The dithiol mechanism of class I glutaredoxins promotes specificity for glutathione as a reducing agent

Abstract

Class I glutaredoxins reversibly reduce glutathione- and nonglutathione disulfides with the help of reduced glutathione (GSH) using either a monothiol mechanism or a dithiol mechanism. The monothiol mechanism exclusively involves a single glutathionylated active-site cysteinyl residue, whereas the dithiol mechanism requires the additional formation of an intramolecular disulfide bond between the active-site cysteinyl residue and a resolving cysteinyl residue. While the oxidation of glutaredoxins by glutathione disulfide substrates has been extensively characterized, the enzyme-substrate interactions for the reduction of S-glutathionylated glutaredoxins or intramolecular glutaredoxin disulfides are still poorly characterized. Here we compared the thiol-specificity for the reduction of S-glutathionylated glutaredoxins and the intramolecular glutaredoxin disulfide. We show that S-glutathionylated glutaredoxins rapidly react with a plethora of thiols and that the 2nd glutathione-interaction site of class I glutaredoxins lacks specificity for GSH as a reducing agent. In contrast, the slower reduction of the partially strained intramolecular glutaredoxin disulfide involves specific interactions with both carboxylate groups of GSH at the 1st glutathione-interaction site. Thus, the dithiol mechanism of class I glutaredoxins promotes specificity for GSH as a reducing agent, which might explain the prevalence of dithiol glutaredoxins in pro- and eukaryotes.

Graphical abstract

We compared the reduction of glutaredoxin disulfides and glutathionylated glutaredoxins revealing different specificities of both enzyme species for GSH as a reducing agent.

Highlights

• •Glutathionylated glutaredoxins (Grx) lack specificity for GSH as a reducing agent.
• •The Grx dithiol mechanism promotes specificity for GSH as a reducing agent.
• •The different specificities result from alternative glutathione interaction sites.
• •The extent of intramolecular disulfide bond formation determines GSH specificity.
• •Grx are optimized for glutathione recruitment while preventing inhibition by GSH.

1. Introduction

The specificity for glutathione (γ-glutamylcysteinyl glycine) distinguishes class I glutaredoxins (EC 1.8.4.1–4) from other thiol:disulfide oxidoreductases of the thioredoxin superfamily [[1], [2], [3], [4], [5], [6]]. While glutaredoxins can reduce some nonglutathione disulfides using reduced glutathione (GSH) as an electron donor in a dithiol mechanism similar to thioredoxins (Fig. 1A) [1,[5], [6], [7], [8], [9], [10], [11], [12]], their preferred disulfide substrates appear to be mixed glutathione disulfides (RSSG) [[2], [3], [4], [5], [6],[13], [14], [15]]. Nonglutathione disulfide substrates include low-molecular-weight model substrates such as bis(2-hydroxyethyl)disulfide and specific intra- or intermolecular protein disulfides such as oxidized Escherichia coli ribonucleotide reductase, whereas mixed glutathione disulfide substrates include low-molecular-weight disulfides such as l-cysteine-glutathione disulfide (GSSCys) and S-glutathionylated proteins [10,11,14,[16], [17], [18]]. The reversible reduction of glutathionylated substrates proceeds via a ping-pong monothiol mechanism and requires only one cysteinyl residue in the glutaredoxin active site (Fig. 1B). During the oxidative half-reaction, the conserved active-site cysteinyl thiolate attacks the disulfide bond of a glutathionylated substrate, releasing a reduced thiol upon formation of the mixed disulfide between glutathione and the glutaredoxin. The S-glutathionylated glutaredoxin then reacts with GSH during the rate-limiting reductive half-reaction, regenerating the reduced enzyme and releasing glutathione disulfide (GSSG) [[3], [4], [5], [6], [7],13,14,[19], [20], [21]]. Distinct glutathione interaction sites for both substrates were proposed to facilitate efficient catalysis and the specificity for mixed glutathione disulfides and GSH during each half-reaction (Fig. 1C and D) [5,[20], [21], [22], [23]].

Fig. 1.

Catalytic mechanism of class I glutaredoxins. A) Dithiol mechanism for the reversible reduction of nonglutathione disulfides. B) Monothiol mechanism for the reversible reduction of glutathionylated substrates. C) Top view of a model of monothiol class I ScGrx7 with glutathione (GS) at the 1st glutathione-interaction site (left). Residues that were shown to contribute to the 1st glutathione-interaction site are highlighted in salmon (middle). Residues and GS that were shown to form the or to recruit GSH to the transient 2nd glutathione-interaction site are shown in blue (right). D) Schematic representation of a side view of the two distinct glutathione-interaction sites that bind or rather encounter the two glutathione moieties during both half-reactions. E) The dithiol mechanism could resolve sterically blocked enzyme intermediates and reintroduce dead-end species in the catalytic cycle. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

The oxidative half-reaction requires the 1st glutathione-interaction site (previously termed glutathione-scaffold site) of the glutaredoxin to interact with the glutathione moiety but not the rest of the mixed glutathione disulfide substrate (Fig. 1B–D) [5,[20], [21], [22], [23], [24]]. These interactions provide an optimized substrate orientation and reaction geometry for the attack of the glutaredoxin cysteinyl thiolate on the sulfur atom of the glutathione moiety [5,[20], [21], [22], [23]]. In this way, the 1st glutathione-interaction site facilitates the high specificity of glutaredoxins for S-glutathionylated proteins and other glutathionylated substrates, the efficient reduction of a plethora of different mixed glutathione disulfides, and the preferential release of RSH instead of GSH during the oxidative half-reaction [14,[25], [26], [27]]. Analyses of the structures of S-glutathionylated glutaredoxins, the binding of glutathionylated substrates to the 1st glutathione-interaction site, and the kinetic parameters of mutant enzymes led to the identification of several residues that are part of this well-defined site (Fig. 1C) [4,5,8,20,21,25,[28], [29], [30]]. GSH can also bind to the 1st glutathione-interaction site in a similar manner [23,31]. Analogous to glutathione transferases, this interaction could activate GSH as a nucleophile by deprotonation to GS^– and facilitate its attack on protein disulfides [16,23,32,33]. In redox-inactive class II glutaredoxins, GS^– also binds to the 1st interaction site, functioning as a ligand for iron-sulfur clusters [5,6,20,29,34,35]. While the initial reaction between class I glutaredoxins and glutathionylated substrates at the 1st glutathione-interaction site has been studied extensively (including GSSCys and S-glutathionylated proteins), two central questions about glutaredoxin catalysis remain unanswered. Why are glutaredoxins specific for GSH as a reducing agent for the reductive half-reaction? Why do most class I glutaredoxins possess a CxxC-motif with a second (resolving) cysteinyl residue following the active-site cysteinyl residue when deglutathionylation reactions only require a monothiol mechanism?

In contrast to the well-characterized 1st glutathione-interaction site and substrate specificity during the oxidative half-reaction, enzyme-substrate interactions during the reductive half-reaction remain to be deciphered. Three challenges have complicated the characterization of the 2nd glutathione-interaction site (previously termed glutathione-activator site) to date: (i) Due to its low pK_a value, the glutaredoxin active-site cysteinyl residue is an ideal leaving group and its mixed disulfide with glutathione is intrinsically highly reactive [4,19,36,37]. Thus, for the reduction of the glutathionylated glutaredoxin, the 2nd glutathione-interaction site is only required to provide transient interactions for efficient recruitment, orientation, and deprotonation of GSH instead of a fixed binding site [4,5,22]. In agreement with this, kinetic measurements with mutant enzymes and MD simulations did not reveal a clearly defined binding site for GSH during the reductive half-reaction [20,21]. However, charged residues in helix 3 have been suggested to play a crucial role for GSH recruitment as part of the 2nd glutathione-interaction site, together with a highly conserved lysyl residue and the already bound glutathione moiety (Fig. 1C) [8,20,21,29]. (ii) Different glutaredoxins vary regarding their apparent K_m values and specificity for GSH [[19], [20], [21], [22],27,36,[38], [39], [40], [41]], suggesting isoform-dependent enzyme-substrate interactions. While human Grx1 and Grx2 can accept various thiols with a 10–20-fold enhancement of deglutathionylation activity for GSH compared to other low-molecular weight thiols [19,36], E. coli Grx1 (EcGrx1) has a deglutathionylation activity of <1 % when the γ-glutamyl residue of GSH is replaced by an α-glutamyl residue [27]. (iii) Most class I glutaredoxins have a CxxC-motif with a second cysteinyl residue close to the active-site cysteinyl residue and can form an intramolecular disulfide during the reduction of glutathionylated substrates (Fig. 1B) [7,10,14,23,25,26,30,42,43]. This intramolecular disulfide is then reduced in consecutive reactions by two molecules of GSH with the S-glutathionylated enzyme species as an intermediate. Depending on the extent of this side reaction, GSH interacts with two different enzyme species (the glutaredoxin disulfide and the 2nd glutathione-interaction site of the S-glutathionylated enzyme) therefore complicating the analysis of kinetic data.

The dithiol mechanism and the side reaction of the monothiol mechanism are a poorly understood property of class I glutaredoxins. While the reduction of E. coli ribonucleotide reductase requires a dithiol-disulfide exchange mechanism [[10], [11], [12]], other protein disulfides such as redox-sensitive green fluorescent protein 2 (roGFP2) or yeast ribonucleotide reductase are efficiently reduced in a monothiol mechanism in vitro and in vivo [23,32,43,44]. Furthermore, for the monothiol mechanism, the formation and resolution of the intramolecular glutaredoxin disulfide introduce two seemingly redundant reaction steps that detract from catalysis [[3], [4], [5],13,14,30,36]. According to the cysteine-resolving model, the side reaction might be able to resolve conformationally trapped mixed glutaredoxin disulfides that cannot be attacked by GSH (Fig. 1E) [5,22,23]. The reintroduction of such dead-end species into the catalytic cycle could come at the expense of optimal catalytic efficiency in the monothiol mechanism. However, detailed reaction kinetics and enzyme-substrate interactions between GSH and intramolecular glutaredoxin disulfides have not been analyzed so far.

In this study, we addressed the thiol-specificity for the reduction of S-glutathionylated glutaredoxins as compared to the intramolecular glutaredoxin disulfide using stopped-flow and steady-state kinetic measurements. Our data reveal strikingly different GSH specificities of both enzyme species. While S-glutathionylated glutaredoxins rapidly react with a plethora of thiols with second-order rate constants around ≤2.5 × 10⁵ M⁻¹s⁻¹, the reduction of the intramolecular glutaredoxin disulfide involves specific GSH interactions at the 1st glutathione-interaction site with a second-order rate constant of (5–7) × 10⁴ M⁻¹s⁻¹. Thus, the dithiol mechanism of class I glutaredoxins promotes specificity for GSH as a reducing agent.

2. Methods

2.1. Materials

CysGly and GABA-SH were synthesized as described previously [45]. Dithiothreitol (DTT), 3-(dimethylcarbamoylimino)-1,1-dimethylurea (diamide), 2-mercaptoethanol (2-ME), thioglycolic acid (TGA), GSH, GSSG, and yeast glutathione reductase (GR) were from Sigma-Aldrich, GSSCys was from Toronto Research Chemicals, diethylenetriaminepentaacetic acid (DTPA) and l-cysteine (Cys) were from Carl Roth, l-cysteine ethyl ester hydrochloride (CysOEt) was from Acros organics, N-acetyl l-cysteine (NAC) was from Alfa Aesar, glutathione ethyl ester (GSHOEt) was from Cayman Chemicals, cysteamine hydrochloride (CA) was from AppliChem, and NADPH was from Gerbu. PCR primers were purchased from Metabion.

2.2. Site-directed mutagenesis

SCGRX7^G107W was generated by site-directed mutagenesis PCR with Phusion HF polymerase (New England Biolabs) using pQE30/SCGRX7^WT as template [38], and mutagenesis primers 5′-GTATTTAGCAAGACTTGGTGCCCATATAGC-3′ and 5′-GCTATATGGGCACCAAGTCTTGCTAAATAC-3'. The methylated template DNA was digested by DpnI (New England Biolabs), and the PCR product was transformed into chemically competent E. coli XL1-Blue cells. Plasmid DNA was isolated by minipreparation and correct mutations and sequences for all constructs were confirmed by Sanger sequencing (Microsynth Seqlab).

2.3. Heterologous expression, protein purification, and sample preparation

Recombinant N-terminally MRGSH₆GS-tagged PfGrx^E28W/C88S, PfGrx^{E28W/C32S/C88S}, ScGrx7^WT, ScGrx7^E147K, and ScGrx7^G107W were produced in E. coli strain XL1-Blue and purified by Ni-NTA affinity chromatography as described previously [20,21,46]. The purity of the eluates was confirmed by analytical SDS-PAGE and the protein concentrations were determined spectrophotometrically at 280 nm. To obtain the fully reduced proteins, the eluate was incubated with 5 mM DTT for 30 min on ice followed by removal of DTT and imidazole on a PD-10 column (Merck). The reduced proteins were eluted with 3.5 mL ice-cold assay buffer (100 mM Na_xH_yPO₄, 0.1 mM DTPA, pH 7.4 at 25 °C). S-glutathionylated PfGrx^{E28W/C32S/C88S} and ScGrx7^G107W were produced by incubation of the reduced enzymes with 5 mM GSSG for 60 min on ice. The intramolecular PfGrx^E28W/C88S disulfide was produced by incubation of the reduced enzyme with 5 mM diamide for 60 min on ice. All oxidized enzyme species were purified again by Ni-NTA affinity chromatography followed by a buffer exchange on a PD-10 column, and final protein concentrations were determined as described above.

2.4. Stopped-flow kinetic measurements

Stopped-flow measurements were performed in a thermostatted SX-20 spectrofluorometer (Applied Photophysics) at 25 °C. The change of tryptophan fluorescence was measured for up to 60 s after mixing (total emission at an excitation wavelength of 295 nm with a slit width of 2 mm). The deglutathionylation of PfGrx^{E28W/C32S/C88S} and ScGrx7^G107W was investigated by mixing 2 μM of S-glutathionylated enzyme in syringe 1 with variable concentrations of low-molecular-weight thiols in assay buffer in syringe 2. For the reduction of the intramolecular PfGrx^E28W/C88S disulfide, 2 μM diamide-oxidized PfGrx^E28W/C88S in syringe 1 was mixed with variable concentrations of low-molecular-weight thiols in syringe 2. Traces of three consecutive measurements were averaged and fitted by single exponential regression (or double exponential regression for GSHOEt) using the Pro-Data SX software (Applied Photophysics) to obtain k_obs values. The k_obs values of three biological replicates were plotted against the substrate concentration in SigmaPlot 13.0 to obtain second order rate constants from the slopes of the linear fits (or apparent second order rate constants from hyperbolic fits for GSH).

2.5. Steady-state kinetic measurements

Oxidoreductase activities of ScGrx7^WT and ScGrx7^E147K were determined in a GSSCys assay in a thermostatted Jasco V-650 UV/Vis spectrophotometer at pH 8.0 and 25 °C as described previously [20,21]. NADPH (0.1 mM), GSH or GABA-SH (up to 1.5 mM), yeast GR (1 U/ml), and ScGrx7^WT or ScGrx7^E147K were mixed in a cuvette and a baseline was recorded for 30 s. The assay was then started by the addition of GSSCys. For GSH, fixed GSSCys concentrations were 25, 50, and 100 μM. For GABA-SH, fixed GSSCys concentrations were 50, 100, and 150 μM. The absorbance of a reference cuvette containing all assay components except the glutaredoxin was measured in parallel and subtracted. Initial slopes of the reaction kinetics were analyzed in SigmaPlot 13.0 according to Michaelis-Menten and Lineweaver-Burke theory.

2.6. Circular dichroism spectroscopy

Circular dichroism spectra were measured in a thermostatted Chirascan V100 CD spectrophotometer (Applied Photophysics) at 25 °C. Reduced and diamide-oxidized PfGrx^E28W/C88S were prepared in ice-cold buffer (10 mM Na_xH_yPO₄, 0.01 mM DTPA, pH 7.4 at 25 °C) and diluted to a protein concentration of 16.5 μM. CD spectra were recorded between 250 and 180 nm and the spectra of two consecutive measurements were averaged.

3. Results

3.1. The 2nd glutathione-interaction site lacks specificity for the reducing agent

Previous studies suggested a preference of class I glutaredoxins for GSH as a reducing agent [19,27,36], although molecular dynamics simulations did not reveal a clear binding pathway and stable enzyme-substrate interactions between S-glutathionylated glutaredoxins and GSH [20]. We therefore analyzed the reductive half-reaction between S-glutathionylated monothiol glutaredoxins and various low-molecular-weight thiols as reducing agents in stopped-flow kinetic measurements (Fig. 2A and B). We used monothiol PfGrx^{E28W/C32S/C88S}, which was generated previously [46], and ScGrx7^G107W, in which we introduced a tryptophan residue as a fluorophore to monitor changes in the redox state of the active-site cysteinyl residue. Residue E28W in PfGrx is at the same position as residue G107W in ScGrx7 (Fig. 1C). Since several class I glutaredoxins have a tryptophan residue at the same position before the active-site cysteinyl residue [39,47], we did not expect drastically altered activities or substrate specificities in the mutant enzymes. Residue C32 is the second cysteinyl residue in the CxxC-motif of PfGrx and C88 is a third cysteinyl residue in a semi-conserved GGC-motif [48]. These two additional cysteinyl residues are missing in monothiol class I glutaredoxin ScGrx7. Mixing of S-glutathionylated PfGrx^{E28W/C32S/C88S} in the first syringe with variable concentrations of GSH in the second syringe resulted in a GSH-dependent monophasic increase in fluorescence with a second-order rate constant of 2.5 × 10⁵ M⁻¹s⁻¹ at pH 7.4 (Fig. 2C and D) in accordance with reciprocal Dalziel coefficients of 1.5 × 10⁵ M⁻¹s⁻¹ from previous steady-state measurements with PfGrx^C32S/C88S at pH 8.0 [21]. We then tested N-acetyl l-cysteine (NAC), l-cysteine ethyl ester (CysOEt), l-cysteine (Cys), cysteamine (CA), thioglycolic acid (TGA), and 2-mercaptoethanol (2-ME) as alternative reducing agents with different sizes, functional groups, charges, and pK_a values. All thiols revealed similar monophasic reaction patterns as observed with GSH (Fig. S1). Plotting the second-order rate constants against the thiol pK_a showed highest activity for CysOEt at pK_a = 7.5 and GSH at pK_a = 9.0, and a decreasing activity for 2-ME > NAC > TGA at increasing pK_a values (Fig. 2E–Table 1). Since the thiolate is the actual nucleophile in this reaction, we also plotted the pH-independent rate constants against the pK_a values to correct for differences in the protonation state [49]. We detected a linear correlation with a Brønsted coefficient β_nuc of 0.65, suggesting a shared reaction mechanism for all thiols. However, the pH-independent second-order rate constant for GSH of 1.0 × 10⁷ M⁻¹s⁻¹ was twice as high as expected, indicating a slight preference of S-glutathionylated PfGrx for GSH as a reducing agent. For S-glutathionylated ScGrx7^G107W we also detected monophasic thiol-dependent increases in fluorescence with similar second-order rate constants (Fig. 2F–H, Fig. S2, Table 1). The second-order rate constant of 2.6 × 10⁵ M⁻¹s⁻¹ for GSH (Fig. 2G) was again in accordance with reciprocal Dalziel coefficients of (2.2–4.7) × 10⁵ M⁻¹s⁻¹ from previous steady-state measurements with ScGrx7 at pH 8.0 [20,21]. However, in contrast to PfGrx^{E28W/C32S/C88S}, S-glutathionylated ScGrx7^G107W showed no preference for GSH as a reducing agent and a higher Brønsted coefficient β_nuc of 0.80 (Fig. 2H). In summary, S-glutathionylated monothiol glutaredoxins are efficiently reduced by a variety of low-molecular-weight thiols, irrespective of their size, charge or specific functional groups. The 2nd glutathione-interaction site of class I glutaredoxins therefore has no or only minor specificity for GSH as a reducing agent.

Fig. 2.

Reductive half-reaction between glutathionylated monothiol class I glutaredoxins and various low-molecular-weight thiols. A) Schematic representation of the predicted transition state of the reductive half-reaction. B) Structures and color coding of the tested thiols. C) Representative monophasic stopped-flow reduction kinetics for the reaction between 1 μM S-glutathionylated monothiol PfGrx^{E28W/C32S/C88S} and variable concentrations of GSH. D) Secondary plot for the k_obs values from the single exponential fits for GSH from panel C. E) Brønsted plot of the second-order rate constants from panel D and Fig. S1 (squares) and normalized, pH-independent second-order rate constants (circles). The nucleophile Brønsted coefficient β_nuc was determined from the slope of the linear fit. F–H) Monophasic stopped-flow reduction kinetics, secondary plot, and Brønsted plot for ScGrx7^G107W as in panels C–E. Kinetic traces and secondary plots for the other thiols are shown in Fig. S1 and Fig. S2. All data sets were generated from at least three independent biological replicates with three technical replicates. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Table 1.

| Rate constants for the reduction of oxidized glutaredoxins at 25 °C.

Glutaredoxin	Substrate	k (M−1s−1) pH-dependent	k (M−1s−1) pH-independent
PfGrxE28W/C32S/C88S(SSG)	GSH	2.5 × 105 M−1s−1	1.0 × 107 M−1s−1
PfGrxE28W/C32S/C88S(SSG)	CysOEt	2.7 × 105 M−1s−1	6.1 × 105 M−1s−1
PfGrxE28W/C32S/C88S(SSG)	Cys	1.7 × 105 M−1s−1	1.7 × 106 M−1s−1
PfGrxE28W/C32S/C88S(SSG)	NAC	7.8 × 104 M−1s−1	1.6 × 107 M−1s−1
PfGrxE28W/C32S/C88S(SSG)	CA	1.8 × 105 M−1s−1	1.3 × 106 M−1s−1
PfGrxE28W/C32S/C88S(SSG)	2-ME	1.1 × 105 M−1s−1	1.8 × 107 M−1s−1
PfGrxE28W/C32S/C88S(SSG)	TGA	4.9 × 104 M−1s−1	1.2 × 107 M−1s−1
ScGrx7G107W(SSG)	GSH	2.6 × 105 M−1s−1	1.0 × 107 M−1s−1
ScGrx7G107W(SSG)	CysOEt	2.4 × 105 M−1s−1	5.4 × 105 M−1s−1
ScGrx7G107W(SSG)	Cys	1.2 × 105 M−1s−1	1.1 × 106 M−1s−1
ScGrx7G107W(SSG)	NAC	1.2 × 105 M−1s−1	2.4 × 107 M−1s−1
ScGrx7WT(SH)	GSSCys (+GSH)	7.8 × 105 M−1s−1a/7.1 × 105 M−1s−1b
ScGrx7WT(SH)	GSSCys (+GABA-SH)	1.8 × 105 M−1s−1a/1.6 × 105 M−1s−1b
ScGrx7WT(SSG)	GSH	3.3 × 105 M−1s−1a	3.6 × 106 M−1s−1
ScGrx7WT(SSG)	GABA-SH	6.3 × 105 M−1s−1a	5.6 × 106 M−1s−1
PfGrxE28W/C88S(S2)	GSH	(5–7) × 104 M−1s−1	(2–3) × 106 M−1s−1
PfGrxE28W/C88S(S2)	GSHOEt	4.2 × 103 M−1s−1	1.1 × 105 M−1s−1
PfGrxE28W/C88S(S2)	GABA-SH	5.1 × 102 M−1s−1	1.7 × 104 M−1s−1
PfGrxE28W/C88S(S2)	NAC	2.3 × 102 M−1s−1	3.7 × 104 M−1s−1
PfGrxE28W/C88S(S2)	CysGly	77 M−1s−1	3.2 × 102 M−1s−1

^aReciprocal Dalziel-coefficients at pH = 8.0.

^bk_cat/K_m values at pH = 8.0. All other rate constants were determined at pH 7.4.

3.2. Glutaredoxins require balanced glutathione interactions to prevent substrate inhibition

We previously showed that D144K and E147K charge inversions in helix 3 of ScGrx7 accelerated the reductive half-reaction with GSH but not the oxidative half-reaction with GSSCys, suggesting an altered recruitment of GSH to the 2nd glutathione-interaction site (Fig. 1C and D) [20]. To address this hypothesis, we performed steady-state measurements for wild-type ScGrx7 and ScGrx7^E147K with GSSCys and either GSH or GABA-SH, in which we replaced the γ-glutamyl residue with a positively charged γ-aminobutyryl group [45] (Fig. 3A). GSH and GABA-SH are thought to interact with the 2nd glutathione-interaction site during the reductive half-reaction [20,21,23]. However, because of their structural similarity, they could also compete with GSSCys for the 1st glutathione-interaction site (Fig. 3B). As shown previously [20,21,38], wild-type ScGrx7 catalysis follows ping-pong kinetics for GSSCys and GSH with k_cat^app/K_m^app values of 7.1 × 10⁵ M⁻¹s⁻¹ and 2.9 × 10⁵ M⁻¹s⁻¹, respectively (Fig. 3C–Table 1). While Lineweaver-Burk plots for GABA-SH also tended to be parallel, in accordance with ping-pong kinetics, a strong decrease of V/[E] was observed at higher GABA-SH concentrations (Fig. 3D). Increasing the GR concentration did not affect the reaction velocity and a limiting GR activity with disulfides of GABA-SH was excluded for the decreased activity. The kinetic pattern is therefore consistent with a substrate inhibition of the oxidative half-reaction by GABA-SH [50]. Accordingly, the inhibitory effect was especially pronounced at lower GSSCys concentrations. Furthermore, while the reductive half-reaction was rate-limiting for GSH, the oxidative half-reaction became rate-limiting for GABA-SH (Fig. 3E). In line with the low specificity of S-glutathionylated glutaredoxins for low-molecular-weight thiols, the k_cat^app/K_m^app values or reciprocal Dalziel coefficients for the reductive half-reactions of wild-type ScGrx7 with GSH and GABA-SH were rather similar with less than twofold increased values for GABA-SH. In contrast, the k_cat^app/K_m^app values or reciprocal Dalziel coefficients for the oxidative half-reaction with GSSCys were more than four times higher for GSH than in the reaction with GABA-SH (Fig. 3E, Fig. S3, Table 1). As reported previously [20], the steady-state kinetics for ScGrx7^E147K with GSH also revealed typical ping-pong patterns with a twofold increased catalytic efficiency for the reductive half-reaction as compared to wild-type ScGrx7 (Fig. 3F). In contrast, a very strong substrate inhibition was observed for the kinetics of ScGrx7^E147K with GABA-SH and no kinetic constants could be determined (Fig. 3G). The slower oxidative half-reaction and substrate inhibition of wild-type ScGrx7 and ScGrx7^E147K by GABA-SH can be explained by a competition between GABA-SH and GSSCys at the 1st glutathione-interaction site. The much stronger substrate inhibition of GABA-SH compared to GSH might either indicate a stronger binding of GABA-SH to the 1st glutathione-interaction site or a more efficient substrate recruitment due to the positively charged γ-aminobutyryl group. Because residue 147 is not directly involved in the 1st glutathione-interaction site (Fig. 1C) [20], the strong effects rather point to an altered recruitment of GABA-SH. In summary, class I glutaredoxins are optimized for efficient glutathione recognition and recruitment while preventing substrate inhibition at physiological GSH concentrations. This optimization can be unmasked by the GSH analogue GABA-SH.

Fig. 3.

GSSCys assay for wild-type ScGrx7 and ScGrx7^E147Kwith GSH or GABA-SH. A) Structures and color coding of GSH and GABA-SH. B) Schematic representations of the predicted transition state of the reductive half-reaction for the S-glutathionylated enzyme (left) and the inhibited reduced enzyme during the oxidative half-reaction (right). C) Michaelis-Menten plot (left) and Lineweaver-Burk plot (right) for wild-type ScGrx7 and variable concentrations of GSH at 25, 50, and 100 μM GSSCys (shown in lilac, light blue, and dark blue, respectively). D) Michaelis-Menten plot (left) and Lineweaver-Burk plot (right) for wild-type ScGrx7 and variable concentrations of GABA-SH at 50, 100, and 150 μM GSSCys (shown in rose, bright red, and dark red, respectively). Squares were omitted for regression analysis. E) Second-order rate constants from panels C and D calculated form averaged k_cat^app/K_m^app values (top) or reciprocal slopes from secondary plots in Fig. S3 (bottom). F) Lineweaver-Burk plot for ScGrx7^E147K and variable concentrations of GSH at 25, 50, and 100 μM GSSCys. G) Michaelis-Menten plot (left) and Lineweaver-Burk plot (right) for ScGrx7^E147K and variable concentrations of GABA-SH at 50, 100, and 150 μM GSSCys. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.3. The 1st glutathione-interaction site is selective for GSH as a reducing agent

Dithiol class I glutaredoxins can also form an intramolecular disulfide bond. Hence, the observed preference of some dithiol glutaredoxins for GSH as a reducing agent [19,27,36] might originate from the GSH-dependent reduction of the intramolecular disulfide bond. We therefore analyzed the reduction of diamide-oxidized dithiol PfGrx^E28W/C88S with GSH and related low-molecular-weight thiols (Fig. 4, Table 1). During the first part of the reductive half-reaction, these thiols are expected to interact with the 1st glutathione-interaction site in the absence of an enzyme-bound glutathione moiety (Fig. 4A) as revealed by structural comparisons between GSH-bound reduced glutaredoxins and glutathionylated glutaredoxins (Fig. S4). We used NAC, glutathione ethyl ester (GSHOEt), and the previously synthesized GSH-analogs L-cysteinyl glycine (CysGly) and GABA-SH (Fig. 4B), as well as Cys and CysOEt. Mixing of PfGrx^E28W/C88S disulfide in the first syringe with variable concentrations of the tripeptides GSH, GSHOEt, and GABA-SH in the second syringe resulted in a thiol-dependent increase in fluorescence (Fig. S5). Secondary plots of the k_obs values revealed a hyperbolic correlation for GSH, biphasic linear correlations for GSHOEt, and a linear correlation for GABA-SH (Fig. 4C–E). The different phases and y-axis intercepts in the secondary plots point towards deviating redox equilibria and backward reactions as discussed below. The apparent second-order rate constant from the initial slope of the hyperbola for GSH as well as the second-order rate constants from the linear fits for GSHOEt and GABA-SH decreased over two orders of magnitude for GSH > GSHOEt > GABA-SH. The highest value of (5–7) × 10⁴ M⁻¹s⁻¹ for the reduction of PfGrx^E28W/C88S disulfide by GSH was four to five times slower than the reduction of S-glutathionlated PfGrx^{E28W/C32S/C88S} (Table 1). The reactivity of PfGrx^E28W/C88S disulfide further decreased for NAC and CysGly as reflected in smaller amplitudes in the fluorescence traces (Fig. S5) as well as lower second-order rate constants (Fig. 4F and G). For Cys and CysOEt no significant change in fluorescence and no activity were detectable (Fig. S5, Fig. 4H).

Fig. 4.

Reductive half-reaction between PfGrx^E28W/C88Sdisulfide and various low-molecular-weight thiols. A) Schematic representation of the predicted first transition state of the reductive half-reaction. B) Structures and color coding of the tested thiols. C–H) Secondary plots of the k_obs values for the indicated thiols from the stopped-flow reduction kinetics in Fig. S5. All data sets were generated from at least three independent biological replicates with three technical replicates. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Next, we analyzed potential correlations between the apparent second-order rate constants for the reduction of PfGrx^E28W/C88S disulfide and the varied structural parameters of the low-molecular-weight thiols (Fig. 5). We identified five key elements that affect the rate constants and the enzyme-substrate interactions at the 1st glutathione-interaction site (Fig. 5A,B, Fig. S4). Removal of both the γ-glutamyl and the glycyl residue of GSH abolished enzymatic activity. Comparison of the rate constants for GSH and GABA-SH revealed that the loss of the γ-glutamyl carboxylate group decreased the second-order rate constant by two orders of magnitude (Fig. 4, Fig. 5C–Table 1). Additional loss of the γ-glutamyl amino group and shortening of the carbon chain in NAC resulted in a two-fold lower apparent second-order rate constant compared to GABA-SH. However, this small effect was reversed given the pH-independent rate constants, suggesting that the amino group and carbon chain of the γ-glutamyl residue have little effect for substrate turnover at the 1st glutathione-interaction site. Modification of the cysteinyl amino group and the formation of a rigid partial double bond are highly relevant for efficient enzyme reduction as revealed by the absent reactivity of Cys compared to NAC. Structural analysis (Fig. 5B, Fig. S4) and previous molecular dynamics simulations showed that the glycyl carboxylate group of glutathione interacts with the conserved lysyl active site residue [20]. Accordingly, alkylation of the glycyl carboxylate group decreased the rate constants of GSHOEt and CysOEt by at least one order of magnitude compared to GSH and CysGly, respectively (Fig. 4C,D,G,H, Fig. 5C–Table 1).

Fig. 5.

Correlations between the thiol structures and second-order rate constants for PfGrx^E28W/C88Sdisulfide reduction. A) Schematic representation of the predicted first transition state of the reductive half-reaction (top) and structures of the tested thiols with relevant functional groups highlighted (bottom). B) Representative structure of S-glutathionylated ScGrx2^C30S (PDB entry 3D5J) [30] with enzyme-glutathione interactions highlighted. C) Apparent second-order rate constants from Fig. 4 (top) and calculated pH-independent second-order rate constants (bottom) reveal crucial roles of the γ-glutamyl and glycyl carboxylate groups. P values from one way Welch-ANOVA analyses were calculated in R 4.4.1 (P ≤ 0.01: ∗∗; P ≤ 0.001: ∗∗∗).

In summary, while S-glutathionylated glutaredoxins are efficiently reduced by a variety of low-molecular-weight thiols including Cys and CysOEt, reduction of PfGrx^E28W/C88S disulfide by GSH is four to five times slower and highly benefits from the presence of the γ-glutamyl and glycyl carboxylate groups as well as modification of the cysteinyl amino group in GSH. Thus, the 1st but not the 2nd glutathione-interaction site of glutaredoxins is selective for GSH as a reducing agent.

3.4. Glutaredoxin reduction correlates with thermodynamic stability of the mixed disulfide

Elgán and Berndt previously analyzed the thermodynamic stability of mixed disulfides between class I EcGrx3^C14S/C65Y and various thiols [28]. To decipher the relevance of our identified enzyme-substrate interactions for glutaredoxin disulfide reduction, we compared the thermodynamic parameters for EcGrx3^C14S/C65Y with our rate constants for PfGrx^E28W/C88S (Fig. 6A). As mentioned above, the reduced thiols occupy the 1st glutathione-interaction site of PfGrx^E28W/C88S disulfide (Fig. 6B). Circular dichroism spectra of reduced and diamide-oxidized PfGrx^E28W/C88S were highly similar in accordance with a high similarity of the 1st glutathione-interaction site in both redox species (Fig. 6C). Plotting the Gibbs free energy contribution of the mixed disulfides of EcGrx3^C14S/C65Y (as compared to the reduced enzyme) [28] (ΔΔG) against the ΔG^‡ values derived from the pH-independent second-order rate constants for PfGrx^E28W/C88S disulfide (as obtained from the Eyring-Polanyi equation) showed a strong correlation (Fig. 6D). Thus, the enzyme-substrate interactions that facilitate efficient reduction of the intramolecular glutaredoxin disulfide are the same that increase the thermodynamic stability of the mixed disulfide.

Fig. 6.

Correlation between the ΔG^‡values for PfGrx^E28W/C88Sdisulfide reduction and the thermodynamic stability of mixed glutaredoxin disulfides. A) Structures and color coding of the tested thiols. B) Schematic representation of the predicted first transition state of the reductive half-reaction. C) Circular dichroism spectra of reduced and diamide-oxidized PfGrx^E28W/C88S. D) Correlation between the ΔG^‡ values derived from the second-order rate constants from Fig. 5 and the Gibbs free energy contribution ΔΔG of the mixed disulfides of EcGrx3^C14S/C65Y from Ref. [28]. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

4. Discussion

In contrast to the well-characterized enzyme-substrate interactions that determine the high specificity of glutaredoxins for glutathione disulfide substrates during the oxidative half-reaction, the specificity for GSH during the reductive half-reaction is poorly understood and appears to differ between glutaredoxins [19,27,36,51]. We would like to suggest that several of the observed differences can be explained by the enzyme-specific (ir)relevance of the dithiol mechanism, which has been deduced from mutagenesis studies, steady-state, stopped- and quenched-flow experiments [7,10,14,23,25,26,30,42,43], and has also been modeled for reactions of glutaredoxins with glutathionylated substrates [52,53]. Using PfGrx^E28W/C88S as dithiol and PfGrx^{E28W/C32S/C88S} or ScGrx7^G107W as monothiol class I glutaredoxins, we show that intramolecular glutaredoxin disulfides and S-glutathionylated glutaredoxins differ greatly in their specificity for the reducing thiol substrate. While S-glutathionylated monothiol PfGrx^{E28W/C32S/C88S} and ScGrx7^G107W both lack GSH-specificity and react efficiently with a variety of low-molecular-weight thiols, the reduction of PfGrx^E28W/C88S disulfide reveals highly specific interactions with the carboxylate groups of GSH. The lack of specificity of S-glutathionylated glutaredoxins is also reflected by the efficient reaction between this enzyme species (but not the intramolecular glutaredoxin disulfide) and reduced redox-sensitive fluorescent proteins [23,43]. After the initial reaction with glutathionylated substrates (such as GSSCys or S-glutathionylated proteins), dithiol glutaredoxins can either react with a reducing thiol or release GSH and form an intramolecular disulfide (Fig. 1B). Thus, dithiol glutaredoxins can partition between the S-glutathionylated enzyme species that is efficiently reduced by various thiols, using the 2nd glutathione-interaction site, and the intramolecular disulfide species that is highly specific for GSH, using the 1st glutathione-interaction site (Fig. 7A). This difference in specificity could explain seemingly contrasting results from previous studies: On the one hand, the reduction of glutathionylated substrates by rat liver glutaredoxin with various low-molecular-weight thiols [51] resembles the low specificity of S-glutathionylated glutaredoxins from our stopped-flow experiments. Similar results were obtained from steady-state assays with human Grx1 and Grx2, although an increase in catalytic efficiency of one order of magnitude was observed for GSH compared to other related thiols such as CysGly [19,36]. The slightly higher specificity of S-glutathionylated human Grx1 and Grx2 compared to PfGrx and ScGrx7 might reflect an altered GSH recruitment or small differences of the transition state involving the 2nd glutathione-interaction site [20,21]. On the other hand, formation of the intramolecular glutaredoxin disulfide promotes GSH specificity independently of the S-glutathionylated enzymes. This is in agreement with the loss of GSH specificity upon removal of the resolving cysteinyl residue in the CxxC-motif of EcGrx1^C14S [27]. Based on our data, the GSH-specificity of wild-type EcGrx1 in contrast to EcGrx1^C14S does not necessarily imply an altered substrate recognition. Rather, EcGrx1 might predominantly form the intramolecular disulfide after reaction with a glutathionylated substrate, which is prevented by the removal of the second cysteinyl residue. In contrast to wild-type EcGrx1, P. falciparum, mammalian, and yeast dithiol class I glutaredoxins do not appear to readily form an intramolecular disulfide, resulting in a low specificity for GSH (Fig. 2, Fig. 7A) [23,27]. A higher tendency to form the intramolecular disulfide has also been proposed to explain the lower activity of ScGrx1 compared to ScGrx2 [30]. Based on our model, we would also like to question the dithiol mechanism that was suggested for the thioredoxin reductase-dependent reduction of human Grx2 [54]. Since the monothiol mutant hGrx2^C40S was more efficiently reduced by thioredoxin reductase than the dithiol wild-type enzyme [54], and because of the reduced substrate specificity of S-glutathionylated glutaredoxins, we would like to suggest that the observed steady-state activities of hGrx2 with RSSG substrates and thioredoxin reductase rather reflect a monothiol than a dithiol mechanism. In summary, the ability of glutaredoxins to react with alternative reducing agents depends on the S-glutathionylated enzyme species, whereas the specificity of dithiol glutaredoxins for GSH as a reducing agent is determined by the extent of intramolecular disulfide bond formation (Fig. 7A).

Fig. 7.

Differences in substrate specificity and reaction mechanism for monothiol and dithiol glutaredoxins. A) S-glutathionylated glutaredoxins react with various thiols at the 2nd glutathione-interaction site. Formation of the intramolecular glutaredoxin disulfide (side-reaction) enables the utilization of the highly GSH-specific 1st glutathione-interaction site. B) The rapid reaction of the intramolecular glutaredoxin disulfide with GSH is facilitated by a conformational strain of the disulfide bond and specific enzyme-substrate interactions at the 1st glutathione-interaction site. C) The rapid reaction of S-glutathionylated glutaredoxins with various thiols is facilitated by the low pK_a value of the leaving group (resulting in an asymmetric transition state) and efficient substrate recruitment to the 2nd glutathione-interaction site.

Our kinetic data also provide novel mechanistic insights regarding the efficient GSH-dependent reduction of the intramolecular glutaredoxin disulfide. With an apparent pH-independent rate constant of 2 × 10⁶ M⁻¹s⁻¹, the reaction of GSH with PfGrx^E28W/C88S disulfide is markedly accelerated compared to common nonenzymatic thiol-disulfide exchange reactions. Generally, this could be facilitated by specific enzyme-substrate interactions involving the 1st glutathione-interaction site and/or an enhanced intrinsic reactivity of the intramolecular glutaredoxin disulfide (Fig. 7B). Comparing the calculated ΔG^‡ values for this reaction and the thermodynamic stability of the different mixed disulfides (ΔΔG) [28] provides the opportunity to quantify both contributing factors (with the caveat that our ΔG^‡ and the ΔΔG values are from two different glutaredoxins). The formation of the mixed disulfide with GABA-SH or NAC does not significantly stabilize or destabilize the enzyme (ΔΔG ≈ 0) [28]. The ΔG^‡ values (calculated using the Eyring-Polanyi equation with the pH-independent rate constants) for the reaction of PfGrx^E28W/C88S disulfide with NAC and GABA-SH are around 40 kJ/mol (Fig. 6D). Assuming a pH-independent rate constant of 10–100 M⁻¹s⁻¹ for nonenzymatic thiol-disulfide exchange reactions [55] yields a ΔG^‡ value of approximately 60 kJ/mol. Thus, the ΔG^‡ value for the reaction between PfGrx^E28W/C88S disulfide and NAC or GABA-SH is lowered by about 20 kJ/mol without significant enzyme-substrate interactions. The geometry of the intramolecular disulfide most likely contributes to this high reactivity (Fig. 7B). We therefore analyzed twelve glutaredoxin crystal or NMR structures revealing a C–S–S–C χ³ angle [56,57] of approximately 70°. This is significantly less than the lowest energy conformation of ±83° [57]. It was estimated that the deviation from the optimal angle increases the conformational energy of the intramolecular EcGrx1 disulfide around 15 kJ/mol [58]. This value is highly similar to the reduction in ΔG^‡, which we ascribe to the increased reactivity of the intramolecular PfGrx^E28W/C88S disulfide. A good correlation between ΔG values and strain energies was also observed for reactions of dithiols and cyclic disulfides [59]. Compared to NAC and GABA-SH, S-glutathionylation of the glutaredoxin leads to a thermodynamic stabilization [28]. Likewise, the ΔG^‡ value for the reaction of GSH with PfGrx^E28W/C88S disulfide of about 30 kJ/mol is further lowered by approximately 10 kJ/mol compared to NAC and GABA-SH. Thus, an increased intrinsic reactivity of the glutaredoxin disulfide as well as specific enzyme-substrate interactions at the 1st glutathione-interaction site both lead to a lowered ΔG^‡ value and therefore faster enzyme catalysis (Fig. 7B). Altered enzyme-substrate interactions at the 1st glutathione-interaction site could also explain the different kinetic patterns for the reaction between the intramolecular disulfide of PfGrx^E28W/C88S and different thiols (Fig. 4C–G). The hyperbolic dependence of k_obs on the GSH concentration (Fig. 4C) can be explained by a two-step binding model with a rapid equilibrium involving the enzyme and GSH prior to the reaction step [60,61]. The form of the hyperbola depends on the individual rate constants of the reaction steps and the dissociation constant of the complex between GSH and the glutaredoxin [61]. A higher dissociation constant for the transient complex between PfGrx^E28W/C88S and GABA-SH, NAC, or CysGly could explain the conversion of the hyperbola for GSH to a linear function [60,61], with GSHOEt having intermediate affinity (Fig. 4D–G). The gradual increase in dissociation constant resembles the trend in apparent second-order rate constants and also the available ΔΔG values of the different thiol substrates with GSH > GSHOEt > GABA-SH ≈ NAC > CysGly. Kinetic models for PfGrx from stopped-flow measurements revealed a dissociation constant for GSH at the 1st glutathione-interaction site k_off/k_on ≈ 10³ s⁻¹/10⁷ M⁻¹s⁻¹ ≈ 0.1 mM [23], which is in accordance with the GSH concentration at half-maximum velocity in Fig. 4C. Although the value is lower than the dissociation constant k_off/k_on ≈ 10⁴ s⁻¹/(6 × 10⁵ M⁻¹s⁻¹) ≈ 15 mM for human Grx1 from NMR measurements [31], both enzyme characterizations reveal that glutaredoxins only form transient enzyme-substrate complexes with GSH at the 1st glutathione-interaction site. This is a prerequisite to prevent substrate inhibition at physiological GSH concentrations, which was unmasked when we replaced GSH with its analogue GABA-SH. In summary, a partially strained disulfide bond and specific but highly transient interactions at the 1st glutathione-interaction site together accelerate the GSH-dependent reduction of the intramolecular disulfide bond of dithiol glutaredoxins (Fig. 7B).

What can we learn from the high Brønsted coefficients for the reduction of our S-glutathionylated glutaredoxins? The rather small or absent deviation of the pH-independent k_obs values for GSH in the Brønsted plots for PfGrx^{E28W/C32S/C88S} and ScGrx7^G107W reveal a small or even absent specificity for GSH at the 2nd glutathione-interaction site. The lack of specificity is similar to the promiscuous reduction of the sulfenic acid of several 1-Cys peroxiredoxins [45,62,63]. Furthermore, the β_nuc values for S-glutathionylated PfGrx^{E28W/C32S/C88S} and ScGrx7^G107W of 0.80 and 0.65 are substantially higher than the β_nuc values for the reaction between thiolates and H₂O₂ around 0.27, [64] the sulfenic acid of the peroxiredoxin PfAOP of 0.36, [45] or small disulfides around 0.5 [55,65]. Although the interpretation of β_nuc values remains controversial [55,64,66], higher β_nuc values could imply a transition state with a much higher partial negative charge on the sulfur atom of the active-site cysteinyl residue and more asymmetric charge distribution for the deglutathionylation of glutaredoxins compared to previously studied thiol-disulfide exchange reactions (Fig. 7C) [55,65]. Consistent with this interpretation, the active-site cysteinyl residue of class I glutaredoxins has an unusually low pK_a value between 3 and 5 [18,20,30,36,41], which could stabilize the negative charge and cause the asymmetric transition state. The low pK_a value of the active-site thiol has already been identified as the major driving force behind the efficient reduction of S-glutathionylated glutaredoxins by providing an ideal leaving group [4,19,36,37]. A possible alternative explanation for higher β_nuc values (without altered transition state structures) could be electrostatic interactions, which can be more pronounced when one reaction partner is a protein instead of a small molecule (Fig. 7C) [66]. For example, the reaction of positively charged cystamine with various charged peptide thiols gave similarly high β_nuc values of 0.75–0.80 despite providing a leaving group with a high pK_a value [66]. An altered GSH recruitment was also suggested as an explanation for the improved activity of charge inversion mutants of ScGrx7 [20]. A better recruitment of GABA-SH to the active site could actually explain the stronger substrate inhibition compared to GSH (Fig. 3C and D). As outlined above, GSH only transiently occupies the 1st glutathione-interaction site of reduced glutaredoxins [23,31], and residue 147 of ScGrx7, which is not part of this interaction site, is likely involved in the recruitment of GSH (Fig. 1C) [23]. Thus, it seems unlikely that differences in binding strength at the 1st glutathione-interaction site explain the much stronger substrate inhibition of ScGrx7^E147K by GABA-SH. More kinetic assays and MD as well as QM/MM simulations with additional GSH analogs, mutants, and mixed disulfides are required to unambiguously address the cause(s) for the high Brønsted coefficients of S-glutathionylated glutaredoxins, the degree of substrate inhibition by GSH and analogs such as GABA-SH, and to fully understand the balance of substrate binding and turnover in both half-reactions.

Why do monothiol and dithiol class I glutaredoxins exist? Traditionally, the monothiol and dithiol mechanism have been viewed as alternative mechanisms for glutathionylated substrates (such as GSSCys and S-glutathionylated proteins) and (intra- or intermolecular) protein disulfides, respectively [[4], [5], [6],17]. For example, specific protein-protein interactions and, as discussed above, the propensity of EcGrx1 to form an intramolecular disulfide explain why EcGrx1 uses a dithiol mechanism for the reduction of E. coli ribonucleotide reductase and 3′-phosphoadenosine 5′-phosphosulfate reductase [11,12]. However, the GSH-dependent reduction of several protein disulfides does not require a (much slower) dithiol mechanism [23,43,44], and a single monothiol class I glutaredoxin can replace all the essential functions of cytosolic thioredoxins and dithiol class I glutaredoxins in S. cerevisiae (including the reduction of ribonucleotide reductase and methionine sulfoxide reductase) [32]. According to the cysteine-resolving model, the second cysteinyl residue and another semi-conserved cysteinyl residue in a GGC-motif could resolve unspecific mixed glutaredoxin disulfides that are kinetically trapped and cannot be attacked by GSH (Fig. 1E) [5,11,17,22,23,48,67]. This model could be particularly applicable if the bound residue is a bulky protein. It was also suggested that the intramolecular disulfide of EcGrx1 is predominantly formed at very high GSSG and very low GSH concentrations, thus protecting EcGrx1 itself and preventing the deglutathionylation of other protected cysteinyl residues [26,53]. Here we present another explanation for the prevalence of dithiol class I glutaredoxins (which does not exclude the other hypotheses): The dithiol mechanism promotes specificity for GSH as a reducing agent. The use of GSH as a reducing agent at the 1st glutathione-interaction site not only taps directly into the glutathione pool of eukaryotes and many prokaryotes [68,69], but also prevents the formation of unspecific mixed glutaredoxin disulfides [23]. How then can yeast cells (and probably other organisms) survive without a dithiol glutaredoxin? An alternative monothiol mechanism can also employ the 1st glutathione-interaction site for the GSH-specific reduction of protein disulfides in a ternary complex [23]. A crucial difference between the dithiol mechanism and the ternary complex monothiol mechanism is the specificity for the protein disulfide substrate. While some glutaredoxins such as EcGrx1 could have a few specific protein disulfide substrates that are reduced by the dithiol mechanism, a ternary complex with GSH could be employed for various unspecific protein disulfide reductions, thus enabling the survival of mutant yeast strains with monothiol glutaredoxins.

CRediT authorship contribution statement

Lukas Lang: Writing – review & editing, Writing – original draft, Investigation, Formal analysis, Conceptualization. Philipp Reinert: Writing – review & editing, Investigation, Formal analysis. Cedric Diaz: Writing – review & editing, Investigation, Formal analysis. Marcel Deponte: Writing – review & editing, Writing – original draft, Supervision, Resources, Conceptualization.

Declaration of competing interest

The authors declare no competing interests. There are no financial/personal interests or beliefs that could affect our objectivity or result in a potential conflict.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Data will be made available on request.