Endothelial Cell Surface Heparan Sulfate Research Articles

Fibroblast Growth Factor-2 Binding to Heparan Sulfate Proteoglycans Varies with Shear Stress in Flow-Adapted Cells.

Fibroblast growth factor 2 (FGF2), an important regulator of angiogenesis, binds to endothelial cell (EC) surface FGF receptors (FGFRs) and heparan sulfate proteoglycans (HSPGs). FGF2 binding kinetics have been predominantly studied in static culture; however, the endothelium is constantly exposed to flow which may affect FGF2 binding. We therefore used experimental and computational techniques to study how EC FGF2 binding changes in flow. ECs adapted to 24h of flow demonstrated biphasic FGF2-HSPG binding, with FGF2-HSPG complexes increasing up to 20dynes/cm2 shear stress and then decreasing at higher shear stresses. To understand how adaptive EC surface remodeling in response to shear stress may affect FGF2 binding to FGFR and HSPG, we implemented a computational model to predict the relative effects of flow-induced surface receptor changes. We then fit the computational model to the experimental data using relationships between HSPG availability and FGF2-HSPG dissociation and flow that were developed from a basement membrane study, as well as including HSPG production. These studies suggest that FGF2 binding kinetics are altered in flow-adapted ECs due to changes in cell surface receptor quantity, availability, and binding kinetics, which may affect cell growth factor response.

ESCRT-I Protein Tsg101 Plays a Role in the Post-macropinocytic Trafficking and Infection of Endothelial Cells by Kaposi's Sarcoma-Associated Herpesvirus.

Kaposi’s sarcoma-associated herpesvirus (KSHV) binding to the endothelial cell surface heparan sulfate is followed by sequential interactions with α3β1, αVβ3 and αVβ5 integrins and Ephrin A2 receptor tyrosine kinase (EphA2R). These interactions activate host cell pre-existing FAK, Src, PI3-K and RhoGTPase signaling cascades, c-Cbl mediated ubiquitination of receptors, recruitment of CIB1, p130Cas and Crk adaptor molecules, and membrane bleb formation leading to lipid raft dependent macropinocytosis of KSHV into human microvascular dermal endothelial (HMVEC-d) cells. The Endosomal Sorting Complexes Required for Transport (ESCRT) proteins, ESCRT-0, -I, -II, and–III, play a central role in clathrin-mediated internalized ubiquitinated receptor endosomal trafficking and sorting. ESCRT proteins have also been shown to play roles in viral egress. We have recently shown that ESCRT-0 component Hrs protein associates with the plasma membrane during macropinocytosis and mediates KSHV entry via ROCK1 mediated phosphorylation of NHE1 and local membrane pH change. Here, we demonstrate that the ESCRT-I complex Tsg101 protein also participates in the macropinocytosis of KSHV and plays a role in KSHV trafficking. Knockdown of Tsg101 did not affect virus entry in HMVEC-d and human umbilical vein endothelial (HUVEC) cells but significantly inhibited the KSHV genome entry into the nucleus and consequently viral gene expression in these cells. Double and triple immunofluorescence, proximity ligation immunofluorescence and co-immuoprecipitation studies revealed the association of Tsg101 with the KSHV containing macropinosomes, and increased levels of Tsg101 association/interactions with EphA2R, c-Cbl, p130Cas and Crk signal molecules, as well as with upstream and downstream ESCRT components such as Hrs (ESCRT-0), EAP45 (ESCRT-II), CHMP6 (ESCRT-III) and CHMP5 (ESCRT-III) in the KSHV infected cells. Tsg101 was also associated with early (Rab5) and late endosomal (Rab7) stages of KSHV intracellular trafficking, and CHMP5 (ESCRT-III) was also associated with Rab 5 and Rab 7. Knockdown of Tsg101 significantly inhibited the transition of virus from early to late endosomes. Collectively, our studies reveal that Tsg101 plays a role in the trafficking of macropinocytosed KSHV in the endothelial cells which is essential for the successful viral genome delivery into the nucleus, viral gene expression and infection. Thus, ESCRT molecules could serve as therapeutic targets to combat KSHV infection.

C-Cbl-Mediated Selective Virus-Receptor Translocations into Lipid Rafts Regulate Productive Kaposi's Sarcoma-Associated Herpesvirus Infection in Endothelial Cells

During target cell entry and infection, many enveloped and nonenveloped viruses utilize cell surface receptors that translocate into lipid rafts (LRs). However, the mechanism behind this translocation is not known. Kaposi's sarcoma-associated herpesvirus (KSHV) interacts with the human microvascular dermal endothelial (HMVEC-d) cell surface heparan sulfate (HS), integrins α3β1, αVβ3, and αVβ5, and the amino acid transporter x-CT protein and enters via c-Cbl-bleb-mediated macropinocytosis (Veettil et al., J. Virol. 82:12126-12144, 2008; Veettil et al., PLoS Pathog. 6:e1001238, 2010). Here we have demonstrated that very early during infection (1 min postinfection), c-Cbl induced the selective translocation of KSHV into the LR along with the α3β1, αVβ3, and x-CT receptors but not αVβ5. Activated c-Cbl localized with LRs at the junctional base of macropinocytic blebs. LR-translocated α3β1 and αVβ3 were monoubiquitinated, leading to productive macropinocytic entry, whereas non-LR-associated αVβ5 was polyubiquitinated, leading to clathrin entry that was targeted to lysosomes. c-Cbl knockdown blocked the macropinocytosis and receptor translocation and diverted KSHV to a clathrin-lysosomal noninfectious pathway. Similar results were also seen by LR disruption with MβCD. These studies provide the first evidence that c-Cbl regulates selective KSHV-α3β1, -αVβ3, and -x-CT receptor translocations into the LRs and differential ubiquitination of receptors which are critical determinants of the macropinocytic entry route and productive infection of KSHV. Our studies suggest that interventions targeting c-Cbl and LRs are potential avenues to block KSHV infection of endothelial cells.

Mechanoregulation of extracellular matrix dynamics by endothelial cell surface heparan sulfate

Decentralized mechanotransmission through endothelial cells (ECs) induces rapid fibronectin (Fn) fibril remodeling in response to hemodynamic shear stress, but mechanisms regulating adaptation of extracellular matrix structure are poorly understood. Since heparan sulfate in the glycocalyx has been proposed to modulate mechanosignaling at the apical cell surface, we hypothesized that it also contributes to mechanotransmission from the basal surface to Fn. Heparan sulfate in primary cultured post‐confluent EC monolayers was primarily in the subendothelial matrix. Heparinase pretreatment decreased initial displacement rate and eliminated directional displacement of Fn fibrils after flow onset. Competitive inhibition of heparan sulfate binding to Fn by heparin allowed coordinated fibril displacement but in random directions rather than in the flow direction. A C‐terminal Fn fragment containing HepII domains but not integrin binding domains reduced heparan sulfate expression but not colocalization with Fn fibrils. Displacement after flow onset remained in the downstream direction but became more heterogeneous, and displacement rate decreased significantly. Thus, heparan sulfate expression and association with Fn fibrils contribute to mechanoregulation of subendothelial matrix dynamics. Supported by NIH grants HL076499, HL071958, and HL080956.

Structural and Functional Studies to Define the Molecular Basis by Which Platelet Factor 4 (PF4) Increases Survival of Mice in Lipopolysaccharide (LPS)-Induced Endotoxicity

Our previous studies have demonstrated that platelet-released PF4 (chemokine CXCL4) promotes survival in a murine LPS-induced endotoxicity model, although the molecular basis for PF4's protective effects was not fully defined. We hypothesized that enhanced generation of cytoprotective activated protein C (APC) by PF4 might contribute to the molecular mechanism of PF4's beneficial effects in vivo, based on the observation that PF4 stimulates protein C (PC) activation by the thrombin/thrombomodulin complex both in vitro and in vivo. Here we show that PF4 in vitro affects human (h) PC activation in the presence of human thrombomodulin (hTM) in a bell-shaped concentration curve, i.e. stimulation at low, but inhibition at high PF4 concentrations with a peak around 3 μM. This curve is similar to that seen with PF4 for surface-bound heparin-induced thrombocytopenia (HIT) antigenicity, suggesting that similar complexes of PF4 with glycosaminoglycans (GAGs) that occur in HIT (termed ultralarge complexes (ULC)) are relevant to PF4's interaction with the hPC/hTM. Addition of heparin blocks PF4 increase of APC generation in a similar fashion as it does for surface-bound ULC. A PF4 variant PF4K50E that poorly forms PF4 tetramers requires 8-fold higher concentrations to enhance APC generation, supporting that PF4 tetramers are central for APC generation as they are for formation of ULC. Neither PF4 nor PF4K50E accelerated in vitro generation of APC in the presence of hTM that was depleted of its chondroitin sulfate chain, suggesting that PF4 binds to this domain on hTM. In vivo studies involving simultaneous infusions of PF4 and thrombin into PF4 knock out (mPF4−/−) mice showed that PF4 leads to enhanced mouse (m) APC generation not seen with infused PF4K50E, consistent with our in vitro studies. We then asked if surface heparan sulfate on the endothelial lining was necessary for the observed PF4 effect on in vivo mAPC formation. We studied mice with a Tie2-Cre conditional knock out of N-deacetylase-N-sulfotransferase-1 activity (NDST-1−/−) that have only 15% of normal endothelial cell surface heparan sulfate content using a similar thrombin/PF4 infusion model. We found that mAPC generation was accelerated by PF4 to the same extent both in NDST1−/−/mPF4−/− and mPF4−/− mice, suggesting that surface GAGs are not involved in the PF4 effect. We have also tested the in vivo effect of PF4 on mAPC formation in TM mutant (TMpro/pro) mice that have impaired capacity for APC formation to further demonstrate that PF4's positive effect in LPS endotoxic shock survival involves enhanced mAPC generation. Upon injection of high doses of thrombin (40 U/kg), mAPC levels are increased to the same extent in WT and TMpro/pro mice. After injection of low amounts of thrombin (8 U/kg), generation of mAPC was impaired in TMpro/pro as compared to WT mice. Concurrent infusion of PF4 increased mAPC formation in TMpro/pro mice after injection of low doses of thrombin, approximately equal to that seen in WT mice with no PF4 injected. As previously described, TMpro/pro mice had increased mortality after injection of LPS as compared to WT mice; however, with concurrent platelet PF4 overexpression, mortality decreased to that seen in WT mice, suggesting that the biological value of PF4 in LPS endotoxicity is related to its effect on the generation of APC. Thus, these studies support enhanced APC generation as the basis for the positive effect of PF4 on LPS endotoxicity and further define the molecular basis for increased APC generation by PF4 by forming ULC with the chondroitin sulfate domain of TM, but not with heparan sulfate on the vascular surface.

Mice Expressing Only Covalent Dimeric Heparin Binding-deficient Lipoprotein Lipase: MUSCLES INEFFICIENTLY SECRETE DIMERIC ENZYME

Lipoprotein lipase (LpL) hydrolyzes triglycerides of circulating lipoproteins while bound as homodimers to endothelial cell surface heparan sulfate proteoglycans. This primarily occurs in the capillary beds of muscle and adipose tissue. By creating a mouse line that expresses covalent dimers of heparin-binding deficient LpL (hLpLHBM-Dimer) in muscle, we confirmed in vivo that linking two LpL monomers in a head to tail configuration creates a functional LpL. The hLpLHBM-Dimer transgene produced abundant activity and protein in muscle, and the LpL was the expected size of a dimer (approximately 110 kDa). Unlike the heparin-binding mutant monomer, hLpLHBM-Dimer had the same stability as nonmutated LpL. The hLpLHBM-Dimer transgene prevented the neonatal demise of LpL knockout mice; however, these mice were hypertriglyceridemic. Postheparin plasma LpL activity was lower than expected with the robust expression in muscle and was no longer covalently linked. Studies in transfected cells showed that Chinese hamster lung cells, but not COS cells, also degraded tandem repeated LpL into monomers. Thus, although muscle can synthesize tethered, dimeric LpL, efficient production of this enzyme leading to secretion, and physiological function appears to favor secretion of a noncovalent dimer composed of monomeric subunits.

Endothelial Cell Surface Heparan Sulfate (ESHS) and Synthetic Heparin Derivatives as Hemocompatible Coating for Biomaterials

In the present overview a coating procedure, that has been developed in our working group for medical devices e. g. implants, which are exposed to permanent blood contact and therefore have to fulfill the highest standard of hemocompatibility is described. For this purpose an endothelial cell surface heparansulfate, which belongs to the class of glycosaminoglycans is used as coating substance. This substance can be isolated from endothelial cell culture, tissue extracts or organ perfusates. Alternatively chemical regio- and stereoselective modified derivatives of the structurally related anticoagulant heparin were brought to action. These substances are anchored covalently or ionically by application of a wide spectrum of immobilization techniques on many different material surfaces. Polymer materials modified as described here have been tested for hemocompatibility in invitro and in invivo experiments with Austrian sheep. The results show, that the described method is an advanced solution for the creation of long term hemocompatible artificial material surfaces.

Surface modification of the polymers present in a polysulfone hollow fiber hemodialyser by covalent binding of heparin or endothelial cell surface heparan sulfate: Flow characteristics and platelet adhesion

The present study addresses the problem of simultaneous surface modification of various polymers, i.e. polysulfone (PSU), polycarbonate (PC), and polyurethane (PU), which constitute the Ultraflux AV 600 S® hollow fibre hemodialyser. An investigation was first made into six different chemical routes aimed at introducing carboxyl groups onto the surface of PSU, PC, and PU model polymers to which heparin (HE) or endothelial cell surface heparan sulfate (ESHS) was covalently bound via the reaction of residual amino groups and a coupling reagent. Carboxyl groups were introduced using three specific reactions based on their nucleophilic or electrophilic introduction into aromatic repeating units of the polymers and three non-specific carboxylation reactions, i.e. UV, heat or redoxactivation via nitrene or radical species. Concentrations of 1-20 nmol COOH groups per cm-2 led to HE or ESHS surface concentrations corresponding to one or several layers. Two nonspecific carboxylation reactions followed by HE- or ESHS-coupling provided the lowest change in membrane pore structure according to cut off, clearance (urea, phosphate, maltose), ultrafiltration, and diafiltration assessments. In some cases the introduction of excess negatively-charged carboxyl groups and HE improved the flux properties of the modified membranes. The various methods were applied to the dialysis module. Platelet adhesion was not observed in the case of the ESHS-coating of PSU membrane at shear rates of 1050 s-1, whereas HE and subendothelial matrix showed 56 and 100% coverage, respectively, under similar conditions. The coating of PSU or of other highflux membranes by ESHS appears a promising method for improving membrane properties and to generate biocompatibility characteristics similar to those of natural blood vessels, i.e. inertness to platelet adhesion and no level effects for complement and intrinsic coagulation cascade activation. The ESHS coating may be used without anticoagulants.

Which glycosaminoglycans are suitable for antithrombogenic or athrombogenic coatings of biomaterials? Part II: Covalently immobilized endothelial cell surface heparan sulfate (ESHS) and heparin (HE) on synthetic polymers and results of animal experiments.

A systematic study was performed on immobilizing unfractionated heparin and endothelial cell surface heparan sulfate covalently with the spacer concept onto two polymer surfaces, followed by characterization of the surface concentration and in vitro and in vivo platelet adhesion properties under comparable high shear rates for microvascular vessels. Oligoamide spacer with a 16-atom chain length on cellulose surface and an 11-atom chain length on silicon surface, respectively, was used for immobilizing HE and ESHS via amino groups of glucosamine to the spacer which was anchored to the polymer surface. The surface concentration was in the range of 7 to 10 pmol/cm2 for HE and 1 to 1.5 pmol/cm2 for ESHS. This is in agreement with a calculated monolayer covering of ESHS and HE. In vitro and preliminary in vivo measurements (beagle, sheep) showed no platelet adhesion on the ESHS coatings, whereas HE showed high platelet adhesion and thrombus formation in vitro as well as in vivo. ESHS coating may be a potential candidate for preparing smooth artificial microvascular blood vessels.

Which glycosaminoglycans are suitable for antithrombogenic or athrombogenic coatings of biomaterials? Part I: Basic concepts of immobilized GAGs on partially cationized cellulose membrane.

Six different GAGs of different natural origin such as unfractionated heparin (HE), chondroitin sulfate (CS), dermatan sulfate (DS), keratan sulfate (KS), endothelial cell surface heparan sulfate (ESHS), and hyaluronan (HA) have been ionically immobilized onto partially cationized cellulose membranes with a substitution degree of 0.06. The GAGs have been characterized in terms of total sulfate content and relative molecular weight. The amount of immobilized GAGs was 10(3) times higher than the theoretical amount for monomolecular side-to-side coordination on polymer surfaces. In a standardized perfusion system with shear rates of 1050 sec-1, 37 degrees C, and 5 minutes with citrated blood, the level of platelet adhesion was 100% for partially cationized membrane, 90% for HA, 75% for each HE, CS, DS, KS, 5% for unmodified cellulose membrane, and 5% for confluent bovine aorta endothelial cells, referred to subendothelial matrix as standard for 100% platelet adhesion. ESHS-coated membranes were completely inert to platelets. The amount of ionically released GAGs during perfusion was also estimated. The partial cationic membrane is a suitable polymer surface in combination with the perfusion system to estimate the hemocompatibility of ionically immobilized water-soluble polyanions in terms of platelet adhesion at defined shear rates. The results of platelet adhesion are discussed in terms of structure and analytical parameter of the immobilized GAGs.

Monoclonal antibody to heparan sulfate from autoimmune tight skin (TSK) mice binds to the endothelial cell surface

Heparan sulfate proteoglycans have pleiotropic functions in the normal vasculature. Autoimmunity to heparan sulfate may play a role in vascular injury. In this study, monoclonal antibody (mb) 28C3–1 to heparan sulfate derived from autoimmune Tight skin (TSK) mice was investigated for its reactivity with endothelial cells. Mb 28C3–1 was previously demonstrated to inhibit the heparin-accelerated formation of antithrombin III-thrombin complexes. In the current studies it is shown that mAb 28C3–1 bound to heparan sulfate proteoglycan with the highest affinity in direct binding solid phase radioimmunoassay. Binding to the heparan sulfate was stronger than binding to the protein core, indicating that the primary epitope of 28C3–1 is the polysaccharide component. Using confocal fluorescent microscopy, mAb 28C3–1 was demonstrated to bind to the endothelial cell surface. Furthermore, treatment of endothelial cells with heparitinase abolished mAb 28C3–1 binding. These studies support the hypothesis that naturally occurring anti-heparan sulfate autoantibodies from autoimmune mice may cause vascular injury by initial interaction with endothelial cell surface heparan sulfate.

Oligosaccharide sequences of endothelial cell surface heparan sulfate proteoglycan with affinity for lipoprotein lipase.

Lipoprotein lipase (LpL) catalyzes the hydrolysis of triglycerides in plasma lipoproteins at the luminal surface of the vascular endothelium. This enzyme is bound via electrostatic interactions to heparan sulfate (HS). The specific endothelial cell surface HS oligosaccharide sequences that are necessary for binding of LpL to HS have not been characterized. To identify this LpL-binding oligosaccharide sequence, oligosaccharides were isolated from bovine aortic endothelial cell-derived HS and assessed for LpL binding properties. Endothelial HS chains that were isolated from endothelial total cell-associated proteoglycans were deacetylated by complete hydrazinolysis, cleaved with nitrous acid (pH 4.5), and reduced with [3H]NaBH4. The resulting fragments composed of N-sulfated glucosamine-rich oligosaccharides terminating with [3H]2,5-anhydromannitol (AManR) were chromatographed on a LpL-Sepharose column. A high affinity decasaccharide was isolated and characterized. Disaccharide analysis of this decasaccharide indicated that it yielded only the disaccharide IdceA(2-SO4)-->AManR(6-SO4) on treatment with nitrous acid at low pH. Therefore, the sequence of the LpL-binding decasaccharide is [IdceA(2-SO4) alpha 1-4GlcNSO4(6-S0(4)) alpha 1-4]4-IdceA(2-SO4) alpha 1-4AManR(6-SO4) and is distinct from those that bind antithrombin and basic fibroblast growth factor. Partial depolymerization of endothelial HS chains with hydrazine/high pH nitrous acid treatment gave rise to lipase-binding oligosaccharides larger than decasaccharide. However, further complete depolymerization of these oligosaccharides resulted in only a high affinity decasaccharide composed of repeating disaccharide units of [IdceA(2-SO4) alpha 1-4GlcNSO4(6-S0(4))]. These results indicate that the decasaccharide is the active fragment that binds to LpL with high affinity. Molecular modeling studies of the decasaccharide indicate that it presents a linear array of negatively charged sulfate groups that may adopt a favorable disposition to bind to peptide region(s) comprised of basic amino acid residues of LpL with high affinity.

ES-HS from blood vessels —A potent substance for inert nonthrombogenic polymers

Two different membranes, a cellulose (visking dialysis tubing) and a polyethercarbonate membrane (Gambrane ®), were chemically modified: 1. (I)partially substituted 2-hydroxy-3-(trimethylammonionochloride)-propylcellulose 2. (II)partially substituted 12-aminododecanyl-desoxiamino-cellulose, partially cationized with 3-chloro-2-hydroxypropyl-trimethylammonium chloride 3. (III) partially substituted (3-aminopropyl)-amidosulfonyl-poly[etherearbonate], partially cationized with 3-chloro-2-hydropropyl-trimethylammoniumchloride. 4. (IV)partially substituted N-(2-aminoethyl)-imidocarbamate-cellulose. 5. (V)partially substituted N-(3-aminopropyl)-amidosulfonyl-poly[ethercarbonate]. A commercially available mucosa heparin (HE) was bound ionically to I–V. An endothelial cell surface heparan sulfate (ES-HS) was bound ionically to I and III. Furthermore, functional groups (COOH) of ionically-immobilized HE on I–V and ES-HS on III were subsequently bound covalently to functional groups (OH, NH 2) of the modified membranes. Platelet adhesion tests in a modified Baumgartner chamber were used to characterize the HE and ES-HS modified membranes. ES-HS ionically bound to I and ES-HS covalently bound to III led to surfaces completely inert to platelet adhesion. In contrast to these results, HE showed 100% and 80% platelet adhesion, respectively. These results show that ES-HS, which is responsible for the very high standard of nonthrombogenicity of the blood vessel wall, is a promising coating substance to achieve nonthrombogenic artificial surfaces.

Partially cationized cellulose for non-thrombogenic membrane in the presence of heparin and endothelial-cell-surface-heparansulfate (ES-HS)

Partially cationized cellulose derivatives ( DS 0.01-0.1) have been synthesized by modification of cotton, Cuprophan® and cellulose dialysis membranes with 3-chloro-2-hydroxypropyltrimethyl-ammoniumchloride and used for the following studies: Adsorption and desorption studies of heparin on partially cationized cellulose in NaHCO 3-solution show that heparin can be drastically enriched with increase of the DS of cellulose, whereas imidazole buffer and a pyridine-water mixture ( DS 0.06) desorbe heparin partially or completely. When partially cationized cellulose was used under suitable conditions for heparin enrichment and additionally four different crosslinking reagents and different reaction steps were applied and compared with non cationzed cellulose, the ionic fixation of heparin was ten to fifty times higher compared with non-cationized cellulose. When tested with blood partially cationized cellulose showed a high platelet adhesion whereas endothelial-cell-surface-heparansulfate (ES-HS), isolated from bovine aorta and immobilized on partially cationized cellulose, was completely inert to platelet adhesion measured in a modified Baumgartner chamber. The latter result seems to give a new promising concept for the synthesis of biomaterials, e.g. membranes with nonthrombogenic surfaces similar to surfaces of blood vessels in nature because the isolation procedure of ES-HS and immobilization of ES-HS on polymer surfaces does not change the inertness of ES-HS to platelet adhesion.

ヒスチジンリッチグリコプロテインと血管内皮細胞ヘパリン様物質間相互反応の欠如

A role of histidine-rich glycoprotein in controlling heparin-like compounds on the endothelial cell surface is yet unclear. Incubation of purified porcine 125I-histidine-rich glycoprotein retaining heparin neutralizing activity with cultured porcine aortic endothelial cells resulted in only unsaturable binding that is not inhibited by excess unlabeled histidine-rich glycoprotein. This binding was not inhibited by excess unlabeled purified porcine antithrombin III or protamine sulfate and by prior treatment of endothelial cells with heparitinase. On the other hand, displacement of 125I-antithrombin III specifically bound to endothelial cells by unlabeled histidine-rich glycoprotein was much less potent than that by unlabeled antithrombin III. The endothelial cell-mediated acceleration of thrombin inactivation by antithrombin III was diminished by protamine sulfate, but was not affected by histidine-rich glycoprotein even at a histidine-rich glycoprotein/antithrombin III molar of -7:1. These data indicate that histidine-rich glycoprotein neither binds to endothelial cell surface heparan sulfate nor interferes with the interaction of this compound with antithrombin III, suggesting that it may not play an important role in the modulation of anticoagulant activity of endothelial cells.