Heparin binding domain of human antithrombin III inferred from the sequential reduction of its three disulfide linkages. An efficient method for structural analysis of partially reduced proteins

Human antithrombin III (AT-III) was partially reduced under mild conditions in the absence or presence of low molecular weight heparin. Quantitation of reduced disulfide bonds was facilitated by the application of a water-soluble color reagent, 4-N,N-dimethylaminoazobenzene-4'-iodoacetamido-2'-sulfonic acid (S-DABIA). The study shows that the three disulfide linkages of AT-III can be sequentially reduced, with Cys8-Cys128 being the most sensitive, followed by Cys21-Cys95, while Cys247-Cys430 is the most resistant to the mild reduction conditions. The rate of reduction of Cys8-Cys128 and Cys21-Cys95 was significantly decreased in the presence of heparin. The reduction of Cys8-Cys128 was also found to correlate quantitatively with the loss of heparin-accelerated antithrombin activity, heparin binding affinity, and heparin-induced fluorescence enhancement. These results suggest that Cys8-Cys128 is required for the integrity of the heparin binding domain of AT-III and support previous findings that lysyl residues surrounding Cys128 (Lys107, Lys114, Lys125, and Lys136) constitute an important part of the heparin binding site in AT-III.     

Non-enzymatic glycation of antithrombin III in vitro

Non-enzymatic glycation of antithrombin III (AT-III) has been proposed as a significant contributor to the increased incidence of thrombo-occlusive events in diabetics. AT-III, isolated from normal human plasma by means of heparin affinity and ion-exchange chromatography, was incubated with 0-0.5 M glucose in neutral phosphate buffer at 37 degrees C. The extent of non-enzymatic glycation could be monitored by uptake of radioactivity as well as by binding to a phenylboronate affinity resin, which effectively retards AT-III containing ketoamine-linked glucose. Non-enzymatically glycated AT-III (approx. 1 mol glucose/mol protein) bound heparin nearly as efficiently as non-glycated AT-III. The two AT-III preparations were equally active in inhibiting thrombin cleavage of chromogenic substrate. Following incubation with [14C]glucose, structural analyses of cyanogen-bromide-cleaved peptides of enzymatically glycated AT-III showed that the [14C]glucose adducts were distributed over many sites on the molecule. This lack of specificity contrasts with the restricted sites of modification on hemoglobin, albumin and ribonuclease A, and explains why non-enzymatic glycation of AT-III has little if any effect on its function.     

Isolation and characterization of an antithrombin III variant with reduced carbohydrate content and enhanced heparin binding

Two distinct forms of antithrombin III were isolated by chromatography of normal human plasma on heparin-Sepharose. The predominant antithrombin species present (AT-III alpha), which eluted from the affinity column in 1 M NaCl, was identified as the antithrombin III form which has been previously characterized. Ionic strength of the buffer was increased to elute a variant form of antithrombin III, designated as AT-III beta. The molecular weight of AT-III beta is less than that of AT-III alpha, but physicochemical studies do not indicate measureable differences in the polypeptide portion of the proteins. Carbohydrate determination revealed the sole detectable structural difference in the two antithrombins: levels of hexosamine, neutral sugars, and sialic acid in AT-III beta were all 25-30% less than in AT-III alpha. Kinetic studies of thrombin inactivation by both antithrombins, in the presence of nonsaturating amounts of heparin, indicated that AT-III beta inhibited thrombin more rapidly. AT-III beta is also distinguishable from AT-III alpha on the basis of heparin-binding affinity estimated from titration of protein fluorescence with heparin. Thus, antithrombin III exists as two molecular entities in human plasma which differ both structurally, in carbohydrate content, and functionally, in their heparin-binding behavior.     

Production of human antithrombin-III in a serum-free culture of CHO cells.

A simple method was developed to establish serum-independent Chinese hamster ovary (CHO) cells that grew and secreted high level of human antithrombin-III (AT-III). First, human AT-III and mouse dihydrofolate reductase (DHFR) cDNAs were transfected into DHFR-deficient CHO cells. Transfected cells were treated with increasing concentrations of methotrexate (MTX) and clones secreting high levels of AT-III (10-20 micrograms/ml/3 day) in a serum-containing medium were obtained. Serum-independent clones were derived from the serum-dependent clones by simply culturing the cells for a few weeks in a serum-free medium. In a serum-free medium the established serum-independent clones grew at normal rate and produced almost equivalent amount of AT-III to that of the serum-dependent, parent clones. In addition, AT-III from the serum-independent clones has specific activity similar to that of plasma-derived AT-III.     

Structure-activity relationship in heparin: A synthetic pentasaccharide with high affinity for antithrombin III and eliciting high anti-factor Xa activity

The structures of the tetrasaccharide (β-D-glucuronic acid)1→4 (N-sulfate-3,6-di-O-sulfate-α-D-glucosamine)1→4(2-O-sulfate-α-L-iduronic acid)1→4(N-sulfate-6-O-sulfate-D-glucosamine) and of the pentasaccharide (N-sulfate-6-O-sulfate-α-D-glucosamine)1→4(β-D-glucuronic acid)1→4(N- sulfate-3,6-di-O-sulfate-α-D-glucosamine)1→4(2-O-sulfate-α-L-iduronic acid)1→4(N-sulfate-6-O-sulfate-D-glucosamine), both prepared for the first time, by chemical synthesis from D-glucose and D-glucosamine, have been confirmed by nuclear magnetic resonance. The synthetic tetrasaccharide neither binds to AT-III nor induces anti-factor Xa activity enhancement of this inhibitor. In contrast, the synthetic pentasaccharide strongly binds to AT-III (Ka : 7.10⁶M^?1) forming an equimolar complex and also enhances the AT-III inhibitory activity towards factor Xa. These results confirm that the synthetic pentasaccharide with the above structure corresponds to the actual minimal sequence required in heparin for binding to AT-III.     

Physiological variant of antithrombin-III lacks carbohydrate sidechain at Asn 135

Both normal antithrombin-III (AT-III alpha) and the high heparin affinity form (AT-III beta) were isolated from pooled human plasma. AT-III beta had a lower negative charge and lower molecular mass than AT-III alpha. Sialidase and endo-F digestion indicated that the inherent difference resided in the oligosaccharide component of the molecule. CNBr fragmentation showed there was an oligosaccharide sidechain missing between residues 104 and 251, subdigestion with trypsin indicated that Asn 135 was not glycosylated in AT-III beta. Chromatography of total tryptic digests on concanavalin A-Sepharose confirmed that the high heparin affinity form of antithrombin lacked an oligosaccharide moiety at Asn 135.     

Production of Human Antithrombin-III in a Serum-free Culture of CHO Cells

A simple method was developed to establish serum-independent Chinese hamster ovary (CHO) cells that grew and secreted high level of human antithrombin-III (AT-III). First, human AT-III and mouse dihydrofolate reductase (DHFR) cDNAs were transfected into DHFR-deficient CHO cells. Transfected cells were treated with increasing concentrations of methotrexate (MTX) and clones secreting high levels of AT-III (10–20 µg/ml/3 day) in a serum-containing medium were obtained. Serum-independent clones were derived from the serum-dependent clones by simply culturing the cells for a few weeks in a serum-free medium. In a serum-free medium the established serum-independent clones grew at normal rate and produced almost equivalent amount of AT-III to that of the serum-dependent, parent clones. In addition, AT-III from the serum-independent clones has specific activity similar to that of plasma-derived AT-III.     

Effect of antithrombin III on glycoprotein Ib/IX/V activation in human platelets: Suppression of thromboxane A(2) generation

We have previously shown that ristocetin, an activator of glycoprotein Ib/IX/V, induces release of soluble CD40 (sCD40) ligand via thromboxane (TX) A(2) production from human platelets. In the present study, we investigated the effect of antithrombin-III (AT-III), an anticoagulant, on the ristocetin-induced glycoprotein Ib/IX/V activation in human platelets. AT-III inhibited ristocetin-stimulated platelet aggregation. The ristocetin-induced production of 11-dehydro-TXB(2), a stable metabolite of TXA(2), and the release of sCD40 ligand were suppressed by AT-III. AT-III also reduced the ristocetin-stimulated secretion of platelet-derived growth factor (PDGF)-AB. AT-III failed to affect U46619-, a TXA(2) receptor agonist, induced levels of p38 mitogen-activated protein kinase phosphorylation or sCD40 ligand release. AT-III reduced the binding of SZ2, a monoclonal antibody to the sulfated sequence in the α-chain of glycoprotein Ib, to the ristocetin-stimulated platelets. These results strongly suggest that AT-III reduced ristocetin-stimulated release of sCD40 ligand due to inhibiting TXA(2) production in human platelets.     

Application of immobilized heparins for isolation of human antithrombin III

Immobilized heparins were prepared by six different methods, and these were utilized for affinity purification of human antithrombin III (AT-III). Affinity support capacities (mg AT-III/g support) were strongly influenced by immobilized active heparin concentrations. In the temperature range 5-37 degrees C, colder temperatures favored affinity adsorption of AT-III as well as nonspecific interactions of all proteins. For representative human-plasma-derived feed solutions the selectivity for AT-III on the affinity support was dependent on relative concentrations of non-AT-III proteins as well as the specific mode of adsorption and elution (batch/continuous).     

Immunoelectrophoretic quantitation of antithrombin III. Electroimmunodiffusion in agarose gels containing heparin.

Antithrombin III (AT-III), being an alpha2-globulin, will have an electrophoretic mobility in the presence of heparin like prealbumin in agarose gels. This phenomenon was utilized to quantitate AT-III from serum and plasma by electroimmunodiffusion (EID) for 90 min agarose gels containing 75 USP units of heparin/ml gel. The method permits a rapid quantitation of AT-III from serum, citrated plasma and EDTA plasma, and a positive correlation was observed between these values and those obtained by single radial immunodiffusion (SRI). This is in contrast to quantitation of AT-III by EID in gels containing no heparin where the values for plasma showed poor correlation with those obtained by SRI.     

Antithrombin III assay using a chromogenic substrate in newborns

The assay of antithrombin III activity (AT-III) was performed by chromogenic synthetic substrate in 42 full-term newborns, 1 to 7 days aged. AT-III levels were lower than normal adults in the first day of life and lessened again in the 3rd day. After this period the levels of AT-III rose slowly and in the 7th day were about half the concentration of adults.     

A model for the stoichiometric regulation of blood coagulation

We have developed a model of the extrinsic blood coagulation system that includes the stoichiometric anticoagulants. The model accounts for the formation, expression, and propagation of the vitamin K-dependent procoagulant complexes and extends our previous model by including: (a) the tissue factor pathway inhibitor (TFPI)-mediated inactivation of tissue factor (TF).VIIa and its product complexes; (b) the antithrombin-III (AT-III)-mediated inactivation of IIa, mIIa, factor VIIa, factor IXa, and factor Xa; (c) the initial activation of factor V and factor VIII by thrombin generated by factor Xa-membrane; (d) factor VIIIa dissociation/activity loss; (e) the binding competition and kinetic activation steps that exist between TF and factors VII and VIIa; and (f) the activation of factor VII by IIa, factor Xa, and factor IXa. These additions to our earlier model generate a model consisting of 34 differential equations with 42 rate constants that together describe the 27 independent equilibrium expressions, which describe the fates of 34 species. Simulations are initiated by "exposing" picomolar concentrations of TF to an electronic milieu consisting of factors II, IX, X, VII, VIIa, V, and VIIII, and the anticoagulants TFPI and AT-III at concentrations found in normal plasma or associated with coagulation pathology. The reaction followed in terms of thrombin generation, proceeds through phases that can be operationally defined as initiation, propagation, and termination. The generation of thrombin displays a nonlinear dependence upon TF, AT-III, and TFPI and the combination of these latter inhibitors displays kinetic thresholds. At subthreshold TF, thrombin production/expression is suppressed by the combination of TFPI and AT-III; for concentrations above the TF threshold, the bolus of thrombin produced is quantitatively equivalent. A comparison of the model with empirical laboratory data illustrates that most experimentally observable parameters are captured, and the pathology that results in enhanced or deficient thrombin generation is accurately described.     

Site-directed mutagenesis of alanine-382 of human antithrombin III

Antithrombin III Hamilton is a structural variant of antithrombin III (AT-III) with normal heparin affinity but impaired serine protease inhibitory activity. The molecular defect of AT-III-Hamilton is a substitution of threonine for alanine at amino acid residue 382. Recently it has been shown that both plasma-derived and cell-free-derived AT-III-Hamilton polypeptides act as substrates rather than inhibitors of thrombin and factor Xa. In the present study, the cell-free expression phagemid vector pGEM-3Zf(+)-AT-III1-432 was mutated at amino acid residue 382 of AT-III to generate 7 cell-free-derived variants. All these cell-free-derived AT-III variants were able to bind heparin as effectively as cell-free-derived normal AT-III. In terms of alpha-thrombin inhibitory activity each variant reacted differently. Variants could be grouped into 3 categories with respect to thrombin-AT-III complex formation: (1) near normal activity (glycine, isoleucine, leucine, valine); (2) low activity (threonine, glutamine); (3) no detectable activity (lysine). These data suggest that mutations at position 382 of AT-III may have a variable effect on protease inhibitory activity, depending on either the stability of the P12-P9 region of the exposed loop of AT-III, or the inability of the amino acid residue at position 382 to interact with a conserved hydrophobic pocket consisting of phenylalanine (at positions 77, 221 and 422) and isoleucine (position 412) residues.     

Evidence for a heparin-induced conformational change on antithrombin III

The effect of heparin on the conformation of antithrombin III (AT-III) was investigated. Solvent perturbation difference spectroscopy shows that the binding of heparin to AT-III results in exposure of two tyrosine residues and a partial burial of a tryptophan residue. The occurrence of a conformational change suggested by this study is also substantiated by circular dichroism (CD) findings in the aromatic and peptide regions. The data in the peptide region show that heparin produces a decrease in the β-structure of AT-III, with a compensatory increase in random coil.     

Antithrombin III,but not C1 esterase inhibitor reduces inflammatory response in lipopolysaccharide-stimulated human monocytes in an ex-vivo whole blood setting

In order to examine the immunomodulatory effects of antithrombin III (AT-III) and C1 esterase inhibitor (C1-INH) in human monocytes, we investigated the intracellular expression of interleukin (IL)-6, IL-8, and tumor necrosis factor (TNF)-α in an ex-vivo laboratory study in a whole blood setting.Heparinized whole blood samples from 23 healthy male and female volunteers (mean age: 27 ± 7 years) were pre-incubated with clinically relevant concentrations of AT-III (n = 11) and C1-INH (n = 12), then stimulated with 0.2 ng/mL lipopolysaccharide (LPS) for 3 h. After phenotyping CD14⁺ monocytes, intracellular expression of IL-6, IL-8, and TNF-α was assessed using flow cytometry. In addition, 12 whole blood samples (AT-III and C1-INH, n = 6 each) were examined using hirudin for anticoagulation; all samples were processed in the same way. To exclude cytotoxicity effects, 7-amino-actinomycin D and Nonidet P40 staining were used to investigate probes.This study is the first to demonstrate the influence of C1-INH and AT-III on the monocytic inflammatory response in a whole blood setting, which mimics the optimal physiological setting. Cells treated with AT-III exhibited significant downregulation of the proportion of gated CD14⁺ monocytes for IL-6 and IL-8, in a dose-dependent manner; downregulation for TNF-α did not reach statistical significance. There were no significant effects on mean fluorescence intensity (MFI). In contrast, C1-INH did not significantly reduce the proportion of gated CD14⁺ monocytes or the MFI regarding IL-6, TNF-α, and IL-8. When using hirudin for anticoagulation, no difference in the anti-inflammatory properties of AT-III and C1-INH in monocytes occurs.Taken together, in contrast to TNF-α, IL-6 and IL-8 were significantly downregulated in monocytes in an ex-vivo setting of human whole blood when treated with AT-III. This finding implicates monocytes as an important point of action regarding the anti-inflammatory properties of AT-III in sepsis. C1-INH was unable to attenuate the monocytic response, which supports the hypothesis that other cellular components in whole blood (e.g., neutrophils) might be responsible for the known effects of C1-INH in inflammation.     

Intron/exon structure of the human gene for the muscle isozyme of glycogen phosphorylase

The intron/exon organization of the human gene for glycogen phosphorylase has been determined. The segments of the polypeptide chain that corresponds to the 19 exons of the gene are examined for relationships between the three-dimensional structure to the protein and gene structure. Only weak correlations are observed between domains of phosphorylase and exons. The nucleotide binding domains that are found in phosphorylase and other glycolytic enzymes are examined for relationships between exons of the genes and structures of the domains. When mapped to the three-dimensional structures, the intron/exon boundaries are shown to be widely distributed in this family of protein domains.