Contributions of substrate binding to the catalytic activity of DsbC.

DsbA and DsbC are involved in protein disulfide bond formation in the periplasm of Gram-negative bacteria. The two proteins are thought to fulfill different functions in vivo, DsbA as a catalyst of disulfide bond formation and DsbC as a catalyst of disulfide bond rearrangement. To explore the basis of this catalytic complementarity, the reaction mechanism of DsbC has been examined using unstructured model peptides that contain only one or two cysteine residues as substrates. The reactions between the various forms of the peptide and DsbC occur at rates up to 10(6)-fold faster than those that involve glutathione and DsbC, and they were constrained to occur at only one sulfur atom of disulfide bonds involving the peptide. Mixed disulfide complexes of DsbC and the peptide were 10(4)-fold more stable than the corresponding mixed disulfides with glutathione. These observations suggest that noncovalent binding interactions occur between the peptide and DsbC, which contribute to the very rapid kinetics of substrate utilization. The interactions between DsbC and the peptide appear to be more substantial than those between DsbA and the same peptide. The differences in the reaction of the peptide at the active sites of DsbA and DsbC provide insight into why DsbC is the better catalyst of disulfide bond rearrangement and how the active site chemistry of these structurally related proteins has been adapted to fulfill complementary functions.

Protein folding coupled to disulphide bond formation.

Protein folding that is coupled to disulphide bond formation has many experimental advantages. In particular, the kinetic roles and importance of all the disulphide intermediates can be determined, usually unambiguously. This contrasts with other types of protein folding, where the roles of any intermediates detected are usually not established. Nevertheless, there is considerable confusion in the literature about even the best-characterized disulphide folding pathways. This article attempts to set the record straight.

Protein folding: does diffusion determine the folding rate?

A major question about protein folding is whether the coming together by diffusion of different segments of the polypeptide chain is rate-determining. This seemingly simple question has been very difficult to answer experimentally, but a positive result has now been obtained with one small model protein.

The folding catalyst protein disulfide isomerase is constructed of active and inactive thioredoxin modules.

BACKGROUND: Protein disulfide isomerase (PDI), a multifunctional protein of the endoplasmic reticulum, catalyzes the formation, breakage and rearrangement of disulfide bonds during protein folding. Dissection of this protein into its individual domains has confirmed the presence of the a and a' domains, which are homologous to thioredoxin, having related structures and activities. The a and a' domains both contain a -Cys-Gly-His-Cys- active-site sequence motif. The remainder of the molecule consists primarily of two further domains, designated b and b' which are thought to be sequence repeats on the basis of a limited sequence similarity. The functions of the b and b' domains are unknown and, until now, the structure of neither domain was known. RESULTS: Heteronuclear nuclear magnetic resonance (NMR) methods have been used to determine the global fold of the PDI b domain. The protein has an alpha/beta fold with the order of the elements of secondary structure being beta1-alpha1-beta2-alpha2-beta3-alpha3-beta4-beta5+ ++-alpha4. The strands are all in a parallel arrangement with respect to each other, except for beta4 which is antiparallel. The arrangement of the secondary structure elements of the b domain is identical to that found in the a domain of PDI and in the ubiquitous redox protein thioredoxin; the three-dimensional folding topology of the b domain is also very similar to that of these proteins. CONCLUSIONS: Our determination of the global fold of the b domain of PDI by NMR reveals that, like the a domain, the b domain contains the thioredoxin motif, even though the b domain has no significant amino-acid sequence similarities to any members of the thioredoxin family. This observation, together with indications that the b' domain adopts a similar fold, suggests that PDI consists of active and inactive thioredoxin modules. These modules may have been adapted during evolution to provide PDI with its complete spectrum of enzymatic activities.

Probing protein folding and stability using disulfide bonds.

Disulfide bonds are required to stabilize the folded conformations of many proteins. The rates and equilibria of processes involved in disulfide bond formation and breakage can be manipulated experimentally and can be used to obtain important information about protein folding and stability. A number of experimental procedures for studying these processes, and approaches to interpreting the resulting data, are described here.

How important is the molten globule for correct protein folding?

The molten globule (MG) state is widely considered to be an important intermediate in protein folding and to have a polypeptide backbone with a native-like topology. The experimental evidence for this view was obtained largely, however, with MG proteins containing native-like constraints. When the four disulphide bonds of alpha-lactalbumin were allowed to rearrange to those favoured by the MG, opposite conclusions were obtained. Consideration of all the experimental data indicates that any tendency of this MG to be native-like is negligible relative to all the other topologies that it can adopt. Furthermore, the experimental data indicate that the MG is not the key to rapid protein folding.

Crystallization of DsbC, the disulfide bond isomerase of Escherichia coli.

DsbC is a 2 x 23 kDa soluble dimeric protein molecule involved in protein disulfide bond formation in the E. coli periplasm, primarily catalyzing disulfide bond rearrangements. Crystals of both the native and selenomethione protein suitable for structure determination were obtained using the hanging-drop vapour-diffusion method. The best crystals were obtained using 18-22% (v/v) polyethylene glycol 550 monomethyl ether in 100 mM Tris-HCl (pH 8.9). Seeding methods were used to produce large crystals diffracting to 2 A resolution, and the detergent n-octyl-beta-glucoside was used to improve crystal quality. Significant variation in cell dimensions and crystal order was observed. Cell dimensions obtained for frozen crystals were in the range a = 58.8 (0.3), b = 78.9 (0.5), c = 95.2 (5.0) A. The lattice is orthorhombic and systematic absences indicate that the space group is P2(1)2(1)2(1).

Electrostatic interactions in the active site of the N-terminal thioredoxin-like domain of protein disulfide isomerase.

Proteins with the thioredoxin fold have widely differing stabilities of the disulfide bond that can be formed between the two cysteines at their active site sequence motif Cys1-Xaa2-Yaa3-Cys4. This is believed to be regulated not by varying the disulfide bond itself, but by modulating the stability of the dithiol form of the protein through interactions with the ionized form of the Cys1 thiol group. A consistent relationship between disulfide bond stability and Cys1 thiol pKa value is found here for DsbA, thioredoxin, and the N-terminal thioredoxin-like domain of protein disulfide isomerase (PDI a), which has a very low thiol pKa value of 4.5. This thiolate anion is stabilized by 5.7 kcal/mol in the dithiol form, giving rise to the corresponding instability of the disulfide bond and the oxidizing properties of PDI a. Electrostatic interactions in the active site of the PDI a-domain have been characterized in order to understand the physical basis of this stabilization. Linkage with the ionization of the imidazole group of His3 in the active site demonstrates that this charge-charge interaction contributes 1.1 kcal/mol. The remainder of the stabilization is believed to be due primarily to interactions with the partial positive charges at the N-terminus of an alpha-helix, which are exceedingly sensitive to charges of surrounding residues.

Identifying and characterizing a structural domain of protein disulfide isomerase.

Protein disulfide isomerase (PDI) appears on the basis of its primary structure to be a multidomain protein, but the number and nature of the domains has been uncertain. Two of the domains, a and a', which are homologous to thioredoxin and active in catalysis of disulfide bond formation, have been identified and characterized previously. Sections of the N-terminal half of the PDI sequence have been expressed and the limits of their folded structures delineated by limited proteolysis. In addition to the a-domain, the boundaries of a domain with no activity on thiol/disulfide groups, designated b, have been identified. This domain has been produced independently; its cooperative unfolding transition and its CD and NMR spectra confirm that it is an autonomously folded structure in isolation and when part of PDI. Fusion of the b-domain to the a-domain, as occurs naturally in the first half of PDI, did not alter substantially the catalytic activity of the a-domain. It still catalyzes only a subset of the thiol/disulfide exchange reactions of intact PDI and has a reduced ability to catalyze protein disulfide rearrangements. The a- and b-domains account structurally for virtually all of the first half of the PDI polypeptide chain, and it is very unlikely that there exists a proposed third domain homologous to the estrogen receptor. The b-domain exhibits some sequence homology to calsequestrin, a calcium binding protein from the sarcoplasmic reticulum of muscle.

Structure determination of the N-terminal thioredoxin-like domain of protein disulfide isomerase using multidimensional heteronuclear 13C/15N NMR spectroscopy.

As a first step in dissecting the structure of human protein disulfide isomerase (PDI), the structure of a fragment corresponding to the first 120 residues of its sequence has been determined using heteronuclear multidimensional NMR techniques. As expected from its primary structure homology, the fragment has the thioredoxin fold. Similarities and differences in their structures help to explain why thioredoxins are reductants, whereas PDI is an oxidant of protein thiol groups. The results confirm that PDI has a modular, multidomain structure, which will facilitate its structural and functional characterization.

On the reactivity and ionization of the active site cysteine residues of Escherichia coli thioredoxin.

Within various proteins of the thioredoxin family, the stability of the disulfide bond formed reversibly between the two active site cysteine residues, one accessible and one buried, varies widely and is directly correlated with the pKa value of the accessible cysteine thiol group. If applicable to thioredoxin, its stable disulfide bond would imply that its accessible thiol group should have a high pKa value, whereas it has long been considered to be about 6.7, largely on the basis of the pH dependence of its reactivity. Such kinetic data are shown to be inconsistent with known pKa values in this case; the rate constants may reflect effects in the transition state for the reaction, which is catalyzed by thioredoxin, rather than the protein itself. Ionization of the thioredoxin thiol groups was measured indirectly by the pH dependence of the equilibrium constant for their reaction with glutathione and directly by detection of the thiolate anion by its UV absorbance. Both observations indicated that both cysteine thiol groups of thioredoxin ionize with apparent pKa values in the region of 9-10 and that their ionization is not linked strongly to that of any other groups. This conclusion is not incompatible with the other data available and would make thioredoxin consistent with the relationship between thiol group ionization and disulfide stability observed in other members of the thioredoxin family.

Comparison of the (30-51, 14-38) two-disulphide folding intermediates of the homologous proteins dendrotoxin K and bovine pancreatic trypsin inhibitor by two-dimensional 1H nuclear magnetic resonance.

The disulphide folding pathway of bovine pancreatic trypsin inhibitor (BPTI) revealed that the native conformation is still stable in each intermediate state with two native disulphide linkages, in the absence of each of the corresponding third disulphide bonds. This is thought to be a consequence of the extreme stability of the native BPTI conformation. The current study addresses the question of whether the native-like conformation would be populated significantly at the two-disulphide stage in disulphide refolding if the final structure is less stable than in the case of BPTI. Dendrotoxin K from black mamba venom provides a good model to test this, since it contains the BPTI fold and was shown to fold predominantly via the same pathway, but its native conformation is stable than that of BPTI. The conformation of a chemically trapped two-disulphide intermediate in the disulphide refolding of dendrotoxin K, with blocking groups on Cys5 and Cys55 and disulphide bonds between Cys30 and Cys51, and Cys14 and Cys38, respectively, has been determined by 1H NMR spectroscopy and compared to those of the native protein and of the corresponding intermediate in BPTI. The analysis reveals that the dendrotoxin K intermediate adopts a partly-folded conformation, in contrast to the quasi-native conformation of the corresponding BPTI intermediate. It is similar to the partly-folded conformation of the BPTI intermediate with just the Cys30-Cys-51 disulphide bond, but with a more fixed conformation in the region of the Cys14-Cys38 disulphide bond. The destabilisation of the fully native conformation of the dendrotoxin K intermediate, relative to BPTI, appears to reduce the cooperativeity of the folding process.

The roles of partly folded intermediates in protein folding.

Proteins can fold very rapidly, undoubtedly because they do not do so simply by random searching. The stable, partly folded species that can be detected during protein refolding are, however, of uncertain kinetic significance. The available kinetic evidence indicates that the intermediates that are most responsible for the rapidity of folding are extremely unstable and not populated detectably; they are less extreme versions of the transition state for folding. Protein folding is most readily studied when it is coupled to disulfide formation, which has the advantages that the intermediates can be characterized in detail and their kinetic roles determined unambiguously. The most important aspects of the disulfide folding pathway of BPTI are understood to at least a first approximation, and several other protein disulfide folding pathways are known in outline. These pathways demonstrate that disulfide folding is not intrinsically different from that not involving disulfide formation. Partly folded conformations can increase the rate of folding somewhat by causing productive disulfide bonds to be populated preferentially, but the most important folding intermediates are not detectable. The essence of folding is to build up the cooperativity between the individual interactions that is necessary for a stable conformation.

Nuclear magnetic resonance characterization of the N-terminal thioredoxin-like domain of protein disulfide isomerase.

A genetically engineered protein consisting of the 120 residues at the N-terminus of human protein disulfide isomerase (PDI) has been characterized by 1H, 13C, and 15N NMR methods. The sequence of this protein is 35% identical to Escherichia coli thioredoxin, and it has been found also to have similar patterns of secondary structure and beta-sheet topology. The results confirm that PDI is a modular, multidomain protein. The last 20 residues of the N-terminal domain of PDI are some of those that are similar to part of the estrogen receptor, yet they appear to be an intrinsic part of the thioredoxin fold. This observation makes it unlikely that any of the segments of PDI with similarities to the estrogen receptor comprise individual domains.

Characterization of the active site cysteine residues of the thioredoxin-like domains of protein disulfide isomerase.

The dithiol/disulfide active sites of each of the two isolated thioredoxin-like domains of protein disulfide isomerase (PDI) expressed in Escherichia coli have been characterized in order to understand their catalytic mechanisms and their functions in PDI. In each of the folded domains, as in other proteins of the thioredoxin family, only one of the cysteine residues of the active site sequence -Cys-Gly-His-Cys- is accessible, and its thiol group is highly reactive and has a low pKa value. The kinetics and equilibria have been measured of the reactions between the active site cysteine residues and glutathione, the predominant thiol/disulfide reagent of the endoplasmic reticulum. A disulfide bond can be formed very rapidly between the pair of cysteine residues of each domain, but each disulfide bond is very unstable and reacts rapidly with reduced glutathione. The very low stabilities of these disulfide bonds, which destabilize the protein structures, account for the efficiency with which PDI and each of the isolated domains can introduce disulfide bonds into proteins. These kinetics and equilibrium data go far in helping to understand the catalytic mechanism of PDI and its individual domains.

Ionisation of cysteine residues at the termini of model alpha-helical peptides. Relevance to unusual thiol pKa values in proteins of the thioredoxin family.

The physical basis of the unusually low pKa values of an active site cysteine thiol group in proteins with the thioredoxin fold is unknown. The electrostatic field associated with an alpha-helix pointing with its N terminus towards the cysteine residue has been implicated to lower the thiol pKa value by up to 5 pH units in glutaredoxin and DsbA. Here, the influence of the presence of an alpha-helical conformation on the ionisation of a cysteine thiol group located at or near the helix terminus is investigated in highly helical synthetic peptides with the generic sequence Ac-AAAAAAAAARAAAARAAAARAA-(NH2). The thiol pKa values have been determined by monitoring the pH dependence of the absorbance at 240 nm, of the alpha-helix content measured by the mean residue ellipticity at 222 nm, and of the chemical shifts of protons close to the sulphur atom of the cysteine residue. The favourable interaction between the thiolate anion at the N terminus and the alpha-helix decreases the thiol pKa value by up to 1.6 pH units when compared to a normal thiol pKa value measured in an unfolded control peptide, corresponding to a stabilisation energy of 2.1 kcal/mol. At the C terminus, the thiol pKa value is increased, but by only 0.2 pH units. The observations are consistent with an interaction of the alpha-helix dipole with the cysteine thiolate anion, involving both its charge and hydrogen-bonding. Subtle conformational effects in different model peptides appear to influence the ionisation of the thiol group significantly, with an N terminal Cys-Pro sequence having the most favourable interaction with the alpha-helix.

Effects of trifluoroethanol on the conformations of peptides representing the entire sequence of bovine pancreatic trypsin inhibitor.

The effects of the cosolvent trifluoroethanol on the conformations of four peptides representing the entire sequence of bovine pancreatic trypsin inhibitor (BPTI) have been measured by CD and NMR. No substantial amounts of helical conformations were induced in one peptide with four proline residues dispersed throughout its sequence, and there were no substantial effects on its average conformational properties or on the interactions between neighboring residues that are normally evident. The other three peptides became helical, although not completely, over their entire lengths. There was a reasonable correlation between the induced content of alpha-helix and the predicted helical propensities of all four peptides. Only one of these peptides is helical in native BPTI; the other two are extended beta-strands. The latter two have an intrinsic propensity for helix formation, but a greater propensity for beta-sheet formation in folded proteins.

Functional properties of the individual thioredoxin-like domains of protein disulfide isomerase.

The two thioredoxin-like domains of human protein disulfide isomerase (PDI) have been produced in bacteria as individual soluble, folded protein molecules, and their functional properties have been compared to those of intact PDI. The two individual domains were very similar in their functional properties, and there were no indications of synergy between them, so it is unlikely that they have intrinsically different functions in PDI. Both domains efficiently introduced disulfide bonds into unfolded model proteins and peptides but were less efficient than PDI with folded substrate protein molecules. Relative to PDI, neither domain had substantial activity in catalyzing disulfide bond isomerization. This pattern of activities is very similar to that of the bacterial catalyst DsbA and probably reflects similarities in the catalytic mechanisms of these proteins. The differences in activity between PDI and its thioredoxin-like domains suggest that other features of the PDI molecule are also required for its complete range of thiol-disulfide exchange activities.

Refolding of bovine pancreatic trypsin inhibitor via non-native disulphide intermediates.

The disulphide folding pathway of bovine pancreatic trypsin inhibitor (BPTI), especially at the two-disulphide stage, has been dissected by replacing one or two particular cysteine residues by serine. This restricts which disulphide species are possible, and the observed kinetics of disulphide-coupled folding reveal the roles of the remaining species. The results obtained confirm the kinetic roles in the original BPTI pathway of the two specific two-disulphide intermediates with non-native second disulphide bonds, (30-51, 5-14) and (30-51, 5-38). Moreover, the rates of folding through each of these intermediates are shown to account quantitatively for the rate of folding of the normal protein; therefore, essentially all the molecules refold through these two particular intermediates. They are amongst the most productive on the folding pathway, and their roles are readily explicable on the basis of their conformations.

Structural and functional characterization of DsbC, a protein involved in disulfide bond formation in Escherichia coli.

DsbC is a soluble protein of the bacterial periplasm that was identified genetically as being involved in protein disulfide formation. The gene sequence was corrected to include an additional proline residue and was then consistent with the molecular weight of the purified protein. Gel filtration and subunit hybridization indicate that DsbC is a stable dimer of identical subunits. Each subunit has a -Cys-Gly-Tyr-Cys- segment that forms an unstable and reactive disulfide bond; only the first cysteine residue is accessible, similar to thioredoxin and DsbA. The other two cysteine residues of DsbC form a buried, structural disulfide bond. The reactivities and stabilities of the active site disulfide bond of DsbC have been characterized and compared to that of DsbA. Both are very unstable and can be transferred rapidly to reduced proteins and peptides, although they differ somewhat in their kinetic reactivities. The two active sites of the DsbC dimer appear to function independently. DsbC is much more active than DsbA in catalyzing protein disulfide rearrangements, and this may be its main function in vivo.