| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,
,||1
* Blood Transfusion Unit,
Division for Experimental Oncology 2, The National Cancer Institute CRO-IRCCS, Aviano (PN) 33081 Italy;
Department for Cell and Molecular Biology, University of Lund, S-22184 Lund, Sweden;
MATI Centre of Excellence, University of Udine, Piazzale Kolbe, Udine, 35100 Italy; and
|| Department of Evolutionary and Functional Biology, University of Parma, 43100 Parma, Italy
1Correspondence: Department of Evolutionary and Functional Biology, The University of Parma, Viale delle Scienze 11/A, 43100 Parma, Italy. E-mail: rperris{at}cro.it; Blood Transfusion Unit, The National Cancer Institute CRO-IRCCS, Aviano 33081, Italy. Email ldemarco{at}cro.it
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
/integrin/selectin-independent platelet ligands in healthy and diseased vascular beds and may be directly responsible for the platelet accruing after rupture of atherosclerotic plaques.—Mazzucato, M., Cozzi, M. R., Pradella, P., Perissinotto, D., Malmström, A., Mörgelin, M., Spessotto, P., Colombatti, A., De Marco, L., Perris, R. Vascular PG-M/versican variants promote platelet adhesion at low shear rates and cooperate with collagens to induce aggregation.
Key Words: proteoglycan • hemostasis • shear stress
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2)
. A certain "pathological" platelet accruing may occur in juxtaposition with atherosclerotic plaques and is thought to be mediated by the combined action of the calcified ECM and acidified low density lipoproteins composing these formations (3
4
5)
. The mechanisms underlying platelet adhesion and aggregation differ in various vascular beds due to the diverse rheological conditions known to operate in the different vascular structures (1)
. Thus, in arteriols and stenotic vessels, where blood flow is elevated and platelets are exposed to a high wall shear rate/stress (1)
, the initial arrest of single platelets is promoted by a stable interaction with von Willebrand factor (vWF) that may be secreted by the endothelium and/or tightly linked to subendothelial ECM components (6
7
8
9)
. This high-avidity interaction is mediated by their specific surface receptor GPIb (10
11
12
; M. Mazzucato et al., unpublished results), whereas the subsequent thrombus growth is permitted by the concomitant action of the activated platelet integrin IIbß3, which binds to soluble and immobilized fibrinogen (6
, 8
, 13)
. In mice, the subtle control of thrombogenesis may involve ECM components with inhibitory functions such as vitronectin and perlecan, which are thought to prevent unwanted occlusive thrombosis in these animals (14
, 15)
.
Based on in vitro studies, interstitial collagen types I and III (Col I and III) possessing high affinity for vWF have been proposed to constitute the natural ligands for soluble vWF and the platelet surface (6
, 10
, 16
, 17)
. However, a precise role for the vWF-collagen interaction has not formally been demonstrated in the context of human platelet aggregation in vivo. Our present knowledge about the topographical organization of the subendothelial ECM would suggest a rather limited binding capability of vWF to the Col I/III component of the fibrillar network, since the surface of the collagen fibrils would not be readily accessible to the circulating vWF in vivo, being physically separated from the subendothelial zone by collagenous and elastic microfilament networks (18
, 19)
and covered by decorin (20)
. Conversely, recent studies from our group indicate that Col VI—a primary element of the above-mentioned anchoring microfilamentous network—may represent a more genuine subendothelial ligand for vWF, as it binds with the same or higher affinity to the glycoprotein and allows it to arrest platelets at equivalent high shear rates (21)
.
Although platelet aggregation via adhesion to vWF-Col VI complexes seems to be the prevailing triggering factor for the human thrombotic cascade under high shear stress, the modes of platelet tethering at low shear rates, such as those found in venules and atherosclerotic coronary arteries, are less well understood. In these normal and pathological vascular beds, platelets may bind directly to Col VI through their collagen receptors 2ß1 (M. Mazzucato et al., unpublished results) and GPVI (22
, 23)
or may interact with basement membrane components such as laminin-8 (24)
, Col IV (25)
, and fibulin-1 (26)
. However, the poor affinity of integrins for their cognate ligands and the low-affinity interaction of fibulin-1 with fibrinogen do not warrant an exclusive role for these putative interactions in low flow. Thus, it is plausible that other components of the subendothelial ECM, which may particularly enriched in certain disease conditions, may be decisive for the site-specific arrest of flowing platelets at lower shear rates. Proteoglycans (PGs) appear as candidate molecules for the platelet-subendothelial ECM interplay as they are markedly concentrated in the subendothelial zone and are strongly up-regulated in endothelial and nonendothelial cells on injury, restenosis, and atherosclerotic formations (27
28
29
30
31)
. We have explored this possibility by assessing the ability of vascular and nonvascular PGs to induce platelet adhesion under dynamic conditions and examining in greater detail the platelet-PG interaction in vitro.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
) can be summarized as follows: mAb 4D1 reacts with native chondroitin-4-sulfate (Ch-4-S); mAbs 5C12, 2C12, 4G5, and 2G5 recognize partially fucosylated O-linked oligosaccharide epitopes prevailing on versicans, weakly expressed on aggrecans, but not found on other PGs tested in this study; mAbs 8D5 and 9G7 are directed against fucosyl capping structures of N- (8D5) and O-linked (9G7) oligosaccharide moieties uniquely expressed by vascular bovine versicans; mAb 2E12 reacts with N-linked oligosaccharides specific of versicans; mAb 4C5 is directed against an unidentified versican core protein epitope whereas mAb 2B3 reacts with a chondroitinase ABC-sensitive epitope on versicans and aggrecans. Several of these mAbs shows species-specific immunoreactivity apparently recognizing solely bovine versicans. Other antibodies used in this study were obtained as follows: an anti-versican antiserum (denoted here as anti-PGM) from Dick Heinegård (University of Lund, Sweden); versican-reactive mAbs 5D5 and 10C2 from Firoz Rahemtulla (University of Alabama, Birmingham, AL; ref 32
); mAb 10E4 against heparan sulfate from Guido David (Center for Human Genetics, University of Leuven, Belgium); mAb 7D5 against human perlecan from Renato Iozzo (Department of Pathology and cell Biology, The Thomas Jefferson University, Philadelphia, PA); mAb MO227 against disulfated chondroitin (ChS-D) and mAb 6B6 against decorin from Seikagaku Corporation (Japan); mAb 373E1 against keratan sulfate (KS) was produced by immunization with a pool of embryonic avian PGs (32)
; mAbs 5D4 against KS, CS56 against CS, 2B6 against Ch-4-S, 1B5 against unsulfated chondroitin and 3B3 against Ch-6-S from Sigma-Aldrich, Milan, Italy; mAb 6C3 against native Ch-6-S from Michael Sorrell (Cleveland, OH); an antiserum against CS, which was determined here to be directed against Ch-6-S and to a lesser extent Ch-4-S, from Chemicon International (Temecula, CA); mAbs R2–7E4 against the 2ß1 integrin, LJ-CP8 against the IIbß3 integrin and LJ-1b1 against the GPIb and rvWF445-733 were received from Zaverio Ruggeri (Scripps Research Institute, La Jolla, CA).
Purification and partial characterization of aortic versicans
The procedure for isolating intimal aortic tissue was as previously described (33
34
35
36)
. The descending thoracic portion of bovine aortas was obtained from the local slaughter house and kept on ice. Aortas were placed in a bacteriological dish and the internal part of the vessel was rinsed extensively with cold PBS. It was placed on a microsurgery plate and the internal layer was finely dissected away from the central media using surgical scalpels and the occasional addition of a stereomicroscope. Segments of thoracic human aortas were obtained from routine autopsies performed at the National Cancer Institute of Aviano and treated in a similar way as the bovine one. No care was taken in either case to remove atherosclerotic plaques or similar sclerotic formations. Versicans were extracted from these arteries by (modified) described procedures (33
34
35
36
; Fig. 1
). Purification steps were monitored by a combination of OD measurements at 206, 260, and 280 nm, immunochemical assays, and SDS-PAGE involving silver/silver-Alcian blue staining. Dissected tissues were homogenized on ice and extracted for 20–24 h at 4°C with 4M GuHCl, 50 mM acetate buffer, pH 5.8, containing 1 mM PMSF, 10 mM each of NEM, caproic acid, benzamidine, and EDTA, and 0.2% Triton X-100. The extract was centrifuged at 14,000 g for 40 min and the supernatant was collected, dialyzed against 50 mM Tris-HCl, pH 8.0, containing 7M urea, 2 mM MgCl2, 0.2% Triton X-100, and the same protease inhibitors, and digested with 50 µL benzoate (Roche) for 1 h at 37°C to allow for efficient removal of contaminating genomic material. The digest was redialyzed and loaded onto a Sephadex G-50 column connected directly to DEAE-Sepharose Fast Flow column (Amersham-Pharmacia Biotech, Little Chalfont, UK). After loading of the sample, the two columns were disconnected and the DEAE-Sepharose column was washed with 5 volumes of washing buffer composed of 50 mM Tris-HCl, pH 8.0, 7M urea, 0.4 M NaCl, 5 mM EDTA, 10 mM aminocaproic acid, and 10 mM benzamidine. The column was eluted with 6M GuHCl in 50 mM acetate buffer, pH 5.8, 50 mM EDTA, 0.1% Triton X-100, and the mixture of protease inhibitors. Fractions containing the PGs of interest were selected by ELISA using mAbs 5D5, 5C12, 2C12, and CS56. Immunoreactive fractions were pooled and loaded onto CL-2B Sepharose columns equilibrated with Tris-urea washing buffer and eluted with the same buffer. PG-containing fractions were pooled and loaded onto DEAE-Sepharose columns eluted with a linear 0.3–2.0 M NaCl gradient in 20 mM Tris-HCl, pH 8.0, and 7M urea. PGs separated by the NaCl gradient were examined by ELISA/dot-blot, SDS-PAGE, SDS-agarose gel electrophoresis, and Western blot using the anti-versican antiserum; mAbs 5D5, 4G5, 2C12, and 5C12. Immunoreactive fractions were further pooled, dialyzed against 10 mM piperazine/perchlorate buffer, pH 5.0, containing 6M urea and 0.1% CHAPS, and further separated by chromatography on a Mono-Q column eluted with a linear 0–0.5M gradient of LiClO4 in the same buffer (Fig. 1)
. Positive fractions were redialyzed against 4M GuHCl and loaded onto Zn2+ chelating columns (Bio-Rad, Hercules, CA) eluted with a pH gradient of 3.5–8.1 already described (37)
. Comparative chromatographic analyses of versican-containing pools enriched by strong ion exchange chromatography on octyl-Sepharose, Sepharose-coupled mannose, and Zn2+-Sepharose showed that the latter resin/chromatographic procedure was the most efficient in achieving a final separation of intimal versicans. For a closer determination of the glycosylation traits of the aorta versicans isolated in this study, PGs were analyzed by ELISA, dot-blot, and Western blot using the battery of antibodies described above in conjunction with single or combined differential digestions with chondroitinases and deglycosylating enzymes, including neuraminidase (from several bacterial sources), endoglycosidase H/F, fucosidase, ß-galactosidase, and N-acetyl-ß-D-glucosaminidase.
|
Other PGs and ECM molecules
Versicans were isolated from human follicular fluid and embryonic chick fibroblasts as described previously (32
, 38)
. A mixture of chick embryonic brain PGs, including versican, neurocan, phosphocan, and purified neurocan, were purchased from Chemicon International. Aggrecans from bovine nasal, chick embryonic, and human articular cartilage, and bovine sclera were obtained according to standard procedures for purification of these PGs. Bovine and human tendon fibromodulin and bovine cartilage link protein were received from Dick Heinegård. Decorin and biglycan from bovine cartilage were purchased from Sigma-Aldrich and rat chondrosarcoma aggrecan from ICN Biochemicals (Irvine, CA). Mimecan, keratocan, and lumican from bovine cornea were obtained from Dr. Jim Funderburgh (University of Pennsylvania, Pittsburgh, PA). Decorin and biglycan were purified in the laboratory from bovine corneas according to previously published procedures (39)
. Human and chick Col VI tetramers were purified as described previously (21)
. High Mr hyaluronan, rat tail Col I, human skin Col III, human placental Col V and human plasma fibronectin were obtained from Sigma and Collaborative Research Biochemicals (Bedford, MA). A recombinant fragment corresponding to the GPIb binding site of vWF (rvWF445-733) was obtained from Zaverio Ruggeri. Well-characterized KS preparations from bovine cornea and skeletal cartilage were received from Ian Niedusinsky (Biological Sciences, Lancaster University, UK). DS from porcine intestinal mucosa, heparan sulfate (HS), low and high Mr heparin, Ch-4-S (bovine trachea), and Ch-6-S (shark cartilage) were purchased from Seikagaku Corporation and Sigma-Aldrich. DS was purified from bovine cornea according to previously published protocols (40)
and separated by gel permeation chromatography. Three different preparations denoted DS-18, DS-36, and DS-50 and having the respective Mr of 17,500, 14,500, and 25,000 daltons were isolated, examined in detail, and established to have compositions in terms of the relative content of hexosamine, uronic acid, D-iduronic acid (D-IdoA), 2-sulfated D-iduronic acid (D-IdoA-2-SO4), and total sulfation (Table 1
).
|
Glycosylation traits of versicans from various tissue sources as well as traits of other PGs used in this study were determined by ELISA and dot-blot using combined digestions with chondroitinase ABC, keratinase I, endo-ß-galactosidase, endoglycosidase H, O-glycosidase, N-glycosidase F, -1,3/4-L-fucosidase, N-acetyl-ß-D-glucosaminidase, various neuraminidases (Roche, ICN Biochemicals, Seikagaku Corporation, Sigma-Aldrich), and the panel of antibodies described above. These assays highlighted marked differences in the oligosaccharide content/composition of the various versicans (not further discussed here) and gave insight into the diversity in their GAG chain composition/structure, reported below.
SDS-PAGE, SDS-agarose gel electrophoresis, and Western blot
Versicans were analyzed in their intact and enzyme-treated form by agarose gel electrophoresis in the presence of SDS, silver/Alcian blue/silver-Alcian blue staining as previously detailed (32)
, SDS-PAGE on 3–8% continuous gradient gels, and 3–6-10% step gradient gels, and by Western blot using the panel of anti-PG and anti-GAG antibodies. For enzyme treatments, versicans were dialyzed against 50 mM Tris-HCl, pH 7.6 (for chondroitinase ABC, ACII, and heparitinase III) or 50 mM Na-acetate buffer, pH 5.8 (for keratinase I and endo-ß-galactosidase) and alternatively digested for 1 h at 37°C with 0.3 U/mL of chondroitinase ABC, 0.3 U/mL of chondroitinase ACII, 0.01 U/mL of keratinase I, 0.01 U/mL of endo-ß-galactosidase, and 0.02 U/mL of heparitinase III. After SDS-PAGE or SDS-agarose electrophoresis, versicans to be immunoblotted were electrolytically transferred onto nylon (Zeta Probe, Bio-Rad), PVDF (Millipore, Bedford, MA), or nitrocellulose (Schleicher & Schuell, Keene, NH) membranes in standard transfer buffer at pH 6.5 for agarose gels and pH 9.0 for polyacrylamide gels. Membranes were saturated with 4% -casein in 0.1M bicarbonate buffer, pH 8.8, containing 1% PVP 40,000 (Sigma) and 10 mM EDTA for 2–12 h at room temperature and incubated with primary antibodies diluted to 1:2 (for supernatants) or 1:1000 for ascites fluids and purified IgGs using the same solution as above containing 1% -casein and 0.15M NaCl. After extensive washings with the antibody incubation buffer supplemented with 0.3% Tween-20, membranes were further incubated with secondary antibodies at 1:1000 dilution for 1 h at room temperature, washed, and processed for chemiluminescent detection using the ECL-Plus chemiluminescence kit (Amersham-Pharmacia). Apparent Mrs above the 205 kDa band of myosin II were determined in parallel run Coomassie blue-stained gels containing reduced and nonreduced EHS laminin-1 (
400 kDa for the 1 chain; 1000 kDa for the heterotrimer), intact collagen type VI tetramers (2400 kDa), a purified IgM-class mouse immunoglobulin (900 kDa), fibronectin (440 kDa), and gel electrophoresis/gel filtration standards (Sigma) comprising thyroglobulin (669 kDa), ferritin (440 kDa), ß-catalase (232 kDa), lactate dehydrogenase (140 kDa), ß-galactosidase (106 kDa), phosphorylase (97 kDa), and BSA (67 kDa).
Ultrastructural analysis of versicans
Purified versicans from various species/tissue sources were examined by transmission electron microscopy involving rotary shadowing as described previously (31
, 35
, 38)
. PG samples (final concentration of 10 µg/mL) were dialyzed overnight against 0.2M ammonium hydrogen carbonate, pH 7.9, in a microdialysis apparatus (5–50 µL, Biowerk), mixed with an equal volume of 80% glycerol and prepared for ultrastructural analysis as described (35)
. Specimens were dried at high vacuum for 1–2 h and rotary shadowed at a 9° angle with platinum/carbon, followed by coating with carbon at a 90° angle. Replicas were floated onto distilled water and picked up on 400 mesh copper grids. Electron micrographs were taken on a Zeiss 10 transmission electron microscope operated at 80 kV accelerating voltage.
Preparation of PG and collagen substrates
A 5 mm-wide central area of rectangular glass coverslips was coated at 4°C overnight with 0–30 µg/mL of various PGs and purified collagens were dissolved in 0.05M bicarbonate buffer, pH 9.6. Saturation of the coated area has been shown to be reached at a coating concentration of 20 µg/mL (21
, 32)
. Coated coverslips were extensively washed, saturated with 1% BSA in the same buffer for 1–2 h at room temperature, and mounted in the parallel flow chamber (see below). Mixed collagen/PG substrates were prepared by incorporating 1–50 µg/mL of selected PGs into the nonpolymerized Col I, III, or V a concentration of 600-1000 µg/mL in PBS, dispensed onto a central area of a coverslip, and allowed to solidify at 37°C for 1 h. In some cases, PG substrates were digested for 2 h at room temperature with chondroitinase ABC, ACII, or B (0.3 U/mL in 50 mM Tris-HCl buffer, pH 7.6), heparitinase III (0.01 U/mL in PBS), endo-ß-galactosidase (0.01 U/mL in 50 mM Na-acetate buffer, pH 5.8), endoglycosidase H/F (0.01 U/mL in 50 mM Na-acetate buffer, pH 5.8), or N-glycanase F (0.01 U/mL in 50 mM Na-citrate buffer with 10% glycerol, pH 5.0). Alternatively, they were chemically desulfated by treatment with 50 mM HCl with 100% methanol for 4 h at room temperature or preincubated with various concentrations of several of the anti-versican and anti-GAG antibodies described above. After these treatments, substrates were extensively washed with PBS before use.
Preparation of human platelets
Blood from healthy donors was obtained after informed consent. All donors denied ingestion of drugs known to interfere with platelet function for a period of at least 2 wk before blood sampling. Blood was collected from the antecubital vein through a 18 gauge needle into syringes containing 400 Units/mL (final concentration) of the thrombin antagonist Hirudin (Iketon, Milan, Italy) as anticoagulant.
Perfusion experiments
The experimental setup for the analysis of platelet substrate adhesion under defined shear stress rates has been described in detail (see ref 21
) and involves the use of a modified Richardson’s parallel flow chamber mounted on an inverted microscope. This was equipped with epi-fluorescent illumination (Diaphot-TMD; Nikon), appropriate fluorescence, and projection lenses, and an intensified CCD video camera (C-2400-87, Hamamatsu Photonics Inc). Photoactivation of platelets by irradiating UV light was prevented by the simultaneous use of two neutral blocking filters with the codes ND32 and ND8 (Nikon). Mepacrine-labeled platelets (5–8 µM) were perfused in 5 mL hirudinated whole blood over a 0.037 substrate area and with a flow path height of 125 µm, as determined by a silicon gasket and after normalization of the volume of flushed blood under each perfusion of 1 min duration. The fluorescent dye mepacrine used to tag live platelets labels leukocytes and thereby allows for the simultaneous monitoring of the behavior of these cells, which can readily be distinguished morphologically from platelets by their larger size and nuclear traits. Perfusion experiments were performed in the presence or absence of 50 µg/mL of mAb R2–7E4 against the 2ß1 integrin, 100 µg/mL of mAbs LJ-CP8, and LJ-Ib1 against IIbß3 integrin and GPIb, respectively, 1 mM EGTA or 100 µg/mL of fragment rvWF445–733, and varying concentrations of purified GAGs.
Image processing and quantitative analysis of the platelet-PG interaction
Images corresponding to representative areas of the substrate were either captured in real-time during perfusion of platelets or retrieved from videotape recordings of the experiments performed at a sampling rate of 25 frames/s by using a computer workstation mounting a TARGA-2000 PLUS board (Truevision). To classify the attached platelets, determine the location along the x and y coordinates of the immobilized ones, estimate their number, measure the surface area covered by the substrate-bound ones, assess the average velocity of substrate dislodgment, and quantify the duration of platelet-substrate contacts, images of predefined optical fields (0.037 mm) (2)
were processed using a specifically designed custom-made software (Casti Imaging, Venice, Italy). Software tracked the area of single platelets and determined the position of the corresponding centroid on all frames collected at a sampling rate of 25/s, i.e., data for measured parameters were obtained every 0.04 s and automatically calculated the relative and mean displacement velocity of up to 150 platelets analyzed simultaneously for each experimental situation (3–10 cases for each experimental condition). Motionless time intervals were calculated as the sum of all frames in which a platelet had a velocity equal to 0 µm/s, i.e., was not displaced from its position according do a previously established definition (6)
. Stable adhesion was hence defined as a velocity of 0 µm/s protracted for 
30 s. In this case, the surface imprint of a platelet in the first frame of the observation period was overlapping at least partially with the imprints in all subsequent 749 frames superimposed to monitor trajectories and velocities of displacing platelets. To define aggregate size, assuming that the average platelet radius is 1.2 µm (measured radius range=0.8–2 µm), substrate-bound particles were arbitrarily deemed as being composed of single platelets (2–13 µm2), microaggregates (13.1–50 µm2), small aggregates (50.1–200 µm2), or large aggregates (>200 µm2). Statistical analysis was carried out by Student’s t test with significance set at P < 0.05.
Surface labeling and affinity chromatography of biotinylated and unlabeled platelet membranes
For surface labeling with biotin (41)
, platelets obtained from ACD-anti-coagulated blood (1:6 v/v) were treated with 1 µL/10 mL plasma of prostaglandin E1, washed extensively in PBS-EDTA (0.02M Na2HPO4, 0.135M NaCl, 2.7 mM KCl, 10 mM EDTA, pH 7.0), resuspended at a density of 109/mL in Tyroid buffer containing 10 mM EDTA, 5 mM NEM, and 1 mM PMSF, and incubated with 12 mM NaIO4 at 0°C for 10 min. Oxidation was blocked by addition of 0.6M glycerol in PBS; platelets were washed twice in PBS-EDTA and incubated with 3 mM biotin-LC-hydrazide (Pierce, Rockford, IL) for 2 h at 22°C. Biotinylated platelets were repeatedly washed by centrifugation, resuspended at a density of 5 x 108/mL in TBS (0.05M Tris-HCl, pH 7.4, 0.15M NaCl) containing 10 mM EDTA, 5 mM NEM, 1 mM PSMF, 200 µg/mL leupeptin, and solubilized with 0.5% Igepal. Insoluble material was removed by centrifugation and platelet-associated IgGs were removed by overnight incubation of the biotinylated platelet membrane preparations with protein G-agarose (Amersham-Pharmacia Biotech). Biotinylated membranes were resuspended in TBS containing 0.5% Igepal and 0.1% BSA; the sample was centrifuged at 5000 g and sequentially applied onto epoxy-activated Sepharose columns (Amersham-Pharmacia) containing immobilized dextran sulfate (Mr 70,000; Sigma) and bovine corneal KS to assure complete elimination of surface components that could bind nonspecifically to highly charged, sulfated GAGs. The flow-through was subsequently applied to a similar column containing immobilized DS and equilibrated in PBS. According to an alternative protocol yielding equivalent results, biotinylated platelet membranes were similarly "precleaned" on dextran sulfate and KS and eventually chromatographed on Sepharose 4B columns (Amersham-Pharmacia Biotech) containing EDC [N-ethyl-N'-(3-dimethyl-aminopropyl)] carbodiimide hydrochloride-coupled DS and prepared according to the manufacturer’s instructions. In both types of chromatographies, bound material was eluted with a linear 0.15–1.0 M NaCl gradient in TBS, pH 7.4 or 8.6 (the latter pH yielding superior results) and dialyzed extensively against TBS. Eluted material was analyzed by SDS-PAGE on 3–10% gradient gels or fixed 7% or 8% gels under reducing and nonreducing conditions. Separated bands were either directly visualized by silver staining or electrotransferred onto nitrocellulose membranes and detected by chemiluminescence by incubation with streptavidin-HRP (Amersham-Pharmacia Biotech) and the chemiluminescent PLUS kit. Biotinylated molecular weight standards were purchased from Bio-Rad and simultaneously visualized by chemiluminescence.
Solid-phase binding assays
Microtiter plates were coated with 0.5 µg/mL bovine vascular, human vascular, ovarian follicular fluid versican, human cartilage aggrecan or decorin diluted in 0.05M bicarbonate buffer, pH 9.6, overnight. Plates were saturated with 2% BSA in the same buffer for 2 h at room temperature and incubated with decreasing concentrations of the biotinylated platelet surface material eluted from DS affinity columns and diluted in PBS with 1 mM CaCl2 at 4°C overnight. Before incubation with PG substrates, the material was concentrated with PEG (Sigma) for 2 h at room temperature or overnight at 4°C to reach an adsorbance value at 280 nm of 0.01. Unbound material was washed off with PBS and the plates were incubated with streptavidin-HRP diluted 1:1000 in PBS, followed by further washing and detection of bound HRP as described previously (32)
.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
34
35
36)
, and equivalent yields were obtained in this study. However, our pilot immunochemical and ultrastructural analyses indicated that versicans isolated from intimae according to these previous protocols were strongly contaminated by other PGs and glycoproteins that copurified with them. Therefore, we designed an alternative purification scheme based on sequential gel filtrations, ion exchange, and Zn2+ chelating chromatography (Fig. 1)
. Combined digestion with lyases and immunoblotting with the anti-versican mAb 5D5 of the fractions eluted by strong ion exchange chromatography identified a heterogeneous population of PG molecules carrying glycan chains that were largely susceptible to digestion with chondroitinase ABC, but not to other GAG-degrading enzymes (not shown). Further digestions with deglycosylating enzymes showed they carried an excessive amount of complex oligosaccharide moieties, which resulted in large variations in the Mr of the isolated PGs. The predominant bulk of versicans eluted from these columns had an Mr estimated to be in the range of 900,000–1,700,000 daltons by agarose gel electrophoresis in which intact Col VI tetramers and EHS laminin-1 were resolved (not shown). Fractions containing versicans, but lacking contaminating PGs, eluted at an unexpectedly high LiClO4 concentration (i.e., 0.415 M; Fig. 1
) and could be further separated by a subsequent Zn2+ chelating chromatography. In fact, such a preparative chromatography step produced two distinct peaks (Fig. 1
, right panel, peaks 2 and 3) that, after step gradient SDS-PAGE under nonreducing conditions, appeared to be composed of macromolecules separable into two major pools: one penetrating the 3% stacking gel and estimated to have a relative Mr of 1,000,000–1,200,000 daltons by agarose gel electrophoresis and the other unable to penetrate the 3% stacking gel and estimated to have an Mr of >1,500,000 daltons (Fig. 1
, lower right). After exposure by chondroitinase ABC digestion, the core proteins of these PGs had a relative Mr compatible with those of isoforms V1 and V2 (Fig. 1
, lower right).
Versicans purified from human aortas showed a similar molecular profile, with two apparent populations, but had slightly different glycanation traits as highlighted by ELISA (not shown) and Western blot using various anti-versican and anti-GAG antibodies (Fig. 2
and data not shown). After chondroitinase ABC digestion, core proteins with an apparent Mr of 460,000 and 310,000 daltons were detected by Western blot with mAb 5D5 (Fig. 2)
. To define the relative isoform composition of the vascular versicans isolated according to the present purification scheme and relate this composition to that of versicans from other sources comparatively assayed in this study, we used TEM/rotary shadowing techniques (32
, 35
, 38)
. This ultrastructural analysis confirmed the purity and retained intactness of the isolated versicans. By direct comparison with previously obtained ultrastructural data on the chick fibroblast V0 isoform (Fig. 3
; ref 32
) and vascular preparations (32
, 35)
, aortic versicans isolated here were found to be composed primarily of the V1 and V2 isoforms, with a slight prevalence of the former isoform (Figs. 1
, 3)
. In contrast, follicular versicans contained molecules with the predicted size of the V1 isoform (Fig. 3
; ref 38
). These findings combined with previously published data, predicted the following isoform composition of the versicans used throughout this study: V0/(V1) (chick fibroblasts), (V0)/V1/(V2) (ovarian follicular fluid), (V0)/(V1)/V2 (embryonic chick brain), and (V1)/V2 (bovine and human intimae), where isoforms indicated in parentheses correspond to the ones deemed to be less represented.
|
|
Vascular versicans isolated here were ascertained to be able to bind to hyaluronan and form well-organized ternary complexes in the concomitant presence of the GAG and link protein (Fig. 3)
. Hence, the PGs had retained at least one of their prevalent functional trait, suggesting they were not structurally and compositionally compromised. Apart from some divergences in the putative isoform ratio in versicans isolated from vascular and nonvascular tissues, major differences could be disclosed in their GAG (Fig. 2)
and oligosaccharide profiles. Thus, aortic versicans were strongly substituted with DS and Ch4S but carried no or only trace amounts of Ch0 (unsulfated chondroitin), Ch6S, ChSD (chondroitin sulfate D), or KS (Fig. 2)
. Thus, GAG composition of these versicans clearly differed from that of the follicular versican known to carry a large proportions of Ch0 (i.e., 54% and 42%, respectively, of the total GAG content; ref 38
) and chick embryonic versicans (32)
. On the other hand, Mr size variations of the intact PGs were consistent with those previously documented and ascribed to the relative abundance and complexity of their N- and O-linked oligosaccharide moieties (36
, 42
43
44)
. The presence of these moieties could be further confirmed here on the basis of the diverse immunoreactivity detected with the anti-oligosaccharide mAbs after differential enzyme digestions (data not shown).
Selective platelet adhesion to vascular versicans at low shear rates
Comparative analysis of the ability of flowing platelets to interact with vascular and nonvascular PGs, including those present in the subendothelial-pericyte and intimal-medial ECM, showed a remarkable specificity for vascular versicans. Nonactivated platelets bound well to bovine and human aorta versicans, weakly to bovine and human decorin and biglycan, but failed to adhere to follicular versican, various aggrecans, other brain, cartilage and corneal PGs, and hyaluronan (Table 2
; Fig. 4
). At maximal coating concentrations, platelet interaction with aortic versicans resulted in a significant substrate tethering frequency, but did not lead to aggregation of the arrested platelets as in the case of collagen-bound platelets (Table 2
; Fig. 5
A, B). Thus, perfused platelets were noted to cover the PG surface rather homogeneously but remained bound as single bodies (Fig. 4)
.
|
|
|
A shear force-dependent analysis of platelet adhesion to vascular versicans further showed that maximal tethering frequency occurred at relatively low shear forces (i.e., 100 s-1) and sharply declined with increasing forces, reaching a >50% reduction at a force of 350 s-1 (Fig. 5A
). Real-time assessment of the relative strength with which single platelets adhered to the versicans at different shear forces demonstrated that binding avidity was markedly reduced with increasing shear rates, being < 20% (of the maximum adhesive capability) at a rate of 1000 s-1 (Fig. 5A
). These findings underscored that platelet interaction with vascular versicans was optimal at rheological conditions characteristic of large arteries and veins, smaller venular vessels, and at the distal portion of atherosclerotic plaques in coronary arteries where the flow rate has been estimated to be in the range of 2.5 dynes/cm2 (44
, 45)
. After initial contact with the versican substrate, 15% of adhering platelets were seen to translocate with a mean velocity of 1.05 ± 0.1 µm/s. Conversely, the majority of the firmly bound platelets remained motionless for undefined periods under the continuous shear rate, though not displaying noticeable shape changes. Platelets in free flow (i.e., contained in the laminar flow immediately above the substrate surface) were estimated to move with a mean velocity of 242 ± 11 µm/s.
For a more nearly accurate characterization of the modes of platelet-versican interaction under shear stress, we performed a time lapse analysis of platelet adhesion to versican substrates at a rate of 100 s-1. This demonstrated that 70% of the platelets forming contact with the substrate detached from it within 5 s of continuous exposure to the applied shear force (Fig. 5B
), whereas <10% of the total amount of platelets analyzed remained anchored to the substrate (Fig. 5B
). The interaction was restricted to single platelets rather than involving cooperative effects exerted by multiple platelets linking to the substrate simultaneously. Finally, leukocytes contained in the perfused blood samples adhered to the versican substrates under the flow conditions used before (Fig. 5B
). In fact, these cells established substrate contacts and exhibited a rolling motion with a mean velocity of 2.67 ± 0.13 µm/s (Fig. 5B
). Detailed examination of the recorded images failed to show stable contact between the flowing platelets and leukocytes in that both cell types bound to versican substrates independent of each other, further supporting the lack of involvement of P-selectin in the process (47)
. Similarly, there was no evidence of putative reciprocal interferences in the adhesive capability of platelets and leukocytes to versicans.
The platelet response to vascular versicans was distinguishable from the response to collagens by other criteria, supporting the idea that the former macromolecules could function as "platelet-capturing" subendothelial agents in the context of vascular injury. Reconstruction of the spatiotemporal movements of individual platelets perfused over versicans substrates at shear forces of < 350 s-1 showed that they established frequent contacts with the PG substrate, but the duration of these contacts was significantly shorter than that measured on collagen substrata (Table 2)
. Accordingly, the frequency of detachment of platelets bound to versicans was markedly high, indicating the formation of substrate contacts of a prevalent transient nature and a corresponding lower retention of bound platelets (Table 2)
.
Potentiation of platelet adhesion and aggregation on polymeric collagen substrata containing immobilized vascular versicans
In a second set of experiments, PGs were immobilized onto polymeric substrates of Col I, III, or V to assay their potential influence on collagen-induced platelet adhesion and aggregation. Both phenomena were observed in these assays and were of significantly greater magnitude than those measured on the collagens alone. Thus, at low shear forces, the number of platelet aggregates and their relative sizes in response to these mixed substrates were superior to those seen on either of the substrates alone (Fig. 6
). There was a >30% increase in surface coverage (i.e., total amount of platelets bound to the substrate) and a significant shift in the relative distribution of single platelets/smaller aggregates vs. large aggregates, with the latter prevailing (Fig. 6)
. Conversely, inclusion of cartilage aggrecans or follicular versicans (not shown) into polymeric Col I did not affect the overall amount of substrate-bound platelets or the relative frequency of single vs. micro-aggregated platelets. In contrast, it induced flowing platelets to form a relatively higher number of larger aggregates (Fig. 6)
. Analogous results were obtained in the presence of antibodies to GPIb and IIbß3 or when perfusing platelets resuspended in buffer lacking vWF and fibrinogen (not shown). Finally, in contrast to the situation where substrates contained versicans alone, a more pronounced accruing of activated leukocytes was observed in the mixed substrates and this seemed to favor an augmented capturing of the flowing platelets. Thus, leukocyte-platelet interaction clearly promoted thrombus growth on these specific collagen-PG substrates.
|
Putative mechanisms of platelet-versican interaction
Platelet binding to vascular versicans was not inhibited by addition of function-blocking antibodies to the vWF platelet receptor GPIb (Fig. 7
A, C, D), a recombinant fragment corresponding to the GPIb binding A1 domain of vWF (rvWF445-733); Fig. 7A
) function-blocking antibodies to the respective fibrinogen and collagen binding platelet integrins IIbß3 and 2ß1 (Fig. 7A, C, E
), or function-blocking antibodies to P-selectin (not shown). In fact, inclusion of some of these reagents in the perfusion chamber instead tended to increase the amount of platelets that bound to the versican substrata. In contrast, platelet tethering to the PGs was completely abrogated in the presence of EGTA (Fig. 7B
) as well as by addition of low and high Mr heparin (data not shown). These findings confirmed the GPIb- and IIbß3-independent nature of the platelet-versican interaction and asserted that it did not involve the primary collagen receptor of these cells. Observations indicated that platelet adhesion to versicans was Ca2+ dependent and mediated by the GAG chains of the PGs.
|
This latter observation prompted us to explore the platelet-GAG interaction in detail by using a combined enzymatic and competition approach involving GAG-degrading enzymes and purified GAGs of known composition. Digestion of aortic versican substrates with chondroitinase ACII (Fig. 7F
), heparitinase III, endo-ß-galactosidase, endoglycosidase H/F, or N-glycanase F (not shown) did not affect the tethering capability of platelets. In contrast, similar digestions with chondrotinases ABC or B markedly reduced platelet binding to the PGs (Fig. 7F
). In accordance with this observation, addition of increasing concentrations of a mixture of Ch4S and Ch6S only partially affected platelet adhesion to versicans, with a definite effect seen only at the highest concentrations. The individual efficiency of the two GAGs was not further investigated. Conversely, addition of a heterogeneous DS preparation strongly competed off platelet adhesion to the substrate (Fig. 7G
).
The inhibitory effect of DS was not dependent on the size of these GAG chains, their degree of iduronic acid sulfation, or their relative ratio of uronic/iduronic acid, because the competitions with three different well-characterized homogeneous preparations of DS (see Materials and Methods) did not yield significant differences when assayed at their maximal inhibitory concentration (i.e., 10 µg/mL; Fig. 7B
). A dose-dependent test of the blocking activity of different DS failed to disclose significant intrinsic differences (not shown). The lack of an overt effect of chemical desulfation of immobilized versicans on platelet adhesion (not shown) was consistent with a nonessential role of the sulfation degree of the DS moieties of versicans for this process. In further support of a prevailing role for the DS chains of aortic versicans were observations that preincubation of versican substrata with the anti-chondroitin sulfate antibodies CS56 and 4D1, as well as mAbs 5C12 and 2E12 recognizing oligosaccharide structures, did not perturb platelet adhesion (not shown). Similar negative results were obtained after various digestions of the immobilized vascular versicans with deglycosylating enzymes, as well as addition of competing glycan moieties (not shown). Thus, there was no indication of a cooperative participation of oligosaccharides in the binding of platelets to these PGs.
Identification of a putative DS binding membrane complex mediating vascular versican-platelet interaction
To identify the putative DS binding cell surface component(s) on human platelets responsible for the shear stress-dependent anchorage of platelets to the GAG chains of aortic versicans, platelet membranes were surface biotinylated and sequentially separated by affinity chromatography on columns containing immobilized KS and DS. Biotinylation of surface proteins allowed to distinguish between external and internal membrane-associated components (41)
that were nonspecifically enriched through GAG chromatography. Salt elution of the DS-bound material yielded molecules with an apparent Mr of 140,000 and 123,000 daltons on SDS-PAGE under reducing conditions and nonreducing conditions and an additional band at 98,000 daltons (Fig. 8
A). Since this latter band was eluted from KS columns it was deemed to not correspond to a DS-specific cell surface ligand. Solid-phase binding assays using vascular and nonvascular PG substrates and the biotinylated material eluted from DS columns showed a marked dose-dependent binding of the putative DS binding platelet membrane complex to aorta versicans, a limited reactivity to decorin, but no significant binding to cartilage aggrecan and the follicular fluid versican (Fig. 8B
).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Experimental elimination of both these molecules by gene deletion in mice does not abrogate thrombus formation (48)
, suggesting that other components of the subendothelium may cooperate with interstitial and microfilamentous collagens in arresting flowing platelets at the sites of normal and abnormal platelet accruing. In this study we have identified the PG versican as a candidate component available for a direct interplay between platelets and the microfibrillar network ECM juxtaposed to the endothelial cell layer (49)
, especially at the lower flow rates found in venular circuits, larger vessels, and in proximity of atherosclerotic plaques. Vascular-specific glycanation variants of the versican isoforms V1/V2 sustained human platelet adhesion under shear stress conditions, mimicking those found in the above type of vessels/conditions; when acting in concert with interstitial collagens, they were capable of enhancing thrombus growth.
The low-affinity platelet "capturing" ability of vascular versicans seemed to be a unique property of these macromolecules because other vascular and nonvascular PGs failed to fully reproduce this effect. These observations indicated that the interaction of platelets with vascular versicans was not merely due to a "nonspecific" cell binding favored by the pronounced negative charge of the macromolecules. As previously proposed (50
, 51)
, several of the nonvascular PGs appeared to be nonpermissive for platelet adhesion and aggregation. However, in contrast to these previous investigations, our data did not corroborate an overall inhibitory function of human cartilage aggrecans under shear stress conditions, though these PGs appeared to exert a modulating effect on the thrombogenic process. Similarly, our data on the platelet-PG interaction under well-defined dynamic conditions in vitro confuted the idea of an inhibitory effect of aorta-derived PGs other than versican on platelet aggregation (51)
. Furthermore, the adhesive interaction of flowing platelets by vascular versicans was not contributed by vWF, fibrinogen, or any of their platelet surface receptors. Finally, there was no direct evidence for "bridging phenomena" involving other ECM components and their putative cognate surface receptors, whereas the platelet binding activity of vascular versicans was cation dependent in a manner compatible with that reported for cell surface interactions with protein- and glycan-type molecular structures.
The notion of a mere cell-arresting function of vascular versican was further sustained by the lack of an overt platelet activation and aggregation after establishment of a stable contact with these PGs. Hence, no evident formation of conspicuous platelet aggregates or intracellular Ca2+ signaling could be detected in single versican-immobilized platelets or smaller platelet clusters. Moreover, in accordance with the prerequired activation of platelets to bring through surface expression of P-selectin and the lack of an effect of anti-selectin antibodies, but despite the reported ability of selectins to bind to certain tumor-derived versicans (53)
, there was no evidence of an involvement of these surface components in the platelet-versican interaction. Similarly, since CD44 is not expressed by human platelets (as suggested by the lack of platelet binding to soluble and immobilized hyaluronan), there was no possibility that this surface receptor could be one implicated in versican binding. However, since in mixed substrates containing versicans and fibrillar collagens, platelet adherence and aggregation were potentiated under low and high shear forces, the possibility remains that de novo expressed surface selectin could in that case participate in regulating platelet aggregate size (54)
. Since the cooperative effect of vascular versicans and collagens was not reproduced by combinations of collagen and follicular versican or aggregans, it is unlikely it could simply be attributed to a structural alteration of the collagen substrate itself. However, it is not possible at this time to exclude that vascular versicans embodied in a 3D substrate could assume a more favorable configuration for platelet binding than when directly immobilized in a putative 2D configuration onto a solid support.
Digestion of bovine aortic versicans with some lyases demonstrated that flowing platelets bound preferentially to the abundant DS chains of these PG variants. Similarly, evidence was obtained in competition experiments with soluble GAGs in which various preparations of Ch4S and Ch6S were found to be active in perturbing platelet binding to the versican substrates only at high concentrations. Conversely, purified DS was significantly more effective in competing off binding. Comparison of DS preparations with different degrees of sulfation and iduronic acid contents had equally efficient inhibitory activities, indicating that platelet interaction was not dependent on those properties of the GAG. This latter finding was supported by the lack of effects of chemical desulfation of the substrates or preblockade of the versicans with anti-Ch4S, anti-Ch6S, and anti-oligosaccharide antibodies.
A certain permissive function in platelet adhesion under flow was noted for other DS-containing PGs known to be similarly enriched in normal and diseased blood vessels, such as decorin and biglycan. It is unclear whether the markedly superior activity of vascular versicans was a pure quantitative effect, i.e., the presentation of more binding sites on their DS chains, or whether it was attributed to qualitative differences in the structural and/or molecular organization of these chains. An additional possibility would be that DS chains of immobilized versicans assumed a more favorable conformation for platelet interaction, which, however, was apparently different from that assumed by DS directly bound to a modified solid support (55)
.
Affinity chromatography on immobilized DS identified a platelet membrane complex composed of two primary polypeptides with an estimated Mr of 120,000 and 140,000. These bound in a dose-dependent and specific manner to immobilized vascular versicans when tested in solid-phase binding assays. The identity of this putative DS platelet receptor complex is so far unknown. However, consistent with the platelet-PG binding data, the components were not recognized by anti-selectin or anti-integrin antibodies (not shown).
Our data are apparently discordant with those of Baumann et al. (55)
, who proposed a high platelet binding efficiency of all GAG types when chemically immobilized onto a cationized cellulose membrane. We believe that this discrepancy could be associated with the artificial nature of the substrates assayed by these investigators and the more physiological traits of intact macromolecules tested in this study. The findings reported in both of these earlier studies may seem to be in direct contrast to the inhibitory effect of DS observed in fibrin-rich thrombus formation in vivo (56)
. On the other hand, in this previous study, high concentrations of soluble DS were used in systemic administrations to a nonprimate thrombosis model in which fibrin-rich venular-type thrombus formation was reproduced in a subcutaneously implanted device. Thus, the finding reported in this earlier investigation may after all be in perfect agreement with our results indicating that soluble DS is a potent competitor of the platelet-versican interaction at low flow rates.
The significance of the low-affinity platelet-arresting function of versicans in conditions of vascular damage may be twofold. First, as discussed above, other ECM components, including microfilamentous and fibrillar collagens, may not be sufficiently accessible for an efficient platelet binding as would the exceptionally long GAG chains of vascular versicans. Second, in low blood flow, there may be a need for a two-phase interaction of platelets with the thrombogenic substrate involving versicans in the first step and collagens in the second. Such a sequential interaction would resemble that mediated by selectins and integrins during extravasation of leukocytes. Third, there is substantial information regarding the concomitant process of ECM remodeling and abnormal platelet accumulation during arterial restenosis and atherosclerosis, especially after plaque rupture (57)
. An increased deposition of versicans with potentially altered glycosylation traits has been observed in the afflicted vascular tissues (27
28
29
30
31
, 42
43
44
, 58
59
60)
, and DS chains of these PGs may cooperate with those of decorin in sequestering low density lipoproteins (61)
. Thus, under pathological cardiovascular conditions, the thrombotic role of versicans may be aberrantly accentuated, providing a putative contribution to the increased accumulation of platelets in intimal lesions of atherosclerotic vessels. Ultimately, it may be envisioned that if some of the versican molecules produced by endothelial cells are secreted at their luminal surface (e.g., in compromised endothelium manifesting a loss of cell polarization), interaction with these PGs may entrap flowing platelets in juxtaposition of a presumptive injury site and thereby create the necessary local reservoir of platelets at that site.
|
ACKNOWLEDGMENTS |
|---|
Received for publication April 22, 2002. Accepted for publication July 30, 2002.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
vß3 on platelets and lymphocytes bind
