Glutaredoxins (Grxs) are a class of GSH (glutathione)-dependent thiol–disulfide oxidoreductase enzymes. They use the cellular redox buffer GSSG (glutathione disulfide)/GSH directly to catalyze these exchange reactions. Grxs feature dithiol active sites and can shuttle rapidly between three oxidation states, namely dithiol Grx(SH)2, mixed disulfide Grx(SH)(SSG) and oxidized disulfide Grx(SS). Each is characterized by a distinct standard reduction potential . The values for the redox couple Grx(SS)/Grx(SH)2 are available, but a recent estimate differs by over 100 mV from the literature values. No estimates are available for for the mixed disulfide couple Grx(SH)(SSG)/(Grx(SH)2 + GSH). This work determined both and for two representative Grx enzymes, Homo sapiens HsGrx1 and Escherichia coli EcGrx1. The empirical approaches were verified rigorously to overcome the sensitivity of these redox-labile enzymes to experimental conditions. The classic method of acid ‘quenching’ was demonstrated to shift the thiol–disulfide redox equilibria. Both enzymes exhibit an (vs. SHE) at a pH of 7.0. Their values (−213 and −230 mV for EcGrx1 and HsGrx1, respectively) are slightly less negative than that () of the redox buffer GSSG/2GSH. Both and vary with log [GSH], but the former more sensitively by a factor of 2. This confers dual catalytic functions to a Grx enzyme as either an oxidase at low [GSH] or as a reductase at high [GSH]. Consequently, these enzymes can participate efficiently in either glutathionylation or deglutathionylation. The catalysis is demonstrated to proceed via a monothiol ping-pong mechanism relying on a single Cys residue only in the dithiol active site.

Introduction

Reversible thiol–disulfide exchange reactions act as versatile redox switches for protein thiols and play important roles in a wide spectrum of integrated cellular functions including redox sensing, cell signaling, antioxidative defense, oxidative protein folding and regulation of protein function [14]. The thioredoxin (Trx) superfamily of dedicated oxidoreductase enzymes has evolved to catalyze these reactions with related but distinct cellular functions. The family includes Trxs, glutaredoxins (Grxs), protein disulfide-isomerases and the disulfide bond protein A (DsbA) [13,5]. They share many features including a thioredoxin fold with a CysN–xx–CysC active-site motif (Figure 1a,c) [6]. The buried CysC (closer to the C-terminus) exhibits a high pKa, while CysN (closer to the N-terminus) is solvent-exposed with an environment that imposes a low pKa. Consequently, its thiolate anion can act as a nucleophile to initiate the exchange processes.

Protein molecular structures.

Figure 1.
Protein molecular structures.

(a) Reduced HsGrx1 (PDB code: 1JHB; thioredoxin fold); (b) GSH bound to the HsGrx1-C8,26,79,83S variant (1B4Q); (c) reduced EcGrx1 (1EGR); (d) CopC-H48C model based on template CopC (2C9Q). Labeled amino acid residues and the GSH fragment are shown as sticks, and protein N- and C-termini are indicated by N and C, respectively.

Figure 1.
Protein molecular structures.

(a) Reduced HsGrx1 (PDB code: 1JHB; thioredoxin fold); (b) GSH bound to the HsGrx1-C8,26,79,83S variant (1B4Q); (c) reduced EcGrx1 (1EGR); (d) CopC-H48C model based on template CopC (2C9Q). Labeled amino acid residues and the GSH fragment are shown as sticks, and protein N- and C-termini are indicated by N and C, respectively.

Grxs are unique as thiol–disulfide oxidoreductase enzymes that employ the cellular redox buffer glutathione disulfide/glutathione (GSSG/GSH) as a direct electron acceptor/donor [3,7,8]. This is attributed to the existence of at least one distinct GSH interaction site that enables them to act as scaffold proteins for the binding and delivery of GSH [9]. X-ray crystal structures of several glutathionylated Grxs are available in which a GSH molecule interacts with a group of highly conserved residues and forms a disulfide bond with the surface-exposed N-terminal reactive Cys (Figure 1b) [1012]. One of the important functions of Grxs is post-translational modification of protein thiols via reversible glutathionylation that is important in regulation of cell viability [4]. Defects in this function have implications in maladies such as Parkinson's and fatty liver diseases [13,14]. In addition, HsGrx1 has also been shown to play a role in copper homeostasis in neuronal cells [15,16].

Dithiol Grxs have been proposed to catalyze thiol–disulfide exchanges via reversible oxidation of the two active-site Cys residues to a disulfide. However, a variety of experiments have demonstrated that the N-terminal reactive Cys residue is all that is required for the catalytic functions of reversible glutathionylation/deglutathionylation of a protein monothiol [17,18]. The C-terminal Cys residue may affect the enzyme activity as a ‘catalytic brake’. Consequently, determination and comparison of the reduction potentials involving both active-site Cys residues and those involving the reactive Cys only are important for an understanding of the catalytic process on a thermodynamic basis. However, while standard reduction potentials for several Grx(SS)/Grx(SH)2 systems have been reported, none have been determined for the glutathionylated mixed disulfide system Grx(SH)(SSG)/(Grx(SH)2 + GSH). In addition, the reported reduction potentials for Grx(SS) are inconsistent and differ by over 100 mV [1922]. This may indicate variation of the reduction potentials with the medium conditions and/or sensitivity of the detection approaches to the redox lability of the enzymes. The present investigation sought to clarify these issues and has determined both types of reduction potentials for two representative dithiol Grx enzymes, Homo sapiens Grx1 (HsGrx1) and Escherichia coli Grx1 (EcGrx1).

Two independent approaches were verified rigorously to overcome the sensitivity of these redox-labile enzymes to experimental conditions and to ensure the reliability of the determinations. The classic approach of ‘quenching’ thiol–disulfide exchange reactions was demonstrated to induce redox shifting. In addition, a solvent-exposed monothiol variant CopC-H48C was generated as a substrate for systematic study of the catalytic function and mechanism of the two Grx enzymes. Both are confirmed to catalyze glutathionylation and deglutathionylation of protein monothiols by a ping-pong mechanism that relies only on the single N-terminal reactive Cys in the active site. The C-terminal Cys may act as a catalytic brake to control the catalysis.

Experimental section

Materials and general methods

General chemicals and reagents were purchased from Sigma–Aldrich, except reductants dithiothreitol (DTT) and tris(2-carboxyethyl)phosphine (Tcep), which were from Astral Scientific and Fluka, respectively. These materials were used as received. Stock solutions of DTT and GSH were prepared in deoxygenated Milli-Q water and stored in an anaerobic glove box. Their concentrations based on quantitative dissolution were confirmed and calibrated with the Ellman thiol assay [23].

To ensure accurate quantification of redox events and redox equilibria, all chemicals and proteins in air-sensitive reduced forms were prepared, stored and handled under anaerobic conditions inside a glove box. Unless indicated otherwise, all experiments on thiol–disulfide exchanges between protein thiols and redox buffer GSSG/2GSH were also conducted under anaerobic conditions using thoroughly deoxygenated buffers.

Protein production and quantification

DNA sequences encoding the intact EcGrx1 and its variant EcGrx1-C14S were synthesized without the stop code, but with restriction sites NdeI and BamHI being incorporated at the N- and C-termini, respectively. The genes were digested and cloned into the pET20b expression vector (Novagen) digested with the same restriction enzymes. To facilitate protein purification with either a Strep tag or a hexa-His tag and then to cleave the affinity tags, a DNA sequence coding for a TEV protease site (underlined) followed by a Strep II affinity tag (in bold) with amino acid sequence ENLYFQGGWSHPQFEKG was inserted between the BamHI and XhoI sites in the pET20b vector. Consequently, both proteins EcGrx1-wt and EcGrx1-C14S were expressed with two consecutive affinity tags (i.e. a Strep II tag followed by a hexa-His tag) in their C-termini. The proteins were purified with metal-affinity resin loaded with Ni2+ ions. Both affinity tags were removed with a TEV protease, producing proteins with eight extra residues (GSENLYFQ) in their C-termini. They were further purified by a size-exclusion column Superdex-75 in MOPS buffer [50 mM (pH 7.0) and 100 mM NaCl].

Expression and purification of other proteins or protein domains were conducted, as reported previously: Atox1, the human copper metallo-chaperone [24]; CopC-H48C, a variant of the copper-binding protein CopC from Pseudomonas syringae [25]; and HsGrx1 and its variants HsGrx1-C23S, HsGrx1-C26S and a triple mutant HsGrx1-C8,79,83S (HsGrx1-tm) [22]. Similarly, a quadruple mutant HsGrx1-C8,26,79,83S (HsGrx1-qm) was generated and isolated. SeDsbA, the DsbA from Salmonella enterica serovar (sv.) Typhimurium, was donated kindly by Dr Begoña Heras (La Trobe University).

The identity and purity of each isolated protein were confirmed by electrospray ionization mass spectrometry (ESI-MS; Table 1) and SDS–PAGE (>95% purity). Each protein was fully reduced with DTT and buffer-exchanged using a desalting column with thoroughly deoxygenated MOPS buffer [50 mM (pH 7.0) and 100 mM NaCl] in an anaerobic glove box. Correct thiol content was determined in each case with Ellman's reagent 5,5-dithiobis(2-nitrobenzoic acid) (DTNB) based on the concentrations estimated from the solution absorbance at 280 nm (Table 1) [23].

Table 1
Characterization data for isolated proteins
Protein Molar mass (Da) ε (280) (M−1 cm−1)1 [–SH]/[protein] 
Found Expected Found Expected 
HsGrx1-wt 11644.7 11644.5 2980 4.7 
HsGrx1-C23S 11628.4 11628.4 2980 4.0 
HsGrx1-C26S 11628.2 11628.4 2980 3.9 
HsGrx1-tm2 11596.3 11596.3 2980 2.0 
HsGrx1-tm(SS) 11594.7 11594.3 3105 
HsGrx1-qm2 11581.0 11580.3 2980 1.0 
EcGrx1-wt3 10624.0 10623.8 12 950 2.0 
EcGrx1(SS) 10622.3 10621.8 12 975 
EcGrx1-C14S3 10609.9 10607.8 12 950 1.2 
Atox14 7271.1; 7403.5 7270.4; 7401.6 2980 3.2 
Atox1(SS)4 7269.2; 7401.2 7268.4; 7399.6 3005 1.0 
CopC-H48C 10501.3 10500.0 87005 1.0 
Protein Molar mass (Da) ε (280) (M−1 cm−1)1 [–SH]/[protein] 
Found Expected Found Expected 
HsGrx1-wt 11644.7 11644.5 2980 4.7 
HsGrx1-C23S 11628.4 11628.4 2980 4.0 
HsGrx1-C26S 11628.2 11628.4 2980 3.9 
HsGrx1-tm2 11596.3 11596.3 2980 2.0 
HsGrx1-tm(SS) 11594.7 11594.3 3105 
HsGrx1-qm2 11581.0 11580.3 2980 1.0 
EcGrx1-wt3 10624.0 10623.8 12 950 2.0 
EcGrx1(SS) 10622.3 10621.8 12 975 
EcGrx1-C14S3 10609.9 10607.8 12 950 1.2 
Atox14 7271.1; 7403.5 7270.4; 7401.6 2980 3.2 
Atox1(SS)4 7269.2; 7401.2 7268.4; 7399.6 3005 1.0 
CopC-H48C 10501.3 10500.0 87005 1.0 
1

Calculated from protein sequences with consideration of the cysteine redox status.

2

HsGrx1-tm and HsGrx1-qm refer to triple mutant HsGrx1-C8,79,83S and quadruple mutant HsGrx1-C8,26,79,83S, respectively.

3

Including an extra eight-residue (GSENLYFQ) C-terminus.

4

A mixture of two forms with and without the N-terminal Met residues.

5

Determined by dry-weight quantification as reported [25], instead by calculation from the protein sequence.

The oxidized proteins Atox1(SS), HsGrx1-tm(SS) and EcGrx1(SS) were generated by reaction with a slight excess of [FeIII(CN)6]3– and then buffer-exchanged with a desalting column to separate the protein component from the Fe complexes. CopC-H48C(SSG) was prepared by incubation of the reduced form with excess GSSG overnight, followed by buffer exchange with a desalting column.

Electrospray ionization mass spectrometry

All experiments were conducted on an Agilent time-of-flight mass spectrometer (TOF-MS; model 6220, Palo Alto, CA) coupled to an Agilent 1200 LC system with details given previously [22]. Control experiments demonstrated that, for the same protein, the integrated mass spectrum intensities for different oxidation states are proportional to their relative concentrations in solution. Consequently, the fraction of each component of the same protein was determined by integration of the mass spectral peak area of that component and then divided by the sum of the total peak areas for all components. All protein thiols were alkylated with excess iodoacetamide (IAA) for ESI-MS analysis. For each thiol group, a net mass of 57 Da for an acetamide group –CH2CONH2 is added to the detected molar mass of the protein target.

Determination of reduction potentials

Theoretical descriptions

The reduction potentials of protein thiols were determined via two approaches: (1) the poised potential method employing GSSG/2GSH as a redox buffer and (2) redox equilibration between two protein dithiols with different starting oxidation states.

For the first, the oxidation products of protein monothiols P(SH) and dithiols P(SH)2 in redox buffer GSSG/2GSH were demonstrated to be dominated by P(SSG) and P(SS), respectively:

 
formula
1
 
formula
2

where reaction (2) was proven to proceed via reactions (2a and 2b) with detectable intermediate species P(SH)(SSG) in most cases:

 
formula
2a
 
formula
2b

The half-cell reaction and reduction potential of the redox buffer GSSG/2GSH are described by the following equations:

 
formula
3a
 
formula
3b

where is the standard reduction potential at a fixed pH such as (vs. SHE) at a pH of 7.0 [26].

The standard reduction potentials of a protein monothiol P(SH) () or dithiol P(SH)2 () in the buffer may be determined according to respective half-cell reactions (4a and 5a), and calculated via eqns (4b and 5b), respectively:

 
formula
4a
 
formula
4b
 
formula
5a
 
formula
5b

In all experiments, the concentrations of GSSG and GSH were set much higher than those of the protein components to ensure that the couple GSSG/2GSH constitutes a redox buffer controlling the reduction potentials of the equilibria of eqns (1 and 2). Then, the terms [GSSG] and [GSH] in the equilibria may be approximated by their respective initial concentrations. The relative equilibrium concentrations for proteins ([P(SH)], [P(SSG)], [P(SH)2], [P(SS)] and [P(SH)(SSG)]) were determined in this work by ESI-MS analysis after the equilibrium position had been fixed by alkylation of the free thiols with excess IAA (IAA/ESI-MS). The alkylation process has been demonstrated previously to be much faster than the thiol–disulfide exchange [22] and is confirmed here again for both DsbA and Grx enzymes, even under the condition of a large thermodynamic driving force (see Supplementary Figure S1). The term Fred is the fraction of fully reduced protein forms to the sum of the fully reduced and the fully oxidized protein forms. The latter refers to P(SSG) for a protein monothiol and P(SS) for protein dithiols where the two thiol groups are vicinal in space to allow formation of a stable internal disulfide bond. In most cases, the semi-oxidized form P(SH)(SSG) has been shown to be an unstable minor component in equilibrium with both fully reduced and fully oxidized forms [22]. Its concentration is canceled out in the calculation of the reduction potentials for the major components P(SS)/P(SH)2 and thus is not included in the calculation [20].

For the second approach of redox equilibration, the difference between the reduction potentials of two protein dithiols P1 and P2 in equilibrium is defined by the following equations:

 
formula
6a
 
formula
6b

If the reduction potential of P1 is known, the reduction potential of P2 and other protein dithiols may be estimated via a series of pairwise redox equilibria according to eqns (6a and 6b). The equilibrium position can be analyzed by the IAA/ESI-MS approach.

Experimental protocols

In the first approach, each protein (∼10 µM) was incubated under anaerobic conditions in a series of redox buffers containing different mole ratios of GSSG/GSH but a fixed total concentration of [GSSG + 2 GSH] = 10 mM in KPi buffer [50 mM (pH 7.0) and 0.1 mM EDTA] at room temperature (∼22°C). The reduction potential of each redox buffer is controlled by the molar ratio GSSG/2GSH and calculated via eqn (3b) with at a pH of 7.0 [26]. The reactions reached equilibrium after 24 h incubation and were quenched by alkylation of free thiols with excess IAA (>10-fold). The ‘quenching’ was also attempted by adding H3PO4 to a final concentration of 300 mM, as commonly employed [1921], followed by further reaction with a mixing alkylating reagent containing IAA and K3PO4 to final concentrations of 100 and 300 mM, respectively. After incubation for >1 h, the oxidation level of each protein was estimated by ESI-MS analysis as detailed above.

For a protein monothiol P(SH), the oxidation product in GSSG/2GSH buffer was confirmed by ESI-MS to be P(SSG). The reduction potential for half-reaction (4a) was calculated via eqn (4b). For a protein dithiol P(SH)2, the dominant oxidation product is P(SS) and the reduction potential for half-reaction (5a) was calculated by eqn (5b).

Under equilibrium conditions, the reduction potential of each half-cell reaction in the same solution is identical, i.e. and . Consequently, the standard reduction potentials and at a pH of 7.0 were derived by curve-fitting of the experimental data to eqns (4b and 5b), respectively.

The second approach applied to protein dithiols only. The proteins in each pair (one fully oxidized and the other fully reduced) were incubated at a 1 : 1 molar ratio (each 10 µM) in deoxygenated KPi buffer (50 mM, pH 7.0) for 24 h under anaerobic conditions. At designated times, a small fraction was taken and mixed with excess IAA for ESI-MS analysis. A thiol–disulfide exchange equilibrium [expressed generally by eqn (6a)] was established in less than 8 h for those protein pairs involving at least one redox enzyme. The difference in reduction potentials for each protein pair was estimated by eqn (6b). The experiments on each system were undertaken from the two opposite directions of eqn (6a) and a consistent outcome guaranteed the reliability of the determination.

Glutathionylation of a protein monothiol in GSSG/2GSH buffer

The reactions were started by adding fully reduced CopC-H48C(SH) (final concentration, 10 µM) to an oxidation buffer containing GSSG (400 µM) and GSH (40 µM) in 50 mM MOPS buffer (pH 7.0), with and without a Grx enzyme (0.1 µM). The reactions were quenched along the reaction time course by transferring an aliquot of the reaction mixture (10 µl) into an IAA solution (5 µl; 50 mM in H2O; >50-fold excess). The protein oxidation product was detected by ESI-MS as the glutathionylated form CopC-H48C(SSG) only, whereas unreacted protein was trapped and detected as the alkylated product CopC-H48C(SA) (A = acetamide CH2CONH2 with a net mass increase by 57 Da). The degree of oxidation at each reaction time point was quantified by integration of the respective ESI-MS peak intensities and expressed as [P(SSG)]/[P]tot (here P = CopC-H48C).

Deglutathionylation of mixed disulfides in GSSG/2GSH buffer

The glutathionylated protein CopC-H48C(SSG) was used for the experiments. It was prepared by incubation of reduced CopC-H48C in excess GSSG overnight and purified by passing through a Bio-Rad P6 desalting column to remove excess GSSG. The reactions were started by adding CopC-H48C(SSG) (final concentration, 10 µM) to a reducing buffer consisting of GSSG (20 µM) and GSH (800 µM) in 50 mM MOPS buffer (pH 7.0), with or without a Grx enzyme (0.1 µM). The reactions were quenched at various reaction time points with IAA and quantified by ESI-MS analysis, as detailed above.

Results

Grx enzymes and protein substrates

HsGrx1 and EcGrx1 were the two representative Grx enzymes employed in this work. Different protein variants were generated for study of their reaction mechanisms. HsGrx1 contains a total of five Cys residues with two in the active-site motif Cys23-Pro-Tyr-Cys26 (Figure 1a,b). It has been demonstrated that (1) the three Cys residues Cys-8, -79, -83 outside the active site have no catalytic function and (2) Cys23 is the reactive thiol for the enzyme function, while the non-catalytic Cys26 may have a certain impact [22]. Consequently, three HsGrx1 protein variants were generated and isolated. They are the single variant HsGrx1-C23S, the triple variant HsGrx1-C8,79,83S (HsGrx1-tm, in which the three non-active-site Cys residues are replaced by Ser) and the quadruple variant HsGrx1-C8,26,79,83S (HsGrx1-qm in which, in addition, the C-terminal Cys of the active site is replaced by Ser).

EcGrx1 contains two Cys residues only, located within the active-site motif Cys11-Pro-Tyr-Cys14 (Figure 1c). The solvent-exposed N-terminal Cys11 is the reactive thiol, but the role of the C-terminal Cys14 is controversial (see later ‘Discussion’) [18,27]. Consequently, a protein variant EcGrx1-C14S was generated. Both EcGrx1-wt and EcGrx1-C14S were expressed and isolated as C-terminal His-tagged proteins, but the affinity tag was removed in the final purified samples.

CopC-H48C, a protein variant of the copper metallo-chaperone CopC from P. syringae, was chosen as a representative enzyme substrate. It contains a single Cys residue in a solvent-exposed flexible protein loop that can be glutathionylated completely by incubation with GSSG overnight (Figure 1d) [25]. The sulfur redox chemistry of copper metallo-chaperone human Atox1 has been studied recently [22]. It features a high-affinity Cu(I)-binding motif Cys12-Gly-Gly-Cys15 and was also isolated and used for comparative studies in selected cases in this work. Fully oxidized protein disulfides [HsGrx1-tm(SS), EcGrx1(SS) and Atox1(SS)] were prepared by quantitative oxidation of the respective reduced form with a slight excess of [FeIII(CN)6]3−, followed by separation from the iron complexes with a desalting column. The identity, purity and oxidation states for each of the above isolated protein samples were confirmed quantitatively by SDS–PAGE and ESI-MS analysis and by a cysteine thiol assay (see Table 1).

Standard reduction potentials for protein dithiols

The standard reduction potentials determined in this work are summarized in Table 2, together with available literature values. The potentials at a pH of 7.0 of the first four proteins in Table 2 have been reported previously. They are the DsbA from Gram-negative bacterium S. enterica serovar Typhimurium (denoted as SeDsbA), Grx enzymes EcGrx1 and HsGrx1-tm and metallo-chaperone Atox1. SeDsbA is included to aid calibration of the reduction potentials, since those of DsbA-type proteins appear to have been determined most reliably (see below). These four proteins feature a redox active CysN–xx–CysC motif with the oxidation products demonstrated to be dominated by the fully oxidized form featuring an internal disulfide bond, i.e. P(SS) (see Figure 2 and refs [2022]). Consequently, the reduction potentials for these proteins may be represented by eqns (5a and 5b) in terms of and . The value determined in this work for SeDsbA (−128 mV) matches that reported previously (−126 mV) [28], but the values for EcGrx1, HsGrx1-tm and Atox1 (−168, −169 and −191 mV, respectively) are somewhat different from literature estimates (see Table 2).

Redox equilibria and the oxidized and the reduced forms of dithiol proteins.

Figure 2.
Redox equilibria and the oxidized and the reduced forms of dithiol proteins.

Each protein (10 µM) was incubated for 24 h in deoxygenated KPi buffer [50 mM (pH 7.0) and 0.1 mM EDTA] containing redox buffer GSSG/GSH with given ratios of GSSG and GSH: (a) SeDsbA with GSSG (4.55 mM)/GSH (0.90 mM) (E, −129 mV); (b and c) HsGrx1-tm (b) and EcGrx1 (c) with GSSG (3.15 mM)/GSH (3.7 mM) (E, −170 mV); (d) Atox1 with GSSG (1.85 mM)/GSH (6.3 mM) (E, −191 mV). Each redox equilibrium was quenched by IAA (∼50 mM), followed by protein speciation analysis with ESI-MS.

Figure 2.
Redox equilibria and the oxidized and the reduced forms of dithiol proteins.

Each protein (10 µM) was incubated for 24 h in deoxygenated KPi buffer [50 mM (pH 7.0) and 0.1 mM EDTA] containing redox buffer GSSG/GSH with given ratios of GSSG and GSH: (a) SeDsbA with GSSG (4.55 mM)/GSH (0.90 mM) (E, −129 mV); (b and c) HsGrx1-tm (b) and EcGrx1 (c) with GSSG (3.15 mM)/GSH (3.7 mM) (E, −170 mV); (d) Atox1 with GSSG (1.85 mM)/GSH (6.3 mM) (E, −191 mV). Each redox equilibrium was quenched by IAA (∼50 mM), followed by protein speciation analysis with ESI-MS.

Table 2
Reduction potentials for monothiols () and dithiols () in certain proteins
Enzyme/protein Organisms Active site pKa7  (mV) at pH 7.0 
    This work8 This work9 Reported 
Dithiols 
 SeDsbA S. enterica Cys30-Pro-His-Cys33 3.3 [28−128 − −126 [28
 EcGrx1 E. coli Cys11-Pro-Tyr-Cys14 <5.0 [29−168 −168 −233 [19
 HsGrx1-tm H. sapiens Cys23-Pro-Tyr-Cys26 3.5 [30−169 −170 −118 [22]; −220 [21]; −232 [20
 Atox1 H. sapiens Cys12-Gly-Gly-Cys15 5.5 [31−191 −193 −188 [22]; −229 [21
Monothiols 
 EcGrx1-C14S E. coli Cys11-Pro-Tyr-Ser14 − −213   
 HsGrx1-qm H. sapiens Cys23-Pro-Tyr-Ser26 − −230   
 CopC-H48C P. syringae Cys48 − −247   
Enzyme/protein Organisms Active site pKa7  (mV) at pH 7.0 
    This work8 This work9 Reported 
Dithiols 
 SeDsbA S. enterica Cys30-Pro-His-Cys33 3.3 [28−128 − −126 [28
 EcGrx1 E. coli Cys11-Pro-Tyr-Cys14 <5.0 [29−168 −168 −233 [19
 HsGrx1-tm H. sapiens Cys23-Pro-Tyr-Cys26 3.5 [30−169 −170 −118 [22]; −220 [21]; −232 [20
 Atox1 H. sapiens Cys12-Gly-Gly-Cys15 5.5 [31−191 −193 −188 [22]; −229 [21
Monothiols 
 EcGrx1-C14S E. coli Cys11-Pro-Tyr-Ser14 − −213   
 HsGrx1-qm H. sapiens Cys23-Pro-Tyr-Ser26 − −230   
 CopC-H48C P. syringae Cys48 − −247   
1

Referred to reactive Cys thiol (highlighted in bold) in the active site.

2

Determined in GSSG/GSH buffer (10 mM) based on (vs. SHE) at a pH of 7.0, for protein monothiols is derived from eqns (4a and 4b), and for protein dithiols is derived from eqns (5a and 5b).

3

Determined from the equilibrium constants of a series of pairwise thiol–disulfide exchange reactions based on for SeDabA (see Table 3).

In each case discussed above, the reduction potentials were determined in GSSG/2GSH redox buffer at a pH of 7.0 via equilibrium 2 based on the same standard potential at a pH of 7.0. However, the approaches to quantify the equilibrium concentrations of the oxidized and the reduced protein forms varied, and this should be the main source of the disparity. Grx enzymes such as HsGrx1 are demonstrated to undertake rapid redox switching with t1/2 < 1 min upon reaction with GSSG or GSH (Supplementary Figure S2). Thus, the most reliable approach appears to be that reported for DsbA-type proteins, as these proteins emit an intense fluorescence at ∼330 nm upon excitation at 295 nm, and reduction of the active-site disulfide increases the fluorescence intensity further by several-fold [32]. Consequently, the reduction potentials of for the DsbA-type proteins can be determined directly and reliably via the change in the fluorescence intensity of a series of equilibrium solutions (eqn 2) without disturbing the redox equilibria. The reduction potentials for SeDsbA and for its E. coli equivalent EcDsbA were determined previously with this approach and are both relatively less negative (−126 and −122 mV, respectively) [28,33], consistent with their general functional role of facilitating oxidative folding of secreted proteins via catalysis of the formation of structural disulfide bonds [5,34,35].

In this work, using the IAA/ESI-MS approach (see Figures 2a and 3a(i)), the reduction potential estimated for SeDsbA (−128 mV) was within experimental error of the literature value (−126 mV). This confirmed previous evidence that alkylation of protein thiols by IAA is much faster than the thiol–disulfide exchange reaction [22]. This was confirmed here by a control experiment in which the addition of excess IAA with and without GSSG into a solution containing either SeDsbA or HsGrx1-tm at a pH of 7.0 detected the same ratio of oxidized to reduced protein forms for each protein (see Supplementary Figure S1). Consequently, the addition of excess IAA into a redox equilibrium system of protein thiols in GSSG/2GSH redox buffer is able to quench the redox reactions rapidly and to cause little disturbance to the established equilibrium.

Determination of reduction potentials in redox buffer GSSG/GSH.

Figure 3.
Determination of reduction potentials in redox buffer GSSG/GSH.

(a) For dithiol proteins: (i) SeDsbA, (ii) HsGrx1-tm and (iii) Atox1; (b) for monothiol proteins: (i) EcGrx1-C14S, (ii) HsGrx1-qm and (iii) CopC-H48C. Conditions: protein (∼10 µM) was incubated overnight under anaerobic conditions in a series of redox buffers GSSG/GSH (total [GSH + 2 GSSG] = 10 mM) in KPi (50 mM, pH 7.0) with reduction potentials defined by eqns (3a and 3b). The protein compositions at redox equilibrium were quantified by IAA/ESI-MS analysis as shown in Figure 2. Each set of experimental data was fitted to eqn (5b) for protein dithiols in (a) and to eqn (4b) for protein monothiols in (b) to derive the respective standard reduction potentials reported in Table 2.

Figure 3.
Determination of reduction potentials in redox buffer GSSG/GSH.

(a) For dithiol proteins: (i) SeDsbA, (ii) HsGrx1-tm and (iii) Atox1; (b) for monothiol proteins: (i) EcGrx1-C14S, (ii) HsGrx1-qm and (iii) CopC-H48C. Conditions: protein (∼10 µM) was incubated overnight under anaerobic conditions in a series of redox buffers GSSG/GSH (total [GSH + 2 GSSG] = 10 mM) in KPi (50 mM, pH 7.0) with reduction potentials defined by eqns (3a and 3b). The protein compositions at redox equilibrium were quantified by IAA/ESI-MS analysis as shown in Figure 2. Each set of experimental data was fitted to eqn (5b) for protein dithiols in (a) and to eqn (4b) for protein monothiols in (b) to derive the respective standard reduction potentials reported in Table 2.

We have recently estimated for Atox1 (−188 mV) and HsGrx1-tm (−118 mV) by the same approach, but in more dilute GSSG/2GSH redox buffer [22]. While the current value for Atox1 is similar to the previous value, the current one for HsGrx1-tm is ∼50 mV more negative. This result may be attributed to the specific binding affinity of HsGrx1 for the GSH molecule (Supplementary Table S1; cf. Figure 1b) [36,37], which would contribute to the value of and cause it to be dependent on the concentration of the redox buffer GSSG/2GSH. Then, it is necessary to consider both the pH and redox buffer concentration in quantitative experiments with the Grx enzymes. In comparison, the standard reduction potentials of the other proteins are less dependent or not dependent on the concentrations of the couple GSSG/2GSH (Supplementary Table S1).

The values reported previously for EcGrx1, HsGrx1-tm and Atox1 are more negative, falling within a narrow range between −220 and −233 mV (Table 2). These values were determined by quenching the reaction of eqn (2) with an acid such as H3PO4 (to a final concentration 300 mM) or trichloroacetic acid (10% w/v), followed by the estimation of the concentrations of oxidized and reduced forms by either HPLC elution or by cysteine-targeted labeling [1921]. We have repeated the ‘quenching’ procedure with H3PO4 and demonstrated that this acid was not able to quench the reaction completely. It promoted a shifting of the redox equilibrium (as anticipated) and caused protein thiol oxidation for all three redox enzymes (Supplementary Figure S3). This led to more negative apparent reduction potentials (Table 2).

The standard reduction potentials of protein dithiols () may also be quantified by direct thiol–disulfide exchange between two protein partners, based on eqns (6a and 6b). A set of pairwise protein–protein interaction experiments demonstrated that the exchange equilibrium can be approached after 24 h incubation for those protein pairs involving at least one thiol–disulfide oxidoreductase enzyme (Supplementary Figure S4). The differences in standard reduction potentials () for these protein pairs estimated via eqns (6a and 6b) matched those of the same protein pairs whose standard reduction potentials were determined independently in GSSG/2GSH redox buffer by the first approach (Tables 2 and 3). Such consistency between two independent determinations underscores the reliability of the approaches employed in this work. The data in Table 2 indicate that the standard reduction potentials for the protein dithiols increase in the following order:

 
formula
7

Standard reduction potentials for protein monothiols

Oxidation of a protein monothiol P(SH) in GSSG/2GSH buffer usually leads to glutathionylation of the isolated cysteinyl thiol (eqn 1). Oxidation of a protein dithiol P(SH)2 in GSSG/2GSH buffer also proceeds dominantly via glutathionylation of a single protein thiol, to yield the intermediate P(SH)(SSG), followed by further thiol–disulfide exchange to generate the fully oxidized protein form P(SS) (eqns 2a and 2b). The intermediate species was trapped and detected in some cases as a minor component (see Figure 2) [21,22].

To explore the catalytic mechanism of Grx enzymes that feature a CysN–x–x–CysC active site, the non-catalytic C-terminal Cys residue in the active-site motif of HsGrx1-tm and EcGrx1 was mutated to Ser to generate monothiol variants HsGrx1-qm and EcGrx1-C14S, respectively. Oxidation of these two monothiol Grx enzymes in GSSG/2GSH buffer generates the stable glutathionylated protein form P(OH)(SSG) only, and this allows reliable determination of the reduction potentials for the semi-oxidized forms, making the reasonable assumption that the potentials of P(SH)(SSG) and P(OH)(SSG) are similar. Equivalent experiments for a non-catalytic monothiol CopC-H48C were conducted for comparison.

Glutathionylation of protein monothiol P(SH) by GSSG.

Figure 4.
Glutathionylation of protein monothiol P(SH) by GSSG.

(a) Reaction progress and protein speciation for CopC-H48C (10 µM) detected by IAA/ESI-MS in MOPS buffer (50 mM, pH 7.0) containing GSSG (400 µM)/GSH (40 µM) and HsGrx1-tm (0.5 µM); (b) catalysis by HsGrx1-tm (0.8 µM) in buffer containing different ratios of GSSG:GSH; (c) catalytic rate is proportional to the Grx enzyme concentrations and a linear plot of the initial glutathionylation rate (µM/min) vs. the enzyme concentrations (µM) allowed estimation of an enzyme activity of 2.6 µM/min/µM enzyme for HsGrx1-tm and (d) catalysis by different enzymes (0.1 µM) under the conditions of (c).

Figure 4.
Glutathionylation of protein monothiol P(SH) by GSSG.

(a) Reaction progress and protein speciation for CopC-H48C (10 µM) detected by IAA/ESI-MS in MOPS buffer (50 mM, pH 7.0) containing GSSG (400 µM)/GSH (40 µM) and HsGrx1-tm (0.5 µM); (b) catalysis by HsGrx1-tm (0.8 µM) in buffer containing different ratios of GSSG:GSH; (c) catalytic rate is proportional to the Grx enzyme concentrations and a linear plot of the initial glutathionylation rate (µM/min) vs. the enzyme concentrations (µM) allowed estimation of an enzyme activity of 2.6 µM/min/µM enzyme for HsGrx1-tm and (d) catalysis by different enzymes (0.1 µM) under the conditions of (c).

The values determined in this work, based on eqns (4a and 4b) and at a pH of 7.0 were −213, −230 and −247 mV for EcGrx1-C14S, HsGrx1-qm and CopC-H48C, respectively (Figure 3b). These values provide the following order of increasing reduction potentials (i.e. order of decreasing stability of the disulfide bonds):

 
formula
8

Glutathionylation of protein monothiols

The variant CopC-H48C contains a single Cys residue in a solvent-exposed flexible loop (Figure 1c) [25]. Reactions were started by the addition of the variant (10 µM) into a solution of GSSG (400 µM) and GSH (40 µM) in MOPS buffer (50 mM, pH 7.0) containing a selected Grx enzyme (0.1–0.5 µM)0001. A control contained no enzyme. The reactions were quenched at various reaction time points by transferring an aliquot of reaction mixture (∼10 µl) into a microtube containing an excess of reagent IAA, followed by ESI-MS analysis of the protein compositions. The results demonstrated that the Grx enzyme catalyzed S-glutathionylation of CopC-H48C(SH) by GSSG according to eqn (9a) under thermodynamic control imposed by eqn (9b) (Figure 4a,b):

 
formula
9a
 
formula
9b
Table 3
Comparison of for protein dithiols determined via direct pairwise protein–protein interaction and via separate reactions with GSSG/GSH redox buffer10
Equilibrium of eqn (6a) Kex  (mV) from 
P1(SS) P2(SH)2 eqn (6b) directly eqn (2) indirectly 
SeDsbA(SS) EcGrx1(SH)2 22.8 40 40 
 HsGrx1-tm(SH)2 26.8 42 41 
EcGrx1(SS) HsGrx1-tm(SH)2 0.85 −2 
 Atox1(SH)2 7.21 25 23 
HsGrx1-tm(SS) EcGrx1(SH)2 0.78 −4 −1 
 Atox1(SH)2 7.01 25 22 
Equilibrium of eqn (6a) Kex  (mV) from 
P1(SS) P2(SH)2 eqn (6b) directly eqn (2) indirectly 
SeDsbA(SS) EcGrx1(SH)2 22.8 40 40 
 HsGrx1-tm(SH)2 26.8 42 41 
EcGrx1(SS) HsGrx1-tm(SH)2 0.85 −2 
 Atox1(SH)2 7.21 25 23 
HsGrx1-tm(SS) EcGrx1(SH)2 0.78 −4 −1 
 Atox1(SH)2 7.01 25 22 
1

Each pair of proteins (10 µM) was incubated in deoxygenated KPi buffer (50 mM, pH 7.0) for 24 h and then the protein compositions were analyzed by the IAA/ESI-MS approach (see Supplementary Figure S4).

The initial catalytic rate is proportional to the enzyme concentration (Figure 4c), allowing estimation of a catalytic activity of ∼2.6 µM/min/µM enzyme for HsGrx1-tm. The protein variant HsGrx1-C23S lacks detectable catalytic activity, while HsGrx1-qm (that retains only a single Cys23 in the active site) exhibited a catalytic activity of ∼7.1 µM/min/µM enzyme, i.e. ∼3 times that of HsGrx1-tm (Table 4). These experiments demonstrated that (1) the solvent-exposed N-terminal Cys23 in the active-site motif of HsGrx1 plays a pivotal role in catalysis of thiol–disulfide exchange and (2) the solvent-shielded C-terminal Cys26 in the active site suppresses enzyme activity, apparently via promoting formation of an internal disulfide with the N-terminal Cys23 thiol.

Table 4
Substrate turnover rates for CopC-H48C11
Enzyme Glutathionylation (µM/min/µM enzyme) Deglutathionylation (µM/min/µM enzyme) 
HsGrx1-C23S ∼0 – 
HsGx1-tm ∼2.6 ∼2.8 
HsGrx1-qm ∼7.1 ∼7.0 
EcGrx1 ∼3.2 ∼2.0 
EcGrx1-C14S ∼7.3 ∼8.5 
Enzyme Glutathionylation (µM/min/µM enzyme) Deglutathionylation (µM/min/µM enzyme) 
HsGrx1-C23S ∼0 – 
HsGx1-tm ∼2.6 ∼2.8 
HsGrx1-qm ∼7.1 ∼7.0 
EcGrx1 ∼3.2 ∼2.0 
EcGrx1-C14S ∼7.3 ∼8.5 
1

Reaction progress and protein speciation were analyzed by IAA/ESI-MS, and the background substrate turnover rate under the same condition without enzyme was subtracted in each case.

The above properties of HsGrx1 are shared by the EcGrx1 enzyme. It catalyzes reaction (9a) with activity comparable to that of HsGrx1-tm, while mutation of the C-terminal Cys14 to Ser increased the catalytic activity to a level comparable also to that of the equivalent human enzyme HsGrx-qm (Figure 4d and Table 4). Apparently, these Grx enzymes both employ a conserved monothiol mechanism for catalysis of glutathionylation of a protein monothiol.

Notably, the copper chaperone Atox1 also features an isolated monothiol (Cys41) that is buried within a panel of four rigid β-sheets [39]. We demonstrated recently that it is difficult to glutathionylate this protein monothiol [22], probably due to limited enzyme access. Therefore, it appears that S-glutathionylation of a protein monothiol P(SH) by a Grx enzyme at the expense of GSSG can only be significant for those protein thiols that are surface-exposed and accessible.

Deglutathionylation of mixed disulfides

In a redox buffer containing GSSG (20 µM) and GSH (800 µM), CopC-H48C(SSG) reacted spontaneously but slowly. The addition of a catalytic amount of Grx enzyme (0.1 µM) accelerated the deglutathionylation process considerably according to the following equation (Figure 5):

 
formula
10

This is the reverse reaction of eqn (9a) with the redox equilibrium still dictated by eqn (9b). In fact, the catalytic properties for reaction (10) of the different enzyme forms parallel those observed for reaction (9a) (cf. Figures 4 vs. 5). Both dithiols HsGrx1 and EcGrx1 catalyze the deglutathionylation reaction of eqn (10) with catalytic activities between 2 and 3 µM/min/µM enzyme. These are comparable to the activities for the reverse reaction (Table 4). Likewise, those of the monothiol variants HsGrx1-qm and EcGrx1-C14S are also comparable to each other and are higher than those of the corresponding dithiol Grxs (Table 4). This observation suggests that Grxs adopt conserved catalytic mechanisms for both glutathionylation and deglutathionylation, and that the N-terminal Cys serves as a catalytic center, while the C-terminal Cys may act as a catalytic brake.

Deglutathionylation of P(SSG) by Grx1/GSH.

Figure 5.
Deglutathionylation of P(SSG) by Grx1/GSH.

CopC-H48C(SSG) (10 µM) was incubated in MOPS buffer (50 mM, pH 7.0) containing GSSG (20 µM)/GSH (800 µM) and a Grx enzyme: (a) typical reaction progress detected by IAA/ESI-MS analysis; (b) catalysis by different enzymes (0.1 µM) under the above condition. Note: the peaks marked with * are due to an unknown modification to CopC protein after overnight incubation with GSSG; see Experimental Section.

Figure 5.
Deglutathionylation of P(SSG) by Grx1/GSH.

CopC-H48C(SSG) (10 µM) was incubated in MOPS buffer (50 mM, pH 7.0) containing GSSG (20 µM)/GSH (800 µM) and a Grx enzyme: (a) typical reaction progress detected by IAA/ESI-MS analysis; (b) catalysis by different enzymes (0.1 µM) under the above condition. Note: the peaks marked with * are due to an unknown modification to CopC protein after overnight incubation with GSSG; see Experimental Section.

Discussion

Major determinants of the reduction potential of a disulfide bond

The reduction potential for a disulfide/thiol couple is determined primarily by the conformation around the disulfide bond in the oxidized form and by the pKa of the thiol in the reduced form [40,41]. A strained disulfide will release conformational energy upon reduction and hence will have a more positive reduction potential [41]. Likewise, a lower thiol pKa indicates a stronger electron-withdrawing effect that will also induce a more positive reduction potential [40]. The dipeptide sequence between the two active-site Cys residues in a dithiol enzyme plays an important role in modulating the stability of the disulfide bond and the pKa of the reactive thiol and, consequently, the reduction potential [33,34]. These general principles help to rationalize the observed order of standard reduction potentials summarized in eqns (7 and 8).

For protein dithiols, an increase in reduction potentials from Atox1 to SeDsbA in eqn (7) correlates qualitatively with the decrease in the pKa of the reactive thiols (Table 2). In fact, mutation of either or both residues of the dipeptide sequence between the two active-site Cys residues in EcDsbA led to increases in the pKa of the reactive Cys thiols and decreases in the standard reduction potentials [33]. In particular, mutation of the active-site sequence Cys-Pro-His-Cys in EcDsbA to that of Cys-Pro-Tyr-Cys (commonly present in Grx proteins) led to a decrease in the reduction potential from −122 to −157 mV [33], matching well with the reduction potentials for the two Grx enzymes determined in this work (Table 2).

The differences in standard reduction potentials for the three glutathionylated disulfides P(SSG) were smaller (Table 2 and Figure 3b). There should be less conformational strain in cleaving this type of disulfide bond than the enclosed one in P(SS), and hence the difference in their reduction potentials will lie mainly on the variations in pKa of the Cys thiol being attacked. The lower pKa for the reactive thiol in Grx enzymes means that it is a better leaving group for reduction of the mixed disulfide, and thus, a disulfide bond involving Grx should have a more positive reduction potential. The standard reduction potential for EcGrx1-C14S(SSG) is somewhat less negative than that of HsGrx1-qm(SSG), but EcGrx1 precipitates readily at pH <5, and hence it was not possible to compare the pKa values of the reactive Cys thiols quantitatively for these two Grx enzymes.

for monothiols vs. for dithiols

For the same Grx protein, is ∼50 mV more negative than (Table 2). However, it must be emphasized that these are the standard reduction potentials for two different types of disulfide bonds (external and internal), and that their values cannot be compared simply without consideration of their definition. The term in eqn (4b) is the formal standard reduction potential for a glutathionylated disulfide bond and is equivalent to the term in eqn (5b) for an internal protein disulfide bond. The former varies with GSH concentration, while the latter does not (cf. eqns 4b and 5b)0002. Therefore, the redox states of related thiol–disulfide components in a redox system are more adequately represented by the half-cell reduction potential E′, calculated from the Nernst equation under specific conditions, and not by , especially for different types of disulfides [43].

We and others have demonstrated that essentially all dithiol Grx enzymes are capable of employing a monothiol mechanism in catalysis of thiol–disulfide exchanges [22,27,4447]. Consequently, the reduction potential will usually be more relevant than for describing and understanding the catalytic functions of these enzymes. Both for the redox buffer GSSG/2GSH and for the disulfide P(SSG)/(P(SH) + GSH) depend on log [GSH], but the former more quantitatively by a factor of 2 (cf. eqns 3b vs. 4b), i.e. the couple GSSG/2GSH provides a driving force for the redox event and Grxs are the relevant catalysts. This helps to explain why a given Grx enzyme may function as an oxidase for disulfide bond formation or as a reductase for disulfide breakage, depending not only on the GSSG/GSH ratio, but also on the GSH concentration.

On the other hand, the contribution of the term (–(RT/nF) ln [GSH]) in eqn (4b) to the reduction potential of makes it more positive than at [GSH] < 50 mM. This rationalizes why Grx(SH)(SSG) is usually not stable under normal cellular conditions and undergoes a spontaneous thiol–disulfide exchange to yield Grx(SS) and GSH. In fact, this spontaneous reaction inhibits the catalytic enzyme activity of Grxs (see below).

Glutathionylation and deglutathionylation of protein monothiols

The experimental evidence presented in Figures 4 and 5 is consistent with the mechanisms of closely related Schemes 1 and 2 for glutathionylation and deglutathionylation, respectively [27,36]. They consist of a series of thiol–disulfide exchanges that are typically second-order reactions involving bimolecular nucleophilic substitution (SN2) that may be expressed in the following equation [3,8]:

 
formula
11

Here, is the attacking thiolate and is the central sulfur group that interacts with the attacking thiolate to yield the product disulfide, while is the leaving group.

For the catalytic glutathionylation of Scheme 1, nucleophilic attack of the N-terminal reactive thiolate in Grx(SH)(S) on the GSSG disulfide bond is the initial process that leads to generation of intermediate Grx(SH)(SSG) via step (i). Following step (ii) formally transfers the activated cation radical [GS·]+ from the Grx enzyme to the protein substrate CopC-H48C(SH) to yield product CopC-H48C(SSG) and to regenerate the reduced enzyme Grx(SH)(S). Step (iii) is a competitive non-catalytic process in which the C-terminal Cys thiol in the active site launches an internal attack on the N-terminal sulfur to yield the inhibitory disulfide species Grx(SS) This species is estimated to be thermodynamically more stable than Grx(SH)(SSG) at [GSH] < 50 mM (vide supra). However, step (iii) is reversible and, although Grx(SS) is detected to be the dominant resting enzyme form (see Figure 2b,c), it is the unstable minor intermediate Grx(SH)(SSG) that plays the active role in catalysis. Mutation of the C-terminal Cys in the active site to Ser eliminates the possibility of formation of this inhibitory internal disulfide bond and enhances the enzyme catalytic activity (Table 4).

For the catalytic deglutathionylation of Scheme 2, the process starts with a nucleophilic attack of the N-terminal active Cys thiolate on the client disulfide bond P(SSG) (rather than on GSSG) to release protein thiol P(SH) and to generate the reactive enzyme intermediate Grx(SH)(SSG) (step i). The latter is unstable to external attack by GSH, and this generates the resting enzyme form Grx(SH)(S) (step ii) to complete the catalytic cycle. Likewise, an inhibitory non-catalytic step (iii) may compete with step (ii) to suppress enzyme activity. Removal of the C-terminal Cys residue in the active site eliminated this effect and enhanced the catalytic activity (Figure 5b and Table 4).

The catalytic processes depicted by Schemes 1 and 2 are closely related but distinct in the direction of reaction. This is decided by the relative concentrations of GSH and GSSG. Both catalytic processes proceed most effectively via a so-called monothiol ping-pong mechanism comprising two separate single redox steps [27,36]. Step (i) is facilitated by specific recognition of the GS fragment by the Grx enzymes, whereas step (ii) is driven by the fact that the reduced form of Grx is a good leaving group [36,37]. The reaction specificity of step (i) in Scheme 2 is consistent in that our experiments were not able to detect the putative protein complex Grx(SH)(SS)P.

The standard reduction potential of a mixed disulfide bond such as that in CopC-H48C(SSG) (−247 mV) is close to that of GSSG (−240 mV; Table 2), suggesting shallow thermodynamic gradients for both glutathionylation and deglutathionylation of a protein monothiol. Accordingly, the substrate turnover rates of these opposing reactions are similar for the same Grx enzyme (Figure 5b and Table 4). This is consistent with the high reversibility of glutathionylation and deglutathionylation. Cellular cytosols are highly reducing and dominated by high concentrations of GSH (1–10 mM) with low concentrations of GSSG (typically GSSG/GSH is ≤0.01) [48,49]. Consequently, cellular deglutathionylation of a protein monothiol should be a thermodynamically favorable process, while glutathionylation will not. However, the latter process may occur via GSH reduction of a protein monothiol that has been oxidized via other pathways [4,50].

Proposed ping-pong mechanism for glutathionylation of protein monothiol P(SH) by Grx/GSSG.

Scheme 1.
Proposed ping-pong mechanism for glutathionylation of protein monothiol P(SH) by Grx/GSSG.

Dashed lines indicate non-catalytic processes. The reactive sulfur atoms are shown in red and the disulfide bonds involved in exchange in orange.

Scheme 1.
Proposed ping-pong mechanism for glutathionylation of protein monothiol P(SH) by Grx/GSSG.

Dashed lines indicate non-catalytic processes. The reactive sulfur atoms are shown in red and the disulfide bonds involved in exchange in orange.

Proposed ping-pong mechanism for deglutathionylation of a protein glutathionyl disulfide P(SSG) by Grx/GSH.

Scheme 2.
Proposed ping-pong mechanism for deglutathionylation of a protein glutathionyl disulfide P(SSG) by Grx/GSH.

For intermediate Grx(SH)(SSG), competing nucleophilic attack by GSH and the C-terminal Cys thiol on the mixed disulfide bond leads to either regeneration of the active enzyme Grx(SH)(S) or formation of the internal disulfide Grx(SS).

Scheme 2.
Proposed ping-pong mechanism for deglutathionylation of a protein glutathionyl disulfide P(SSG) by Grx/GSH.

For intermediate Grx(SH)(SSG), competing nucleophilic attack by GSH and the C-terminal Cys thiol on the mixed disulfide bond leads to either regeneration of the active enzyme Grx(SH)(S) or formation of the internal disulfide Grx(SS).

Conclusions

This work determined the reduction potentials for protein disulfides P(SS)/P(SH)2 and for glutathionylated mixed disulfides P(SSG)/(P(SH)2 + GSH), and studied the glutathionylation and deglutathionylation functions of two representative Grx enzymes with the following conclusions:

  • (1) The two active-site Cys residues in both HsGrx1 and EcGrx1 shuttle rapidly between the reduced dithiol form Grx(SH)2 and the oxidized disulfide form Grx(SS) in GSSG/GSH buffer (Supplementary Figure S2), consistent with their oxidoreductase enzyme function in catalyzing GSH-dependent thiol–disulfide exchange reactions. A standard reduction potential of for this process was determined for both HsGrx1 and EcGrx1 (Figures 2 and 3). It differs from most literature values, but was confirmed by two independent determinations. The foundation for these approaches, namely quenching a thiol–disulfide redox equilibrium with excess IAA without disturbing the equilibrium, was confirmed (Supplementary Figure S1). The reduction potentials for Grx enzymes lie between those for the reducing thioredoxin enzymes (typically approximately −270 mV) and those for the oxidizing DsbA enzymes (typically approximately −125 mV) and are consistent with the dual functional roles of Grxs as either reductases or oxidases, depending on the driving force of the cellular redox buffer GSSG/GSH. The traditional ‘quenching’ approach with acid was shown to induce protein thiol oxidation that leads to more negative but misleading reduction potentials.

  • (2) The two Grx enzymes examined in this work catalyzed glutathionylation and deglutathionylation of protein monothiols more efficiently upon substitution of the C-terminal Cys residue in the active site with Ser, demonstrating that a monothiol ping-pong mechanism is employed by these enzymes (Schemes 1 and 2). Consequently, although the fully oxidized form Grx(SS) was detected as the dominant resting oxidized enzyme form (Figure 2), it is the unstable mono-glutathionylated intermediate Grx(SH)(SSG) that is the catalytically active form of the oxidized enzyme. The C-terminal Cys residue contributes negatively to the catalysis by allowing trapping of the oxidized enzyme in the inhibitory internal disulfide form.

  • (3) The reduction potential of the redox couple Grx(SH)(SSG)/[Grx(SH)2 + GSH] is most relevant for understanding the catalytic properties of Grxs employing GSSG/GSH as the co-substrates. This work estimated this value for the first time by employing variants HsGrx1-qm and EcGrx1-C14S that retain only a single reactive Cys residue. As expected, both values are slightly less negative than that for the redox couple GSSG/2GSH (−240 mV at pH 7). It is noted that, theoretically, the reduction potential for glutathionylated Grx(SSG) is dependent on GSH concentration, while for the disulfide Grx(SS) is not, but in practice, they both are (see Supplementary Table S1). This seems to rationalize the capacity of a Grx enzyme to act as either an oxidase or a reductase, depending on the solution conditions. It is necessary to consider both the pH and redox buffer concentration in quantitative experiments with the Grx enzymes. They have specific binding site(s) for GSH and GSSG, and this binding free energy contributes to the reduction potentials, causing them to be dependent on the concentration of the redox buffer GSSG/2GSH.

  • (4) The surface-exposed single Cys48 in CopC-H48C can be glutathionylated and deglutathionylated effectively by Grx enzymes, while the internal Cys41 residue in Atox1 cannot. Consequently, access to the Cys residue substrate is one of the key determinants in the catalysis of thiol–disulfide exchange reactions.

Abbreviations

     
  • DTT

    dithiothreitol

  •  
  • EcGrx1

    Escherichia coli Grx1

  •  
  • ESI-MS

    electrospray ionization mass spectrometry

  •  
  • Grx

    glutaredoxin

  •  
  • HsGrx1-tm

    a HsGrx1 triple mutant C8,79,83S

  •  
  • HsGrx1-qm

    a HsGrx1 quadruple mutant HsGrx1-C8,26,79,83S

  •  
  • GSH

    glutathione

  •  
  • GSSG

    glutathione disulfide

  •  
  • IAA

    iodoacetamide

Author Contribution

Z.X. and A.W. conceived the study. Z.X. and A.U. designed the experiments and A.U. performed them. A.U., A.B., A.W. and Z.X. discussed the experiments and interpreted the experimental data. Z.X. drafted the manuscript with discussion and contributions from all authors.

Funding

This work was supported by funds from the Australian Research Council Grant DP130100728. All ESI-MS analysis was conducted at Bio21 Mass Spectrometry and Proteomics Facility.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

*

The inclusion of a low concentration of product GSH in the reaction buffer avoids a very high initial driving force in the starting solution and promotes a steady oxidation process. Likewise, for reduction with GSH as reductant, a low level of GSSG was also included for similar reasons.

*

However, in practice, we detected that the reduction potentials of for the latter redox couple for Grx enzymes (and for HsGrx1 in particular) also depend somewhat on the GSH buffer concentration (see Table S1), likely due to the specific interaction between Grx and GSH.

References

References
1
Meyer
,
Y.
,
Buchanan
,
B.B.
,
Vignols
,
F.
and
Reichheld
,
J.-P.
(
2009
)
Thioredoxins and glutaredoxins: unifying elements in redox biology
.
Annu. Rev. Genet.
43
,
335
367
2
Hanschmann
,
E.-M.
,
Godoy
,
J.R.
,
Berndt
,
C.
,
Hudemann
,
C.
and
Lillig
,
C.H.
(
2013
)
Thioredoxins, glutaredoxins, and peroxiredoxins--molecular mechanisms and health significance: from cofactors to antioxidants to redox signaling
.
Antioxid. Redox Signal.
19
,
1539
1605
3
Netto
,
L.E.S.
,
de Oliveira
,
M.A.
,
Tairum
,
C.A.
and
da Silva Neto
,
J.F.
(
2016
)
Conferring specificity in redox pathways by enzymatic thiol/disulfide exchange reactions
.
Free Radic. Res.
50
,
206
245
4
Allen
,
E.M.G.
and
Mieyal
,
J.J.
(
2012
)
Protein-thiol oxidation and cell death: regulatory role of glutaredoxins
.
Antioxid. Redox Signal.
17
,
1748
1763
5
Shouldice
,
S.R.
,
Heras
,
B.
,
Walden
,
P.M.
,
Totsika
,
M.
,
Schembri
,
M.A.
and
Martin
,
J.L.
(
2011
)
Structure and function of DsbA, a key bacterial oxidative folding catalyst
.
Antioxid. Redox Signal.
14
,
1729
1760
6
Qi
,
Y.
and
Grishin
,
N.V.
(
2005
)
Structural classification of thioredoxin-like fold proteins
.
Proteins Struct. Funct. Bioinf.
58
,
376
388
7
Lillig
,
C.H.
,
Berndt
,
C.
and
Holmgren
,
A.
(
2008
)
Glutaredoxin systems
.
Biochim. Biophys. Acta, Gen. Subj.
1780
,
1304
1317
8
Deponte
,
M.
(
2013
)
Glutathione catalysis and the reaction mechanisms of glutathione-dependent enzymes
.
Biochim. Biophys. Acta, Gen. Subj.
1830
,
3217
3266
9
Begas
,
P.
,
Liedgens
,
L.
,
Moseler
,
A.
,
Meyer
,
A.J.
and
Deponte
,
M.
(
2017
)
Glutaredoxin catalysis requires two distinct glutathione interaction sites
.
Nat. Commun.
8
,
14835
10
Yang
,
Y.
,
Jao
,
S.-
,
Nanduri
,
S.
,
Starke
,
D.W.
,
Mieyal
,
J.J.
and
Qin
,
J.
(
1998
)
Reactivity of the human thioltransferase (glutaredoxin) C7S, C25S, C78S, C82S mutant and NMR solution structure of its glutathionyl mixed disulfide intermediate reflect catalytic specificity
.
Biochemistry.
37
,
17145
17156
11
Discola
,
K.F.
,
de Oliveira
,
M.A.
,
Rosa Cussiol
,
J.R.
,
Monteiro
,
G.
,
Bárcena
,
J.A.
,
Porras
,
P.
et al. 
(
2009
)
Structural aspects of the distinct biochemical properties of glutaredoxin 1 and glutaredoxin 2 from Saccharomyces cerevisiae
.
J. Mol. Biol.
385
,
889
901
12
Zhang
,
X.
,
Wang
,
W.
,
Li
,
C.
,
Zhao
,
Y.
,
Yuan
,
H.
,
Tan
,
X.
et al. 
(
2017
)
Structural insights into the binding of buckwheat glutaredoxin with GSH and regulation of its catalytic activity
.
J. Inorg. Biochem.
173
,
21
27
13
Johnson
,
W.M.
,
Golczak
,
M.
,
Choe
,
K.
,
Currran
,
P.L.
,
Gorelenkova Miller
,
O.
,
Yao
,
C.
et al. 
(
2016
)
Regulation of DJ-1 by glutaredoxin 1 in vivo - implications for Parkinson's disease
.
Biochemistry
PMID:
[PubMed]
14
Shao
,
D.
,
Han
,
J.
,
Hou
,
X.
,
Fry
,
J.
,
Behring
,
J.B.
,
Seta
,
F.
et al. 
(
2017
)
Glutaredoxin-1 deficiency causes fatty liver and dyslipidemia by inhibiting Sirtuin-1
.
Antioxid. Redox Signal.
PMID:
[PubMed]
15
De Benedetto
,
M.L.
,
Capo
,
C.R.
,
Ferri
,
A.
,
Valle
,
C.
,
Polimanti
,
R.
,
Carrì
,
M.T.
et al. 
(
2014
)
Glutaredoxin 1 is a major player in copper metabolism in neuroblastoma cells
.
Biochim. Biophys. Acta
1840
,
255
261
16
Cater
,
M.A.
,
Materia
,
S.
,
Xiao
,
Z.
,
Wolyniec
,
K.
,
Ackland
,
S.M.
,
Yap
,
Y.W.
et al. 
(
2014
)
Glutaredoxin1 protects neuronal cells from copper-induced toxicity
.
BioMetals
27
,
661
672
17
Lundström-Ljung
,
J.
and
Holmgren
,
A.
(
1995
)
Glutaredoxin accelerates glutathione-dependent folding of reduced ribonuclease A together with protein disulfide-isomerase
.
J. Biol. Chem.
270
,
7822
7828
18
Saaranen
,
M.J.
,
Salo
,
K.E.H.
,
Latva-Ranta
,
M.K.
,
Kinnula
,
V.L.
and
Ruddock
,
L.W.
(
2009
)
The C-terminal active site cysteine of Escherichia coli glutaredoxin 1 determines the glutathione specificity of the second step of peptide deglutathionylation
.
Antioxid. Redox Signal.
11
,
1819
1828
19
Åslund
,
F.
,
Berndt
,
K.D.
and
Holmgren
,
A.
(
1997
)
Redox potentials of glutaredoxins and other thiol-disulfide oxidoreductases of the thioredoxin superfamily determined by direct protein-protein redox equilibria
.
J. Biol. Chem.
272
,
30780
30786
20
Sagemark
,
J.
,
Elgán
,
T.H.
,
Bürglin
,
T.R.
,
Johansson
,
C.
,
Holmgren
,
A.
and
Berndt
,
K.D.
(
2007
)
Redox properties and evolution of human glutaredoxins
.
Proteins Struct. Funct. Bioinf.
68
,
879
892
21
Hatori
,
Y.
,
Clasen
,
S.
,
Hasan
,
N.M.
,
Barry
,
A.N.
and
Lutsenko
,
S.
(
2012
)
Functional partnership of the copper export machinery and glutathione balance in human cells
.
J. Biol. Chem.
287
,
26678
26687
22
Brose
,
J.
,
La Fontaine
,
S.
,
Wedd
,
A.G.
and
Xiao
,
Z.
(
2014
)
Redox sulfur chemistry of the copper chaperone Atox1 is regulated by the enzyme glutaredoxin 1, the reduction potential of the glutathione couple GSSG/2GSH and the availability of Cu(I)
.
Metallomics
6
,
793
808
23
Ellman
,
G.L.
(
1959
)
Tissue sulfhydryl groups
.
Arch. Biochem. Biophys.
82
,
70
77
24
Xiao
,
Z.
,
Brose
,
J.
,
Schimo
,
S.
,
Ackland
,
S.M.
,
La Fontaine
,
S.
and
Wedd
,
A.G.
(
2011
)
Unification of the copper(I) binding affinities of the metallo-chaperones Atx1, Atox1 and related proteins: detection probes and affinity standards
.
J. Biol. Chem.
286
,
11047
11055
25
Zhang
,
L.
,
Koay
,
M.
,
Maher
,
M.J.
,
Xiao
,
Z.
and
Wedd
,
A.G.
(
2006
)
Intermolecular transfer of copper ions from the CopC protein of Pseudomonas syringae. Crystal structures of fully loaded Cu(I)Cu(II) forms
.
J. Am. Chem. Soc.
128
,
5834
5850
PMID:
[PubMed]
26
Schafer
,
F.Q.
and
Buettner
,
G.R.
(
2001
)
Redox environment of the cell as viewed through the redox state of the glutathione disulfide/glutathione couple
.
Free Radic. Biol. Med.
30
,
1191
1212
27
Bushweller
,
J.H.
,
Aslund
,
F.
,
Wuthrich
,
K.
and
Holmgren
,
A.
(
1992
)
Structural and functional characterization of the mutant Escherichia coli glutaredoxin (C14.fwdarw.S) and its mixed disulfide with glutathione
.
Biochemistry
31
,
9288
9293
28
Heras
,
B.
,
Totsika
,
M.
,
Jarrott
,
R.
,
Shouldice
,
S.R.
,
Gunčar
,
G.
,
Achard
,
M.E.S.
et al. 
(
2010
)
Structural and functional characterization of three DsbA Paralogues from Salmonella enterica Serovar Typhimurium
.
J. Biol. Chem.
285
,
18423
18432
29
Aslund
,
F.
,
Ehn
,
B.
,
Miranda-Vizuete
,
A.
,
Pueyo
,
C.
and
Holmgren
,
A.
(
1994
)
Two additional glutaredoxins exist in escherichia coli: Glutaredoxin 3 is a hydrogen donor for ribonucleotide reductase in a thioredoxin/Glutaredoxin 1 double mutant
.
Proc. Natl Acad. Sci. U.S.A.
91
,
9813
9817
30
Mieyal
,
J.J.
,
Starke
,
D.W.
,
Gravina
,
S.A.
and
Hocevar
,
B.A.
(
1991
)
Thiotransferase in human red blood cells: kinetics and equilibrium
.
Biochemistry
30
,
8883
8891
31
Badarau
,
A.
and
Dennison
,
C.
(
2011
)
Copper trafficking mechanism of CXXC-containing domains: insight from the pH-dependence of their Cu(I) affinities
.
J. Am. Chem. Soc.
133
,
2983
2988
32
Wunderlich
,
M.
and
Glockshuber
,
R.
(
1993
)
Redox properties of protein disulfide isomerase (dsba) from Escherichia coli
.
Protein Sci.
2
,
717
726
33
Huber-Wunderlich
,
M.
and
Glockshuber
,
R.
(
1998
)
A single dipeptide sequence modulates the redox properties of a whole enzyme family
.
Fold. Des.
3
,
161
171
34
Grauschopf
,
U.
,
Winther
,
J.R.
,
Korber
,
P.
,
Zander
,
T.
,
Dallinger
,
P.
and
Bardwell
,
J.C.A.
(
1995
)
Why is DsbA such an oxidizing disulfide catalyst
?
Cell
83
,
947
955
35
Kadokura
,
H.
,
Katzen
,
F.
and
Beckwith
,
J.
(
2003
)
Protein disulfide bond formation in prokaryotes
.
Annu. Rev. Biochem.
72
,
111
135
36
Gravina
,
S.A.
and
Mieyal
,
J.J.
(
1993
)
Thioltransferase is a specific glutathionyl mixed-disulfide oxidoreductase
.
Biochemistry
32
,
3368
3376
37
Peltoniemi
,
M.J.
,
Karala
,
A.-R.
,
Jurvansuu
,
J.K.
,
Kinnula
,
V.L.
and
Ruddock
,
L.W.
(
2006
)
Insights into deglutathionylation reactions: different intermediates in the glutaredoxin and protein disulfide isomerase catalyzed reactions are defined by the γ-linkage present in glutathione
.
J. Biol. Chem.
281
,
33107
33114
38
Reference deleted.
39
Anastassopoulou
,
I.
,
Banci
,
L.
,
Bertini
,
I.
,
Cantini
,
F.
,
Katsari
,
E.
and
Rosato
,
A.
(
2004
)
Solution structure of the Apo and copper(I)-loaded human metallochaperone HAH1
.
Biochemistry
43
,
13046
13053
40
Roos
,
G.
,
Foloppe
,
N.
and
Messens
,
J.
(
2013
)
Understanding the pKa of redox cysteines: the key role of hydrogen bonding
.
Antioxid. Redox Signal.
18
,
94
127
41
Roos
,
G.
,
Fonseca Guerra
,
C.
and
Bickelhaupt
,
F.M.
(
2015
)
How the disulfide conformation determines the disulfide/thiol redox potential
.
J. Biomol. Struct. Dyn.
33
,
93
103
42
However, in practice, we detected that the reduction potentials of for the latter redox couple for Grx enzymes (and for HsGrx1 in particular) also depend somewhat on the GSH buffer concentration (see Table S1), likely due to the specific interaction between Grx and GSH
43
Watson
,
W.H.
,
Chen
,
Y.
and
Jones
,
D.P.
(
2003
)
Redox state of glutathione and thioredoxin in differentiation and apoptosis
.
BioFactors
17
,
307
314
44
Yang
,
Y.F.
and
Wells
,
W.W.
(
1991
)
Identification and characterization of the functional amino acids at the active center of pig liver thioltransferase by site-directed mutagenesis
.
J. Biol. Chem.
266
,
12759
12765
PMID:
[PubMed]
45
Yang
,
Y.F.
and
Wells
,
W.W.
(
1991
)
Catalytic mechanism of thioltransferase
.
J. Biol. Chem.
266
,
12766
12771
PMID:
[PubMed]
46
Du
,
Y.
,
Zhang
,
H.
,
Zhang
,
X.
,
Lu
,
J.
and
Holmgren
,
A.
(
2013
)
Thioredoxin 1 is inactivated due to oxidation induced by peroxiredoxin under oxidative stress and reactivated by the glutaredoxin system
.
J. Biol. Chem.
288
,
32241
32247
47
Bouldin
,
S.D.
,
Darch
,
M.A.
,
Hart
,
P.J.
and
Outten
,
C.E.
(
2012
)
Redox properties of the disulfide bond of human Cu,Zn superoxide dismutase and the effects of human glutaredoxin 1
.
Biochem. J.
446
,
59
67
48
Hwang
,
C.
,
Sinskey
,
A.J.
and
Lodish
,
H.F.
(
1992
)
Oxidized redox state of glutathione in the endoplasmic reticulum
.
Science
257
,
1496
1502
49
Morgan
,
B.
,
Ezerin¸a
,
D.
,
Amoako
,
T.N.E.
,
Riemer
,
J.
,
Seedorf
,
M.
and
Dick
,
T.P.
(
2013
)
Multiple glutathione disulfide removal pathways mediate cytosolic redox homeostasis
.
Nat. Chem. Biol.
9
,
119
125
50
Gallogly
,
M.M.
and
Mieyal
,
J.J.
(
2007
)
Mechanisms of reversible protein glutathionylation in redox signaling and oxidative stress
.
Curr. Opin. Pharmacol.
7
,
381
391