Myofibrillar protein structure and assembly during idiopathic dilated cardiomyopathy.

A neutral protease, mekratin, active in human hearts at end stage idiopathic dilated cardiomyopathy (IDC), mediates the breakdown of cardiac myosin LC2. Myosin purified from IDC heart tissue forms unusually short synthetic thick filaments. Therefore, determination of filament length and mekratin distribution in IDC heart muscle were initiated. Native thick filaments were prepared directly from control and IDC tissues and analyzed. Also, paraffin-embedded tissue sections were stained with a fluorescently-labeled anti-protease antibody to establish its distribution in myocardial tissues. Control sections had only very weak, background levels of fluorescence whereas IDC sections stained intensely throughout, indicating a wide ranging distribution of the protease within the myocyte cytoplasm. SDS-PAGE revealed LC2 to be present in stoichiometric amounts in control but greatly reduced in IDC heart muscle. Native thick filaments from control myocardium were structurally stable. They had a median length of 1.65 microm with well-defined bare zones and displayed the 43 nm helical periodicity typical of the relaxed arrangement of myosin heads close to the filaments' shafts. In contrast, native IDC filaments were less stable, and had a median length of 0.9 microm. These filaments were highly disordered: they had no surface periodicity and myosin heads were positioned away from the filaments' shafts. The shorter, less stable, aperiodic thick filaments from IDC hearts appear to result from depletion of LC2 caused by increased activity of mekratin in the IDC myocardium.

Calcium regulation in the human myocardium affected by dilated cardiomyopathy: a structural basis for impaired Ca2+-sensitivity.

Calcium regulation in the human heart is impaired during idiopathic dilated cardiomyopathy (IDC). Here, we analyze the structural basis for impairment in the regulatory mechanism. Regulation of contractility was monitored by MgATPase and Ca2+-binding assays as a function of calcium. Myofibrillar proteolysis and expression of troponin T isoforms were established by gel electrophoresis and by Western blots. Myofibrillar ATPase assays in low salt however, revealed a drastic lowering of calcium sensitivity in IDC myofibrils as indicated by reductions in both activation by high calcium and in EGTA-mediated inhibition of MgATPase. Structural changes in myofilament proteins were found in most IDC hearts, specifically proteolysis of myosin light chain 2 (LC2), troponin T and I (TnT and TnI), and sometimes a large isoform shift in TnT. IDC did not induce mutations in LC2 and troponin C (TnC), as established by cDNA sequence data from IDC cases, thus, calcium binding to IDC myofibrils was unaffected. Reassociation of IDC myofibrils with native LC2 raised MgATPase activation at high Ca2+ to control levels, while repletion with intact, canine TnI/TnT restored inhibition at low Ca2+. A model, identifying possible steps in the steric blocking mechanism of regulation, is proposed to explain IDC-induced changes in Ca2+-regulation. Moreover, shifts in TnT isoforms may imply either a genetic or a compensatory factor in the development and pathogenesis of some forms of IDC.

Cloning of the cDNA and nucleotide sequence of a skeletal muscle protease from myopathic hamsters.

A neutral protease with an estimated Mr of about 26 kD and responsible for cleavage ofmyosin LC2 was isolated from hamster skeletal muscle. Complementary DNAs were generated by RT-PCR using total hamster muscle RNA and degenerate oligonucleotide primers based on the sequences of two internal peptides. The nucleotide sequences of the resultant cDNAs were subsequently determined and the complete amino acid sequence of the protease deduced. Although the hamster protein shared 63-85% identity in nucleotide and amino acid sequences with rat and mouse mast cell proteases, it had a higher degree of specificity for myosin LC2 than mast cell proteases which also digested myosin LC1 and myosin heavy chains. As a result, the hamster protease was designated mekratin because of its unique enzymatic specificities to distinguish it from other mast cell proteases. A polyclonal antibody was raised specific to the hamster muscle and human cardiac muscle mekratins without apparent cross-reaction with rat mast cell proteases. We have earlier demonstrated the presence in excess of a neutral protease that specifically cleaves LC2 in human hearts obtained at end stage idiopathic dilated cardiomyopathy (IDC). Western analyses revealed that heart tissue from patients with IDC contained 5-10 fold more mekratin than control samples. Furthermore, the level of the protease in human IDC tissues was similar to that seen in myopathic hamster skeletal muscle. No bands were recognized by the antibody when IDC myofibrils were probed due to the removal of soluble proteins during sample preparation. Thus, these results strongly suggest that the anti-mekratin antibody will provide positive identification of IDC in many cases and diagnosis by exclusion may be replaced.

Continuous detection of radiation or metal generated hydroxyl radicals within core chromatin particles.

PURPOSE: The aim of this work was to adapt a recently developed fluorometric method for use in the detection of hydroxyl radical (HO.) generated in the immediate vicinity of chromatin core particles reconstituted from pUC19 plasmid DNA and isolated core histones. MATERIALS AND METHODS: The procedure followed involves labelling nucleosomal histones with SECCA, a non-fluorescent coumarin derivative that generates the fluorescent 7-hydroxy-SECCA after reaction with HO.. Core particles are formed using histones and pUC19 DNA in a salt-dialysis procedure. RESULTS: Electron microscopy and micrococcal nuclease digestion are consistent with successful formation of core particles. No significant differences between core particle formation in the unlabelled and SECCA-labelled samples were detected. Exposure to HO. generated by radiation or copper--ascorbic acid--hydrogen peroxide results in a gradual induction of fluorescence. Studies using dimethyl sulphoxide (DMSO) demonstrate that, unlike HO. produced by radiation, the majority of HO. generated by copper--ascorbic acid--hydrogen peroxide occurs primarily in the immediate vicinity of core particles and DNA and cannot be scavenged. CONCLUSIONS: The present procedure demonstrates the feasibility to quantitate HO. generated by several agents in the immediate vicinity of nucleosomes (chromatin-associated HO.) or associated with specific regions within the genome.

Antineutrophil cytoplasmic autoantibodies interact with primary granule constituents on the surface of apoptotic neutrophils in the absence of neutrophil priming.

The pathogenic role of antineutrophil cytoplasmic autoantibodies (ANCA) remains controversial because of the difficulty in explaining how extracellular ANCA can interact with intracellular primary granule constituents. It has been postulated that cytokine priming of neutrophils (PMN), as may occur during a prodromal infection, is an important trigger for mobilization of granules to the cell surface, where they may interact with ANCA. We show by electron microscopy that apoptosis of unprimed PMN is also associated with the translocation of cytoplasmic granules to the cell surface and alignment just beneath an intact cell membrane. Immunofluorescent microscopy and FACS analysis demonstrate reactivity of ANCA-positive sera and antimyeloperoxidase antibodies with apoptotic PMN, but not with viable PMN. Moreover, we show that apoptotic PMN may be divided into two subsets, based on the presence or absence of granular translocation, and that surface immunogold labeling of myeloperoxidase occurs only in the subset of PMN showing translocation. These results provide a novel mechanism that is independent of priming, by which ANCA may gain access to PMN granule components during ANCA-associated vasculitis.

Structures of class A macrophage scavenger receptors. Electron microscopic study of flexible, multidomain, fibrous proteins and determination of the disulfide bond pattern of the scavenger receptor cysteine-rich domain.

Structures of secreted forms of the human type I and II class A macrophage scavenger receptors were studied using biochemical and biophysical methods. Proteolytic analysis was used to determine the intramolecular disulfide bonds in the type I-specific scavenger receptor cysteine-rich (SRCR) domain: Cys2-Cys7, Cys3-Cys8, and Cys5-Cys6. This pattern is likely to be shared by the highly homologous domains in the many other members of the SRCR domain superfamily. Electron microscopy using rotary shadowing and negative staining showed that the type I and II receptors are extended molecules whose contour lengths are approximately 440 A. They comprised two adjacent fibrous segments, an alpha-helical coiled-coil ( approximately 230 A, including a contribution from the N-terminal spacer domain) and a collagenous triple helix ( approximately 210 A). The type I molecules also contained a C-terminal globular structure ( approximately 58 x 76 A) composed of three SRCR domains. The fibrous domains were joined by an extremely flexible hinge. The angle between these domains varied from 0 to 180 degrees and depended on the conditions of sample preparation. Unexpectedly, at physiologic pH, the prevalent angle seen using rotary shadowing was 0 degrees , resulting in a structure that is significantly more compact than previously suggested. The apparent juxtaposition of the fibrous domains at neutral pH provides a framework for future structure-function studies of these unusual multiligand receptors.

Human cardiac myosin light chains: sequence comparisons between myosin LC1 and LC2 from normal and idiopathic dilated cardiomyopathic hearts.

The primary structures of light chains isolated from the human myocardium with idiopathic dilated cardiomyopathy (IDC) were determined and compared with the sequence structures of myosin light chains obtained from control human heart myosin. Sequences were determined by chemical analysis and the identity of N-terminal residues established by mass spectrometry. The N-terminal residues in essential (ELC) and regulatory (RLC) light chains were blocked and were identified to be trimethyl alanine. The amino acid sequences of ELC and RLC from control human myosin revealed a high degree of homology with those purified from rat and chicken cardiac myosin. Comparison with a published partial chemical sequence of the human heart myosin light chains revealed significant variations. However, there was very good agreement with published sequences obtained by molecular biological techniques. Sequences of the light chains from cardiomyopathic myosin revealed no difference in the primary structures when compared with control human heart myosin light chains indicating IDC had no influence on, nor was caused by, altered myosin light chain gene expression.

Immunocytochemical localization of endogenous anti-thrombin III in the vasculature of rat tissues reveals locations of anticoagulantly active heparan sulfate proteoglycans.

We localized endogenous anti-thrombin III (ATIII) by light and electron microscopic immunocytochemical staining in cryostat and ultra-thin frozen sections of 10 different rat tissues, using rabbit alpha-human ATIII antibody that was shown to crossreact strongly with rat ATIII. EM immunocytochemical methods revealed discrete deposits of endogenous ATIII (absent after heparinase treatment), and thus by inference anticoagulantly active heparan sulfate proteoglycans (HSPGs) at a resolution of 10-20 nm, or an order of magnitude better than autoradiography or LM. ATIII was found in variable amounts almost entirely in the subendothelial space of blood vessels in various rat tissues. In kidney, ATIII was found immediately beneath the endothelium, in concentrated clusters associated with the vascular basement membrane. Equally important is the observed variation in expression of ATIII in the various tissues studied (i.e., kidney > liver, aorta, lung, spleen, adrenal > intestine, muscle, brain). On the basis of these observations, we confirm a model in which vascular abluminal and, perhaps to a much smaller extent, luminal anticoagulantly active HSPGs regulate coagulation mechanism activity, either by serving as a reserve of anticoagulant or by modulating the ambient function of the coagulation cascade.

Mapping the heparin-binding sites on type I collagen monomers and fibrils.

The glycosaminoglycan chains of cell surface heparan sulfate proteoglycans are believed to regulate cell adhesion, proliferation, and extracellular matrix assembly, through their interactions with heparin-binding proteins (for review see Ruoslahti, E. 1988. Annu. Rev. Cell Biol. 4:229-255; and Bernfield, M., R. Kokenyesi, M. Kato, M. T. Hinkes, J. Spring, R. L. Gallo, and E. J. Lose. 1992. Annu. Rev. Cell Biol. 8:365-393). Heparin-binding sites on many extracellular matrix proteins have been described; however, the heparin-binding site on type I collagen, a ubiquitous heparin-binding protein of the extracellular matrix, remains undescribed. Here we used heparin, a structural and functional analogue of heparan sulfate, as a probe to study the nature of the heparan sulfate proteoglycan-binding site on type I collagen. We used affinity coelectrophoresis to study the binding of heparin to various forms of type I collagen, and electron microscopy to visualize the site(s) of interaction of heparin with type I collagen monomers and fibrils. Using affinity coelectrophoresis it was found that heparin has similar affinities for both procollagen and collagen fibrils (Kd's approximately 60-80 nM), suggesting that functionally similar heparin-binding sites exist in type I collagen independent of its aggregation state. Complexes of heparin-albumin-gold particles and procollagen were visualized by rotary shadowing and electron microscopy, and a preferred site of heparin binding was observed near the NH2 terminus of procollagen. Native or reconstituted type I collagen fibrils showed one region of significant heparin-gold binding within each 67-nm period, present near the division between the overlap and gap zones, within the "a" bands region. According to an accepted model of collagen fibril structure, our data are consistent with the presence of a single preferred heparin-binding site near the NH2 terminus of the collagen monomer. Correlating these data with known type I collagen sequences, we suggest that the heparin-binding site in type I collagen may consist of a highly basic triple helical domain, including several amino acids known sometimes to function as disaccharide acceptor sites. We propose that the heparin-binding site of type I collagen may play a key role in cell adhesion and migration within connective tissues, or in the cell-directed assembly or restructuring of the collagenous extracellular matrix.

Autocrine release of angiotensin II mediates stretch-induced hypertrophy of cardiac myocytes in vitro.

Hypertrophy is a fundamental adaptive process employed by postmitotic cardiac and skeletal muscle in response to mechanical load. How muscle cells convert mechanical stimuli into growth signals has been a long-standing question. Using an in vitro model of load (stretch)-induced cardiac hypertrophy, we demonstrate that mechanical stretch causes release of angiotensin II (Ang II) from cardiac myocytes and that Ang II acts as an initial mediator of the stretch-induced hypertrophic response. The results not only provide direct evidence for the autocrine mechanism in load-induced growth of cardiac muscle cells, but also define the pathophysiological role of the local (cardiac) renin-angiotensin system.

Physical characterization of recombinant tissue plasminogen activator.

Electron microscopic and physical-chemical properties of one- and two-chain tissue plasminogen activator (t-PA) were studied. The molecular weight of one-chain t-PA obtained by both sedimentation equilibrium and SDS-PAGE was estimated to be about 65,000, while both chains in the reduced two-chain form were in the range of 35,000-40,000. Sedimentation coefficients were identical for both forms of t-PA (S(0)20,w = 4.12). The two forms of t-PA were indistinguishable by electron microscopic analysis, which confirmed the sedimentation results, and showed that they were ellipsoidal and relatively compact. The major and minor axes were approx. 13 nm and approx. 10 nm and f/f0 was 1.36. The individual domains of t-PA are relatively small and are folded within the molecule, so that the overall appearance is globular.

Functional effects of LC1-reassociation with cardiac papain Mg.S1.

The effect of LC1 on cardiac myosin structure and activity was investigated using as a model S1 prepared by papain digestion in the presence of Mg2+. The resulting S1 contained LC2 but a part of the N-terminal region of LC1 was cleaved. Sequencing the N-terminal part of the band migrating below LC1 on SDS gels revealed it to consist of alternating alanyl and prolyl residues thus establishing LC1 as the origin of this band. However, Western blots did not reveal any LC1 while radioimmunoassays indicated it to be present at the 5% level suggesting the anti-LC1 antibody used in these experiments did not recognize the C-terminal portion of LC1 still attached to Mg.S1. Mixing a 10-15 M excess of isolated light chains with Mg.S1 in the presence of 10 mM ATP, 12 mM MgCl2, 4.7 M NH4Cl allowed LC1 to recombine with LC1-deficient Mg.S1. Equilibrium ultracentrifugation analysis revealed a highly heterogeneous LC1-deficient S1 which upon recombination with intact LC1 became monodisperse as indicated by the superimposition of molecular weight averages all across the centrifuge cell. LC1-deficient Mg.S1 had a Vm of 0.4 s-1, Ka of 30 microM and a Kbind of 28 microM. In the presence of intact LC1, Vm rose to 0.8 s-1 while Ka and Kbind were reduced to 7.5 and 12 microM, respectively. The fourfold decrease in Ka strongly indicated an increased affinity for actin by Mg.S1 in the presence of uncleaved LC1. Also, Ca(2+)-regulation of dog heart myofibrils was suppressed when Ca(2+)-activated MgATPase assays, as a function of Ca2+, were performed in the presence of anti-LC1 antibodies. These observations suggest the presence of intact, uncleaved LC1 in S1 is required for the stability of S1 heavy chains and proper Ca(2+)-regulation.

Light chain 2 profile and activity of human ventricular myosin during dilated cardiomyopathy. Identification of a causal agent for impaired myocardial function.

BACKGROUND: A number of parameters reflecting the effects of idiopathic dilated cardiomyopathy (IDC) on the structure and function of myosin from the human myocardium were analyzed. METHODS AND RESULTS: The content of the regulatory light chain, LC2, was reduced in myopathic heart myosin in contrast to the controls in which it was present in stoichiometric amounts relative to the essential light chain, LC1. In IDC hearts, the absence or significant reduction in amount of LC2 was related to the presence of an active protease, which was isolated and purified about 130-fold. The protease exhibited a significant degree of specificity: It cleaved LC2 almost totally (but not the heavy chains) in human control heart myosin but only partially cleaved LC2 in canine heart or in rabbit skeletal muscle myosins. The protease was present at a very low level or was inactive in control heart tissue. When the LC1/LC2 molar ratio was calculated, it was found to be 1:1.0 in control heart myosin and remained constant in various samples analyzed, whereas in myopathic myosin from different individuals, this ratio varied from 1:0.1 to 1:0.69. The rates of ATP binding to control and myopathic myosins were similar, whereas the Vm of actin-activated ATPase of myopathic myosin was about 25% less than that of the control. However, ATP binding and its hydrolysis by control S1, i.e., the myosin head, were faster by a factor of 2 than that of the myopathic S1. In addition, control myosin synthetic thick filament length as well as turbidity in solution, measured by light scattering, were twice as large as those of the myopathic heart myosin. These effects induced by myopathy in both filament assembly and turbidity were reversed upon reassociation of IDC myosin with LC2. CONCLUSIONS: The changes in myosin structure and function were linked to a protease-mediated cleavage of LC2 in myosin; a possible role for the protease in the degenerative effects of idiopathic dilated cardiomyopathy is thus defined.

Intestinal mucin of germ-free rats. Biochemical and electron-microscopic characterization.

Purified germ-free rat intestinal mucin was found by chemical analysis to contain 25% protein, enriched in serine, threonine, and proline, 75% carbohydrate, and no nucleic acid. It was analyzed by darkfield electron microscopy and found to consist of long filamentous molecules with a maximum length of approximately 740 nm, a mean length of 456 nm, and a mean width of 7 nm. Given reasonable assumptions derived from earlier work on other well-characterized mucins, the molecular weight of the peptide, calculated by the length from electron microscopy, was 200,000, and, given the chemical composition, the molecular weight of the entire mucin molecule was calculated to be approximately 800,000.

Platelet activation by a synthetic hydrophobic polymer, polymethylmethacrylate.

Platelets adhere to artificial surfaces in the initial stage of thrombus formation, but the subsequent steps in signal transduction that lead to platelet activation by artificial surfaces are not understood. When 0.325-micron diameter beads composed of a hydrophobic polymer, polymethylmethacrylate (PMMA), were added to gel-filtered aequorin-loaded platelets suspended in media containing Ca2+, the platelets aggregated; addition of fibrinogen was not required. Platelet aggregation was preceded by an increase in cytoplasmic Ca2+ and was accompanied by phosphorylation of the 47-Kd substrate of protein kinase C (PKC), 5-hydroxytryptamine (5-HT) release, and accumulation of phosphatidic acid. All these effects were partially inhibited by apyrase and aspirin. Monoclonal antibodies (MoAbs) 7E3 and M148 and the synthetic peptides RGDS and fibrinogen gamma chain fragment 400-411, all of which bind to the platelet fibrinogen receptor glycoprotein IIb-IIIa (GPIIb-IIIa) and inhibit fibrinogen binding, prevented PMMA-induced aggregation but did not inhibit the Ca2+ increase. Chymotrypsin-treated platelets aggregated after addition of fibrinogen, but not PMMA. We conclude that platelets interact initially with PMMA at membrane sites other than those required for fibrinogen binding, leading to activation of membrane phospholipases and PKC, an increase in cytoplasmic Ca2+, release of 5-HT, ADP, and fibrinogen from storage granules, and to platelet aggregation.

The effect of the carbohydrate moiety upon the size and conformation of human plasma galactoglycoprotein as judged by electron microscopy and circular dichroism. Structural studies of a glycoprotein after stepwise enzymic carbohydrate removal.

Galactoglycoprotein is a unique human plasma protein [76% carbohydrate (23% N-acetylneuraminic acid, 20% galactose, 3% mannose, 1% fucose and 29% N-acetylgalactosamine plus N-acetylglucosamine) and 24% polypeptide, a single polypeptide chain of about 200 amino acid residues that is high in serine and threonine content] [Schmid, Mao, Kimura, Hayashi & Binette (1980) J. Biol. Chem. 255, 3221-3226]. Highly purified exoglycosidases with well-defined specificities were used to prepare five derivatives of galactoglycoprotein in which sequential residues of N-acetylneuraminic acid, galactose, N-acetylglucosamine, a second galactose and N-acetylgalactosamine were removed with 83% of the total carbohydrate cleaved. C.d. shows that native galactoglycoprotein and all derivatives in aqueous buffer are predominantly random coil, suggesting that removal of a large number of electrostatic net charges, as well as the major portion of the carbohydrate moiety, does not alter the secondary structure of the polypeptide chain. Examination of the size and conformation of tungsten-shadowed galactoglycoprotein and asialo and agalacto derivatives by electron microscopy shows the size and conformation of all three preparations to be similar, with only minor differences in particle length and width.

Intracellular pathway of a mucin-type membrane glycoprotein in mouse mammary tumor cells.

Epiglycanin, a mucin-type glycoprotein, was found by immunoelectron microscopy to be located in cytoplasmic compartments, as well as at the surface of the TA3-Ha mammary carcinoma ascites cell. The glycoprotein was identified by means of gold-labeled secondary antibody bound to a primary anti-epiglycanin monoclonal antibody or by lectins specific for carbohydrate structures in epiglycanin. The primary antibody recognized a glycopeptide component containing a beta-D-(1----3)-D-GalNAc chain attached to a serine or threonine residue. Two routes to the cell surface from epiglycanins's first-recognized location in the trans-Golgi reticulum were suggested. Its presence in vesicles, which fuse with the cell surface, would explain the presence of epiglycanin as an integral membrane protein. Some of these observed vesicles, however, may be endocytotic in character. Epiglycanin was also found in large multivesiculate sacs which were observed on occasion to be open to the extracellular milieu. This finding, as well as the observed fusion of small vesicles from the trans-Golgi network with the sacs, strongly suggested exocytotic migration for the large sacs. Endocytotic migration may also be possible, although incubation of viable cells with gold-labeled antiepiglycanin antibody resulted in minimal uptake within the intracellular sacs, and incubation with [125I]-epiglycanin under metabolic conditions resulted in no detectable uptake of radiolabel by the cells.

Influence of the cardiac myosin hinge region on contractile activity.

The participation of cardiac myosin hinge in contractility was investigated by in vitro motility and ATPase assays and by measurements of sarcomere shortening. The effect on contractile activity was analyzed using an antibody directed against a 20-amino acid peptide within the hinge region of myosin. This antibody bound specifically at the hinge at a distance of 55 nm from the S1/S2 junction, was specific to human, dog, and rat cardiac myosins, did not crossreact with gizzard or skeletal myosin, and had no effect on ATPase activity of purified S1 and myofibrils. However, it completely suppressed the movement of actin filaments in in vitro motility assays and reduced active shortening of sarcomeres of skinned cardiac myocytes by half. Suppression of motion by the anti-hinge antibody may reflect a mechanical constraint imposed by the antibody upon the mobility of the S2 region of myosin. The results suggest that the steps in the mechanochemical energy transduction can be separately influenced through S2.

Localization of anticoagulantly active heparan sulfate proteoglycans in vascular endothelium: antithrombin binding on cultured endothelial cells and perfused rat aorta.

We have studied the interaction of 125I-antithrombin (125I-AT) with microvascular endothelial cells (RFPEC) to localize the cellular site of anticoagulantly active heparan sulfate proteoglycans (HSPG). The radiolabeled protease inhibitor bound specifically to the above HSPG with a Kd of approximately 50 nM. Confluent monolayer RFPEC cultures exhibited a linear increase in the amount of AT bound per cell for up to 16 d, whereas suspension RFPEC cultures possessed a constant number of protease inhibitor binding sites per cell for up to 5 d. These results suggest that monolayer RFPEC cultures secrete anticoagulantly active HSPG, which then accumulate in the extracellular matrix. This hypothesis was confirmed by quantitative light and EM level autoradiography which demonstrated that the AT binding sites are predominantly located in the extracellular matrix with only small quantities of protease inhibitor complexed to the cell surface. We have also pinpointed the in vivo position of anticoagulantly active HSPG within the blood vessel wall. Rat aortas were perfused, in situ, with 125I-AT, and bound labeled protease inhibitor was localized by light and EM autoradiography. The anticoagulantly active HSPG were concentrated immediately beneath the aortic and vasa vasorum endothelium with only a very small extent of labeling noted on the luminal surface of the endothelial cells. Based upon the above data, we propose a model whereby luminal and abluminal anticoagulantly active HSPG regulate coagulation mechanism activity.