Sulfur Denitrosylation by an Engineered Trx-like DsbG Enzyme Identifies Nucleophilic Cysteine Hydrogen Bonds as Key Functional Determinant

Abstract

Exposure of bacteria to NO results in the nitrosylation of cysteine thiols in proteins and low molecular weight thiols such as GSH. The cells possess enzymatic systems that catalyze the denitrosylation of these modified sulfurs. An important player in these systems is thioredoxin (Trx), a ubiquitous, cytoplasmic oxidoreductase that can denitrosylate proteins in vivo and S-nitrosoglutathione (GSNO) in vitro. However, a periplasmic or extracellular denitrosylase has not been identified, raising the question of how extracytoplasmic proteins are repaired after nitrosative damage. In this study, we tested whether DsbG and DsbC, two Trx family proteins that function in reducing pathways in the Escherichia coli periplasm, also possess denitrosylating activity. Both DsbG and DsbC are poorly reactive toward GSNO. Moreover, DsbG is unable to denitrosylate its specific substrate protein, YbiS. Remarkably, by borrowing the CGPC active site of E. coli Trx-1 in combination with a T200M point mutation, we transformed DsbG into an enzyme highly reactive toward GSNO and YbiS. The pK_a of the nucleophilic cysteine, as well as the redox and thermodynamic properties of the engineered DsbG are dramatically changed and become similar to those of E. coli Trx-1. X-ray structural insights suggest that this results from a loss of two direct hydrogen bonds to the nucleophilic cysteine sulfur in the DsbG mutant. Our results highlight the plasticity of the Trx structural fold and reveal that the subtle change of the number of hydrogen bonds in the active site of Trx-like proteins is the key factor that thermodynamically controls reactivity toward nitrosylated compounds.

Introduction

In eukaryotes, NO, a lipophilic gas generated by nitric-oxide synthases, plays an important role as a signaling molecule, in large part through the S-nitrosylation of cysteine thiols present in proteins (SNO proteins).⁴ For instance, the reversible S-nitrosylation of caspase 3 has been shown to regulate the activity of this pro-apoptotic protein (1). In addition to protein thiols, activated NO intermediates also react with low molecular weight thiols, such as GSH, resulting in the formation of S-nitrosoglutathione (GSNO). GSNO can then in turn cause the direct nitrosylation of proteins via transnitrosylation reactions (2).

Bacteria also encounter NO physiologically, when it is produced at low concentrations during anaerobic growth to serve as a terminal electron acceptor. Under these conditions, intracellular proteins can be nitrosylated, as recently reported in Escherichia coli (3). In this bacterium, the transcription factor OxyR, which controls an oxidative stress response, also serves as a regulator of protein S-nitrosylation and protects against endogenous nitrosative stress (3). Bacteria also face high levels of NO generated by the induction of nitric-oxide synthase in macrophages as part of the defense mechanisms of the host against invading microorganisms (4). These high NO levels cause a nitrosative stress and potentially lead to cell death. Detoxification of NO, as well as SNO, has also been shown to contribute to pathogenicity of bacteria (5).

Two enzymatic pathways play a major role in denitrosylation and the GSNO reductase and the thioredoxin (Trx) systems (1, 6). These systems are important both to regulate SNO-based signaling and to protect microorganisms from nitrosative stress. GSNO reductase is a NADH-dependent enzyme that is conserved from bacteria to human and that metabolizes GSNO to oxidized glutathione and NH₃ (6). GSNO reductase does not directly denitrosylate SNO proteins. However, because the GSNO pool is in equilibrium with protein thiols (2), reduction of GSNO by the GSNO reductase indirectly results in protein denitrosylation.

Trx is a ubiquitous oxidoreductase whose catalytic activity depends on a conserved CXXC motif that is maintained reduced by thioredoxin reductase at the expense of NADPH, although the best characterized function of Trx is the reduction of disulfide bonds that form in proteins during catalysis or as a result of oxidative stress (7), and Trx has recently been shown to also catalyze the direct denitrosylation of proteins in vivo (1). Trx is also able to reduce GSNO and other nitrosylated low molecular weight thiols in vitro, but the physiological relevance of this activity remains to be shown (8).

The fact that the GSNOR and Trx systems exclusively function in the bacterial cytoplasm raises the question of how the proteins that are present in the periplasm are protected from reactive nitrogen species. To address this problem, we decided to investigate the role of DsbC and DsbG, two periplasmic oxidoreductases from the Trx family, in the protection of periplasmic proteins against nitrosative stress. DsbG and DsbC were selected because they both function in a reductive pathway that also involves DsbD, an inner membrane protein that transfers electrons from the thioredoxin system to the periplasm. Both DsbC and DsbG protect periplasmic proteins containing single cysteine residues from oxidation (9, 10), although DsbC mainly functions as a periplasmic disulfide bond isomerase (10–12).

Here, we show that DsbG and DsbC are poorly reactive toward low molecular weight nitrosothiols (GSNO and CysNO). Moreover, DsbG is also unable to catalyze the denitrosylation of its specific substrate protein YbiS. This prompted us to investigate the molecular factors that would allow DsbG to function in denitrosylation like E. coli Trx-1. Using site-directed mutagenesis, we transformed DsbG into a protein highly reactive toward GSNO and YbiS. We provide structural insights explaining how the mutation of only three amino acids alters the pK_a, redox, and thermodynamic properties to resemble these of Trx-1, thereby completely changing the function of DsbG.

Results

DsbG and DsbC Are Poor Denitrosylating Enzymes

As a first step, we asked whether DsbC and DsbG are able, like Trx-1, to catalyze the in vitro reduction of GSNO. If DsbC and DsbG are able to reduce GSNO, this should lead to the oxidation of their catalytic CXXC motifs, which can be monitored using AMS trapping experiments. AMS is a 490-Da reagent that covalently reacts with free thiol groups, leading to a mobility shift respective of the modified protein on SDS-PAGE gels (Fig. 1A). E. coli DsbG, DsbC, and Trx-1 were expressed and purified, using a His tag present at the C terminus and the purified proteins incubated with an excess of GSNO. As shown in Fig. 1A, both DsbG and DsbC remain reduced, even when the proteins are incubated with high concentrations of GSNO, in contrast to Trx-1, which becomes fully oxidized. In parallel, we tested the ability of DsbG and Trx-1 to reduce GSNO using a fluorescence assay, following the oxidation of the proteins by GSNO (Fig. 1B). This assay is based on the different fluorescence spectra of oxidized and reduced DsbG and Trx-1. Trx-1 rapidly reacts with GSNO, leading to the oxidation of Trx-1, in agreement with the data reported by Nikitovic and Holmgren (8). In contrast, DsbG reacts ∼100 times slower with GSNO, confirming the results from the AMS trapping experiments. We also compared the reactivity of both enzymes to CysNO, an unstable low molecular weight thiol and obtained similar results (Fig. 1B). Thus, DsbG and DsbC are poorly reactive toward LMW nitrosothiols, in contrast to Trx-1.

FIGURE 1.

Oxidation of DsbG and DsbC with low molecular weight S-nitrosothiols. A, E. coli DsbG, DsbC, and Trx-1 (1 μm) were incubated with 10 μm of GSNO or CysNO at room temperature in the dark. At different incubation times, the redox state of the proteins was determined by AMS gel shift. B, progress curves of the oxidation of DsbG and Trx-1 by GSNO and CysNO, as assessed by the decrease in fluorescence emission at 334 nm (excitation at 280 nm), are shown. The reaction was started by mixing 1 μm of enzyme with 10 μm of either GSNO or CysNO. The fraction of reduced protein was plotted as a function of time. Progress curves were fitted with an exponential decay, resulting in the following rate constants: CysNO: k_obs of 10⁻² min⁻¹ for DsbG and 2.1 min⁻¹ for Trx-1; GSNO: k_obs of 3 × 10⁻² min⁻¹ for DsbG and 3.6 min⁻¹ for Trx-1. red, reduced; ox, oxidized.

Because Trx has been shown to catalyze the direct denitrosylation of proteins, we then sought to determine whether the periplasmic oxidoreductases are able to catalyze the denitrosylation of substrates proteins by determining the ability of DsbG to denitrosylate YbiS, one of its substrate proteins (9). YbiS was first S-nitrosylated by incubation with high concentrations of GSNO and then incubated with a 10-fold molar excess of reduced DsbG or Trx-1. The ability of DsbG or Trx-1 to denitrosylate YbiS was determined by following NO release using the Saville-Griess assay (Fig. 2). We found that DsbG is only poorly reactive toward YbiS-NO, in contrast to Trx-1, which is able to efficiently denitrosylate this nitrosylated protein.

FIGURE 2.

Denitrosylation of YbiS-SNO by E. coli DsbG and Trx-1. The reactions were initiated by mixing a 10-fold molar excess of reduced DsbG or Trx-1 with Ybis-SNO in the dark at room temperature. At different times, the amount of NO released was measured using the Saville-Griess assay. The data are means ± S.E., n = 3.

Engineering DsbG into a Protein Reactive toward GSNO

To gain insights into the intrinsic properties of DsbG that contribute to its resistance against GSNO, we decided to engineer this oxidoreductase into a GSNO reducing enzyme. We first assessed the effect of the redox potential on the GSNO reducing activity. As the XX dipeptides flanked by the active site cysteine residues are known to be important determinants of the redox potential of proteins from the thioredoxin superfamily (7), we replaced the CGPC active site motif of E. coli Trx-1. The mutant protein, DsbG_CGPC was expressed and purified, and we determined a redox potential of −195 mV (Fig. 3 and Table 1), which is already significantly more reducing than wild type DsbG (−125 mV) (13).

FIGURE 3.

Redox potential of WT and mutant DsbG. The redox potential of WT DsbG and DsbG_CGPC was determined in equilibrium with GSH/GSSG, whereas the redox potential of DsbG_CGPC-T200M and Trx-1 was determined in equilibrium with DTTred/DTTox. The percentage of reduced protein is plotted as a function of the redox potential.

TABLE 1.

Properties of DsbG wild type and the mutants

red, reduced; ox, oxidized.

DsbG	pKa Cys109	No. H-bonds S109γa	Tm			Eo′
			red	ox	Δred-ox
			°C			mV
WT	3.5 (14)	5	70	58	12	−125 (13)
T200M	4.2 (13)	3	68	62	6	−181 (13)
CGPC	5	4	66	64	2	−195
CGPC + T200M	6.1	1	59	70	−11	−280

^a Based on the reduced form in the respective structures and using a cutoff of 3.5 Å.

Because the redox potential is directly linked to the Gibbs free energy ΔG, which can be regarded as the thermodynamic driving force of the reaction, we determined the difference in stability between the oxidized and reduced forms of WT DsbG and DsbG_CGPC by measuring their melting temperatures in a CD experiment (Fig. 4). For the WT, we measured a Δ_red-oxT_m of 12 °C, the reduced form being more stable. For DsbG_CGPC, the difference decreases to only 2 °C. Determination of the pK_a of the nucleophilic cysteine shows a shift from a pK_a of 3.5 for the wild type (14) to a value of 5 for DsbG_CGPC (Table 1). Importantly, DsbG_CGPC becomes also more reactive toward GSNO and CysNO than wild type DsbG (Fig. 5A), whereas introducing the CPYC active site of DsbG in Trx-1 makes Trx-1 less reactive (Fig. 5B). Importantly, DsbG_CGPC also becomes reactive toward nitrosylated YbiS (Fig. 2). This experiment clearly shows the importance of the CGPC active site for the denitrosylation activity.

FIGURE 4.

Unfolding curves of WT and mutant DsbG. The unfolding of the proteins was followed by circular dichroism. The fraction of unfolded protein was plotted in function of the temperature. The data for the T200M DsbG mutant are not shown but give similar results as for the DsbG_CGPC (see Table 1). A, WT DsbG. B, DsbG_CGPC. C, DsbG_CGPC-T200M.

FIGURE 5.

The active site determines the reactivity toward LMW nitrosothiols. A, progress curves of the oxidation of DsbG and its CGPC mutant are shown. B, progress curves of Trx-1 and its CPYC mutant are shown. C, progress curves of the oxidation of WT and mutant DsbG by GSNO. D, progress curves of the oxidation of WT and mutant DsbG by CysNO. The reaction was followed as the decrease in fluorescence emission at 334 nm (excitation at 280 nm). Curves are fitted with a single exponential decay (dotted lines).

Combining the Trx Active Site CGPC Motif with a T200M Mutation Makes DsbG Highly Reactive toward LMW Nitrosothiols

Encouraged by these results, we decided to introduce an additional mutation in DsbG_CGPC to further lower the redox potential. We chose to replace Thr²⁰⁰ by Met because this mutation has been shown by Hiniker et al. (13) to lower the redox potential of DsbG. We found that the resulting mutant has a strikingly lowered redox potential (−280 mV; Fig. 3 and Table 1), which is similar to that of Trx-1 (15). Remarkably, whereas for the WT, the reduced form is more stable than the oxidized form, the oxidized form of the double mutant becomes more stable (T_m = 70 °C) than the reduced form (T_m = 59 °C) (Fig. 4 and Table 1). The pK_a of the nucleophilic cysteine is also further increased to 6.1. Thus, the triple mutant displays redox and thermodynamic properties similar to these of Trx-1 (15). This double mutant also becomes highly reactive toward LMW nitrosothiols (Fig. 5). Based on pseudo first order kinetics in the presence of an excess of substrate, we estimated from t _½ that this double mutant reduces GSNO and CysNO with a second order rate constant in the range of ∼10² m⁻ s⁻¹ (Fig. 5).

Hydrogen Bonds Influence Cysteine Reactivity

To gain more insight in the underlying mechanism by which these subtle mutations in DsbG cause such a drastic change in activity, we solved the crystal structures of DsbG_CGPC and DsbG_CGPC-T200M. Both proteins crystallized in similar conditions as the WT and in the same space group C2 (Table 2) (16, 17). The structures of DsbG_CGPC and DsbG_CGPC-T200M were solved to 1.7 and 1.96 Å, respectively. The overall structure of the mutants and the WT are the same with root mean square deviations between the Cα atoms of the wild type and the mutants of 0.25 Å (DsbG_CGPC) and 0.39 Å (DsbG_CGPC-T200M). The overall structure has also been reported to be unchanged for T200M (root mean square deviation 0.45 Å) and other DsbG mutants (13). As observed for WT DsbG (17), the loop between β3 and β4 (residues 43–48) was difficult to model, especially in DsbG_CGPC, where a double conformation of the loop was modeled in.

The short distance between the Sγ of Cys¹⁰⁹ and the backbone nitrogen of Cys¹¹² in oxidized WT DsbG (3.1 Å) has been reported to be one of the reasons why the reduced form is more favored than the oxidized form (17). Whereas DsbG_CGPC is found in the reduced form in both chains in the asymmetric unit of the crystal structure (Protein Data Bank code 5G1L), both chains A and B of the DsbG_CGPC-T200M structure (Protein Data Bank code 5G1K) show a mixture of the oxidized and reduced form of the protein (27% oxidized, 67% reduced in chain A, 38% oxidized, 62% reduced in chain B, as determined from the refined occupancies). However, the distance between the Sγ of Cys¹⁰⁹ and the backbone nitrogen of Cys¹¹² (2.9 Å) is similar to the WT DsbG one (Fig. 6, A and B).

FIGURE 6.

The oxidation state and the hydrogen-bonding pattern of Cys109 in WT DsbG, DsbG_CGPC and DsbG_CGPC-T200M. A and B, the x-ray structure of DsbG_CGPC (A) shows the reduced form of the protein, whereas the structure of DsbG_CGPC-T200M (B) shows a mixture of the reduced and oxidized forms. Cys¹⁰⁹ and Cys¹¹² are shown in light yellow and light blue stick representation for DsbG_CGPC and DsbG_CGPC-T200M, respectively. The 2F_o − F_c electron density map is shown in blue mesh, contoured at 1.2 σ. C, the H-bonds between the S₁₀₉γ and the cis-Pro loop are shown for WT DsbG (left, shown in gray stick representation) and the DsbG_CGPC-T200M mutant (right, shown in light blue stick representation). A shift of ∼1 Å is observed in the cis-Pro loop between DsbG_CGPC-T200M and WT DsbG, causing the loss of two hydrogen bonds between S₁₀₉γ and residue 200 (Thr in WT DsbG and Met in the DsbG_CGPC-T200M). Hydrogen bonds are shown as red dashed lines. The H-bond between the S₁₀₉γ and the backbone oxygen of Met²⁰⁰ is longer than 3.5 Å and is therefore shown faded. The Protein Data Bank entry 1V57 was used for the WT DsbG, because both chains of this entry also contain a mixture of both redox states.

Change of the wild type DsbG CPYC motif to the Trx1 CGPC active site does not change the position of the Cα atoms in the active site motifs in both DsbG_CGPC and DsbG_CGPC-T200M. This has also been shown in other studies where the XX dipeptide is changed (14, 18). In DsbG_CGPC-T200M where Thr²⁰⁰ is replaced by Met, however, there is a ∼1 Å displacement of the Cα atoms at residues 198–201 comprising the cis-Pro loop. This displacement causes the backbone oxygen of Met²⁰⁰ to move further away from the disulfide bond of DsbG_CGPC-T200M. As a consequence, two hydrogen bonds to the nucleophilic cysteine are lost (using a cutoff of 3.5 Å for a hydrogen bond) (Fig. 6C and Table 1): one between the backbone oxygen of Met²⁰⁰ and the sulfur of nucleophilic Cys¹⁰⁹ and one caused by absence of the side chain of the Thr. In the oxidized and reduced form of wild type DsbG, the main chain O and Oγ of Thr²⁰⁰ are within H-bond distance with the Sγ of Cys¹⁰⁹ (Fig. 6), which has been shown to destabilize the oxidized form of the protein (14, 17). The lack of this interaction from Met²⁰⁰ in the DsbG_CGPC-T200M mutant could lead to the observed stabilization of the oxidized form.

Discussion

S-Nitrosylation, resulting from the covalent attachment of an NO group to a cysteine or LMW thiol, has been regarded as a redox-based mechanism, allowing the post-translational regulation of all main classes of proteins (19). In addition, it also alters the intracellular thiol pool under nitrosative stress. Alterations in cellular NO and GSNO levels and in protein S-nitrosylation patterns have been shown to contribute to pathological conditions, e.g. in neurodegenerative diseases or bacterial pathogenesis (5, 20). Two major systems have been described to be involved in the denitrosylation of proteins and LMW nitrosothiols: the Trx system and the GSNO reductase system (1, 6). The latter cannot directly reduce nitrosylated proteins, but because of the equilibrium between GSNO and nitrosylated proteins, its reduction of GSNO indirectly leads to reduction of nitrosylated proteins (6). On the other hand, the Trx system directly denitrosylases by using its catalytic center with both cysteines in their reduced state. Trx establishes a disulfide bridge with the substrate SNO protein, allowing the release of NO, which will subsequently be protonated to HNO. An intramolecular disulfide bond is then formed between the Trx active site cysteines, releasing the substrate protein in the reduced state. Next, thioredoxin reductase restores the reduced state of the Trx catalytic cysteines through a NADPH-dependent reduction of the Trx intramolecular disulfide.

Here, we questioned whether the DsbG and DsbC oxidoreductases could be responsible for the reduction of periplasmic nitrosothiols in the E. coli periplasm. Upon the finding that neither could reduce the LMW nitrosothiols GSNO or Cys-SNO (Fig. 1) and that DsbG was not even able to denitrosylate its specific substrate protein YbiS (Fig. 2), we decided to investigate the molecular factors that would allow DsbG to function in denitrosylation like Trx-1.

The CXXC active site motifs of Trx family proteins determine to a large extent their specific redox properties, and several studies have assessed the effect of changing the XX dipeptide within the CXXC catalytic signature motif of one thiol-disulfide oxidoreductase with that of another. Generally, this results in an oxidoreductase with redox properties similar to that of the oxidoreductase from where the dipeptide originated (18, 21–23). To this end, in two state-of-the-art papers, the Glockshuber group changed the XX dipeptide of the prototype reductase Trx (CGPC) and oxidase DsbA (CPHC) into the protein-disulfide isomerase type (CGHC), the glutaredoxin (Grx) type (CPYC), and the DsbA or Trx type, respectively (22, 23). The redox potential of Trx gradually increases from −270 mV for the WT enzyme, over −221 mV for the protein-disulfide isomerase type, to −204 and −195 mV for the DsbA and Grx types, respectively. Conversely, the DsbA redox potential decreased from −122 mV for the WT enzyme over −146 mV for the protein-disulfide isomerase type mutant, −157 mV for the Grx type, and −214 mV for the Trx type. Similarly, changing the dipeptide of the DsbG CPYC motif by that of Trx-1, CGPC, lowered the redox potential by 70 mV, making it more reducing (−195 mV in comparison to −125 mV for the WT enzyme (13)). In addition to affecting the redox potential, changes in the XX dipeptide of the CXXC active site motif also influence the pK_a of the nucleophilic cysteine. For instance, the Trx CXXC motif conversion in the protein-disulfide isomerase/Grx/DsbA types resulted in a maximal ΔpK_a of 1.2 units, whereas the DsbA conversion to the protein-disulfide isomerase/Grx/Trx types led to a ΔpK_a of 2.93 units (22, 23). Here, we observe a pK_a increase of 1.5 units between the WT DsbG and the CGPC DsbG (Fig. 3 and Table 1).

Most importantly, the DsbG_CGPC mutant shows increased denitrosylating activity (Fig. 5), toward both LMW nitrosothiols and nitrosylated YbiS. However, the XX active site dipeptide is not the only important feature determining the function of Trx family proteins. The loop containing the conserved cis-proline, which is in close proximity to the active site CXXC motif, also affects the function and the redox properties (14). A DsbG mutant where the residue located N-terminally from the proline in the cis-Pro loop was mutated, T200M, showed a more reducing redox potential of −181 mV and was shown to behave more like the disulfide isomerase DsbC (13). We prepared a DsbG mutant combining the CGPC Trx active site with this T200M mutation, DsbG_CGPC-T200M. DsbG_CGPC-T200M has a redox potential of −280 mV, even more reducing than WT Trx-1, −271 mV (15), and a nucleophilic Cys pK_a of 6.1 (Table 1). A similar result was obtained for E. coli Trx-1, where combining the CPHC active site of DsbA with an I75T (the residue before the cis-Pro in E. coli Trx) mutation shifted the redox potential to −180 mV, 91 mV more oxidizing than the WT Trx (14). Even a single mutation, P31T, in the active site of Staphylococcus aureus Trx, in which the surrounding of the nucleophilic cysteine becomes more hydrophilic, resulted in a decrease of the pK_a value of the cysteine by approximately 1 pH unit, whereas its redox potential increased by 30 mV (24). The introduced mutations seem to have a clear additive effect as we have also shown here for DsbG (Table 1). The increase in pK_a of the nucleophilic cysteine follows the decrease of the redox potential toward a more reducing value (−280 mV). The increase of the pK_a of the nucleophilic cysteine closer toward physiological conditions increases its reactivity (25).

In addition, mutations in the CXXC motif and the cis-Pro loop have been shown to change the thermodynamic properties of Trx family proteins (7). In general, the oxidized form of a reductase, such as Trx is more stable than the reduced form (7), whereas, conversely, the reduced form of an oxidase such as DsbA is more stable than its oxidized form (23). Mutations in the Trx CXXC motif to the protein-disulfide isomerase/Grx/DsbA type leads to a decrease in the stability of the oxidized form, whereas the conversion of the DsbA CXXC motif in the protein-disulfide isomerase/Grx/Trx type leads to an increase in the stability of the oxidized form. Similarly, here we see that whereas for the WT DsbG, the reduced form is more stable (Δ_red-oxT_m of 12 °C), for DsbG_CGGC this difference is only 2 °C. Remarkably, for DsbG_CGPC-T200M, the oxidized form becomes 11 °C more stable than the reduced one (Fig. 4 and Table 1). Thus, the thermodynamic difference in stability between the oxidized and reduced form drives the reaction toward the reduction of nitrosylated substrates.

It is important to mention that the number of hydrogen bonds received by the sulfur cysteine determines the stability of the thiolate, a key factor that modulates the equilibrium between thiol (SH) and thiolate (S⁻) (25). The crystal structure of DsbG_CGPC-T200M indeed shows that introducing the T200M mutation leads to the loss of two hydrogen bonds to Cys¹⁰⁹ (Fig. 6C and Table 1). This loss of hydrogen bonding can also be linked to the stability of the Cys¹⁰⁹–Cys¹¹² disulfide bond. This disulfide bond has previously been shown to be unstable (17). Even upon oxidation using (1,10 phenanthroline) copper(II), exposure of WT DsbG crystals to synchrotron radiation leads to the reduction of the disulfide bond. Heras et al. (17) showed, however, that this is not merely due to radiation but is rather an intrinsic property of this disulfide bond. Thus, the fact that we observe the disulfide bond in the crystal structure of DsbG_CGPC-T200M, even though we did not chemically oxidize the proteins before crystallization and we used synchrotron radiation, indicates that the mutations indeed stabilize the oxidized form.

Heras et al. (17) argue that the short distance between the Sγ of Cys¹⁰⁹ and the backbone nitrogen of Cys¹¹² in oxidized WT DsbG (3.1 Å) could explain the destabilization of the oxidized form. However, whereas we actually clearly see a stabilization of the oxidized form for the mutants, especially for DsbG_CGPC-T200M mutants, for which the oxidized form is 23 °C more stable than the WT oxidized form (in comparison with the respective reduced forms), this distance is the same as in the WT. Heras et al. (17) had also already hinted at the destabilizing effect of the Thr²⁰⁰–Cys¹⁰⁹ hydrogen bonds on the oxidized form. Indeed, as mentioned above, the crystal structure of DsbG_CGPC-T200M shows the loss of two hydrogen bonds upon introducing the T200M mutation. Similarly, the I75T mutation in Trx-1 led to a more oxidizing form, thus stabilizing the reduced form. This effect was attributed to the formation of an extra hydrogen bond between the nucleophilic Cys and the T75 side chain, with respect to the WT enzyme (14).

Again, most importantly, the activity of the DsbG_CGPC-T200M double mutant was drastically changed, changing a protein with hardly any denitrosylation activity into an enzyme with an estimated second order rate constant for the denitrosylation of GSNO and CysNO of ∼10² m⁻ s⁻¹ (Fig. 5). This rate is comparable with the published second order rate constant for GSNO reduction by E. coli Trx-1 (760 m⁻ s⁻¹) (8).

In conclusion, we present the first evidence for the importance of the CXXC motif and of the cis-Pro residue in determining the activity of a Trx family oxidoreductase toward nitrosylated substrates. This property had never been investigated for any of the CXXC DsbA mutants. In addition, a recent paper describing the denitrosylating properties of the human Trx-related protein TRP14 does not study the role of the CXXC motif or the redox properties of this enzyme in relation to its function (26). In general, our study is a beautiful example of the plasticity of the Trx fold. We gave evolution a helping hand by showing how DsbG can be modified relatively easily from an enzyme that has been shown to reduce sulfenic acids on proteins (9) to an enzyme that is capable of reducing nitrosylated targets. Evolution of DsbG toward a denitrosylating enzyme could affect the pathogenicity of E. coli. In addition, our findings could contribute to the design of other mutant proteins with therapeutic potential, e.g. in neurodegenerative diseases.

Experimental Procedures

Plasmids and Site-directed Mutagenesis

The plasmids used for the expression of wild type E. coli DsbC, DsbG, YbiS, and Trx-1 have been described previously (9, 10, 21). The P110GY111P (CGPC) and T200M mutations were inserted in DsbG using the Stratagene QuikChange mutagenesis kit, using the CGPC-fw (3′-gcc gat ccg ttc tgc gga cct tgt aaa cag ttc tgg-5′) and CGPC-rv (3′-cca gaa ctg ttt aça agg tcc gca gaa cgg atc ggc-5′) primers for the CGPC mutant and the T200M-fw (3′-gca aat gtc atg ccg gct atc-5′) and T200M-rv (3′-gat agc cgg cat gac att tgc-5′) primers for the T200M mutant. The insertion of the desired mutations was confirmed by DNA sequencing. Construction of the Trx CPYC mutant was described previously (21).

Protein Overexpression and Purification

WT E. coli DsbG, DsbC, YbiS, and Trx-1 were overexpressed in BL21 cells and purified as described previously (9, 10, 21). Their corresponding mutants were expressed and purified using the same protocol.

Protein Reduction and Oxidation

DsbG, DsbC, Trx-1, YbiS, and their mutants were reduced by incubation in 10 mm DTT for 1 h at 4 °C. The reactions were then desalted on NAP-5 columns (GE Healthcare) and equilibrated in 50 mm phosphate buffer, pH 7.0, 50 mm NaCl and 1 mm EDTA.

Synthesis of LMW SNOs

S-Nitrosocysteine (CysNO) and GSNO were made freshly by reacting 50 mm cysteine or GSH, respectively, with 50 mm NaNO₂ in 0.5 m HCl for 20 min in the dark at room temperature as described elsewhere (8, 27). GSNO and CysNO concentrations were determined by absorption spectroscopy using the published extinction coefficient (ϵ₃₃₄ = 922 m⁻¹ cm⁻¹) (8, 28).

Synthesis of YbiS-SNO

YbiS was nitrosylated as previously described (29). Briefly, purified YbiS (90 μm) was first treated with DTT (10 mm) in buffer A (50 mm sodium phosphate, pH 8.0, 150 mm NaCl) 30 min at room temperature to reduce the Cys thiol group and then desalted using a Nap-5 column equilibrated with buffer B (10 mm HEPES, pH 7.7, 0.1 mm EDTA). Reduced YbiS was then incubated with a large excess (500 μm) of CysNO (prepared as described above) at room temperature for 45 min in the dark to form YbiS-SNO with a yield of about 60% (with respect to reactive thiols). Excess of CysNO was removed using a NAP-5 column equilibrated with buffer B. YbiS-SNO was stored at −20 °C.

Denitrosylation Activity of DsbG and Trx-1

The denitrosylase activity of Trx-1 and DsbG was determined using the Saville-Griess assay as described previously (29, 30). This assay allows the measurement of S-nitrosothiol and nitrite levels in a sample. Briefly, samples (volume ≤ 50 μl) diluted into 500 μl of aceto-acetic buffer, pH 2.5, were mixed with 100 μl of sulfanilamide/HgCl₂ solution (0.6% sulfanilamide, 0.2% HgCl₂ in 0.4 m HCl), or sulfanilamide alone. After 5 min of incubation in the dark at room temperature, 25 μl of 0.6% N-(1-naphthyl)ethenediamine in 0.4 m HCl was added to the reaction mixture, and the sample was incubated in the dark for 5 min. Following the second incubation, the absorbance at 540 nm was measured. Subtraction of appropriate blanks and of values for samples treated without HgCl₂ determined the levels of S-nitrosothiol without nitrite. The concentration of S-nitrosothiol or nitrite was calculated with standards of known concentrations of GSNO.

The ability of DsbG and Trx-1 to denitrosylate S-nitrosoproteins was investigated by monitoring the release of nitrite in the presence of these enzymes and Ybis-SNO using the Saville-Griess assay. The reaction is initiated by mixing a 10-fold molar excess of DsbG or Trx-1 with 25 μm of Ybis-SNO in 50 mm phosphate buffer, pH 8.0, 150 mm NaCl, and 1 mm EDTA, in the dark at room temperature. At different time points (5, 10, 15, 30, and 60 min), the amount of released NO was determined using the Saville-Griess assay as described above.

Determination of the Protein Redox State by AMS Gel Shift

1 μm of each freshly reduced oxidoreductase (Trx-1, DsbG, and DsbC) was incubated with 10 μm of GSNO or CysNO at room temperature in the dark. At different incubation times, the redox state of the proteins was determined by AMS gel shift, by alkylating the free thiol groups with AMS (Molecular Probes) as described (31, 32). AMS reacts with free thiol groups, resulting in a molecular mass increase of 490 Da. A 100-μl sample of each enzyme concentrated at 2 μm was precipitated by adding 10% TCA, incubated 15 min on ice, and centrifuged at 13,000 rpm for 5 min. The resulting pellet was then washed with acetone and resuspended in 30 μl of 50 mm Tris-HCl, pH 7.5, 20 mm AMS, 0.1% SDS, 10 mm EDTA. The sample was evaluated on 15% SDS-PAGE under non-reducing conditions.

Kinetics of Oxidation of Trx-1, DsbG, and Their Mutants by LMW Nitrosothiols

The ability of Trx-1, DsbG, and their mutants to be oxidized with GSNO or CysNO was investigated by following the decrease of fluorescence at the emission wavelength of 334 nm (excitation at 280 nm) in function of time by using a SLM-Aminco 8000C spectrofluorometer. The reaction was initiated after mixing 1 μm of each freshly reduced proteins with GSNO or CysNO (1 or 10 μm) in a total volume of 3 ml.

Redox Potential Determination

Equilibrium against Glutathione

The redox potentials of WT DsbG and DsbG_CGPC were determined by fluorimetry by measuring the equilibrium constant between WT or CGPC mutant DsbG and glutathione as described previously (33). Briefly, 1 μm of oxidized protein was incubated for 12 h at 293 K under a nitrogen atmosphere with 0.1 mm GSSG mixed with different concentrations of reduced GSH (0.2 μm to 1 mm) in 50 mm sodium phosphate, pH 7.0, 1 mm EDTA. The relative amount of reduced protein at equilibrium was measured using the specific fluorescence at 330 nm (excitation at 280 nm).

Equilibrium against DTT_red/DTT_ox

The redox potential of the DsbG double mutant CGPC-T200M and Trx-1 as control were determined by equilibrium against different DTT_red/DTT_ox ratios as described previously (34). Briefly, 1 μm of oxidized protein was incubated for 4 h at 293 K under a nitrogen atmosphere with 10 mm DTT_ox mixed with different concentrations of DTT_red (2 μm to 200 mm) in 50 mm sodium phosphate, pH 7.0, 1 mm EDTA. After incubation, the redox state of each sample was investigated by AMS gel shift (see above). The ratio of the reduced and oxidized forms under each DTT_ox/DTT_red conditions on Coomassie Brilliant Blue-stained SDS-PAGE (15%) was quantified with ImageJ software. The redox potential was calculated from the midpoint of the curves as described (35).

Determination of the pK_a of the Nucleophilic Cysteine

The pH-dependent protonation of the nucleophilic cysteine thiolate was followed by the specific absorbance of the thiolate anion at 240 nm (36). As a reference, the pH-dependent absorbance for the oxidized form of the protein was monitored. All of the measurements were carried out at 298 K in buffer consisting of 10 mm K₂HPO₄, 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 μm. The oxidized and reduced protein samples were prepared as described above. The pH of the protein solution was lowered stepwise to 2.2 by the addition of 0.2 or 0.1 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 = (A₂₄₀/A₂₈₀)_reduced/(A₂₄₀/A₂₈₀)_oxidized) was fitted according to the Henderson-Hasselbalch equation, A_exp = A_SH + (A_S− − A_SH)/(1 + 10^(pK_a−pH)), in which A_exp = A₂₄₀/A₂₈₀ stands for the experimental determined value, A_SH is the = A₂₄₀/A₂₈₀ value for the protonated form, and A_S− = A₂₄₀/A₂₈₀ for the deprotonated form (25).

Unfolding and Thermal Stability

Oxidized and reduced wild type and mutant DsbG were prepared by incubation at room temperature with a 10 molar excess of diamide for 30 min or with 10 mm DTT for 30 min, respectively. The excess of diamide or DTT was removed by size exclusion chromatography on a Superdex 75 HR (10/30) column equilibrated in 50 mm boric acid, pH 8.0, 100 mm KF for the oxidized form and with the same buffer solution containing 1 mm DTT for the reduced form. For the thermal unfolding, samples were diluted to 0.2 mg/ml in the respective boric acid buffer solutions and slowly heated to 95 °C. The fraction of folded DsbG was recorded using CD on a J-715 spectropolarimeter (Jasco, Tokyo, Japan) at 220 nm for reduced wild type DsbG and the mutants. Oxidized wild type DsbG and the mutants were recorded at 232 nm. The change in ellipticity was plotted in function of temperature. The reversibility of the unfolding transitions was checked by cooling down to 4 °C and by rescanning the far UV spectrum.

Crystallization, X-ray Diffraction, and Structure Solution

DsbG_CGPC and DsbG_CGPC-T200M, in 10 mm HEPES, pH 7, 100 mm NaCl, were crystallized at 298 K in 0.1 m sodium citrate pH 3.75, 0.2 m ammonium sulfate, 20% PEG 4000 by the hanging drop method as described for wild type DsbG and other DsbG mutants (13, 16, 17), at a protein concentration of 5 mg/ml. The crystals were dehydrated by increasing the PEG4000 concentration to 30% and cryoprotected by adding 10% glycerol before flash freezing in liquid nitrogen (16). Diffraction data were collected at the Proxima 1 Beamline at the SOLEIL synchrotron facility. The data were processed using the XDS package (37). The structures were solved by molecular replacement using the wild type DsbG structure, Protein Data Bank code 1V57, as a template, using the Phaser program (38) within the CCP4 suite (39). Further manual building and refinement were done using Coot, Refmac, and phenix.refine (40–42). The latter was used from the Phenix suite (42, 43). X-ray data collection parameters, processing, and refinement statistics are summarized in Table 2.

Author Contributions

All authors contributed to the design of the experiments, which were performed by C. L., I. V. M., V. T. D., K. W., A. B., and P. L. Then I. V. M., J.-F. C., and J. M. wrote the manuscript.