The N-terminal Sequence (Residues 1–65) Is Essential for Dimerization, Activities, and Peptide Binding of Escherichia coli DsbC*

Limited proteolysis of DsbC with trypsin resulted in a compact and stable C-terminal fragment (residues 66–216), fDsbC, which retains the active site sequence, -Cys98-Gly-Tyr-Cys101-, and shows only minor differences in conformation compared with that of the intact molecule. The pK a of active site thiol and theK SS with glutathione are very close to that of DsbC, respectively; however, fDsbC is inactive as an isomerase in catalyzing the formation of correct disulfide bonds in scrambled RNase A and denatured and reduced bovine pancreatic trypsin inhibitor and shows only 13% thiol-protein oxidoreductase activity (TPOR) of DsbC. In contrast to the dimeric DsbC, fDsbC exists as a monomer and has no chaperone activity in assisting the reactivation of denaturedd-glyceraldehyde-3-phosphate dehydrogenase. The heterodimer of DsbC with the inactive DsbC carboxymethylated at both active site thiols shows about 50% TPOR activity of DsbC but no isomerase activity, indicating that the DsbC subunit in the heterodimer displays full TPOR activity but little, if any, isomerase activity. It is concluded that the N-terminal sequence (residues 1–65) is essential for dimer formation and chaperone activity of DsbC. The active sites in both subunits of the dimeric DsbC appear to be essential for its isomerase activity.

The Dsb protein family in bacterial periplasm has been recently characterized to be responsible for the formation of native disulfide bonds of newly synthesized polypeptides (1). At least six proteins, DsbA, DsbB, DsbC, DsbD, DsbE, and DsbG, have been identified, and they work in conjunction with one another (2). DsbC is a soluble protein catalyzing the rearrangement of disulfide bonds (3) and therefore considered as the prokaryotic counterpart of the eukaryotic protein-disulfide isomerase (PDI) 1 (3,4). PDI has recently been shown to have not only enzyme activities but also chaperone activity (5)(6)(7). Compared with PDI, DsbC shows lower isomerase activity but even more marked chaperone activity (8).
As a member of the thioredoxin superfamily, DsbC shares 24% sequence identity with DsbG (9) but no apparent sequence homology with other members of the Dsb family or with other members of the thioredoxin superfamily except for the motif of Phe-X-X-X-X-Cys-X-X-Cys in the active site (10). DsbC is a homodimer (3), and each of its 23.4-kDa subunits is composed of 216 amino acid residues with four cysteines, two in the active site sequence of -Cys 98 -Gly-Tyr-Cys 101 -and the other two at positions 141 and 163, believed to form a disulfide bond with a structural function (3). It was predicted that the molecule appeared to consist of C-and N-terminal domains of similar length, with the C-terminal half recognized as another variation on the thioredoxin theme (10). The crystal structure of oxidized DsbC shows that the molecule has a V-shape with each arm for one subunit, and each subunit consists of a Cterminal thioredoxin-like domain connected by a hinged linker helix to an N-terminal domain responsible for dimerization (9).
Limited proteolysis has been widely used in protein studies for the exploration of surface regions, ligand-induced conformational changes, domain boundaries, and protein unfolding/refolding (11). Based on the profile of limited digestion by trypsin and V8 protease, the architecture of PDI appears to be consistent with a model of four consecutive domains, each with a thioredoxin fold (12).
In this paper, we report the properties of a C-terminal fragment (residues 66 -216) of DsbC generated by limited trypsin digestion. This fragment, fDsbC, is monomeric in contrast to the dimeric intact molecule and has a compact structure. Although containing all four cysteine residues with very similar chemical properties to those of DsbC, it shows low thiol-protein oxidoreductase (TPOR) activity and neither isomerase nor chaperone activity. Hybrid experiments suggest that the dimeric DsbC requires both of its active sites to be intact for its isomerase activity.
Preparation-The expression plasmid pDsbC containing the fulllength DsbC precursor gene is a generous gift from Dr. Rudi Glockshuber (Eidgenössische Technische Hochschule, Hönggerberg, Switzerland). DsbC was prepared according to Missiakas et al. (13) and Chen et al. (8). D-Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) from rabbit muscle (14) was kindly provided by J. Li and N. X. Zhang of this laboratory. Scrambled RNase A (sRNase) was prepared essentially according to Hillson et al. (15). S-Carboxymethylated RNase A (mR-Nase) was prepared by overnight incubation of RNase A at 1.8 mM with 130 mM DTT and 6 M GdnHCl in 0.4 M Tris-HCl buffer (pH 8.0) at room temperature and then modification with a 10-fold excess of iodoacetic acid in 1.5 M Tris-HCl with 6 M GdnHCl (pH 9) at room temperature for 1 h. The reaction mixture was adjusted to pH 4 -4.5 with glacial acetic acid, dialyzed thoroughly against 0.02 M acetic acid at 4°C, and lyophilized.
fDsbC was prepared by limited digestion of DsbC with trypsin. DsbC at 1 mg/ml was incubated with trypsin at 0.01 mg/ml in 0.1 M ammonium bicarbonate buffer (pH 8.1) at 37°C for 2-3 h. The reaction mixture was then loaded on a Bio-Scale Q 2 column (Bio-Rad) equilibrated with 50 mM Tris-HCl buffer (pH 8.0) and eluted with a linear gradient of 0 -0.2 M NaCl in 20 ml of the same buffer. The main peak was collected, dialyzed thoroughly against 50 mM NH 4 HCO 3 at 4°C, and lyophilized. The condition for trypsin digestion was optimized by examination of the reaction products on 15% SDS-polyacrylamide gel electrophoresis (PAGE).
DsbC and fDsbC carboxymethylated at the N-terminal thiol and at both thiols (mmDsbC and mmfDsbC) in their active site sequence -Cys 98 -Gly-Tyr-Cys 101 -, respectively, were prepared according to Zapun et al. (3) and were determined to be all devoid of enzyme activity.
Determinations-Concentrations of proteins were determined spectrophotometrically at 280 nm with the following absorption coefficients (A 1 cm 0.1% ): 0.66 for bovine serum albumin, 0.98 for GAPDH (16), and 0.7 for DsbC (3). The A 1 cm 0.1% value of fDsbC was determined to be 0.87 by protein concentration determinations by the Bradford method (17) using bovine serum albumin as a standard. For the convenience of comparison, homotetrameric GAPDH and homodimeric DsbC were considered as protomers in the calculation of molar ratios. Thiol groups were determined with 5,5-dithiobis (2-nitrobenzoic acid) in the presence of 6.4 M GdnHCl (18).
The N-terminal amino acid sequence of fDsbC was determined on a MilliGen/Biosearch model 6600 ProSequencer linked to a Waters high pressure liquid chromatography system with simultaneous detection at 269 and 313 nm. The mass of fDsbC was determined by electrospray ionization mass spectrometry on a VG Platform spectrometer. The sizes of native fDsbC and DsbC were estimated by gel filtration on a Superose 12 HR 10/30 column (Amersham Pharmacia Biotech) calibrated with GAPDH, bovine serum albumin, ovalbumin, chymotrypsinogen, and cytochrome c as mass markers. Hybridization and nondenaturing PAGE on a 6% gel were performed exactly as described by Zapun et al. (3) to identify the oligomeric state of DsbC and fDsbC. The ratios of the protein bands on the gel were scanned by a Bio-Rad DS-670 densitometer. Protein was recovered from the gel by slicing out the band and cutting into small pieces before soaking in 50 mM sodium phosphate buffer (pH 7.5) overnight, desalting of the supernatant, and lyophilization.
CD spectra from 200 to 250 nm were determined with a Jasco 500 spectropolarimeter at 25°C. Fluorescence spectra were measured at 25°C in a Hitachi F-4010 spectrofluorimeter with excitation wavelengths of 280 and 295 nm for the intrinsic and 373 nm for the ANS spectra. The sample buffer was 0.1 M potassium phosphate with 2.5 mM EDTA (pH 7.5), and 1 mM DTT was also used for reduced proteins.
pK a values of the nucleophilic cysteine (Cys 98 ) were determined by following the increase of thiolate-specific absorbance at 240 nm with the change of pH at 30°C in buffer A (10 mM KH 2 PO 4 , 10 mM K 2 HPO 4 , 10 mM sodium citrate, 10 mM Tris, 200 mM KCl, and 1 mM EDTA, pH 7.7). DsbC and fDsbC (ϳ200 M) were first reduced with 10 mM DTT in 0.1 M Tris-HCl (pH 8.0) for 10 min at 25°C, and the excess of DTT was removed using a Superose 12 HR 10/30 column equilibrated with buffer A. The oxidized proteins treated in the same way without DTT were used as references. The pH values of 2-ml protein solutions at ϳ25 M were measured by an Orion pH meter with stepwise addition of 200 mM HCl in portions of 25 l, and the absorbance values at 280 and 240 nm were recorded on a UV-1601 UV-visible spectrophotometer (SHI-MADZU). Since the absorbance of DsbC at 280 nm was determined to be pH-independent (data not shown), (A 240 red /A 280 red )/(A 240 oxi /A 280 oxi ) was used as a measure of the fraction of the Cys 98 thiolate in DsbC and fDsbC. The data were fitted according to the Henderson-Hasselbach equation (19).
The equilibrium constants for disulfide bond formation in the active site of fDsbC and DsbC, with glutathione as thiol-disulfide reagent, were determined according to a method well established by Zapun et al. (3). The samples were analyzed by reverse phase FPLC on a PepRPC TM HR 5/5 column (Amersham Pharmacia Biotech) at room temperature with a gradient of 30 -38% (v/v) acetonitrile in 20 ml of 0.1% trifluoroacetic acid.
Activity Assay-Activities of disulfide isomerase based on the isomerization of sRNase to the native enzyme and TPOR using the reduction of insulin with a linkage to glutathione reductase were assayed according to Lambert and Freedman (20), protein-disulfide oxidoreductase (PDOR) was determined based on the increases of fluorescence emission at 519 nm upon the reduction of intercatenary disulfide bonds of difluoresceinthiocarbamyl-insulin (21). The effects of DsbC and fDsbC on the disulfide-coupled refolding reaction of fully reduced BPTI were performed basically according to Zapun and Creighton (22). The refolding reaction mixture was incubated at 25°C for 20 min before the addition of reduced BPTI to initiate the reaction in order to ensure the redox equilibrium in the active site of DsbC and fDsbC under the used concentration of GSSG/GSH. The acid-trapped samples were analyzed by FPLC on a PepRPC TM HR 5/5 reverse phase column at room temperature with a gradient of 9 -36% (v/v) acetonitrile in 50 ml of 0.1% trifluoroacetic acid.
Denaturation and Renaturation of GAPDH-Denaturation with GdnHCl and assisted reactivation of GAPDH upon dilution in the presence of DsbC and fDsbC were carried out according to Chen et al. (8). Aggregation during refolding of denatured GAPDH was followed continuously at 25°C by 90°light scattering at 488 nm in a Hitachi F-4010 spectrofluorimeter. Fig. 1, by digestion of 1 mg/ml DsbC at 37°C with trypsin at a weight ratio of 0.01, the DsbC band disappeared completely after 100 min, and after 3 h a fragment of 17 kDa was the only digestion product detectable on the gel. This component was purified through one-step anion exchange chromatography and named as fDsbC.

Limited Digestion of DsbC by Trypsin-As shown in
Identification of fDsbC-The first three amino acid residues at the N terminus of fDsbC have been determined to be Met-Leu-Leu, indicating that fDsbC results from the cleavage of DsbC between Lys 65 and Met 66 by trypsin (Scheme I). The mass of fDsbC determined by mass spectrometry is 16,683 Da, in agreement with the calculated value of 16,687 Da for the fragment of 66 -216.
Oligomeric State-Upon size exclusion chromatography, DsbC was eluted at the position corresponding to an apparent mass of 58 kDa, similar to the 67 kDa reported by Zapun et al. (3), and fDsbC was eluted at 22 kDa, suggesting that the former exists as a dimer and the latter as a monomer. It should be noted that both of the apparent mass values are higher than the expected values of 46.9 and 16.7 kDa, respectively and the reduced forms of DsbC and fDsbC, mmDsbC and mmfDsbC, showed exactly the same elution position as those of the respective oxidized forms, indicating that redox state and thiol modification do not affect the association properties. Thus, mmDsbC is a dimer, and mmfDsbC is a monomer.
As shown in Fig. 2, the nondenaturing PAGE of the hybrid mixtures of DsbC/mmDsbC at all ratios contain three bands (Fig. 2, lanes 2-6) for the native protein, the heterodimer, and the modified protein, while the mixture of fDsbC/mmfDsbC shows only two bands with no heterodimer between the two homodimers (lane 9). In addition, no band of heterodimeric hybrid appeared for DsbC/mmfDsbC (lane 11) and DsbC/fDsbC (lane 12). The above finding indicates the dimeric state of DsbC and the monomeric state of fDsbC and mmfDsbC, in agreement with the results of gel filtration.
Cysteine Thiol Groups of fDsbC-fDsbC freshly prepared, like DsbC, is in the oxidized form, since no thiol group was detected by Ellman's assay when it was denatured with 6.4 M GdnHCl.
DsbC and fDsbC, carboxymethylated at the N-terminal thiol in the active site sequence -Cys 98 -Gly-Tyr-Cys 101 -, show 0.8 and 0.9 thiol groups, respectively, after denaturation, suggesting that in fDsbC, like in DsbC (8), only the N-terminal thiol group (Cys 98 ) of the active site was alkylated in the native conformation, so that only Cys 101 can be detected after denaturation. The other two non-active site thiol groups form a disulfide bond.
There is no significant change in reactivity of Cys 98 in DsbC caused by removing the N-terminal segment (residues 1-65) as shown by the pK a values, 4.3 Ϯ 0.2 (n ϭ 3) for fDsbC as compared with 4.1 Ϯ 0.3 (n ϭ 6) for DsbC (Fig. 3). As shown in Fig. 4, the equilibrium constant K SS for the formation of the accessible disulfide bond of fDsbC with glutathione is 205 M, which is close to the K SS of 273 M of DsbC (3), indicating that the disulfide bond of fDsbC at the active site is nearly as unstable as that of DsbC.
Fluorescence and CD Spectra-DsbC contains one Trp at position 140 and eight Tyr residues, whereas fDsbC keeps Trp 140 but has only six Tyr residues. The intrinsic fluorescence spectra of fDsbC with excitation wavelengths at either 280 or 295 nm are similar to that of the intact molecule in terms of the emission maximum but with lowered emission intensity (Fig. 5,  A and B). The ANS fluorescence spectra are nearly the same for DsbC and fDsbC (Fig. 5C). The CD spectrum of fDsbC is also similar to that of DsbC (Fig. 5D).
For oxidized and reduced fDsbC, like DsbC (3) but unlike DsbA (23), there is no difference between the intrinsic fluorescence spectra with excitation wavelength at either 280 or 295 nm, ANS fluorescence spectra, and CD spectra (data not shown), indicating that the oxidized and reduced fDsbC have essentially the same fold.
Carboxymethylation at the N-terminal thiol in the active site sequence -Cys 98 -Gly-Tyr-Cys 101 -of DsbC and fDsbC shows little effect on its intrinsic fluorescence spectra with excitation at either 280 or 295 nm and ANS fluorescence spectra (data not shown). However, mmDsbC and mmfDsbC show a 3-nm blueshifted emission maximum in their intrinsic fluorescence spectra with excitation at 280 or 295 nm (Fig. 5, A and B). The ANS fluorescence spectra for mmDsbC and mmfDsbC show a more marked, 15-nm, blue shift in emission and markedly higher intensity (Fig. 5C). The above suggest that the alkylation of both thiols in the active site perturbs the structure of DsbC and fDsbC more markedly than the modification at the protruding thiol group (Cys 98 ) alone.
Enzymatic Activities-Like DsbC, fDsbC catalyzes the reduction of insulin in its TPOR activity but with a considerably lower activity, 13.5 Ϯ 1.4% of that of DsbC (n ϭ 3). As shown in the Table I for the kinetic constants of insulin reduction catalyzed by fDsbC and DsbC, the K m for fDsbC is 2.58-fold that for DsbC, and the k cat of fDsbC is 12.4% that of DsbC. Thus the factor of k cat /K m of fDsbC is only 4.8% that of DsbC. However, fDsbC shows no isomerase activity using sRNase as a substrate compared with DsbC with an isomerase activity of 590 units/g. The kinetics of disappearance of the fully reduced unfolded BPTI and the appearance of the native three disulfide-bonded BPTI catalyzed by DsbC and fDsbC was compared in order to provide more information about the enzyme activity of fDsbC (Fig. 6). In the absence of any catalyst, the half-time for the disappearance of the reduced BPTI, about 4.2 min, was decreased to 3.2 min by fDsbC and further to 2 min by DsbC. This indicated that fDsbC catalyzes the oxidation of the thiols in reduced BPTI like DsbC but at a lower rate. However, in contrast to DsbC, which catalyzed 50% of BPTI to be reactivated in 13 min, fDsbC had little effect on the formation of SCHEME 1. Potential sites for trypsin attack in the DsbC sequence. All Lys and Arg residues in the DsbC sequence (10)  native BPTI just like the case without any catalyst. The above finding indicates that fDsbC showed no isomerase activity as well with reduced BPTI as a substrate.
Enzymatic Activities of Hybrid Mixtures of DsbC/mmDsbC- Table II shows the relative intensity of the three bands for the native protein, the heterodimer, and the inactive modified protein, the TPOR activity, and the isomerase activity of the hybrid mixtures with different ratios of DsbC/mmDsbC in Fig.  2. The native species, DsbC, is fully active, and the modified mmDsbC is inactive. It is noted that the experimentally determined TPOR activities of hybrid mixtures agree well with the sums of the relative intensity of the native and the half-intensity of the heterodimer bands. The isomerase activities experimentally determined are very close to the corresponding values of the relative band intensity of the native species. The above finding suggests that the heterodimer of DsbC/mmDsbC shows little (if any) isomerase activity but almost 50% TPOR activity of DsbC; i.e. the DsbC subunit in the heterodimer displays full TPOR activity but little (if any) isomerase activity. The isomerase activities of DsbC have been determined in the absence and presence of the same concentration of mmDsbC and found to be identical (530 units/g). This eliminated the possibility of unproductive binding of the substrate by mmDsbC that might contribute to the large decrease of the isomerase activity of the hybrid mixtures of DsbC and mmDsbC.
Effects of fDsbC on Reactivation and Aggregation of Denatured GAPDH-Reactivation of denatured GAPDH was used to examine the chaperone activity of fDsbC independent to its thiol-protein oxidoreductase activity (5). As shown in Fig. 7A, reactivation of GAPDH in the presence of DsbC increases from 6 to 30% as the molar ratio of DsbC to GAPDH increases to 25. However, fDsbC at the same range of molar ratios shows no effect on the reactivation of the denatured GAPDH. In the presence of a 20-fold molar excess of DsbC, the aggregation of   denatured GAPDH during refolding decreases significantly in both rate and extent, but fDsbC at the same concentration has no effect (Fig. 7B).
Effects of Nonnative RNase A on PDOR Activities of DsbC and fDsbC-The PDOR assay recently developed using a fluorescent derivative of insulin as the substrate (21) was employed for measuring the effects of peptides on the reductase activity of fDsbC for its higher sensitivity than the conventional TPOR assay. As shown in Fig. 8A, sRNase, as a big misfolded peptide and a good substrate for DsbC, inhibits the PDOR activity of both DsbC and fDsbC. However, mRNase at higher concentrations inhibits the PDOR activity of DsbC but has no effect on the activity of fDsbC at the same concentrations (Fig. 8B). DISCUSSION Compact Structure of fDsbC-Although there are 21 potential sites for trypsin cleavage in DsbC, as shown in Scheme I, the protein is only cleaved at a limited number of sites under the employed conditions with the site between Lys 65 and Met 66 , one of the most liable to be cleaved. All of the 16 sites within the fDsbC fragment seem to be unaffected during the first 3 h of digestion at 37°C even with 1:10 (weight ratio) trypsin or prolonged digestion time of 20 h with 1:100 trypsin at 37°C.
The resistance toward trypsin proteolysis of the C-terminal fragment suggests a compact structure of this part of the DsbC molecule. The peptide segment around the nick site between Lys 65 and Met 66 appears to be a flexible and exposed loop or a mobile linker region between domains (11). The nick site is indeed located within the movable hinged linker ␣-helix, which connects the N-terminal domain (residues 1-61) and the Cterminal domain (residues 78 -216) with a thioredoxin fold (9) and is most accessible for trypsin attack. The fragment fDsbC is thus just the C-terminal thioredoxin domain with 12 residues in the linker at the N terminus.
The N-terminal domain, consisting of six-stranded anti-parallel ␤-sheets (9), is obviously much less compact than the C-terminal region for its greater susceptibility to digestion, since in the first 50 min of digestion two additional weak bands with a mass of about 21 kDa (cleaved between Lys 28 and Thr 29 ) and 19 kDa (cleaved between Lys 44 and His 45 ) appeared in addition to the band of the intact molecules and the main band with a mass of 17 kDa (cleaved between Lys 65 and Met 66 ) and  7. The effects of DsbC and fDsbC on the reactivation (A) and aggregation (B) during refolding of denatured GAPDH. The refolding was initiated by a 50-fold dilution of GdnHCl-denatured GAPDH to 2.8 M in 0.1 M phosphate buffer (pH 7.5) containing 5 mM DTT, 2.5 mM EDTA, and DsbC (q) or fDsbC (E) at different molar ratios to GAPDH as indicated. The reactivation mixture was first kept at 4°C for 30 min and then for additional 3 h at 25°C before an aliquot containing 2 g of GAPDH was taken for activity assay at 25°C. Aggregation was followed by light scattering at 488 nm after dilution of GAPDH to 2.8 M in the absence (dashed line) and presence of DsbC (solid curve 1) or fDsbC (solid curve 2) at a ratio to GAPDH of 20. is digested to small pieces during trypsin treatment.
fDsbC Is Monomeric in Contrast to the Dimeric DsbC-By size exclusion chromatography and nondenaturing electrophoresis of hybrid mixtures, it is concluded that fDsbC is monomeric. The fact that fDsbC retains the general conformation as in the intact molecule led us to suggest that the N-terminal fragment of 65 residues is responsible for the dimerization of DsbC subunits, and this is now confirmed by the crystal structure (9). The two subunits of DsbC form a V-shaped molecule with each arm consisting of two separate domains connected by a hinged linker ␣-helix (see Scheme II). The N-terminal domains (residues 1-61) from each monomer form a pair of sixstranded anti-parallel ␤-sheets that form the dimer interface at the base of the V shaped molecule. The N-terminal domain is thus named the dimerization domain. It becomes very clear that removal of the dimerization domain results in fDsbC as two separate monomeric C-terminal thioredoxin domains.
In nondenaturing PAGE of a hybrid mixture of DsbC/ mmDsbC at a ratio of 1:1, the relative proportion of the native protein, the heterodimer, and the inactive modified protein is not 1:2:1 as would be expected from random association of the subunits but actually approximately 1:1:1. It is interesting to find that a rerun of nondenaturing PAGE of the recovered heterodimeric protein of DsbC/mmDsbC from the gel shows again three bands with a ratio of approximately 1:1:1 (data not shown). The above result indicates a lower affinity to form heterodimers compared with that of homodimers and an equilibrium between the heterodimer and the homodimers. According to the crystal structure, the dimerization is essentially via the hydrogen bonds between the ␤-sheets of the N-terminal domains and the two C-terminal thioredoxin domains with the active sites are normally 38 Å apart across the cleft. In addition, the linker helix is sufficiently long that the N-terminal association domain and the C-terminal catalytic domain within a monomer also have minimal interaction. It is hard to understand why the modification at the active site of one subunit affects the dimerization with a normal subunit unless the moderate perturbed conformation of the modified subunit is responsible. We also found that the heterodimer appeared from a mixture of DsbC and mmDsbC under nondenaturing conditions, indicating that native DsbC and mmDsbC also exist in equilibrium of dissociation and association (data not shown). However, the process under nondenaturing conditions is very slow and takes about 24 h to reach equilibrium.
Enzyme and Chaperone Activities-fDsbC, as a monomer, shows no isomerase activity in catalyzing the formation of native molecules either from sRNase or from reduced BPTI but retains 13.5% TPOR activity and part of the oxidase activity of homodimeric DsbC.
It is suggested that the combination of the two basic structural features, i.e. a large uncharged surface (40 ϫ 40 ϫ 25 Å) consisting of the cleft of the V for substrate binding and a thiolate active site, account for its isomerase activity. The broad uncharged cleft in DsbC molecule is sufficient to allow the binding of target protein and may be involved in both the chaperone and isomerase activities (9). Moreover, the hinged linker helix may provide sufficient flexibility to allow the binding of different sized substrates (9).
Since fDsbC has an intact active site with unchanged chemical properties (the protruded Cys 98 , pK a , and redox constant K SS ) and a slightly perturbed molecular conformation, the absence of the large uncharged surface within the V-shaped dimeric molecule resulting from the removal of the N-terminal dimerization domains very likely leads to the absence of isomerase activity. In this connection, the K m of fDsbC for insulin is 2.58-fold higher than that of DsbC, indicating a poorer substrate binding with fDsbC than with DsbC. The fact that the DsbC monomer in the heterodimer of DsbC/mmDsbC shows no isomerase activity but full TPOR activity of DsbC suggests that both of the active sites of dimeric DsbC may be necessary for its isomerase activity and that the monomeric state is sufficient for its TPOR activity. Our present data provide an experimental demonstration of the speculation by Zapun et al. (3) that in the catalyzed isomerization of mispaired disulfides of a scrambled protein with no free thiols, the simultaneous break of the another disulfide may be needed, so that a different disulfide can be formed via disulfide interchange. Similar to PDI, the two active sites with one in each domain of DsbC are obviously favorable and probably necessary to catalyze the isomerization of scrambled substrate. The catalyzed reduction of a disulfide only needs one active site of the enzyme to function. In this respect, it is speculated that in the recycling of DsbC in cells the reduction of the disulfide in each domain of DsbC by DsbD may be independent. In the determination of reaction kinetics of DsbC with glutathione, Zapun et al. (3) proposed that the two active sites of DsbC function independently, since no indication of interactions between the two monomers of the DsbC dimer was found. Meanwhile, the sufficient area for substrate binding is also necessary for TPOR and oxidase activities, since fDsbC lacking the N-terminal domain has much lower activities. The possibility is not excluded that the conformational change in modified subunit could be responsible for the activities of the heterodimer. It is also to be recalled that in contrast to the dimeric DsbC and PDI with substantial isomerase activity, thioredoxin is monomeric and shows little isomerase activity (24). DsbA is also monomeric but has an additional helical insert, which, together with the thioredoxin domain, provides a much more extended hydrophobic surface than that of thioredoxin for substrate binding, and therefore DsbA shows low isomerase activity of about 5% of that PDI (25) and 20% of DsbC (3). In this connection, it is known that the active sites of some enzymes are shared between different subunits (26,27). For DsbA and the monomeric domain a of PDI with 14% isomerase activity (28) of PDI, their small size may allow two enzyme molecules to attack two disulfide bonds of a substrate simultaneously.
Since the chaperone and PDOR activities of DsbC can be inhibited by sRNase and mRNase at high concentrations but not by small peptides, it has been suggested that DsbC has an extended surface for peptide binding so that only a relatively large unfolded peptide is able to compete with the substrate or the target folding intermediate for binding to DsbC (8). fDsbC is no longer active as a chaperone on the reactivation and prevention of aggregation of denatured GAPDH, and its PDOR SCHEME 2. Schematic presentation of the domain structure of the DsbC molecule. DsbC is a V-shaped dimeric molecule with each arm of the V for one subunit. Each subunit consists of a C-terminal thioredoxin-like domain (C) and an N-terminal domain (N) responsible for dimerization. The two domains are connected by a hinged linker helix. The arrow represents the cleavage site of DsbC by trypsin. Thiols at active sites are shown as balls. The dimer interface consists essentially of hydrogen bonds between the two N domains. There is a broad uncharged cleft between the two domains. activity is not inhibited by mRNase, suggesting that fDsbC has lost the ability for large peptide binding and consequently the activity as a chaperone. This has now also been supported by the three-dimensional structure (9). In contrast to mRNase, sRNase decreases PDOR activity of DsbC and fDsbC at very low concentrations, since DTT present in the determination of PDOR activity may reduce the disulfide of sRNase and lead to disulfide exchange with DsbC and fDsbC, and results in the inhibition of the PDOR activity. Many chaperones have ring structures to form a peptide binding site at the end of the ring contributed by each subunit (29); e.g. GroEL has two sevenmember rings, the eukaryotic cytosolic chaperonin, CCT, and archaebacteria thermosome have two eight-member rings, and archaebacteria TF55 has two rings of nine subunits. In contrast, in other chaperones, like dimeric PDI (30) and DsbC, the multidomain structure seems to contribute to form an extended site for substrate binding.