Adhesion of Blood Platelets Is Inhibited by VCL, a Recombinant Fragment (Leucine⁵⁰⁴ to Lysine⁷²⁸) of von Willebrand Factor

Construction, Synthesis, and Purification of Peptide

The sequence encoding Leu⁵⁰⁴ to Lys⁷²⁸ (VCL) was derived from a human endothelial cDNA library (Clontech). The coding sequence was placed into a vector under control of an inducible λ P_L promoter²⁵ and the deo P1P2 RBS.²⁶ The peptide was purified from Escherichia coli after induction by ion exchange chromatography with CM-Sepharose, yielding essentially pure material. The disulfide bond between Cys⁵⁰⁹ and Cys⁶⁹⁵ was formed by oxidation. Purity and oxidation were verified with SDS-PAGE under reducing and nonreducing conditions as well as by high-performance liquid chromatography with a Superose 12 column. Note that the peptide contains an extra N-terminal methionine.

Functional Studies

The functional activity of VCL was determined in aggregation studies. These studies were performed as follows. Platelet-rich plasma (250 000/μL) was incubated for 30 minutes at 20°C with various concentrations of VCL. Aggregation was performed in an aggregometer (Bio/Data Corp) with 500 μL platelet-rich plasma that was prewarmed for 1 minute at 37°C. The appropriate aggregating agents such as ADP (FC, 5 μmol/L), collagen (equine tendon collagen type I, Hormonchemie; FC, 4 μg/mL), botrocetin purified from Bothrops jararaca venom (Sigma Chemical Co) by DEAE chromatography as described by Fujimura et al²⁷ (FC, 35 μg/mL), and ristocetin (H. Lundbeck & Co; FC, 1.0 mg/mL) were then added. Aggregation studies were performed at 37°C and 900 rpm stirring speed.

Perfusion Studies

Perfusion studies were carried out in a parallel-plate perfusion chamber with well-defined rheological characteristics devised to accommodate duplicate protein-coated glass microscope coverslips.^{28 29} Whole blood obtained by venipuncture from healthy volunteer donors was anticoagulated with 1/10 vol 110 mmol/L trisodium citrate (citrate blood) or with 1/10 vol 200 U/mL LMWH blood (Fragmin, Kabi Pharmacia). The LMWH blood was used for perfusion studies over laminin, since adhesion to laminin is dependent on the presence of divalent cations.⁴ VCL was added to whole blood as a 64-μmol/L stock solution in HBS in varying amounts to reach the appropriate concentration. The same amount of HBS was added to the controls. The blood (15 mL) was then prewarmed at 37°C for 5 minutes and recirculated through the perfusion chamber for 5 minutes at wall shear rates ranging from 300 to 1800 s⁻¹. The coverslips were removed, rinsed with 10 mmol/L HEPES buffer, pH 7.4, containing 150 mmol/L NaCl (HBS), fixed in 0.5% glutardialdehyde in PBS, dehydrated in methanol, and stained with May-Grünwald-Giemsa as previously described.²⁹ Platelet adhesion was evaluated with a light microscope, and the coverage was measured with an image analyzer (AMS 40-10). Platelet adhesion was expressed as the percentage of the surface covered with platelets.

Adhesion to Cryostat Cross Sections of Coronary Arteries With Atherosclerotic Plaques

Glass coverslips (18×18 mm; Menzel Gläser) were cleaned by being soaked overnight in chromosulfuric acid and rinsed thoroughly with deionized water. They were air dried and coated with Denhardt’s solution (0.02% Ficoll 70 [Pharmacia AB], 0.02% BSA fraction V, and 0.02% polyvinylpyrrolidone 360 [Sigma] in deionized water) for 3 hours at 68°C. The coverslips were rinsed with deionized water (20 seconds), fixed in ethanol/acetic acid (3/1; 20 minutes at room temperature), air dried, and baked for 3 hours at 180°C.

Frozen cross sections of postmortem (obtained within 24 hours after death) coronary arteries were cut at 6-μm thickness at 20°C and were mounted on Denhardt’s solution–coated glass coverslips.³⁰ The Denhardt’s coating was used to prevent detachment of the tissue section in flow. Tissue sections stuck very well on this nonthrombogenic coating. After air drying (2 hours), the sections were incubated in a 4% human albumin (Behringwerke AG) solution in HBS (12 to 16 hours at 4°C) to block nonspecific binding. The coverslips were then rinsed with HBS and inserted into the perfusion chamber. A minimum of six serial sections per frozen artery segment was used for perfusions. After perfusion, the coverslips were removed from the chamber, rinsed with HBS, and fixed with 2% paraformaldehyde in PBS for 5 minutes. Platelet deposition was demonstrated with a biotinylated monoclonal antibody directed against GPIb that is specific for platelets, followed by a streptavidin–biotinylated horseradish peroxidase complex (Dakopatts) as described previously.³⁰

Purified Proteins and Coating Procedure

Mouse laminin (A₁,B₁,B₂; Gibco) was analyzed on a 6% SDS-PAGE gel. It yielded the characteristic single band of about 800 kD on unreduced gels and three bands of 400 000, 215 000, and 205 000 kD under reducing conditions. vWF was purified from control cryoprecipitates by gel filtration on Sepharose 4B (Pharmacia AB) as described earlier.^{31 32} vWF in the void volume was precipitated by dialysis against 1.9 mol/L ammonium sulfate, pH 7.0, at 4°C and stored as ammonium sulfate suspension at 4°C until use. Precipitated protein was collected by centrifugation (2 minutes, 10 000g). The pellet was dissolved in 0.05 mol/L Tris-HCl and 0.1 mol/L NaCl, pH 7.4 (TBS), and dialyzed against the same buffer. The characteristic multimer pattern of vWF was seen after agarose electrophoresis. The characteristic pattern of a band at 270 000 and two minor proteolytic bands at 170 000 and 140 000 kD was found in SDS-PAGE of reduced vWF. Fibronectin was isolated from human plasma by affinity chromatography on gelatin-Sepharose as previously described.³² SDS-PAGE after reduction showed the characteristic doublet at 220 000 kD. Fibrinogen was purchased from Kabi Pharmacia. SDS-PAGE after reduction yielded the characteristic triplet at 62 000, 58 000, and 51 000 kD. Small quantities of fibronectin and vWF were found by ELISA. These quantities gave no platelet adhesion by themselves when they were applied to a coverslip (experiments not shown). Collagen type I (calf skin) was purchased from Sigma and equine tendon from Hormonchemie. Both were dissolved at 1.4 mg/mL in 50 mmol/L acetic acid.

Glass coverslips (18×18 mm, Menzel Gläser) were cleaned by being soaked overnight in 80% ethanol and rinsed thoroughly with distilled water. They were coated by incubation for 1 hour at room temperature with laminin (100 μg/mL in 10 mmol/L phosphate buffer, pH 7.4, 0.15 mol/L NaCl [PBS]) and vWF (10 μg/mL in PBS). After coating, the glass coverslips were incubated with 1% human albumin in PBS for 2 hours. Fibronectin, fibrinogen, and collagen type I were sprayed onto glass coverslips (cleaned as described above) with a retouching airbrush (Badger Model 100 Il, Badger Air Brush Co) connected to a nitrogen cylinder operating at a pressure of 1.5 atm. The fine droplets dried instantaneously on the glass surface at room temperature.³³ The collagens were coated at 30 μg/cm², fibronectin at 20 μg/cm², and fibrinogen at 20 μg/cm².

Coverslips were coated with VCL by spraying quantities ranging between 10 and 100 μg per coverslip with the retouching airbrush. The best results were obtained with VCL in HBS. All spray-coated coverslips were blocked with a 1% human albumin solution as described above. Control studies showed no platelet adhesion to glass coverslips incubated with 1% human albumin solution in PBS for 2 hours.

The roles of GPIb and GPIIb-IIIa were investigated with monoclonal antibodies. AK1 and AK2 directed against GPIb were a kind gift from Dr M. Berndt, Melbourne, Australia³⁴ ; 6D1 directed against GPIb was a kind gift from Dr B. Coller, New York, NY³⁵ ; and C17 directed against GPIIIa inhibiting platelet adhesion and aggregation via GPIIb-IIIa was a generous gift from Dr A. von dem Borne, Amsterdam, Netherlands.³⁶ CLB-RAg 201 directed against vWF and inhibiting binding to collagen^{37 38} was a kind gift from Dr Jan van Mourik, Amsterdam, Netherlands.

Preparation of Endothelial Cell and Fibroblast Matrices

Human endothelial cells derived from umbilical veins were isolated and cultured as described before.^{39 40} For the experiments described, endothelial cells were cultured on glass coverslips previously coated with gelatin. To isolate the extracellular matrix, cells were grown to confluence and exposed to 0.1 mol/L NH₄OH for 30 minutes at room temperature. The cell layer was completely removed by this procedure, leaving the extracellular matrix intact.⁴¹ The extracellular matrix was washed three times with PBS before use. Fibroblasts derived from human lung were cultured as described.⁴² The matrix was prepared as described above for the endothelial cells.

Binding Studies With VCL

VCL was radiolabeled with ¹²⁵I using the lactoperoxidase–glucose oxidase beads (Enzymobeads, Bio-Rad) according to the instructions of the manufacturer. Noncovalently linked ¹²⁵I was removed by dialysis against PBS. The specific activity ranged between 20 and 50 cpm/ng. Trichloroacetic acid (10%) precipitated 93% to 95% of the radioactivity. The effect of radiolabeling on VCL was studied by diluting the ¹²⁵I-labeled VCL with various concentrations of unlabeled material and studying it in relevant binding assays. Radiolabeling did not affect binding of VCL. vWF was labeled with ¹²⁵I using the lactoperoxidase–glucose oxidase beads (Enzymobeads) according to the instructions of the manufacturer. Radiolabeled ¹²⁵I-vWF contained 14% free radiolabel. It was labeled to a specificity ranging between 250 and 300 cpm/ng. Radiolabeled vWF showed the same binding characteristics as unlabeled vWF in mixing experiments.

Platelet binding studies were performed with platelets washed with Krebs-Ringer buffer at pH 6.5.⁴³ Platelets were resuspended to 200 000/μL in PBS with 1% human serum albumin and varying amounts of radiolabeled VCL. Ristocetin (FC, 1 mg/mL) or buffer was added to a volume of 650 μL. This was layered on top of 25% sucrose in PBS and incubated for 1 hour at room temperature. The tube was then spun for 2 minutes at 10 000g at room temperature, and the pellet was counted in a gamma counter. All assays were performed in triplicate. In some experiments, anti-GPIb antibody (AK2 or 6D1) was added as ascites in the indicated dilution and incubated for 1 hour at room temperature before the radiolabeled VCL was added.

Binding of radiolabeled VCL or vWF to collagen type III was studied with the following technique. Fibrillar collagen type III from human placenta (Sigma) was dissolved in 50 mmol/L acetic acid at 1 mg/mL, dialyzed against PBS for 48 hours at 4°C, and diluted with PBS to a concentration of 100 μg/mL. This suspension (125 μL) was pipetted into wells of a 96-well ELISA tray (Costar) and spun at 1500 rpm for 15 minutes in an ELISA tray rotor to coat the wells. Supernatants were carefully removed, and the wells were cautiously rinsed and blocked for 1 hour at room temperature with TBS containing 3% BSA and 0.1% Tween 20. Samples for binding were added in 100 μL TBS and incubated for 2 hours at room temperature. The supernatant was then removed, and the wells were rinsed three times with 200 μL TBS. The tips of the wells were cut off and counted for 5 minutes in a gamma counter. All assays were performed in triplicate. Collagen type III was used for technical reasons. Previous studies have shown similar binding characteristics of vWF to fibrillar collagen types I and III. Nonspecific binding of VCL was determined in the presence of a 10-fold excess of unlabeled VCL and subtracted. Nonspecific binding of vWF was determined in the presence of 5 μmol/L recombinant leech antiplatelet protein,^{44 45} a specific inhibitor of vWF binding to collagen.⁴⁶

Characteristics of VCL

On high-performance liquid chromatography Superose 12 analysis, VCL showed a single sharp peak indicative of homogeneous material. SDS-PAGE gave a single band with an M_r of 24 kD reduced and 25 kD unreduced. The functional activity of VCL was studied in aggregation tests. No effect was observed on ADP- or collagen-induced aggregation. Ristocetin-induced aggregation was completely inhibited at 0.5 μmol/L with half maximal inhibition at 0.2 μmol/L. Inhibition of botrocetin-induced aggregation has been published before by one of us (L.I.G.) in a study by Gralnick et al.²⁴ It was completely inhibited at 0.5 μmol/L with half maximal inhibition at 0.08 μmol/L. Reduced VCL was not studied because it precipitates as soon as it is reduced at neutral pH.

Inhibition of Adhesion to ECM

The effect of VCL on adhesion to ECM was studied at 300 and 1600 s⁻¹ (Fig 1⇓). Maximal inhibition of 60% was found at concentrations of 5 μmol/L, with half maximal inhibition at 1.5 μmol/L at 1600 s⁻¹. Maximal inhibition of 40% was found at 5 μmol/L, with half maximal inhibition at 1.5 μmol/L when the perfusions were performed at 300 s⁻¹. Because hemostasis is determined by adhesion to the matrix of perivascular fibroblasts as well as to ECM, we also studied the inhibition of adhesion to a fibroblast matrix. At a final VCL concentration of 2 μmol/L, platelet adhesion was not inhibited at a shear rate of 300 s⁻¹ but was inhibited by 40% at a shear rate of 1600 s⁻¹ (Table 1⇓). The inhibition at 1600 s⁻¹ was of the same order of magnitude as that to the ECM.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Graph shows inhibition of platelet adhesion to ECM by VCL. The final concentration of VCL in whole citrated blood (x axis) was plotted against the percent inhibition of platelet adhesion after 5 minutes of perfusion (y axis). Solid line indicates shear rate of 300 s⁻¹; dotted line, shear rate of 1600 s⁻¹. Values are the mean of three separate experiments for each two perfusions with each two coverslips per concentration of VCL. Platelet adhesion at 300 s⁻¹ was 35.8±7.2 and at 1600 s⁻¹ 45.5±5.3 (mean±SEM).

Inhibition of Adhesion to Matrix Proteins

Platelet adhesion to the vessel wall or ECM is determined by a series of ligand-receptor interactions. The most important adhesive ligands are collagen, fibronectin, vWF, and laminin. To clarify further the inhibitory effect of VCL, we studied its effect on the adhesion to these proteins. We included fibrinogen, which is not a proper matrix protein but is also important, because platelet adhesion to fibrinogen/fibrin is essential for the propagation of a thrombus. The results are summarized in Table 1⇑. Adhesion to laminin was studied only at a shear rate of 300 s⁻¹, because this adhesion is strongly dependent on shear and very little or no adhesion is observed at 1600 s⁻¹. VCL had no effect on adhesion to laminin. Both the adhesion to fibrinogen and fibronectin were inhibited more strongly at 1600 s⁻¹ than 300 s⁻¹. These results are in agreement with our previous observation that adhesion to these ligands at 1600 s⁻¹ is supported by vWF.^{46 47} The adhesion to vWF was not inhibited at either 300 or 1600 s⁻¹. This lack of inhibition might have been caused by the high density of vWF on the coverslip. We therefore also studied the effect of 2 μmol/L VCL on adhesion to glass coverslips incubated with either 3 or 6 μg of vWF and compared that with the effects on coverslips incubated with 10 μg vWF. Platelet adhesion increased as expected at higher vWF concentrations, but VCL had no significant effect whether at high or low surface density.

Inhibition of Adhesion to Collagen

Perfusion studies were performed with calf skin and equine tendon collagen type I. The results of these studies are summarized in Table 1⇑. The inhibition of adhesion to calf skin collagen type I was 35% at 300 s⁻¹ and 75% at 1600 s⁻¹. Adhesion to equine tendon collagen was inhibited by 15% at 300 s⁻¹ and 25% at 1600 s⁻¹. This inhibition is in agreement with earlier observations, showing that adhesion to collagen is dependent on vWF.³² The lower inhibition of adhesion to equine tendon collagen type I may indicate that the effect of VCL is less when the collagen is very reactive and when platelet vWF is thus released from α-granules.⁴⁸ Because platelets adhere to VCL itself (see below) and VCL has a binding site for collagen, it was of interest to study the effect of preincubation of collagen with VCL. Collagen type I was sprayed on a coverslip, and this was incubated for 60 minutes at room temperature with VCL in a concentration of 2 μmol/L in PBS. Preincubation of collagen type I caused a slight but insignificant increase in adhesion of about 15% for equine tendon collagen type I. Preincubation of calf skin collagen type I caused no increase in adhesion, and preincubation of ECM also had no effect on adhesion.

Effect of VCL on Adhesion to an Atherosclerotic Plaque

Perfusion studies of atherosclerotic plaques were performed at 2250 s⁻¹, which caused formation of platelet thrombi on the atherosclerotic lesion.³⁰ VCL at 5 μmol/L concentration caused a decrease in thrombus size (Fig 2⇓). This effect was reproducible, but there was considerable variation in the degree of inhibition among different blood donors.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Photomicrographs show inhibition of thrombus formation on an atherosclerotic plaque. Cryostat sections of atherosclerotic coronary arteries were mounted on glass coverslips and exposed to circulating citrated whole blood with and without 5 μmol/L VCL at a shear rate of 2250 s⁻¹ for a perfusion time of 5 minutes. Thrombi were stained with biotinylated anti-GPIb followed by streptavidin-peroxidase.³⁰ Top, Control: large thrombi are present on the connective tissue of an atherosclerotic plaque. Bottom, 5 μmol/L VCL: a pronounced reduction in thrombus size was observed after perfusion in the presence of 5 μmol/L VCL.

Adhesion to VCL

Because VCL possesses the GPIb-binding domain of vWF, it was of interest to see whether platelets adhere to this fragment itself. VCL was sprayed onto glass coverslips in concentrations varying between 10 and 100 μg per coverslip and then blocked by human serum albumin as described in “Methods.” The results are shown in Fig 3⇓. Platelet adhesion increased with increasing surface concentrations of VCL. The results often fluctuated from one concentration to another without evident explanation. The effect of monoclonal antibodies against GPIb and GPIIb-IIIa on adhesion to VCL was tested in separate experiments. The inhibitory antibodies against GPIb, AK2, and 6D1 caused complete inhibition, whereas the noninhibitory antibody AK1 had much less effect; anti–GPIIb-IIIa (C17) had no effect at all. VCL also has the heparin-binding site, which is essential for the binding of vWF to heparin.^{14 49} The effect of heparin preincubation on the adhesion to VCL was therefore studied in separate experiments. Glass coverslips were coated with VCL, and these coverslips were preincubated with UFH (Thromboliquine, Organon) and LMWH (Fragmin, Kabi Pharmacia) at concentrations of 0.1, 0.3, and 1.0 U/mL for 2 hours at room temperature. Adhesion was already strongly inhibited by 0.1 U/mL UFH (90%) or LMWH (80%) and completely inhibited by 0.3 U/mL or higher UFH or LMWH.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Graph shows platelet adhesion to VCL. Adhesion was studied at shear rates of 300 (solid line) and 1600 (dotted line) s⁻¹. The curves are representative for three similar experiments. Every single point is the mean of two perfusions each with two coverslips in the chamber. Percent adhesion indicates the percentage of the coverslip covered by platelets.

Preincubation of a VCL-coated coverslip with UFH completely inhibited adhesion. This raised the question of whether UFH might influence the inhibitory action of VCL. For this purpose we studied the inhibitory action of VCL on platelet adhesion in flowing heparinized and citrated blood to ECM, calf skin collagen, and fibrinogen. VCL was also a good inhibitor in blood anticoagulated with UFH (results not shown).

Binding Studies With VCL

To improve our understanding of the mechanism of action of VCL, we radiolabeled it with ¹²⁵I and performed binding studies. VCL binding to platelets was studied in the presence and absence of ristocetin. The results of a typical experiment are presented in Fig 4⇓, top. Ristocetin-induced binding was not saturable up to a concentration of 10 μmol/L VCL. vWF competed for this binding (Fig 4⇓, middle). VCL binding in the presence of ristocetin was inhibited by anti-GPIb (Fig 4⇓, bottom). Binding in the absence of ristocetin was much lower and not inhibited by vWF and only weakly by anti-GPIb.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Graphs show binding of ¹²⁵I-VCL to platelets. Top, Binding of ¹²⁵I-VCL to washed platelets was performed as described in “Methods.” Binding in the presence of ristocetin (▵——▵) was much higher than binding in its absence (▴——▴). Middle, Dose-dependent inhibition of VCL binding to washed platelets by vWF was observed in the presence of ristocetin (▵——▵) but not in its absence (•——•). The VCL concentration was 0.5 μmol/L in the presence of ristocetin and 2 μmol/L in the absence of ristocetin (to obtain enough counts). The binding was expressed as a percentage of the value with no added vWF (control). Bottom, Inhibition of ¹²⁵I-VCL binding with anti-GPIb, which was expressed as a percentage of the binding without addition of anti-GPIb (AK2) in the presence or absence of ristocetin. Similar data were obtained with monoclonal antibody 6D1.

Collagen binding studies were performed using collagen type III. VCL bound much less to collagen than to vWF itself, with only 12 fmol bound at 0.07 μmol/L VCL versus 230 fmol vWF at 0.07 μmol/L vWF (Fig 5⇓, top). The binding of VCL was inhibited by VCL itself, whereas vWF had about five times less effect (Fig 5⇓, middle). vWF binding was inhibited by excess nonlabeled vWF or a specific monoclonal antibody, CLB-RAg 201, which has been described before,^{37 38} but not by a 5000-fold excess of VCL, which is in agreement with the fact that the collagen-binding site in the A3 domain of vWF is essential for the binding of vWF to collagen⁴⁶ (Table 2⇓).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Graphs show binding of VCL and vWF to collagen type III. Top, ¹²⁵I-VCL binding was performed as described in “Methods.” Data on the y axis describe the amount bound per well. The binding curve shows the specific binding. Middle, ¹²⁵I-vWF binding was performed as described in “Methods.” Note the difference in units on the x axis with Fig 5⇓, top. Bottom, Competition of binding of ¹²⁵I-VCL (0.038 μmol/L) by VCL (▴——▴) or vWF (▵——▵).