Staphylococcus aureus DsbA Does Not Have a Destabilizing Disulfide

In Gram-negative bacteria, the introduction of disulfide bonds into folding proteins occurs in the periplasm and is catalyzed by donation of an energetically unstable disulfide from DsbA, which is subsequently re-oxidized through interaction with DsbB. Gram-positive bacteria lack a classic periplasm but nonetheless encode Dsb-like proteins. Staphylococcus aureus encodes just one Dsb protein, a DsbA, and no DsbB. Here we report the crystal structure of S. aureus DsbA (SaDsbA), which incorporates a thioredoxin fold with an inserted helical domain, like its Escherichia coli counterpart EcDsbA, but it lacks the characteristic hydrophobic patch and has a truncated binding groove near the active site. These findings suggest that SaDsbA has a different substrate specificity than EcDsbA. Thermodynamic studies indicate that the oxidized and reduced forms of SaDsbA are energetically equivalent, in contrast to the energetically unstable disulfide form of EcDsbA. Further, the partial complementation of EcDsbA by SaDsbA is independent of EcDsbB and biochemical assays show that SaDsbA does not interact with EcDsbB. The identical stabilities of oxidized and reduced SaDsbA may facilitate direct re-oxidation of the protein by extracellular oxidants, without the need for DsbB.

The formation of native disulfide bonds through air oxidation of cysteine pairs is a slow reaction and organisms ranging from bacteria to humans encode enzymatic systems to catalyze the process. In eukaryotes, oxidative folding in the endoplasmic reticulum is primarily catalyzed by protein-disulfide isomerases that are reoxidized by Ero1p/Erv2p proteins (reviewed in Ref. 1). In Escherichia coli, dithiol oxidation takes place in the periplasm through the action of the Dsb 3 (Disulfide bond) family of proteins. Dsb proteins form two distinct pathways, the DsbA-DsbB (oxidative) pathway that introduces disulfides indiscriminately, and the DsbC/DsbG-DsbD (isomerization) pathway that shuffles incorrect disulfides (2,3).
Probably the best-studied Dsb protein is EcDsbA, the primary disulfide catalyst in E. coli (reviewed in Refs. 2,4). This promiscuously oxidizing protein is a 21-kDa monomer containing a CPHC active site in a thioredoxin (TRX) fold with an inserted helical domain of ϳ80 residues. Upon catalyzing disulfide bond formation in substrate proteins, reduced EcDsbA relies on EcDsbB, a quinone reductase, to recover its catalytically active, and higher energy, oxidized form (5). Orthologs of DsbA are encoded in many Gram-negative bacteria and have been shown to play an essential role in the biogenesis of virulence factors and toxins in pathogenic organisms (6 -12).
Although Gram-positive organisms do not have a conventional periplasm, they contain a compartment confined between the plasma membrane and the outer cell wall (13,14) and genomic analysis indicates that they do encode cell surface anchored Dsb-like proteins (15). Functional orthologs of Dsb proteins in Gram-positive organisms include Bacillus brevis Bdb (a DsbA ortholog) (16), Bacillus subtilis BdbB and BdbC (orthologs of DsbB), B. subtilis BdbD (a DsbA ortholog) (17)(18)(19)(20)(21) and Mycobacterium tuberculosis DsbE (structurally a homolog of DsbE, but functionally an oxidant like DsbA) (22). Moreover, functional virulence factors produced by Gram-positive bacteria, such as the human pathogen Staphylococcus aureus, contain disulfide bonds that are introduced during folding ( [23][24][25]. S. aureus and other pathogens such as Listeria monocytogenes and Enterococcus faecalis belong to a subgroup of Grampositive bacteria that encode only a homolog of DsbA and lack other Dsb proteins (15). S. aureus DsbA (SaDsbA) (26), is a 23-kDa lipoprotein that has low sequence homology with EcDsbA (15% amino acid identity) and lacks residues thought to be important for DsbA function in Gram-negative organisms, including a His in the CXHC active site motif and a Val preceding the cis-Pro (VcP) active site motif. Instead, SaDsbA has residues in these two motifs that are more typical of the Gram-negative disulfide isomerases DsbC and DsbG (CXYC, TcP). Despite these differences, the soluble form of SaDsbA can partially complement EcDsbA in a ⌬dsbA background (26). Moreover, SaDsbA can replace each of the disulfide oxidoreductases of B. subtilis (BdbA-D) that catalyze the oxidative folding of specific extracytoplasmic proteins (15). Despite some ability to complement DsbA function in these organisms, its low sequence identity suggests that SaDsbA has a different purpose, most likely to act on different, as yet unidentified target protein(s). Notably, the activity of SaDsbA in both E. coli and B. subtilis does not depend on re-oxidation by a DsbB-like disulfide oxidoreductase (15,26). Indeed, S. aureus does not encode a DsbB nor any other Dsb-like proteins.
To investigate how SaDsbA catalyzes disulfide bond formation without the need for a partner disulfide shuttle protein, we determined the crystal structure of SaDsbA, representing the first such Gram-positive DsbA structure, and analyzed its biochemical and functional properties. Our findings show that the physicochemical basis for oxidative folding in S. aureus is very different to that underpinning Gram-negative oxidative folding.
Selenomethionine (SeMet)-labeled SaDsbA was expressed from BL21(DE3) pLysS strain in minimal medium containing SeMet (L/D mixture) using methods similar to those described previously (28). SaDsbA variants were generated using a Stratagene Quikchange kit (Stratagene, CA) and verified by DNA sequencing. SeMet SaDsbA and all SaDsbA variants were expressed and purified following the same procedures described for native SaDsbA.
Crystallization and Diffraction Data Measurement-Native and selenomethionine (SeMet)-labeled SaDsbA crystals were obtained from 28 -30% PEG 3350. MAD data were collected at the 8.3.1 beam line at the Advance Light Source (ALS) in Berkeley. Diffraction data were integrated and scaled with HKL2000 (29). Native, SaDsbA E96Q, and SaDsbA T153V data were measured using a Rigaku FR-E copper rotating anode generator operating at 45 kV, 45 mA with Osmic Confocal Max-Flu TM optics (either HiRes 2 or Maxscreen). Reflections were measured with an R-AXIS IV ϩϩ imaging plate area detector (Rigaku Americas). A Cryo Industries CryoCool LN2 was used for cooling the crystals during data measurement. Data were processed using Crystal Clear (Rigaku).
Structure Determination-The structure of SaDsbA was solved by MAD phasing of the SaDsbA-SeMet derivative. The two possible selenium positions were located in the asymmetric unit with the program SOLVE/RESOLVE (30). The resulting phases were used in the program ARP/WARP for automated building of the protein structure (31). The structure was completed by manual building using the programs O (32) and Coot (33).
Refinement was performed using maximum likelihood in CNS (34) on the native 1.81-Å resolution dataset. The final model corresponds to residues 14 -177 of the deduced mature protein. Some surface residues with weak electron density were modeled with reduced occupancies for the side chain atoms (Glu 104 , Lys 133 , Glu 170 , Lys 174 ).
The structure of SaDsbA variants were solved by difference Fourier or molecular replacement (Phaser (35)) methods by using the structure of native SaDsbA. Superposition of molecules was carried out using the LSQ options from the programs O (32) and Coot (33).
Molecular figures were generated using MolScript (36) and PyMOL (37) and figures of the electrostatic potential were generated using GRASP (38).
GdmCl-induced Unfolding Equilibria-For the chemical unfolding experiments, proteins were diluted to 1.5 M in 20 mM Hepes-NaOH, 170 mM NaCl, 0.1 mM EDTA, pH 7.0 buffer containing increasing concentrations of GdmCl and incubated at room temperature for 24 h. The reduced proteins were unfolded under identical conditions but in the presence of 0.75 mM DTT. For the refolding experiments oxidized and reduced protein stock solutions (37.5 M) were unfolded in 20 mM Hepes-NaOH, 170 mM NaCl, 0.1 mM EDTA (pH 7.0), 6 M GdmCl, (containing 20 mM DTT in the case of reduced proteins) for 16 h at room temperature. Then 20 l of these solutions were mixed with 480 l of buffers containing different concentrations of GdmCl and incubated for 24 h at room temperature. Transitions were measured fluorimetrically by the change in the fluorescence at 330 nm using an excitation wavelength of 280 nm. Data were evaluated according to the twostate model of folding using a six parameter fit (39).
Thermal Unfolding-Temperature-induced unfolding curves of native SaDsbA and variants were monitored by the change in the far-UV circular dichroism (CD) signal. Far-UV CD spectra (from 250 to 190 nm) of 10 M protein solutions were recorded at 25 and 95°C on a Jasco J-715 spectropolarimeter. The wavelength with the largest change in signal was determined from the differential spectrum at 25 and 95°C. Thermal unfolding was monitored at 223 nm from 25-90°C with a heating-rate of 1°C/min. All measurements were carried out in 100 mM phosphate buffer, 1 mM EDTA pH 7.0 (in the presence of 0.75 mM DTT for reduced proteins) using a 1-mm quartz cuvette. Transitions were normalized by assuming a linear dependence of the spectroscopic signal of the native and unfolded states on temperature.
Determination of the Equilibrium Constants with Glutathione-The redox equilibrium of SaDsbA and SaDsbA variants was determined as described previously (40). Briefly, 1.5 M protein was incubated for 16 h at 25°C in 100 mM sodium phosphate, 0.1 mM EDTA (pH 7.0) containing 1 mM GSSG and increasing concentrations of GSH. The redox state of the SaDsbA was followed by fluorescence emission at 335 nm. The equilibrium constant K eq and redox potential were determined according to standard thermodynamic equations (41).
Determination of pK a Values-The pH-dependent ionization of the nucleophilic cysteine thiol was followed by the specific absorbance of the thiolate anion at 240 nm (42). As a reference, the pH-dependent absorbance for the oxidized form of the protein was monitored. All measurements were carried out at 25°C in pK a buffer consisting of 10 mM K 2 HPO 4 , 10 mM boric acid, 10 mM sodium succinate, 1 mM EDTA, and 200 mM KCl, pH 7.5 and an average initial protein concentration of 20 M. The oxidized and reduced protein samples were prepared by incubating the proteins with 1.7 mM copper(II)[1,10-phenanthroline] and 10 mM DTT, respectively. The oxidizing and reducing agents were eliminated by gel filtration using a PD10 column equilibrated in pK a buffer. The pH of the protein solution was lowered to 2.2 stepwise by the addition of aliquots of 0.2 M HCl. The absorbance at 240 and 280 nm was recorded on a Cary 50BIO UV-visible spectrophotometer and corrected for the volume increase. The pH dependence of the thiolate-specific absorbance signal (S ϭ (A 240 /A 280 ) reduced / (A 240 /A 280 ) oxidized ) was fitted according to the Henderson-Hasselbalch equation.
Insulin Reduction Assay-The ability of SaDsbA and SaDsbA variants to catalyze insulin reduction in the presence of DTT was determined as previously described (43). Briefly, reaction mixtures were prepared in cuvettes containing 131 M insulin, and various concentrations of oxidized protein catalyst (5-10 M) in 0.1 M phosphate buffer, pH 7.0, 2 mM EDTA. Reactions were started by adding DTT to a final concentration of 0.35 mM. After thorough mixing, the optical density at 650 nm was recorded every 30 s. The non-catalyzed reduction of insulin by DTT was monitored in a control reaction without catalyst.
Ubiquinone Reduction Assay-To determine the EcDsbB-catalyzed oxidation of SaDsbA by ubiquinone Q1 in vitro, 15 M reduced SaDsbA in 50 mM sodium phosphate, 300 mM NaCl, 0.1% (w/v) n-dodecyl-D-maltoside, pH 6.0, were incubated at 30°C with 15 M ubiquinone-1 (Q1; Sigma) (44). The reaction was started by the addition of DsbB to a final concentration of 0.1 M and the reduction of Q1 was monitored through the decrease in absorbance at 275 nm. EcDsbA was used as a positive control.  FEBRUARY 15, 2008 • VOLUME 283 • NUMBER 7

Structural and Biochemical Characterization of DsbA from S. aureus
Refolding of Scrambled RNase A-The in vitro assay of refolding of scrambled RNase was used to monitor the isomerase activity of SaDsbA (45). scRNase was prepared from reduced denatured RNase by incubating the enzyme (0.5 mg/ml) in 50 mM Tris-HCl, pH 8.5 and 6 M GdmCl for at least 3 days at room temperature and in the dark. After acidifying the solution as previously described, the fully oxidized state of RNase A was confirmed by Ellman's assay. Reshuffling of scRNase (40 M) was carried out by incubation in 100 mM phosphoric acid-NaOH, pH 7.0, 1 mM EDTA, with 10 M oxidized DsbC or oxidized SaDsbA. The reactions were started by addition of DTT to a final concentration of 10 M. As positive and negative controls we carried out two additional reactions using folded RNase and scRNase, respectively, and without any Dsb protein.
The assay was performed at 25°C and at several time points samples were withdrawn and assayed for RNase A activity by monitoring cCMP hydrolysis.
Coordinates and structure factors for SaDsbA, SaDsbA E96Q, and SaDsbA T153V have been deposited in the Protein Data Bank (PDB ID codes 3BCI, 3BD2, and 3BCK, respectively).

RESULTS
Structure of SaDsbA-The crystal structure of soluble SaDsbA (Fig. 1A), the deduced mature protein, was determined by multiwavelength anomalous diffraction methods (46,47) and refined at 1.81-Å resolution to R-factor and R-free values of 19.8 and 21.9%, respectively ( Table 1). The structure of SaDsbA reveals a fold similar to that of EcDsbA comprising a thioredoxin (TRX) domain and an embedded helical domain (Fig.  1A). The classic TRX fold is composed of residues 17-58 forming the ␤␣␤ motif and residues 155-181 forming the ␤␤␣ motif with the connecting helix formed by residues 142-154. The 82-residue insertion in the TRX fold of SaDsbA (residues 59 -141) forms an antiparallel three-heli-cal bundle (␣2-␣4), an additional helix (␣5) and an extension to helix ␣6. As in all redox-active TRX-like proteins, the TRX fold of SaDsbA includes a CXXC motif (Cys 26 -Pro 27 -Tyr 28 -Cys 29 ) located at the N terminus of the first helix in the TRX fold (Fig. 1A).
Comparison with E. coli DsbA-The overall fold of SaDsbA resembles that of EcDsbA; the C␣ atoms of 110 residues can be superimposed with an r.m.s.d. of 1.9 Å, and the sequence identity is 15%. Despite the structural similarity, the two proteins have critical differences in regions surrounding the catalytic CXXC motif that could impact on substrate specificity and activity. These regions include the ␤3␤4␣7 (␤␤␣) motif of the TRX fold (Fig. 1B) and loops connecting helices in the inserted helical domain, primarily the loop between ␣3 and ␣4 (Fig. 1C). The ␤3␤4␣7 motif at the C-terminal end of the polypeptide chain produces a dramatic difference between SaDsbA and EcDsbA (Fig. 1B). This region forms one edge of a hydrophobic peptide-binding groove in EcDsbA (48) that binds to DsbB (49) and is proposed to interact with unfolded protein substrates. In both proteins this region is hydrophobic, however in SaDsbA it is 9 residues shorter, resulting in a shorter loop connecting ␤4 and ␣7 and a shorter ␣7 helix (by two turns) (Fig. 1B). Based on sequence alignment, this deletion appears to be conserved in DsbAs from Gram-positive organisms and, to a lesser extent in some Gram-negative DsbA proteins such as TcpG from Vibrio cholerae ( Fig. 2A).
Another notable difference between SaDsbA and EcDsbA occurs in the loop connecting helices ␣3 and ␣4. Compared with EcDsbA, this loop has an extra residue in SaDsbA and is oriented toward the CXXC active site, positioning the acidic side chain of Glu 96 near to the nucleophilic Cys 26 (6.9 Å) (Fig.  1C). Sequence alignment shows that acidic residues are highly abundant in this loop in Gram-positive DsbAs but where F o and F c are the observed and calculated structure-factor amplitudes for each reflection "h". e R free was calculated with 10% of the diffraction data selected randomly and excluded from refinement.  FEBRUARY 15, 2008 • VOLUME 283 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4265 absent in E. coli DsbA (Fig. 2A). The proximity of Glu 96 to the redox active site of SaDsbA (Fig.  2B) suggests a possible role in substrate specificity or in modulating the redox characteristics.

Structural and Biochemical Characterization of DsbA from S. aureus
A cis-Pro loop is conserved among redox-active TRX-fold containing proteins and has been shown to be involved in substrate recognition (50 -52). SaDsbA contains a Thr preceding the cis-Pro residue (TcP) a motif that is highly conserved in Gram-negative disulfide isomerases (DsbC and DsbG) but not in Gram-negative oxidases like EcDsbA (which generally have VcP). Superposition of SaDsbA and EcDsbA structures reveals a shift in the position of the ␣c loop that precedes the cis-Pro loop (Fig. 1B). Similar changes in this loop are associated with the different substrate specificities of the disulfide isomerases DsbC and DsbG (53).
Surface Features-EcDsbA has a hydrophobic patch and a peptidebinding groove incorporating a hydrophobic pocket (Fig. 3A) that are thought to be important for activity (48). By contrast, the electrostatic surface of SaDsbA shows that it lacks these features. First, the hydrophobic patch adjacent to the active site in EcDsbA is absent in SaDsbA (Fig. 3A). Hydrophobic residues that contribute to this patch in EcDsbA (Phe 29 preceding the CXXC active site, Phe 63 -Met 64 -Gly 65 -Gly 66 in the loop connecting the TRX and helical domains) are replaced or covered with charged residues in SaDsbA (supplemental Fig. S1). Thus, a basic rather than a hydrophobic residue (Lys 25 ) precedes the CXXC active site and, although residues of the interdomain loop (Lys 62 -Asp 63 ) are mainly hydrophobic, the different conformation of the ␣3-␣4 loop in SaDsbA results in charged residues (Lys 98 , Glu 99 ) covering this area which flanks the active site. Moreover, the different position of the ␣3-␣4 loop in SaDsbA also results in an acidic protrusion (generated by E96) near the redox active site ( Fig. 3A and supplemental Fig. S1).
The 9-residue deletion at the C terminus of SaDsbA affects the surface properties in that the peptide binding groove is truncated compared with EcDsbA (Fig. 3A). This groove is used by EcDsbA to interact with its partner protein DsbB (49) and has been proposed to be important for substrate binding interactions (54). The altered groove and the lack of a hydrophobic patch in SaDsbA suggest a differing substrate specificity to EcDsbA and indicates that SaDsbA may be unable to interact with DsbB-like quinone oxidoreductases. Indeed, when we modeled the DsbB periplasmic loop of the EcDsbA:DsbB complex (49) into the structure of SaDsbA, steric clashes are predicted as a consequence of the truncated peptide binding groove (Fig. 3B). For example, Tyr 167 in the ␤4-␣7 loop in SaDsbA clashes with residues Pro 100 and Phe 101 in DsbB (Fig.  3B). Furthermore, when we tested in vitro the ability of DsbB to oxidize SaDsbA by monitoring EcDsbB-catalyzed ubiquinone reduction (44), we found that EcDsbB was able to oxidize EcDsbA in a ubiquinone-dependent manner, as expected, but we could not detect SaDsbA oxidation (Fig. 3C). These results are in agreement with a novel mechanism of DsbA oxidation in S. aureus differing from that performed by DsbB in E. coli.
SaDsbA Variants-Two of the most striking sequence and structural differences between SaDsbA and EcDsbA localize to the redox active site; namely, the acidic residue Glu 96 close to the CXXC motif and the TcP (rather than VcP) in the cis-Pro loop. We therefore generated two variants of SaDsbA, E96Q (to remove the acidic residue) and T153V (to generate the same cis-Pro sequence as in EcDsbA) for further studies. Their structures were solved showing that the mutations did not affect the overall structure. Thus, the wt and variant SaDsbA structures can be superposed with an average r.m.s.d. of 0.14 Å and 0.15 Å for T153V and E96Q, respectively, and the two variants can be superposed with an average r.m.s.d. of 0.16 Å.
SaDsbA Does Not Have a Destabilizing Disulfide-Although most protein disulfides stabilize their protein fold, the catalytic disulfide of EcDsbA is destabilizing (55,56) This unusual characteristic drives the EcDsbA catalyzed thiol-disulfide exchange reaction in substrate proteins, and is reflected in the highly oxidizing nature of EcDsbA. This characteristic requires that EcDsbA is re-oxidized by DsbB to the more unfavorable, but catalytically active, oxidized form. We determined the thermodynamic stabilities (⌬G stab ) of oxidized and reduced SaDsbA and SaDsbA variants with equilibrium unfolding/refolding experiments induced by guanidinium chloride. The transitions were fully reversible for all three SaDsbA proteins (Fig. 4A) and the unfolding curves were all highly cooperative and fitted a two state model (57). Surprisingly, the data showed that the oxidized and reduced forms of all three SaDsbA proteins have similar stabilities (Table 2). This was confirmed by temperature-induced unfolding experiments (Fig. 4B). For these exper-iments, EcDsbA was used as a control and the unfolding curves confirmed previous findings that reduced EcDsbA is more stable than oxidized EcDsbA (Table 2). However, oxidized and reduced SaDsbA were found to have similar stabilities (Table 2).
Redox Potential of the Active Site Disulfide-Previous work has shown that SaDsbA (EЈ 0 ϭ Ϫ131 mV (26) is not as strong an oxidant as EcDsbA (EЈ 0 ϭ Ϫ122 mV (40)). We measured the redox potential for the two SaDsbA variants, using wt SaDsbA (26) as a control (Fig. 4C). Mutation of the Thr in the cis-Pro loop of SaDsbA to Val (T153V, the residue present in EcDsbA), resulted in a redox potential similar to SaDsbA. However, the E96Q mutant yielded a redox potential a little more oxidizing than that of EcDsbA ( Fig. 4C and Table 2B).
pK a of the Nucleophilic Cysteine-An important characteristic of thiol-disulfide oxidoreductases is the lowered pK a of the reactive cysteine, which determines reactivity in thiol-disulfide exchange reactions. Thus, the nucleophilic cysteine of EcDsbA has a pK a of 3.3, much lower than the typical value for cysteine residues of ϳ8.5 (42). We measured the pK a of Cys 26 , the nucleophilic cysteine of SaDsbA (Fig. 4D), and found it to be 3.37, similar to that of EcDsbA. The E96Q variant had a slightly lower pK a of 3.09. However, the pK a for the SaDsbA T153V variant was dramatically different with a value of 5.21. To confirm these pK a values we also measured the pH dependence of the nucleophilic cysteine reactivity with iodoacetamide (IAM) (42). Protein samples were incubated with IAM and after different incubation times the reaction was quenched and reaction products were analyzed by reverse phase HPLC. For wt SaDsbA and SaDsbA E96Q the reaction at pH 3.5 was over in less than 2 min, whereas for SaDsbA T153V the reaction took 1 h to complete (data not shown), supporting the finding that it has a higher pK a .
Disulfide Reductase Activity-Insulin contains two intramolecular disulfide bonds that link the A and B chains. Reduction of these disulfides causes the two chains to dissociate and the insoluble B chain to precipitate. Disulfide reductase activity can thus be assessed by monitoring the increase in turbidity of an insulin solution in the presence and absence of possible reductases. Most TRX-like oxidoreductases, including EcDsbA and EcDsbC, are active in this assay, though EcDsbA has only about 10% the activity of DsbC. We determined the rate of insulin reduction by DTT catalyzed by SaDsbA. Unlike EcDsbA, SaDsbA did not show any activity suggesting that its less hydrophobic surface features do not allow interaction with insulin (Fig. 5A). The T153V variant that emulates EcDsbA in the residue preceding the cis-Pro at the active site, partially restores reductase activity (Fig. 5A). Conversely, the equivalent replacement on EcDsbA (V150T) abolishes EcDsbA insulin reductase activity (data not shown), supporting the  FEBRUARY 15, 2008 • VOLUME 283 • NUMBER 7

JOURNAL OF BIOLOGICAL CHEMISTRY 4267
notion that this residue is involved in substrate interactions and that Thr and Val residues preceding cis-Pro confer different binding characteristics.
Disulfide Isomerase Activity-SaDsbA contains residues characteristic of the disulfide isomerases DsbC and DsbG (D and Y in the DXXCXYC motif, Thr in cis-Pro loop). To assess whether these residues confer isomerase activity, we determined the SaDsbA catalyzed recovery of active RNase A from oxidized, scrambled RNase A (scRNase) (45). In this assay, the isomerase DsbC yields 80% of active molecules (Fig. 5B) whereas SaDsbA has an activity similar to that of the oxidase EcDsbA, with 20% active RNase A after 300 min under the applied conditions (Fig.  5B). SaDsbA mutants did not show any activity in this assay.

DISCUSSION
Disulfide bonds between cysteine residues contribute to the stability and activity of many secretory proteins, and organisms contain complex enzymatic systems to promote catalysis of disulfide bond formation. These enzymatic pathways have been extensively studied in E. coli (reviewed in Ref. 3,58). Functional homologues of Dsb proteins have also been found in Gram-positive bacteria (15)(16)(17)(18)(19)(20)(21)(22); however our knowledge of the processes of oxidative folding in these organisms is scarce. Moreover, investigations of Gram-positive Dsb proteins indicate that we may not be able to directly extrapolate the mechanisms observed in E. coli to these organisms. For example, functional and structural characterization of M. tuberculosis DsbE (22) showed that it is an oxidant even though its structure and sequence resemble that of CcmG (DsbE), a Gram-negative disulfide reductant.   Our work also shows that despite the overall structural similarity, SaDsbA does not operate in the same way as EcDsbA. The structural fold of SaDsbA is typical of DsbA disulfide oxidants from Gram-negative bacteria (Fig. 1A). However, it has a truncated binding groove and a charged rather than hydrophobic surface surrounding the redox active site (Figs. 1, B and C  and 3A). These structural differences suggest that the two proteins have different substrate specificities. This was confirmed in that EcDsbA catalyzes insulin reduction whereas SaDsbA does not. A single SaDsbA mutation (T153V) in the cis-Pro loop (which is known to be involved in substrate binding (50 -52)) partially restored activity in this reaction.
Sequence alignment showed that these characteristics of SaDsbA are generally conserved among Gram-positive DsbAs ( Fig. 2A). Most Gram-positive sequences have a similar deletion at the C-terminal segment compared with EcDsbA, and therefore are also likely to have a truncated groove. Also, Grampositive DsbAs have acidic residues in the loop connecting helices ␣3 and ␣4 that are located close to the active site.
SaDsbA, like EcDsbA, has an oxidizing redox potential (EЈ 0 Ϫ131 mV and Ϫ122 mV, respectively) (26,40). Interestingly, the acidic residue Glu 96 in the ␣3-␣4 loop adjacent to the active site, decreases the redox potential. When SaDsbA Glu 96 is mutated to Q, the redox potential increases from Ϫ131 mV (26) to Ϫ111 mV (Fig. 4C). This suggests that Glu 96 plays an important role in controlling the oxidative strength of SaDsbA.
A major characteristic of EcDsbA that contributes to its highly oxidizing nature is its unstable disulfide bond (56) as well as an unusually low pK a for the nucleophilic cysteine (42). Surprisingly, we found that the strongly oxidizing redox potential of SaDsbA does not arise from an unstable disulfide. Both chemically and thermally induced unfolding experiments showed that oxidized and reduced SaDsbA have equivalent thermodynamic stabilities (Fig. 4, A and B and Table 2).
The redox potential of SaDsbA might derive from the low pK a (3.37) of the nucleophilic cysteine Cys 26 ( Fig. 4D and Table  2B), which is similar to that of EcDsbA (3.3 (40)). One factor contributing to the low pK a for Cys 30 in EcDsbA, is stabilization of the thiolate by hydrogen bonding to the His side chain in the CHXC motif (40,59). SaDsbA has a Tyr rather than a His, but the catalytic CXXC is flanked on either side by basic residues that could help to stabilize a thiolate anion (Fig. 2B). However, an acidic residue close to the active site (e.g. Glu 96 ), destabilizes the thiolate as has been shown for other redox-active proteins (60). This was broadly borne out, in that replacement of the acidic residue with a polar residue (E96Q), dropped the pK a of the nucleophylic cysteine even further (pK a 3.09) and increased the redox potential to Ϫ111 mV ( Fig. 4C and Table 2B). The Thr in the cis-Pro loop was found to have a major effect on pK a in that mutation to Val increased it from 3.37 to 5.21 (Fig. 4D). This change to the pK a may be due to removal of a water-mediated hydrogen bond to the Cys 26 thiolate in the T153V variant as observed in the crystal structure (Fig. 2B).
A major difference between SaDsbA and EcDsbA is that SaDsbA does not require a DsbB for activity (15,26). Indeed, S. aureus does not encode a DsbB (26). In agreement with these findings we showed that the structure of SaDsbA cannot interact with DsbB in the same way that EcDsbA does (Fig. 3B) (49), and a direct assay of DsbB activity showed that EcDsbA causes quinone reduction by EcDsbB but SaDsbA does not (Fig. 3C). An important implication of these findings is that S. aureus uses a very different mechanism to that of E. coli for forming disulfides in secreted proteins, relying on the equivalent stabilities of the oxidized and reduced forms so that oxidizing agents such as molecular oxygen in the extracellular milieu may be all that is necessary to convert the reduced form to the functionally active oxidized form.
In conclusion, despite similar folds, oxidizing potentials and nucleophilic cysteine pK a s, SaDsbA from the Gram-positive pathogen S. aureus has a different mechanism than EcDsbA to recover its active oxidized form. This mechanism is likely to be based on the similar thermodynamic stabilities of the oxidized and reduced forms of SaDsbA, that could ensure efficient recovery of the oxidized form through extracellular oxidants. SaDsbA thus needs only the presence of oxidizing agents in the extracellular environment to access the catalytically active oxidized form. Moreover, this mechanism could be common to an entire subgroup of Gram-positive organisms that encode a DsbA but not a DsbB, including the important pathogens E. faecalis and L. monocytogenes.