Effect of cell density on thrombin binding to a specific site on bovine vascular endothelial cells.

We studied thrombin binding to proliferating and confluent endothelial cells derived from bovine vascular endothelium. [125]thrombin was incubated with nonconfluent or confluent endothelial cells and both the total amount bound and the amount linked in a 77,000-dalton thrombin-cell complex were determined. Approximately 230,000 molecules of thrombin bound per cell in nonconfluent cultures compared to 12,800 molecules per cell in confluent cultures. Approximately 67,7000 thrombin molecules were bound in an apparently covalent complex, Mr = 77,000, with each cell in sparse cultures, whereas only 4,600 thrombin molecules per cell were bound in this complex with confluent cultures. Similar studies with [125I]thrombin and endothelial cells derived from bovine cornea revealed no difference either in the total amount of thrombin bound or in the amount bound in the 77,000-dalton complex using sparse or confluent cultures. When confluent vascular endothelial cultures were wounded, additional cellular binding sites for the 77,000-dalton complex with thrombin appeared within 24 h. A 237% increase in the amount of thrombin bound to these sites was induced by a wound which resulted in a 20% decrease in cell number in the monolayer. There was no significant increase in thrombin binding to other cellular sites at 24 h. These experiments provide evidence that the first change in thrombin binding after injury is an increase in the cellular sites involved in the 77,000-dalton complex, and suggest that thrombin binding to endothelial cells may be important in the vascular response to injury.

Redistribution of alpha-granules and their contents in thrombin-stimulated platelets.

The redistribution of beta-thromboglobulin (beta TG), platelet Factor 4 (PF4), and fibrinogen from the alpha granules of the platelet after stimulation with thrombin was studied by morphologic and immunocytochemical techniques. The use of tannic acid stain and quick-freeze techniques revealed several thrombin-induced morphologic changes. First, the normally discoid platelet became rounder in form, with filopodia, and the granules clustered in its center. The granules then fused with one another and with elements of the surface-connected canalicular system (SCCS) to form large vacuoles in the center of the cell and near the periphery. Neither these vacuoles nor the alpha granules appeared to fuse with the plasma membrane, but the vacuoles were connected to the extracellular space by wide necks, presumably formed by enlargement of the narrow necks connecting the SCCS to the surface of the unstimulated cell. The presence of fibrinogen, beta TG, and PF4 in corresponding large intracellular vacuoles and along the platelet plasma membrane after thrombin stimulation was demonstrated by immunocytochemical techniques in saponin-permeabilized and nonpermeabilized platelets. Immunocytochemical labeling of the three proteins on frozen thin sections of thrombin-stimulated platelets confirmed these findings and showed that all three proteins reached the plasma membrane by the same pathway. We conclude that thrombin stimulation of platelets causes at least some of the fibrinogen, beta TG, and PF4 stored in their alpha granules to be redistributed to their plasma membranes by way of surface-connected vacuoles formed by fusion of the alpha granules with elements of the SCCS.

The platelet fibrinogen receptor: an immunogold-surface replica study of agonist-induced ligand binding and receptor clustering.

Platelet aggregation requires the binding of fibrinogen to its receptor, a heterodimer consisting of the plasma-membrane glycoproteins (GP) IIb and IIIa. Although the GPIIb-IIIa complex is present on the surface of unstimulated platelets, it binds fibrinogen only after platelet activation. We have used an immunogold-surface replica technique to study the distribution of GPIIb-IIIa and bound fibrinogen over broad areas of surface membranes in unstimulated, as well as thrombin-activated and ADP-activated human platelets. We found that the immunogold-labeled GPIIb-IIIa was monodispersed over the surface of unstimulated platelets, although the cell surface lacked immunoreactive fibrinogen. On thrombin-stimulated platelets, approximately 65% of the GPIIb-IIIa molecules were in clusters within the plane of the membrane. Fibrinogen, which had been released from the alpha-granules of these cells, bound to GPIIb-IIIa on the cell surface and was similarly clustered. To determine whether the receptors clustered before ligand binding, or as a consequence thereof, we studied the surface distribution of GPIIb-IIIa after stimulation with ADP, which causes activation of the fibrinogen receptor function of GPIIb-IIIa without inducing the release of fibrinogen. In the absence of added fibrinogen, the unoccupied, yet binding-competent receptors on ADP-stimulated platelets were monodispersed. The addition of fibrinogen caused the GPIIb-IIIa molecules to cluster on the cell surface. Clustering was also induced by the addition of the GPIIb-IIIa-binding domains of fibrinogen, namely the tetrapeptide Arg-Gly-Asp-Ser on the alpha-chain or the gamma-chain decapeptide gamma 402-411. These results show that receptor occupancy causes clustering of GPIIb-IIIa in activated platelets.

A platelet alpha-granule membrane protein (GMP-140) is expressed on the plasma membrane after activation.

We have previously characterized a monoclonal antibody, S12, that binds only to activated platelets (McEver, R.P., and M.N. Martin, 1984, J. Biol. Chem., 259:9799-9804). It identifies a platelet membrane protein of Mr 140,000, which we have designated as GMP-140. Using immunocytochemical techniques we have now localized this protein in unstimulated and thrombin-stimulated platelets. Polyclonal antibodies to purified GMP-140 were used to enhance the sensitivity of detection. Nonpermeabilized, unstimulated platelets, incubated with anti-GMP-140 antibodies, and then with IgG-gold probes, showed very little label for GMP-140 along their plasma membranes. In contrast, thrombin-stimulated platelets exhibited at least a 50-fold increase in the amount of label along the plasma membrane. On frozen thin sections of unstimulated platelets we observed immunogold label along the alpha-granule membranes. We also employed the more sensitive technique of permeabilizing with saponin unstimulated platelets in suspension, and then incubating the cells with polyclonal anti-GMP-140 antibodies and Fab-peroxidase conjugate. Alpha-granule membranes showed heavy reaction product, but no other intracellular organelles were specifically labeled. These results demonstrate that GMP-140 is an alpha-granule membrane protein that is expressed on the platelet plasma membrane during degranulation.

Identification of a missense mutation in the factor VIII gene of a mild hemophiliac.

DNA probes derived from the cloned factor VIII gene can be used to detect mutations in the factor VIII gene of hemophiliacs. DNA hybridization analysis led to the identification of two contrasting point mutations in the same codon. In a severe hemophiliac with no detectable factor VIII activity, the normal arginine codon (number 2307) is converted to a stop codon, while in a mild hemophiliac with 10 percent of normal activity, this same codon is converted to glutamine.

Thrombin generation and secretion of platelet Factor 4 during blood clotting.

We have studied the platelet release reaction and thrombin generation during the spontaneous clotting of whole blood in vitro. Both thrombin formation and secretion of platelet Factor 4 were detected at least 12 min before clotting (clotting time, 22--26 min). Initially, at low thrombin concentrations (2--5 ng/ml), there is a small increase in plasma platelet Factor 4 (less than 1% of the amount present in serum). This is followed by a gradual increase in both platelet Factor 4 and thrombin concentrations over a 12 to 20-min interval. Finally, 5 min 5 before clotting, there is a rapid increase in both thrombin generation and platelet secretion. Thus, we have shown that the release of platelet Factor 4 is a prolonged reactoin and the extent to which it occurs parallel thrombin generation. It is only when thrombin concentrations are high (45--90) ng/ml)--during the period of clot formation--that the major part of platelet Factor 4 secretion occurs. Release of platelet Factor 4, like fibrin formation, occurs in the last step of in vitro coagulation.

Relationship between secretion of platelet Factor 4 and thrombin generation during in vitro blood clotting.

We have studied the effects of both impaired prothrombin activation and direct inhibition of thrombin on the platelet release reaction in clotting blood to determine the role of thrombin in this process. In blood from two patients with congenital Factor V deficiency, prothrombin activation during spontaneous in vitro clotting was delayed and decreased. Secretion of platelet Factor 4 was also delayed and was detected only after thrombin formation was initiated. Addition of a small amount of normal plasma to the patients' blood in vitro corrected the abnormalities in both thrombin formation and the platelet release reaction in parallel fashion. A delay in the onset of secretion of platelet Factor 4 was also observed when thrombin generated in normal blood during spontaneous in vitro clotting was inhibited by either purified hirudin or anti-thrombin Fab. These observations suggest that thrombin is the essential stimulus for platelet secretion during in vitro blood clotting. The effect of inhibitors of the platelet release reaction on prothrombin activation during in vitro blood clotting was also studied. When either prostacyclin or the combination of prostaglandin E(1) and N(6)O(2')-dibutyryl cyclic AMP was added, secretion of platelet Factor 4 was inhibited 85-95%. We were unable to detect any inhibition of initiation of prothrombin activation or inhibition of that part of thrombin generation associated with clotting. These results indicate either that the platelet release reaction may not be required for the initiation of prothrombin activation or only a very limited amount of secretion may be necessary for normal generation of thrombin to occur.

The measurement of thrombin in clotting blood by radioimmunoassay.

We have developed a radioimmunoassay for human thrombin using rabbit anti-human thrombin IgG. The assay can measure 2 ng thrombin/ml plasma, 500-fold more sensitive than clotting assays. Human prothrombin is less reactive in the assay than thrombin by at least four orders of magnitude, and there is no demonstrable cross-reactivity with human factor Xa, the clotting factor structurally most similar to thrombin. The assay does not detect thrombin bound to anthithrombin III. Using the assay, we have demonstrated that plasma from 20 normal subjects does not contain detectable thrombin. We measured thrombin generation in clotting blood in polypropylene tubes and observed that thrombin appears (approximately equal to 3 ng/ml) within 45 S-5 min after venipuncture. This material is thrombin, not intermediates of prothrombin activation, since it disappears after addition of heparin, which promotes thrombin antithrombin III complex formation. After a plateau of 2-10 min, there is further thrombin generation, which results in clotting after 15-27 min at a level of 40-50 ng thrombin/ml. The thrombin generated 9-25 min before clotting may activate factors V and VIII and stimulate platelet aggregation and release. In contrast, the cascade hypothesis assigns a role for thrombin only late in blood clotting. Radioimmunoassay of thrombin and other clotting factors will be useful for clinical and physiological studies of blood clotting especially since the assay seems specific for thrombin and is independent of other activities that affect bioassays.

The perturbation of thrombin binding to human platelets by anions.

Thrombin binds with high affinity to specific cell-surface receptors on washed human platelets. We present experiments indicating that thrombin binding correlates withe the release reaction when binding is perturbed by anions. Marked differences in the affinity of human 125I-thrombin for platelets wer observed in various isotonic buffers at pH 7.4. At low concentrations of thrombin (0.001-0.01 U/ml), binding was 5-fold greater in Tris-sodium acetate and 12-fold greater in Tris-sodium cacodylate than in Tris-sodium chloride. These anion-induced changes in 125I-thrombin binding paralleled changes in [14C] serotonin release when both parameters were measured in the same platelets. Thus, equivalent release occurred for equal amounts of thrombin bound in all buffers, even though the thrombin concentration varied by up to 30-fold. After approximately 100 molecules of thrombin bound per platelet, complete release occurred in all buffers in 2 min. The effect of anions was specific for the thrombin-receptor interaction as there was no corresponding effect on the binding of erythroagglutinating phytohemagglutinin (E-PHA) to platelets nor on E-PHA or collagen-induced serotonin release. The various anions did not alter platelet morphology as judged by electron microscopy. The anions had no effect on thrombin esterase catalytic activity. In addition, the total number of thrombin receptors per platelet was approximately the same in all buffers. Thus anions alter the affinity between platelet thrombin receptors and a site on thrombin distinct from the catalytic site. We conclude that the thrombin receptor is essential for thrombin-induced platelet reactions.

Incorporation of intravenously injected albumin, immunoglobulin G, and fibrinogen in guinea pig megakaryocyte granules.

In a previous study we provide evidence for a circuitous pathway by which circulating plasma proteins enter megakaryocyte granules by an endocytic mechanism and are returned to the circulation in platelets (1987. Proc. Natl. Acad. Sci. USA. 84:861-865). Horseradish peroxidase (40,000 mol wt) was injected into guinea pigs and its uptake into megakaryocyte organelles examined by electron microscopy and cytochemistry. In the present study we tested the ability of guinea pig megakaryocytes to take up intravenously injected albumin, IgG, and fibrinogen. We used two types of proteins to study the endocytic pathway: (a) heterologous human proteins, which were detected immunohistochemically using antibodies that do not crossreact with the native guinea pig counterparts; and (b) human and guinea pig proteins labeled with the small (250 mol wt), inert molecule, biotin, which were detected using an antibody against biotin. We detected all three of the injected proteins in bone marrow megakaryocytes in patterns identical to those of native counterparts. The injected protein consistently appeared in platelets 24 h later and was secreted in response to thrombin. We conclude that there are at least two mechanisms by which guinea pig megakaryocyte granules acquire proteins (a) endogenous synthesis, as demonstrated by others, and (b) endocytosis of plasma proteins synthesized by other types of cells.

Thrombin-induced platelet secretion. Further evidence for a specific pathway.

We have studied the interaction between thrombin and washed, human platelets using prostacyclin, a reversible inhibitor of platelet secretion. The effect of thrombin is limited to those reactions that are not inhibited by an increased concentration of platelet cyclic adenosine 3',5'-monophosphate, because prostacyclin is a potent inducer of the latter. Prostacyclin-treated platelets were briefly (15-30 s) exposed to low concentrations of human thrombin (0.01-0.2 U/ml). After removal of the prostacyclin and thrombin, the platelets were incubated with fresh thrombin. Although they had not undergone the release reaction after the first thrombin incubation, these platelets had a diminished capacity to secrete [(3)H]serotonin when exposed to thrombin the second time. Refractoriness was concentration dependent: the higher the initial thrombin concentration, the greater the degree of inhibition of serotonin secretion on subsequent thrombin exposure. Inhibition was closely related to the ability of thrombin to induce platelet secretion and not to its esterase or fibrinogen clotting activity. Diisopropyl fluorophosphate-inactive thrombin did not induce refractoriness. Refractoriness to thrombin did not increase when the time of the initial incubation with thrombin was lengthened, nor was it reversible.INHIBITION WAS THROMBIN SPECIFIC: serotonin secretion induced by collagen, wheat germ agglutinin, and the ionophore A23187 was minimally affected. For an equivalent amount of thrombin bound, a decrease was observed in serotonin secretion by thrombin-pretreated platelets compared to control platelets. Thus, there is at least one step in the secretory pathway between thrombin binding and regulation of adenylate cyclase. This step appears to transmit the signal that leads to extrusion of intracellular granular contents.

Evidence that a 210,000-molecular-weight glycoprotein (GP 210) serves as a platelet Fc receptor.

We previously identified a 210,000-mol-wt platelet glycoprotein (GP 210) that is missing from Bernard-Soulier platelets, and found that an antibody against GP 210 inhibits ristocetin-induced platelet agglutination. We now show by immunoblotting that GP 210 binds heat-aggregated rabbit and human IgG, as well as keyhole limpet hemocyanin (KLH)-anti-KLH and ovalbumin (OA)-anti-OA immune complexes. Immune complex binding to GP 210 was preserved on chymotrypsin-treated platelets that lacked glycoprotein Ib (GP Ib). In contrast, ristocetin-induced platelet agglutination resulted in disappearance of immunologically detectable GP 210 and loss of immune complex binding, even though GP Ib remained intact. Purified Fc fragments inhibited binding of anti-GP 210 antibody to intact platelets and to GP 210 on immunoblots. The Fc fragments also blocked immune complex binding to GP 210. Conversely, anti-GP 210 antiserum and F(ab)2 fragments inhibited binding of fluorescein-labeled Fc fragments to intact platelets. We conclude that GP 210 functions as a platelet Fc receptor.

Acquired Bernard-Soulier syndrome. Evidence for the role of a 210,000-molecular weight protein in the interaction of platelets with von Willebrand factor.

A patient with a lymphoproliferative disorder developed bleeding associated with a prolonged bleeding time and a selective defect of platelet aggregation in response to ristocetin. The patient's purified IgG was shown to inhibit aggregation of washed normal platelets by ristocetin and von Willebrand factor (F VIII:vWF). By Western blotting, it was shown that antibody bound specifically to an antigen of Mr 210,000 present on normal platelets but missing on platelets from patients with congenital Bernard-Soulier syndrome (BSS). Binding was effected by the F(ab)2 portion of the IgG, indicating the presence of an autoantibody rather than an immune complex. These results suggest that the 210,000-Mr protein is involved in the interaction of F VIII:vWF with platelets. Furthermore, we have demonstrated the apparent absence of an additional protein on congenital BSS platelets. Heat-aggregated IgG was also shown to bind to the 210,000-Mr protein, suggesting that this protein may function as an Fc receptor on platelets. The relationship of the 210,000-Mr protein to glycoprotein Ib and the precise role of this protein in the interaction of platelets with F VIII:vWF need to be characterized.

Cleavage of blood coagulation factor XIII and fibrinogen by thrombin during in vitro clotting.

Thrombin cleavage of blood coagulation Factor XIII (a2b2) and fibrinogen was studied during in vitro clotting to determine the physiologic sequence of these events. First, the time course of fibrin formation and cleavage of Factor XIII was measured in platelet-rich plasma. Cleavage of fibrinogen was measured by using a radioimmunoassay for fibrinopeptide A. Conversion of trace amounts of radioiodinated a-chains of 125I-Factor XIII to thrombin-modified a-chains was measured in unreduced 10% sodium dodecyl sulfate-polyacrylamide gels. During spontaneous clotting, a similar percentage of 125I-Factor XIII and fibrinogen was cleaved at each time point. Visible gelation of polymerized fibrin monomer occurred when 24 +/- 8% of fibrinogen was cleaved and 21 +/- 6% of Factor XIII was converted to Factor XIII'. Thrombin cleavage of Factor XIII and fibrinogen was also studied in platelet-poor plasma to which thrombin was added. In order to measure Factor XIIIa activity, fibrin polymerization was completely inhibited by the addition of Gly-Pro-Arg-Pro. Factor XIIIa formation was measured by the incorporation of [3H]putrescine into casein. The concentration of added thrombin required to cleave 50% of fibrinogen and Factor XIII was 0.65 U/ml and 0.35 U/ml, respectively. The rate of cleavage of fibrinogen by thrombin was 43-fold greater than cleavage of Factor XIII. Lower Gly-Pro-Arg-Pro concentrations were used to determine the effects of incompletely inhibiting fibrin polymerization on cleavage of Factor XIII and fibrinogen. Thrombin cleavage of Factor XIII but not fibrinogen was dependent on the extent of fibrin polymerization. The more marked the degree of inhibition of fibrin polymerization, the slower the rate of Factor XIIIa formation. Thus, in platelet-rich plasma, thrombin cleavage of Factor XIII and fibrinogen are closely related events during spontaneous clotting. Furthermore, cleavage of Factor XIII during clotting is enhanced by fibrin polymerization in platelet-poor plasma.

Localization of the gene for coagulation factor XIII a-chain to chromosome 6 and identification of sites of synthesis.

Factor XIII, the clotting factor essential for covalent stabilization of the fibrin clot, is a heterodimer consisting of a2 and b2 subunits, with catalytic function residing in the a-chain. In order to address questions regarding sites of synthesis and chromosomal localization of the Factor XIII a-chain, cDNA was cloned from a lambda gt11 human placental cDNA library. Nucleotide and amino acid sequences were determined from the cDNA. Amino acid sequencing of purified platelet Factor XIII a-chains confirmed the authenticity of the lambda gt11 clone. The gene for Factor XIII a-chain was mapped uniquely to chromosome 6. Northern blot analysis of human placental and U937 (monocytelike) cell poly (A)+ mRNA showed a single approximately 4.0-kb message for the Factor XIII a-chain. These results provide conclusive evidence that the a-chain is synthesized by placenta and monocyte cell lines.

Kistrin, an integrin antagonist, blocks endocytosis of fibrinogen into guinea pig megakaryocyte and platelet alpha-granules.

Recent data indicate that megakaryocyte/platelet alpha-granule fibrinogen is endocytosed from plasma. Because fibrinogen is the major platelet protein present in high concentrations in alpha-granules, fibrinogen uptake into alpha-granules may occur via specific receptors. In that cells of the megakaryocyte/platelet lineage contain two integrins--alpha IIb beta 3 (GP IIb-IIIa) and the vitronectin receptor (alpha v beta 3)--that can bind fibrinogen, one or both of these receptors may mediate the endocytic uptake of fibrinogen. To test this hypothesis, we examined the effect of Kistrin, an RGD-containing protein purified from the venom of Agkistrodon rhodostoma that inhibits fibrinogen binding to human platelet receptors, on endocytosis of fibrinogen by megakaryocytes and platelets. Continuous intravenous infusion of kistrin into guinea pigs (200 micrograms/h) over a 24-h period inhibited collagen-induced platelet aggregation. When biotinylated fibrinogen was injected intravenously into animals receiving Kistrin, megakaryocytes failed to endocytose the labeled fibrinogen. Endocytosis of fibrinogen into platelets was also inhibited in these animals. In contrast, platelets and megakaryocytes obtained from sham-infused control animals contained the injected biotinylated fibrinogen. We conclude that, in addition to the well-known extracellular function of cell adhesion, integrins can also act as receptors that mediate endocytosis of exogenous proteins and incorporate them into regulated secretory granules.

Platelet glycoprotein IIb. Chromosomal localization and tissue expression.

The GPIIb-IIIa complex functions as a receptor for cytoadhesive proteins on the platelet surface. Both GPIIb and GPIIIa are synthesized by a human erythroleukemia (HEL) cell line. We isolated several cDNA clones by screening a HEL cell cDNA library with an oligonucleotide derived from amino acid sequence of GPIIb. Nucleotide and amino acid sequences were determined from 703 bp of one of these clones. Amino acid sequence of purified platelet GPIIb peptides confirmed the identity of the clone. The cDNA encodes the carboxyl terminus of the large (alpha) subunit of GPIIb and all of the smaller (beta) subunit of GPIIb. By hybridizing the cDNA directly to chromosomes separated by dual laser chromosome sorting, the gene for GPIIb was mapped to chromosome 17. Northern blot analysis showed a approximately 3.4-kb GPIIb mRNA in HEL cells. We also compared the amino acid sequences determined from eight additional platelet GPIIb peptides with the derived amino acids from a published HEL cell GPIIb cDNA, and the platelet and HEL cell proteins appear to be the same. Despite previous reports that vascular endothelial cells and monocytes contain GPIIb, no GPIIb mRNA was observed in either type of cell. Thus, GPIIb appears to be specific for the platelet-megakaryocyte membrane and is distinct from the alpha subunits of the adhesion receptors in other normal tissues.

Vascular endothelial growth factor promotes tumor dissemination by a mechanism distinct from its effect on primary tumor growth.

Tumor growth is dependent on new blood vessel formation. Inhibition of vascular endothelial growth factor (VEGF), an endothelial cell mitogen and angiogenic factor secreted by a variety of tumors and tumor cell lines, is sufficient to inhibit primary tumor growth. In the present study, we examined the effect of inhibiting VEGF on tumor cell micrometastasis. A transfectant of A431 (a human epidermoid carcinoma cell line) expressing chloramphenicol acetyltransferase (CAT) was injected s.c. into severe combined immunodeficiency (scid) mice, which were then sacrificed after 6 weeks. The presence of A431 metastases at distant sites was demonstrated by detection of CAT activity in whole-organ lysates. Treatment of animals with VEGF-neutralizing antibodies not only inhibited primary tumor growth but also suppressed metastases, as determined by CAT activity in organ lysates. In experiments to determine the mechanism by which anti-VEGF antibody inhibited metastasis, control animals were sacrificed when their tumors had reached the same size as tumors in VEGF antibody-treated animals. Metastases were uniformly present in these control animals. These findings show that inhibition of VEGF alone is sufficient to prevent tumor growth and dissemination in vivo. The inhibitory effect on metastases appears to be distinct from that on primary tumor growth.

Inhibition of prostate cancer neovascularization and growth by urokinase-plasminogen activator receptor blockade.

Binding of the serine protease urokinase (u-PA) to its receptor on tumor cell surfaces facilitates proteolysis and tumor invasion. We undertook this study to determine whether the role of u-PA in prostate cancer induced angiogenesis and secondary tumor growth by developing a homologous, immunocompetent in vivo model in which the tumors cells secrete an inhibitor of the murine u-PA receptor. A mutant recombinant murine u-PA that retains receptor binding but not proteolytic activity was made by PCR mutagenesis. Mutant u-PA and a reporter gene pRK luciferase were transfected and stably expressed in the highly metastatic rat Dunning MAT-LyLu prostate cancer cell line. Several clones expressing mutant u-PA and luciferase were identified by Western blotting, plasminogen zymography, and reverse transcription-PCR. One of these clones, 5C4, was injected s.c. into Copenhagen rats. Compared to animals injected with clones expressing pRK luciferase alone, tumors in animals injected with 5C4 cells were significantly smaller. Moreover, there were fewer lung micrometastases in the 5C4 animals. Primary tumor angiogenesis was measured by microvessel quantification of tissue stained with antibodies against von Willebrand factor. Mean microvessel density in 5C4 tumors was 4.3-fold lower than that in animals with tumors derived from the control tumor cell line (P < 0.0001). Significant inhibition of tumor growth was also observed for two additional MAT-LyLu cell lines expressing mutant u-PA. These findings suggest that cell surface u-PA contributes to prostate cancer growth by enhancing angiogenesis.

Vascular endothelial growth factor and basic fibroblast growth factor urine levels as predictors of outcome in hormone-refractory prostate cancer patients: a cancer and leukemia group B study.

Better prognostic markers are needed for hormone-refractory prostate cancer (HRPC) patients. No single biochemical or clinical parameter can reliably predict patient response to therapy or rapidity of disease progression. Peptide factors involved in major cancer growth pathways, such as tumor angiogenesis, are attractive candidates as markers of low- and high-risk HRPC patients. We analyzed prospectively collected urine specimens from 100 of 390 HRPC patients undergoing therapy with the growth factor antagonist suramin as part of CALGB 9480. Levels of vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF) were assessed from day 1 of therapy (D1) and day 29 (D29) urine samples from this subset of 100 randomly selected patients. Growth factor levels were determined by standardized ELISA microtiter plate assays from a commercial (bFGF) or proprietary (VEGF) source. Pretreatment urine VEGF levels were predictive of survival. In univariate analysis, patients whose baseline urine VEGF level was < or =28 pg/ml (the median level) had an average survival of 17 months; those with baseline VEGF >28 pg/ml had a significantly shorter survival of 10 months (P = 0.024). This difference corresponded to a 60% increased risk of dying for the higher urine VEGF patients (hazard ratio, 1.62; P = 0.03) and remained significant in multivariate analysis (hazard ratio, 1.72, P = 0.02). No significant correlations between urine bFGF level or change in bFGF levels and survival were found. These results support the notion that certain peptide growth factor-mediated, mitogenic pathways are important in HRPC and that their levels can predict outcome.