Amino acid sequence and location of the disulfide bonds in bovine beta 2 glycoprotein I: the presence of five Sushi domains.

beta 2 glycoprotein I is a plasma protein with the ability to bind with various kinds of negatively charged substances. The complete amino acid sequence and the location of all the disulfide bonds of bovine beta 2 glycoprotein I were determined. Bovine beta 2 glycoprotein I consists of 326 amino acid residues with five asparagine-linked carbohydrate chains. Homology with the human protein was calculated to be 83%. Eleven disulfide bonds in bovine beta 2 glycoprotein I constitute four characteristic domains, Sushi domains, and one modified form of a Sushi domain.     

The interaction of beta(2)-glycoprotein I domain V with chaperonin GroEL: the similarity with the domain V and membrane interaction

To clarify the mechanism of interaction between chaperonin GroEL and substrate proteins, we studied the conformational changes; of the fifth domain of human beta(2)-glycoprotein I upon binding to GroEL. The fifth domain has a large flexible loop, containing several hydrophobic residues surrounded by positively charged residues, which has been proposed to be responsible for the binding of beta(2)-glycoprotein I to negatively charged phospholipid membranes. The reduction by dithiothreitol of the three intramolecular disulfide bonds of the fifth domain was accelerated in the presence of stoichiometric amounts of GroEL, indicating that the fifth domain was destabilized upon interaction with GroEL. To clarify the GroEL-induced destabilization at the atomic level, we performed H/(2)H exchange of amide protons using heteronuclear NMR spectroscopy. The presence of GroEL promoted the H/(2)H exchange of most of the protected amide protons, suggesting that, although the flexible loop of the fifth domain is likely to be responsible for the initiation of binding to GroEL, the interaction with GroEL destabilizes the overall conformation of the fifth domain.     

An extra disulfide bridge in the constant domain of Rana catesbeiana immunoglobulin light chains

The immunoglobulins of the bullfrog Rana catesbeiana are unusual in that, in all classes, the light chains are not disulfide bonded to heavy chains or to other light chains. Moreover, the light chains contain six, rather than the usual five, residues of half-cystine. As none of these half-cystines is in the sulfhydryl form or is alkylated after mild reduction, we suggested that the light chains probably contain three intrachain disulfide bridges. We have now carried out experiments to confirm the existence of an extra intrachain disulfide bridge in Rana catesbeiana light chains and to determine its location. Disulfide bridge assignments were based on 1) isolation and sequence analysis of S-(carboxymethyl)cysteine-containing peptides and 2) isolation, from unreduced light chains, of peptides containing a disulfide bridge. Half-cystine residues were found at positions 134 and 194, and these were shown to be joined in the conserved intradomain disulfide bridge. In addition, we found that a residue of half-cystine, located at the third position from the carboxy-terminus, forms a disulfide bridge with a half-cystine at position 119, near the amino-terminus of the domain, the latter residue replacing a proline that has been found at this position in all other light chains. An intrachain disulfide bridge has not been found at this location in any other light chain.     

Location of the disulfide bonds in human plasma prekallikrein: the presence of four novel apple domains in the amino-terminal portion of the molecule

The location of 16 of the 18 disulfide bonds in human plasma prekallikrein was determined by amino acid sequence analysis of cystinyl peptides produced by chemical and enzymatic digestions. A unique structure, named the apple domain, was established for each of the four tandem repeats in the amino-terminal portion of the molecule. The apple domains (90 or 91 amino acids) contain 3 highly conserved disulfide bonds linking the first and sixth, second and fifth, and third and fourth half-cystine residues present in each repeat. The fourth tandem repeat contains an extra disulfide bond that forms a second small loop within the apple domain. The carboxyl-terminal portion of plasma prekallikrein containing the catalytic region of the molecule was found to have disulfide bonds located in positions similar to those of other serine proteases.     

Identification of disulfide bonds among the nine core 2 N-acetylglucosaminyltransferase-M cysteines conserved in the mucin beta6-N-acetylglucosaminyltransferase family

Bovine core 2 beta1,6-N-acetylglucosaminyltransferase-M (bC2GnT-M) catalyzes the formation of all mucin beta1,6-N-acetylglucosaminides, including core 2, core 4, and blood group I structures. These structures expand the complexity of mucin carbohydrate structure and thus the functional potential of mucins. The four known mucin beta1,6-N-acetylglucosaminyltransferases contain nine conserved cysteines. We determined the disulfide bond assignments of these cysteines in [(35)S]cysteine-labeled bC2GnT-M isolated from the serum-free conditioned medium of Chinese hamster ovary cells stably transfected with a pSecTag plasmid. This plasmid contains bC2GnT-M cDNA devoid of the 5'-sequence coding the cytoplasmic tail and transmembrane domain. The C18 reversed phase high performance liquid chromatographic profile of the tryptic peptides of reduced-alkylated (35)S-labeled C2GnT-M was established using microsequencing. Each cystine pair was identified by rechromatography of the C8 high performance liquid chromatographic radiolabeled tryptic peptides of alkylated bC2GnT-M on C18 column. Among the conserved cysteines in bC2GnT-M, the second (Cys(113)) was a free thiol, whereas the other eight cysteines formed four disulfide bridges, which included the first (Cys(73)) and sixth (Cys(230)), third (Cys(164)) and seventh (Cys(384)), fourth (Cys(185)) and fifth (Cys(212)), and eighth (Cys(393)) and ninth (Cys(425)) cysteine residues. This pattern of disulfide bond formation differs from that of mouse C2GnT-L, which may contribute to the difference in substrate specificity between these two enzymes. Molecular modeling using disulfide bond assignments and the fold recognition/threading method to search the Protein Data Bank found a match with aspartate aminotransferase structure. This structure is different from the two major protein folds proposed for glycosyltransferases.     

Adhesion mechanism of human beta(2)-glycoprotein I to phospholipids based on its crystal structure.

Human beta(2)-glycoprotein I is a heavily glycosylated five-domain plasma membrane-adhesion protein, which has been implicated in blood coagulation and clearance of apoptotic bodies from the circulation. It is also the key antigen in the autoimmune disease anti-phospholipid syndrome. The crystal structure of beta(2)-glycoprotein I isolated from human plasma reveals an elongated fish-hook-like arrangement of the globular short consensus repeat domains. Half of the C-terminal fifth domain deviates strongly from the standard fold, as observed in domains one to four. This aberrant half forms a specific phospholipid-binding site. A large patch of 14 positively charged residues provides electrostatic interactions with anionic phospholipid headgroups and an exposed membrane-insertion loop yields specificity for lipid layers. The observed spatial arrangement of the five domains suggests a functional partitioning of protein adhesion and membrane adhesion over the N- and C-terminal domains, respectively, separated by glycosylated bridging domains. Coordinates are in the Protein Data Bank (accession No. 1QUB).     

Complete determination of disulfide bonds localized within the short consensus repeat units of decay accelerating factor (CD55 antigen).

Decay accelerating factor (DAF) has 4 SCR (short consensus repeat) units. Each SCR unit consists of approx. 60 amino acids characterized by having four conserved cysteine residues and several other highly conserved residues which include proline, tryptophan, tyrosine/phenylalanine and glycine. To determine the disulfide-bonding pattern, we used the urine form of DAF. After thermolysin and trypsin digestion, we isolated seven disulfide-linked peptides by HPLC purification. Because all of the cysteine residues are disulfide-bonded, DAF should contain eight disulfide bonds. After subtilisin and trypsin digestion, we isolated the eighth disulfide-bonded peptides by HPLC purification. From sequence analyses of these peptides, we could identify all disulfide bonds in the 4 SCR units of DAF as being between the first and the third and between the second and the fourth half-cystines within each SCR unit.     

Location of disulfide bonds within the sequence of human serum cholinesterase

Human serum cholinesterase was digested with pepsin under conditions which left disulfide bonds intact. Peptides were isolated by high pressure liquid chromatography, and those containing disulfide bonds were identified by a color assay. Peptides were characterized by amino acid sequencing and composition analysis. Human serum cholinesterase contains 8 half-cystines in each subunit of 574 amino acids. Six of these form three internal disulfide bridges: between Cys65-Cys92, Cys252-Cys263, and Cys400-Cys519. A disulfide bond with Cys65 rather than Cys66 was inferred by homology with Torpedo acetylcholinesterase. Cys571 forms a disulfide bridge with Cys571 of an identical subunit. This interchain disulfide bridge is four amino acids from the carboxyl terminus. A peptide containing the interchain disulfide is readily cleaved from cholinesterase by trypsin (Lockridge, O., and La Du, B. N. (1982) J. Biol. Chem. 257, 12012-12018), suggesting that the carboxyl terminus is near the surface of the globular tetrameric protein. The disulfide bridges in human cholinesterase have exactly the same location as in Torpedo californica acetylcholinesterase. There is one potential free sulfhydryl in human cholinesterase at Cys66, but this sulfhydryl could not be alkylated. Comparison of human cholinesterase, and Torpedo and Drosophila acetylcholinesterases to the serine proteases suggests that the cholinesterases constitute a separate family of serine esterases, distinct from the trypsin family and from subtilisin.     

Disulfide assignments in recombinant mouse and human interleukin 4

The disulfide pairings of mouse and human interleukin 4 (IL-4) proteins have been determined. The purified proteins, synthesized by recombinant DNA technology, are fully active as judged by their ability to stimulate an appropriate biological response in a variety of functional assays. Peptide maps were produced by digesting the proteins with pepsin and separating the resulting fragments by reverse-phase HPLC using linear acetonitrile-TFA gradients. Cystine-containing peptides were identified by determining which reverse-phase peaks showed an altered elution pattern after reduction. These peptides were purified further and defined by composition and sequence analysis. Three sets of disulfide-linked peptides were consistently identified for each protein. For mouse IL-4, the first and fifth, second and fourth, and third and sixth cysteines are joined. The disulfide bonds in human IL-4 are between the first and sixth, second and fourth, and third and fifth cysteines. A large double-loop region within the central three-fifths of each protein is stabilized by these bonds. Sequence analysis of the peptides containing the third and fifth cysteines of human IL-4 also demonstrated that only one of the potential N-glycosylation sites is used by C127 mammary tumor cells. Complete alkylation of mouse IL-4 under mild conditions completely destroyed its biological activity in a hematopoietic precursor cell proliferation assay.     

A second disulfide bridge from the N-terminal domain to extracellular loop 2 dampens receptor activity in GPR39

A highly conserved feature across all families of 7TM receptors is a disulfide bridge between a Cys residue located at the extracellular end of transmembrane segment III (TM-III) and one in extracellular loop 2 (ECL-2). The zinc sensor GPR39 contains four Cys residues in the extracellular domains. By using mutagenesis, treatment with the reducing agent TCEP, and a labeling procedure for free sulfhydryl groups, we identify the pairing of these Cys residues in two disulfide bridges: the prototypical bridge between Cys (108) in TM-III and Cys (210) in ECL-2 and a second disulfide bridge connecting Cys (11) in the N-terminal domain with Cys (191) in ECL-2. Disruption of the conserved disulfide bond by mutagenesis greatly reduced the level of cell surface expression and eliminated agonist-induced increases in inositol phosphate production but surprisingly enhanced constitutive signaling. Disruption of the nonconserved disulfide bridge by mutagenesis led to an increase in the Zn (2+) potency. This phenotype, with an approximate 10-fold increase in agonist potency and a slight increase in E max, was mimicked by treatment of the wild-type receptor with TCEP at low concentrations, which had no effect on the receptor already lacking the second disulfide bridge and already displaying a high Zn (2+) potency. We conclude that the second disulfide bridge, which according to the beta2-adrenergic structure will form a covalent link across the entrance to the main ligand binding pocket, serves to dampen GPR39 activation. We suggest that formation of extra disulfide bridges may be an important general mechanism for regulating the activity of 7TM receptors.     

The covalent structure of the elastase inhibitor from Anemonia sulcata--a "non-classical" Kazal-type protein

The amino-acid sequence of the proteinase inhibitor specific for elastases from the sea anemone Anemonia sulcata was determined from performic-acid oxidized inhibitor and from three cyanogen bromide fragments of reduced and carboxymethylated inhibitor. The molecule consists of a single polypeptide chain formed from 48 amino-acid residues and is stabilized by three intramolecular disulfide bridges. After cyanogen bromide cleavage of the native protein at methionines 10 and 28 followed by chymotryptic cleavage two fragments each containing a single disulfide bridge were isolated. These indicated the location of three intramolecular disulfide linkages between Cys4 and Cys34 (part of A-loop), Cys8 and Cys27 (B-loop) and Cys16 and Cys48 (C-loop). The sequential homology and the disulfide pattern identified the elastase inhibitor as a Kazal-type inhibitor in which, however, not only the CysI-CysII segment is rather short but interestingly the Cys4-Cys34 disulfide anchoring point (i.e. CysI-CysV) in the C-loop is shifted by one turn in the alpha-helical segment towards the C-terminus. Thus, the elastase inhibitor is a non-classical Kazal-type inhibitor with respect to the positioning of the half-cystines. The inhibitor molecule was modelled based on the known three-dimensional structure of the silver pheasant ovomucoid third domain. The shortened amino-terminal segment was arranged in such a manner to allow disulfide bridge formation between the first cysteine Cys4 and the replaced Cys34 under maintenance of a suitable binding loop conformation. The characteristic ovomucoid scaffold consisting of a central alpha-helix, an adjacent three-stranded beta-sheet and the proteinase-binding loop cross-connected through disulfide bridges CysI-CysV and CysIII-CysVI was conserved.     

Transmembrane domain interactions and residue proline 378 are essential for proper structure, especially disulfide bond formation, in the human vitamin K-dependent gamma-glutamyl carboxylase

We used recombinant techniques to create a two-chain form (residues 1-345 and residues 346-758) of the vitamin K-dependent gamma-glutamyl carboxylase, a glycoprotein located in the endoplasmic reticulum containing five transmembrane domains. The two-chain carboxylase had carboxylase and epoxidase activities similar to those of one-chain carboxylase. In addition, it had normal affinity for the propeptide of factor IX. We employed this molecule to investigate formation of the one disulfide bond in carboxylase, the transmembrane structure of carboxylase, and the potential interactions among the carboxylase's transmembrane domains. Our results indicate that the two peptides of the two-chain carboxylase are joined by a disulfide bond. Proline 378 is important for the structure necessary for disulfide formation. Results with the P378L carboxylase indicate that noncovalent bonds maintain the two-chain structure even when the disulfide bond is disrupted. As we had previously proposed, the fifth transmembrane domain of carboxylase is the last and only transmembrane domain in the C-terminal peptide of the two-chain carboxylase. We show that the noncovalent association between the two chains of carboxylase involves an interaction between the fifth transmembrane domain and the second transmembrane domain. Results of a homology model of transmembrane domains 2 and 5 suggest that not only do these two domains associate but that transmembrane domain 2 may interact with another transmembrane domain. This latter interaction may be mediated at least in part by a motif of glycine residues in the second transmembrane domain.     

The disulfide structure of mouse lysosome-associated membrane protein 1

The disulfide structure of mouse lysosome-associated membrane protein 1 has been determined by reverse-phase isolation and sequence analysis of the cysteine-containing tryptic fragments of the reduced and non-reduced deglycosylated protein. Half-cystines were distinguished (a) by their localization within tryptic or chymotryptic peptides that formed reverse-phase peaks unique to the reduced digests and (b) by their 3H-carboxymethylation only after reduction of the protein. The disulfide arrangement of the cysteines was assigned after isolation of disulfide-linked peptide pairs. Each pair chromatographed as a peak present in the nonreduced (but not the corresponding reduced) tryptic digest. NH2-terminal sequencing as well as reduction, alkylation, and rechromatography of the disulfide-linked fragments led to the following assignment of disulfide bonds: Cys11 and Cys50, Cys125 and Cys161, Cys198 and Cys235, and Cys303 and Cys340. This structure creates four 36-38-residue loops that are symmetrically placed within the two halves of the protein's intraluminal domain. The loops formed by the Cys11-Cys50 and Cys198-Cys235 bridges are homologous, and the Cys125-Cys161 and Cys303-cys340 loops form a second set of homologous domains. The conservation of cysteine residues among lysosome-associated membrane proteins 1 and 2 suggests that this disulfide arrangement is common to both members of this family of lysosomal membrane glycoproteins.     

Identification of the ligand-binding regions in the macrophage colony-stimulating factor receptor extracellular domain.

The c-fms gene encodes the receptor for the macrophage colony-stimulating factor (M-CSF), and its extracellular domain consists of five immunoglobulin-like subdomains. To identify which of the five immunoglobulin-like regions are involved in ligand binding, we polymerase chain reaction-cloned five segments of the extracellular domain of the murine c-fms gene, each starting with the normal initiation codon and containing successive additions of the immunoglobulin-like subdomains. These protein segments are designated A, B, C, D, and E and contain, from the N-terminal end, either one, two, three, four, or all five immunoglobulin-like subdomains, respectively. Each segment was expressed as a secreted soluble protein from a baculovirus expression vector in Sf9 insect cells. In addition, segments A, B, C, and E were produced as soluble alkaline phosphatase fusion proteins, as was a segment containing only the fourth and fifth immunoglobulin domains. These segments of the Fms extracellular domain were used to assess M-CSF binding by competition radioimmunoassays, plate binding immunoassays, and immunoprecipitation analyses. The results indicated that the first two N-terminal immunoglobulin-like domains did not interact with M-CSF but, in combination with the third immunoglobulin-like domain, provided high-affinity M-CSF binding. The fourth and fifth immunoglobulin-like domains near the cell membrane did not exhibit M-CSF binding and may inhibit interaction of M-CSF with the first three immunoglobulin domains. These results suggest that the three N-terminal immunoglobulin-like domains constitute the high-affinity M-CSF binding region and that the fourth and fifth immunoglobulin-like domains may perform functions other than ligand binding.     

Use of single point mutations in domain I of beta 2-glycoprotein I to determine fine antigenic specificity of antiphospholipid autoantibodies

Autoantibodies against beta(2)-glycoprotein I (beta(2)GPI) appear to be a critical feature of the antiphospholipid syndrome (APS). As determined using domain deletion mutants, human autoantibodies bind to the first of five domains present in beta(2)GPI. In this study the fine detail of the domain I epitope has been examined using 10 selected mutants of whole beta(2)GPI containing single point mutations in the first domain. The binding to beta(2)GPI was significantly affected by a number of single point mutations in domain I, particularly by mutations in the region of aa 40-43. Molecular modeling predicted these mutations to affect the surface shape and electrostatic charge of a facet of domain I. Mutation K19E also had an effect, albeit one less severe and involving fewer patients. Similar results were obtained in two different laboratories using affinity-purified anti-beta(2)GPI in a competitive inhibition ELISA and with whole serum in a direct binding ELISA. This study confirms that anti-beta(2)GPI autoantibodies bind to domain I, and that the charged surface patch defined by residues 40-43 contributes to a dominant target epitope.     

Complete primary structure of a galactose-specific lectin from the venom of the rattlesnake Crotalus atrox. Homologies with Ca2(+)-dependent-type lectins

The complete primary structure of a galactose-specific lectin contained in the venom of the rattlesnake, Crotalus atrox, was determined. The lectin is composed of two covalently linked, identical subunits, each consisting of 135 amino acid residues. Under physiological conditions the lectin proved to be highly aggregated. The venom lectin contained 9 half-cystines, 8 of which formed four intrasubunit disulfide bridges (Cys3-Cys14, Cys31-Cys131, Cys38-Cys133, and Cys106-Cys123), while Cys86 was involved in an intersubunit disulfide bridge. Because of the high content of disulfide bridges, the intact lectin was extremely resistant to tryptic digestion. The determined amino acid sequence was found to be homologous with those of the so-called carbohydrate recognition domains of Ca2(+)-dependent-type lectins in animal. Among them, 8 amino acid residues (Cys31, Gly69, Trp92, Pro97, Cys106, Asp120, Cys123, and Cys131) were completely conserved. Leu40, Trp67, and Trp81 were also well conserved. The rattlesnake venom lectin showed high hemagglutinating activity. These results, together with the occurrence of similar lectins in crotalid venoms, suggest that these lectins have evolved in order to make the venom a more effective weapon to capture prey animals.     

The structure of the murine Fc receptor for IgG. Assignment of intrachain disulfide bonds, identification of N-linked glycosylation sites, and evidence for a fourth form of Fc receptor

The Fc receptor (Fc gamma R) of the murine macrophage cell line, J774, was purified by immunoaffinity chromatography then subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and amino-terminal sequencing. FcR material judged to be pure by these criteria was digested with a number of enzymes to identify the cysteine residues engaged in disulfide bonds within the native structure. The results clearly establish that the mouse macrophage Fc gamma R contains two intrachain disulfide bonds, each of which connects adjacent cysteine residues within the two putative extracellular domains of the molecule. In addition, each disulfide-bonded domain was shown to contain two authentic sites of N-linked glycosylation. Extensive peptide sequencing resulted in the unexpected identification of peptide fragments from a fourth Fc gamma R whose sequences were highly homologous to sequences surrounding the two Cys residues in the amino-terminal domain of both alpha and beta 1 Fc gamma R. The fourth Fc gamma R contains a disulfide-bonded amino-terminal domain similar to beta 1 Fc gamma R.     

Human tissue factor contains thioester-linked palmitate and stearate on the cytoplasmic half-cystine

The state of the five half-cystine residues in human tissue factor (TF) has been characterized. The results indicate that the four half-cystines in the extracellular domain of TF form two disulfide bonds and the half-cystine in the cytoplasmic region is acylated by palmitic acid and stearic acid. The extracellular disulfide cross-links, Cys49-Cys57 and Cys186-Cys209, were deduced from the analysis of tryptic peptides. Acylation of the cytoplasmic half-cystine was demonstrated by purifying and characterizing fibroblast TF from cells labeled with [3H]palmitic acid. Radiolabeled fibroblast TF was observed by autoradiography following sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The tritiated material covalently bound to the protein was identified as [3H]palmitate and [3H]stearate by reverse-phase high-pressure liquid chromatography. Deacylation of TF with hydroxylamine resulted in the spontaneous generation of disulfide-linked TF dimers. This result suggests that the disulfide-linked TF dimer, a minor component of most TF preparations, and the recently described heterodimeric form of TF are artifacts produced by deacylation of Cys245 and subsequent interchain disulfide bond formation.     

Substructural analysis of the insulin receptor by microsequence analyses of limited tryptic fragments isolated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis in the absence or presence of dithiothreitol

Human placental insulin receptor contains 47 Cys per an alpha beta dimer. Most of the 94 Cys in an intact alpha 2 beta 2 receptor are expected to form interchain or intrachain disulfide bonds, since there appears to be only one free cysteine residue in each beta subunit. In order to gain more insight into the three-dimensional organization of the insulin receptor, we have used limited trypsin digestion, SDS-PAGE, and protein microsequencing. The present study revealed the following; major tryptic cleavages occurred at alpha 164, alpha 270, alpha 582, and beta 1115, generating Mr 175,000, 130,000, 100,000, 70,000, and 55,000 disulfide-linked complexes. Under reducing conditions, tryptic fragments of Mr values = 30,000, 70,000, 20,000, 55,000, and 20,000 were identified to be alpha(1-164), alpha(165-582), alpha(165-270), alpha(271-582), and alpha(583-C-terminal), respectively. The major beta subunit tryptic fragment of Mr = 55,000 was assumed to have beta(724-1115) or beta(N-terminal-392). The Mr 175,000 complex appeared to contain two alpha(1-164) and two alpha(165-582), whereas the Mr 70,000 complex contained alpha(583-C-terminal) and beta(724-1115). Tryptic cleavage at alpha 582 apparently produced one Mr 175,000 and two Mr 70,000 complexes, suggesting that the alpha(583-C-terminal) domain interacts with the extracellular domain of the beta subunit by disulfide bonds. Tryptic cleavage at alpha 270 resulting in a formation of one Mr 100,000 complex consisting of two alpha(1-270) and two Mr 130,000 complexes consisting of alpha(271-C-terminal) and beta(724-1115) suggests that Cys residues involved with disulfide bonds between the two alpha subunits are located in the alpha(1-270) domain. The identification of the Mr 55,000 complex consisting of small tryptic fragments between alpha(122-270) indicates that 40 Cys residues in the two alpha(122-270) domains are inter- and intramolecularly associated by disulfide bonds. The alpha(1-121) domain does not appear to be linked to any other domains by disulfide bonds. These results are consistent with the structural model that the N-terminal domains of alpha subunits (122-270) are disulfide-linked together while the C-terminal domain (583-C-terminal) of the alpha subunit is linked to the N-terminal domain of the beta subunit by disulfide bonds.     

Further localization of binding sites for thrombin and protein C in human thrombomodulin

To elucidate the binding sites for thrombin and protein C in the six epidermal growth factor (EGF) domains of human thrombomodulin, recombinant mutant proteins were expressed in COS-1 cells. Mutant protein EGF456, which contains the fourth, fifth, and sixth EGF domains from the NH2 terminus of thrombomodulin, showed complete cofactor activity in thrombin-catalyzed protein C activation, as did intact thrombomodulin or elastase-digested thrombomodulin. EGF56, containing the fifth and sixth EGF domains, did not have cofactor activity; but EGF45, containing the fourth and fifth EGF domains, had about one-tenth of the cofactor activity of EGF456. Thrombin binding to attached recombinant thrombomodulin (D123) was inhibited by EGF45 as well as by EGF56. A synthetic peptide (ECPEGYILDDGFICTDIDE), corresponding to Glu-408 to Glu-426 in the fifth EGF domain, inhibited thrombin binding to attached thrombomodulin (D123) with an apparent Ki of 95 microM. At Ca2+ concentrations of 0.25-0.3 mM, intact protein C was maximally activated by thrombin in the presence of EGF45, EGF456, or EGF1-6, which contains the first to sixth EGF domains; but such maximum cofactor activity was not observed when gamma-carboxyglutamic acid-domainless protein C was used. These findings suggest that: 1) thrombin binds to the latter half of the fifth EGF domain; and 2) protein C binds to the fourth EGF domain of thrombomodulin through Ca2+ ions.