Real-time Monitoring of Intermediates Reveals the Reaction Pathway in the Thiol-Disulfide Exchange between Disulfide Bond Formation Protein A (DsbA) and B (DsbB) on a Membrane-immobilized Quartz Crystal Microbalance (QCM) System*

Background: A device of QCM can be used in the transient kinetics of oxidation of a pair of cysteine residues in DsbA by DsbB. Results: The obtained kinetic parameters indicate that the two pairs of cysteine residues in DsbB are important. Conclusion: The reaction pathway of almost all DsbA oxidation processes would proceed through the stable intermediate. Significance: The transient kinetics of the reaction intermediate is important. Disulfide bond formation protein B (DsbBS-S,S-S) is an inner membrane protein in Escherichia coli that has two disulfide bonds (S-S, S-S) that play a role in oxidization of a pair of cysteine residues (SH, SH) in disulfide bond formation protein A (DsbASH,SH). The oxidized DsbAS-S, with one disulfide bond (S-S), can oxidize proteins with SH groups for maturation of a folding preprotein. Here, we have described the transient kinetics of the oxidation reaction between DsbASH,SH and DsbBS-S,S-S. We immobilized DsbBS-S,S-S embedded in lipid bilayers on the surface of a 27-MHz quartz crystal microbalance (QCM) device to detect both formation and degradation of the reaction intermediate (DsbA-DsbB), formed via intermolecular disulfide bonds, as a mass change in real time. The obtained kinetic parameters (intermediate formation, reverse, and oxidation rate constants (kf, kr, and kcat, respectively) indicated that the two pairs of cysteine residues in DsbBS-S,S-S were more important for the stability of the DsbA-DsbB intermediate than ubiquinone, an electron acceptor for DsbBS-S,S-S. Our data suggested that the reaction pathway of almost all DsbASH,SH oxidation processes would proceed through this stable intermediate, avoiding the requirement for ubiquinone.

Transport reactions that occur inside a cell, such as electron transport reactions, are catalyzed by specific enzymes (1). The mechanism of protein interaction-based transport systems is different from that of electron transport reactions of general chemical compounds. The disulfide bond formation (Dsb) system is a catalytic system that accelerates the oxidative formation of protein disulfide bonds in Escherichia coli and involves a number of protein factors, including the Dsb proteins DsbA, 2 DsbB, DsbC, DsbD, and DsbG (2)(3)(4). Together, DsbA and DsbB function as a disulfide-introducing unit in the bacterial periplasmic space. DsbA belongs to the thioredoxin superfamily and contains a disulfide bond in the thioredoxin domain (Cys-30 -X-X-Cys-33) that is known to act as an acceptor of two electrons (and two protons) from a preprotein containing cysteine residues during the transfer of the disulfide bond from the oxidized DsbA S-S to the preprotein ( Fig. 1) (2)(3)(4)(5). To repeat this disulfide bond-introducing reaction in the next catalytic cycle, re-oxidation of DsbA SH,SH is required. DsbB S-S,S-S is an inner membrane protein in E. coli that has four transmembrane segments containing four cysteine residues that form two disulfide bonds (Cys-41 with Cys-44, and Cys-104 with Cys-130) within the periplasmic space. Reduced DsbA SH,SH is reoxidized by the transfer of two electrons (and two protons) to the two thiol groups of the disulfide bonds on DsbB S-S,S-S , resulting in disulfide exchange. Finally, the two electrons (and two protons) are transferred to lipophilic ubiquinone in the inner membrane, where DsbB functions as a quinone reductase, catalyzing the conversion of ubiquinone to ubiquinol (3).
The mechanism and direction of electron transfer between chemical compounds can be explained in terms of its oxidation-reduction potential. In the case of electron transport based on protein factors, such as the DsbA-DsbB system, proteinprotein interactions and intermediate complexes are also important for understanding the mechanism of the electron transport reaction. Various methods have been used to investigate the electron transport mechanisms associated with thioldisulfide exchange in the DsbA-DsbB system. Nonreducing SDS-PAGE analysis has been used to estimate the amounts of reduced DsbA SH,SH and oxidized DsbA S-S after modification by a thiol-reacting alkylating agent (6 -8). Previous investigations also used x-ray crystal structure analysis to determine the structures of DsbB and the intermediate DsbB-DsbA complex (inactivated with mutation); this analysis revealed the positional relationship of cysteine residues and ubiquinone within the complex (9,10). Snapshots of nuclear magnetic resonance solution structures of DsbB and its variant have proven useful for understanding electron movement from DsbA to ubiquinone (11). Moreover, fluorescence measurements have enabled the observation of the oxidation reaction of DsbA over time on the basis of the change in the tryptophan fluorescence of DsbA (12). Stopped-flow absorbance measurements have also been conducted to monitor DsbB-ubiquinone complexes in the process of electron transfer from DsbA to ubiquinone (13).
Based on these previous studies, two potential models of the reaction pathway have been proposed ( Fig. 1) (3). In the first step the interprotein disulfide bond between Cys-30(DsbA) and Cys-104(DsbB) is formed by the nucleophilic attack of Cys-30 from DsbA SH,SH on the Cys-104 -Cys-130 disulfide bond of DsbB S-S,S-S . Then, Cys-33(DsbA) intramolecularly attacks Cys-30, and oxidized DsbA S-S is released from the reduced DsbB S-S,SH,SH (pathway I, referred to as the rapid pathway) (6,7). In this pathway, ubiquinone reduction occurs in the reduced DsbB SH,SH,S-S -ubiquinone complex, independent of DsbA. In the alternative model, after the formation of the Cys-30(DsbA)-Cys-104(DsbB) complex, the newly reduced Cys-130(DsbB) immediately attacks Cys-41(DsbB) in an intramolecular isomerization reaction to form the interloop disulfide bond Cys-41-Cys-130 in DsbB (the stable intermediate). The consequent thiolate of Cys-44(DsbB) interacts with ubiquinone in DsbB to form the electron transfer complex (7,13,14). Then, DsbA in the complex is oxidized by the intramolecular attack of Cys-33 on Cys-30 of the Cys-30 -Cys-104 disulfide bond and is released (pathway II, referred to as the slow pathway) (8,13). These reaction models are still currently under discussion (3). Although the structure of the DsbB-DsbA complex is known, the reaction pathway has not yet been elucidated because it involves an intermediate structure that is at the branching point of the two proposed reaction pathways. Thus, identification of the reaction pathway requires quantitative kinetic studies of the behavior of intermediates as key molecules.
Quartz crystal microbalance (QCM) devices can measure nanogram-scale mass changes based on the altered resonance frequency (⌬F) of a quartz-plate oscillator, which indicates mass changes on the plate (15)(16)(17). Oscillation of QCMs in aqueous solutions has been applied to nonlabeling biosensors that can detect biomolecular interactions as mass changes on the sensor surface in real time (18). We previously used this technique to detect various biomolecular interactions, such as DNA-DNA, DNA-peptide, and peptide-peptide interactions using a ligand-immobilized quartz plate (19 -21). Recently, enzyme reactions, such as DNA polymerase reactions (22), DNA cleavage reactions (23), and protease reactions (24,25), were carried out on substrate-immobilized QCM plates, allowing the formation and decomposition of the enzyme-substrate complex (intermediate) to be monitored as a detectable mass   change. In addition, transient kinetic analysis could be conducted on the basis of the behavior of the intermediate (24,25). In this paper we have described the immobilization of a membrane protein, DsbB, embedded in supported lipid bilayers on a QCM plate and the detection of an intermediate complex between DsbA SH,SH and DsbB S-S,S-S according to mass changes on the QCM device (Fig. 2). A QCM with a sensor surface consisting of a planar gold electrode was used for immobilization of DsbB membrane proteins in supported lipid bilayers to obtain kinetic parameters under nearly native conditions rather than in surfactant-solubilized DsbB in bulk solutions. Three kinetic parameters were measured, the formation and reverse rate constants (k f and k r , respectively) of the intermolecular disulfide bond in the DsbA-DsbB intermediate and the release rate constant (k cat ) of oxidized DsbA S-S from reduced DsbB S-S,SH,SH in the membrane. These kinetic parameters were compared with those of the mutated variants DsbA(C33A) SH , DsbA(C30A) SH , and DsbB(C41A,C44A) S-S in the presence or absence of ubiquinone. Based on the stability of the DsbA-DsbB intermediate estimated by the obtained kinetic parameters, we concluded that DsbB S-S,S-S mainly oxidized DsbA SH,SH in a ubiquinoneindependent manner through a stable intermediate in which delocalization of the transferred electron on cysteine residues of DsbB was important (modified pathway II).
DsbA SH,SH , DsbA(C33A) SH , and DsbA(C30A) SH proteins were expressed and purified according to a previously reported method, with minor modifications (26). E. coli JM109 (DE3) harboring each expression vector was grown in 1 liter of LB broth with ampicillin at 26°C, and then 1 mM isopropyl-␤-Dthio-galactopyranoside was added into the medium for protein induction at an A 600 of 1.0 followed by additional incubation for 16 h. The cells were harvested and then suspended in 20 ml of a buffer (200 mM boric acid-NaOH, pH 8.0, 160 mM NaCl, 5 mM EDTA) to allow conversion of the cells into spheroplasts. The suspension was stirred on ice for 2 h and centrifuged (48,000 ϫ g, 30 min, 4°C) to remove the spheroplasts, and the collected supernatant containing the periplasmic contents was repeatedly dialyzed against 2 liters of Buffer A (10 mM MOPS-NaOH, pH 7.0). The solution was added onto a 5-ml DEAE FF anion exchange column (GE Healthcare), and target proteins were obtained by elution with a linear gradient of 0 -500 mM NaCl in A buffer. Typically, 10 mg of purified DsbA per 1 liter of LB broth was obtained and stored at Ϫ80°C until use.
Preparation of Biotin-BCCP-DsbB S-S,S-S and Its Variant-An amplified fragment encoding DsbB (EC 1.8.4.2) was obtained in the same manner as DsbA SH,SH by PCR with the primers 5Ј-GGAGATATACATATGTTGCGATTTTTGAACCAATGT-TCACAAGGC-3Ј and 5Ј-GAGCTCGAATTCTTAGCGACC-GAACAGATCACGTTTTTTCGC-3Ј and cloned into pET22b(ϩ) through the restriction sites NdeI and EcoRI. The codons for the two nonessential cysteine residues, Cys-8 and Cys-49, were then replaced by alanine and valine codons, respectively, using the QuikChange method to prevent disulfide-linked oligomerization (7,8). A fragment encoding the biotin carboxyl carrier protein (BCCP) tag was PCR-amplified from Pinpoint Xa vector and then inserted into the XhoI-HindIII digested vector coding the improved DsbB by using In-Fusion Advantage PCR Cloning Kit (Takara Bio Inc., Shiga, Japan). A vector for the double-mutated DsbB(C41A,C44A) S-S protein was prepared by the QuikChange method.
Biotin-BCCP-DsbB S-S,S-S and biotin-BCCP-DsbB(C41A, C44A) S-S proteins were also expressed and purified as described previously (12,27). E. coli C41 (DE3) harboring each expression vector was grown in 3 liters of LB broth with ampicillin and biotin at 37°C, and then 15 M isopropyl-␤-D-thiogalactopyranoside was added into the medium for protein induction at an A 600 of 0.6 followed by the additional incubation for 4 h. The cells were harvested and suspended in a buffer (50 mM Tris-HCl, pH 8.0, 300 mM NaCl, 1 mM PMSF) and then passed through a French press. After precentrifugation for 30 min at 10,000 ϫ g to remove untreated cells, the cell lysate was centrifuged for 1 h at 100,000 ϫ g to collect membrane fractions as a pellet. The membrane pellet was homogenized in a Dounce homogenizer and passed through a syringe attached to a needle (0.7 ϫ 50 mm). After the DDM solution was added to the suspension at a final concentration of 1%, the suspension was ultracentrifuged (100,000 ϫ g, 30 min) to remove insoluble membrane fractions. The supernatant was loaded on a 5-ml nickel-nitrilotriacetic acid column (GE Healthcare) equilibrated with a buffer (50 mM phosphate, pH 8.0, 300 mM NaCl, 0.1% DDM), and the target protein was eluted from the column by a 0 -0.5 M imidazole gradient at pH 6.0. The collected protein was additionally purified on a gel filtration column (Superdex 200 pg, GE Healthcare) in Buffer B (50 mM phosphate, pH 6.0, 300 mM NaCl, 0.1% DDM) and stored at Ϫ80°C until use.
Setup of 27-MHz QCM and Immobilization of Biotin-BCCP-DsbB on a NeutrAvidin-modified QCM Plate-An AFFINIX Q4 system used as the QCM instrument (Initium Co. Ltd., Tokyo, Japan) is shown in Fig. 2A. The QCM instrument has four 500-l cells equipped with a 27-MHz QCM plate (60-mthick and 8.7-mm-diameter AT-cut shear-mode quartz plate and gold electrode with a 5.7 mm 2 area) at the bottom of each cell, a stirring bar, and a temperature controlling system (23)(24)(25). NeutrAvidin (60 kDa) was immobilized covalently on the QCM plate as described previously (19,20,23). Briefly, 3,3Јdithiodipropionic acid was immobilized on a cleaned bare gold electrode, and then carboxylic acids were activated as N-hydroxysuccinimidyl esters on the surface using 1-ethyl-3-[3-(dimethylamino)propyl]carbodiimide. NeutrAvidin was reacted with the activated esters by mounting aqueous solutions on the QCM plate. The binding behaviors of biotin-BCCP-DsbB S-S,S-S to the NeutrAvidin-immobilized QCM were followed in DDM(ϩ) buffer (50 mM HEPES-NaOH, pH 7.5, 100 mM NaCl, 0.1% DDM at 25°C) as frequency changes (⌬F water ) with time. When the ⌬F water value corresponding to an amount of an anchored biotin-BCCP-DsbB S-S,S-S solubilized with DDM reached Ϫ2000 Hz (a predetermined value), 10 M free biotin was added to regulate the immobilization amount. At Ϫ2000 Hz, the amount of biotin-BCCP-DsbB S-S,S-S solubilized with DDM was estimated to be 560 ng⅐cm Ϫ2 (16 pmol cm Ϫ2 ) based on a calibration value of 0.28 ng⅐cm Ϫ2 ⅐Hz Ϫ1 (28 -30). The relationship between frequency changes and mass changes for calibration is described elsewhere (28 -30). The coverage of immobilized biotin-BCCP-DsbB S-S,S-S was calculated to be ϳ68% with 7.1 nm 2 molecule Ϫ1 of DsbB S-S,S-S area on the QCM electrode (5.7 mm 2 ).
Preparation of DsbB Embedded in Lipid Bilayers on a QCM Plate-The procedure for DsbB preparation is illustrated in Fig.  2B. A 200-l chloroform solution of total lipid extracts from E. coli (10 mg/ml) was evaporated in a recovery flask under light shielding. In the preparation of lipid bilayers containing excess ubiquinone-10, a mixed solution of the above-mentioned lipid extracts and a 200-l chloroform solution of ubiquinone-10 (1.45 mg/ml, corresponding to 11 mol%) was used. A mixed micelle solution of total lipids from E. coli and DDM was prepared by the addition of 398 l of DDM(ϩ) buffer (50 mM Tris-HCl, pH 7.5, 150 mM KCl, and 0.44% DDM) into the flask. After immobilization of biotin-BCCP-DsbB S-S,S-S in 500 l of DDM(ϩ) buffer onto a NeutrAvidin-modified QCM plate by the above-mentioned procedure, 450 l of DDM(ϩ) buffer was replaced by 20 l of the mixed micelle solution and 430 l of DDM(Ϫ) buffer (50 mM HEPES-NaOH, pH 7.5, 100 mM NaCl), resulting in a final DDM concentration of 0.028%, which was slightly higher than the critical micelle concentration of DDM (0.01%). After incubating for 1 h to form a lipid bilayer around the immobilized DsbB S-S,S-S , the remaining DDM was removed from the QCM cell by replenishing the DDM(Ϫ) buffer several times. Finally, the QCM cell was filled with 500 l of a reaction buffer (50 mM citrate, pH 6.0, with 100 mM NaCl), which was adjusted to the optimum pH of the reaction (12). Immobilization of biotin-BCCP-DsbB(C41A,C44A) S-S was also performed in the same manner.
Observation of the DsbA SH,SH Oxidation Reaction on the DsbB S-S,S-S -immobilized QCM Plate-DsbA was reduced with 30 mol eq of tris(2-carboxyethyl)phosphine for 30 min, and reduced DsbA SH,SH was then recovered using desalting spin columns just before the QCM experiments. The DsbB S-S,S-Simmobilized QCM was thermally pre-equilibrated at 25°C until the resonance frequency was held constant (Ϯ 1 Hz over 15 min), and frequency changes in response to the addition of DsbA SH,SH or DsbA variants were then measured over time.

Observation of the DsbA SH,SH Oxidation Reaction on DsbB S-S,S-S in the Absence of Endogenous Ubiquinone on a QCM Plate-
The DsbB S-S,S-S -immobilized QCM cell was prepared according to the above-mentioned procedure and was filled with 500 l of a reaction buffer (50 mM citrate, pH 6.0, 100 mM NaCl) at 25°C. An excess amount of reduced DsbA SH,SH (400 nM) was added into the DsbB S-S,S-S -immobilized QCM cell and incubated for 30 min to completely reduce DsbB S-S,S-S and endogenous ubiquinone. Then the QCM plate was incubated for 2 h with redox buffer (50 mM phosphate, pH 8.0, 100 mM NaCl, 100 mM oxidized DTT, and 1 M reduced DTT) that had a redox potential (Ϫ162 mV) between that of DsbB S-S,S-S (Ϫ224 to Ϫ207 mV) and ubiquinone (ϩ110 mV) to oxidize only DsbB S-S,SH,SH but not ubiquinol (UQ 2Ϫ ) (31). The QCM cell containing oxidative DsbB S-S,S-S and ubiquinol was filled with 500 l of reaction buffer, and frequency changes in response to the addition of DsbA SH,SH or the DsbA(C33A) SH variant were followed over time.
Quantification of Oxidized DsbA S-S by SDS-PAGE-In a 0.5-ml tube, 17 M tris(2-carboxyethyl)phosphine-reduced DsbA SH,SH was incubated with 34 M surfactant-solubilized DsbB S-S,S-S or the DsbB(C41A,C44A) S-S variant in 100 mM citrate, pH 6.0, with 0.1% DDM for 3 min at room temperature. DsbA was recovered by trichloroacetic acid precipitation into a pellet, which was then dissolved with 50 mM Tris-HCl, pH 8.0, containing 1% SDS. The free thiol of the unreacted DsbA SH,SH was modified with AMS (32). Oxidized DsbA S-S (21 kDa) and AMS-derivatized DsbA (22 kDa, the unreacted DsbA) were separated on 12% nonreducing gels by SDS-PAGE and identified using the Gel Doc EZ System (Bio-Rad) after staining with Coomassie Brilliant Blue. Kinetic Q ϭ k f k cat ͓DsbA͔ 0 (Eq. 8) We followed the formation and degradation of the DsbA-DsbB intermediate on the QCM as frequency (mass) changes over time (curve a in Fig. 3A). Then we excluded nonspecific binding (curve b in Fig. 3A) from the ladle-shaped curve before curve-fitting. The formation and reverse rate constants (k f and k r , respectively) of the intermediate and the release rate constant (k cat ) were obtained from curve fittings on the basis of Equation 3. More detailed information about transient kinetic analysis on a QCM has been described in previous studies (24,25).

Construction of DsbB-embedded Supported Lipid Bilayers on
QCM-In vitro experiments on membrane proteins, the activities of hydrophobic membrane proteins are usually measured in a surfactant-solubilized form (6,11,13,33). In the presence of a surfactant, however, enzymatic activity may be reduced, due to the creation of an unnatural environment for the membrane protein. Although using a proteoliposome is another method for solubilizing membrane proteins in solution, it is unsuitable for quantitative measurements due to the small and unclear amounts of active membrane proteins in bulk solutions. To observe and quantify thiol-disulfide exchange reactions between soluble DsbA SH,SH and membrane-embedded DsbB S-S,S-S , we examined the construction of DsbB S-S,S-S -embedded supported lipid bilayers on a QCM plate (Fig. 2). We first prepared biotin-BCCP-DsbB using an E. coli protein expression system as a fusion protein consisting of the BCCP tag, which included a region biotinylated by an endogenous biotin ligase in E. coli. Biotin-BCCP-DsbB was immobilized on a NeutrAvidin-modified QCM plate through biotin-avidin linking in the presence of DDM. This immobilization method has the advantage of regulating the amount of protein bound onto the QCM and the unidirectional orientation of membrane proteins (Fig. 2B). Next, lipid bilayers were formed around the biotin-BCCP-DsbB immobilized on the QCM plate by the addition of lipid-surfactant-mixed micelles followed by slow removal of the surfactants. We observed frequency decreases (mass increases) due to the binding of the mixed micelle around the DsbB (data not shown). After saturation of the frequency decrease, the buffer was replaced by a buffer without DDM several times to achieve a lower DDM concentration than the critical micelle concentration of DDM. To confirm the formation of lipid bilayers on the QCM plate, we also prepared a lipid bilayer on the QCM plate in the same manner by using a mixture of E. coli lipid extracts, DDM, and biotinylated lipid (5 mol%), and we subsequently observed the binding of avidin onto the biotinylated lipid bilayer (data not shown). The amount of bound avidin was consistent with the amount of avidin adsorbed onto a planar substrate, such as the gold electrode of a QCM, indicating that planar lipid bilayers around DsbB were formed on the QCM plate.
Observation of Reactions between DsbA and DsbB- Fig. 3A indicates typical frequency changes as a function of time for 27-MHz QCM on which DsbB S-S,S-S embedded in lipid bilayers was immobilized, in response to the addition of native DsbA SH,SH , DsbA(C30A) SH , and DsbA(C33A) SH variants. After the addition of DsbA SH,SH , the frequency decreased (i.e. the mass increased) rapidly and then increased gradually to return to the original frequency (curve a in Fig. 3A). This behavior corresponds to the formation of the DsbA-DsbB intermediate followed by the release of oxidized DsbA S-S from reduced DsbB S-S,SH,SH in the membrane (Fig. 2C). When the DsbA(C30A) SH variant was added onto the DsbB S-S,S-S -immobilized QCM plate, only a very slight change was observed (curve b in Fig. 3A), indicating that Cys-30 of DsbA was essential for the formation of the DsbA-DsbB intermediate. This is consistent with x-ray crystallography reports demonstrating that a disulfide bond is formed between Cys-30 of DsbA SH,SH and Cys-104 of DsbB S-S,S-S (Fig. 2C) (3). We also confirmed that oxidized DsbA S-S could not interact with DsbB S-S,S-S or reduced DsbB SH,SH,SH,SH on the QCM (curves a and b in Fig.  3B). When the DsbA(C33A) SH variant was added, it could bind to, but not be released from DsbB S-S,S-S (curve c in Fig. 3A). This indicated that Cys-30 of DsbA SH,SH formed an interprotein disulfide bond with Cys-104 of DsbB S-S,S-S and that Cys-33 of DsbA SH,SH was essential for the oxidation of DsbA SH,SH to allow it to be released from reduced DsbB S-S,SH,SH (Fig. 2C).
Thus, we could detect the intermediate of the covalently bonded DsbA-DsbB complex on a DsbB-immobilized QCM.
We also confirmed that supported bilayer structures are important for formation of DsbA-DsbB intermediates. When DsbB S-S,S-S was covered with DDM micelles, the amounts and formation rates of DsbA-DsbB intermediates produced by the addition of DsbA SH,SH (curve a in Fig. 3C) and DsbA(C33A) SH (curve c) were very small and slow compared with those shown in Fig. 3A. This clearly indicated that DDM inhibited interactions between DsbA SH,SH and DsbB S-S,S-S and/or that lipid bilayers were required to activate DsbB S-S,S-S .
Importance of the Two Cysteine Residues (Cys-41 and Cys-44) of DsbB S-S,S-S for the Reaction with DsbA SH,SH -Previously, two potential models were proposed to represent the reaction pathway of the oxidation mechanism of DsbA SH,SH by DsbB S-S,S-S as follows. 1) Oxidized DsbA S-S is released from the DsbA-DsbB intermediate before intraprotein isomerization in DsbB (attack of the reduced Cys-130 on Cys-41 of DsbB, pathway I in Fig. 1 SDS-PAGE Analysis of Oxidation Products from Reactions between DsbA and DsbB in the Bulk Solution-Although we could follow the oxidation reaction of DsbA SH,SH with DsbB S-S,S-S embedded in supported bilayers on the QCM as changes in frequency, we needed to confirm the actual products formed in this reaction. Therefore, we next performed nonreducing SDS-PAGE analysis for the reaction between DsbA SH,SH and DsbB S-S,S-S in the DDM micellar solution. After reactions of DsbA SH,SH with DsbB S-S,S-S or DsbB(C41A,C44A) SH in the bulk solution, AMS was added to modify the free thiol of the unreacted DsbA SH,SH , as described in previous reports (32). As shown in lane 1 in Fig. 6, we confirmed that oxidized DsbA S-S appeared as the reaction product (reaction time ϭ 3 min) at 21 kDa together with unreacted DsbA SH,SH appearing as a reaction product with AMS (22 kDa). The oxidation was completed within 30 min (data not shown). In addition, very little oxidized DsbA S-S was found in the reaction with DsbB(C41A,C44A) S-S as compared with that found in the reaction with DsbB S-S,S-S (lane 2). This also suggested that the two additional cysteine residues (Cys-41 and Cys-44) in DsbB S-S,S-S were essential for achieving a good yield of oxidized DsbA S-S .    Fig. 3A) is described by Equations 1-3 under "Experimental Procedures" (24,25). Kinetic parameters, i.e. the formation and reverse rate constants (k f and k r , respectively) and the oxidation rate constant (k cat ), were obtained from curvefittings as optimized coefficients in these equations. Experi-mental curves, generated using different concentrations of DsbA SH,SH (Fig. 7A) were fitted well with theoretical curves described by Equations 2 and 3. The k f and k cat values were obtained from one experimental curve, and the average of values at three different concentrations of DsbA SH,SH in Fig.  7A were calculated. The obtained binding rate constants (k f ϭ 5.9 ϫ 10 5 M Ϫ1 s Ϫ1 ) and catalytic rate constants (k cat ϭ 0.12 s Ϫ1 ) are summarized in Table 1. The k r value could not be accurately measured because it was negligibly smaller than the k cat value.  DECEMBER 13, 2013 • VOLUME 288 • NUMBER 50

JOURNAL OF BIOLOGICAL CHEMISTRY 35977
The reaction between DsbA(C33A) SH and DsbB S-S,S-S or the DsbB(C41A,C44A) S-S variant showed simple binding behaviors (Figs. 3A and 4A) and is described by Equations 9 -12 under "Experimental Procedures." The relaxation time () was calculated from curve-fittings of frequency changes at various concentrations, as shown in Fig. 7, B and C. The formation and reverse rate constants (k f and k r , respectively) for the intermediate were calculated from the slope and intercept of Equation 12 (Fig. 8). The parameter corresponding to the dissociation constants (K d ) was obtained as the ratio of k r to k f . The obtained kinetic parameters are summarized in Table 1. The k f value (2.3 ϫ 10 5 M Ϫ1 s Ϫ1 ) obtained from the reaction of the DsbA(C33A) SH variant and DsbB S-S,S-S was close to k f ϭ 5.9 ϫ 10 5 M Ϫ1 s Ϫ1 , which was obtained from the reaction of DsbA SH,SH and DsbB S-S,S-S , suggesting that the DsbA(C33A) SH variant maintained almost the same activity as DsbA SH,SH .
Because the k f values of DsbA SH,SH and the DsbA(C33A) SH variant were close, the k r value for DsbA SH,SH , which could not be obtained (Table 1) Table 1). This may be explained by the unexpected effect of C41A and C44A mutations on the attack of DsbA on DsbB.
In addition, we also performed kinetic analysis of the reactions of DsbA SH,SH or the DsbA(C33A) SH variant with DsbB S-S,S-S in the absence of endogenous ubiquinone (Fig. 7, D and E) and in the presence of excess ubiquinone-10 ( Fig. 7F) in the same manner. The obtained kinetic parameters are also summarized in Table 1. Despite variance in the formation rate of the DsbA-DsbB intermediate (k f ), these parameters indicated that k f values and the reaction rates of DsbA SH,SH oxidation (k cat values) were mostly independent of ubiquinone, as compared with the effects of Ala mutations at Cys-41 and Cys-44 in DsbB on the reduction of k r values (Fig. 8). These kinetic values are also shown in the reaction scheme in Fig. 9.

DISCUSSION
In the thiol-disulfide exchange reactions of DsbA SH,SH on DsbB S-S,S-S embedded in supported lipid bilayers bound to a QCM plate, we could observe the covalently bonded DsbA-DsbB intermediate in real time. This allowed us to obtain kinetic parameters of the reaction, including the formation and reverse rate constants (k f and k r , respectively) of the intermediate and the DsbA oxidation rate constant (k cat ). Absorbance measurements based on the electronic states of ubiquinone in DsbB have previously been used to monitor reactions involving sequential electron transfers from DsbB S-S,S-S to ubiquinone. In contrast, the kinetic parameters obtained using our QCM method reflected electron transfer from DsbA SH,SH to DsbB S-S,S-S , thereby providing additional information on this reaction step.

Comparison of Previous Absorbance Measurements with the Kinetic Parameters Obtained in the Current Study-Previous
kinetic characterization of the oxidation reaction of DsbA with DsbB solubilized by DDM micelles was performed by stoppedflow absorbance measurements (13). Despite different experimental conditions, the k f value (5.9 ϫ 10 5 M Ϫ1 s Ϫ1 ) obtained showing purple color, see Stable intermediate in Fig. 9B from the formation of the covalent DsbA-DsbB complex in this study was similar to the rate constant (ϳ5 ϫ 10 5 M Ϫ1 s Ϫ1 ) of formation of the DsbB-ubiquinone complex (purple complex) in the process of electron transfer from DsbA to ubiquinone reported in the previous study (reaction conditions: 50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 0.04% DDM at 10°C) (13). This suggests that the rate-limiting step is likely the covalent binding of DsbA to DsbB. On the other hand, the rate of DsbA oxidation (ϳ2 s Ϫ1 at pH 8.0), which was found to be the same as the rate of quinone reduction in the previous work, is larger than the k cat value (0.12 s Ϫ1 at pH 6.0) obtained in the current work. This may be due to the higher pH used in the previous work, as the presence of the thiolate anion at Cys-33(DsbA) increases with increasing pH. We confirmed the similar k f value and increased k cat value at pH 8.0 in our other studies (34). Unlike absorbance   . 7E) (d). Reactions were performed in 50 mM citrate, pH 6.0, with 100 mM NaCl at 25°C.

TABLE 1 Kinetic parameters of the thiol-disulfide exchange reaction between DsbA and DsbB embedded in supported bilayers on the 27-MHz QCM
The experimental conditions were 50 mM citrate, pH 6.0, 100 mM NaCl, 25°C. Each kinetic parameter was obtained from Equations 2 and 3 and from 4 -6 and is reported with experimental errors. measurements, where the solution condition requires a pH greater than 8.0 to obtain the purple complex of DsbB-ubiquinone by deprotonation of Cys-44(DsbB) (14), the current method allows us to perform experiments at the optimum pH for the specific reaction (here, pH 6.0) (12) (Fig. 9B). The Mechanism of DsbA/DsbB Reaction Based on Kinetic Parameters-The destiny of the DsbA-DsbB intermediate can be determined by both the k r and k cat values (Fig. 8B). On comparing the first-order rate constants, k r and k cat , between DsbA SH,SH and DsbB S-S,S-S , we found that the k cat value (0.12 s Ϫ1 ) was 60 times larger than the estimated k r value (0.0020 s Ϫ1 ). This suggested that the reaction of DsbA SH,SH with DsbB S-S,S-S was a consecutive reaction. Moreover, the k r value in the reaction between DsbA SH,SH and the DsbB(C41A,C44A) S-S variant (0.12 s Ϫ1 ) was also 60 times larger than the k r value in the reaction between DsbA SH,SH and DsbB S-S,S-S (0.0020 s Ϫ1 ; Fig. 9A). These results clearly suggested that the stability of the DsbA-DsbB(C41A,C44A) S-S intermediate was largely reduced due to the lack of these two cysteine residues.

DsbA
In the thiol-disulfide exchange reaction between DsbA SH,SH and DsbB S-S,S-S , the interprotein disulfide bond of Cys-30 with Cys-104 is first formed by nucleophilic attack of Cys-30 of DsbA SH,SH on Cys-104 of DsbB S-S,S-S . Then, reduced Cys-130 attacks Cys-104 or Cys-41 of the Cys-41-Cys-44 disulfide bond of DsbB. The attack on Cys-104 results in the reverse reaction, whereas the attack on Cys-41 results in the formation of the interloop Cys-41-Cys-130 disulfide bond in pathway II. The formation of the interloop Cys-41-Cys-130 prevents the reverse reaction caused by the attack of Cys-130 on Cys-104 as a stable intermediate. We found that the formation of the stable intermediate reduced the rate of the reverse reaction from k r ϭ 0.12 s Ϫ1 to k r ϭ 0.0020 s Ϫ1 . This allowed the forward reaction to proceed (Fig. 9B). Thus, our data demonstrated that the thiol-disulfide exchange reaction between DsbA SH,SH and DsbB S-S,S-S could occur efficiently through the stable intermediate, as indicated by pathway II.
Ubiquinone-independent Oxidation of DsbA on DsbB-We expected that ubiquinone in DsbB could contribute to the reduction of the reverse rate constant (k r ) in the stable intermediate due to delocalization of the electrons transferred from DsbA because ubiquinone can interact with the Cys-44 thiolate anion produced by the cleavage of the Cys-41-Cys-44 disulfide bond of DsbB (14). However, we obtained kinetic parameters