In Gram-negative bacteria, the Dsb (disulfide bond) family of proteins with the CXXC motif in the thioredoxin (TRX)-fold domain is involved in the formation and/or rearrangement of disulfide bonds of substrate proteins located in the periplasm (Cho and Collet, 2013; Martin, 1995). In Escherichia coli, there are two distinct pathways of the Dsb family of proteins in the periplasm, namely, the DsbA-DsbB oxidative pathway and DsbC-DsbD isomerization/reductive pathway (Cho and Collet, 2013; Nakamoto and Bardwell, 2004). DsbA introduces disulfide bonds into folding proteins via the donation of the energetically unstable disulfide bond of DsbA, which in turn is reoxidized by the action of the membrane protein DsbB, a ubiquinone reductase (Hatahet et al., 2014). The crystal structure of DsbA from Escherichia coli (EcDsbA) revealed a monomeric TRX-like fold domain with an inserted helical domain (Cho and Collet, 2013; Martin et al., 1993). EcDsbA consists of an unstable disulfide at the active site CXXC motif which is characterized by a low pKa of the first nucleophilic cysteine residue in the CXXC motif and a high redox potential (Huber-Wunderlich and Glockshuber, 1998). In contrast, the V-shaped dimeric protein DsbC isomerizes or reduces incorrect disulfides in misfolded substrate proteins through the action of the reactive thiol motif (usually CXYC motif), which is kept in the reduced state by the membrane protein DsbD (Goldstone et al., 2001). DsbA has been considered a virulence factor as it catalyzes the folding of a number of bacterial virulence proteins (Heras et al., 2009).

Gram-positive bacteria do not have a conventional periplasm, but they do have a counterpart compartment confined between the plasma membrane and the cell wall (Matias and Beveridge, 2006). DsbA homologue proteins have been identified and characterized in some Gram-positive bacteria, and appear to be anchored to the outer leaflet of the plasma membrane (Kouwen et al., 2007). On the other hand, no DsbC homologues have been identified in Gram-positive bacteria. Staphylococcus aureus, a human pathogen, encodes only a DsbA-like protein (SaDsbA) and lacks other Dsb proteins (Dumoulin et al., 2005), whereas Bacillus substilis have DsbB-like proteins (BdbC) as well as a DsbA-like protein (BdbD) (Dorenbos et al., 2002; Erlendsson and Hederstedt, 2002; Meima et al., 2002; Sarvas et al., 2004; Stein, 2005). The crystal structures of SaDsbA and BdbD are similar to that of EcDsbA, despite low sequence identities (17%, 11%, respectively) (Crow et al., 2009; Heras et al., 2008). However a tyrosine at the CXXC motif (resulting in CXYC) and Thr preceding the cis-Pro motif present in SaDsbA, are more commonly found in DsbC homologues. Similar to EcDsbA, BdbD has an unstable disulfide bond in the CXXC motif, which is reformed through the action of BdbC, Conversely, the catalytic disulfide bond in SaDsbA is relatively stable (Crow et al., 2009; Heras et al., 2008).

The host defense system produces hydrogen peroxide and superoxide anion via NADPH oxidase in the phagosomes of macrophages (Babior, 1999; Haraga et al., 2008). Recent studies have highlighted the importance of copper ions in the killing of invading pathogens in the phagosomes of activated macrophages and neutrophils (White et al., 2009). Macrophage phagosomes accumulate copper ion during bacterial infections, and these copper ions mediate the conversion of hydrogen peroxide to hydroxyl radicals via the Fenton reaction (Achard et al., 2012). However, some intracellular bacteria can counteract this copper-mediated immune response by up-regulating copper export and detoxifying genes. Furthermore, these activities have been suggested to be important determinants of pathogen virulence (Hodgkinson and Petris, 2012; Wolschendorf et al., 2011). The periplasmic copper binding protein CueP was initially found in Salmonella enterica serovar Typhimurium, and has been implicated in bacterial pathogenesis because of its ability to scavenge ROS both directly and through periplasmic Cu, Znsuperoxide dismutase (Osman et al., 2013; Yoon et al., 2013; 2014b). The crystal structure and subsequent biochemical studies suggested that the cysteine-rich cluster, consisting of three conserved cysteine residues, plays an important role in the function of CueP, and needs to be kept reduced in order to function properly (Yoon et al., 2013; 2014b). It was further reported that DsbC maintains CueP in an active form by reducing the disulfide bonds of conserved cysteine residues (Yoon et al., 2014a).

Corynebacterium diphtheriae is a Gram-positive bacterium that evolved to survive in macrophage phagosomes and can cause diphtheria. In this study, we identified genes encoding a DsbA homologous protein (CdDsbA) and a CueP homologous protein (CdCueP) in the C. diphtheriae genome and determined the crystal structure of CdDsbA at the reduced state as well as the functional relationship between the two proteins.

Cloning, expression, and protein production

A DNA fragment encoding a truncated forms (residues 32?289) of CdDsbA (lacking the predicted N-terminal transmembrane helix) was amplified by PCR using the genomic DNA of C. diphtheriae as a template (Supplementary Table S1). The PCR products were ligated into cut pProEX-HTA vector (Invitrogen, USA) and transformed into E. coli XL1-blue cell. DNA cloning, protein expression, and protein purification of S. enterica serovar Typhimurium CueP (residues 22?179 based on the numbering of the precursor CueP) have been previously described (Yun et al., 2011). Briefly, CdDsbA containing a N-terminal hexahistidine tag was expressed in E. coli BL21 (DE3) cells cultured in 1 L LB media at 37°C and induced at OD₆₀₀ = 0.7 by adding IPTG (0.5 mM final concentration). After incubation for 5 h at 30°C, cells were harvested and resuspended in lysis buffer containing 20 mM Tris (pH 8.0), 150 mM NaCl, and 2 mM 2-mercaptoethanol. The resuspended lysate was disrupted by sonication, and cell debris was removed by centrifugation. The supernatant was mixed with Ni-NTA affinity resin (Qiagen, USA) pre-incubated with lysis buffer, and the resulting mixture was stirred for 1.5 h at 4°C. After the slurry was loaded onto the column, unbound proteins were washed off with lysis buffer supplemented with 20 mM imidazole. Recombinant proteins were eluted with 30 ml of lysis buffer supplemented with 250 mM imidazole. Fractions containing CdDsbA were pooled, and the hexahistidine tag was cleaved with recombinant TEV protease by incubation at room temperature overnight. CdDsbA protein was purified using HiLoad Superdex 200 columns (GE Healthcare, USA) pre-equilibrated with lysis buffer. The purified protein was concentrated to 15 mg/ml in 20 mM Tris buffer (pH 8.0) containing 150 mM NaCl and 2 mM 2-mercaptoethanol and then crystalized. For CdDsbA^Δ32?95 (residues 96?289), a DNA fragment encoding the corresponding region was amplified by PCR using the resulting plasmid for CdDsbA (residues 32?289) as a template (Supplementary Table S1). The same procedures as used for the CdDsbA (residues 32?289) protein were applied to express and purify the protein.

To obtain selenomethionine-labeled CdDsbA, E. coli strain B834 (DE3) cells expressing the recombinant plasmid were cultured overnight in adapted M9 medium, which is containing 47 mM Na₂HPO₄, 22 mM KH₂PO₄, 18.7 mM NH₄Cl, 8 mM NaCl and 30 mg/ml L-selenomethinone (TCI, Japan). The cells were collected by centrifugation (Supra 22K; Hanil Centrifuges, Korea). The pellet was resuspended in a lysis buffer containing 20 mM Tris (pH 8.0), 150 mM NaCl, and 2 mM 2-mercaptoethanol. Gel-filtration chromatography using HiLoad Superdex 200 columns (GE Healthcare, USA) was subsequently performed. The purity of the protein samples was examined by SDS-PAGE.

Crystallization, data collection, and structure determination of selenomethionine-labeled CdDsbA

Selenomethionine-derivatized crystals of CdDsbA were obtained in a precipitation solution containing 0.2 M sodium acetate, 0.1 M Tris-HCl pH 8.5, and 30% PEG 2K by mixing 1 μl of protein solution with 1 μl of precipitant solution. X-ray diffraction data of the crystals were collected at the 5C beamline of the Pohang Accelerator Laboratory (PAL, Korea) using a QUANTUM 315 CCD detector (ADSC). The diffraction data sets were handled and scaled with the HKL2000 package (Otwinosky and Minor, 1997). The crystal belonged to the space group P2₁2₁2₁ with cell measurements of a = 48.6 ?, b = 68.2 ?, and c = 69.0 ?. Initial phases were determined by the single-wavelength anomalous dispersion (SAD) method (Table 1). Three selenium sites were found using the program AUTOSOL in the PHENIX program suite (Adams et al., 2002), and we obtained the excellent electron density map using the AUTOSOL SAD phasing at 1.8 ? resolution.

Crystallization, data collection, and structure determination of CdDsbA

CdDsbA protein (15 mg/ml) was crystallized by the same method as selenomethionine-derivatized CdDsbA. The crystal belonged to the space group P2₁2₁2₁ with cell dimensions of a = 48.8 ?, b = 68.4 ?, c = 69.1 ?, which was similar to that of selenomethionine-derivatized CdDsbA. The structure was solved using coordinates obtained from the selenomethionine-derivatized CdDsbA structure (Table 1).

Disulfide isomerase activity assay

S. enterica serovar Typhimurium DsbC (StDsbC) was purified described previously (Jiao et al., 2013). Bovine pancreatic RNase A (Sigma-Aldrich, USA) was reduced and denatured in 7 M guanidium chloride, 40 mM Hepes (pH 7.5), 120 mM DTT, and 0.2 mM EDTA at 25°C for 24 h, and buffer-exchanged into 0.1% acetic acid by dialysis. The thiol content of reduced DsbC and DsbA proteins and the denatured and reduced RNase A (drRNase) was measured with DTNB as described (Jiao et al., 2013), and they were found to be more than 90% reduced. Proteins were frozen and stored at ?70°C until use. To prepared RNase A with scrambled disulfide bonds (scRNase), the drRNase was oxidized with copper/H₂O₂; drRNase was diluted into 50 mM sodium phosphate buffer (pH 6.0) containing 300 mM NaCl to a final concentration of 50 μM in the presence of 50 μM CuCl₂ and 2 mM H₂O₂, or the buffer alone and incubated at 25°C for 30 min. CuCl₂ and H₂O₂ were removed using a desalting column (HiPrep 26/10 Dsealting; GE Healthcare) in 50 mM sodium phosphate buffer (pH 7.5) containing 300 mM NaCl. To test the disulfide isomerase activities of DsbC, DsbA, or 2-mercaptorethanol, scRNase (5 μM) was incubated with the reduced forms of DsbC (20 μM), DsbA (20 μM), or 2-mercaptorethanol (20 μM) at the room temperature. We performed an RNase A activity assay based on cCMP as a substrate (Hiniker et al., 2005) from 200 μl aliquots of the samples which were taken in given times after the incubation. The aliquot (200 μl) was added to 4 mM cCMP (final concentration) in 400 μl of 300 mM NaCl, 50 mM sodium phosphate (pH 7.5) to measure the RNase A activity. The hydrolysis of cCMP was followed at 296 nm for 30 s. The protein concentrations were determined by the specific absorbance at 280 nm.

Reductase assay of CdDsbA using NTB-labeled CdCueP

S. enterica serovar Typhimurium CueP (StCueP) was purified described previously (Yun et al., 2011). CdCueP was cloned, expressed, and purified using the similar procedures as used for StCueP. Briefly, a DNA fragment of CdCueP (residues 29?211), amplified from a genomic DNA of C. diphtheriae, was ligated into the BamHI and HindIII sites of pProEX-HTA vector (Invitrogen, USA) (Supplementary Table S1). To prepare the reduced forms of CueP proteins and CdDsbA proteins in a reducing agent-free buffer, purified protein samples were dialyzed against degassed 20 mM Tris buffer (pH 8.0) containing 150 mM NaCl twice for 2.5 h, and then immediately concentrated using a Centricon device (GE Healthcare, USA). To prepare NTB-labeled proteins, CueP proteins (20 μM, 5 ml) were first incubated with a 100-fold excess (2 mM) of DTT for 1 h at room temperature. Excess DTT was then removed using a HiPrep 26/10 desalting column (GE Healthcare) pre-equilibrated with 20 mM Tris HCl (pH 8.0) and 150 mM NaCl. Proteins were then incubated with 4 mM of the Ellman’s reagent (DTNB) for 2 h at room temperature to produce 2-nitro-5-thiobenzoic acid (NTB)-labeled CueP proteins. Unreacted DNTB and free thiobis-(2-ntirobenzoic acid) or TNB were removed using HiPrep 26/10 desalting columns with 20 mM Tris?HCl (pH 8.0) and 150 mM NaCl. Upon reduction of the mixed disulfide by CdDsbA or DTT, yellowish TNB was released. In the reductase assay, the NTB-labeled CueP or CdDsbA proteins (5.0 μM) were reacted with reduced CdDsbA protein or CdCueP, or DTT (15 μM) at room temperature. Absorbance of liberated TNB was measured at 412 nm.

Differential scanning calorimetry (DSC)

DSC measurements were obtained using a VP-DSC Micro-calorimeter (Microcal, USA) with 0.51471 cm³ twin cells for the reference and sample solutions. Prior to DSC measurements, the sample and reference were degassed under vacuum while being stirred. Next, 20 mM Tris-HCl (pH 8.0) containing 150 mM NaCl and 2 mM 2-mercaptoethanol prepared for dialysis were used as references to obtain baselines. Measurements were repeated three times at a scan rate of 1°C/min for each sample. The temperature ranged from 15 to 110°C and the final thermogram was processed by subtracting the baseline from the sample thermogram. Thermal compensation curves were evaluated using the Microcal LLC DSC plug-in the Origin 7.0 software package provided with the DSC instrument. Transition melting points were calculated using the same software package.

Structural determination

It was previously reported that periplasmic isomerase DsbC plays a key role in maintaining CueP protein in a reduced state in S. enterica (Yoon et al., 2014a). C. dipththeriae does not have a DsbC homologue in the genome like other Gram-positive bacteria (Kouwen et al., 2007). We identified a gene that encodes a DsbA homologous protein (CdDsbA) in the C. diphtheriae genome as a putative functional substitute for DsbC.

DsbA from C. diphtheriae encodes a protein of 289 amino acid residues with a putative N-terminal signal peptide (residues 1?21) for export out of the cytoplasmic membrane, which was predicted by SignalP (Petersen et al., 2011). The transmembrane prediction program TMHMM suggested that residues 12?31 were located in the transmembrane region, while residues 32 to the C-terminus were located on the outer plasma membrane (Chen et al., 2003). Moreover, BdbD, a functional and structural homologue of DsbA, was shown to be localized outside of the cytoplasmic membrane by its N-terminal transmembrane segment (Crow et al., 2009). When we expressed a truncated protein lacking the predicted N-terminal signal peptide and transmembrane segment, the protein was produced in a soluble form. Taken together, CdDsbA is likely anchored to the cytoplasmic membrane via its N-terminal transmembrane segment similar to BdbD, although this remains to be validated experimentally.

We obtained crystals of the truncated form of CdDsbA (residues 32?289) in the reduced condition. The crystal structure was determined using the SAD method with selenomethionine-labeled crystals (Rice et al., 2000). A high-quality electron density map was produced, and the majority of the model could be traced. The crystals belonged to the P2₁2₁2₁ space group with a single molecule in the asymmetric unit (Table 1). The final model was refined against the 1.5 ? resolution native dataset of both crystals in reduced form, and contained all residues between 38?287. The structure was refined to a free R value of 19.5% with good stereochemistry. Further details on the structure determination and refinement are given in Table 1.

Overall structure

CdDsbA exhibited a one-domain structure with extended loops at both termini (Fig. 1A). The core DsbA-like domain of CdDsbA (residues 115?277) could be divided into two regions: a classical TRX-fold region (residues 115?156 and residues 229?277; colored in green), and an α-helical region (residues 157?228; colored in salmon). The first part (residues 115?156) of the classical TRX-fold region formed a βαβ component and the second part (residues 229?277) formed αββα components. The first and second parts of the TRX-fold region were interrupted by an α-helical region (residues 157?228) forming four α-helices. The helix α7 served as an extension of the helix α8 in the second part of the TRX-fold region (Fig. 1A). The N-terminal extended region (residues 32?114) consisted mainly of loops with two short α-helices (α1 and α2) attached to the core DsbA-like domain via a few polar interactions between the backbone and side chains or backbone. The side chain of Arg172 from α5 in the core domain formed a polar interaction with the backbone carbonyl groups of Ala80 and Ile82 from the N-terminal region. Likewise, the backbone NH group of Ala176 in the core domain formed a hydrogen bond with the backbone carbonyl group of Phe78 in the N-terminal region (Fig. 1B). In addition, the C-terminal stretched region (residues 278?289; Fig. 1A colored in red) protruded from the core domain.

Comparison with other DsbA family proteins

We performed a sequence alignment with the DsbA family proteins, including the structurally and biochemically characterized EcDsbA, BdbD, and SaDsbA proteins (Crow et al., 2009; Heras et al., 2008; Martin et al., 1993) (Fig. 2A). We found that the N-terminal extended region (residues 32?114; colored in cyan) and the C-terminal stretched region (residues 278?289) were unique to CdDsbA. In particular, the N-terminal extended region was unusually long and its interactions with the core DsbA-like domain were not extensive. The homologous and core domain (residues 115?277; Fig. 2A) of the CdDsbA structure was superposed onto other DsbA family proteins, yielding rmsd values for BdbD, SaDsbA, and EcDsbA of 1.2, 2.3, and 4.8 ?, respectively. The structural superposition revealed a notable conformational difference in the loop connecting α-helices α5 and α6 (residues 190?202; Figs. 2A and 2B; blue box). Specifically, the conformations of this connecting loop were divergent among DsbA family proteins (Figs. 2B and 2C). The connecting loop of CdDsbA was located in closest proximity to the CXXC motif (Fig. 2C).

In the case of BdbD, a calcium binding site was identified that appeared to be involved in modulation of the oxidative properties of the active site cysteine residues in a calcium dependent manner (Crow et al., 2009). However, no such calcium binding site was found in CdDsbA.

Structural features of the CdDsbA active site

The cysteine residues in the active site CXXC motif were present in the reduced form, as expected (Fig. 3A). The CXXC motif was located at the third α-helix (α3), and was found to be long and highly curved like other DsbA proteins (Shouldice et al., 2011). This helical curve seems to be caused by a proline residue (Pro138) in the middle of the helix. The sequence of the CXXC motif in CdDsbA was ¹²⁶CPFC¹²⁹ (Fig. 2A). Interestingly, the second residue Pro127 in the CXXC motif was conserved in BdbD, SaDsbA, and EcDsbA (CPSC in BdbD, CPYC in SaDsbA, and CPHC in EcDsbA). It was unusual that the hydrophobic residue Phe128 in the CdDsbA CXXC motif was exposed to solvent unlike other DsbA proteins (Fig. 3B). The phenylalanine residue does not block the active site cysteine residues in the crystal structure. We speculate that Phe128 might play a role in recognition of the hydrophobic region on the surface of the substrates.

Like other TRX superfamily proteins, a conserved cis-Pro loop containing Pro249 was found in close proximity to the nucleophilic Cys residue (Cys126) (Fig. 3). It has been suggested that the residue preceding the cis-Pro loop is related to substrate specificity (Huber-Wunderlich and Glockshuber, 1998), and also that the residue preceding the cis-Pro is specifically involved in modulating the redox properties of TRX-like proteins. CdDsbA, SaDsbA, and BdbD all contain the Thr-cis-Pro motif, whereas EcDsbA has a Val-cis-Pro motif (Ren et al., 2009). Taken together, our findings suggest that CdDsbA is more functionally related to SaDsbA and BdbD than EcDsbA.

It was reported that the thiolate forms at the nucleophilic cysteine residues (Cys30 in the EcDsbA numbering) are stabilized in reduced forms of EcDsbA and Vibrio cholera DsbA (Guddat et al., 1998; Mohanty et al., 2012). The stabilization of the thiolate forms were suggested to result from the interactions within the CXXC moitf (Guddat et al., 1998). Thiol group and the backbone amide group of the resolving cysteine residue (Cys32 in the EcDsbA numbering), and the imidazole ring and the backbone amide groups of the third residue (His31 in EcDsbA) in this motif form hydrogen bonds with Sγ of the nucleophilic cysteine residue. In the CdDsbA structure, the similar interactions within the CXXC motifs were found except for the imidazole ring-mediated interaction because His31 in EcDsbA is replaced with phenylalanine residue (Phe128) in CdDsbA (Fig. 3). Cys126 of CdDsbA makes five hydrogen bonds with adjacent backbone atoms (Fig. 3B). Additional polar interactions with the Thr248 in the Thr-cis-Pro motif were found in CdDsbA. The sidechain hydroxyl group and backbone carbonyl group of Thr248 make hydrogen bonds with Sγ of Cys126 (Fig. 3A). From these interactions, we could speculate that these extensive polar interactions might be involved in stabilization of Cys126 thiolate form in CdDsbA like EcDsbA.

CdDsbA plays a role in the isomerization/reductive pathway similar to DsbC in Gram-negative bacteria

It was previously reported that SaDsbA functions as a disulfide bond isomerase/reductase like DsbC in Gram-negative bacteria (Heras et al., 2008). To investigate the activity of CdDsbA, we performed a disulfide bond isomerase/reductase assay using denatured RNase A with scrambled disulfide bonds as a substrate. CdDsbA mediated the reductive refolding of the denatured RNase A via its disulfide bond isomerase/reductase activity, which was validated with DsbC protein from S. enterica serovar Typhimurium and 2-mercaptoethanol employed as controls (Fig. 4A and Supplementary Fig. S1). Thus, our observation indicates that CdDsbA is involved in isomerization/reductive pathways like SaDsbA and DsbC of Gram-negative bacteria.

To investigate the role of the N-terminal extended region, we produced a variant protein whose N-terminal extended region was mostly deleted. The expression level of the mutant protein was very high, and the protein exhibited comparable activity to that of the intact soluble domain (residues 32?289) of CdDsbA (Fig. 4A and Supplementary Fig. S1). These results demonstrated that the N-terminal extended region is not essential for the overall folding of the core DsbA-like domain and the isomerase/reductase activity of CdDsbA. We next carried out differential scanning calorimetry (DSC) to evaluate the role of the N-terminal region on the structural stability of the proteins. Specifically, wild-type CdDsbA and the N-terminal region deleted variant proteins were applied to DSC. Figure 4B shows the T_m values of the proteins (wild-type protein: 43.2°C; deletion variant: 40.7°C). The decreased T_m of 2.5°C of the deletion variant versus the wild-type protein indicated that the N-terminal extended region partly contributed to the structural stability of the core DsbA-like domain.

CdCueP is a putative substrate of CdDsbA

In the Gram-negative bacteria S. enterica serovar Typhimurium, the reduced form of CueP scavenges the production of hydroxyl radicals in the copper-mediated Fenton reaction, and DsbC is responsible for keeping CueP in the reduced, active form (Yoon et al., 2013; 2014a; 2014b). Given the disulfide bond isomerase/reductase activity of CdDsbA, we investigated whether CdDsbA shares a similar function with respect to reduction of CdCueP (accession number WP_003849917), which was discovered by the sequence search program BLAST (Altschul et al., 1990) (Supplementary Fig. S2). Specifically, we tested the activity of CdDsbA using NTB-labeled CdCueP as a substrate. The NTB group was attached to the cysteine residues of CdCueP using Ellman’s reagent (the disulfide-containing agent 5,5′-dithiobis-(2-nitrobenzoic acid) or DTNB), which can selectively react with the free cysteine (Yoon et al., 2014a). When freshly-prepared CdDsbA protein was incubated with NTB-labeled CdCueP, its reducing activity on CdCueP was as efficient as that of the strong reducing agent DTT (Fig. 5). In contrast, CdDsbA failed to reduce the NTB-labeled S. enterica CueP, which indicated that CdDsbA recognized CdCueP via a specific protein-protein interaction. We next tested whether the reduced form of CdCueP can reduce CdDsbA using the NTB-labeled CdDsbA and the reduced form of CdCueP. Any significant reaction was not observed, indicating that CdDsbA is more reductive than CdCueP (Fig. 5). Thus, our findings suggest that DsbA is a putative reductant of CueP in the Gram-positive bacterium C. diphtheriae.

In this study, we identified genes encoding a DsbA homologous protein (CdDsbA) and a CueP homologous protein (CdCueP) in the C. diphtheriae genome and determined the crystal structure of CdDsbA as well as the functional relationship between the two proteins. Despite structural differences, CdDsbA exhibited disulfide bond isomerase/reductase activity similar to that of DsbC from Gram-negative bacteria. Notably, a unique long N-terminal extended region was found in CdDsbA, and the region was dispensable for the activity of the core DsbA-like domain. More importantly, this study showed that CdDsbA could reduce CueP of C. diphtheriae, but not of Salmonella, suggesting that CdCueP is a substrate of DsbA in C. diphtheria. However, further study is needed to know whether CdCueP is a physiological substrate of CdDsbA.

We noted that the long N-terminal extended region of CdDsbA did not form notable hydrophobic interactions with the core DsbA-like domain. Without the N-terminal region, the protein could be expressed in a soluble form with comparable enzymatic activity. Furthermore, the N-terminal region only partly contributed to the structural stability of the protein. Thus, it is also expected that the N-terminal extended region may be easily unfolded since it does not have an apparent secondary structure. This observation led us to ask what role the N-terminal extended region might play. We speculated that the N-terminal region may stretch to a long string (over 300 ?) from the anchored plasma membrane if the region is unfolded. Since DsbA is a chaperone protein that helps correct the folding of a diverse set of proteins, this long string from the anchoring membrane may provide a significant advantage in reaching substrates on the membrane.

How can CdDsbA maintain the reductase activity like DsbC of Gram-negative bacteria? To answer this question, we noted Etrx proteins from Streptococcus pneumoniae, which are extracellular surface thioredoxin-fold lipoprotein and functional orthologues of CdDsbaA. The Etrx proteins maintain methionine sulfoxide reductases with their reducing power, which plays an important role in extracellular ROS stress resistance of S. pneumoniae (Saleh et al., 2013). Moreover, the Etrx proteins is in turn kept reduced by the membrane protein CcdA, which has a low sequence similarity to dsbD that supplies the reducing power to DsbC in Gram-negative bacteria (Saleh et al., 2013). The sequence analysis of C. diphtheriae revealed a putative CcdA-like protein (31% sequence identity), suggesting that CdDsbA is kept in the reduced state by the CcdA-like proteins.

In S. enterica servar Typhimurium, DsbC and CueP are co-localized in the periplasm, and can make cooperative interactions in this confined space to scavenge the generation of hydroxyl radicals from H₂O₂ in the presence of copper ion (Yoon et al., 2014b). Since the Gram-positive bacterium C. diphtheriae does not have a conventional periplasmic space due to the lack of an outer membrane, many extracellular proteins are thought to be tethered at the extracellular side of the cytoplasmic membrane in order to function near the cell and make molecular interactions. Consistent with this possibility, the two C. diphtheriae proteins appeared to be tethered at the cytoplasmic membrane by a predicted uncleavable signal peptide or transmembrane helix. Given that both proteins seem to be anchored to the membrane, they are likely able to interact sufficiently to perform their functions. C. diphtheriae can survive in macrophage phagosomes, which is directly related to the pathogenesis of this bacterium. The role of DsbA in reduction of CueP in the extracellular region of C. diphtheriae might be involved in survival in macrophage phagosomes, since CueP is related to the survival of Salmonella in macrophage phagosomes (Yoon et al., 2014b).

In this study, the crystal structure of DsbA from C. diphtheriae was determined, which revealed a core DsbA-like domain and long extended N-terminal region. The core domain of CdDsbA exhibited disulfide bond isomerase/reductase activity, while the N-terminal extended region appeared to serve as a long anchor in the membrane. Furthermore, CdDsbA also showed reducing activity specifically toward CueP from C. diphtheriae.