Internalization of vitronectin-thrombin-antithrombin complex by endothelial cells leads to deposition of the complex into the subendothelial matrix.

Internalization of the ternary vitronectin-thrombin-antithrombin (VN-TAT) complex by human umbilical vein endothelial cells was investigated. Radiolabeled VN-TAT was bound to the cell surface at 4 degrees C, and internalization was initiated by increasing the temperature to 37 degrees C. After 30 min about half of the VN-TAT complex disappeared from the cell surface and accumulated in the subendothelial matrix. Translocation of VN-TAT complex from the luminal to the basolateral side was confirmed by electron microscopic evaluation of cross-sections of endothelial cells incubated with gold-conjugated VN-TAT complex. Furthermore, cells cultured in VN-TAT deficient serum, incubated with purified VN-TAT, and subsequently assayed for fluorescent staining using a monoclonal antibody directed against thrombin-modified antithrombin and a polyclonal antibody against vitronectin showed co-localization of both antibodies in punctates. Punctates were randomly distributed in both the xy and xz plane of endothelial cells as evidenced by confocal laser scanning microscopy. Trichloroacetic acid precipitation and SDS-polyacrylamide gel electrophoresis showed that VN-TAT was not degraded during translocation and inhibition of the microfilament system reduced release of VN-TAT to the matrix, indicating that transcytosis was responsible for translocation. These findings emphasize that VN-TAT complex is taken up by endothelial cells, not only leading to the removal of inactivated thrombin from the circulation but also to deposition of VN into the subendothelial matrix.

sulting in the formation of a ternary vitronectin-thrombinantithrombin (VN-TAT) complex. Upon complex formation the normally folded plasma form of VN is conformationally altered, leading to exposure of multiple domains (Tomasini and Mosher, 1988) such as a heparin-binding site and a collagen binding domain (Gebb et al., 1986). The exposure of the heparin binding domain has been found to be a prerequisite for some of the physiological properties ascribed to VN, such as binding and stabilization of plasminogen activator inhibitor (PAI-1) (Declerck et al., 1988;Wiman et al., 1988;Salonen et al., 1989) and scavenging and inactivation of the nascent C5b-9 complex of the complement cascade (Podack and Mü ller-Eberhard, 1979;Tschopp et al., 1988). Therefore the extended form of VN is considered to be an "activated" form of VN and TAT complex a physiological inducer of VN extension.
In a previous report we have shown that the heparin binding domain also mediates binding of VN-TAT to EC (de Boer et al., 1992). However, these experiments did not directly address the metabolic fate of VN-TAT complex bound to the endothelial cell surface. The present study provides evidence that following binding, VN-TAT is translocated through the EC and becomes deposited into the extracellular matrix. The mechanism described represents a route by which VN reaches the subendothelial matrix where it may be involved in a number of important physiological functions.

EXPERIMENTAL PROCEDURES
Materials-All chemicals obtained from commercial sources were of the highest grade available. Culture plastics (96-well plates containing 12 strips consisting of eight disconnectable wells and six-well plates) were purchased from Costar (Cambridge, MA). RPMI 1640 medium, penicillin/streptomycin, and fungizone were obtained from Gibco (Biocult, Paisley, United Kingdom). Primaquine, chloroquine, monensin, and colchicine were obtained from Sigma. Ammonium chloride was purchased from Baker (Deventer, The Netherlands) and cytochalasin B from Aldrich. Unfractionated heparin was obtained from Organon (Oss, The Netherlands).
Proteins and Antibodies-VN-TAT complex was purified from human serum as described earlier (de Boer et al., 1992). The polyclonal antibody against VN was raised in rabbits, and the IgG fraction was isolated using protein G affinity chromatography according to the instructions of the manufacturer (Pharmacia, Uppsala, Sweden). Monoclonal antibody ␣-AT mod directed against thrombin-modified antithrombin was kindly provided by Dr. H. Pannekoek (University of Amsterdam, The Netherlands). Fibronectin was a generous gift of Dr. J. A. van Mourik (Central Laboratory of Blood Transfusion, Amsterdam, The Netherlands).
Radiolabeling of VN-TAT-VN-TAT was labeled with [ 125 I]Na (Amersham, United Kingdom) using IODO-BEADS (Pierce) and separated from free [ 125 I]Na by gel filtration on Sephadex G-10 (Pharmacia), equilibrated with phosphate-buffered saline (PBS) containing 0.1% bovine serum albumin (BSA, Fraction V, A-7906, Sigma). The specific radioactivity was 3-6 Ci/g protein. The amount of free [ 125 I]Na was determined by trichloroacetic acid precipitation and accounted for less than 4%.
Cell Culture-Human umbilical vein endothelial cells (EC) were isolated from umbilical cords according to the method described by Jaffe et al. (1973). The culture medium contained 20% (v/v) normal human serum pool (of 20 healthy donors) in RPMI 1640 supplemented with antibiotics penicillin (100 units/ml), streptomycin (100 g/ml), and fungizone (4 g/ml).
Distribution Assay-Distribution studies were performed in Hepes buffer (10 mM Hepes, 137 mM NaCl, 4 mM KCl, 2 mM CaCl 2 , 15 mM glucose, 5 mg/ml BSA, pH 7.4). In all experiments specific binding is shown, defined as the difference in signal in the absence and presence of 100-fold molar excess of unlabeled VN-TAT complex. To each well was added 50.000 -100.000 cpm of radiolabel in 50 l, which accounted for 1.2-2.7 nM radiolabeled VN-TAT (dependent on the specific radioactivity of the radiolabel). Measurements were performed in triplicate on two different cell batches.
Time-dependent binding and internalization was studied over a time period of 30 min to 4 h. EC were cultured in strips containing eight disconnectable wells and were incubated with radiolabeled VN-TAT (10 ng/well, 1.25 nM) at 4 and 37°C. Following washes with cold Hepes buffer, cells were separated from the matrix by ammonia extraction (0.1 M) for 15 min at 4°C. Cell lysates were collected, and radioactivity was determined in a ␥-scintillation counter. The wells, containing the remaining extracellular matrix, were washed three times with cold Hepes buffer, disconnected, and counted. Distribution of radiolabel is shown as specific binding, expressed in nanograms of VN-TAT/10 6 cells.
To study the effect of heparin on internalization, 13 ng/well (1.65 nM) of radiolabeled ternary complex was incubated for 3 h at 4°C in the absence or presence of 100-fold molar excess of unlabeled VN-TAT or unfractionated heparin (0.01-100 units/ml) and label distribution was examined after ammonia extraction. Bound radiolabel was expressed as specific signal (nanograms of VN-TAT/10 6 cells). BSA-coated wells were used as control for binding of radioligand to plastic in the absence of EC.
To compare different extraction buffers, confluent EC cultured in eight-well strips were incubated at 4 or 37°C with 21.4 ng/well (2.7 nM) of 125 I-labeled VN-TAT in a total volume of 50 l of Hepes buffer. After 1 h the cells were washed three times with cold Hepes buffer to eliminate non-bound label. Removal of the cells was performed by incubation with ammonia (0.1 M in water, pH 11), EDTA (10 mM), or urea (2 M) in Hepes buffer (50 l/well, pH 7.4) for 15 min at 4°C. Cell lysates (in the case of ammonia) or cell suspensions (in the case of urea or EDTA) were collected, and radioactivity was determined in a ␥-scintillation counter. The wells, containing the isolated EC matrices, were washed three times and disconnected, and their radioactivity was measured in a ␥-scintillation counter. Results are expressed as percentage of total cell-associated signal.
Pulse-chase Assay-Monolayers of EC were incubated with radiolabeled VN-TAT (35 ng/well, 4.4 nM) at 4°C for 2 h and washed three times with cold Hepes buffer. This incubation period was designated time point 0. Medium containing non-labeled VN-TAT (40 ng/well) was added and the incubation temperature was raised to 37°C. At the time points indicated, medium (40 l) was removed and the cell monolayers were washed with cold Hepes buffer. Cells were either ammoniatreated and assayed as described above or incubated with 50 l/well trypsin/EDTA (Life Technologies, Inc.) containing proteinase K (Sigma) (concentrations 0.5 g/liter, 2 g/liter and 50 g/ml, respectively) for 30 min at 4°C as described by Chappell et al. (1992). Trypsin treatment detached the cells from the wells. Cell suspensions were transferred to microcentrifuge tubes with narrowed tip containing 100 l of sucrose (25%, w/v) and centrifuged for 2 min at 10,000 rpm. The cell pellet was obtained by cutting of the microcentrifuge tip. Radioactivity associated with the cell pellet and the supernatant fluid was measured separately. Surface-bound ligand was defined as ligand released from the cells by trypsin/EDTA/proteinase K treatment. Ligand not sensitive to enzymatic digestion represented internalized ligand.
Of the saved media, 40 l was mixed with 10 l of pasteurized protein solution and 100 l of trichloroacetic acid (final concentration 20%, w/v). Trichloroacetic acid-insoluble label was separated from trichloroacetic acid-soluble label by centrifugation and the pellet and supernatant were counted. Degradation was defined as the amount of radioactivity in the medium that was soluble in 20% trichloroacetic acid. Matrix-deposited label was defined as radioactivity remaining associated with the wells after removal of the cells either by ammonia or trypsin.
For SDS-PAGE analysis, cell lysates of four wells were obtained by ammonia extraction. Lysates were pooled to obtain a sufficient signal, the pH was neutralized and SDS-containing sample buffer (Laemmli, 1970) was added. The matrices were collected by scraping in sample buffer. The scraped wells were checked for radiolabel left behind. Proteins were separated on 4 -15% gradient acrylamide gels followed by autoradiography.
Binding of Radiolabeled VN-TAT to Endothelial Cell Matrix (ECM)-Endothelial cell matrix was preisolated by ammonia extraction (designated as preisolated matrices ϭ PI-ma) and used in a binding assay in comparison with monolayers of intact EC. Wells were incubated with radiolabeled VN-TAT (35 ng/well, 4.4 nM) for 1 h at 4°C in triplicate wells in the absence or presence of 100-fold molar excess of unlabeled VN-TAT. Monolayers of intact cells and PI-ma were washed three times, and radioactivity associated with the intact EC-monolayer was determined. Subsequently cells and PI-ma were treated with ammonia and radioactivity was measured in the ammonia extracts and in the wells as described under "Distribution Assay." Bound VN-TAT is shown as specific signal and expressed in cpm/well.
Phase-contrast or Electron Microscopic Evaluation-VN-TAT colloidal gold conjugates were prepared using purified VN-TAT complex and 17-nm gold particles essentially as described earlier (Völker et al., 1991). EC cultured in six-well plates were preincubated with 150 l of the VN-TAT-gold suspension in 1 ml of serum-free medium at 4°C for 1 h and washed three times with ice-cold serum-free medium. Nonspecific binding was determined by competition of binding of the conjugate by heparin (100 units/ml). Fixation was performed with 1% formaldehyde and 0.1% glutaraldehyde in PBS. Silver-enhancement of golddecorated EC was performed as described earlier (Völker et al., 1991), phase-contrast illuminated in an inverted microscope, and photographed. For uptake studies, cell cultures were incubated with goldconjugated VN-TAT at 4°C, washed as described above, and incubated for various periods of time at 37°C to allow uptake of surface-bound gold-conjugated VN-TAT complexes. Specimens were stained with 2% osmium tetroxide for overall visualization of intracellular structures. Ultrathin sections were investigated in a transmission electron microscope. The basolateral surface of cells was identified by its close proximity to the substrate, a thin coat of adhesive material attached to the bottom of the culture dish. Representative micrographs are shown.
Immunofluorescent Localization of VN-TAT-EC were cultured on glass coverslips in medium supplemented with VN-TAT-deficient serum (10% v/v), prepared by passing the serum over heparin-Sepharose. The amount of VN-TAT was checked by ELISA and accounted for less than 1%. At confluence, deficient cultures were incubated for 4 h at 37°C with deficient medium to which purified VN-TAT (5 g/ml) was added and washed with PBS. Cells were fixed in methanol (100%) for 10 min at 20°C and co-incubated with a monospecific polyclonal antibody directed against VN and a monoclonal antibody directed against thrombin-modified antithrombin (␣AT mod ). All antibodies were diluted in PBS containing 3% (w/v) BSA, and each step in this procedure was followed by extensive washing with PBS. Fluorescent label was applied to the first antibodies using IgG directed against mouse (FITC-conjugated) and rabbit (TRITC-conjugated) antibody (both from CLB, Amsterdam, The Netherlands). Fluorescence was analyzed using a confocal laser scanning microscope (Leitz, Heidelberg) equipped with an argon/krypton mixed gas laser and an oil-immersion objective (63 ϫ 1.4). Final images were merged using Photoshop software on an Apple computer and photographed directly from screen using a digital camera (Agfa).
Treatment of Endothelial Cells with Metabolic Inhibitors-The following inhibitors were added to serum free medium: ammonium chloride (20 mM), chloroquine (100 M), primaquine (100 M), monensin (5 M), cytochalasin B (10 M and 400 M), and colchicine (0.5 M). Inhibitors were added to the cultures 15 min prior to the binding assay and were present during the experiment, unless stated otherwise. After 1 h at 37°C, differentiation between internalized and matrix-deposited label was performed using 0.1 M ammonia as described.

Distribution of VN-TAT Complex into the Subendothelial Matrix
VN-TAT by EC was investigated at 4 and 37°C. EC were cultured on disconnectable wells and grown to confluence. Cells were incubated with radiolabeled VN-TAT (10 ng/well) over a time period of 30 min to 4 h in the absence or presence of 100-fold molar excess of unlabeled VN-TAT. The results were expressed as specific binding and uptake, which was defined as the difference in signals in the absence or presence of unlabeled VN-TAT. Aspecific binding accounted for approximately 10% of the total signal (not shown). Following binding, EC were separated from their matrix by ammonia extraction. Radioactivity associated with cell lysate represented bound radioligand when incubated at 4°C and bound and internalized radioligand when incubated at 37°C. Matrix-deposited label was defined as radioactivity remaining associated with the wells after removal of the cells. Fig. 1a shows specific binding after incubation at 4°C. The majority of bound radiolabel was found associated with the cell compartment, whereas hardly any label was detectable in the matrix. Incubation of radiolabel at 37°C resulted in time-dependent deposition of radiolabel into the endothelial cell matrix (ECM) (Fig. 1b).
To investigate the involvement of heparin-like structures on the endothelial cell surface, VN-TAT (13 ng/well) was incubated for 3 h in the absence or presence of increasing amounts of heparin (0.05-500 g/ml) and binding at 4°C or uptake at 37°C was measured. At 4°C, binding of VN-TAT complex to the cell surface was inhibited in the presence of heparin (Fig.  1c). Half-maximal inhibition was obtained in the presence of 200 g/ml heparin. At 37°C, heparin inhibited VN-TAT asso-ciation with the cell compartment and decreased deposition of VN-TAT into the matrix proportionally (Fig. 1d). Half-maximal inhibition was achieved in the presence of 3 and 20 g/ml of heparin, respectively. No binding occurred to BSA-coated wells.
To compare different extraction methods, detachment of the cells was performed with 0.1 M ammonia, 2 M urea, or 10 mM EDTA. Removal of the cells was checked microscopically; with ammonia the cells were lysed, whereas incubation with urea or EDTA removed the cells predominantly intact (not shown). At 4°C, 90 -93% of the cell-associated label was found associated with the cell compartment, whereas 7-10% was found in the matrix. At 37°C, 50 -59% was found in the cell compartment and 41-50% was associated with the ECM (Table I).
Internalization Assay-To follow intracellular routing, radiolabeled VN-TAT (35 ng/well, 4.4 nM) was incubated with an EC monolayer for 1 h at 4°C (designated time point 0) and cells were washed to remove non-bound material. The radioactive signal was chased by incubating the cells with an equal amount of unlabeled VN-TAT (40 ng/well). Subsequently cells were warmed to 37°C for the time course indicated on the x axis. After 1 h of incubation at 4°C (designated time point 0), 0.96 ng of radiolabel/well was associated with the endothelial cells (Fig.  2a). In the first 30 min of pulse-chase, about half of the initially surface-bound label was released to the medium, due to the establishment of a new equilibrium in the presence of excess of unlabeled VN-TAT complex (40 ng/well) in the medium (Fig.  2a). This amount of radiolabel in solution stayed constant during the remaining incubation period. To investigate whether VN-TAT was degraded, aliquots of the media were subjected to trichloroacetic acid precipitation. The amount of free iodine in the medium after 1 h incubation at 4°C (time point 0) accounted for 3.9%. During incubation at 37°C, free iodine in the media did not increase (Fig. 2b), indicating that the radiolabel was not proteolytically degraded during cellular processing. In the first 30 min at 37°C, half of the initially surface-bound ligand was shed to the medium, indicating that the other half was bound to the cell monolayer irreversibly.
Division into cell-associated or ECM-associated label by ammonia treatment showed that after 1 h of incubation at 4°C (time point 0) 88% was associated with the cell compartment and 12% with ECM. Raising the temperature showed that in time the cell compartment lost label, whereas ECM was enriched (Fig. 2c). To further subdivide cell-associated ligand into surface-bound and intracellular ligand, limited trypsin digestion according to Chappell et al. (1992) was performed. Surfacebound ligand was defined as ligand released from the ECmonolayer by trypsin/EDTA/proteinase K treatment. Ligand not sensitive to enzymatic digestion represented internalized  and d) and association of label with the cell compartment (q or å) or with the extracellular matrix (E or Ç) was determined as described under "Distribution Assay." Specific binding is shown, expressed in nanograms/ 10 6 cells (Ϯ standard error of the mean), which was defined as the difference in binding in the absence and presence of 100-fold molar excess of unlabeled VN-TAT complex. Data presented are representative results of three independent experiments performed in triplicate wells. Aspecific binding of radiolabeled VN-TAT on BSA-coated wells (ࡗ) is shown in panels c and d. When symbols lack standard error bars, they are too small to extend the symbols. ligand. Partitions, expressed as percentage of the sum of celland matrix-associated label, are shown in Fig. 2d. After 1 h at 4°C (time point 0), 90% was associated with the cell surface, 8% with the matrix, and 2% could be detected intracellularly. During incubation at 37°C, the amount of surface-bound label decreased in time, whereas the amount of intracellular ligand increased to 19% and matrix-deposited ligand increased up to 50%.
Binding of Gold-conjugated VN-TAT to EC Monolayer-To check whether incubation at 4°C influences the integrity of the monolayer, EC were incubated with gold-conjugated VN-TAT for 1 h at 4°C and cells were examined by phase-contrast light microscopy. Fig. 3a shows a random distribution of gold particles on the intact monolayer of EC, which was dramatically reduced in the presence of 500 g/ml heparin (Fig. 3b).
Binding of Radiolabel to Endothelial Cell Matrix (ECM)-Direct binding of radiolabeled VN-TAT to ECM was studied. For this purpose, matrix was preisolated from EC-monolayers by ammonia extraction (designated "preisolated matrix" or PIma) and assayed in comparison with intact EC monolayer.
Wells were incubated with radiolabeled VN-TAT (4.4 nM, 35 ng/well) in the absence or presence of 100-fold molar excess of unlabeled VN-TAT for 1 h at 4°C, washed, treated with ammonia, and washed. Radiolabel associated with EC monolayer before ammonia treatment, with the cell lysate, and with the matrix was measured. Defining specific binding to EC monolayer prior to ammonia treatment as 100% (ec), 89% of this label was associated with the cell compartment (lys), whereas only 5% was deposited into the underlying matrix (ma, Fig. 4). Six percent of radiolabel originally associated with the cell monolayer was lost during the extraction procedure. Direct binding to preisolated matrix (PI-ma) was 23 times higher compared to matrix (ma) isolated after radiolabel incubation. The second incubation of the preisolated matrix with ammonia did not release matrix-bound label (not shown).
Electrophoretic Analysis of Internalized VN-TAT Complex-Radiolabeled VN-TAT, subjected to SDS-PAGE and autoradiography was detectable as a prominent band at 158 kDa with two additional bands, representing multimeric forms of the complex (Fig. 5, lane 1). To analyze the appearance of the complex after internalization and deposition in the matrix, cell lysates and matrices obtained by ammonia extraction were subjected to SDS-PAGE followed by autoradiography. After incubation with EC for 1 h by 4°C, radiolabeled VN-TAT was bound to the cell compartment (Fig. 5, lane 2), whereas no label was detectable in the matrix (lane 6). The composition of cellbound label was not modified. After raising the temperature to 37°C, radiolabel disappeared from the cell compartment (lanes 3-5) and accumulated in the matrix (lanes 7-9). In all cases, matrix-deposited ternary complex retained its original molecular mass of 158 kDa.
Translocation of Gold-conjugated VN-TAT-Binding of VN-TAT complex was visualized by electron microscopic analysis in ultrathin cross-sections of EC. Cells were preloaded with goldconjugated ternary complex for 1 h at 4°C and washed to remove non-bound material. Fig. 6a shows predominantly surface-bound gold particles organized in clusters (Fig. 6b, at higher magnification). Hardly any gold particles were associated with the basolateral side of the endothelial cells. Raising the temperature to 37°C induced internalization of the VN-TAT-gold conjugates with ultimate deposition into the subendothelial matrix (Fig. 6c). Translocation of VN-TAT-gold conjugates occurred in endosomal structures (panels c and d). Incubation of EC with gold particles alone did not lead to internalization (not shown).
Immunofluorescent Analysis of Internalized VN-TAT Complex-Internalization of VN-TAT was shown in a fluorescence experiment. EC were cultured in VN-TAT deficient serum. At confluence, cells were incubated for 3 h at 37°C with medium containing VN-TAT deficient serum to which 5 g/ml VN-TAT was added and cells were prepared for fluorescent staining (see materials & methods). The presence of VN-TAT complex was detected with a monoclonal antibody against thrombin-modified antithrombin (␣AT mod ) and a monospecific polyclonal antibody directed against VN. Fluorescence was visualized using confocal laser scanning microscopy. This equipment is able to scan fluorescent signals sequentially from the top of the cells into the extracellular matrix. Fluorographs can than be generated en phase (a-f) or traverse (g-l). Fig. 7 (a-f) shows en phase fluorographs, representing TAT (a-c, detected with ␣AT mod ) or vitronectin (d-f). In the upper section of the cell (a and d), some fluorescent punctates are detectable but most punctates are visible in the middles section of the cell (b and e). The subendothelial matrix is diffusely stained for both TAT (c) and vitronectin (f). Distribution of punctates in traversal sections (g-l) taken at the location designated by the broken white line in . Cells were incubated with unlabeled VN-TAT (4.4 nM) to chase the signal, and the temperature was raised to 37°C. At the time-points indicated, radioactivity associated with the well prior to cell removal and in the media was measured. Incubation at 37°C in the presence of unlabeled VN-TAT resulted in release of radioligand from the cell monolayer (å) into the medium (É). b, free iodine in the media was measured by trichloroacetic acid precipitation (expressed as percentage of total counts/min). During incubation at 37°C (15-120 min) the amount of free iodine (Ⅺ) did not exceed the amount present in the medium after 1 h at 4°C (time point 0). c, at the time points indicated cells were washed and incubated with ammonia (0.1 M for 15 min at 4°C). Radioactivity detectable in ammonia lysates represented ligand associated with the cell compartment (surface ϩ intracellar, É). Matrixdeposited label was defined as radioactivity remaining associated with the wells after removal of the cells (q). d, cells were treated with trypsin/EDTA/proteinase K (30 min at 4°C). Surface-bound ligand was defined as ligand released from the cells by trypsin (E), whereas ligand not sensitive to trypsin treatment represented internalized ligand (Ç). Matrix-deposited label was defined as radioactivity remaining associated with the wells after removal of the cells (q). Results in panels c and d are expressed as percentage of cell ϩ ECM-associated ligand (Ϯ standard error of the mean). Representative results of three independent experiments are shown. When symbols lack standard error bars, they are too small to extend the symbols. micrograph d, confirm the en phase results: punctates are visible throughout the cell compartment and associated with the extracellular matrix, which is diffusely stained. Co-localization of TAT and vitronectin is shown in a traversal detail which was scanned for TAT (FITC channel, j) and for VN (TRITC channel, k) and then these signals were merged (l). The black and white prints show identical patterns, whereas under the microscope green (FITC channel) and red signals (TRITC channel) turned into yellow signals, indicating that TAT and vitronectin co-localized in the punctates. Incubation of the cells with medium containing human serum instead of purified VN-TAT showed identical results (not shown). Cells cultured in VN-TAT deficient conditions did not show any fluorescent signal (not shown).
Treatment of the Cells with Specific Inhibitors of Endocytosis or Transcytosis-To investigate the intracellular route involved in the translocation of VN-TAT from the luminal to the basolateral side of EC, inhibitors of cellular processes were added. Ammonium chloride, chloroquine, primaquine and monensin were added to the cell cultures 15 min before the label was added to the wells and were present during the binding experiment. Cytochalasin B and colchicine were only added during the binding assay, which was performed for 30 min, since longer incubation times damaged the integrity of the EC monolayer. The role of lysosomal processing was examined using NH 4 Cl (20 mM), primaquine (100 M), or chloroquine (100 M), weak bases that become concentrated in lysosomes and raise their pH (Maxfield, 1982) or monensin (5 M), a proton ionophore which raises endocytotic vesicle pH (Harford et al., 1983). Neither of these compounds affected release of VN-TAT to the matrix (Table II). The role of transcytotic processes was investigated using cytochalasin B, which inhibits the microfilament system (Sandvig and Deurs, 1990) or colchicine, which interferes with microtubule polymerization (Sackett and Varma, 1993). Cytochalasin B (10 and 400 M) inhibited the incorporation of VN-TAT into the matrix by 20 and 50%, respectively, and colchicine (0.5 M) by about 25%.

DISCUSSION
The clearance of the equimolar thrombin-antithrombin (TAT) complex from the circulation and its distribution into extravascular tissue is coupled to ternary complex formation with a third glycoprotein in plasma, vitronectin (de Boer et al., 1993). In the present study we provide evidence for the presence of these components in the vascular wall and define requirements that lead to translocation of the luminally bound ternary complex to the basolateral side of endothelial cells (EC). Recently, we have shown that VN-TAT complex binds rapidly to EC and that the binding domain of the complex is located in the heparin binding region of the VN moiety (de Boer

FIG. 4. Binding of VN-TAT to preisolated matrix.
Preisolated matrix (PI-ma) was prepared by ammonia extraction. EC monolayer and PI-ma was incubated for 1 h at 4°C with 4.4 nM radiolabeled VN-TAT in the absence or presence of 100-fold molar excess of unlabeled VN-TAT, washed and treated with ammonia. Radioligand associated with the cell monolayer (ec) prior to ammonia treatment, with ammonia lysate (lys), its isolated matrix (ma) and with PI-ma was measured. Specific signals are shown (expressed in cpm/well Ϯ standard error of the mean), defined as the difference in radioactivity in the absence or presence of unlabeled ligand. Experiments were performed on triplicate wells on three different cell batches and a representative experiment is shown. PI-ma bound 23 times more radioligand compared to matrix isolated after incubation with radiolabel (ma).

FIG. 5. Electrophoretic analysis of translocated VN-TAT.
Label associated with the cell compartment or deposited into the extracellular matrix was subjected to 4 -15% SDS-PAGE followed by autoradiography. Radiolabeled VN-TAT is shown (lane 1) associated with the cell compartment (lanes 2-5) or associated with the matrix (lanes 6 -9) after 1 h of preincubation at 4°C (time point 0 min) or after warming the cells to 37°C for 15, 60, and 180 min, respectively. Molecular markers in thousands are shown on the left margin. et al., 1992). Exposure of the heparin binding region occurs upon a conformational transition; native VN has no affinity for heparin. In vitro, this transition can be achieved by denaturation using chaotropes, detergent, low pH, or binding to plastic.
In vivo, TAT complex may serve as a "physiological activator" of VN, since the interaction between TAT and VN leads to a similar conformational change in VN.
In the present report, we studied the destination of radiola-FIG. 6. Electron microscopic evaluation of VN-TAT translocation. Timedependent internalization and translocation of VN-TAT gold-conjugates was visualized by electron microscopy in ultrathin sections of EC. Incubation of EC for 1 h at 4°C reveals clusters of VN-TATgold conjugates close to the cell surface (a) and the same situation at higher magnification (b). After preincubation at 4°C, the cells were warmed to 37°C for 1 h. The gold marker is seen in endosomes (c and d) and deposited into the subendothelial matrix (c). Representative micrographs are shown. Bars represent 1.0 m.

FIG. 7. Confocal microscopy of intracellular and subcellular distribution of VN-TAT.
EC were cultured on glass coverslips in VN-TAT deficient medium. At confluence, cells were incubated for 3 h at 37°C with medium to which purified VN-TAT was added. After washing, cells were fixed and permeabilized with methanol enabling the antibodies to enter the cell compartment. Internalized VN-TAT was detected with monoclonal antibody ␣AT mod directed against thrombin-modified antithrombin and a monospecific polyclonal antibody against VN, followed by anti-mouse IgG conjugated with FITC and anti-rabbit IgG conjugated with TRITC fluorescent label. Fluorescent signal was visualized using convocal laser scanning microscopy and en phase (a-f) or traverse (g-l) micrographs were generated. Panel a-f shows en phase micrographs scanned from the top of the cell (a and d) into the subendothelial matrix (c and f) and in an intermediate section (b and e) for the presence of thrombin-modified antithrombin (FITC channel, a-c) or vitronectin (TRITC channel, d-f). In panels g-i triplicate micrographs are shown of the part of the cell marked by the broken line in micrograph d, scanned for the presence thrombin-modified antithrombin. Panels j-l represent a detail of some punctates scanned for thrombin-modified antithrombin (j) or vitronectin (k) and the merged signals of micrographs j and k (l), showing complete co-localization of thrombin-modified antithrombin and vitronectin. Representative micrographs are shown (amplification 63 ϫ 1.4).
beled VN-TAT complex incubated on metabolically inactive EC (4°C) or metabolically active cells (37°C). An assay was designed that discriminates between VN-TAT associated with the cell compartment or with extracellular matrix. For this purpose endothelial cells were cultured on disconnectable wells, which could be placed in a ␥-scintillation counter and measured separately. Ammonia extraction, a standard technique in our laboratory (Sixma et al., 1987), was used to lyse the cells. This method of cell removal keeps the matrix intact and firmly attached to the entire area of the culture wells (Vlodavsky et al. 1987). VN-TAT-associated radioactivity detectable in the cell lysate represented cell-associated label, whereas matrix-associated label was defined as the radioactivity left behind on the culture wells after ammonia extraction. Other extraction buffers containing urea or EDTA, which removed the cells from the matrix intact, gave similar results indicating that measurement of matrix-deposited label was not dependent on the extraction method.
VN-TAT binding to EC incubated at 4°C occurred in a cellspecific, ligand-specific, time-dependent, and heparin-dependent manner, which is in accordance with previous observations (de Boer et al., 1992). EC incubated at 37°C bound VN-TAT with similar binding characteristics, but additionally radiolabel was delivered to the extracellular matrix. The appearance of radiolabeled VN-TAT in the subendothelial matrix was not due to direct binding of VN-TAT to exposed extracellular matrix, since the integrity of the monolayer remained intact during the binding assay.
This was deduced from two control experiments. (a) When matrices were preisolated by ammonia extraction and subsequently incubated with radiolabel for 1 h at 4°C, this matrix contained 27 times more radiolabel compared to the matrix isolated from an EC monolayer which had been incubated with radioligand prior to ammonia extraction. (b) Light-microscopic evaluation of EC incubated with gold-conjugated VN-TAT for 1 h at 4°C showed a fully intact EC monolayer on which radiolabel was evenly distributed.
To study internalization in more detail, a pulse-chase set-up was used in which the surface of EC was preloaded with VN-TAT at 4°C. The radioactive signal was then chased by adding unlabeled VN-TAT, and the cells were metabolically activated by raising the temperature to 37°C. In time, the cell compartment lost radiolabel, whereas the matrix was enriched with ligand. During the pulse-chase experiment, radiolabel initially bound to the cell surface was released to the medium due to the establishment of a new equilibrium in the presence of an excess amount of unlabeled VN-TAT in the medium. Radioactivity remained constant throughout the incubation period, indicating that matrix-deposited radioligand was not derived from the medium, but originated from intracellular pools. To obtain additional data on kinetics of internalization, we performed a cell removal technique using limited proteolytic trypsin cleavage (Chappell et al., 1992) which discriminates between surface-bound and intracellular radioligand. We could trace an intracellular pool of radioligand; internalization reached a steady state level after 90 min of incubation at 37°C. The trypsin method detected about 10 -20% less VN-TAT in the matrix compared to ammonia treatment. Apparently some VN-TAT was liberated from the matrix during the enzymatic treatment with trypsin.
The presence of an intracellular pool was also evidenced by immunofluorescent staining of endothelial cells cultured under VN-TAT deficient conditions and subsequently incubated with medium containing purified VN-TAT. Since no antibodies are available that directly recognize the ternary VN-TAT complex, cells were incubated with a monoclonal antibody directed against thrombin-modified antithrombin and a polyclonal antibody against vitronectin and double-fluorescent staining was performed. Fluorescent label co-localized in a punctated pattern randomly spread intracellularly and associated with the extracellular matrix.
These findings were complemented by electron microscopic evaluation of ultrathin sections of EC incubated with goldconjugated VN-TAT. EM micrographs illustrated the translocation phenomenon and revealed the presence of gold conjugates in transcytotic endosomes as well as in association with the subendothelial matrix. We cannot rule out transport of VN-TAT complex via cell junctions or through lateral diffusion, although indications for this possibility could not be found in the EM micrographs.
Transcytosis rather than endocytosis seemed to be involved in the translocation of the complex from the luminal to the basolateral side of the EC as was concluded upon several observations. (a) Endosomal structures were involved in the translocation from the luminal to the basolateral side of the EC. (b) Electrophoretic analysis of matrix-deposited complex showed that the complex was released to the matrix fully intact. (c) Trichloroacetic acid precipitation of media collected during the pulse-chase experiment showed a constant level of free iodine, indicating that VN-TAT was not degraded during internalization. (c) Specific inhibitors of lysosomal function and endocytosis such as ammonium chloride, chloroquine, primaquine, or monensin did not influence matrix deposition, whereas specific inhibitors of transcytosis such as cytochalasin B and colchicine partially decreased matrix deposition. It should be noted that treatment with cytochalasin B and colchicine, which may induce cell death in time, was carried out only for a short period of time in order to minimize loss of integrity of the EC monolayer. This may explain their relatively mild effect on matrix deposition compared to the high concentrations used.
Various plasma molecules cross the endothelium by receptormediated transcytosis. For instance, insulin (King and Johnson, 1985) as well as some carrier proteins like albumin (Ghitescu et al., 1986), transferrin (Jefferies et al., 1984), or lipoprotein lipase (Saxena et al., 1990) are transported through the endothelium very efficiently without degradation. Especially the processing of lipoprotein lipase by endothelial cells is of interest, since striking similarities are apparent compared to VN-TAT; binding is heparin-dependent, internalization reaches a steady state level (Saxena et al., 1990), and during internalization no degradation occurs. Translocation of lipoprotein lipase is decreased in the presence of cytochalasin B but not affected by chloroquine, indicating that transcytosis is involved as well. The establishment of a steady state level of internalized ligand may be characteristic for transcytosis and may be due to very rapid translocation of ligand. In the case of VN-TAT, label appeared in the matrix almost immediately and exceeded the amount of intracellular ligand. When incubated on fibroblasts, lipoprotein lipase is internalized and subsequently degraded through a lysosomal pathway (Chappell et al., 1992). This was shown to occur via the low density lipoprotein receptor-related protein pathway.
Lipoprotein receptor-related protein cannot be responsible for internalization of VN-TAT by endothelial cells, since HU-VEC do not express this receptor (Godyna et al., 1995). Interestingly, fibroblasts are able to process and degrade conformationally altered, non-complexed VN (Panetti and McKeown-Longo, 1993). This was shown to be mediated by the ␣ v ␤ 5 integrin receptor. This receptor cannot be involved in the above described pathway either, since RGD peptides have no effect on binding of VN-TAT to HUVEC (de Boer et al., 1992). Recently Waltz and Chapman (1994) showed that the urokinase plasminogen activator receptor binds conformationally altered VN. The involvement of this receptor and possibly others in binding and internalization of VN-TAT by endothelial cells remains to be established.
In conclusion, our results indicate that the direct association of TAT with VN will cause a conformational transition(s) of VN, leading to the exposure of the heparin binding site, which enables the complex to bind to EC through high affinity binding sites with subsequent internalization. This process not only leads to the clearance of TAT complex from the circulation, but also promotes deposition of VN-TAT into the extracellular matrix. This mechanism may supply the matrix with VN, which must originate from plasma, since EC neither synthesize VN (Seiffert et al., 1991) nor bind native VN . Alternatively, heparin-binding VN multimers found in platelet releasate  and at low concentrations in plasma may cross the EC monolayer in an equivalent manner, as suggested from in vitro experiments Völker et al., 1993). Two important features of VN in the matrix are its abilities to promote cell attachment and to stabilize PAI-1, a major inhibitor of plasminogen activation. Purified VN-TAT complex is still able to bind PAI-1 and promote cell attachment, 2 indicating that the ternary complex may fulfil crucial functions at its final destination in the vessel wall.