Buried S-nitrosocysteine revealed in crystal structures of human thioredoxin

We have determined the 1.65 A crystal structure of human thioredoxin-1 after treatment with S-nitrosoglutathione, providing a high-resolution view of this important protein modification and mechanistic insight into protein transnitrosation. Thioredoxin-1 appears to play an intermediary role in cellular S-nitrosylation and is important in numerous biological and pathobiological activities. S-Nitroso modifications of cysteines 62 and 69 are clearly visible in the structure and display planar cis geometries, whereas cysteines 32, 35, and 73 form intra- and intermolecular disulfide bonds. Surprisingly, the Cys 62 nitroso group is completely buried and pointing to the protein interior yet is the most readily formed at neutral pH. The Cys 69 nitroso group is also protected but requires a higher pH for stable formation. The helix intervening between residues 62 and 69 shifts by approximately 0.5 A to accommodate the SNO groups. The crystallographic asymmetric unit contains three independent molecules of thioredoxin, providing three views of the nitrosated protein. The three molecules are in general agreement but display subtle differences, including both cis and trans conformers for Cys 69 SNO in molecule C, and greater disorder in the Cys 62-Cys 69 helix in molecule B. Possible mechanisms for protein transnitrosation with specific geometric requirements and charge stabilization of the nitroxyl disulfide reaction intermediate are discussed.     

Cysteine‐free rop: A four‐helix bundle core mutant has wild‐type stability and structure but dramatically different unfolding kinetics

Cysteine residues can complicate the folding and storage of proteins due to improper formation of disulfide bonds or oxidation of residues that are natively reduced. Wild‐type Rop is a homodimeric four‐helix bundle protein and an important model for protein design in the understanding of protein stability, structure and folding kinetics. In the native state, Rop has two buried, reduced cysteine residues in its core, but these are prone to oxidation in destabilized variants, particularly upon extended storage. To circumvent this problem, we designed and characterized a Cys‐free variant of Rop, including solving the 2.3 Å X‐ray crystal structure. We show that the C38A C52V variant has similar structure, stability and in vivo activity to wild‐type Rop, but that it has dramatically faster unfolding kinetics like virtually every other mutant of Rop that has been characterized. This cysteine‐free Rop has already proven useful for studies on solution topology and on the relationship of core mutations to stability. It also suggests a general strategy for removal of pairs of Cys residues in proteins, both to make them more experimentally tractable and to improve their storage properties for therapeutic or industrial purposes.     

Structure of human thioredoxin exhibits a large conformational change

Thioredoxin is an oxidoreductase, which is ubiquitously present across phyla from humans to plants and bacteria. Thioredoxin reduces a variety of substrates through active site Cys 32, which is subsequently oxidized to form the intramolecular disulphide with Cys 35. The thioredoxin fold is known to be highly stable and conformational changes in the active site loops and residues Cys 32, Cys 35 have been characterized between ligand bound and free structures. We have determined a novel 2.0 Å resolution crystal structure for a human thioredoxin, which reveals a much larger conformational change than previously characterized. The principal change involves unraveling of a helix to form an extended loop that is linked to secondary changes in further loop regions and the wider area of the active site Cys 32. This gives rise to a more open conformation and an elongated hydrophobic pocket results in place of the helix. Buried residue Cys 62 from this helix becomes exposed in the open conformation. This provides a structural basis for observations that the Cys 62 sidechain can form mixed disulphides and be modified by thiol reactive small molecules.     

Cleavage and methylation of DNA by the restriction endonuclease HinfIII isolated from Haemophilus influenzae Rf

Oxidized uteroglobin, in the C222₁ crystal form, has been analyzed by X-ray diffraction at a resolution of 2.2 Å.Uteroglobin is a dimer, possessing in this crystal form a true binary axis symmetry. It is built from two identical polypeptidic chains of 70 residues each, held together by antiparallel (Cys3—Cys69′, Cys3′—Cys69) disulfide bridges. The observed structure is in agreement with one of the amino acid sequence determinations previously described. It is a monodomain globular protein which contains a high proportion (~70%) of α helix and no β sheets. An oblong hydrophobic pocket is located in a central position around the binary axis. This cavity has the features expected from a progesterone-binding site. The possible mechanisms of hormone binding to uteroglobin are discussed.     

Assignment of the disulfide bonds in the sweet protein brazzein

The thermostable sweet protein brazzein consists of 54 amino acid residues and has four intramolecular disulfide bonds, the location of which is unknown. We found that brazzein resists enzymatic hydrolysis at enzyme/substrate ratios (w/w) of 1:100-1:10 at 35–40°C for 24–48 h. Brazzein was hydrolyzed using thermolysin at an enzyme/substrate ratio of 1:1 (w/w) in water, pH 5.5. for 6 h and at 50°C. The disulfide bonds were determined, by a combination of mass spectrometric analysis and amino acid sequencing of cystine-containing peptides, to be between Cys4-Cys52, Cys16-Cys37, Cys22-Cys47, and Cys26-Cys49. These disulfide bonds contribute to its thermostability. © 1996 John Wiley & Sons, Inc.     

Evidence for difference in the roles of two cysteine residues involved in disulfide bond formation in the folding of human lysozyme

Human lysozyme is made up of 130 amino acid residues and has four disulfide bonds at Cys6-Cys128, Cys30-Cys116, Cys65-Cys81, and Cys77-Cys95. Our previous results using the Saccharomyces cerevisiae secretion system indicate that the individual disulfide bonds of human lysozyme have different functions in the correct in vivo folding and enzymatic activity of the protein (Taniyama, Y., Yamamoto, Y., Nakao, M., Kikuchi, M., and Ikehara, M. (1988) Biochem. Biophys. Res. Commun. 152, 962-967). In this paper, we report the results of experiments that were focused on the roles of Cys65 and Cys81 in the folding of human lysozyme protein in yeast. A mutant protein (C81A), in which Cys81 was replaced with Ala, had almost the same enzymatic activity and conformation as those of the native enzyme. On the other hand, another mutant (C65A), in which Cys65 was replaced with Ala, was not found to fold correctly. These results indicate that Cys81 is not a requisite for both correct folding and activity, whereas Cys65 is indispensable. The mutant protein C81A is seen to contain a new, non-native disulfide bond at Cys65-Cys77. The possible occurrence of disulfide bond interchange during our mapping experiments cannot be ruled out by the experimental techniques presently available, but characterization of other mutant proteins and computer analysis suggest that the intramolecular exchange of disulfide bonds is present in the folding pathway of human lysozyme in vivo.     

Role of simian virus 40 Vp1 cysteines in virion infectivity

We have developed a new nonoverlapping infectious viral genome (NO-SV40) in order to facilitate structure-based analysis of the simian virus 40 (SV40) life cycle. We first tested the role of cysteine residues in the formation of infectious virions by individually mutating the seven cysteines in the major capsid protein, Vp1. All seven cysteine mutants-C9A, C49A, C87A, C104A, C207S, C254A, and C267L-retained viability. In the crystal structure of SV40, disulfide bridges are formed between certain Cys104 residues on neighboring pentamers. However, our results show that none of these disulfide bonds are required for virion infectivity in culture. We also introduced five different mutations into Cys254, the most strictly conserved cysteine across the polyomavirus family. We found that C254L, C254S, C254G, C254Q, and C254R mutants all showed greatly reduced (around 100,000-fold) plaque-forming ability. These mutants had no apparent defect in viral DNA replication. Mutant Vp1's, as well as wild-type Vp2/3, were mostly localized in the nucleus. Further analysis of the C254L mutant revealed that the mutant Vp1 was able to form pentamers in vitro. DNase I-resistant virion-like particles were present in NO-SV40-C254L-transfected cell lysate, but at about 1/18 the amount in wild-type-transfected lysate. An examination of the three-dimensional structure reveals that Cys254 is buried near the surface of Vp1, so that it cannot form disulfide bonds, and is not involved in intrapentamer interactions, consistent with the normal pentamer formation by the C254L mutant. It is, however, located at a critical junction between three pentamers, on a conserved loop (G2H) that packs against the dual interpentamer Ca(2+)-binding sites and the invading C-terminal helix of an adjacent pentamer. The substitution by the larger side chains is predicted to cause a localized shift in the G2H loop, which may disrupt Ca(2+) ion coordination and the packing of the invading helix, consistent with the defect in virion assembly. Our experimental system thus allows dissection of structure-function relationships during the distinct steps of the SV40 life cycle.     

Assignment of all four disulfide Bridges in echistatin

Echistatin is a 49-amino-acid protein fromEchis carinatus venom. It contains four disulfide bonds. Since the disulfide bonding is critical for biological activity, it is very important to assign the disulfide linkage in this protein. Echistatin was incubated in 250 mM oxalic acid at 100°C for 4 hr under nitrogen. Under these conditions, many overlapping disulfide-containing peptides were identified by ionspray mass spectrometry. Ionspray MS/MS data indicate that the four disulfide bonds are Cys 2–Cys 11, Cys 7–Cys 32, Cys 8–Cys 37, and Cys 20–Cys 39. To our knowledge, this is the first time all four disulfide bonds in echistatin have been assigned in one experiment without disulfide bond exchange. This approach, which combines oxalic acid hydrolysis and ionspray MS/MS, may be very useful for assigning disulfide bridges in other proteins from the disintegrin family.     

Role of individual cysteine residues and disulfide bonds in the structure and function of Aspergillus ribonucleolytic toxin restrictocin.

Restrictocin, produced by the fungus Aspergillus restrictus, belongs to the group of ribonucleolytic toxins called ribotoxins. It specifically cleaves a single phosphodiester bond in a conserved stem and loop structure in the 28S rRNA of large ribosomal subunit and potently inhibits eukaryotic protein synthesis. Restrictocin contains 149 amino acid residues and includes four cysteines at positions 5, 75, 131, and 147. These cysteine residues are involved in the formation of two disulfide bonds, one between Cys 5 and Cys 147 and another between Cys 75 and Cys 131. In the current study, all four cysteine residues were changed to alanine individually and in different combinations by site-directed mutagenesis so as to remove one or both the disulfides. The mutants were expressed and purified from Escherichia coli. Removal of any cysteine or any one of the disulfide bonds individually did not affect the ability of the toxin to specifically cleave the 28S rRNA or to inhibit protein synthesis in vitro. However, the toxin without both disulfide bonds completely lost both ribonucleolytic and protein synthesis inhibition activities. The active mutants, containing only one disulfide bond, exhibited relatively high susceptibility to trypsin digestion. Thus, none of the four cysteine residues is directly involved in restrictocin catalysis; however, the presence of any one of the two disulfide bonds is absolutely essential and sufficient to maintain the enzymatically active conformation of restrictocin. For maintenance of the unique stability displayed by the native toxin, both disulfide bonds are required.     

2,2′‐Dipyridyl diselenide: A chemoselective tool for cysteine deprotection and disulfide bond formation

There are many examples of bioactive, disulfide‐rich peptides and proteins whose biological activity relies on proper disulfide connectivity. Regioselective disulfide bond formation is a strategy for the synthesis of these bioactive peptides, but many of these methods suffer from a lack of orthogonality between pairs of protected cysteine (Cys) residues, efficiency, and high yields. Here, we show the utilization of 2,2′‐dipyridyl diselenide (PySeSePy) as a chemical tool for the removal of Cys‐protecting groups and regioselective formation of disulfide bonds in peptides. We found that peptides containing either Cys(Mob) or Cys(Acm) groups treated with PySeSePy in trifluoroacetic acid (TFA) (with or without triisopropylsilane (TIS) were converted to Cys‐S–SePy adducts at 37 °C and various incubation times. This novel Cys‐S–SePy adduct is able to be chemoselectively reduced by five‐fold excess ascorbate at pH 4.5, a condition that should spare already installed peptide disulfide bonds from reduction. This chemoselective reduction by ascorbate will undoubtedly find utility in numerous biotechnological applications. We applied our new chemistry to the iodine‐free synthesis of the human intestinal hormone guanylin, which contains two disulfide bonds. While we originally envisioned using ascorbate to chemoselectively reduce one of the formed Cys‐S–SePy adducts to catalyze disulfide bond formation, we found that when pairs of Cys(Acm) residues were treated with PySeSePy in TFA, the second disulfide bond formed spontaneously. Spontaneous formation of the second disulfide is most likely driven by the formation of the thermodynamically favored diselenide (PySeSePy) from the two Cys‐S–SePy adducts. Thus, we have developed a one‐pot method for concomitant deprotection and disulfide bond formation of Cys(Acm) pairs in the presence of an existing disulfide bond.     

Crystal structure of TGF-β2 refined at 1.8 Å resolution

The crystal structure of TGF-β2 has been refined using data collected with synchrotron radiation (CHESS) to 1.8 Å resolution with a residual R (= ∑ | |F_o| ? |F_c| | /∑ |F_o|) factor of 17.3%. The model consists of 890 protein atoms from all 112 residues and 59 water molecules. The monomer of TGF-β2 assumes a rather extended conformation and lacks a well-defined hydrophobic core. Surface accessibility calculations show only 44% of the nonpolar surface is buried in the monomer. In contrast, 55.8% of the nonpolar surface area is buried when the two monomers from a dimer, a typical value for globular proteins. This includes a 1300 Ų buried interface area that is largely hydrophobic. Sequence comparisons using a profile derived from the refined TGF-β2 structure suggest that the cluster of four disulfides (three intramonomeric disulfide bonds 15–78, 44–109, 48–111 forming a disulfide knot, and one intermonomeric disulfide 77–77) together with the extended β strand region constitutes the conserved structural motif for the TGF-β superfamily. This structural motif, without the 77–77 disulfide bond, defines also the common fold for a general family of growth factors, including the nerve growth factor and platelet-derived growth factor families. The fold is conserved only at the monomer level, while the active forms are dimers, suggesting that dimerization plays an important role in regulating the binding of these cytokines to their receptors and in modulating the biological responses. © 1993 Wiley-Liss, Inc.

1 This article is a US Government work and, as such, is in the public domain in the United States of America.

     

The structure of hepadnaviral core antigens. Identification of free thiols and determination of the disulfide bonding pattern.

A set of wild-type and mutant human, woodchuck, and duck hepatitis viral core proteins have been prepared and used to study the free thiol groups and the disulfide bonding pattern present within the core particle. Human (HBcAg) and woodchuck (WHcAg) core proteins contain 4 cysteine residues, whereas duck (DHcAg) core protein contains a single cysteine residue. Each of the cysteines of HBcAg has been eliminated, either singly or in combinations, by a two-step mutagenesis procedure. All of the proteins were shown to have very similar physical and immunochemical properties. All assemble into essentially identical core particle structures. Therefore disulfide bonds are not essential for core particle formation. No intra-chain disulfide bonds occur. Cys107 is a free thiol buried within the particle structure, whereas Cys48 is present partly as a free sulfhydryl which is exposed at the surface of the particle. Cys61 is always and Cys48 is partly involved in interchain disulfide bonds with the identical residues of another monomer, whereas Cys183 is always involved in a disulfide bond with the Cys183 of a different monomer. WHcAg has the same pattern of bonding, whereas DHcAg lacks any disulfide bonds, and the single free sulfhydryl, Cys153 which is equivalent to Cys107 of HBcAg, is buried.     

Mechanism of the electron transfer catalyst DsbB from Escherichia coli

The membrane protein DsbB from Escherichia coli is essential for disulfide bond formation and catalyses the oxidation of the periplasmic dithiol oxidase DsbA by ubiquinone. DsbB contains two catalytic disulfide bonds, Cys41-Cys44 and Cys104-Cys130. We show that DsbB directly oxidizes one molar equivalent of DsbA in the absence of ubiquinone via disulfide exchange with the 104-130 disulfide bond, with a rate constant of 2.7 x 10 M(-1) x s(-1). This reaction occurs although the 104-130 disulfide is less oxidizing than the catalytic disulfide bond of DsbA (E(o)' = -186 and -122 mV, respectively). This is because the 41-44 disulfide, which is only accessible to ubiquinone but not to DsbA, is the most oxidizing disulfide bond in a protein described so far, with a redox potential of -69 mV. Rapid intramolecular disulfide exchange in partially reduced DsbB converts the enzyme into a state in which Cys41 and Cys44 are reduced and thus accessible for reoxidation by ubiquinone. This demonstrates that the high catalytic efficiency of DsbB results from the extreme intrinsic oxidative force of the enzyme.     

Disulfide transfer between two conserved cysteine pairs imparts selectivity to protein oxidation by Ero1

The membrane-associated flavoprotein Ero1p promotes disulfide bond formation in the endoplasmic reticulum (ER) by selectively oxidizing the soluble oxidoreductase protein disulfide isomerase (Pdi1p), which in turn can directly oxidize secretory proteins. Two redox-active disulfide bonds are essential for Ero1p oxidase activity: Cys100-Cys105 and Cys352-Cys355. Genetic and structural data indicate a disulfide bond is transferred from Cys100-Cys105 directly to Pdi1p, whereas a Cys352-Cys355 disulfide bond is used to reoxidize the reduced Cys100-Cys105 pair through an internal thiol-transfer reaction. Electron transfer from Cys352-Cys355 to molecular oxygen, by way of a flavin cofactor, maintains Cys352-Cys355 in an oxidized form. Herein, we identify a mixed disulfide species that confirms the Ero1p intercysteine thiol-transfer relay in vivo and identify Cys105 and Cys352 as the cysteines that mediate thiol-disulfide exchange. Moreover, we describe Ero1p mutants that have the surprising ability to oxidize substrates in the absence of Cys100-Cys105. We show the oxidase activity of these mutants results from structural changes in Ero1p that allow substrates increased access to Cys352-Cys355, which are normally buried beneath the protein surface. The altered activity of these Ero1p mutants toward selected substrates leads us to propose the catalytic mechanism involving transfer between cysteine pairs evolved to impart substrate specificity to Ero1p.     

Crystal structure of the DsbB-DsbA complex reveals a mechanism of disulfide bond generation

Oxidation of cysteine pairs to disulfide requires cellular factors present in the bacterial periplasmic space. DsbB is an E. coli membrane protein that oxidizes DsbA, a periplasmic dithiol oxidase. To gain insight into disulfide bond formation, we determined the crystal structure of the DsbB-DsbA complex at 3.7 A resolution. The structure of DsbB revealed four transmembrane helices and one short horizontal helix juxtaposed with Cys130 in the mobile periplasmic loop. Whereas DsbB in the resting state contains a Cys104-Cys130 disulfide, Cys104 in the binary complex is engaged in the intermolecular disulfide bond and captured by the hydrophobic groove of DsbA, resulting in separation from Cys130. This cysteine relocation prevents the backward resolution of the complex and allows Cys130 to approach and activate the disulfide-generating reaction center composed of Cys41, Cys44, Arg48, and ubiquinone. We propose that DsbB is converted by its specific substrate, DsbA, to a superoxidizing enzyme, capable of oxidizing this extremely oxidizing oxidase.     

Role of disulfide bonds in goose-type lysozyme

The role of the two disulfide bonds (Cys4-Cys60 and Cys18-Cys29) in the activity and stability of goose-type (G-type) lysozyme was investigated using ostrich egg-white lysozyme as a model. Each of the two disulfide bonds was deleted separately or simultaneously by substituting both Cys residues with either Ser or Ala. No remarkable differences in secondary structure or catalytic activity were observed between the wild-type and mutant proteins. However, thermal and guanidine hydrochloride unfolding experiments revealed that the stabilities of mutants lacking one or both of the disulfide bonds were significantly decreased relative to those of the wild-type. The destabilization energies of mutant proteins agreed well with those predicted from entropic effects in the denatured state. The effects of deleting each disulfide bond on protein stability were found to be approximately additive, indicating that the individual disulfide bonds contribute to the stability of G-type lysozyme in an independent manner. Under reducing conditions, the thermal stability of the wild-type was decreased to a level nearly equivalent to that of a Cys-free mutant (C4S/C18S/C29S/C60S) in which all Cys residues were replaced by Ser. Moreover, the optimum temperature of the catalytic activity for the Cys-free mutant was downshifted by about 20 degrees C as compared with that of the wild-type. These results indicate that the formation of the two disulfide bonds is not essential for the correct folding into the catalytically active conformation, but is crucial for the structural stability of G-type lysozyme.     

Dynamic nature of disulphide bond formation catalysts revealed by crystal structures of DsbB

In the Escherichia coli system catalysing oxidative protein folding, disulphide bonds are generated by the cooperation of DsbB and ubiquinone and transferred to substrate proteins through DsbA. The structures solved so far for different forms of DsbB lack the Cys104–Cys130 initial‐state disulphide that is directly donated to DsbA. Here, we report the 3.4 Å crystal structure of a DsbB–Fab complex, in which DsbB has this principal disulphide. Its comparison with the updated structure of the DsbB–DsbA complex as well as with the recently reported NMR structure of a DsbB variant having the rearranged Cys41–Cys130 disulphide illuminated conformational transitions of DsbB induced by the binding and release of DsbA. Mutational studies revealed that the membrane‐parallel short α‐helix of DsbB has a key function in physiological electron flow, presumably by controlling the positioning of the Cys130‐containing loop. These findings demonstrate that DsbB has developed the elaborate conformational dynamism to oxidize DsbA for continuous protein disulphide bond formation in the cell.     

Crystal structure of a dimeric oxidized form of human peroxiredoxin 5

Peroxiredoxin 5 is the last discovered mammalian member of an ubiquitous family of peroxidases widely distributed among prokaryotes and eukaryotes. Mammalian peroxiredoxin 5 has been recently classified as an atypical 2-Cys peroxiredoxin due to the presence of a conserved peroxidatic N-terminal cysteine (Cys47) and an unconserved resolving C-terminal cysteine residue (Cys151) forming an intramolecular disulfide intermediate in the oxidized enzyme. We have recently reported the crystal structure of human peroxiredoxin 5 in its reduced form. Here, a new crystal form of human peroxiredoxin 5 is described at 2.0 A resolution. The asymmetric unit contains three polypeptide chains. Surprisingly, beside two reduced chains, the third one is oxidized although the enzyme was crystallized under initial reducing conditions in the presence of 1 mM 1,4-dithio-dl-threitol. The oxidized polypeptide chain forms an homodimer with a symmetry-related one through intermolecular disulfide bonds between Cys47 and Cys151. The formation of these disulfide bonds is accompanied by the partial unwinding of the N-terminal parts of the alpha2 helix, which, in the reduced form, contains the peroxidatic Cys47 and the alpha6 helix, which is sequentially close to the resolving residue Cys151. In each monomer of the oxidized chain, the C-terminal part including the alpha6 helix is completely reorganized and is isolated from the rest of the protein on an extended arm. In the oxidized dimer, the arm belonging to the first monomer now appears at the surface of the second subunit and vice versa.     

Unfolding of an α-helix in peptide crystals by solvation: Conformational fragility in a heptapeptide

The structure of the peptide Boc-Val-Ala-Leu-Aib-Val-Ala-Leu-OMe has been determined in crystals obtained from a dimethylsulfoxide–isopropanol mixture. Crystal parameters are as follows: C₃₈H₆₉N₇O₁₀ · H₂O · 2C₃H₇OH, space group P2₁, a = 10.350 (2) Å, b = 26.084 (4) Å, c = 10.395(2) Å, β = 96.87(12), Z = 2, R = 8.7% for 2686 reflections observed > 3.0 σ (F). A single 5 → 1 hydrogen bond is observed at the N-terminus, while two 4 → 1 hydrogen bonds characteristic of a 3₁₀-helix are seen in the central segment. The C-terminus residues, Ala(6) and Leu(7) are expended, while Val(5) is considerably distorted from a helical conformation. Two isopropanol molecules make hydrogen bonds to the C-terminal segment, while a water molecule interacts with the N-terminus. The structure is in contrast to that obtained for the same peptide in crystals from methanol-water [ I. L. Karle, J. L. Flippen-Anderson, K. Uma, and P. Balaram (1990) Proteins: Structure, Function and Genetics, Vol. 7, pp. 62–73] in which two independent molecules reveal an almost perfect α-helix and a helix penetrated by a water molecule. A comparison of the three structures provides a snapshot of the progressive effects of solvation leading to helix unwinding. The fragility of the heptapeptide helix in solution is demonstrated by nmr studies in CDC1₃ and (CD₃)₂SO. A helical conformation is supported in the apolar solvent CDCl₃, whereas almost complete unfolding is observed in the strongly solvating medium (CD₃)₂SO. © 1993 John Wiley & Sons, Inc.     

DsbA and DsbC-catalyzed oxidative folding of proteins with complex disulfide bridge patterns in vitro and in vivo

Oxidative protein folding in the periplasm of Escherichia coli is catalyzed by the thiol-disulfide oxidoreductases DsbA and DsbC. We investigated the catalytic efficiency of these enzymes during folding of proteins with a very complex disulfide pattern in vivo and in vitro, using the Ragi bifunctional inhibitor (RBI) as model substrate. RBI is a 13.1 kDa protein with five overlapping disulfide bonds. We show that reduced RBI can be refolded quantitatively in glutathione redox buffers in vitro and spontaneously adopts the single correct conformation out of 750 possible species with five disulfide bonds. Under oxidizing redox conditions, however, RBI folding is hampered by accumulation of a large number of intermediates with non-native disulfide bonds, while a surprisingly low number of intermediates accumulates under optimal or reducing redox conditions. DsbC catalyzes folding of RBI under all redox conditions in vitro, but is particularly efficient in rearranging buried, non-native disulfide bonds formed under oxidizing conditions. In contrast, the influence of DsbA on the refolding reaction is essentially restricted to reducing redox conditions where disulfide formation is rate limiting. The effects of DsbA and DsbC on folding of RBI in E.coli are very similar to those observed in vitro. Whereas overexpression of DsbA has no effect on the amount of correctly folded RBI, co-expression of DsbC enhanced the efficiency of RBI folding in the periplasm of E.coli about 14-fold. Addition of reduced glutathione to the growth medium together with DsbC overexpression further increased the folding yield of RBI in vivo to 26-fold. This shows that DsbC is the bacterial enzyme of choice for improving the periplasmic folding yields of proteins with very complex disulfide bond patterns.