Localization of disulfide bonds in the cystine knot domain of human von Willebrand factor

von Willebrand factor (VWF) is a multimeric glycoprotein that is required for normal hemostasis. After translocation into the endoplasmic reticulum, proVWF subunits dimerize through disulfide bonds between their C-terminal cystine knot-like (CK) domains. CK domains are characterized by six conserved cysteines. Disulfide bonds between cysteines 2 and 5 and between cysteines 3 and 6 define a ring that is penetrated by a disulfide bond between cysteines 1 and 4. Dimerization often is mediated by additional cysteines that differ among CK domain subfamilies. When expressed in a baculovirus system, recombinant VWF CK domains (residues 1957-2050) were secreted as dimers that were converted to monomers by selective reduction and alkylation of three unconserved cysteine residues: Cys(2008), Cys(2010), and Cys(2048). By partial reduction and alkylation, chemical and proteolytic digestion, mass spectrometry, and amino acid sequencing, the remaining intrachain disulfide bonds were characterized: Cys(1961)-Cys(2011) (), Cys(1987)-Cys(2041) (), Cys(1991)-Cys(2043) (), and Cys(1976)-Cys(2025). The mutation C2008A or C2010A prevented dimerization, whereas the mutation C2048A did not. Symmetry considerations and molecular modeling based on the structure of transforming growth factor-beta suggest that one or three of residues Cys(2008), Cys(2010), and Cys(2048) in each subunit mediate the covalent dimerization of proVWF.     

Factor XIa dimer in the activation of factor IX

Factor XI, unlike other coagulation proteins, is a homodimer of two identical subunits linked by a single disulfide bond formed by Cys321. The present study was undertaken to understand the physiological significance of the dimeric nature of factor XI. We have expressed a mutant FXI/G326C in which the Gly326 residue of factor XI has been mutated to Cys326, reasoning that Cys321 would form an intrachain disulfide bond with Cys326 as in prekallikrein, a plasma protein that exists as a monomer even with 58% amino acid sequence identity and a domain structure very similar to factor XI. No free thiol could be detected in the expressed protein, and it migrated as a monomer on nonreduced SDS-PAGE. In physiological buffer, however, the protein was found to exist in a state of monomer-dimer equilibrium as assessed by gel-filtration chromatography and ultracentrifugation studies (K(d) approximately 36 nM). Functional studies revealed that FXI/G326C was indistinguishable from plasma factor XI in a plasma-clotting assay and in a factor IX activation assay both in the presence and absence of activated platelets even at concentrations at which less than 5% of the mutant exists as dimers. We conclude that, for optimal function in the presence of activated platelets, a preformed dimer of factor XI is not required.     

Sequential comparison of peptides containing half-cystine residues from ovalbumins of six avian species

Hen ovalbumin contains one cystine disulfide (Cys73-Cys120) and four cysteine sulfhydryl groups (Cys11, Cys30, Cys367, and Cys382) in a single polypeptide chain of 385 amino acid residues. To investigate whether or not such a structure is shared by related avian species, the contents of disulfide-involved half-cystine residues and their positions in the primary structure of ovalbumins from five species were compared with those of hen ovalbumin. Ovalbumins were alkylated with a fluorescent dye, IAEDANS, under disulfide-reduced and disulfide-intact conditions and digested with a number of proteolytic enzymes. The sequences were deduced from peptides containing half-cystine residues labeled with the fluorescent dye. The results showed that the number of free cysteine sulfhydryl groups of ovalbumins was different among the species, three for guinea fowl and turkey (Cys11, Cys367, and Cys382); and two for Pekin duck, mallard duck, and Emden goose (Cys11 and Cys331). On the other hand, a single intrachain disulfide bond could be identified from ovalbumins of five species using a combination of peptide mapping and N-terminal amino acid sequencing analysis under reduced and non-reduced conditions, in which the intrachain disulfide bond was like that of hen ovalbumin (Cys73-Cys120). The results also indicated that the variations in amino acid sequences on these peptides containing half-cystine residues bear a close relationship with the phylogeny of the six species.     

Catalysis of creatine kinase refolding by protein disulfide isomerase involves disulfide cross-link and dimer to tetramer switch

Protein disulfide isomerase (PDI) functions as an isomerase to catalyze thiol:disulfide exchange, as a chaperone to assist protein folding, and as a subunit of prolyl-4-hydroxylase and microsomal triglyceride transfer protein. At a lower concentration of 0.2 microm, PDI facilitated the aggregation of unfolded rabbit muscle creatine kinase (CK) and exhibited anti-chaperone activity, which was shown to be mainly due to the hydrophobic interactions between PDI and CK and was independent of the cross-linking of disulfide bonds. At concentrations above 1 microm, PDI acted as a protector against aggregation but an inhibitor of reactivation during CK refolding. The inhibition effect of PDI on CK reactivation was further characterized as due to the formation of PDI-CK complexes through intermolecular disulfide bonds, a process involving Cys-36 and Cys-295 of PDI. Two disulfide-linked complexes containing both PDI and CK were obtained, and the large, soluble aggregates around 400 kDa were composed of 1 molecule of tetrameric PDI and 2 molecules of inactive intermediate dimeric CK, whereas the smaller one, around 200 kDa, was formed by 1 dimeric PDI and 1 dimeric CK. To our knowledge this is the first study revealing that PDI could switch its conformation from dimer to tetramer in its functions as a foldase. According to the observations in this research and our previous study of the folding pathways of CK, a working model was proposed for the molecular mechanism of CK refolding catalyzed by PDI.     

Amino acid sequence and carbohydrate-binding analysis of the N-acetyl-D-galactosamine-specific C-type lectin, CEL-I, from the Holothuroidea, Cucumaria echinata

CEL-I is one of the Ca2+-dependent lectins that has been isolated from the sea cucumber, Cucumaria echinata. This protein is composed of two identical subunits held by a single disulfide bond. The complete amino acid sequence of CEL-I was determined by sequencing the peptides produced by proteolytic fragmentation of S-pyridylethylated CEL-I. A subunit of CEL-I is composed of 140 amino acid residues. Two intrachain (Cys3-Cys14 and Cys31-Cys135) and one interchain (Cys36) disulfide bonds were also identified from an analysis of the cystine-containing peptides obtained from the intact protein. The similarity between the sequence of CEL-I and that of other C-type lectins was low, while the C-terminal region, including the putative Ca2+ and carbohydrate-binding sites, was relatively well conserved. When the carbohydrate-binding activity was examined by a solid-phase microplate assay, CEL-I showed much higher affinity for N-acetyl-D-galactosamine than for other galactose-related carbohydrates. The association constant of CEL-I for p-nitrophenyl N-acetyl-beta-D-galactosaminide (NP-GalNAc) was determined to be 2.3 x 10(4) M(-1), and the maximum number of bound NP-GalNAc was estimated to be 1.6 by an equilibrium dialysis experiment.     

MURF-1 and MURF-2 target a specific subset of myofibrillar proteins redundantly: towards understanding MURF-dependent muscle ubiquitination

MURF-1, MURF-2 and MURF-3 are a specific class of RING finger proteins that are expressed in striated muscle tissues. MURF-1 has been suggested to act as an ubiquitin ligase, thereby controlling proteasome-dependent degradation of muscle proteins. Here, we performed yeast two-hybrid (YTH) screens of skeletal muscle cDNA libraries with MURF-1 baits to identify potential myocellular targets of MURF-1-dependent ubiquitination. This identified eight myofibrillar proteins as binding partners of MURF-1: titin, nebulin, the nebulin-related protein NRAP, troponin-I (TnI), troponin-T (TnT), myosin light chain 2 (MLC-2), myotilin and T-cap. YTH mating studies with MURF-1,2,3 baits indicated that these eight myofibrillar proteins are all targeted redundantly by both MURF-1 and MURF-2. Western blot studies on cardiac tissues from wild-type and MURF-1-deficient mice suggested that titin and nebulin were ubiquitinated at similar levels, and MLC-2 and TnI at reduced levels in MURF-1 KO mice. Mapping of the TnI and titin binding sites on MURF-1 peptide scans demonstrated their binding to motifs highly conserved between MURF-1 and MURF-2. Our data are consistent with a model in which MURF-1 and MURF-2 together target a specific set of myofibrillar proteins redundantly, most likely to control their ubiquitination-dependent degradation. Finally, our YTH screens identified the interaction of MURF-1 with 11 enzymes required for ATP/energy production in muscle including the mitochondrial ATP synthase and cytoplasmic creatine kinase. These data raise the possibility that MURF-1 may coordinately regulate the energy metabolism of mitochondrial and cytoplasmic compartments.     

Structure and function in rhodopsin: effects of disulfide cross-links in the cytoplasmic face of rhodopsin on transducin activation and phosphorylation by rhodopsin kinase.

Six rhodopsin mutants containing disulfide cross-links between different cytoplasmic regions were prepared: disulfide bond 1, between Cys65 (interhelical loop I-II) and Cys316 (end of helix VII); disulfide bond 2, between Cys246 (end of helix VI) and Cys312 (end of helix VII); disulfide bond 3, between Cys139 (end of helix III) and Cys248 (end of helix VI); disulfide bond 4, between Cys139 (end of helix III) and Cys250 (end of helix VI); disulfide bond 5, between Cys135 (end of helix III) and Cys250 (end of helix VI); and disulfide bond 6, between Cys245 (end of helix VI) and Cys338 (C-terminus). The effects of local restrictions caused by the cross-links on transducin (G(T)) activation and phosphorylation by rhodopsin kinase (RK) following illumination were studied. Disulfide bond 1 showed little effect on either G(T) activation or phosphorylation by RK, suggesting that the relative motion between interhelical loop I-II and helix VII is not crucial for recognition by G(T) or by RK. In contrast, disulfide bonds 2-5 abolished both G(T) activation and phosphorylation by RK. Disulfide bond 6 resulted in enhanced G(T) activation but abolished phosphorylation by RK, suggesting the structure recognized by G(T) was stabilized in this mutant by cross-linking of the C-terminus to the cytoplasmic end of helix VI. Thus, the consequences of the disulfide cross-links depended on the location of the restriction. In particular, relative motions of helix VI, with respect to both helices III and VII upon light activation, are required for recognition of rhodopsin by both G(T) and RK. Further, the conformational changes in the cytoplasmic face that are necessary for protein-protein interactions need not be cooperative, and may be segmental.     

Generation of the oxidized form protects human brain type creatine kinase against cystine-induced inactivation

Cystine accumulation in cystinotic patients has been reported to inhibit brain type creatine kinase (BBCK), an important thiol-containing enzyme in energy homeostasis. In this research, we found that the oxidized form of BBCK (O-BBCK) was induced by cystine, and the intramolecular disulfide bond of O-BBCK was formed between Cys74 and Cys254. The wild type BBCK was found to be more resistant to the inactivation induced by cystine when compared to the single point mutant C74S or C254S. Meanwhile, the existence of GSH could protect the wild type BBCK more efficiently than the mutants. These observations suggested that the ability to generate the oxidized form could protect BBCK against the intracellular oxidative stress.     

Theophylline-induced apoptosis is paralleled by protein kinase A-dependent tissue transglutaminase activation in cancer cells

beta(2)-Microglobulin (beta2M), the light chain of the type I major histocompatibility complex, is a major component of dialysis-related amyloid fibrils. beta2M in the native state has a typical immunoglobulin fold with a buried intrachain disulfide bond. The conformation and stability of recombinant beta2M in which the intrachain disulfide bond was reduced were studied by CD, tryptophan fluorescence, and one-dimensional NMR. The conformation of the reduced beta2M in the absence of denaturant at pH 8.5 was similar to that of the intact protein unless the thiol groups were modified. However, reduction of the disulfide bond decreased the stability as measured by denaturation in guanidine hydrochloride. Intact beta2M formed amyloid fibrils at pH 2.5 by extension reaction using sonicated amyloid fibrils as seeds. Under the same conditions, reduced beta2M did not form typical amyloid fibrils, although it inhibited fibril extension competitively, suggesting that the conformation defined by the disulfide bond is important for amyloid fibril formation of beta2M.     

Impact of intra-subunit domain-domain interactions on creatine kinase activity and stability

Creatine kinase (CK) is a key enzyme in vertebrate excitable tissues. In this research, five conserved residues located on the intra-subunit domain-domain interface were mutated to explore their role in the activity and structural stability of CK. The mutations of Val72 and Gly73 decreased both the activity and stability of CK. The mutations of Cys74 and Val75, which had no significant effect on CK activity and structure, gradually decreased the stability and reactivation of CK. Our results suggested that the mutations might modify the correct positioning of the loop contributing to domain-domain interactions, and result in decreased stability against denaturation.     

Nonclottable fibrin obtained from partially reduced fibrinogen: characterization and tissue plasminogen activator stimulation.

Out of 29 disulfide bonds in human fibrinogen, 7 were cleaved during limited reduction under nondenaturing conditions in calcium-free buffer: 2 A alpha 442Cys-A alpha 472Cys and 2 gamma 326Cys-gamma 339Cys intrachain disulfide bonds in the carboxy-terminal ends of the A alpha- and gamma-chains and the symmetrical disulfide bonds at gamma 8Cys, gamma 9Cys, and A alpha 28Cys. We studied the loss of thrombin clottability that followed limited reduction and the increase in the susceptibility of the fibrinogen A alpha 19-A alpha 20 bond to hydrolysis by thrombin. Using differential scanning calorimetry, we show that the extent of unfolding and denaturation of specific domains following limited reduction is small. Heat absorption peaks corresponding to the melting of the major regions of compact structure give high calorimetric enthalpies, as in untreated nonreduced fibrinogen, indicating that substantial regions of native structure are still present in partially reduced fibrinogen. Thrombin releases fibrinopeptide A at an identical rate as in nonreduced fibrinogen while fibrinopeptide B release is slower. Sedimentation velocity studies show that thrombin treatment leads to complex formation; however, gelation does not occur. Amino-terminal analysis indicates that the second thrombin cleavage in the A alpha-chain at A alpha 19-A alpha 20 takes place only after fibrinopeptide A release. Thus, the loss of clottability appears to result from perturbation of carboxy-terminal polymerization sites, probably a consequence of gamma 326Cys-gamma 339Cys intrachain disulfide bond cleavage. The thrombin-treated partially reduced fibrinogen remains soluble in buffered saline and fully expresses at least one epitope, B beta 15-21, unique to fibrin. Furthermore, this nonclottable form accelerates the tissue plasminogen activator dependent conversion of plasminogen to plasmin.     

Extrinsic 33-kilodalton protein of spinach oxygen-evolving complexes: kinetic studies of folding and disulfide reduction

The 33-kDa protein is one of the three extrinsic proteins in the oxygen-evolving photosystem II complexes. The protein has one intrachain disulfide bond. On reduction of this disulfide bond, the protein was unfolded and lost its activity. On the basis of the unfolding equilibrium curve obtained by using guanidine hydrochloride, the free energy change of unfolding in the absence of guanidine hydrochloride was estimated to be 4.4 kcal/mol using the Tanford method [Tanford, C. (1970) Adv. Protein Chem. 24, 1-95] and 2.8 kcal/mol using the linear extrapolation method. The unfolding of the 33-kDa protein caused by reduction was explained in terms of the entropy change associated with reduction of the intrachain disulfide bond. The kinetics of the reduction of the disulfide bond using dithiothreitol were studied at various concentrations of guanidine hydrochloride at pH 7.5 and 25 degrees C. The disulfide bond was reduced even in the absence of guanidine hydrochloride. The unfolding and refolding kinetics of the 33-kDa protein using guanidine hydrochloride were also studied under the same conditions, and the results were compared with those for the reduction kinetics. It was shown that the reduction of the disulfide bond proceeds through a species in which the disulfide bond is exposed by local fluctuations.     

Localization of creatine kinase isoenzymes in myofibrils. I. Chicken skeletal muscle

Purified, repeatedly washed, skeletal muscle myofibrils contain approx. 0.2 U of creatine kinase (CK) activity (equivalent to 2.5 micrograms CK) per milligram dry weight; this firmly bound CK activity is estimated to represent 3-5% of the total cellular CK. It had been shown previously that the myofibrillar CK, which can be quantitatively extracted at low ionic strength and purified to homogeneity, is very similar, if not identical, to the bulk MM-CK. It is shown that the two protein preparations also have the same peptide pattern after cyanogen bromide fractionation and very similar specific activities, confirming their identity. The earlier demonstration that the bound CK is specifically located at the M-lines of isolated myofibrils has been confirmed by immunofluorescence. Antibodies directed against purified MM- and BB-CK were used in the indirect fluorescent antibody technique to study the specificity of myofibril binding sites for different forms of CK. With myofibrils from adult muscle, which has only MM-CK, as well as from early developmental stages in which BB-CK is the predominant isoenzyme, M-type CK was localized exclusively at the M-line, while greater or lesser amounts of B-type CK were found at the Z-line. The data provide strong evidence that the MM-CK at the M-lines in skeletal myofibrils is not adventitiously bound but is rather an integral element in the M-line structure. The amount of CK bound is reasonably consistent with the earlier proposal that the CK molecules might be the transverse M-bridges and appears to be sufficient to regenerate all of the ATP hydrolyzed during muscle contraction.     

Replacement of the interchain disulfide bridge-forming amino acids A7 and B7 by glutamate impairs the structure and activity of insulin

Insulin contains three disulfide bonds, one intrachain bond, A6-A11, and two interchain bonds, A7-B7 and A20-B19. Site-directed mutagenesis results (the two cysteine residues of disulfide A7-B7 were replaced by serine) showed that disulfide A7-B7 is crucial to both the structure and activity of insulin. However, chemical modification results showed that the insulin analogs still retained relatively high biological activity when A7Cys and B7Cys were modified by chemical groups with a negative charge. Did the negative charge of the modification groups restore the loss of activity and/or the disturbance of structure of these insulin analogs caused by deletion of disulfide A7-B7? To answer this question, an insulin analog with both A7Cys and B7Cys replaced by Glu, which has a long side-chain and a negative charge, was prepared by protein engineering, and its structure and activity were analyzed. Both the structure and activity of the present analog are very similar to that of the mutant with disulfide A7-B7 replaced by Ser, but significantly different from that of wild-type insulin. The present results suggest that removal of disulfide A7-B7 will result in serious loss of biological activity and the native conformation of insulin, even if the disulfide is replaced by residues with a negative charge.     

Cysteine residues required for the attachment of the light chain in human IgA2

In humans, there are two subclasses of IgA, IgA1 and IgA2, with IgA2 existing as three allotypes, IgA2m(1), IgA2m(2) and IgA2(n). In IgA1, Cys(133) in C(H)1 forms the disulfide bond to the L chain. Our previous studies indicated that in IgA2 lacking Cys(133), a disulfide bond forms between the alpha-chain and the L chain when Cys(220) is followed by Arg(221), but not when Cys(220) is followed by Pro(221), suggesting that the Cys in C(H)1 might be involved in disulfide bonding to the L chain. However, here we show that covalent assembly of the H and L chains in IgA2(n) requires hinge-proximal Cys(241) and Cys(242) in C(H)2 and not Cys(196) or Cys(220) in C(H)1. Using pulse-chase experiments, we have demonstrated that wild-type IgA2(n) with Arg(221) and Cys(241) and Cys(242) assembles through a disulfide-bonded HL intermediate. In contrast, the major intermediate for IgA2 m(1) with Pro(221) assembly was H(2) even though both Cys(241) and Cys(242) were present. Only a small fraction of IgA2 m(1) assembles through disulfide-bonded HL. Overall, our studies indicate that for IgA2 covalent assembly of the H and L chains requires the hinge-proximal cysteines in C(H)2 and that the structure of C(H)1 influences the efficiency with which this covalent bond forms.     

Effect of a disulfide bond on mevalonate kinase

Mevalonate kinase is one of ATP-dependent enzymes in the mevalonate pathway and catalyzes the phosphorylation of mevalonate to form mevalonate 5-phosphate. In animal cells, it plays a key role in regulating biosynthesis of cholesterol, while in microorganisms and plants, it is involved in the biosynthesis of isoprenoid derivatives that are one of the largest groups of natural products. Crystal structure and sequence alignment show that a unique disulfide bond exists in mevalonate kinase of thermostable species Methanococcus jannaschii, but not in rat mevalonate kinase. In the present study, we investigated the effect of the disulfide bond in M. jannaschii mevalonate kinase and an engineered disulfide bond in rat mevalonate kinase mutant A141C on the properties of enzymes through characterization of their wild-type and variant enzymes. Our result suggests that the Cys107-Cys281 disulfide bond is important for maintaining the conformation and the thermal activity of M. jannaschii mevalonate kinase. Other interactions could also have contributions. The thiol-titration and fluorescence experiment further indicate that rat mevalonate kinase A141C variant enzyme has a new disulfide bond, which makes the variant protein enhance its thermal activity and resist to urea denaturation.     

Thioredoxin-dependent enzymatic activation of mercaptopyruvate sulfurtransferase. An intersubunit disulfide bond serves as a redox switch for activation

Rat 3-mercaptopyruvate sulfurtransferase (MST) contains three exposed cysteines as follows: a catalytic site cysteine, Cys(247), in the active site and Cys(154) and Cys(263) on the surface of MST. The corresponding cysteine to Cys(263) is conserved in mammalian MSTs, and Cys(154) is a unique cysteine. MST has monomer-dimer equilibrium with the assistance of oxidants and reductants. The monomer to dimer ratio is maintained at approximately 92:8 in 0.2 m potassium phosphate buffer containing no reductants under air-saturated conditions; the dimer might be symmetrical via an intersubunit disulfide bond between Cys(154) and Cys(154) and between Cys(263) and Cys(263), or asymmetrical via an intersubunit disulfide bond between Cys(154) and Cys(263). Escherichia coli reduced thioredoxin (Trx) cleaved the intersubunit disulfide bond to activate MST to 2.3- and 4.9-fold the levels of activation of dithiothreitol (DTT)-treated and DTT-untreated MST, respectively. Rat Trx also activated MST. On the other hand, reduced glutathione did not affect MST activity. E. coli C35S Trx, in which Cys(35) was replaced with Ser, formed some adducts with MST and activated MST after treatment with DTT. Thus, Cys(32) of E. coli Trx reacted with the redox-active cysteines, Cys(154) and Cys(263), by forming an intersubunit disulfide bond and a sulfenyl Cys(247). A consecutively formed disulfide bond between Trx and MST must be cleaved for the activation. E. coli C32S Trx, however, did not activate MST. Reduced Trx turns on a redox switch for the enzymatic activation of MST, which contributes to the maintenance of cellular redox homeostasis.     

Stability and structure-forming properties of the two disulfide bonds of alpha-conotoxin GI

alpha-Conotoxin GI is a 13 residue snail toxin peptide cross-linked by Cys2-Cys7 and Cys3-Cys13 disulfide bridges. The formation of the two disulfide bonds by thiol/disulfide exchange with oxidized glutathione (GSSG) has been characterized. To characterize formation of the first disulfide bond in each of the two pathways by which the two disulfide bonds can form, two model peptides were synthesized in which Cys3 and Cys13 (Cono-1) or Cys2 and Cys7 (Cono-2) were replaced by alanines. Equilibrium constants were determined for formation of the single disulfide bonds of Cono-1 and Cono-2, and an overall equilibrium constant was measured for formation of the two disulfide bonds of alpha-conotoxin GI in pH 7.00 buffer and in pH 7. 00 buffer plus 8 M urea using concentrations obtained by HPLC analysis of equilibrium thiol/disulfide exchange reaction mixtures. The results indicate a modest amount of cooperativity in the formation of the second disulfide bond in both of the two-step pathways by which alpha-conotoxin GI folds into its native structure at pH 7.00. However, when considered in terms of the reactive thiolate species, the results indicate substantial cooperativity in formation of the second disulfide bond. The solution conformational and structural properties of Cono-1, Cono-2, and alpha-conotoxin GI were studied by 1H NMR to identify structural features which might facilitate formation of the disulfide bonds or are induced by formation of the disulfide bonds. The NMR data indicate that both Cono-1 and Cono-2 have some secondary structure in solution, including some of the same secondary structure as alpha-conotoxin GI, which facilitates formation of the second disulfide bond by thiol/disulfide exchange. However, both Cono-1 and Cono-2 are considerably less structured than alpha-conotoxin GI, which indicates that formation of the second disulfide bond to give the Cys2-Cys7, Cys3-Cys13 pairing induces considerable structure into the backbone of the peptide.     

The status of half-cystine residues and locations of N-glycosylated asparagine residues in human eosinophil peroxidase

The determination by protein chemistry methods of the half-cystine status in human eosinophil peroxidase (EPO) is reported. EPO is two-chained and has a total of 14 half-cystine residues. Cys141 and Cys152 form an intrachain bridge in the light chain of EPO. Disulfide bridges connect Cys253 and Cys263, Cys257 and Cys287, Cys359 and Cys370, Cys570 and Cys635, and Cys676 and Cys701, forming five intrachain disulfide bridges in the heavy chain of EPO. Cys291 and Cys455 are found to be unpaired, containing free sulfhydryl groups. The pattern of disulfide bridges is in agreement with that predicted from the X-ray crystallographic structure of canine myeloperoxidase (MPO) (Zeng, J., and Fenna, R. E. (1992) J. Mol. Biol. 226, 185-207) to be general for the class of mammalian peroxidases, including EPO, MPO, lactoperoxidase (LPO), and thyroid peroxidase (TPO). Of four candidate sites in EPO for attachment of glucosamine-based carbohydrate, Asn327 and Asn363 are occupied, whereas Asn700 and Asn708 are unsubstituted. Furthermore, a discrepancy in the literature regarding the sequence of residues 645-659 is resolved.