Biochemical Characterization of Yeast Mitochondrial Grx5 Monothiol Glutaredoxin * .

of reaction of


Summary
Grx5 is a yeast mitochondrial protein involved in iron-sulfur biogenesis that belongs to a recently described family of monothiolic glutaredoxin-like proteins. No member of this family has been previously biochemically characterized. Grx5 contains a conserved cysteine residue (Cys60) and a non-conserved one (Cys117). In this work we have purified wild type and mutant C60S and C117S proteins and characterized their biochemical properties. A redox potential of -175 mV was calculated for WT-Grx5. The pK a values obtained by titration of mutant proteins with iodoacetamide at different pHs were 5.0 for Cys60 and 8.2 for Cys117. When Grx5 was incubated with glutathione disulfide a transient mixed disulfide was formed between glutathione and the cystein 60 of the protein due to its low pK a . Binding of glutathione to Cys60 promoted a decrease in Cys117 pK a value that triggered the formation of a disulfide bond between both cysteine residues of the protein indicating that Cys117 plays an essential role in the catalytic mechanism of Grx5. The disulfide bond in Grx5 could be reduced by GSH but at a rate at least 20 times slower than that observed for reduction of glutaredoxin 1 from E. coli, a dithiolic glutaredoxin. This slow reduction rate could suggest that GSH

Introduction
Glutaredoxins are small proteins with thiolreductase activity that are required for maintaining protein cysteines in reduced form. In contrast to thioredoxins, glutaredoxins require the reduced form of glutathione (GSH) 1 as electron donor (1)(2)(3). Previously characterized glutaredoxins contain an active site that includes two conserved cysteine residues with two non-conserved residues between them (4)(5)(6). Mutagenic studies have shown that both residues are required for reducing protein disulfides. However, only the amino-terminal cysteine may be essential for the reduction of mixed disulfides of proteins with glutathione (6)(7)(8). In Saccharomyces cerevisiae five different glutaredoxins have been described. Two of them (Grx1-2) are classic dithiolic glutaredoxins containing both conserved cysteine residues and have been already biochemically characterized (9)(10)(11). On the basis of sequence analysis, a new family of monothiolic glutaredoxins has been recently described. These proteins are highly homologous to glutaredoxins but contain only one cysteine residue in its putative active site (12). Members of this family are found elsewhere, from bacteria to mammals, including human (13). To date, none of them has been properly biochemically characterized. 1 Three monothiolic glutaredoxins are found in yeast (Grx3- 5). No clear phenotypes have been described in yeast cells lacking Grx3 and Grx4 and consistently no specific role has been assigned to any of these proteins. In contrast, the absence of Grx5 induces severe growth defects (12). Cells lacking Grx5 are not able to grow on minimal medium nor in the presence of non-fermentable carbon-sources, they accumulate iron in the mitochondria and show decreased activities of iron-sulfur containing enzymes. These characteristics are common to other genes involved in the synthesis and assembly of Fe/S clusters such as SSQ1, JAC1, ATM1, NFU, YAH1, ARH1, ISU1-2 (14), ISA1-2 (15) NFS1, YFH1 (16) and ERV1 (17). Recently, we have shown that Grx5 is a mitochondrial protein involved in iron-sulfur biogenesis (18). A three dimensional model of Grx5 was recently presented based on the known structure of several dithiolic glutaredoxins (13). Grx5 shows a classic thioredoxin fold structure, with the putative catalytic cysteine (Cys60) lying opposite to another conserved motif that could be involved in the formation of a glutathione cleft. Beside this motif, another non-conserved cysteine is found (Cys117). Site directed mutagenesis studies suggest that this cysteine is not essential for the biological activity of the protein (13).
Despite these observations there is no evidence that Grx5 works as a thiolreductase.
Also, the specific role of Grx5 in iron-sulfur biogenesis is still not clear. Shenton et al.
showed that in cells lacking Grx5 the cytosolic enzyme glyceraldehyde-3-phosphatedehydrogenase was glutathiolated and suggested that Grx5 could work as a deglutathiolase (19). However the recent finding that Grx5 is a mitochondrial enzyme (18) suggests that this glutathiolation may be rather related to the oxidative stress conditions generated by iron accumulation in ∆grx5 cells than to the direct effect of Grx5, a mitochondrial protein, on glyceraldehyde-3-phosphate-dehydrogenase, a by guest on July 9, 2020 http://www.jbc.org/ Downloaded from cytosolic enzyme. In this work we address the biochemical characterization of Grx5, including determination of cysteine pK a value and redox potential. Based on these results we propose a mechanism of action for Grx5 protein. This is the first characterization of a monothiolic glutaredoxin and constitutes the first evidence that these proteins can work as thioloxidoreductases.
Strains and plasmids-Plasmid pMM192 contains the GRX5 open reading frame without the region coding from amino acid 2 to 29 (PCR-amplified from S. cerevisiae genomic DNA), cloned between the Nde I and BamH I unique sites of the E. coli expression vector pET-21a ( Novagen). Point mutations in GRX5 that yielded the different amino acid replacements were constructed by the ExSite method (20), using pMMM192 as template. Oligonucleotides for the introduction of the point mutations were designated in such a way that a restriction site that did not alter the translation product was introduced near to the desired point mutation and used as a marker for the DNA sequencing. Plasmids were maintained and amplified in E. coli BL21 cells (Novagen). Analyses-Protein concentration was determined by the Bradford method (21).
Titration of free sulfhydryl groups with 5,5'-dithiobis-2-nitrobenzoic acid (DTNB) was performed as described (22). Briefly, 20-50 µg of protein were incubated for 10 minutes Activities-Reduction of the mixed disulfide formed between HED and glutathione (low molecular weight mixed disulfide reduction assay or HED assay) was assayed as At different times, 23 µl of the preparation were removed through the septum of the tubes using a degassed 25 µl Hamilton gastight syringe and mixed with 2 µl of 10% trifluoroacetic acid to stop the reaction. Protein mixtures were analyzed by HPLC as described above except that a linear gradient from 30 to 50 % acetonitrile was used.

Results
Expression and purification of Grx5 proteins-In a previous work we had shown that  (Table I).
Grx5 is not active in the HED assay-Several reactions can be catalysed by dithiolic glutaredoxins. The most widely used form to asses glutaredoxin activity is the glutathione:HED transhydrogenase assay. In this assay glutaredoxin catalyses the reduction of a mixed disulfide between glutathione and HED (2). Dithiolic glutaredoxins lacking one of the two conserved cysteines are still capable to catalyse this reaction (6). Glutaredoxin activity of WT-Grx5, C117S-Grx5 and C60S-Grx5 was assayed with the HED assay. No activity could be detected even when a wide range of pH values (7-9.5) and GSH concentrations (0.6-40 mM) were used. Additionally, dehydroascorbate reductase and glutathione peroxidase activities, which have been described for dithiolic glutaredoxins (23,24), were also tested. None of the Grx5 variants showed detectable activity in these assays.
Determination of Cys60 and Cys117 pK a value-Reactivity of thiol groups in proteins highly depends on its pK a value. Active cysteines from dithiolic glutaredoxins have pK a values close to 4 (9,26,27). In order to determine the pK a value of both cysteine residues in Grx5 we measured the rate of alkylation of Grx5 with iodoacetamide at different pHs. This reaction occurs only when cysteines are in the ionized thiolate anion state (28). Thus, reduced WT, C117S and C60S Grx5 proteins were incubated with 0. Reaction rates showed a sigmoidal dependence on pH value at pHs around 5 (WT and C117S) and 8 (WT and C60S). From these data it can be deduced that the increases in reaction rates at low and high pH were respectively a consequence of the ionisation of Cys60 and Cys117. Using the Henderson-Hasselbach equation (28) thiol pK a values of 5.0 ± 0.1 and 8.2 ± 0.1 were calculated for Cys60 and Cys117 respectively ( Table I).
Reaction of Grx5 with GSSG-Reactivity of reduced Grx5 with GSSG was tested at pH 8.0 in 0.1M Tris-HCl buffer because mitochondrial pH is close to this value (30). WT-Grx5 was incubated with increasing concentrations of GSSG for 15 minutes at 20ºC and the products of the reaction were separated by HPLC. Figure 2 A shows that four new peaks corresponding to oxidized forms of the protein appeared. When the mutant proteins were incubated with GSSG, only one new peak appeared (Fig. 2 B and C).
The characterization of these peaks is summarized in Table II. From mass spectrometry data it can be deduced that two additional glutathione molecules were present in peak 1 compared to peak 5 (reduced form) while only one additional glutathione molecule was present in both peak 3 and 4. Peak 1 corresponded to a protein glutathiolated at both cysteines while peak 2 was a protein presenting a disulfide bond between both cysteines. This was deduced from the following observations: i) no free thiols were detected when peak 2 protein was incubated with DTNB ; ii) it presented a mass of 13,484 Da (as the reduced form); iii) it was also the major peak obtained when reduced Grx5 was incubated with several oxidants such as H 2 O 2 , cystine or oxidized proteins; iv) peak 2 was the end product obtained either from peak 3 or from peak 4 when these peaks were collected, dried and solubilized at pH 8.0; v) reduced Grx5 (peak 5) was obtained by incubation of peak 2 with DTT.
Concerning the monoglutathiolated forms of the protein, our results indicate that peak 3 corresponded to a protein glutathiolated at Cys60 while peak 4 corresponded to a protein glutathiolated at Cys117. According to the pK a values previously calculated for Cys60 and Cys117, the pH dependence of the appearance of peak 3 and 4 was consistent with this assumption (Fig 3 A). This was confirmed by analyzing the rate of glutathiolation of C117S and C60S mutant proteins at different pHs (Fig. 3 B).
One interesting result was the observation that the rates of carboxymethylation and glutathiolation did not follow the same pH dependence in the WT protein, while in mutant proteins they were nearly the same (compare figures 1 and 3). Thus, introduction of the first glutathione in the WT protein may increase the reactivity of the Cys117. This suggested the idea that one glutathione molecule could be transferred from Cys60 to Cys117 in the wild type protein during the reaction, being glutathiolation of Cys117 an intermediate step before formation of the disulfide bond. To test this hypothesis peak 4 was collected, dried, rehydrated with Tris-HCl buffer at pH 8.0, and incubated at 4ºC. Figure 4 shows the percentage of each peak found at different incubation times as determined by HPLC analysis. It can be observed that peak 3 appeared mainly at short incubation times, as it would be expected for an intermediary product of the transformation of peak 4 to peak 2. µM and 1.34 mM (Fig. 2 D). These results reinforced the idea that interaction between both cysteines occurred, and that the presence of Cys117 enhanced Grx5 reactivity.
Nevertheless reactivity of Cys117 alone (in the C60S protein) was very poor.  (Fig. 5 A). In addition, we compared the rate of reduction of the oxidized Grx5 with that of Grx1 from E. coli, a dithiolic glutaredoxin active on the HED assay. Both proteins were incubated at fixed concentrations of 1 and 2 mM GSH for different times. Figure 5 B shows that even at the shorter incubation times (30 seconds) the reaction of Grx1 with GSH reached the equilibrium. Instead, reaction of Grx5 with GSH required 1 hour to reach the equilibrium. Thus, the rate reduction of Grx5 was at least 20 times slower than that of Grx1. These results indicated that reduction of Grx5 by GSH can be a limiting step for its thiolreductase activity. The absence of detectable HED activity in monothiolic glutaredoxins may thus be related to the unefficient reduction of these proteins.

Reduction of Grx5 by GSH-
Determination of the redox potential of Grx5-The redox potential of Grx5 was determined by direct protein-protein equilibration with E. coli Grx1 (25). Reduced Grx5 and oxidized Grx1 were incubated at 25ºC under anaerobic conditions. HPLC separation and quantification of the four protein species was performed after incubation for 1, 2, 4, 8 and 12 hours (Fig. 6). The redox equilibrium was obtained after 4 hours of incubation, as indicated by a stable ratio of the four protein species. Same results were obtained when oxidized Grx1 and reduced Grx5 were used as the starting material.
The redox potential of Grx5 was calculated from the following Nernst equation:

Discussion
Grx3, 4 and 5 from S. cerevisiae were the first described members of a new family of proteins with glutaredoxin signature. These proteins contain one conserved cysteine residue at the putative active site (12), and they have been found in all types of organisms from bacteria to humans (13). Very few of them have been studied and only two of them have an assigned function. The human PICOT protein has been proposed to be a modulator of the protein kinase C-θ pathway (35). We have recently shown that Grx5 from yeast is located in the mitochondria and is involved in the maturation of Fe/S cluster-containing proteins (18). The glutaredoxin-like protein GLP1 from P. falciparum has also been cloned and purified, but it has no specific assigned role (31).
Despite these observations, there was no consistent biochemical data supporting the involvement of monothiolic glutaredoxins in thiol redox reactions and consequently no mechanism of action had been proposed for the members of this family. Bushweller et al. described that mutant dithiolic glutaredoxins lacking the second conserved cysteine residue were still able to catalyse the reduction of the HED-GSH mixed disulfide (6).
The mechanism proposed for this reaction (summarized in figure 8 A) involved the formation of a mixed disulfide between glutathione and the cysteine located at the active site. This mixed disulfide could be cleft by GSH yielding reduced glutaredoxin and GSSG. It has been suggested that monothiolic glutaredoxins could follow this same scheme (3). However, this was a controversial issue. First, Rahlfs et al. purified and partially characterized PfGLP1 from P. falciparum and concluded that it could not be reduced by GSH (31). However, this was probably because PfGLP1 was already reduced after purification, as occurs with Grx5. Second, neither Grx5 nor PfGLP1 are active in the HED assay, although dithiolic glutaredoxins lacking the C-terminal cysteine are still active in this assay (31, 19).
The results from this work demonstrate that Grx5 is a thiolreductase that can participate in thiol redox reactions. Several evidences support this idea: first, Cys60 presents a low pK a , close to the pK a values of reactive cysteines in dithiolic glutaredoxins (9,26,27); second, Grx5 has the potential to form a mixed disulfide with glutathione with high affinity; finally, Grx5 has the ability to reduce a glutathiolated protein such carbonic anhydrase, indicating that its redox potential is low enough to act as an electron donor in redox reactions involving oxidized proteins. We propose a mechanism of action for the reduction of mixed disulfides by Grx5 based on the reaction of Grx5 with GSSG (summarized in figure 8 B). First, a mixed disulfide will be formed between Cys60 and glutathione. This would induce a decrease in the Cys117 pK a value that will trigger the formation of a disulfide bond between both cysteines and yield reduced glutathione. However, it is not clear how Grx5 may be reduced in vivo because the reduction rate of Grx5 by GSH may not be fast enough to allow the efficient reduction of oxidized Grx5. Thus, involvement of other mitochondrial reducing agent(s) in this last step should be considered in further investigations. In this context it is interesting to note that E. coli thioredoxin efficiently reduces Grx5 (data not shown).
Finally, it should be noted that the absence of activity of Grx5 in the HED assay may be a consequence of its inefficient reduction by GSH, but also of its redox potential that would not be low enough to efficiently reduce the mixed HED-GSH disulfide. The redox potential of Grx5 (-175mV) is higher than that of dithiolic glutaredoxins, which range from -198 to -233 mV (25). However, it can be low enough to reduce other disulfide bonds, as indicated by our results with glutathiolated carbonic anhydrase.
Another important conclusion derived from this work is the relevance of Cys117 for in ∆grx5 yeast cells was investigated (13). Although being less efficient than WT-Grx5, the C117S protein may display enough activity to suppress the severe growth defects found in a ∆grx5 strain by the monothiolic mechanism described in figure 8 A.
Genetic and biochemical results obtained with yeast cells depleted in Grx5 have linked this protein to the process of iron-sulfur assembly (18). Now it is clear that its role may be related to its thiolreductase activity. However, its physiological substrate remains unknown. Several steps in the process of Fe/S assembly may require the presence of a thiol reductase. Recent works in this field indicate that the bacterial proteins IscU and IscA (homologous to Isa and Isu proteins in yeast) serve as scaffolds for the assembly of iron/sulfur clusters (36,37). The first step in this process is a sulfur transfer from the cysteine desulfurase IscS (NifS in yeast) to IscU or IscA (38,39). Later, iron is incorporated and a transient [2Fe2S] center is formed in IscA/U proteins. Although the exact mechanism is still controversial, it seems clear that reducing equivalents required for this process would be provided by the formation of a disulfide bond between two cysteines in IscA/U and/or IscS proteins (40,41). Grx5 would be required for the reduction of these cysteine residues and constitute an essential enzyme for the turnover of the whole process. Another possibility may consider that Grx5 would act as a general mitochondrial thiolreductase, being one of the steps in iron-sulfur assembly more dramatically affected for its absence than any other biological process. However, it is important to note that Grx5 is not the most abundant thiolreductase in mitochondria, where the presence of thioredoxin 3 and Grx2 have also been described (11,42). Thus, a specific role for Grx5 seems quite possible. Further research will determine whether this specificity is a consequence of Grx5 redox potential or of the recognition by Grx5 of specific regions in target proteins.         by guest on July 9, 2020