Tissue factor is the receptor for plasminogen type 1 on 1-LN human prostate cancer cells

Tissue factor (TF), a 45-kd transmembrane glycoprotein, initiates the extrinsic pathway of coagulation through the formation of an enzymatic complex with factor VII/VIIa.1,2 TF also binds to and promotes plasminogen (Pg) activation to plasmin (Pm) through a site distinct from the binding site for factor VIIa.3 TF is involved in angiogenesis,4 cell migration,5 and tumor cell invasion.6 Pg/Pm stimulates tumor cell invasion7,8; however, their elastase-derived fragments act as antagonists of this process.10,11 The diverging roles of Pg/Pm or its fragments suggest a complex network of receptors regulating Pg binding and activation on the cell surface. In the circulation, human Pg exists as 2 glycoforms, Pg type 1 (Pg 1) and Pg type 2 (Pg 2).12 Pg 1 contains one O-glycan (Thr₃₄₅) and one biantennary N-glycan (Asn₂₈₈), whereas Pg 2 contains only the O-glycan chain at Thr₃₄₅.13,14 In previous studies, we found that different cell types bind Pg 2 preferentially.15,16We have characterized and identified the membrane glycoprotein dipeptidyl peptidase IV (CD26) as the receptor for Pg 2 in the highly invasive 1-LN human prostate tumor cell line.17Furthermore, we also demonstrated that Pg 2 is involved in regulating the expression of matrix metalloproteinase-9 (MMP-9) by these cells.17 

In the current study, we demonstrate that Pg 1 binds to TF on the surfaces of 1-LN cells and to TF isolated from 1-LN cell membranes. Pg 1 binding to TF is mediated by l-lysine binding sites and the d-mannosyl residues of the biantennary N-glycan linked to Asn₂₈₈. We also show that binding of Pg 1 to TF on 1-LN cells induces a rise in cytosolic-free Ca⁺⁺; however, this signaling cascade does not stimulate the expression of MMP-9 or enhance 1-LN cell invasiveness of the synthetic basement membrane, Matrigel, as observed when Pg 2 binds to the same cell line.17 These studies suggest that Pg 1 and Pg 2 bind independently to separate receptors on the surfaces of 1-LN cells and may have different physiological functions.

Culture media were purchased from Life Technologies (Gaithersburg, MD). Fura 2 acetoxymethyl ester (fura-2 am) was obtained from Molecular Probes (Eugene, OR). The chromogenic plasmin substrate Val-Leu-Lys-p-nitroanilide (S-2251) was purchased from Sigma Chemical (St Louis, MO). The homobifunctional cross-linker 3, 3′-Dithiobis (sulfosuccinimidylpropionate) (DTSSP) was purchased from Pierce (Rockford, IL). Other reagents used were of the highest grade available.

Goat anti-human Pg IgG was provided by Dr Kenneth C. Robbins (Northwestern University, Chicago, IL). Goat anti-human tissue factor, anti-human urinary-type Pg activator (u-PA), and anti-human t-PA IgG were purchased from American Diagnostica (Greenwich, CT). The anti–MMP-9 monoclonal antibody (mAb), clone 6-6B, was purchased from Oncogene Research Products (Cambridge, MA). Anti-human Pg or anti-human TF F(ab')₂ fragments were prepared by digestion of goat anti-human Pg or anti-human TF IgG (1 mg) with pepsin (0.1 mg/mL) in 100 mM sodium citrate, pH 3.5, for 18 hours at 37°C. These fragments were separated from any remaining intact IgG or Fc fragments by chromatography on protein A–Sepharose.18 Alkaline phosphatase–conjugated rabbit F(ab')₂ with a specificity for goat F(ab')₂ fragments was purchased from ICN-Cappel (Aurora, OH).

Pg 1 and Pg 2 were purified from human plasma using a combination of Lys-Sepharose and concanavalin A–Sepharose affinity chromatographies, as previously described.17Radioiodination was carried out by the method of Markwell.19 Radioactivity was measured in a Pharmacia LKB Biotechnology 1272-gamma counter (Pharmacia, Rockville, MD). Incorporation of sodium iodide I 125 was at a level of approximately 8 × 10⁶ cpm/nmol protein. ¹²⁵I-labeled Pg was purified by affinity chromatography onl-lysine–Sepharose and then used for the binding experiments.

1-LN cell membranes were prepared as previously described.17 TF was sequentially purified to homogeneity using chromatography on concanavalin A-Sepharose,20followed by immunoaffinity chromatography with a goat anti-human TF IgG covalently attached to Sepharose.21 These steps yielded approximately 200 μg TF (approximately 50 μg/10⁹cells).

The human prostate tumor cell line 1-LN was grown in RPMI 1640 supplemented with 10% (vol/vol) fetal bovine serum, 100 U/mL penicillin G, and 100 ng/mL streptomycin as previously described.17 These cells were the kind gift of Dr Philip Walther (Division of Urology, Department of Surgery, Duke University Medical Center).

Cells were grown in tissue culture plates until the monolayers were confluent. Before use in binding assays, the cells were washed in Hanks balanced salt solution. All binding assays were performed at 4°C in RPMI 1640 containing 2% (wt/vol) bovine serum albumin (BSA). Increasing concentrations of ¹²⁵I-labeled Pg were incubated with cells for 60 minutes in 48-well culture plates. Free ligand was separated from bound ligand by aspiration, followed by rinsing 3 times the cell monolayers with RPMI 1640 containing 2% BSA. The cells were then lysed with 0.1 M NaOH, and bound radioactivity was determined in a Pharmacia LKB Biotechnology 1272-gamma counter. Molecules of ligand bound were calculated after subtraction of nonspecific binding measured in the presence of 100 μM unlabeled Pg. Estimates of dissociation constant (K_d) values and maximal binding (B_max) were determined by fitting data directly to the Langmuir isotherm using the statistical program SYStat for Windows.

To study specific binding of Pg 1 and Pg 2 glycoforms to immobilized TF isolated from 1-LN cell membranes, 96-well strip plates were coated with 200 μL of a solution containing TF (1 μg/mL) in 0.1 M sodium carbonate, pH 9.6, at 37°C for 2 hours. After coating, plates were extensively washed with 10 mM sodium phosphate, 100 mM NaCl, pH 7.4, containing 0.05% Tween-80 (phosphate-buffered salt solution [PBS]–Tween) to remove unbound protein. Nonspecific sites were blocked by incubation with PBS-Tween containing 2% (wt/vol) BSA at room temperature for 1 hour. Plates were rinsed twice with 200 μL PBS-Tween, air dried, and stored at 4°C. The amount of TF immobilized on the plates (100 ng/well) was determined using an enzyme-linked immunosorbent assay procedure as follows: increasing amounts of goat anti-human TF F(ab')₂ fragments were incubated in TF-containing wells in triplicate at 37°C for 1 hour, followed by the addition of an anti-goat alkaline phosphatase–conjugated rabbit F(ab')₂ fragment (1:400 dilution). A calibration curve was constructed by plotting the amount of anti-TF F(ab')₂fragment versus the rate of hydrolysis of the alkaline phosphatase substrate p-nitrophenyl phosphate (2 mM in 50 mM Tris-HCl, pH 9.0) at a wavelength of 405_nm (rA_{405 nm}/min) using an Anthos Labtec kinetic plate reader. For Pg binding assays, increasing concentrations of ¹²⁵I-labeled Pg 1 and Pg 2 glycoforms, with or without a 50-fold excess of unlabeled ligand, were added to triplicate TF-containing wells and incubated at 37°C for 1 hour. Following incubation, the supernatants were removed, and the plates were rinsed 3 times with 200 μL PBS-Tween. Wells were stripped from the plates, and radioactivity was measured. Specific binding was calculated by subtraction of nonspecific binding measured in the presence of unlabeled ligand.

Pm generation was measured on confluent 1-LN cell monolayers grown in 96-well plates. Cells were incubated in serum-free RPMI 1640 medium, without phenol red, with Pg for 30 minutes at 37°C. Excess Pg was removed, the cells were rinsed with serum-free RPMI 1640, and 200 μL of the Pm substratel-valine-l-lysine-l-glycine-p-nitroanilide (S-2251) at a final concentration of 0.3 mM in RPMI 1640 was added to the wells. The hydrolysis of the substrate was monitored at a wavelength of 405 nm using an Anthos Labtec kinetic plate reader. Pm activity was expressed as ΔA₄₀₅ nm/min. When inhibitory antibodies were used, the cells were incubated with the antibody for 30 minutes before Pg was added.

Cytosolic-free Ca⁺⁺,[Ca⁺⁺]_i, was measured by digital imaging microscopy using the fluorescent indicator fura-2am as previously described.16 

Electrophoresis was performed on polyacrylamide gels (1.2-mm thick; 14 × 10 cm) containing 0.1% sodium dodecyl sulfate (SDS). A discontinuous Laemmli buffer system was used.22 Transfer to nitrocellulose membranes was carried out by the Western blot method.23 Electroblots were reacted with anti–MMP-9 mAb, purified goat anti-human Pg, or anti-TF F(ab')₂ fragments. The dye-conjugated M_r markers (Bio-Rad, Richmond, CA) used were myosin (M_r = 218 000), β-galactosidase (M_r = 134 000), BSA (M_r = 86 000), carbonic anhydrase (M_r = 44 000), and soybean trypsin inhibitor (M_r = 32 000).

Protein samples were electrophoresed on gelatin-containing 0.75-mm-thick 10% (wt/vol) polyacrylamide gels in the presence of SDS under nonreducing conditions.24 After completion of the electrophoretic run, the gels were incubated with 2 changes of 2.5% (vol/vol) Triton X-100 for 1 hour, followed by incubation for 18 hours at 37°C in 0.1 M glycine–NaOH, pH 8.3, containing 1 mM CaCl₂ and 0.1 M ZnCl₂, before staining with 0.25% Coomassie Brilliant Blue R-250 in 45% (vol/vol) methanol–10% (vol/vol) acetic acid to visualize the lysis bands.

Cross-linking of Pg 1 to 1-LN cell membrane proteins was performed by incubating first 1-mL cell suspensions (at a cell density of 5 × 10⁶ cells/mL) in serum-free RPMI 1640 culture medium with Pg 1 (1 μM) at 37°C for 20 minutes. Excess ligand was removed by centrifugation. After rinsing the cells with serum-free RPMI 1640, they were resuspended in 1 mL ice-cold PBS, and the homobifunctional cross-linker DTSSP was added at a final concentration of 10 mM. The cross-linking reaction was conducted on ice for 60 minutes and terminated by the addition of 1/10 volume 1 M glycine, pH 7.5, and 1/10 volume of 1 M Tris-HCl, pH 7.5. The cells were then lysed with 100 mM N-octylglucopyranoside in 50 mM Tris-HCl, pH 7.5. After mixing, the samples were incubated at room temperature for 30 minutes to extract the membrane proteins. Cell debris was removed by centrifugation at 100 000g for 30 minutes at 4°C. Supernatants containing protein cross-linked to TF were purified by immunoaffinity with anti-human Pg goat IgG covalently attached to Sepharose. After elution with 1 M guanidine-HCl in 50 mM Tris-HCl, pH 8.0, and extensive dialysis against 50 mM Tris-HCl, pH 8.0, the protein solution was concentrated to 0.5 mL with an Amicon concentration cell. Cross-linked proteins were resolved by electrophoresis in 10% SDS–polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions, which breaks the DTSSP bridge between Pg 1 and TF. After separation, the proteins are electroblotted to a nitrocellulose membrane and individually probed with goat anti-human Pg or anti-human TF F(ab')₂ fragments, followed by reaction with specific rabbit anti-goat alkaline phosphatase-conjugated F(ab')₂ fragments.

Invasive activity in vitro was assessed by determining the ability of 1-LN cells to penetrate the synthetic basement membrane Matrigel.25 Polycarbonate filters (8 μm pore size; Becton Dickinson, Bedford, MA) were coated with Matrigel (12 μg/filter) and placed in a modified Boyden Chamber. Cells (1 × 10⁵) were added to the upper chamber in serum-free RPMI 1640 medium or in medium containing purified Pg 1 or Pg 2 or a combination of both glycoforms and were incubated at 37°C for 24 hours in a humidified atmosphere. Following incubation, noninvading cells were removed from the upper chamber with a cotton swab, and filters were excised and stained with Cyto-Quik (Fisher Scientific, Fair Lawn, NJ). Cells on the lower surface of the filter were enumerated using an ocular micrometer; at least 5 high-powered fields were counted. Each experiment was performed twice with triplicate samples.

¹²⁵I-labeled Pg 1 binds to 1-LN cells in a dose-dependent manner (Figure 1A). Analysis of the ligand binding isotherm shows that Pg 1 binds to these cells with high affinity (K_d of 15 nM) and to a large number of sites (7.5 × 10⁵ binding sites/cell). Because TF was previously identified as a Pg receptor,3 we purified TF from 1-LN cells as described in “Materials and methods.” Coomassie Brilliant Blue-R-250 stain of the electrophoresed material (Figure 1B, inset, lane 1) demonstrates a major protein band in the M_r approximately 44 000 size range. A blot binding assay with a goat anti-human TF F(ab')₂ fragment showed reactivity only with the M_r approximately 44 000 protein band (Figure 1B, inset, lane 2)). Binding studies of¹²⁵I-labeled Pg 1 and Pg 2 to TF, immobilized on cell culture plates, demonstrate that only Pg 1 binds to TF in a dose-dependent and saturable manner (Figure 1B). Little specific binding was observed with Pg 2. Analysis of the binding isotherm shows that ¹²⁵I-labeled Pg 1 binds to TF with high affinity (K_d of 17 nM) and maximal binding (B_max of approximately 0.8 nmol Pg 1/nmol TF). The close correspondence in affinity between the kinetic parameters of Pg 1 binding to 1-LN cells and purified TF suggests this molecule as a unique receptor for Pg 1 on the 1-LN cell surface. Binding of¹²⁵I-labeled Pg 1 to TF was progressively inhibited by nonlabeled Pg 1, whereas nonlabeled Pg 2 did not affect its binding (Figure 2A). Anti-TF F(ab')₂fragments, 6-aminohexanoic acid (6-AHA) and α-methyl-d-mannopyranoside (αMGP) were also able to inhibit binding of Pg 1 to TF, whereas factor VIIa did not produce any effect (Figure 2B). Similar results (data not shown) were obtained after binding of Pg 1, in the presence of these inhibitors, to 1-LN cells. These observations not only suggest that Pg 1 binds to a site in TF distinct from that in which factor VIIa binds, it suggests that the interaction is also highly specific involving its l-lysine binding sites and the d-mannosyl residues of the biantennary N-glycan linked to Asn_288. To further determine the identity of the receptor for Pg 1, 1-LN cells were treated with Pg 1 and cross-linked to cell-surface proteins with DTSSP, followed by immunoaffinity chromatography specific for Pg, as described in “Materials and methods.” Electrophoretic separation is performed under reducing conditions, which breaks the DTSSP bridge between Pg 1 and cell surface proteins, allowing each protein to be resolved separately in the gel, before analysis by the Western blot method. Individual analysis of the proteins cross-linked to Pg with anti-Pg F(ab')₂ fragments shows protein bands corresponding to Pg/Pm (Figure 3, lane 1). The same materials show a protein corresponding to TF when the blot is individually reacted with anti-TF F(ab')₂ fragments (Figure 3, lane 5). These results further suggest TF as the receptor for PG 1 on the surfaces of 1-LN cells.

Pg 1 is rapidly converted to Pm on the surfaces of 1-LN cells (Figure4). Preincubation of the cells with anti-TF F(ab')₂ fragments and anti-uPA or anti-tPA inhibitory IgG demonstrates that only anti-uPA or anti-TF antibodies effectively inhibit Pg 1 activation (Figure 4), suggesting uPA as the sole Pg 1 activator when this Pg glycoform is bound to TF on the surfaces of these cells. The specificity of the plasminogen glycoform bound to TF on 1-LN cells and activation by uPA was further analyzed with anti-human TF F(ab')₂ fragments and both Pg glycoforms (Figure 5A). These experiments show that increasing amounts of the anti-TF antibody inhibit binding and activation of only Pg 1 in a dose-dependent manner, whereas binding and activation of Pg 2 to these cells are not affected by the anti-TF antibody. Pg 1 activation by uPA, in the presence of increasing concentrations of isolated 1-LN cell TF, was also assessed. The results (Figure 5B) clearly demonstrate that TF stimulates Pg 1 activation by uPA in a dose-dependent manner.

We measured changes in cytosolic-free Ca⁺⁺ after the addition of Pg 1 to 1-LN cells preincubated in serum-free RPMI 1640 culture medium. Addition of Pg 1 at 90 seconds induced a rise in Ca⁺⁺ that lasted for 120 seconds before returning to baseline (Figure 6A). Addition of Pg 2 also induced a rise in Ca⁺⁺ and did not prevent a second response by the cells when Pg 1 was added to the system (Figure 6B). Preincubation of the cells with an anti-TF F(ab')₂fragments and the addition of Pg 1 inhibited the response of 1-LN cells to this ligand (Figure 6C), whereas the response of 1-LN cells to Pg 2 was not inhibited by this antibody (Figure 6D). These results further confirm that TF is the receptor for Pg 1 on the surfaces of 1-LN cells.

Cells were seeded into 48-well culture plates and grown in RPMI 1640 medium containing 10% (vol/vol) fetal bovine serum. Before the experiment, confluent cell monolayers were incubated for 24 hours with quiescent culture medium containing RPMI 1640 and 0.5% (vol/vol) fetal bovine serum. Pg 1 and Pg 2 glycoforms (0.1 μM) were added in triplicate to cell monolayers in 300 μL serum-free RPMI 1640 and were incubated for 24 hours at 37°C. Culture medium was collected to measure secretion of MMP-9. Gelatin zymography (Figure7A) shows a major band of active protein of M_r approximately 85 000 in medium from cells incubated with Pg 2 (Figure 7A, lane 2) and decreased bands of activity, in the same size range, in media from cells incubated with Pg 1 and serum-free medium alone (Figure 7A, lanes 1 and 3, respectively). Samples of these conditioned media were also subjected to SDS-PAGE under reducing conditions, electroblotted onto nitrocellulose membranes, and treated with an anti–MMP-9 mAb (Figure 7B). This study also shows an increase in the secretion of MMP-9 in conditioned medium from cells incubated with Pg 2 (Figure 7B, lane 2) when compared with media from cells incubated with Pg 1 or serum-free medium alone (Figure 7B, lanes 1 and 3, respectively).

Incubation of 1-LN cells with Pg 2 enhances the ability of these cells to penetrate the synthetic basement membrane Matrigel. This effect of Pg 2 is mediated by a rise in cytosolic-free Ca⁺⁺, which leads to an enhancement in the expression of MMP-9.17 Because Pg 1 also induces a rise in cytosolic-free Ca⁺⁺, we tested its capacity to stimulate the invasiveness of 1-LN cells to Matrigel. Analysis of the data (Table1) shows that Pg 1 does not stimulate 1-LN cell invasiveness. Consistent with these data, we have previously shown that 1-LN invasiveness requires stimulation of MMP-9 secretion.17 Taken together, our results suggest that though both Pg 1 and Pg 2 bind to the surfaces of these cells, interaction of Pg 1 with TF on 1-LN cells supports fibrinolysis alone, whereas interaction of Pg 2 with CD26 is pivotal for the penetration of these cells into basement membranes.

Tissue factor, the 45-kd transmembrane protein activator of blood coagulation, has additional biologic functions, including the promotion of tumor angiogenesis4,26 and cell adhesion.27 A recent report suggested that TF binds Pg with high affinity,3 promoting the activation of Pg by u-PA.3 These additional functions may contribute to the roles that TF and Pg play in the enhancement of tumor growth and metastasis.7,8,28 TF promotes angiogenesis through a mechanism involving the formation of a complex with activated factor VII (FVIIa), which generates activated factor X (FXa), leading to the activation of thrombin that then stimulates the production of vascular endothelial growth factor.29 The physiologic function of Pg binding to TF is unclear; however, it is known that Pg binds to a region in TF distinct from that interacting with factor VII,3 and it involves Pg kringles 1-3,3 the Pg fragment with the greatest capacity to inhibit angiogenesis.9 Given these observations and given that Pg oligosaccharide chains are important for Pg binding to cellular receptors,17 we investigated the capacity of Pg 1 and Pg 2 glycoforms to bind to TF isolated from 1-LN cells. We found that only Pg 1 binds to this TF. Furthermore, Pg 1 also binds to cell surface–bound TF on 1-LN human prostate tumor cells with high affinity, and binding is not inhibited by Pg 2. Bound Pg 1 is rapidly converted to Pm by receptor-bound u-PA on the surfaces of 1-LN cells. Binding of Pg 1 to 1-LN cells induces a rise in cytosolic-free Ca⁺⁺; however, this signaling cascade does not influence secretion of MMP-9 or enhanced cell invasiveness to Matrigel, as observed when Pg 2 binds to CD26 in the same cells.17 

In previous studies, we found that Pg 2 was the preferred ligand for cellular receptors.15,16 The concentration of Pg 2 in the extravascular space is 6-fold greater than that of Pg 1,30suggesting Pg 2 as the major form for Pm formation during metabolism on the cell surface.31 However, because Pg 2 does not interfere with binding of Pg 1 to TF, our results suggest that Pm-1 may be generated even under a limited availability of this glycoform in the extravascular compartment. Our experiments also suggest that the biantennary N-glycan attached to Asn₂₈₈ plays an important role in this interaction.

Because the engagement of L-lysine binding sites enhances the activation of Pg 1 more than that of Pg 2 by their physiologic activators,32 we suggest that the binding and activation of Pg 1 linked to TF concentrates, on the cell surface, a Pg glycoform that otherwise is present in limited amounts. Paradoxically, TF has been reported to inhibit fibrinolysis.3 Because Pm lysine-binding sites are occupied by TF, the catalytic function of Pm can be inhibited by fibrinogen or fibrin,33 thereby excluding TF-linked Pm from fibrinolytic functions. However, under low concentrations of fibrinogen or fibrin, the generation of Pm from Pg 1 linked to TF may down-regulate the functions of TF. This hypothesis is supported by results suggesting that proteolysis by Pm inactivates tissue factor pathway inhibitor34 and coagulation factor X,35 both associated with the TF-dependent extrinsic pathway of coagulation. Because tissue factor pathway inhibitor associated with components of the extracellular matrix supports cell adhesion through a TF-VIIa complex,36 these actions of Pm 1 should also enhance the capacity of tumor cells to migrate and penetrate basal membranes in the cellular microenvironment.