycated ratio of radiolabeled fibrinogens associated with the platelet monolayer was 0.94, whereas for the damaged subendothelium the ratio was 1.20 (p < 0.005). We conclude that glycation in vivo changes several metabolic characteristics of fibrinogen in the normal and diabetic rabbit.

Molecular cloning and expression of rabbit antithrombin III.

A cDNA containing the complete open-reading frame encoding rabbit antithrombin III (AT-III) was isolated from a rabbit liver cDNA expression library, using a specific antibody as a probe. Sequence analysis showed 84% identity between the deduced amino acid sequences of the rabbit and human proteins. A previously described cell-free expression system was used to verify the identity of the clone. The full-length cDNA was inserted into an expression vector, and messenger RNA (mRNA) transcripts generated. In vitro translation of these transcripts, in the presence of [35S]methionine, in an mRNA-dependent rabbit reticulocyte lysate system resulted in the synthesis of a 51-Kd polypeptide, as shown by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). This nonglycosylated protein was capable of forming SDS-stable complexes with human alpha-thrombin. Complex formation was significantly enhanced following the deletion of nucleotides encoding the signal peptide, and the resultant generation of a 47-Kd nonglycosylated mature protein product. When the template DNA giving rise to this product was internally truncated, two rabbit AT-III deletion mutants were generated that lacked the ability to interact with thrombin, but retained the ability to bind heparin. Cell-free expression plasmids encoding the human and rabbit AT-III mature molecules were manipulated to produce two interspecies fusion proteins. For the first, human codons were used to replace rabbit codons from residue 369-433, while in the second human codons replaced rabbit codons from residue 217-433. Both fusion proteins exhibited less efficient thrombin-complexing ability than the original cell-free-derived mature rabbit AT-III. Thus, portions of AT-III molecules from the two species, despite their high degree of homology, are not interchangeable. Knowledge of the structure of rabbit AT-III, combined with the availability of the rabbit cDNA, will permit defined experimentation aimed at understanding antithrombin III structure relative to its function in vivo.

The structural heterogeneity of the carbohydrate moiety of desialylated human transferrin.

Human transferrin consists of a single chain polypeptide which supports two N-glycosidically linked glycans at sequons a and b. Glycopeptides were released from human transferrin by proteolytic digestion, desialylated by mild acid hydrolysis, and then isolated by chromatographic methods. The structures of the glycans located on each sequon were determined by a combination of analytical techniques including Smith degradation, permethylation, and enzymic degradation. Approximately 79% of the total glycan from sequon a was of the biantennary type as previously described by Dorland and his colleagues (FEBS Lett. 77, 15-20 (1977)). The remaining 21% consisted of a mixture of triantennary and tetraantennary glycans, each amounting to approximately 10% of the total glycan for this sequon. The triantennary structure resembled that described for the N-glycosidic triantennary glycans of bovine fetuin by Nilsson and his colleagues (J. Biol. Chem. 254, 4545-4553 (1979)). Of the tetraantennary glycan, approximately half of the structures were incomplete, i.e., one antenna terminated by N-acetylglucosamine. On sequon b, 81% of the glycan was biantennary, identical to those biantennary glycans of sequon a, and the reminder was triantennary, also of the fetuin type. The glycan structures and their locations on the polypeptide are related to the known subpopulations of human transferrin.

Bovine serum transferrin phenotypes AA, D1D1, D2D2, EE: their carbohydrate compositions and electrophoretic multiplicity.

Samples of homozygous bovine serum transferrins have been prepared and their purity has been ascertained by immunological techniques and electrophoretic analysis in SDS. Measurements of carbohydrate composition show that no significant differences exist among the phenotype variants AA, D1D1, D2D2, and EE. Chromatography of transferrin AA on DEAE-cellulose separated four subfractions, each of which corresponded well with one band obtained by polyacrylamide gel electrophoresis. Carbohydrate analyses of the individual subfractions did not show significant differences in sialic acid, hexose, or hexosamine contents. After desialylation with neuraminidase, each subfraction was converted to a major band and a minor band on gel electrophoresis. From the relative band positions of the desialylated transferrins, it was concluded that possession of sialyl residued by bovine transferrin is not the primary cause of electrophoretic multiplicity. Rather, sialic acid masks an underlying heterogeneity which most likely resides within the polypeptide chain. Further characterization of this heterogeneity will best be undertaken with the isolated asialotransferrin subfractions.

Studies on the coagulant enzyme from Agkistrodon rhodostoma venom. Isolation and some properties of the enzyme.

1. Arvin, a commercial preparation of the coagulant activity from the venom of Agkistrodon rhodostoma, is shown to contain a non-coagulant caseinolytic fraction. 2. A method is described for the purification of the coagulant enzyme free from any detectable contaminating protein. 3. The coagulant enzyme is identified as a glycoprotein which probably consists of a single polypeptide chain containing approx. 29% by weight of carbohydrate. Amino acid and carbohydrate analyses are reported and the N- and C-terminal amino acid residues identified. 4. Electrophoresis on polyacrylamide gel reveals the polymorphic nature of the glycoprotein. Five forms of the enzyme are observed. 5. The coagulant action is correlated with an arginine esterase activity and kinetic properties are studied with both arginine and lysine esters as substrates. The inhibitory nature of guanidine and arginine toward the esterase activity is reported.

The proteolytic action of Arvin on human fibrinogen.

1. Human fibrinogen was subjected to proteolysis by enzyme preparations (clinical Arvin and IRC-50 Arvin) from the venom of Agkistrodon rhodostoma. 2. IRC-50 Arvin releases three peptides from fibrinogen, and these were identified as fibrinopeptides AP, AY and A. 3. The less purified ;clinical' Arvin releases, in addition to fibrinopeptides AP, AY and A, small amounts of two heptapeptides derived from fibrinopeptides AP and A, probably because it contains another enzyme as well as Arvin. 4. No fibrinopeptide B is released by either Arvin preparation. 5. Thus, although Arvin is known to differ from ;reptilase' from Bothrops jararaca in that it does not activate the enzyme that cross-links fibrin (fibrin-stabilizing factor), it is identical with reptilase with respect to the peptides that it liberates from fibrinogen.